Crystallinity Enhancement of Nafion Electrolyte Membranes Assisted

ARTICLE pubs.acs.org/Langmuir

Crystallinity Enhancement of Nafion Electrolyte Membranes Assisted by a Molecular Gelator Wenjing Zhang, Po-Lock Yue, and Ping Gao* Department of Chemical and Biomolecular Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, People’s Republic of China

bS Supporting Information ABSTRACT: Nanocrystallites, acting as physical cross-links in Nafion membranes, play a crucial role in building blocks for improving mechanical durability and stopping fuel crossover. However, Nafion membranes suffer from low crystallinity due to the irregular pendent side chains, which hinder self-aggregation of the poly(tetrafluoroethylene) (PTFE) backbones. For the first time, a molecular gelator was introduced in the membrane casting process to enhance the rate of self-assembly of PTFE backbones so as to increase the membrane’s crystallinity as well as proton conductivity without sacrificing the purity of Nafion. The molecular gelator used was 3,4-dimethylbenzaldehyde (DMBA). Addition of 0.5 wt % DMBA led to a 42% increase in crystallinity, a 32% increase in yield strength, a 22% increase in tensile modulus and an 18% increase in proton conductivity at 60
C and 90% relative humidity. Additionally, the membrane electrode assembly (MEA) prepared from the membranes cast from the solution containing 0.5 wt % DMBA also showed an increase of 17% in maximum power density in comparison to the MEA prepared from pure Nafion membrane in a single cell polarization test without any external humidification. Transmission electron microscopy (TEM) and molecular dynamics simulation were used to elucidate the structural changes in Nafion membrane due to the introduction of DMBA. It was observed that the presence of DMBA gives wider crystalline regions under TEM. The molecular dynamics simulation at 500 K shows that the PTFE backbones become elongated in the presence of DMBA due to the enhanced mobility. This is consistent with the observed increase in crystallinity in the membrane as it means reduced entropic change upon crystallization.

1. INTRODUCTION Being the heart of a fuel cell, the proton-exchange membrane (PEM)1 is undoubtedly considered as a key and performancelimiting component.2 Perfluorosulfonic acid ionomers, such as Nafion, are commonly employed as the electrolyte in PEM fuel cells. Nafion is a polymer consisting of a poly(tetrafluoroethylene) (PTFE) backbone with pendent side chains terminated with sulfuric acid groups, as shown in Figure 1. The extreme difference in polarity of backbone and sulfuric acid groups results in the formation of an inhomogeneous, hydrophilic
hydrophobic phaseseparated structure. The hydrophilic regions are quite hygroscopic and readily absorb water. Upon hydration, an interconnected network is formed that allows protons to flow with the water being the medium.3 The transport of protons through the Nafion membrane governs the overall proton conductivity. On the other hand, the hydrophobic nanocrystallites are the only physical crosslinks, which prevent Nafion from dissolving and maintain mechanical stability. However, the crystallinity of Nafion (EW = 1100) remains very low [5
20% based on wide-angle X-ray diffraction (WAXD) tests]4 due to the lack of long-range structural order, which leads to very critical challenges for fuel cell commercialization, such as high hydrogen/methanol crossover and poor durability. r 2011 American Chemical Society

Over past decades, there have been considerable research efforts focusing on improving crystallinity of Nafion membranes.5
7 Inorganic/organic fillers were used to enhance crystallization of Nafion membranes, stop fuel crossover,8
11 and improve proton conductivity at low humidity.12,13 In most composite systems, the size of the fillers is much bigger than that of crystalline domains (∼5 nm). Therefore, proton conductivity was often found to be sacrificed to mechanical properties and crystallinity.10 A solution processing procedure, developed by Moore and Martin,14 opened a new way to increase crystallinity of Nafion without additional fillers. In their work, membranes casted from Nafion solutions in ethanol/water or 1-propanol/water mixture solvents without high temperature annealing were brittle and had poor mechanical properties, while the Nafion membrane annealed in the presence of high boiling point solvents had higher crystallinity and was insoluble in all solvents at temperatures below about 200
C.14 It indicates that Nafion’s performance and intrinsic properties are dependent not only on its chemical identity, but also on the method of Received: March 20, 2011 Revised: June 10, 2011 Published: June 17, 2011 9520

dx.doi.org/10.1021/la201047k | Langmuir 2011, 27, 9520–9527

Langmuir

Figure 1. Chemical formulas of Nafion 1100EW (n = 7, m = 1) and 3,4dimethylbenzaldehyde.

membrane casting process and thermal history.15 More recently a parallel cylindrical pore model was proposed by Schmidt-Rohr and Chen,2 where nanosized Nafion crystallites (∼5 nm) are elongated and parallel to water channels with an average diameter of 2.4 nm. The challenge remaining is the precise control of crystallization of Nafion without disturbing proton conduction through water channels that are embedded in the crystalline phase. Low molecular weight organic gelator was reported to be capable of constructing a three-dimensional network by selforganizing into finely dispersed anisotropic aggregates.16
20 The ability of self-assembly to form nanosized structure by very specific interactions of the gelator such as hydrogen bonding, hydrophobic interaction, and π
π interaction triggered our interest to explore the effect of molecular gelator on selfassembled crystallization of Nafion. For the first time, 3,4-dimethylbenzaldehyde (DMBA) was introduced as a molecular gelator to promote crystallization of Nafion membranes during the casting process. Intrinsically different from conventional fillers, molecular gelator DMBA has comparable size with crystalline domains of Nafion. The hydrophobic benzyl rings of DMBA in combination with the polar aldehyde groups facilitate its self-organization into the anisotropic phase of Nafion, where benzyl rings formed hydrophobic templates to promote PTFE backbone aggregation at the temperature above Tg and aldehyde groups point to ionic cylindrical pores driven by hydrogen bonding between
SO3H and
CO
. Both molecular dynamics simulation and experimental characterization show that the use of this molecular gelator leads to a significant improvement in crystallinity. High ordering of backbones makes the membrane mechanically robust, while sulfuric acid groups selfassemble into wider hydrophilic channels to facilitate proton conduction. Moreover, DMBA was only present during the casting process and was removed from the system during high-temperature annealing, resulting in pristine Nafion membrane with the combination of desired physical and electrochemical properties. In addition, the use of DMBA also led to membranes with a thermodynamically more stable microstructure due to the enhanced self-assembly rate of PTFE backbones during crystallization processes. Therefore, long-term durability of the membranes is also expected. This is in a clear contrast to those composite membranes whose microstructures are time-dependent due to their strong deviations from thermodynamic equilibrium and potential for leaching in use; long-term durability is a concern for their use in fuel cells.

2. EXPERIMENTAL SECTION 2.1. Materials. Nafion solution, 5 wt % polymer (EW = 1100) dissolved in aliphatic alcohols, was purchased from Aldrich. E.I. du Pont de Nemours & Co. supplied the Nafion 112 membrane for this

ARTICLE

experiment. Gas diffusion electrodes (GDE) with 0.25 mg/cm2 Pt loading were used as received from E-TEK. N,N-Dimethylformamide (99.5%, DMF) and N,N-Dimethylacetamide (99%, DMAc) were used as cosolvents to solution-cast Nafion membranes. Analytical-grade hydrogen and oxygen gases were used as received. 3,4-Dimethylbenzaldehyde (98%, DMBA) was obtained from Aldrich. 2.2. Membrane Preparation. Nafion-DMBA membranes were prepared by the same procedure as that for pure Nafion membranes. The mixture of commercial 5 wt % Nafion solution, DMF, and DMAc in a volume ratio of 4:1:1 was mixed with a certain desired mass concentration of DMBA (based on the mass of dry Nafion membrane). The entire solution mixture was sonicated by use of an ultrasonication probe running in pulsation mode at ambient temperature for 30 min. Afterward, the entire solution was cast onto a glass Petri dish and dried at 80
C in a nonconvection oven for 8 h. Further drying and thermal annealing at T = 190
C in vacuum for 10 h was applied to remove the residues and increase crystallinity. After thermal treatment, the sample was cooled down slowly to room temperature inside the vacuum oven. After rinsing with water, the membrane was treated in a 5 vol % H2O2 solution at 60
C for 1 h and boiled in a 0.5 M H2SO4 solution for 1 h with boiling in distilled water in between. For comparison, pure Nafion membranes were also prepared by the same protocol. The membrane electrode assembly (MEA) was then prepared by uniaxially hot-pressing two E-TEK GDE onto both sides of the membrane at 135
C and 4.0 MPa for 90 s. A Nafion and 2-propanol solution (1:4 volume ratio) is sprayed on the surfaces of GDE to enhance the adhesion between GDEs and membranes. Nafion ionomer loading was optimized to be 33 wt % based on total weight of Pt/C catalysts.21 2.3. Wide-Angle X-ray Diffraction. WAXD spectrum was collected on a Philips PW 1825 diffractometer with Cu KR radiation (40 kV, 50 mA). Angular scanning was performed in the range of 2
< 2θ < 50
at the rate of 2
/min. Crystallinity was obtained through a crystalline
amorphous peak deconvolution process using Gaussian functions from the originally convoluted peak.22

2.4. Positive Staining and Transmission Electron Microscopic Analysis. A 0.5 wt % aqueous solution of ruthenium tetraoxide (RuO4) was prepared for staining of Nafion membranes. Excessive NaIO4 powder (1 g) was put into a Petri dish. The membranes were first cut into sections with an ultramicrotome (Leica) to a thickness of 70 nm and then collected on a Cu grid. The cutting direction is perpendicular to the surface of bulk membranes. After 2 mil of RuO2 aqueous solution was dropped on the powder of NaIO4, the whole container was immediately sealed. The appearance of a golden color indicated the formation of RuO4. Staining duration was optimized as 15 min at room temperature. For fieldemission transmission electron microscopy (TEM), the JEOL 2010 microscope was used to study the nanostructure of the membranes. 2.5. Tensile Tests. Tensile properties of the membranes were measured on the Advanced Rheometrics Expansion System (ARES) at an extension rate of 2 mm/min at room temperature. Each film was cut into parallel-sided strips with dimensions of 10  3 mm2. Before tensile tests, all membrane samples were predried in a vacuum oven at 105
C for 24 h and then cooled down to room temperature. To eliminate the influence of relative humidity and temperature on mechanical properties,23 the tests were performed immediately after the samples were taken out of the vacuum oven. The laboratory was conditioned at 22
C and 40% relative humidity. The yield strength was obtained at 2% strain offset. Ten specimens were used for each test. 2.6. Single Fuel Cell Tests. All MEAs were evaluated on a commercial fuel cell test station (FCTS, Arbin). The fuel cell test fixture accommodated a 5 cm2 electrode and contained single anode and cathode serpentine flow channels. Dry H2 and O2 gases were fed to the cell at the constant flow rates of 100 sccm. The current
voltage (I
V) curves were measured at back-pressures of 0.1 MPa for O2 and 0.05 MPa for H2 with a cell temperature of 60
C. 9521

dx.doi.org/10.1021/la201047k |Langmuir 2011, 27, 9520–9527

Langmuir

ARTICLE

Table 1. Thicknesses, Chemical Compositions, and Crystallinity of Nafion 112, Pure Nafion, and N-DMBA Membranesa Nafion sample DMBA content (wt %)

112 0

pure

N-

N-

Nafion DMBA-0.1 DMBA-0.5 0

0.1

0.5

NDMBA-1.0 1

thickness (μm)

57

70

70

70

70

crystalline peak

17.7

17.8

17.8

17.8

18.0

18

24

23

34

33

(deg) crystallinity (%) a

Calculated from the WAXD spectra shown in Figure 2.

2.7. Thermogravimetric Analysis. Thermal stability of the membranes was examined by use of a thermogravimetric analyzer (TA Hi-Res 2950). The membrane was first heated from 30 to 120
C and held for 2 h to remove water, followed by a temperature scan from 120 to 600
C at a heating rate of 5
C/min under argon flow. 2.8. Model Preparation and Simulation Details. To build the models of Nafion and Nafion
3,4-dimethylbenzaldehyde (N-DMBA) membranes, we used a regular repeat of 20 units of the Nafion structure as shown in Figure 1. The atom typing and partial charges of Nafion and H2O were assigned following those of the model used by Wescott et al.24 For the model of DMBA, the default COMPASS atom typing and partial charges were used. The initial microstructure of the model containing Nafion or N-DMBA was built by use of the “Amorphous Cell” module in Material Studio Software (Accelrys Inc.). The compositions of two simulation cells were Nafion with 5 wt % H2O (designated as Nafion) and Nafion with 5 wt % H2O and 3 wt % DMBA (designated as Nafion-DMBA). Periodic boundary conditions were imposed on the cubic unit cell in order to eliminate surface effects. The cubic unit cell was visualized as being surrounded on all sides by replicas of itself, in this way forming an infinite three-dimensional microlattice. The molar ratio of repeat units of Nafion, hydronium ions, and H2O molecules was set at 1:1:3 for Nafion model, corresponding to 5 wt % H2O in Nafion membrane. This ratio of repeat units of Nafion, hydronium ions, H2O molecules, and DMBA was set as 4:4:12:1 for Nafion-DMBA model, resulting in 3 wt % DMBA based on the dry weight of Nafion membrane. Following the construction of amorphous cell, minimization of the potential energy of the whole cell was achieved by using Discover module. Constant total energy with simulation time was considered as an indication of equilibration of the cell. The equilibration step was accomplished via NVT ensemble simulation to eliminate the overlapping or close contact. The simulation time of 50 ps was proven to be sufficient for maintaining the fluctuation of energies constant. The resulting atomistic structures were subsequently treated by an annealing procedure. Both Nafion and Nafion-DMBA models were heated from 300 to 500 K at intervals of 50 K and then cooled down to 300 K. At each step, 50 ps NPT dynamics were applied on the cells. Afterward, two NVT dynamics (T = 500 K and T = 300 K) were performed at the end point of the NPT run to obtain equilibrium molecular structures. The atomic trajectories were recorded at each picosecond for subsequent analysis for each model. Considering high water uptake of Nafion under fully humidified conditions, we also performed the same simulations of Nafion and Nafion-3 wt % DMBA at 20 wt % H2O content.

3. RESULTS AND DISCUSSION 3.1. Selection of DMBA Concentration. Nafion-DMBA membranes at different concentrations of DMBA were fabricated. Table 1 lists the details of membrane composition and

Figure 2. Wide-angle X-ray diffraction spectra for Nafion 112, solutioncast pure Nafion, and Nafion-DMBA membranes (N-DMBA-0.1, N-DMBA-0.5, and N-DMBA-1.0). Diffraction patterns were corrected for background scatter, and crystalline/amorphous peaks were fit to Pearson VII distribution functions, all of which had correlation coefficients greater than 99%.

thickness. Figure 2 shows X-ray diffraction spectra of pure Nafion, Nafion-DMBA, and commercial Nafion 112 membranes. To assay for crystallinity, the deconvolution of curves shows a broad peak at 2θ < 16.0
for amorphous region and a distinct sharp peak at 2θ > 17.8
for crystalline regions (see Figure S1 in Supporting Information). The calculated results of crystallinity of membranes are listed in Table 1. The crystallinity (Wcr) is obtained according to Z 21 Icr ð2θÞ dð2θÞ 9 " # ð1Þ Wcr ¼ Z Z 21

9

21

Icr ð2θÞ dð2θÞ +

Iam ð2θÞ dð2θÞ

9

where I(2θ) = intensity of scattering. Consistent with previous research reports,4,25,26 as-received Nafion 112 shows a crystalline reflection superimposed as a shoulder on an amorphous halo. The calculated crystallinity is 16%. Solution-cast pure Nafion membrane with thermal annealing treatment develops a sharper shoulder at 2θ = 17.8
, indicating higher crystallinity of pure Nafion. With a small amount of DMBA (0.1
1 wt %), the crystallinity of N-DMBA membranes increases from 23% to 34% with the concentration of DMBA. This represents a significant crystallinity increase of 48%. Improvement of crystallinity was also achieved by adding inorganic fillers. Park et al.27 developed Nafion/CHP organic
inorganic composite membrane with high crystallinity. The strong polymer
inorganic interaction was considered as the driving force to change crystallization, resulting in the formation of different morphology and the increase in crystallinity. However, this enhancement was only possible at the interface between polymer chains and inorganic fillers. The particle size was around 125 nm, which is much larger than the characteristic size of the crystalline region in Nafion (5 nm). Introducing more crystalline regions inevitably means adding more inorganic particles, which was found to hinder the proton conductivity or, as the additives acted as defects, to degrade the overall performance at high concentration levels. Therefore, in situ enhancement of crystallization, achieved by molecular gelator in this work, is particularly 9522

dx.doi.org/10.1021/la201047k |Langmuir 2011, 27, 9520–9527

Langmuir

ARTICLE

Table 2. Mechanical Properties of Nafion 112, Pure Nafion, and N-DMBA Membranes max

energy to

modulus

max

yield stress

stress

fracture

(MPa)

strain (%)

(MPa)

(MPa)

(MPa)

Nafion 112

228.5

304

4.0

27.5

49.6

pure Nafion

199.3

434

3.7

25.1

74

N-DMBA-

209.4

343

4.4

25.2

57.5

243.5

360

4.9

29.0

67

272.4

375

5.5

32.2

80.2

0.1 N-DMBA0.5 N-DMBA1.0

interesting because Nafion nature can be preserved perfectly after all the solvents and DMBA were completely removed after the thermal annealing process at high temperature. It should be emphasized here that all Nafion membranes were cast in the presence of cosolvents DMF and DMAc, and annealing was applied. Hence, the 48% increase in crystallinity is due to the presence of the molecular gelator DMBA. Because Nafion is one of the most fragile components in PEM fuel cells, mechanical properties of Nafion are crucial for durability. Higher crystallinity usually gives better mechanical properties since these crystalline regions act as physical cross-links and maintain their integrity toward external stresses.25 Therefore, tensile tests were performed to evaluate the mechanical robustness of N-DMBA membranes. Because linear deformation occurs at very low strains, we define the modulus as the ratio of the stress to the strain at 0.2% strain. The data in Table 2 show that modulus, yield stress, and energy to fracture increase with the concentration of DMBA in Nafion membranes. Compared to pure Nafion, N-DMBA-1.0 (1 wt % DMBA) shows 49% improvement in tensile yield strength and 37% improvement in tensile modulus. These are again consistent with the observed X-ray crystallinity results shown in Table 1 assisted by the molecular gelator DMBA. It is also interesting to compare the mechanical properties of the solvent-cast membranes with those of Nafion 112 prepared by melt processing. The tensile modulus and yield strength of the membranes show, respectively, 19% and 38% increase. This is a significant achievement, as the solvent casting process tends to reduce the mechanical performance while raising proton conductivity. The nature of the Nafion membranes governs the electrochemical and mechanical properties in the practical operation of PEM fuel cells. It is widely agreed that hydrophobic PTFE region is responsible for mechanical strength of the membrane, while the hydrophilic region accommodates the majority of the absorbed solvent and is, therefore, critical to ionic and solvent transport characteristics. In principle, the crystallinity of Nafion is mutually exclusive with proton conductivity, since rigid hydrophobic regions are not soluble in water at temperatures lower than 200
C and not conductive to protons. Therefore, it is necessary to strike a balance between mechanical robustness and electrochemical performance of fuel cells. Polarization curves, obtained from single fuel cell tests containing various membranes, are shown in Figure 3. The performance obtained here on Nafion 112 is consistent with the results

Figure 3. Polarization curves of MEAs employing commercial Nafion 112, pure Nafion, and N-DMBA membranes with dry H2 and O2. The flow rates were set at 100 sccm for both H2 and O2. Fuel cell performance was measured at 60
C with back-pressure of 0.1 MPa for O2 and 0.05 MPa for H2.

obtained by other groups.28,29 The measurement was carried out at 60
C, as the most favorable working temperature of a PEM fuel cell is usually from 60 to 80
C.30 The objective of this study is to elucidate the gelation effects of DMBA on fuel cell performance or the membrane’s electrochemical properties. The power output of fuel cells employing N-DMBA membranes increases with decreasing DMBA concentrations. This result is consistent with the crystallinity of membranes measured by WAXD and other works,31 where the proton conductivity of membranes compromised crystallinity. However, it is particularly interesting that N-DMBA-0.1 and N-DMBA-0.5 performed better than pure Nafion in terms of both mechanical robustness and maximum power density in the fuel cell tests. N-DMBA-0.5, cast with 0.5 wt % DMBA, shows a maximum power density of 486 mW/cm2, which is about 17% higher than the pure Nafion membrane and 36% higher than Nafion 112. This is related to the higher proton conductivity of N-DMBA-0.5, as shown in Figure S2 (see Supporting Information). 3.2. Characterization of N-DMBA-0.5 Membrane. The combined mechanical and electrochemical performance tests presented above suggest that N-DMBA-0.5 membrane, cast with 0.5 wt % DMBA, may be an optimal candidate for use in fuel cell MEAs as it shows better mechanical properties than Nafion 112 and also a significant improvement in polarization performance over solution-cast pure Nafion. To elucidate nanostructural features, we carried out a TEM study of Nafion and N-DMBA membranes. Direct experimental visualization of the clusters and crystalline networks is challenged by the poor image contrast and the need to create extremely thin membrane samples.32,33 Therefore, staining of sectioned membranes was performed,34 where RuO4 vapor reacts with water in the cluster regions of Nafion to produce RuO2 particles. Under TEM, dark and nearly spherical spots (RuO2) represent the hydrophilic cluster regions of Nafion membranes, while white regions with no presence of water represent the hydrophobic crystalline domains. TEM images of solution-cast pure Nafion, N-DMBA-0.1, N-DMBA-0.5, and N-DMBA-1.0 are shown in Figure 4. For pure Nafion, cluster domains (spherical black spots) are randomly distributed in the membrane with diameter of 2
5 nm, 9523

dx.doi.org/10.1021/la201047k |Langmuir 2011, 27, 9520–9527

Langmuir

ARTICLE

Figure 5. Plots of derivative of weight loss vs temperature for Nafion 112, pure Nafion, and N-DMBA-0.5 in N2. Figure 4. TEM pictures of (a) pure Nafion, (b) N-DMBA-0.1, (c) N-DMBA-0.5, and (d) N-DMBA-1.0 membranes. All samples were poststained with RuO4 to enhance the contrast.

consistent with the model prediction and the SAXS results.35 The growth of ionic clusters and crystalline regions were significantly developed with the presence of molecular gelator DMBA, giving larger hydrophobic crystalline domains and clusters aggregates compared to those in pure Nafion. With the increase of DMBA concentration, one can clearly observe the sharper interface between the hydrophilic clusters (black) and hydrophobic domains (white) in terms of both color and size. The structure is consistent with the higher crystallinity and better mechanical properties measured by WAXD and tensile test. The enhanced crystalline domain size is attributed to enhanced ordering in the PTFE backbones induced by the hydrophobic
hydrophobic interactions, where self-aggregation of PTFE backbone and ionic clusters of Nafion were promoted by DMBA during hightemperature annealing process. Thermal stability of all membranes was characterized by thermogravimetric analysis (TGA).12 It is generally agreed that the thermal degradation of Nafion takes place in three stages: desulfonation, degradation of side chains, and degradation of PTFE backbones.36,37 However, the degradation temperature range for each stage may vary depending on the solution casting methods, sample pretreatment, and TGA operation conditions. Tiwari et al.36 and Park et al.37 showed that the degradation temperature range for desulfonation was 250
400
C, and the other two stages are 400
500 and 500
600
C for the degradation of side chains and PTFE backbone, respectively. The analysis of TGA in conjunction with Fourier transform infrared spectroscopy (FTIR)37 also showed the release of sulfur dioxide from Nafion only within the temperature range 275
400
C. The thermal degradation behaviors for membranes studied here show consistent degradation mechanisms and also occurs in three stages.36,37 As shown in Figure 5, the first stage ranged from 280 to 360
C, corresponding to a desulfonation process. The second derivative peak exhibited a shoulder from 380 to 470
C, and the third one is from 470 to 600
C. These two stages can be assigned to decomposition of side chains and subsequent total degradation of PTFE backbones. For the solution-cast pure Nafion, the thermal degradation behavior is slightly different, where the onset temperatures of the first peak for desulfonation and the second shoulder for decomposition of side chains are higher, but the decomposition peak of PTFE backbones shifted

slightly to lower temperature than that for commercial Nafion 112. These phenomena are due to the effect of processing method since pure Nafion is a solution-cast membrane while commercial Nafion 112 is an extruded thin film. The N-DMBA0.5 membrane shows almost similar thermal stability to that of the pure Nafion membrane, suggesting that the use of DMBA will not lead to any adverse effects on fuel cell performance. It should be noted that there is no evidence of DMBA trace in the TGA result for N-DMBA-0.5, indicating complete evaporation of DMBA after thermal annealing process at high temperature. 3.3. Simulation Results and Discussion. Crystallization of Nafion is a highly nonequilibrium process during the thermal annealing process.38 Solution-cast membranes were slightly crystalline. When they were subjected to annealing, the membranes showed evidence of higher crystallinity.39 High-temperature drying under vacuum changed the water sorption of Nafion, while drying at room temperature under vacuum did not alter the membranes’ equilibrium water uptake.40 However, there is no universally accepted statement of the thermal and solvent effects on Nafion crystallization behavior due to the lack of experimental technique to directly observe the structure and dynamics of the polymer chains during the casting and annealing treatments. Therefore, we performed molecular dynamics simulation to reveal the evolution of the structure and properties of Nafion polymer during the casting and annealing stages at the atomistic level. The simulated model was validated by calculating the equilibrium density via NPT molecular dynamics simulations. Minimization (50 ps) was used to collect the average density for each model. For the model of Nafion, the calculated density was 2.04 g/cm3. This value is 4.6% higher than the experimental value of 1.95 g/cm3, which is acceptable for a small model.24 Annealing at high temperature resulted in higher crystallinity in Nafion membranes. The crystallization process involves the aggregation of hydrophobic backbones of Nafion. Therefore, chain mobility is highly related to the crystallinity of polymer during the annealing treatment. Polymer chain mobility can be analyzed by the mean square displacement (MSD) of polymer chains: MSD ¼ ƽri ðtÞ
ri ð0Þ2 æ

ð2Þ

where ri(t) and ri(0) are the positions of atom i at times t and 0, respectively. The bracket denotes the ensemble average, NVT in this simulation. In the presence of true solvents, higher mobility will result in higher MSD value. 9524

dx.doi.org/10.1021/la201047k |Langmuir 2011, 27, 9520–9527

Langmuir

ARTICLE

Figure 6. Mean square displacements of backbones of Nafion at (a) 500 K and (b)300 K after the annealing treatments.

Figure 7. Snapshots of configurations at 20 wt % water content at the end of the production run for (a) Nafion and (b) Nafion-3 wt % DMBA. Orange, sulfur; red, oxygen atom of H2O or SO3
; green, oxygen atom of H3O+; white, hydrogen. (c) Snapshot of chain structures of Nafion-3 wt % DMBA-20 wt % H2O at the end of energy minimization. Yellow, DMBA; red, oxygen; green, sulfur; gray, carbon atoms of Nafion.

Scheme 1. Schematic Representation of Crystallization Enhancement of Nafion Assisted by DMBA during the Annealing Process at High Temperaturea

a

The proposed mechanism is based on the inverted micelle model of Nafion.2 The PTFE backbones of Nafion were stretched and aligned into crystalline domains with the assistance of a hydrophobic template consisting of benzene rings of DMBA molecules. The polarized side chains point to the ionic clusters.

Our particular interest is the DMBA effect on the backbone mobility of Nafion during the annealing treatment. Thus, the MSD of Nafion backbones in both Nafion and Nafion-DMBA models are shown in Figure 6. At 500 K, the mean square end-to-end

distance in Nafion-DMBA is larger than that in Nafion, implying that the PTFE backbones are elongated and highly aligned due to the presence of DMBA at the annealing temperature. The same simulation was done for the mixture of Nafion, DMF, and DMAc with the same concentration (3 wt % cosolvent with DMF:DMAc weight ratio of 1:1). The result showed negligible effect of DMF and DMAc on the PTFE mobility, which further confirms that the remarkable enhancement can only be assisted by DMBA. Details about the simulation and material characterizations will be published separately. On the other hand, MSD is lower in the blend when samples are cooled down from 500 K to room temperature, as shown in Figure 6b. At 300 K, more PTFE backbones are able to self-aggregate into crystalline domains in Nafion-DMBA, showing higher stability of PTFE backbones compared to Nafion. Considering the higher water uptake of Nafion membranes under fully humidified conditions, we also conducted simulations on Nafion and Nafion-3 wt % DMBA at 20 wt % water content. The same results were observed with more distinguished phase separation and higher MSD. Snapshots of configurations at the end of the run are shown in Figure 7a,b. The snapshot of chain structures of Nafion-3 wt % DMBA at 20 wt % water content after energy minimization (Figure 7c) shows the preferential ordering and selfassembly of PTFE backbones in the vicinity of DMBA inclusions. Thus, a mechanism of crystallization enhancement of Nafion assisted by DMBA is described in Scheme 1, in which DMBA molecules act as templates, improving the mobility of Nafion backbones and promoting the self-aggregation of both crystalline and cluster regions in Nafion-DMBA model. The average distance 9525

dx.doi.org/10.1021/la201047k |Langmuir 2011, 27, 9520–9527

Langmuir among PTFE backbones in the crystalline domain at 500 K is around 3 Å, calculated from simulation results. This size is consistent with the molecular size of DMBA.

4. CONCLUSION For the first time, the concept of molecular gelator is introduced to improve crystallinity of Nafion membranes. To prepare highly crystallized Nafion membranes facilitated by molecular gelator, the concentration of DMBA in Nafion membrane was carefully chosen by consideration of both the crystallinity and electrochemical performance. The higher concentration of DMBA results in higher crystallinity of Nafion membranes, while higher crystallinity also induces lower output power density during the single fuel cell test. Therefore, N-DMBA with 0.5 wt % DMBA in Nafion is recommended to improve the crystallinity and mechanical robustness without sacrificing fuel cell performance. By TEM analysis, we found the morphology of N-DMBA membranes is different from that of pure Nafion prepared without the molecular gelator DMBA. Increasing DMBA concentration increases rates of self-assembly of the PTFE aggregates and simultaneously increases crystallinity and mechanical robustness. Hydrophilic clusters (composed of continuous ionic thin channels) and hydrophobic domains (composed of PTFE aggregates) in the N-DMBA membranes are much bigger than the Nafion membranes without DMBA. However, the relatively smaller crystalline domains seem to be randomly distributed over the amorphous ionic clusters in pure Nafion membrane. Molecular dynamics simulation was carried out to investigate the effects of the molecular gelator on dynamic behaviors of PTFE backbones of Nafion during solution casting and annealing process at atomistic level. With the Discover and Amorphous Cell modules of Material Studio, simulation analysis showed that the backbones of Nafion in the presence of the molecular gelator DMBA were significantly stretched out at the annealing temperature, whereas they are confined tightly at room temperature. This subsequently reduces the energy barrier for formation of nanocrystallites in Nafion membranes. Therefore, the presence of the molecular gelator DMBA acted as a template for hydrophilic
hydrophobic structure, promoted Nafion’s PTFE backbone self-aggregation at the annealing temperature, and consequently facilitated crystallization when the membrane was gradually cooled down to room temperature. On the basis of morphological features and simulation analysis, DMBA was proved to be a novel molecular gelator to improve mechanical properties and electrochemical performance of Nafion membrane without sacrificing its structural purity. This is part of ongoing research to address one of the most critical challenges of Nafion. We are the first group in the literature to introduce the concept of molecular gelator in Nafion casting. Differing from conventional approaches to enhance the crystallization of Nafion in the past decades, our study can be used to manipulate the crystallization and clustering of Nafion at nanoscale without bringing any additional filler. With the assistance of a small amount of DMBA, both mechanical properties and electrochemical performances in a fuel cell were improved simultaneously. Microscopic structure evolution and underlying mechanism were studied through both experimental analysis and dynamic simulation. The study here presented a fundamental basis for the design of thermodynamically more stable membranes for fuel cell applications.

ARTICLE

’ ASSOCIATED CONTENT

bS

Supporting Information. Additional text and two figures showing deconvolution of WAXD curves of Nafion 112 and N-DMBA-1.0 and proton conductivity of pure Nafion and N-DMBA-0.5. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT The project was sponsored by the Research Grant Council of Hong Kong with an earmarked grant for research (612805). ’ REFERENCES (1) Sethuraman, V. A.; Weidner, J. W.; Haug, A. T.; Pemberton, M.; Protsailo, L. V. Electrochim. Acta 2009, 54 (23), 5571–5582. (2) Schmidt-Rohr, K.; Chen, Q. Nat. Mater. 2008, 7 (1), 75–83. (3) Spry, D. B.; Goun, A.; Glusac, K.; Moilanen, D. E.; Fayer, M. D. J. Am. Chem. Soc. 2007, 129 (26), 8122–8130. (4) Gierke, T. D.; Munn, G. E.; Wilson, F. C. J. Polym. Sci. Polym. Phys. 1981, 19 (11), 1687–1704. (5) Lin, J.; Wu, P. H.; Wycisk, R.; Trivisonno, A.; Pintauro, P. N. J. Power Sources 2008, 183 (2), 491–497. (6) Lin, J.; Wu, P. H.; Wycisk, R.; Pintauro, P. N.; Shi, Z. Q. Macromolecules 2008, 41 (12), 4284–4289. (7) Lin, J.; Wycisk, R.; Pintauro, P. N.; Kellner, M. Electrochem. Solid State 2007, 10 (1), B19–B22. (8) Li, C. N.; Sun, G. Q.; Ren, S. Z.; Liu, J.; Wang, Q.; Wu, Z. M.; Sun, H.; Jin, W. J. Membr. Sci. 2006, 272 (1
2), 50–57. (9) Song, M. K.; Park, S. B.; Kim, Y. T.; Kim, K. H.; Min, S. K.; Rhee, H. W. Electrochim. Acta 2004, 50 (2
3), 639–643. (10) Byun, S. C.; Jeong, Y. J.; Park, J. W.; Kim, S. D.; Ha, H. Y.; Kim, W. J. Solid State Ionics 2006, 177 (37
38), 3233–3243. (11) Zhang, W. J.; Li, M. K. S.; Yue, P. L.; Gao, P. Langmuir 2008, 24 (6), 2663–2670. (12) Di Noto, V.; Boaretto, N.; Negro, E.; Pace, G. J Power Sources 2010, 195 (23), 7734–7742. (13) Di Noto, V.; Piga, M.; Lavina, S.; Negro, E.; Yoshida, K.; Ito, R.; Furukawa, T. Electrochim. Acta 2010, 55 (4), 1431–1444. (14) Moore, R. B., III; Martin, C. R. Anal. Chem. 1986, 58 (12), 2569–2570. (15) Luan, Y.; Zhang, H.; Zhang, Y.; Li, L.; Li, H.; Liu, Y. J. Membr. Sci. 2008, 319 (1
2), 91–101. (16) Mohmeyer, N.; Wang, P.; Schmidt, H. W.; Zakeeruddin, S. M.; Gratzel, M. J. Mater. Chem. 2004, 14 (12), 1905–1909. (17) Huo, Z. P.; Zhang, C. N.; Fang, X. Q.; Cai, M. L.; Dai, S. Y.; Wang, K. J. J. Power Sources 2010, 195 (13), 4384–4390. (18) Huo, Z. P.; Dai, S. Y.; Zhang, C. G.; Kong, F. T.; Fang, X. Q.; Guo, L.; Liu, W. Q.; Hu, L. H.; Pan, X.; Wang, K. J. J. Phys. Chem. B 2008, 112 (41), 12927–12933. (19) Abdallah, D. J.; Sirchio, S. A.; Weiss, R. G. Langmuir 2000, 16 (20), 7558–7561. (20) Lescanne, M.; Colin, A.; Mondain-Monval, O.; Heuze, K.; Fages, F.; Pozzo, J. L. Langmuir 2002, 18 (19), 7151–7153. (21) Antolini, E.; Giorgi, L.; Pozio, A.; Passalacqua, E. J. Power Sources 1999, 77 (2), 136–142. (22) Park, Y. S.; Yamazaki, Y. Eur. Polym. J. 2006, 42 (2), 375–387. (23) Satterfield, M. B.; Benziger, J. B. J. Polym. Sci. Polym. Phys. 2009, 47 (1), 11–24. (24) Wescott, J. T.; Qi, Y.; Subramanian, L.; Capehart, T. W. J. Chem. Phys. 2006, 124, No. 134702. 9526

dx.doi.org/10.1021/la201047k |Langmuir 2011, 27, 9520–9527

Langmuir

ARTICLE

(25) Moore, R. B., III; Martin, C. R. Macromolecules 1988, 21 (5), 1334–1339. (26) Fujimura, M.; Hashimoto, T.; Kawai, H. Macromolecules 1981, 14 (5), 1309–1315. (27) Park, Y. S.; Yamazaki, Y. J. Membr. Sci. 2005, 261 (1
2), 58–66. (28) Hung, T. F.; Liao, S. H.; Li, C. Y.; Chen-Yang, Y. W. J. Power Sources 2011, 196 (1), 126–132. (29) Watanabe, M.; Uchida, H.; Emori, M. J. Phys. Chem. B 1998, 102 (17), 3129–3137. (30) Wu, J. F.; Yuan, X. Z.; Martin, J. J.; Wang, H. J.; Zhang, J. J.; Shen, J.; Wu, S. H.; Merida, W. J. Power Sources 2008, 184 (1), 104–119. (31) Emery, M.; Frey, M.; Guerra, M.; Haugen, G.; Hintzer, K.; Lochhaas, K. H.; Pham, P.; Pierpont, D.; Schaberg, M.; Thaler, A.; Yandrasits, M.; Hamrock, S. ECS Trans. 2007, 11 (1), 3–14. (32) Porat, Z.; Fryer, J. R.; Huxham, M.; Rubinstein, I. J. Phys. Chem. 1995, 99 (13), 4667–4671. (33) Rollins, H. W.; Lin, F.; Johnson, J.; Ma, J. J.; Liu, J. T.; Tu, M. H.; DesMarteau, D. D.; Sun, Y. P. Langmuir 2000, 16 (21), 8031–8036. (34) Xue, T.; Trent, J. S.; Osseo-Asare, K. J. Membr. Sci. 1989, 45 (3), 261–271. (35) Mauritz, K. A.; Moore, R. B. Chem. Rev. 2004, 104 (10), 4535– 4585. (36) Tiwari, S. K.; Nema, S. K.; Agarwal, Y. K. Thermochim. Acta 1998, 317 (2), 175–182. (37) Park, H. S.; Kim, Y. J.; Hong, W. H.; Lee, H. K. J. Membr. Sci. 2006, 272 (1
2), 28–36. (38) Hensley, J. E.; Way, J. D.; Dec, S. F.; Abney, K. D. J. Membr. Sci. 2007, 298 (1
2), 190–201. (39) Gebel, G.; Aldebert, P.; Pineri, M. Macromolecules 1987, 20 (6), 1425–1428. (40) Zawodzinski, T. A., Jr.; Derouin, C.; Radzinski, S.; Sherman, R. J.; Smith, V. T.; Springer, T. E.; Gottesfeld, S. J. Electrochem. Soc. 1993, 140 (4), 1041–1047.

9527

dx.doi.org/10.1021/la201047k |Langmuir 2011, 27, 9520–9527