Modeling the Nanophase Structural Dynamics of Phenylated

ARTICLE pubs.acs.org/JPCB

Modeling the Nanophase Structural Dynamics of Phenylated Sulfonated Poly Ether Ether Ketone Ketone (Ph-SPEEKK) Membranes As a Function of Hydration Roberto D. Lins,† Ram Devanathan,*,‡ and Michel Dupuis‡ † ‡

Department of Fundamental Chemistry, Federal University of Pernambuco (UFPE), Recife, PE, 50740-540, Brazil Chemical & Materials Sciences Division, Pacific Northwest National Laboratory, Richland, Washington 99352, United States ABSTRACT: Solvated phenylated sulfonated poly ether ether ketone ketone (Ph-SPEEKK) membranes in the presence of hydronium ions were modeled by classical molecular dynamics simulations. The characterization of the nanophase structure and dynamics of such membranes was carried out as a function of the water content λ, where λ is the number of water molecules per sulfonate group, for λ values of 3.5, 6, 11, 25, and 40. Analysis of pair correlation functions supports the experimental observation of membrane swelling upon hydration as well the increase in water and hydronium ion diffusion with increasing λ. Whereas the average number of hydrogen bonds between hydronium ions and sulfonate groups is dramatically affected by the hydration level, the average lifetime of the hydrogen bonds remains essentially constant. The membrane is found to be relatively rigid, and its overall flexibility shows little dependence on water content. Compared with Nafion, water and ion diffusion coefficients are considerably smaller at lower hydration levels and room temperature. However, at higher λ values of 25 and 40, these coefficients are comparable to those in Nafion at a λ value of 16. This study also shows that water diffusion in Ph-SPEEKK membranes at low hydration levels can be significantly improved by raising the temperature with important implications for proton conductivity.

I. INTRODUCTION Proton exchange membrane fuel cells (PEMFCs) and direct methanol fuel cells (DMFCs) are energy conversion devices that can help reduce global dependence on energy from fossil fuels. They convert the chemical energy of a fuel, such as hydrogen, methanol, or formic acid, into electrical energy with high efficiency and minimal environmental impact. The utility of PEMFCs has been demonstrated in automotive and portable power applications. However, progress in their widespread use has been hampered by the need for improvements in performance, durability, and service life of fuel cells while operating under challenging electrochemical conditions. These challenges are compounded by high production cost. A key component of PEMFCs is a membrane electrode assembly (MEA) containing a polymer membrane that separates the reactant gases and selectively conducts protons.1 Perfluorosulfonic acid (PFSA)-based membranes, such as Nafion developed by DuPont, are the most widely used in PEMFCs.2 PFSA membranes perform well below 90 °C in the presence of water, which is a reaction product that also serves as proton shuttle and coolant. However, operation of the PEMFC under anhydrous conditions at temperatures above 120 °C is desired to limit poisoning of the electrode catalysts by CO, enable the use of inexpensive catalysts, and enhance reaction kinetics. This has created the need to develop membranes with high proton conductivity under low hydration levels and thermal, mechanical, and chemical stability during prolonged operation at elevated r 2011 American Chemical Society

temperature. Decades of PEM research have led mostly to a phenomenological understanding of membrane performance. An alternative to the traditional trial and error approach to membrane development is to use the power of modern computational approaches and resources to gain a fundamental understanding of membrane chemistry, morphology, and dynamical properties of protons and water molecules to advance this technology. In particular, the transport of water molecules and protons confined within nanoscale regions of PFSA membranes, especially at low hydration levels, has yet to be well characterized.3 Despite recent efforts to develop alternative membranes,4-7 Nafion remains the benchmark for PEMFC membranes, and it continues to be the focus of many studies. Mauritz and Moore8 have reviewed the extensive experimental database on the morphology of Nafion membranes. Experimental studies have used a variety of X-ray and neutron scattering and diffraction techniques, electron microscopy and atomic force microscopy, and spectroscopic techniques to characterize the molecular structure of hydrated PFSA membranes. Because of the inability of experiments to observe directly the structure and dynamics of the membrane under different levels of hydration, there is no universally accepted model of the structure of the extensively studied Nafion membrane. Received: October 28, 2010 Revised: January 4, 2011 Published: February 3, 2011 1817

dx.doi.org/10.1021/jp110331m | J. Phys. Chem. B 2011, 115, 1817–1824

The Journal of Physical Chemistry B

ARTICLE

Figure 1. Depiction of Ph-SPEEKK monomer molecular structure and the calculated atomic point charges in e (as described in the methodology section). Hydrogen atoms are not shown for clarity.

The level of understanding is even much poorer regarding alternative polymer membranes proposed in the literature. The recently proposed ionic liquids7 hold great promise for use in PEMFCs, but there has been only a limited number of experimental and theoretical studies of ionic liquids relevant to this application. In Nafion, a hydrophobic CF2 backbone and a highly hydrophilic sulfonic acid-terminated side chain are known to form nanoscale domains within which ionic transport occurs.8,9 The acidic functional groups have great propensity for donating their protons in fully or partially solvated conditions. Besides sulfonic acid, phosphonic acid and imidazole functionalized compounds have been proposed as the protogenic group in polymer electrolyte membranes.10,11 There is considerable potential for optimization of conductivity, durability, and other membrane properties by modifying existing PFSA membranes, for instance by changing the protogenic group or modifying the side-chain length and separation between side chains along the backbone.12 Theory and computer simulation have a crucial role to play in such an optimization effort given the limitations of experiment and the lack of fundamental understanding of proposed membranes. Aromatic polymers with sulfonate groups in the pendants are promising candidates for higher-temperature low-humidity PEMFC membranes because they exhibit good conductivity, thermal stability, and chemical stability.4,13 Membranes based on sulfonated poly ether ether ketone (SPEEK) have been proposed for their potential to lower the cost of the MEA. However, conductivity has been found to be considerably less efficient than that in Nafion. On the other hand, blending sulfonated PEEK (SPEEK) with selected inert and apolar compounds has been found to yield SPEEKs with relative selectivity (proton conductivity divided by methanol permeability) comparable to that of Nafion. Recently, Liu et al.13 have synthesized phenylated sulfonated poly ether ether ketone ketone (Ph-SPEEKK) membranes that exhibit lower methanol permeability with proton conductivity comparable to that of Nafion at 100 °C. The chemical structure of this membrane is shown in Figure 1. The morphology of Ph-SPEEKK and its properties, especially the influence of water content on proton conductivity, are, however, poorly understood. Studies comparing Nafion and PEEK-based membranes have proposed that pKa, flexibility, hydrophilic clustering, and the distances between sulfonate groups may affect membrane efficiency.10,14 It has also been hypothesized that less hydrophobic/hydrophilic separation results in highly branched structures that produce lower electro-osmotic drag and permeation

coefficient.14 The present study makes use of molecular dynamics (MD) simulations to characterize the nanophase structure of a fully sulfonated Ph-SPEEKK 48-mer chain and the diffusion of water and hydronium ions throughout its matrix as a function of solvation.

II. DETAILS OF THE SIMULATION The present study involved atomistic simulations by classical MD, informed by quantum mechanical analysis of the partial atomic charges on the atoms of the polymeric unit. A 3D molecular model of a unit of Ph-SPEEKK, as shown in Figure 1, was built, and the geometry was optimized at the HF/631G* level using the NWChem program.15 Atomic charges were calculated by a RESP fitting16 and used in conjunction with the AMBER parm99 force field parameter set17 in the subsequent MD simulations. This methodology follows the approach in our previous theoretical studies of Nafion so that a comparison between the two membranes can be made. The polymer membrane model is made of a fully sulfonated 48-mer of Ph-SPEEKK, 48 H3Oþ ions, and a number of water molecules defining six levels of hydration, namely, λ = 3.5, 6, 11, 16, 25, and 40, where λ is the number of water molecules per sulfonate group. The initial model was built as a straight polymer chain, uniformly solvated at λ = 40 and with hydronium ions randomly distributed at a distance no shorter than 0.8 nm from any sulfonate group. The system was energy minimized using 10 000 steps of the steepest descent algorithm, and 20 ns MD-simulated annealing (MD/SA) cycles, over a 5 ns simulation, were used to collapse the PhSPEEKK chain into an amorphous membrane. Each MD/SA cycle consisted of a total of 250 ps divided into four annealing points, where the system was heated from 300 to 1500 K in 50 ps and maintained at the high temperature for an additional time of 150 ps before being cooled to 300 K over a period of 50 ps. The final structure obtained from the MD/SA was used as a starting point for all simulations. Water molecules were randomly removed for the systems bearing smaller λ values. A total of 50 ns MD simulation was carried out as production for each system in the isothermal-isobaric (NPT) ensemble using the leapfrog algorithm with a 2 fs integration step. The configurations were saved at every 0.5 ps for analyses. Temperature was maintained by coupling solute and solvent separately to the Berendsen thermostat with a relaxation time of 0.2 ps, and pressure was kept at 1 bar by coupling to a Berendsen barostat via isotropic coordinate scaling with a relaxation time of 0.5 ps 1818

dx.doi.org/10.1021/jp110331m |J. Phys. Chem. B 2011, 115, 1817–1824


Figure 2. Monomer- and sulfonate-averaged RMSF values over the 50 ns simulations as a function of hydration level. The solid line corresponds to the linear regression for the monomer-averaged values, whereas the dashed line is the linear regression for the sulfonateaveraged values. (Correlation coefficients are 0.77 and 0.85, respectively.)

and compressibility of 4.5 10-6 (kJ 3 mol-1 3 nm-3)-1. The SPC/E water model18 was used with its stretching and bending motions constrained using the LINCS algorithm.19 The Kusaka model20 was used for the force field parameters and partial charges for hydronium ions. A 1.4 nm cutoff was used for the short-range electrostatic and van der Waals interactions. Longrange electrostatic contributions were treated via the generalized reaction field method with ε = 66. All simulations were carried out using the GROMACS 4 program.21 A few caveats are in order here. The simulation time in this work was not sufficiently long to study polymer chain dynamics. Because the sulfonic acid group in Ph-SPEEKK lies further from the backbone oxygen relative to that in SPEEK, the degree of dissociation could be different. Effects such as proton dissociation and proton hopping cannot be modeled because nonreactive force fields were used. Moreover, in keeping with our aim of studying a simple model system for 50 ns, we modeled only a single chain in view of computational resource limits on the number of chains we could simulate for multiple hydration levels and temperatures. The chain was allowed to collapse on itself, and the simulation was carried out using periodic boundary conditions so that the polymer would be expected to behave as in a “bulk-like” system. This is ensured by the fact that the smallest box size dimension in the smallest system (λ = 3.5) is at least 2.5 times larger than the short cutoff and more than four times larger for the larger systems. The limits for convergence have not been systematically studied for these polymers, in terms of neither number of chains nor simulation time. The desire to start with simple models is also why we adopted an “unrealistic” degree of sulfonation of 100%. At this high degree of sulfonation, SPEEK would become soluble in water and would not be useful in a practical fuel cell. The stability of Ph-SPEEKK at this degree of sulfonation is not known. We hope to build more complexity and realism into the model in future.

III. RESULTS AND DISCUSSION Swelling, flexibility, hydrophilic clustering and the distances between sulfonate groups are some of the main properties

ARTICLE

Figure 3. Contact map for sulfonate groups in Ph-SPEEKK at different hydration levels. Data are averaged over the last 10 ns of the simulations, that is, 40-50 ns. (The term residue corresponds to one Ph-SPEEKK monomer.)

thought to affect membrane efficiency. To the best of our knowledge, experimental information about these quantities for Ph-SPEEKK is currently unavailable. In the present study, trajectory analyses have been carried out to evaluate the effect of increased hydration on the nanophase structure and dynamics of Ph-SPEEKK membranes. Our previous simulations of Nafion membranes22-24 have shown excellent agreement with experimental data; therefore, our previous work is used here as a benchmark for comparison when appropriate. A. Membrane Morphology and Flexibility As a Function of Hydration Level. The aromatic-rich structure of Ph-SPEEKK is naturally expected to display less flexibility when compared with Nafion membranes. However, it is not certain whether the backbone rigidity is correlated to the overall motion of the sulfonate groups. The root-mean-square fluctuation (RMSF) over the entire 50 ns simulation was calculated for all atoms and averaged over each residue and for the sulfonate groups only. A comparison of the overall motion is shown in Figure 2. It can be seen that an increase in the solvation level leads to a higher membrane overall flexibility. However, the effect is considerably larger for the sulfonate groups. At lower hydration levels (λ = 3.5 and 6), the sulfonate groups display a smaller fluctuation from initial position than the monomers they are attached to. An inversion on this behavior is observed for λ values of 11 and higher. The flexibility trend as a function of the hydration level is illustrated as the linear regressions of the RMSF based on the positions of the residue-average and the sulfonate group in Figure 2. An increase in hydration is also translated into a larger distance between sulfonate groups in the Ph-SPEEKK membrane. The average distance of the sulfonate groups was calculated for the last 10 ns of MD simulation for each λ value, and it is displayed in Figure 3 as contact maps. The contact maps represent the average distance between a pair of sulfonate groups over a 10 ns interval from 40 to 50 ns. The abundance of blue, green, and yellow patches is indicative of proximity among the sulfonate groups. Dispersed yellow spots in a red background indicate no proximity among sulfonates. Distances between groups span from 0 (represented as blue) to 1.5 nm (shown as red). The diagonal is blue and has no significance because it represents the 1819



Figure 4. Radial distribution function between the sulfonate groups of Ph-SPEEKK (represented by the sulfur atom) and (a) water oxygen, (b) hydronium oxygen, and (c) itself as a function of hydration level.

distance between a sulfonate group and itself. The maps show a tendency of the sulfonate groups to cluster at lower hydration levels. This pattern is nearly absent for λ values of 16 and higher, which indicates that the hydration level corresponding to a λ value between 11 and 16 is necessary to break sulfonate clustering in Ph-SPEEKK at room temperature. The corresponding λ value in Nafion is ∼5. Anion clustering is related to binding of hydronium ions by anions and inhibition of proton transport. The low-hydration clustering of sulfonate groups is further confirmed by analysis of the radial distribution functions g(r)

ARTICLE

between sulfonate groups (Figure 4c). An increase in hydration level leads to broadening of the well-defined g(r) peaks characteristic of low hydration levels. Also, a considerable shift in the g(r) is also observed. The first peak is located at 0.37 nm for λ = 3.5, whereas it is at 0.57 nm for λ = 40. Increased hydration leads to an increase in the separation distance between sulfonate groups. The sulfonate groups clustered at low hydration levels become dispersed at high hydration levels. Similar behavior was also observed for two independent simulations of SPEEK membranes at 40 and 50% degree of sulfonation.25-27 Because we are not aware of theoretical studies of Ph-SPEEKK in the literature, we will compare the present results with the findings of these studies of SPEEK with the caveat that there are clear structural differences between Ph-SPEEKK and SPEEK. For instance, there are two ketone groups in the former and one in the latter, and the sulfonic acid group resides farther from the backbone in the former. Brunello et al.25 found that the distance between sulfonate groups increases as a function of water content. They report the first peak in the sulfonate g(r) at distances of 0.44, 0.48, and 0.54 nm for λ values of 4.9, 6.7, and 11.1, respectively. Their results also show the loss of a defined S-S peak at λ = 11, in contrast with the findings of Mahajan and Ganesan,26,27 where, apart from relative intensities, a S-S peak was observed around 0.55 nm for all hydration levels studied (in the latter, λ equals 1.94, 4.12, 9.25, 16, and 25.05). This discrepancy may arise from structural differences between Ph-SPEEKK and SPEEK mentioned above. To a smaller extent, peak broadening is also observed in the radial distribution function between sulfonate groups and water molecules represented by the g(r) between sulfur atoms of the sulfonate group and the oxygen atoms of water molecules in Figure 4a. However, the first peak remains virtually unchanged at ca. 0.39 nm for all λ values. The significant decrease in peak amplitude with increasing hydration is consistent with previous findings in Nafion,22-24 although the structure beyond 0.5 nm is quite different in Ph-SPEEKK and Nafion. The g(r) between the sulfur atom and oxygen atoms of hydronium ions shows no difference as a function of hydration other than a modest reduction in amplitude (Figure 4b). The λ-independent values of 0.37 nm and ca. 0.56 nm correspond to the first and second peaks for hydronium oxygen atoms in all simulations. This shows that hydronium ions are consistently bound to sulfonate groups, even at hydration levels equivalent to λ = 40. Mahajan and Ganesan27 have previously observed similar binding of hydronium ion with the sulfonate anion in SPEEK. To evaluate the influence of hydration on the binding of hydronium ions to sulfonate groups, we performed hydrogen bond analyses between these groups as a function of hydration. A hydrogen bond D-H 3 3 3 A is assumed to be present when the distance H 3 3 3 A is 135°. The data summarized in Table 1 show that the average number of hydrogen bonds between hydronium ions and sulfonate groups decreases as solvation increases. At λ = 3.5, each hydronium ion makes on average two hydrogen bonds with sulfonate groups, whereas a single hydrogen bond is detected between these two groups when the hydration level is equivalent to λ = 40. The lifetime of these interactions seems to be independent of the hydration level with an average value of ca. 2 ns across all simulations. In the future, we will perform a detailed analysis of the residence times of H2O and H3Oþ around SO3- as a function of λ. Another measure of the number of hydronium ions and water molecules around sulfonate groups is the coordination number (CN), 1820



ARTICLE

Table 1. Hydrogen Bonds between Hydronium Ions and Sulfonate Groups hydrogen bond

average number of

hydration level (λ)

lifetime (ns)a

hydrogen bonds/H3Oþ

3.5 6

2.2 1.7

1.89 1.73

11

2.3

1.48

16

2.0

1.22

25

2.2

1.13

40

2.0

1.02

a

Hydrogen bond lifetime (or residence time) is expressed as the integral of the autocorrelation function. Data shown were calculated for the 25-50 ns interval.

which is estimated by integrating the first peaks in the pair correlation functions. The limits of integration are from the start of the peak to the first minimum. Figure 5 shows that the position of such minimum does not change significantly as a function of water content; therefore, for the sake of consistency, an average cutoff value of 0.47 nm was adopted and used as standard in all simulations. The same cutoff value was also used by Brunello and coworkers in an MD study of SPEEK membranes.25 The CN results for Ph-SPEEKK are presented in Figure 5. In agreement with previously reported studies for SPEEK,25-27 these results show that hydronium ions move further apart from the sulfonate groups with increasing hydration. As expected, the total CN increases with hydration at the determined cutoff value. Recently, Karo et al.28 determined that the terminal sulfonate groups in Nafion and Hyflon per fluoro sulfonic acid membrane is surrounded on the average by six water molecules and a hydronium ion for λ = 15, which is comparable to the present results for λ = 16. However, the RDF of oxygen of sulfonate anion and that of hydronium ion at low temperature in this membrane is remarkably different than that in Nafion, where the hydronium detaches from the sulfonate anion efficiently with increasing water content. It is worth noting that the CNs for hydronium ions with respect to sulfonate groups are only marginally higher than the calculated number of hydrogen bonds between these two groups (Figure 5 and Table 1, respectively). This result suggests that every hydronium ion near a sulfonate group forms a hydrogen bond with the latter. The number of sulfonate-coordinated hydronium ions is roughly 2 at λ 3.5 and decreases to ca. 1 only when λ is equal to 40. The diffusion coefficient for water molecules and hydronium ions in PEMFCs can be directly correlated to the proton exchange efficiency, especially the vehicular component of proton transport. The structural transport of protons by the Grotthuss mechanism is beyond the scope of the present work. Table 2 shows the calculated diffusion coefficient for water molecules and hydronium ions in Ph-SPEEKK as a function of the hydration level over the entire 50 ns period. Values for Nafion from our previous simulation of a Nafion membrane24 at λ equal to 3.5, 6, 11, and 16 are listed for comparison. Experimental diffusion coefficients for water in SPEEK at 353 K for λ values of 6.7 and 11 are (0.01 and 0.2) 10-5 cm2/s, respectively.14 At λ = 11 and 350 K, the present simulation yields 0.237 10-5 cm2/s. These results indicate that at room temperature the proton exchange efficiency of Ph-SPEEKK membrane does not compare well to that of Nafion at the same hydration level. Diffusion coefficients are one order of magnitude smaller in all cases.

Figure 5. Coordination numbers for water molecules and hydronium ions around Ph-SPEEKK sulfonate groups as a function of hydration level at a 0.47 nm cutoff value.

However, at higher hydration levels (λ = 25 and 40), these values reach the same magnitude observed for Nafion at lower hydration levels (λ from 3.5 to 16). Another important aspect of PEMFCs is the water-swelling propensity characteristic of each material. To evaluate this property, we calculated the radius of gyration as a function of λ. Figure 6a shows these values as a function of the 50 ns trajectory. This value converges after 10 ns of simulation in all simulations. However, the most important finding is that the internal membrane structure of Ph-SPEEKK is only perturbed by hydration levels above λ = 16. A ca. 10% increase in radius of gyration is observed when Ph-SPEEKK is hydrated to λ values of 25 and 40. The fact that no significant difference is observed for λ in the range from 3.5 to 16 and from 25 to 40 suggests that water swelling in Ph-SPEEKK is a discrete property. Final configurations of the simulations for λ = 11 and 40 are shown in Figure 6b,c, respectively. Water channels are represented via van der Walls surfaces, illustrating the difference in solvation upon a 10% increase in the radius of gyration of the membrane. The same structures are displayed in Figure 7, where water molecules were removed to illustrate the internal structure and nanophase segregation of the material upon the increase in swelling. The simulation at λ = 11 shows a higher incidence of sulfonate clustering compared with a better solvated membrane (λ = 40), therefore explaining the lower diffusion coefficients observed for Ph-SPEEKK at lower hydration levels. Such high levels of sulfonate clustering involving three to four SO3- groups located within a distance of 0.8 nm have been observed by Karo et al.28 recently in Hyflon perfluoro sulfonic acid membrane hydrated to a level of λ = 15. These authors also observed clustering of H3Oþ near SO3- groups. B. Effect of Temperature on the Internal Structure of PhSPEEKK. A significant change in nanophase segregation, solvation, and diffusion coefficients for water molecules and hydroniun ions is observed upon transition from λ = 11 to 40. To evaluate the effect of temperature on Ph-SPEEK membrane, we calculated the same quantities for two 50 ns runs at temperatures of 350 and 400 K and λ = 11. The results are compared with the runs at 300 K and with our previous simulations of Nafion at 350 K.24 The radius of gyration as a function of temperature for λ = 11 remained constant with average values of 1.98, 1.91, and 1.94 nm 1821



ARTICLE

Table 2. Comparison of the Diffusion Coefficients for H2O and H3Oþ in Ph-SPEEKK and Nafion As a Function of Hydration Level

a

hydration level (λ)

D(H2O) ( 10-5 cm2/s)

D(H2O)Nafion ( 10-5 cm2/s)a

D(H3Oþ) (x10-5 cm2/s)

D(H3Oþ)Nafion ( 10-5 cm2/s)a

3.5

0.014

0.329

0.005

0.030

6 11

0.020 0.084

0.311 0.464

0.003 0.004

0.022 0.048

16

0.176

0.589

0.008

0.068

25

0.417

0.025

40

0.884

0.058

Values from Venkatnathan et al., 2007.24

Figure 6. (a) Radius of gyration for Ph-SPEEKK membranes as a function of hydration level. (b,c) Final structures from the simulations at λ = 11 and 40, respectively. Water channel network is illustrated by the connecting vdW surface in gray. Ph-SPEEKK backbone (herein defined as the PEEKK chain) is shown as blue ribbons, and the sulfonate groups and hydronium ions are shown in stick model. Atoms are color-coded as cyan: carbon, yellow: sulfur, red: oxygen, and white: hydrogen.

for temperatures of 300, 350, and 400 K, respectively. This finding suggests that membrane swelling has a much higher dependence on water content than on temperature. The calculated contact map for sulfonate groups in Figure 8 shows only minor changes in the spatial distribution of these groups with temperature, which suggests that the membrane structure is not significantly changed by increasing the temperature from 300 to 400 K. Average densities were also calculated and are shown in Figure 9. Linear regression for the data points at 300 K is shown

as a solid line. An increase in temperature results in lower densities as seen for λ = 11, while increasing hydration leads to overall higher density values. However, the average density seems to converge for higher values λ > 25 (Figure 9). Experimental data for Ph-SPEEKK is not available for comparison. Compiled data for RMSF, hydrogen-bond data, and diffusion coefficients for water molecules and hydronium ions at λ = 11 and temperatures of 300, 350, and 400 K are presented in Table 3. Overall structure fluctuation is affected at only 400 K, in contrast 1822



ARTICLE

Table 3. RMSF, Hydrogen-Bond Data, and Diffusion Coefficients for Water and Hydronium Ions in Ph-SPEEKK Simulations at λ = 11 As a Function of Temperature property/λ

a

Figure 7. Final structures from the simulations at (a) λ = 11 and (b) λ = 40, the same as shown in Figure 6. Red circles highlight the presence of sulfonate clusters formed by at least four sulfonate groups bridged by one water/hydronium ion.

Figure 8. Contact map for sulfonate groups in Ph-SPEEKK at λ = 11 as function of the temperature: 300, 350, and 400 K. Data are shown as averaged over the last 10 ns of the simulations, that is, 40-50 ns. (The term residue corresponds to one Ph-SPEEKK monomer.)

Figure 9. Average densities for the simulated hydrated Ph-SPEEKK systems as a function of hydration level. Values are shown as averages over the entire trajectory. Error bars represent the average partial density variation throughout the boxes.

with hydrogen-bonding properties. An increase in temperature does not affect the average number of hydrogen bonds between

11(300 K)

11(350 K)

11(400 K)

RMSF residue average (nm)

0.157

0.168

0.207

RMSF SO3- (nm)

0.171

0.173

0.236

H-bond lifetime (ns)

2.3

1.6

1.5

H-bond average number

1.48

1.59

1.56 0.760

D(H2O) ( 10-5 cm2/s)

0.084

0.237

D(H2O)Nafion ( 10-5 cm2/s)a

0.464

1.093

D(H3Oþ) ( 10-5 cm2/s)

0.004

0.011

D(H3Oþ)Nafion ( 10-5 cm2/s)a

0.048

0.119

0.015

Values from Venkatnathan et al., 2007.24

sulfonate groups and hydronium ions; however, a decrease in their lifetime of ca. 35% is observed for simulations at 350 and 400 K. This shows that an increase in temperature improves the dynamics of exchange of sulfonate-hydronium ion partners. This is supported by the increase of one order of magnitude in the diffusion coefficients for water molecules and hydronium ions at higher temperatures (Table 3). This result is promising because it shows that highly hydrated Ph-SPEEKK at 350 K holds similar proton-exchange potential as Nafion at 300 K (data can be verified by comparing values in Tables 2 and 3). In fact, the high efficiency of Ph-SPEEKK membranes at higher temperatures has been recently described by a set of experiments carried out by Liu and coworkers.29 In a study involving two series of SPEEK-based membranes, Ph-SPEEK presented the best balance of properties (dimensional stability, methanol permeability, and high proton conductivity), achieving higher current densities than Nafion.

IV. CONCLUSIONS Molecular dynamics simulations of fully sulfonated Ph-SPEEKK membranes were carried out to evaluate the influence of hydration level on its internal structure, nanophase segregration, and potential for proton exchange. The membrane displayed low flexibility and a well-defined solvation structure, both properties reasonably independent of the amount of water present. An increase in hydration level led to an increase in the overall distance between sulfonate groups. However, water swelling seems to be a discrete property in Ph-SPEEKK. At 300 K and low hydration levels (from λ = 3.5 to 16), the diffusion coefficient of water molecules and hydronium ions was found to be about an order of magnitude lower than the corresponding values reported for Nafion. These diffusion coefficients in Ph-SPEEKK at high hydration levels (λ = 25 and 40) are comparable to those in Nafion at lower hydration levels (λ from 3.5 to 16). Alternatively, a 50 K increase in temperature of Ph-SPEEKK at λ = 11 results in solvent diffusion coefficient data that are similar to the ones observed for Nafion on the same hydration level at room temperature. This result is supported by recent experimental results for Ph-SPEEKK at elevated temperatures, where higher overall performance is reported over Nafion.28 In light of the known advantages of SPEEK membranes such as low methanol permeability, low cost, and ease of synthesis, the present results place Ph-SPEEKK-based membranes as promising candidates for DMFC applications at elevated temperatures. 1823



’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Tel: 1-509-371-6487. Fax: 1-509-371-6242.

’ ACKNOWLEDGMENT R.D.L. acknowledges CNPq, INCT-INAMI, and FACEPE for partial support. Computer time was provided by the Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by the Department of Energy’s (DOE) Office of Biological and Environmental Research located at Pacific Northwest National Laboratory, and the National Energy Research Scientific Computing Center (NERSC), which is supported by the Office of Science of DOE under contract no. DE-AC02-05CH11231. This work is supported by the DOE Office of Basic Energy Sciences, Chemical Sciences, Geosciences, and Biosciences Division. Pacific Northwest National Laboratory is operated for the DOE by Battelle Memorial Institute under contract DE-AC05-76RLO1830.

ARTICLE

(20) Kusaka, I.; Wang, Z.-G.; Seinfeld, J.-H. J. Chem. Phys. 1998, 108, 6829. (21) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008, 4, 435. (22) Devanathan, R.; Venkatnathan, A.; Dupuis, M. J. Phys. Chem. B 2007, 111, 8069. (23) Devanathan, R.; Venkatnathan, A.; Dupuis, M. J. Phys. Chem. B 2007, 111, 13006. (24) Venkatnathan, A.; Devanathan, R.; Dupuis, M. J. Phys. Chem. B 2007, 111, 7234. (25) Brunello, G.; Lee, S. G.; Jang, S. S.; Qi, Y. J. Renewable Sustainable Energy 2009, 1, 033101–1. (26) Mahajan, C. V.; Ganesan, V. J. Phys. Chem. B 2010, 114, 8357. (27) Mahajan, C. V.; Ganesan, V. J. Phys. Chem. B 2010, 114, 8367. (28) Karo, J.; Aabloo, A.; Thomas, J. O.; Brandell, D. J. Phys. Chem. B 2010, 114, 6056. (29) Liu, B.; Kim, Y. S.; Hu, W.; Robertson, G. P.; Pivovar, B. S.; Guiver, M. D. J. Power Sources 2008, 185, 899–903.

’ REFERENCES (1) Devanathan, R. Energy Environ. Sci. 2008, 1, 101. (2) Kreuer, K. D.; Paddison, S. J.; Spohr, E.; Schuster, M. Chem. Rev. 2004, 104, 4637. (3) Paciaroni, A.; Casciola, M.; Cornicchi, E.; Marconi, M.; Onori, G.; Pica, M.; Narducci, R. J. Phys. Chem. B 2006, 110, 13769. (4) Hickner, M. A.; Ghassemi, H.; Kim, Y. S.; Einsla, B. R.; McGrath, J. E. Chem. Rev. 2004, 104, 4587. (5) Hogarth, W. H. J.; da Costa, J. C. D.; Lu, G. Q. J. Power Sources 2005, 142, 223. (6) Xu, W.; Angell, C. A. Science 2003, 302, 422–425. (7) Nakamoto, H.; Watanabe, M. Chem. Commun. 2007, 24, 2539. (8) Mauritz, K. A.; Moore, R. B. Chem. Rev. 2004, 104, 4535. (9) Eikerling, M.; Kornyshev, A. A.; Kucernak, A. R. Phys. Today 2006, 59, 38. (10) Schuster, M.; Rager, T.; Noda, A.; Kreuer, K. D.; Maier, J. Fuel Cells 2005, 5, 355. (11) Paddison, S. J.; Kreuer, K. D.; Maier, J. Phys. Chem. Chem. Phys. 2006, 8, 4530. (12) Paddison, S. J.; Elliott, J. A. Solid State Ionics 2006, 177, 2385. (13) Liu, B.; Robertson, G. P.; Kim, D. S.; Guiver, M. D.; Hu, W.; Jiang, Z. Macromolecules 2007, 40, 1934. (14) Kreuer, K. D. J. Membr. Sci. 2001, 185, 29. (15) Apra, E.; Windus, T. L.; Straatsma, T. P.; E.J., B.; de Jong, W.; Kowalski, K.; Hirata, S.; Valiev, M.; Hackler, M. T.; Zhao, Y.; Harrison, R. J.; Dupuis, M.; Smith, D. M. A.; Nieplocha, J.; Tipparaju, V.; Krishnan, M.; Auer, A. A.; Brown, E.; Cisneros, G.; Fann, G. I.; Frucht, H.; Garza, J.; Hirao, K.; Kendal, R.; Nichols, J. A.; Tsemekhaman, K.; Wolinski, K.; Anchell, J.; Bernhold, D.; Borowski, P.; Clark, T.; Clerc, D.; Dachsel, H.; Deegan, M.; Dyall, K.; Elwood, D.; Glendening, E.; Gutowski, M.; Hess, A.; Jaffe, J.; Johnson, B.; Ju, J.; Kobayashi, R.; Kutteh, R.; Lin, Z.; Littlefield, R.; Long, X.; Meng, B.; Nakajima, T.; Niu, S.; Pollack, L.; Rosing, M.; Sandrone, G.; Stave, M.; Taylor, H.; Thomas, G.; van Lenthe, J.; Wong, A.; Zhang, Z. NWChem, a Computational Chemistry Package for Parallel Computers, version 4.7; Pacific Northwest National Laboratory: Richland, Washington, 2005. (16) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. J. Phys. Chem. 1993, 97, 10269. (17) Cheatham, T. E.; Cieplak, P.; Kollman, P. A. J. Biomol. Struct. Dyn. 1999, 16, 845. (18) Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P. J. Phys. Chem. 1987, 91, 6269. (19) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. J. Comput. Chem. 1997, 18, 1463. 1824


Modeling the Nanophase Structural Dynamics of Phenylated

Recommend Documents