Chemical Degradation of Nafion Membranes under Mimic Fuel Cell

Aug 5, 2010 - Open Access .... Membrane Degradation Correlating Accelerated Stress Testing and Lifetime .... Journal of Power Sources 2016 304, 207-23...
0 downloads 0 Views 3MB Size
J. Phys. Chem. C 2010, 114, 14635–14645

14635

Chemical Degradation of Nafion Membranes under Mimic Fuel Cell Conditions as Investigated by Solid-State NMR Spectroscopy Lida Ghassemzadeh,†,‡ Klaus-Dieter Kreuer,† Joachim Maier,† and Klaus Mu¨ller*,§ Max-Planck-Institut fu¨r Festko¨rperforschung, Heisenbergstrasse 1, D-70569 Stuttgart, Germany, Institut fu¨r Physikalische Chemie, UniVersita¨t Stuttgart, Pfaffenwaldring 55, D-70569 Stuttgart, Germany, and Dipartimento di Ingegneria dei Materiali e Tecnologie Industriali, UniVersita` degli Studi di Trento, Via Mesiano 77, I-38100, Trento, Italy and INSTM, UdR Trento, Italy ReceiVed: March 20, 2010; ReVised Manuscript ReceiVed: July 21, 2010

A new ex situ method has been developed to mimic the degradation of the polymer membranes in polymer electrolyte membrane fuel cells (PEMFCs), caused by the cross-leakage of H2 and O2. In this ex situ setup, it is possible to expose membranes to flows of different gases with a controlled temperature and humidity. H+-form Nafion films with and without an electrode layer (Pt) have been treated in the presence of different gases in order to simulate the anode and cathode side of a PEMFC. The changes of the chemical structure occurring during the degradation tests were primarily examined by solid-state 19F NMR spectroscopy. For completion, liquid-state NMR studies and ion-exchange capacity measurements were performed. The molecular mobility changes of the ionomer membrane upon degradation were examined for the first time by variabletemperature 19F NMR line-shape, T1 and T1F relaxation experiments. It was found that degradation occurs only when both H2 and O2 are present (condition of gas cross-leakage) and when the membrane is coated with a Pt catalyst. The chemical degradation rate is found to be highest for H2-rich mixtures of H2 and O2, which corresponds to the anode under OCV conditions. It is further shown that side-chain disintegration is very important for chemical degradation, although backbone decomposition also takes place. The temperaturedependent line-width and spectral anisotropy alterations were explained by the reduction of static disorder in the Nafion membrane. From the relaxation data, there is evidence for structural annealing, which is independent of the chemical degradation. Chemical degradation is considered to reduce the chain flexibility, as expressed by smaller motional amplitudes, most probably due to chain cross-linking. Introduction The most significant problem for commercialization of polymer electrolyte membrane fuel cells (PEMFCs) is their relatively short lifetime. Extension of the lifetime for all fuel cell components is, therefore, absolutely mandatory for the use of fuel cells as economical alternative energy conversion systems, with high reliability and low maintenance costs. It has been demonstrated that the failure of fuel cells results from the interplay of different degradation routes within the various fuel cell components. Hence, dissolution of the platinum catalyst particles,1-3 corrosion of the carbon catalyst support,4-7 delamination of the electrode and membrane layers,8,9 decrease of the electrochemically active surface areas of the catalyst, and the increase of the Pt particle size10 are frequently discussed. However, a major reason for the fuel cell failure is the degradation of the polymer electrolyte membrane. There are several requirements that have to be met for a successful application as fuel cell membranes.11-13 Sulfonated fluoropolymers or aromatic polymers were found to be suitable candidates, out of which Nafion (see Figure 1) is still a kind of benchmark material.14 When it comes to membrane breakdown, three contributions, namely, mechanical, thermal, and chemical degradation, are * To whom correspondence should be addressed. Phone: (+39) 0461 282439. Fax: (+39) 0461 281977. E-mail: [email protected]. † Max-Planck-Institut fu¨r Festko¨rperforschung. ‡ Universita¨t Stuttgart. § Universita´ degli Studi di Trento and INSTM.

Figure 1. Chemical structure of Nafion 117 in the sulfonic acid form.

distinguished. Hence, various mechanical degradation modes were proposed to explain experimental observations, such as released fluoride detected in the wastewater,15-22 membrane thinning,23 and pinhole/crack formation.21,22,24-28 Likewise, extended durability studies on hydrated membranes revealed structural and chemical changes even at ambient temperature,29 although pure Nafion is reported to be thermally stable up to about 280 °C. Chemical degradation of membranes is generally thought to play the most important role for fuel cell failures.9,15,16,23,29,30 Experimental observations, such as decrease in ion-exchange capacity,28 changes in pH,16 conductivity,16,31 and membrane thickness,23 and release of fluoride,15-23 are typical indicators for chemical membrane degradation. In the meantime, a variety of experimental techniques, such as broad-band dielectric spectroscopy,32 EPR,8,9,30,33-40 FT-IR,15,20,24,27,33,41-44 Raman,33 UV-visible,33 liquid24,45,46 and solid-state20,24,42,47 NMR, mass spectroscopy,45,46 XPS,10,26,41,48,49 wide-angle and small-angle X-ray diffraction,10,26 TGA,27,28,50 TEM,51 and SEM,24 have been used for identification of the degradation products. Membrane degradation studies are done under in situ or ex situ conditions. In the first scenario, the membrane operates in

10.1021/jp102533v  2010 American Chemical Society Published on Web 08/05/2010

14636

J. Phys. Chem. C, Vol. 114, No. 34, 2010

a fuel cell as part of the membrane electrode assembly (MEA), and degradation is studied directly during the fuel cell operation, or at the end of the test procedure. In ex situ tests the polymer electrolyte membrane is subjected to suitable test conditions to mimic the situation during fuel cell operation, often with considerable reduction of the testing time. However, due to the complexity of the reaction schemes, a generalization of the conclusions has to be done with great care. Therefore, the combined use of in situ and ex situ protocols may lead to a more detailed and reliable understanding of the degradation reactions in PEM fuel cells. Finding suitable ex situ tests in which the system parameters are well-controlled appears, therefore, an interesting option for obtaining complementary information. In a running fuel cell, the membrane is subjected to a chemically oxidizing environment on the cathode side and a chemically reducing environment on the anode side, and gas crossover may lead to radical formation. In fact, it is the attack by radicals that is considered as the main source for chemical degradation of PEMFC membranes, as directly supported by a recent study that proved the formation of HO• and HOO• radicals during in situ fuel cell experiments.30 They originate from both electrochemical and chemical reactions and occur on the anode as well as on the cathode side. Also, radical centers at the tertiary backone carbons were reported for Nafion membranes.9,37 Chain-end radicals with structures such as -O-CF2-CF2• were identified by EPR, after UV irradiation of Nafion membranes in a H2O2 solution containing metal ions.37 During other EPR studies on X-ray-irradiated Nafion, peroxy radicals were detected.33 XPS investigations confirmed that the polymer backbone and the sulfonic acid groups in the side chains are susceptible to X-ray damage, with a faster decomposition of the side chains.48-50 Samples exposed to the Fenton’s test showed reduced intensities of the fluorine and sulfur signals in the XPS spectrum, which is consistent with the detection of released fluoride and sulfate ions.41 Likewise, FTIR spectroscopy proved the formation of CdO and S-O-S units in degraded membranes,41,43 the latter of which reflecting a cross-linking of the sulfonic acid groups in the side chains.42,43 19F NMR and mass spectrometric analyses of the degradation solution exhibited fluorinated fragments largely resembling the Nafion sidechain structure.45 Nevertheless, the various reactions and structural changes during chemical degradation are far from being completely understood. Therefore, we report here on a new type of ex situ test setup that is generally applicable for membrane degradation studies, where external parameters, such as gas atmosphere, humidity, and temperature, can be varied. In the present work, the effect of gas composition and the absence/presence of the Pt catalyst on the Nafion structure and mobility are examined by solid-state 19F NMR spectroscopy. As demonstrated by numerous applications on quite different types of materials, solid-state NMR spectroscopy is a powerful tool for elucidation of the molecular features of almost any type of inorganic, organic, polymeric, and biological material.52-56 The 19F nucleus, examined in the present work, is characterized by a spin I of 1/2, 100% natural abundance, and a large gyromagnetic ratio, which make it a very favorable and sensitive nucleus for NMR studies. It possesses a large isotropic chemical shift range (more than 1000 ppm) and typically a large chemical shift anisotropy and is thus very sensitive to chemical and conformational changes. For materials with a high fluorine content, the respective solid-state NMR spectra show additional broadening due to strong dipolar couplings, as known from 1H

Ghassemzadeh et al. NMR spectra of highly protonated materials.57,58 Identification of different structural units is possible by fast magic-angle spinning (MAS),59,60 with spinning frequencies well above 10 kHz. Solid-state 19F NMR spectroscopy was used earlier for the study of fluorinated fuel cell membranes. Broad-line 19F NMR techniques (with very limited spectral resolution) were applied to derive structural information as well as the dynamical properties of such materials.61,62 More recently, high-resolution 19 F MAS NMR techniques provided detailed insights into the structural features because it is possible to distinguish between the various main- and side-chain segments. So far, only few applications of 19F NMR spectroscopy in connection with polymer degradation studies were reported.20,42,45,47,63 For instance, in a recent study on polymer membranes subjected to in situ tests, it was shown that the degradation primarily affects the side-chain segments.47 In the present work, solid-state 19F NMR spectra are used for direct examination of the structural alterations in Nafion membranes that were subjected to different ex situ test conditions. The investigations are completed by dynamic NMR studies (i.e., relaxation experiments, line width, and overall spectral widths), which provide information about changes in the molecular mobility of the different chain segments, which, in turn, reflect structural changes due to membrane degradation. On the basis of these experimental results, possible mechanisms for radical formation and radical attacks are discussed. Experimental Section Sample Preparation and Degradation Test. Membrane Pretreating. Pieces of Nafion 117 (size ) 9 × 9 cm) were washed in H2O2 (3 v/v %) at 80 °C for 1 h and were rinsed in deionized water for another hour at T ) 80 °C. After this procedure, the membranes were completely transparent. For exchanging all ions by protons, the Nafion membrane was exposed to a solution of HNO3 (5 w/w %) at 80 °C for 4 h. The membranes were then rinsed in deionized water at T ) 80 °C. The water was renewed after each hour for at least four times. In a last step, the membranes were kept in deionized water at room temperature overnight. Preparing Membranes with Catalyst Layer. Each piece of hydrated membrane was sandwiched between two glass plates and was fixed with some clips at the corners. The glass plates were then kept in the oven at T ) 100 °C overnight. After separating the glass plates, a completely flat dried membrane was obtained (Figure 2a). In a second step, a certain amount of Nafion dispersion in water and ethanol (5 w/w %) was added to a weighted beaker (6 mL). The beaker was kept in a water bath at T ) 80 °C until the total amount of solvent was evaporated. After weighing the beaker again, a small amount of dimethylformamide (DMF) was added to the beaker (∼3 mL). When the Nafion was completely dissolved in DMF, 0.2 g of platinum (40% on carbon) was added to the beaker and was mixed well with the DMF solution containing the Nafion. This mixture was used as an electrode ink. The flat dried membrane from the first step was then laid on a glass plate, and all edges of the membrane were fixed to the glass by self-adhesive tape. After putting the glass plate in the oven at T ) 80 °C, the surface of the membrane was painted with the electrode ink by means of a small paint brush. The reason for using only a small amount of DMF solvent in the electrode ink is that a too high amount of DMF in the electrode ink prevents Pt from covering the surface of the membrane in a homogeneous way; it rather flows over the membrane surface,

Degradation of Nafion Membranes in PEMFCs

J. Phys. Chem. C, Vol. 114, No. 34, 2010 14637

Figure 2. Membrane preparation: (a) membrane flattening, (b) pasting the electrode layer, (c) putting the membrane in the frame.

forming flow marks. After painting, the oven door was closed, and the glass plate was kept in the oven at T ) 80 °C until the DMF over the membrane was evaporated completely. The membrane coated with a Pt layer was obtained (Figure 2b). During the painting process, the membrane may deform, but as it is fixed to the glass, it will appear flat after drying. By weighing the dry membrane before and after the electrode painting, it was possible to calculate the amount of Pt that finally covered the surface of the membrane. The amount of Pt for the present membranes was between 1.9 and 2.0 w/w % of the total weight of the Nafion membrane (1.25 mg of Pt per cm2 of membrane). Preparing the Membrane for Aging Tests. The dried membrane was sandwiched between two glass rings. The glass rings with the membrane in between were tightened with four stainless steel screws. Extra membrane material around the rings was removed with a cutter and kept as a reference sample without degradation. For each experiment, one membrane without a Pt layer and one with a Pt layer were prepared (Figure 2c). The sample holder containing the membrane was then hooked inside the ex situ setup cell with a Nylon string.

Ex Situ Test Setup. The experimental setup is shown in Figure 3. Temperature and humidity were chosen to be constant and close to typical fuel cell operating conditions (T ) 80 °C and RH ) 70-80%). The two compartments of the setup are kept at different temperatures. The bottom of the cell (lower-temperature compartment) was placed on a hot plate to keep the water inside the cell on a fixed temperature. To keep the level of the water constant during a long-term experiment, a water refilling system was connected to the cell. The water temperature was controlled to be T ) 70 °C, which, at thermal equilibrium, corresponds to a relative humidity of 65% at an ambient pressure of 1 atm. Because the upper compartment, containing the sample, was kept at T ) 80 ( 1 °C by a temperature-controlled homemade oven, the actual value for the relative humidity might deviate from the aforementioned equilibrium value. It should be, however, very close and the same for all experiments done in this work. The setup was equipped with a gas inlet and a gas outlet. The inlet gas was passed through the water to ensure fast humidification, and the outlet gas was connected to a dish to

14638

J. Phys. Chem. C, Vol. 114, No. 34, 2010

Ghassemzadeh et al.

Figure 3. Setup of ex situ degradation test cell.

TABLE 1: Experimental Conditions for the Preparation of the Samples inlet gas cathode (O2) cathode (O2-rich) anode (H2) anode (H2-rich)

T (°C)

RH (%)

80 80 80 80 80 80 80 80

65 65 65 65 65 65 65 65

sample Nafion Nafion Nafion Nafion Nafion Nafion Nafion Nafion

condense and collect the release water from the test as well as any fragments of the sample that might be released during the experiment. To simulate anodic and cathodic conditions, including the possibility of gas crossover, four different sets of experiments were chosen (Table 1). The total inlet gas flow was fixed to mimic a current density of 1 A/cm2 in a real fuel cell. Therefore, for each 10 mg of the sample, the flow rate was chosen to around j ) 10 mL/min. For each sample, NMR measurements were done before and after the degradation test, which took 160 h. For solid-state NMR investigations, two pieces of each sample were taken, one of which was dried at 110 °C for 12 h, and kept inside a desiccator until the NMR measurements were started. The other piece was kept in water at 70 °C overnight and then in the same water environment at room temperature before the NMR measurements. These two hydrated states were denoted as “dry” and “wet” samples. Changes in the chemical structure of the membranes, as derived from the solid-state NMR experiments, were also compared with the results from liquid-state NMR measurements that were done on the water released from the cell during the different test runs. For this purpose, the released water from each test was collected and kept at room temperature under a slow N2 gas flow for about 2 days. During this time, water

117 117 117 117 117 117 117 117

+ Pt coating + Pt coating + Pt coating + Pt coating

1 O2 O2 O2 O2 H2 H2 H2 H2

2

3

(20%) (20%)

H2 (2%) H2 (2%)

Ar (78%) Ar (78%)

(90%) (90%)

O2 (2%) O2 (2%)

Ar (8%) Ar (8%)

evaporated and the total amount of the solution decreased to about 10 mL, which was then used for liquid-state 19F NMR measurements. NMR Spectroscopy. Solid-state NMR experiments were done on a Varian InfinityPlus 400 MHz spectrometer (Varian, Palo Alto, CA) operating at a static magnetic field of 9.4 T. A 4 mm four-channel HFXY MAS probe with a Vespel spinning module and zirconia rotors with Vespel drive tips, caps, and plug-ins were used in order to avoid fluorine background. The pretreated Nafion membranes were cut into narrow strips (size ) ca. 1 × 10 mm) and were packed into the NMR rotors. The spectra were recorded at 379.09 MHz for 19F nuclei at spinning rates of 10.5 and 15 kHz. The lower spinning rate was used to study the changes in the spinning sideband intensities of the different peaks corresponding to different parts of the polymer and for the line-width analysis. The higher spinning rate was used for all other studies. Variable-temperature 19F NMR experiments (25, 40, 60, 80, 95, and 120 °C) were done using the Varian temperature control unit. 19 F MAS NMR spectra were recorded with a 90° pulse length of 2.8 µs, a recycle delay of 2 s, and a dwell time of 5 µs; 256 transients were recorded. Each transient was acquired for 10.24 ms with a spectral width of 200 kHz. The NMR signals were processed with the Spinsight software (Varian), and the FID was Fourier transformed without any additional line broadening.

Degradation of Nafion Membranes in PEMFCs

J. Phys. Chem. C, Vol. 114, No. 34, 2010 14639 TABLE 2: Chemical Shift Values (in ppm) and Assignment of the 19F Resonances in Nafion Membranes

Figure 4. Solid-state 19F NMR spectra for samples from different ex situ tests in the dry and wet state. Asterisks indicate spinning sidebands. The given spectra are normalized to the dominant peak of the backbone CF2 groups. The sample spinning was 15 kHz. The peak assignment is shown in the figure (isotropic chemical shift values are given in ppm).

The chemical shifts were calibrated with respect to polytetrafluoroethylene (Teflon) as a secondary external standard, whose resonance was set to -121 ppm.64 Spectral deconvolution was done by using the Spinsight software. 19 F spin-lattice relaxation times in the laboratory frame, T1, were measured by using the inversion recovery experiment.65 19 F spin-lattice relaxation times in the rotating frame, T1F, were obtained by observing the decay of the spin locked polarization generated by a π/2 pulse (spin-lock field ) ν1(19F) ) 90 kHz). Sixteen transients were recorded for each experiment with a repetition delay of 5 s. The processing steps for the T1 and T1F experiments were the same as mentioned for the 19F MAS NMR spectra, and the magnetization decays were then fitted with the appropriate one-component equations. 19 F liquid-state NMR spectra were recorded (without sample spinning) with the 4 mm MAS probe and the same parameters as for the MAS NMR experiments. In this case, 1024 transients were recorded and averaged. EquiWalent Weight (EW). The amount of sulfonic acid groups available for ion exchange was determined by acid-base titration. The membranes were soaked in 1 M NaCl solution for 12 h before titrating with 0.01 M NaOH (Mettler Titrator DL21). The samples (in their Na+ form) were dried after titration in a vacuum oven at 140 °C for 4 h. Ion-exchange capacities (IEC) are given in units of moles of titratable protons per gram of dry ionomer with the equivalent weight being the inverse of the IEC. Results 19 F MAS NMR. Figure 4 compares 19F NMR spectra of different samples before and after the ex situ tests for different hydration states. The spectra were recorded at T ) 80 °C because, at this elevated temperature, the reduced line widths allowed for an enhanced

CF (side chain)

CF (main chain)

(CF2)n

SCF2

OCF2

CF3

-144

-138

-121

-117

-80.1 79.9

-80

resolution. The given signal assignment (Table 2) is based on former solid-state66 and liquid-state NMR investigations.67 The 19F NMR spectrum of Nafion shows distinct signals characteristic for either the side chain or the polymer backbone. The CF2 groups of the backbone give rise to a resonance at -121 ppm similar to a signal of 19F NMR in Teflon, whereas the 19F resonance at -138 ppm is associated with the backbone CF group at which the side chain is attached. The signal of the CF group in the side chain appears at -144 ppm. The 19F resonance at -117 ppm reflects the SCF2 groups, and the two peaks originating from the two OCF2 groups and the CF3 group of the side chain appear at about -80 ppm. The spectra shown in Figure 4 were taken at a spinning speed of 15 kHz, which was the upper speed that was reached safely for this sample in the 4 mm MAS probe. The low field signal (-80 ppm) originating from the OCF2 and CF3 groups appears broadened because of its overlap with the spinning sidebands of the backbone CF2 group and the side-chain SCF2 group signals (asterisks in Figure 4). For examination of the NMR line widths, the sample spinning speed was, therefore, reduced to 10.5 kHz, at which these latter signals are not affected by spinning sidebands from other resonances. The two sets of spectra given in Figure 4 refer to dry and wet samples (see the Experimental Section), the latter of which was prepared to rule out potential condensation reactions that might occur during the normal workup procedure of the Nafion membranes (dry samples). As can be seen, the line widths are generally smaller for wet than for dry membranes, and the lines are slightly broadened for samples with the Pt coating. The comparison of the spectra of the samples before and after the ex situ tests, but with the same hydration state and coatings, reveals a reduced intensity for the polymer side-chain signals in the treated samples. This effect is most pronounced for the intensities of the OCF2 and SCF2 peaks at -80 and -117 ppm, respectively. To quantify the spectral changes, all spectra were deconvoluted and normalized to the integral intensity of the CF2 backbone signal. Figure 5 shows the normalized integrals of the 19F NMR signals for the OCF2 and SCF2 groups before and after the ex situ tests. Apart from the aforementioned line widths and spectral resolution differences, the areas under the spectra of the corresponding wet and dry samples (subjected to the same ex situ treatments) are identical, which makes us confident that deconvolution allows for quantitative analysis, indeed. This also proves that the sample workup procedure does not affect the results from membrane degradation. As a general result, practically no intensity changes are observed for samples without a Pt coating after any treatment (Figure 5, left). For membranes coated with the Pt catalyst, however, pronounced reductions in signal intensity after treatment with H2/O2 gas mixtures are registered. The largest effect occurs in the case of H2-rich conditions, and the SCF2 group signal appears to be somewhat more affected than the OCF2 group signal (Figure 5, right). Liquid-State 19F NMR. To identify possible reaction products released via the humidified gas stream, liquid-state 19F NMR measurements were done on the water collected during the tests. No 19F NMR signals are detected if, during the tests, the cell is flushed by either H2 or O2 only, regardless of whether the

14640

J. Phys. Chem. C, Vol. 114, No. 34, 2010

Ghassemzadeh et al.

Figure 5. Peak integrals of the side-chain groups, OCF2 and SCF2, in the solid-state 19F NMR spectra before and after the ex situ tests (spinning speed ) 15 kHz). The integral of the backbone CF2 peak at -121 ppm was put to unity. The experimental error is (3%. Figure 7. Variable-temperature solid-state 19F NMR spectra of Nafion at a spinning speed of 10.5 kHz. Asterisks indicate the spinning sidebands used for the data plotted in Figure 8. The sidebands and the respective central peaks are shown by the same gray scale.

Figure 6. Liquid-state 19F NMR spectrum for the water extracted from the H2-rich cell. The peak assignment is given in the figure. The numbers are the isotropic chemical shift values in ppm.

samples are coated with Pt or not. However, after exposure of the membranes to mixtures of both gases, 19F NMR signals are clearly detectable, and the pH of the collected water is significantly reduced (pH ∼ 4.5). The latter may be the result of HF release, which is also indicated by the appearance of a strong fluoride (F-) signal in the NMR spectra. The representative 19F NMR spectrum shown in Figure 6 (taken on the water released from the H2-rich condition ex situ test) exhibits this signal (-127.6 pm) along with 19F resonances originating from CF3 and OCF2 (-76.8 ppm), SCF2 (-116.5 ppm), and CF (-144 ppm), following the assignment in the literature.45 It should be noted that the resolution of the present spectra is rather poor because the 19F NMR spectra were recorded with the MAS probe and without any special precautions for highresolution conditions. Sideband Intensity and Line-Width Analysis. A series of variable-temperature 19F MAS NMR spectra of Nafion (prior to the ex situ test) at a sample spinning frequency of 10.5 kHz are shown in Figure 7. At this spinning frequency, the 19F NMR spectra contain spinning sidebands that stem from the large chemical shift anisotropies and strong dipolar couplings of the 19F nuclei. These spectra were analyzed (on a semiquantitative level) by considering the line widths and intensities of the spinning sidebands, which can be related to structural and dynamic changes within the various samples.

Figure 8. 19F NMR line widths and relative sideband intensities (normalized to the respective central peak) for the polymer backbone and side chain at a sample spinning speed of 10.5 kHz for the samples before and after the H2-rich condition test. The estimated error is less than (3%.

Because of the limited resolution, only the line widths for the central lines of the backbone CF2 and side-chain OCF2/CF3 resonances were analyzed. These resonances were assumed to consist of a single component each, although, from a former 1D and 2D NMR study, it is known that several spectral components have to be considered for both signals.66 For a

Degradation of Nafion Membranes in PEMFCs qualitative discussion of the 19F NMR line widths, however, the present simplifying approach appears to be justified. Widths and intensities of peaks associated with both the backbone and the side chain of samples treated under H2-rich conditions (Figure 8) decrease with temperature. Coating the membranes with Pt increases the 19F NMR line widths, which might stem from susceptibility or packing effects due to the presence of the metallic Pt particles. The NMR line widths of side-chain signals are generally smaller than those of the main chain, pointing toward higher side-chain mobility. In general, the degradation test is accompanied by only small line-width alterations. As a general trend, somewhat larger line widths are registered for the treated samples. The same 19F NMR spectra were used to derive the relative intensities of the spinning sidebands,68 which contain qualitative information about changes of the underlying chemical shift anisotropies and dipolar interactions as a function of various experimental parameters. The sideband intensities for both the side- and the main-chain signals (normalized to the intensities of their respective central peaks) also decrease with increasing temperature (Figure 8), although the differences between the samples with and without a Pt coating are very small. The presence of Pt is found to slightly increase the sideband intensity, at least at lower temperatures. These findings point to an increased chemical shift anisotropy due to the interaction with the metallic Pt particles, in analogy to the aforementioned NMR line widths. The relative intensities of the spinning sidebands of the backbone CF2 signal are significantly higher than those of the OCF2/CF3 side-chain peaks, once more, suggesting smaller anisotropies of the involved magnetic interactions (i.e., chemical shift, dipolar couplings). 19 F Relaxation. To obtain further insights into the chain mobility, 19F relaxation data (T1 and T1F) were recorded at a sample spinning frequency of 15 kHz for a membrane that had been subjected to a degradation test under H2-rich conditions. The observed T1 values are almost identical for the side-chain and backbone segments and slightly decrease with increasing sample temperature (Figure 9). This observation is in line with a former relaxation study on Nafion and is explainable by spin diffusion due to the strongly dipolar coupled 19F network.69 The Pt coating of the membrane causes a general shortening of the T1 values. After the ex situ degradation test, a general increase of the T1 values is observed, regardless of whether the membrane was coated with Pt or not. For the T1F data, the situation appears to be more complex. For all signals, a maximum of the relaxation curves is found at around T ) 60 °C. Unlike the quite uniform T1 data, T1F values are generally longer for the backbone signals than for the side-chain resonances. Within the side chain, shorter T1F values are found at the end of the side chain (see data for OCF2 and SCF2 groups). The effect of the Pt coating on the T1F values is the same as found for the T1 data; that is, upon addition of Pt, the T1F values become shorter. The degradation test, using the H2-rich condition, in general, is accompanied by an increase of the T1F values. The absolute changes, however, also depend on both the segmental position and the presence or absence of a Pt coating. Equivalent Weight and Water Uptake. Table 3 summarizes EW, λ, and φ values for samples before and after the degradation tests. In agreement with the above results, only the samples treated in the H2/O2 mixtures exhibit some changes of the EW values, which are even larger for the samples with a Pt coating. For all samples, an increase of the equivalent weight is accompanied by a decrease of the water volume fraction φ, while the hydration

J. Phys. Chem. C, Vol. 114, No. 34, 2010 14641

Figure 9. 19F T1 and T1F values of the samples before and after the H2-rich condition test (dry samples, sample spinning speed ) 15 kHz). The estimated error is less than (3%.

number λ remains constant or slightly increases. Again, the most significant effects on EW and φ are detected for the membrane that was treated under H2-rich conditions. Discussion The experimental techniques of the present work, comprising solid-state and liquid-state NMR along with ion-exchange capacity and water uptake data, clearly prove that membrane degradation only takes place if O2-rich or H2-rich conditions (i.e., in the presence of both gases) are chosen and the samples contain a Pt coating. According to the literature,9,20,22,36,37,70-73 Nafion membrane degradation occurs via radical attack. Inside the fuel cell, there are different sources for radical formation. One relevant reaction is the formation of hydrogen peroxide by oxygen reduction.16,30,74

2H+ + O2 + 2e- f H2O2

(1)

Hydrogen peroxide leads to radical formation in the presence of heat or traces of transition-metal cations:

H2O2 f 2HO•

(by heat)

(2)

or

H2O2 + M2+ f M3+ + HO• + OH(M ) transition-metal cation) (3) H2O2 + HO• f HOO• + H2O

(4)

Because the radical precursor (H2O2) is formed in an electrochemical reaction, this source of radicals cannot be considered for the present experiments, in which no current is

14642

J. Phys. Chem. C, Vol. 114, No. 34, 2010

Ghassemzadeh et al.

TABLE 3: Equivalent Weight (EW), Number of Water Molecules Per Sulfonic Acid Group (λ), and Water Volume Fraction (φ) of the Nafion 117 Membrane before and after the Degradation Tests sample EW λ φ

Nafion 117 + Pt coating

Nafion 117

tests 1108 21.2 0.34

O2

H2

O2-rich

H2-rich

1123 21.5 0.34

1109 21 0.34

1155 21.5 0.33

1333 21.6 0.27

applied. In a running fuel cell, however, in which electrochemical oxygen reduction takes place, the above radicals may form as side products. The second possibility for radical formation is the direct reaction of H2 and O2, at the surface of the Pt catalyst, creating membrane degrading species:19,23,40,75

H2 + O2 f 2HO•

(at catalyst surface)

(5)

HO• + H2 f H2O + H•

(6)

H• + O2 f HOO•

(7)

The rates of the above reactions depend on the catalyst surface properties and the relative concentrations of H2 and O2 at the catalyst surface.19,23,30 This mechanism is of a chemical nature, and it can, therefore, proceed in the present ex situ experiments where H2, O2, and the Pt catalyst are present. In a running fuel cell, this route of radical formation may occur on the surface of the anode or cathode catalyst depending on the current density. Only at OCV or very low current densities, oxygen dissolves in the aqueous phase of the membrane and diffuses to the anode side, where it may participate in reaction 5. If a protonic current is flowing from the anode to the cathode, however, the electroosmotic water drag is anticipated to transport dissolved hydrogen from the anode to the cathode, where reactions 6 and 7 may then lead to the formation of radicals. In the present ex situ tests, degradation is observed only if H2, O2 and the Pt catalyst are present (Figure 5), and this suggests that (i) the chemical formation of radicals is occurring, indeed, and that (ii) the presence of these radicals leads to bondcleavage processes. These reactions are most pronounced for H2-rich conditions, suggesting that H• is reacting with a higher rate than HOO• radicals. As for the in situ degradation tests,47 side-chain signals are more affected by degradation than signals originating from the main chain (note that the spectra shown in Figure 4 are normalized to the backbone CF2 signal). The presence of sidechain fragments in the collected water (Figure 6), the reduction of the IEC (Table 3), and the reduction of the hydrophilic character independently confirm side-chain cleavage. Backbone degradation via a main-chain unzipping mechanism76 (Figure 10) is frequently considered in the literature.76 This reaction starts at the carboxylic acid end groups (-COOH), which are unintentionally introduced during the manufacturing process of Nafion via hydrolysis of the persulfate initiators used for the polymerization process,77 and which releases fluoride, F-. Postfluorination of backbone end groups reduces degradation,76 although some residual fluorine release remains even when reactive end groups are virtually eliminated,46,76 pointing toward additional mechanisms, for instance, from side-chain degradation. The present ex situ test results suggest predominant sidechain degradation, in accordance with earlier findings on low

1106 21.2 0.34

O2

H2

O2-rich

H2-rich

1111 21.2 0.34

1098 21.5 0.34

1443 23.2 0.28

1979 24.9 0.22

carboxyl content commercial PEM membranes.30,46,47 Although carboxylic groups are found to be more reactive than ether linkages,46 the latter are higher in concentration by at least 2-3 orders of magnitude (every side chain has two bonds of this type), which makes side-chain degradation more likely.78 With this, one may even expect an increasing side chain-to-backbone degradation rate with increasing temperature. What are the mechanisms of side-chain degradation by radicals? One possibility is the radical attack to the C-S bond at the end of the side chain (Figure 11), which is actually the weakest bond within this segment. From there, the reaction may proceed, which progressively leads to a side-chain shortening along with the formation of acid (HF), that is, fluoride release, as observed experimentally. This mechanism is in agreement with the detection of -O-CF2-CF2-SO3• and -O-CF2-CF2• radicals in EPR experiments on UV-induced Fenton treated membranes.9,37 The side-chain degradation may also start at the tertiary backbone carbon bearing the side chain or the tertiary carbon between two ether groups within the side chain. These reactions can lead to complete or partial release of the side chains46,78 (Figure 12), and again release HF.

Figure 10. Main-chain degradation mechanism, unzipping of the main chain by radical attacks to COOH groups.

Figure 11. Side-chain degradation mechanism by unzipping the side chain via radical attack to the C-S bond.

Degradation of Nafion Membranes in PEMFCs

Figure 12. Side-chain degradation mechanism by radical attack to the CF backbone carbon. The same mechanism can be assumed for the attack to the tertiary carbon in the side chain.

The effects of degradation on the membrane dynamics are further examined by T1 and T1F relaxation data and NMR lineshape effects. 19F and 1H NMR T1, T1F and T2 relaxation studies of unaged samples as a function of preparation route and water content have already been reported in the literature.69,79 Also, spin relaxation and spectral anisotropies, in the presence of MAS, were explored for Nafion containing different counterions.66,80 It was found that T1 relaxation is dominated by spin diffusion, preventing a distinction of backbone and the side-chain dynamics, but T1F data revealed a dynamical coupling of the backbone and the side-chain segments. An extensive investigation, based on one and two-dimensional 19F and 13C NMR studies on stationary and spinning samples, covered both structural aspects and the chain dynamics of Nafion membranes.61,62 Analysis of these data proved fast uniaxial rotations of the helical backbone, like in polytetrafluoroethylene. There was even evidence for the movement of the backbone axis as well as motions in the side chains with maximum amplitude at the center of the side chain. 13C and 19F NMR line widths, obtained from fast MAS spectra, were traced back to the presence of static disorder in the Nafion backbone in the vicinity of the branching site. The presence of water not only has a plasticizing effect76 but also leads to cross-relaxation effects between the 19F and 1H spin system reservoirs, complicating the interpretation of NMR data.65,78 The study of the present work is, therefore, only based on relaxation data recorded for the dry samples. The experimental 19F NMR line widths showed (i) a decrease with increasing temperature, (ii) larger values for the backbone, and (iii) an increase for the samples with a Pt coating. The latter effect might be related to susceptibility or packing effects due to the presence of the metallic Pt particles. On the basis of earlier high-speed MAS investigations, the line widths can be understood by static disorder in the vicinity of the chain-branching CF segment and also within the Nafion side chain. Moreover, restriced backbone reorientation and higher motional freedom of the side-chain segments explain the differences in line widths. Because the overall chain motions occur in the fast motional limit, at about 100 kHz or above,69 changes of the motional rates cannot be responsible for the temperature-dependent linewidth reductions. Rather, they reflect a thermally activated increase of the fluctuation amplitudes that cause a reduction in static disorder. The stronger temperature dependence of the respective line widths is explainable by the higher mobility, that is, minor spatial hindrance, of the side chains. Likewise, the

J. Phys. Chem. C, Vol. 114, No. 34, 2010 14643 successive convergence of the line widths for the samples with and without a Pt coating can be understood by the thermally activated changes in static disorder. The same assumptions can be used to explain the experimental findings for the overall spectral widths (due to chemical shift anisotropy and homonuclear dipolar coupling), which are reflected by the relative intensities of the spinning sidebands. The significantly smaller sideband intensities for the side-chain segments are thus consistent with a higher fluctuation amplitude. The relative changes upon temperature increase are larger for the side-chain segments than for the backbone, and the differences between the samples with and without a Pt coating vanish at elevated temperatures, in agreement with the aforementioned line-width effects. However, chemical degradation, in general, is found to have only little (for the line widths) or no effect on the NMR line-shape parameters. The results and trends for the derived T1 data (see Figure 9) are in agreement with previous investigations. The observation of identical 19F T1 values for all positions in the sample can be understood by the presence of spin diffusion due to the strong homonuclear dipolar interactions between the 19F nuclei, which equalizes the spin-lattice relaxation values for the different chain positions.61,69 Moreover, all experimental T1 values lie on the low-temperature side of the T1 curve, that is, below the curve minimum. A general decrease of the 19F T1 values for the samples with a Pt coating is observed, which, as mentioned during the discussion of the NMR line widths, is related to an additional contribution due to the presence of the metallic Pt particles. A general shift of the 19F T1 values toward higher values is observed for the samples after the degradation test. Surprisingly, this also holds for the sample without a Pt coating, which, as discussed above, should not be affected by chemical degradation. At present, this latter finding is not fully understood. It is very likely that the degradation tests give rise to some polymer aging, which is not related to chemical degradation. That is, structural annealing may occur with alterations of the magnetic interactions responsible for spin-lattice relaxation. Hence, a reduction for the homonuclear dipolar couplings can explain the aforementioned increase of the 19F T1 values. Likewise, the chain motions may become more restricted, which would cause the same trend for the T1 relaxation data. The derived T1F data are more complex than the T1 data, as the values (i) change with the chain position (i.e., decrease from backbone to side chain, and toward the end of the side chains; T1F (CF2) > T1F (OCF2) > T1F (SCF2)), (ii) decrease by the addition of a Pt coating, (iii) increase after the degradation tests, and (iv) exhibit a maximum at around 60 °C. Obviously, spin diffusion does not affect the T1F data, and the values for the individual segments can be clearly distinguished. Moreover, the trends by the presence of the Pt coating and after the degradation test are the same as found for the T1 data. The observation of a maximum in the T1F curves is in agreement with previous studies and was attributed to the appearance of a new motional mechanism that dominates T1F relaxation at higher temperatures.61 It was reported that the position of the maximum is strongly dependent on the counterion in the Nafion sample and the water content. In agreement with earlier work,80 the additional mechanism was attributed to the R-process that yields a T1F minimum above 100 °C, which is further related to the glass transition. The motional mechanism that is responsible for T1F relaxation at temperatures below the curve maximum is attributed to the aforementioned rotation and reorientational motions of the

14644

J. Phys. Chem. C, Vol. 114, No. 34, 2010

backbone and the side-chain segments, which, at room temperature and above, occur in the fast motional limit, that is, at frequencies g 100 kHz.81 In fact, a further variable-temperature T1F study showed a low-temperature T1F minimum at around -20 °C, at which the frequency of the underlying motional process should be 40 kHz ()strength of experimental spin-lock field in the experiment).81 Accordingly, the aforementioned decrease of the T1F values toward the side-chain ends reflects an increase of the motional amplitude, due to higher chain flexibility, in the same direction. The slight increase of the T1F values for the sample without a Pt coating after the degradation test is again attributed to a decrease of the motional amplitude and/or reduction of the dipolar interactions due to structural annealing, as mentioned during the discussion of the T1 data. Upon Pt coating, a much more pronounced increase of the T1F values after the chemical degradation test is registered, which is in line with the aforementioned reduction of the motional amplitude. In summary, the same motional modes are assumed to be responsible for T1F and T1 relaxation. It is certainly desirable to provide a more quantitative analysis of such relaxation data, by separating the various contributions from degradation and structural annealing and by identifying the underlying motional processes. To do so, experiments on well-defined model systems would be necessary. The stronger rise of the relaxation data for the Pt-coated sample after the degradation test may originate from chemical and thus structural changes of the Nafion membrane. In this context, chain cross-linking, as a result of the degradation process, may take place, which directly reduces the motional amplitudes and, hence, the chain flexibility. From the present NMR studies, there was no direct proof for such chain crosslinking reactions. This might be related to the limited spectral resolution that could be achieved or to the low concentration for such cross-linking points. In former solid-state 19F NMR studies on electron-beam or γ-irradiated fluoropolymers, employing fast MAS techniques, it was possible to prove chain cross-linking by the appearance and identification of new spectral components.82-84 Similar experiments are also planned for the present Nafion materials. Conclusions Chemical degradation of Nafion membranes has been studied by treatments in an ex situ setup and subsequent NMR analysis of the membranes and the water, which had been in contact with the membrane during the tests. The temperature and gas composition (H2O, H2, and O2) were controlled and varied independently during the tests, and as samples, membranes with and without a Pt catalyst coating were used. In this way, in situ fuel cell operating conditions could be mimicked, and the controlled variation of single parameters allowed for the identification of the parameters controlling chemical aging. Because no current was applied during the aging process, electrochemical reactions could not contribute to the observed effects. However, the fact that in situ degradation effects were reproduced by the present ex situ experiments suggests that membrane degradation in a running fuel cell is mainly the consequence of chemical aging. Post-aging solid-state 19F NMR spectroscopy clearly shows that remarkable degradation only takes place if the membrane is coated with Pt and if both gases, H2 and O2, are present. This observation points toward the importance of radicals in the degradation process, which, in a running fuel cell (in situ conditions), may only form in the presence of some gas crossover allowing H2 and O2 to react at the Pt catalyst of the anode or cathode structure.

Ghassemzadeh et al. Solid-state NMR of the membrane before and after aging also shows that side-chain degradation is prevailing. Side-chain fragments are actually identified in the collected water, but this also contains significant amounts of fluoride ions, which may, to some extent, originate from backbone degradation. 19 F relaxation and NMR line-shape data were analyzed to get further insights into the dynamic properties of the Nafion membranes along with the alterations upon degradation. The temperature-dependent decrease in NMR line widths and spectral anisotropies points to a reduction of the static disorder in the Nafion membrane upon sample heating. The relaxation data give evidence for structural annealing, as expressed by a reduction of the motional amplitudes and/or modulated dipolar interactions, which is not related to chemical degradation. Chemical degradation amplifies this effect, which may result from chain cross-linking. Further work along these lines is in progress. Acknowledgment. The authors thank Helmut Kammerlander for making all of the glassware needed in the ex situ setup test and Udo Klock for his technical support. L.G. would like to thank the International Max-Planck Research School for Advanced Materials for the fellowship. References and Notes (1) Darling, R. M.; Meyers, J. P. J. Electrochem. Soc. 2005, 152, A242. (2) Bi, W.; Gray, G. E.; Fuller, T. F. Electrochem. Solid-State Lett. 2007, 10, B101. (3) Darling, R. M.; Meyers, J. P. J. Electrochem. Soc. 2003, 150, A1523. (4) Meyers, J. P.; Darling, R. M. J. Electrochem. Soc. 2006, 153, A1432. (5) Tang, H.; Qi, Z. G.; Ramani, M.; Elter, J. F. J. Power Sources 2006, 158, 1306. (6) Luo, Z.; Li, D.; Tang, H.; Pan, M.; Ruan, R. Int. J. Hydrogen Energy 2006, 31, 1831. (7) Stevens, D. A.; Hicks, M. T.; Haugen, G. M.; Dahn, J. R. J. Electrochem. Soc. 2005, 152, A2309. (8) Bosnjakovic, A.; Schlick, S. J. Phys. Chem. B 2006, 110, 10720. (9) Bosnjakovic, A.; Kadirov, M. K.; Schlick, S. Res. Chem. Intermed. 2007, 33, 677. (10) Wang, Z. B.; Zuo, P. J.; Chu, Y. Y.; Shao, Y. Y.; Yin, G. P. Int. J. Hydrogen Energy 2009, 34, 4387. (11) Smitha, B.; Sridhar, S.; Khan, A. A. J. Membr. Sci. 2005, 259, 10. (12) Neburchilov, V.; Martin, J.; Wang, H. J.; Zhang, J. J. J. Power Sources 2007, 169, 221. (13) Kreuer, K. D. J. Membr. Sci. 2001, 185, 29. (14) Mauritz, K. A.; Moore, R. B. Chem. ReV. 2004, 104, 4535. (15) Balko, E. N.; Chaklos, J. T. J. Appl. Polym. Sci. 1981, 26, 1519. (16) Pozio, A.; Silva, R. F.; De Francesco, M.; Giorgi, L. Electrochim. Acta 2003, 48, 1543. (17) Aoki, M.; Uchida, H.; Watanabe, M. Electrochem. Commun. 2005, 7, 1434. (18) Mittal, V. O.; Kunz, H. R.; Fenton, J. M. Electrochem. Solid-State Lett. 2006, 9, A299. (19) Mittal, V. O.; Kunz, H. R.; Fenton, J. M. J. Electrochem. Soc. 2006, 153, A1755. (20) Kinumoto, T.; Inaba, M.; Nakayama, Y.; Ogata, K.; Umebayashi, R.; Tasaka, A.; Iriyama, Y.; Abe, T.; Ogumi, Z. J. Power Sources 2006, 158, 1222. (21) Liu, W.; Ruth, K.; Rusch, G. J. New Mater. Electrochem. Syst. 2001, 4, 227. (22) Aoki, M.; Uchida, H.; Watanabe, M. Electrochem. Commun. 2006, 8, 1509. (23) Mittal, V. O.; Kunz, H. R.; Fenton, J. M. J. Electrochem. Soc. 2007, 154, B652. (24) Tang, H. L.; Shen, P. K.; Jiang, S. P.; Fang, W.; Mu, P. J. Power Sources 2007, 170, 85. (25) Collier, A.; Wang, H. J.; Yuan, X. Z.; Zhang, J. J.; Wilkinson, D. P. Int. J. Hydrogen Energy 2006, 31, 1838. (26) Huang, C. D.; Tan, K. S.; Lin, H. Y.; Tan, K. L. Chem. Phys. Lett. 2003, 371, 80. (27) Mittleman, M. L.; Thomsen, J. R.; Wilkie, C. A. Chem. Abstr. 1990, 200, 211. (28) Surowiec, J.; Bogoczek, R. J. Therm. Anal. 1988, 33, 1097.

Degradation of Nafion Membranes in PEMFCs (29) Beyazyildirim-So¨rgel, S.; Ghassemzadeh, L.; Kreuer, K. D.; Telfah, A.; Cavalca de Araujo, C.; Lyonnard, S.; Mu¨ller, K.; Maier, J. J. Mater. Chem. 2010. Manuscript in preparation. (30) Danilczuk, M.; Coms, F. D.; Schlick, S. J. Phys. Chem. B 2009, 113, 8031. (31) Schulze, M.; Wagner, N.; Kaz, T.; Friedrich, K. A. Electrochim. Acta 2007, 52, 2328. (32) Rhoades, D. W.; Hassan, M. K.; Osborn, S. J.; Moore, R. B.; Mauritz, K. A. J. Power Sources 2007, 172, 72. (33) Almeida, S. H.; Kawano, Y. Polym. Degrad. Stab. 1998, 62, 291. (34) Panchenko, A.; Dilger, H.; Kerres, J.; Hein, M.; Ullrich, A.; Kaz, T.; Roduner, E. Phys. Chem. Chem. Phys. 2004, 6, 2891. (35) Vogel, B.; Aleksandrova, E.; Mitov, S.; Krafft, M.; Dreizler, A.; Kerres, J.; Hein, M.; Roduner, E. J. Electrochem. Soc. 2008, 155, B570. (36) Danilczuk, M.; Bosnjakovic, A.; Kadirov, M. K.; Schlick, S. J. Power Sources 2007, 172, 78. (37) Kadirov, M. K.; Bosnjakovic, A.; Schlick, S. J. Phys. Chem. B 2005, 109, 7664. (38) Panchenko, A.; Dilger, H.; Moller, E.; Sixt, T.; Roduner, E. J. Power Sources 2004, 127, 325. (39) Lund, A.; Macomber, L. D.; Danilczuk, M.; Stevens, J. E.; Schlick, S. J. Phys. Chem. B 2007, 111, 9484. (40) Endoh, E. Electrochem. Solid-State Lett. 2004, 7, A209. (41) Chen, C.; Levitin, G.; Hess, D. W.; Fuller, T. F. J. Power Sources 2007, 169, 288–295. (42) Collette, F. M.; Lorentz, C.; Gebel, G.; Thominette, F. J. Membr. Sci. 2009, 330, 21. (43) Qiao, J. L.; Saito, M.; Hayamizu, K.; Okada, T. J. Electrochem. Soc. 2006, 153, A967. (44) Xie, T.; Hayden, C. A. Polymer 2007, 48, 5497. (45) Healy, J.; Hayden, C.; Xie, T.; Olson, K.; Waldo, R.; Brundage, A.; Gasteiger, H.; Abbott, J. Fuel Cells 2005, 5, 302. (46) Zhou, C.; Guerra, M. A.; Qiu, Z. M.; Zawodzinski, T. A.; Schiraldi, D. A. Macromolecules 2007, 40, 8695. (47) Ghassemzadeh, L.; Marrony, M.; Barrera, R.; Kreuer, K. D.; Maier, J.; Mu¨ller, K. J. Power Sources 2009, 186, 334. (48) Schulze, M.; Lorenz, M.; Wagner, N.; Gulzow, E. Fresenius’ J. Anal. Chem. 1999, 365, 106. (49) Militello, M. C.; Gaarenstroom, S. W. Surf. Sci. Spectra 2005, 10, 117. (50) Jalani, N. H.; Dunn, K.; Datta, R. Electrochim. Acta 2005, 51, 553. (51) Xie, J.; Wood, D. L.; More, K. L.; Atanassov, P.; Borup, R. L. J. Electrochem. Soc. 2005, 152, A1011. (52) Fyfe, C. A. Solid State NMR for Chemists; C.F.C. Press: Guelph, Ontario, Canada, 1983. (53) Duer, M. J. Solid-State NMR Spectroscopy: Principles and Applications; Blackwell Science: Malden, MA, 2002. (54) MacKenzie, K. J. D.; Smith, M. E. Multinuclear Solid-State NMR of Inorganic Materials, 1st ed.; Pergamon: Oxford, U.K., 2002. (55) Schmidt-Rohr, K.; Spiess, H. W. Multidimensional Solid-State NMR and Polymers; Academic Press: London, 1994.

J. Phys. Chem. C, Vol. 114, No. 34, 2010 14645 (56) Stejskal, E. O.; Memory, J. D. High Resolution NMR in the Solid State: Fundamentals Of CP/MAS; Oxford University Press, Inc.: New York, 1994. (57) Laws, D. D.; Bitter, H. M. L.; Jerschow, A. Angew. Chem., Int. Ed. 2002, 41, 3096. (58) Brown, S. P.; Spiess, H. W. Chem. ReV. 2001, 101, 4125. (59) Lowe, I. J. Phys. ReV. Lett. 1959, 2, 285. (60) Andrew, E. R.; Bradbury, A.; Eades, R. G. Nature 1958, 182, 1659. (61) Boyle, N. G.; Mcbrierty, V. J.; Eisenberg, A. Macromolecules 1983, 16, 80. (62) MacMillan, B.; Sharp, A. R.; Armstrong, R. L. Polymer 1999, 40, 2471. (63) Wang, F.; Tang, H.; Pan, M.; Li, D. Int. J. Hydrogen Energy 2008, 33, 2283. (64) Isbester, P. K.; Brandt, J. L.; Kestner, T. A.; Munson, E. J. Macromolecules 1998, 31, 8192. (65) Farrar, T. C.; Becker, E. D. Pulse and Fourier Transform NMR: Introduction to Theory and Methods; Academic Press: New York, 1971. (66) Chen, Q.; Schmidt-Rohr, K. Macromolecules 2004, 37, 5995. (67) Takasaki, M.; Kimura, K.; Kawaguchi, K.; Abe, A.; Katagiri, G. Macromolecules 2005, 38, 6031. (68) Shao, L.; Titman, J. J. Prog. Nucl. Magn. Reson. Spectrosc. 2007, 51, 103. (69) Page, K. A.; Jarrett, W.; Moore, R. B. J. Polym. Sci., Part B: Polym. Phys. 2007, 45, 2177. (70) Bosnjakovic, A.; Schlick, S. J. Phys. Chem. B 2004, 108, 4332. (71) Merlo, L.; Ghielmi, A.; Cirillo, L.; Gebert, M.; Arcella, V. J. Power Sources 2007, 171, 140. (72) Mitov, S.; Vogel, B.; Roduner, E.; Zhang, H.; Zhu, X.; Gogel, V.; Jorissen, L.; Hein, M.; Xing, D.; Schonberger, F.; Kerres, J. Fuel Cells 2006, 6, 413. (73) Panchenko, A. J. Membr. Sci. 2006, 278, 269. (74) Okada, T.; Satou, H.; Yuasa, M. Langmuir 2003, 19, 2325. (75) Xie, J.; Wood, D. L.; Wayne, D. M.; Zawodzinski, T. A.; Atanassov, P.; Borup, R. L. J. Electrochem. Soc. 2005, 152, A104. (76) Curtin, D. E.; Lousenberg, R. D.; Henry, T. J.; Tangeman, P. C.; Tisack, M. E. J. Power Sources 2004, 131, 41. (77) Bro, M. I.; Sperati, C. A. J. Polym. Sci. 1959, 38, 289. (78) Coms, F. D. ECS Trans. 2008, 16, 235. (79) Page, K. A.; Cable, K. M.; Moore, R. B. Macromolecules 2005, 38, 6472. (80) Chen, Q.; Schmidt-Rohr, K. Macromol. Chem. Phys. 2007, 208, 2189. (81) Boyle, N. G.; Mcbrierty, V. J.; Douglass, D. C. Macromolecules 1983, 16, 75. (82) Fuchs, B.; Scheler, U. Appl. Magn. Reson. 2004, 27, 435. (83) Oshima, A.; Washio, M. Polymer Durability And Radiation Effects; ACS Symposium Series; 2008; Vol. 978, p 204. (84) Dargaville, T. R.; George, G. A.; Hill, D. J. T.; Scheler, U.; Whittaker, A. K. Macromolecules 2002, 35, 5544.

JP102533V