J. Phys. Chem. B 2009, 113, 10679–10685
10679
Investigation of Water and Methanol Sorption in Monovalent- and Multivalent-Ion-Exchanged Nafion Membranes Using Electron Spin Resonance Jamie S. Lawton and David E. Budil* Department of Chemistry and Chemical Biology, Northeastern UniVersity, Boston, Massachusetts 02115 ReceiVed: March 26, 2009; ReVised Manuscript ReceiVed: June 18, 2009
Electron spin resonance (ESR) spectroscopy was used to monitor the local environment of 2,2,6,6-tetramethyl4-piperidone N-oxide (TEMPONE) spin probe in Li+, Ca2+, and Al3+ ion-exchanged Nafion 117 membranes swollen with mixed methanol/water solvent at varying compositions. The 14N hyperfine splitting, aN, which reflects the local polarity of the nitroxide probe, remains nearly steady at higher solvent contents but increases substantially at lower solvent contents, reflecting close contact with the ions. The rotational rate (R) of the probe increased with solvent content, depending strongly on the amount of solvent at low contents but increasing more gradually at higher solvent contents, similar to the behavior of previously measured solvent translation diffusion coefficients. The rotational rate data from water-containing membranes were fitted using the Fujita free-volume diffusion model, which indicated that multivalent ions tend to increase the free volume fraction of the polymer while decreasing that of the solvent phase. Methanol-containing membranes exhibited greater variation with different exchange ions, but the data could not be fit using the free-volume model, suggesting that the assumption of two phases underlying the free-volume model might not apply to this case. The difference in the trends of swelling between water and methanol is consistent with previous results that have indicated different patterns of penetration for the two solvents. The results are interpreted in terms of changes in membrane morphology with higher-valence ions. Introduction Permselective membranes play a central role in a wide variety of devices including fuel cells, permselective materials for chemical reactions, water treatment,1 and barriers in textiles that selectively block the permeation of organic molecules such as toxins or warfare agents.2-5 In addition, there has been interest in such membranes in aqueous-organic electrolyte solutions because of their possible industrial and medical applications.6,7 The most commonly used cation-exchange membrane for many of these applications is Nafion, a perfluorosulfonic acid polymer manufactured by DuPont. Currently, its major application is as a proton-exchange membrane for fuel cells including direct methanol fuel cells (DMFCs).8 Because the structures through which diffusants permeate are determined by the phase behavior of Nafion rather than by chemical cross-linking, membrane permeability can depend strongly on the method of preconditioning. Cation exchange has frequently been explored as a means of tailoring membrane properties to the needs of specific applications. Thus, exchange with different cations has been used to modify the permeability and uptake properties of Nafion. Although fully cation-exchanged membranes are not optimal for use in fuel cells because of their decreased conductivity, fundamental studies of morphological changes caused by multivalent ions are nonetheless important for understanding how the leaching of metal ions from catalyst layers into the membrane can alter the fuel cell environment.9 Such studies are also relevant for applications as permselective materials for controlling chemical reactions, water treatment, and protective clothing. Although a number of models describing the structure of hydrated Nafion have been investigated, the effects of * To whom correspondence should be addressed. E-mail:
[email protected]. Fax: 617-373-3697.
multivalent ions and organic solvents on the morphology of the membrane are still to be fully elucidated. It has been established that, in the hydrated membrane, the ionic groups at the end of the side chains of the perfluorinated backbone form clusters within the perfluorinated polymer matrix and feature an extended phase of water in hydrated membranes.10,11 The earliest model of the morphology of the hydrated Nafion membrane is from the small-angle X-ray scattering (SAXS) studies of Gierke et al.,12 who proposed a cluster-network model with a hydrophilic phase characterized by 4-nm-diameter clusters of sulfonate-capped ether groups in the form of inverted micelles filled with water that are connected by 1-nm-diameter channels lined with sulfonic acid groups. Shortly thereafter, Yeager et al.13 proposed an interfacial region between the hydrophobic and hydrophilic phases that was described as mostly vacant space with side-chain material and residual water and sulfonate groups that were unincorporated into the hydrophilic phase. In their review, Mauritz and Moore14 concluded that the most plausible model is that with three phases with hydrophilic and hydrophobic phases characterized by nebulous boundaries and buffered by an interfacial region. Previous studies have investigated effects of solvent uptake, including water and methanol, in Nafion membranes exchanged with monovalent15 and divalent16,17 ions. The exchange of monovalent and divalent ions into the Nafion membrane has been studied considerably more than the exchange of trivalent ions. Some works involving Al3+ exchange in the membrane include investigations into minimization of methanol crossover in a DMFC,18 water treatment,1 membrane ion-exchange properties,19,20 membrane degradation, and isotectic polymerization processes by Ziegler-Natta catalysts.21 Hydration studies have investigated changes in membrane properties with respect to water content. The dependence of the
10.1021/jp902750j CCC: $40.75 2009 American Chemical Society Published on Web 07/10/2009
10680
J. Phys. Chem. B, Vol. 113, No. 31, 2009
Lawton and Budil such a model. The implications of these results for membrane morphology are discussed. Experimental Section
Figure 1. Rotational diffusion rate constant (top) and nitrogen hyperfine splitting (bottom) of TEMPONE spin probe in ion-exchanged Nafion membranes soaked in varying concentrations of methanol in water. Diamonds, squares, circles, and triangles represent H+, Li+exchanged, Ca2+-exchanged, and Al3+-exchanged Nafion respectively.
cluster size on hydration level has been investigated by a number of techniques including SAXS,22 nuclear magnetic resonance (NMR) cryoporometry,23 and atomic force microscopy (AFM). NMR cryoporometry uses the Gibbs-Thomson equation, which relates the melting point depression of a confined liquid to the diameter of the confinement.23 Atomistic simulations of Nafion membranes of varying water contents showed an increase in average water-sulfur distance with increased water content24 and larger diffusion coefficients for water and hydronium ions at higher water contents.25 NMR studies of water self-diffusion in Nafion 117 membranes showed that the self-diffusion coefficient of water increases with increasing water content, with a rapid increase at low water contents and a lower rate of increase at higher water contents.26 Electron spin resonance (ESR) spectroscopy has been a major tool for elucidating structural characteristics of ion-exchange membranes.27-29 Recently, the effects of methanol on the microviscosity, local ordering, and local polarity in Li+exchanged Nafion 117 membranes were investigated by ESR spectroscopy.30,31 Here, we extend these initial investigations to characterize fundamental morphological changes in the membrane that can be induced by multivalent ions. In particular, we investigate the effects of multivalent ion exchange on the rotational rate of the spin probe and solvent uptake. We find clear correlations between the solvation of Nafion membranes, the identity of the exchange ion, and the rotational rate of the spin probe. In the case of water-containing membranes, these correlations can be interpreted in terms of a free-volume model by analogy with translational diffusion measurements. However, membranes containing methanol and monovalent ions exhibit qualitatively different behavior that cannot be explained with
Nafion 117 membranes were first purified by being heated to 75 °C for 1 h in 3% hydrogen peroxide, followed by 1 h in deionized water, 1 h in 0.5 M sulfuric acid, and 1 h in deionized water. Nafion membranes were ion-exchanged by being soaked in 0.06 M chloride salt solutions for 2 weeks. The salts used included LiCl (Sigma-Aldrich), CaCl2 (Sigma-Aldrich), and AlCl3 (Sigma-Aldrich). Solution uptake measurements were carried out gravimetrically using a Cahn C-33 microbalance. Membranes were dried under a vacuum at 80 °C for 24 h, weighed to determine the initial mass, and equilibrated in solvent for 24 h at ambient temperature. Excess solution was wiped from the surface of each membrane before it was weighed to determine the fully swollen mass. It is important to mention that a number of studies have shown that the solvent uptake in Nafion membranes varies significantly with different environmental factors.32-35 Therefore, great care was taken to treat each membrane sample identically throughout the process of purifying, exchanging, drying, and soaking. Energy-dispersive x-ray spectroscopy (EDS) measurements were taken on a Hitachi S4800 scanning electron microscope equipped with an EDAX EDS detector. An accelerating voltage of 10 kV was used with 10 mA probe current. EDS data were collected at a magnification of 400× to use a wide window to detect the dispersed ions in the membrane matrix. The aluminum sample stage was covered with a copper tape to avoid interference with the sample signal of the Al3+-exchanged Nafion sample. EDS spectra for Nafion exchanged with different ions are given in the Supporting Information. Hydration measurements were accomplished by soaking the exchanged membranes in 0.1 mM 2,2,6,6-tetramethyl-4-piperidone N-oxide (TEMPONE) spin probe in water or methanol. The solvent uptake was calculated using the equation
λtotal )
Wwet - Wdry EWmembrane Wdry MWsolvent
(1)
where λ is the number of water molecules per equivalent of membrane; Wwet and Wdry are the weights of the membrane when wet and dry, respectively; EWmembrane is the equivalent weight of the membrane; and MWsolvent is the molecular weight of the swelling solvent. The low concentration of probe was selected so that the mass of probe in the membrane would be below the limit of detection of the microbalance and would not interfere with determining the solvent content in the membrane. H+ Nafion samples were not used in these studies because the probe lifetime was not long enough in the acidic environment to gather data over the full hydration range. For studies of varying methanol content, mixed solvents were prepared with different concentrations of methanol in water over a range of 0%, 30%, 70%, and 100% methanol by volume, with 0.25 mM 2,2,6,6-tetramethyl-4-piperidone N-oxide (TEMPONE) spin probe. The TEMPONE spin probe was selected both to enable direct comparison with previous spin-probe studies and because, in contrast to other spin probes, TEMPONE is soluble only in the water phase of the membrane, producing singlecomponent spectra that can be analyzed unambiguously. In addition, the hydrogen bonding of TEMPONE is weak enough to allow interpretation of rotational diffusion in terms of a freevolume model.
Investigation of Water and Methanol Sorption
J. Phys. Chem. B, Vol. 113, No. 31, 2009 10681
Figure 2. Rotational rate R (upper plots) and isotropic hyperfine splitting aN (lower plots) of TEMPONE spin probe in Nafion swollen with water (left side) and methanol (right side) plotted versus number of moles of solvent present per sulfonic acid group. Triangles, squares, and circles represent Nafion exchanged with Li+, Ca2+, and Al3+, respectively.
All samples were deoxygenated in a glovebag by bubbling with ultra-high-purity argon that had been passed through an oxygen/moisture trap (model OT-4-SS, Agilent Technologies). The deoxygenated samples were placed in a sealed quartz sample tube, and ESR spectra were obtained on an X-band Bruker EMX spectrometer. For each spectrum, three scans of 2048 points each were averaged using magnetic field modulation of 0.02 mT at 100 kHz, a time constant of 20.48 ms, and a conversion time of 81.92 ms. The fitting method utilized a MATLAB (MathWorks) based version of EPRLL, the slow-motional line shape program of Freed and co-workers,36,37 which can include the microscopic order, macroscopic disorder (MOMD) model in the case of local ordering of the spin probe. The parameters that were varied during the fitting procedure included the isotropic Gaussian inhomogeneous broadening ∆ib, the Azz tensor component, and the orienting potential parameter c20. The Azz tensor component was then used to determine values for gxx and gyy, as described previously.30 Several groups have observed that the relation between the isotropic g value, giso, and aN (or that between gxx and Azz) is approximately linear.38-41 This relationship was used to scale gxx and gyy with respect to Azz because the only tensor components that are appreciably affected by the local electric field are gxx, gyy, and Azz.42 Reported slopes of g0 versus aN in protic solvents38-41 range from -3.2 × 10-4 to -4.0 × 10-4 G-1. We then made the approximation that the quantity ∆gx ) (gx - gz) is linearly related to the quantity ∆gy ) (gy - gz). This linear relationship is evident from the extensive data tabulation of Lebedev for different nitroxide spin probes,43 for which a linear fit gives ∆gy ≈ 0.20∆gx. The remaining parameters were fixed at the values gzz ) 2.0023, Axx/γe ) 5.0 G, and Ayy/γe ) 5.5 G. Also varied in the least-squares fitting procedure was the isotropic rotational diffusion constant R, which is related to the rotational correlation time by τc ) 1/(6R). Selected examples of spectra from TEMPONE in different membranes, together with the least-squares calculated line shapes, are shown in the Supporting Information.
TABLE 1: Water and Methanol Uptakes for Ion-Exchanged Nafion Equilibrated in Solventa ion
water uptake
methanol uptake
H+ Li+ Ca2+ Al3+
22.7 ( 0.3 18.7 ( 0.4 16.1 ( 0.2 7.9 ( 0.6
34.1 ( 0.6 36.9 ( 0.9 9.0 ( 0.6 6.9 ( 0.8
a Reported uncertainties represent standard deviations of at least three separate measurements.
Results The measured uptakes of water and methanol are reported in Table 1 for each of the ion-exchanged membranes. In general, membranes exchanged with monovalent ions exhibited a greater uptake of water than those exchanged with multivalent ions. A similar trend was seen for the membranes equilibrated with methanol. Although there is evidence that the solvent composition inside the membrane is the same as that of the equilibrating solution for H+ and Li+ membranes,44 this is not necessarily true for the multivalent cations. Therefore, we did not gravimetrically determine solvent uptake for water-methanol mixtures. Figure 1 shows the dependence of the probe rotational rate R (upper plot) and the isotropic hyperfine coupling aN (lower plot) on the composition of the water-methanol solvent phase for membranes exchanged with H+, Li+, Ca2+, and Al3+ ions. As has been shown previously, the R parameter reflects the microscopic viscosity experienced by the spin probe, which is, in turn, related to the free volume available to the spin probe for rotation, whereas aN reflects the effective local polarity experienced by the probe. Both of these parameters exhibit a significant dependence on the solvent composition and the identity of the exchange ion. For the H+- and Li+-exchanged membranes, the rotation increased significantly with methanol concentration, as has been reported previously.30 This increase was much less pronounced for Ca2+-exchanged membranes and essentially absent for Al3+
10682
J. Phys. Chem. B, Vol. 113, No. 31, 2009
Lawton and Budil
membranes. As shown in the lower plot of Figure 1, the observed dependence of aN on solvent composition for the monovalent ions H+ and Li+ is similar to that previously reported,30 and there does not appear to be a significant effect of the ion valence on the polarity sensed by the probe over the full range of methanol concentrations. Figure 2 shows the probe rotational rate and the isotropic hyperfine splitting, aN, as functions of the solvent content for Nafion membranes exchanged with each of the ions studied and equilibrated with water (left) and with methanol (right). In water, the rotational rate of the probe increases rapidly with the addition of solvent at low water content and then levels off somewhat at water contents above about 7.5 water molecules per sulfonic acid group. Corresponding to the transition in the rotational behavior observed at ∼7.5 water molecules per sulfonate group, a transition is also apparent in the dependence of aN on solvent at a similar water content (cf. lower left-hand plot of Figure 2). At higher water contents, all samples exhibit an average splitting of aN/γe ) 15.9 G, comparable to the splitting observed in bulk water and independent of the identity of the exchange ion. At lower water contents, the hyperfine splitting increases dramatically as water is removed, and the effective local polarity becomes strongly dependent on the identity of the exchange ion, with Ca2+ producing the highest effective polarity. The overall dependence of the probe rotational rate on methanol sorption (top right, Figure 2) is qualitatively different from that observed for water. The trends are somewhat similar at low methanol levels, although the influence of the ions is reversed, with Li+-exchanged membranes exhibiting the lowest rotational rates, comparable to those observed in water (note the difference in vertical scales between the two upper plots in Figure 2), and significantly faster rotation in the presence of Ca2+. The most dramatic differences between water and methanol are observed for the Li+-exchanged Nafion. The trend of R is completely reversed at higher solvent contents, and probe rotation increases dramatically at methanol contents between 13 and 25 mol per equivalent, eventually leveling off in a sigmoidal fashion at the highest methanol contents. Similar trends in effective solvent diffusion were reported by Riven et al.45 in Nafion swollen with methanol, ethanol, and propanol. Although free-volume theories are most typically applied to describe translational diffusion in polymer-solvent systems,46-49 spin-probe rotational diffusion can also be described by such a model.50,51 This approach is supported by the qualitative similarities between the solvent dependence of the rotational diffusion in the present work and the reported solvent dependence of translational diffusion.45 The rotational diffusion data were therefore analyzed in terms of a simplified form of the Fujita free-volume theory,48,49,52 as described below. According to this model, the rotational diffusion constant can be written as
( )
R ) Ad exp
-Bd Vf
(2)
ion
ln(Ad/s)
B
G
+
18.61 ( 0.03 18.64 ( 0.06 18.65 ( 0.18
2.97 ( 0.73 1.63 ( 0.12 1.66 ( 0.09
37.09 ( 13.9 13.28 ( 3.2 12.62 ( 5.5
Li Ca2+ Al3+
a Reported uncertainties estimated from the curvature matrix of χ2 and the t-distribution for the 95% confidence level.
Figure 3. Rotational rate of TEMPONE probe in Nafion swollen with water plotted versus volume fraction of solvent. Triangles, squares, and circles represent Li+-exchanged, Ca2+-exchanged, and Al3+exchanged Nafion, respectively. Lines represent least-squares curves calculated according to eq 5 with parameters given in Table 2.
In eq 2, Bd is a parameter that depends on the relative sizes of the penetrant and the segment of the polymer involved in viscous flow,53 and Ad is related to the rotational diffusion of bulk penetrant [R(1)] as follows
( )
R(1) ) Ad exp
-Bd Vf1
(4)
Schneider et al.54 simplified the model of Stern and co-workers53,55 to three parameters, Ad, B, and G
(
R ) Ad exp
-B φ2 + Gφ1
)
(5)
where B ) Bd/Vf2 and G ) Vf1/Vf2. Equation 5 was used to fit the free-volume model to the experimental measurements of spin-probe R value versus solvent volume fraction. Values for φ1 and φ2 were calculated from the measured solvent uptake using the nearly linear relationship between density and solvent content reported for swollen ionexchanged polymers by Nandan et al.16 The results for the fitting parameters Ad, B, and G are reported in Table 2, and the leastsquares fits are shown by solid lines in Figure 3. Discussion
where Vf is the average fractional free volume of the system, which can be calculated from the fractional-free-volume contributions from the penetrant (Vf1) and the polymer (Vf2) and their respective volume fractions φ1 and φ2 as
Vf ) Vf2φ2 + Vf1φ1
TABLE 2: Least-Squares Free-Volume Parameters for Ion-Exchanged Nafion in Watera
(3)
Solvent Sorption and Ionic Radius. The observed values for water and methanol uptake in Li+- and Ca2+-exchanged Nafion reported in Table 1 are slightly higher than those calculated from solubility measurements by Villaluenga et al.17 and Nandan et al.;16 those for H+ and Li+ slightly lower than those determined by Evans et al.15 and comparable to the values calculated by Sun et al.56 for H+ Nafion equilibrated with water
Investigation of Water and Methanol Sorption and methanol. Differences in sorption values in the literature have been attributed to differences in the treatment of the membrane such as predrying, boiling, or equilibration method.35 Studies of solvent sorption in Nafion exchanged with different cations by Villaluenga et al.17 suggested that water uptake depends more on ion radius than charge, although this trend was less clear in the presence of methanol. Nandan et al.16 reported similar findings; however, neither Villaluenga et al.17 nor Nandan et al.16 considered any trivalent ions. In contrast, we found the least water uptake in the Al3+-exchanged membrane (cf. Table 1), even though the radii of the ions used in this work follow the sequence Al3+ < Li+ < Ca2+.57 One possible explanation for this observation in terms of ionsize effects is that the aluminum ion drags chloride ions from the ion-exchange solution into the membrane, leading to an effectively larger ion with a lower effective valence. Young et al.20 detected trace amounts of residual Cl in ion-exchanged Nafion using prompt-gamma neutron activation analysis (PGAA). To assess this possibility, EDS analysis was carried out (see EDS spectra in the Supporting Information). Whereas EDS revealed detectable levels of Al and Ca in the membrane, no Cl signal was observed. Although this does not rule out the presence of small amounts of Cl below the detection limit of the instrument, it does indicate that most of the Al3+ ions are not associated with Cl- ions. We conclude that ion-radius effects are most likely not responsible for the reduced solvent sorption in the presence of Al3+. Spin-Probe Rotational Diffusion and Local Polarity. The trends in probe rotation rate and polarity with solvent composition shown in Figure 1 suggest that probe rotation is influenced both by the volume of the solvent phase and by changes in the fundamental morphology of the membrane. Previously, we observed that the difference in spin-probe rotation in Li+-exchanged membranes equilibrated with pure water versus pure methanol was much larger than the difference between the rotational rates measured in the pure solvents alone.30 Thus, the membrane structure has a much stronger influence on the free volume accessible to the probe than the solvent alone. The higher uptake of methanol compared to water in the Li+-exchanged membrane also suggests a more open membrane structure with a much larger fluid-phase volume that could enhance probe rotation. However, Ca2+-exchanged membranes also exhibit a significant increase of probe rotation in methanol, even though they absorb significantly less methanol per equivalent than water. The observed behavior of rotational rate with solvent sorption in the different ion-exchanged membranes (cf. Figure 2) is qualitatively quite similar to that observed in a number of rotational diffusion studies in microporous materials, including ESR studies of spin probes in homoporous silica gels by Martini et al.58 and in zeolites by Damian et al.,59 as well as NMR studies of water in Nafion by Macmillan et al.60 and Greenbaum and co-workers.26 These studies all showed that solvent or probe rotation initially increases rapidly with water content at low hydration levels, transitioning to a region with a significantly weaker dependence at higher hydration. In combination, the correlated dependences of R and aN on solvent sorption in both water- and methanol-containing membranes show that, above a certain critical volume, the regions of the membrane occupied by solvent are large enough to exhibit some of the properties of a bulk solvent phase. Below this volume, the probe is more strongly influenced by the membrane structure and much more closely associated with the membrane polymer. Interestingly, membranes equilibrated with Al3+ cannot
J. Phys. Chem. B, Vol. 113, No. 31, 2009 10683 imbibe more than approximately 7.9 solvent molecules per sulfonate group, indicating that more extended solvent-containing regions do not form in such membranes. The behavior of rotational diffusion with solvent content has frequently been interpreted in terms of a model that explicitly accounts for the size of the pores occupied by the solvent. The studies of Martini et al.58 directly considered such effects by examining samples with well-defined pore sizes. In the membrane studies, rotational rates were measured as a function of water content, and the dependence on pore size was inferred from the linear relationship between water content and pore size that had previously been observed in the SAXS study of Gierke et al.12 McMillan et al. applied an explicit empirical relationship to characterize the dependence of solvent rotational correlation time on pore size. However, the strong similarities between the dependences of probe rotational diffusion and solvent translational diffusion on solvent content suggest that a free-volume model is most appropriate for the present application. Such a model avoids specific assumptions about the geometry of the membrane and its internal phases, and it offers a unified framework that can account for different kinds of diffusive behavior on a variety of distance scales. Free-Volume Effects. Because the parameter Bd in eq 4 depends largely on the minimum cross-sectional area required for the diffusion of the penetrant,61 it is mainly determined by the size of the penetrant molecule, which is the same for all samples equilibrated with a given solvent. Therefore, changes in the B fitting parameter in eq 5 mainly reflect changes in the free volume fraction of the polymer, Vf2. Because Vf2 is in the denominator of B, the lower B values in the Ca2+- and Al3+exchanged membranes indicate higher values of Vf2 in the presence of these ions. The modeling works of Boyd and Pant62,63 suggest a link between the free volume of a polymer and the molecular surface, such that the drop in Vf2 in the Ca2+and Al3+-exchanged membranes corresponds to an increase in the effective polymer surface area, possibly as a result of the polymer contorting to accommodate the interaction of multiple sulfonate groups with the multivalent ions. The value of B falls by 45% upon going from the monovalent ion-exchanged membrane to the multivalent ion-exchanged membranes, whereas the value of the G parameter decreases by 64%. This indicates that the rise in Vf2 is accompanied a decrease in Vf1, possibly resulting from increased clustering of the solvent as a result of electrostatic interactions with the more highly charged ions. Although the dependence of probe rotation on solvent content is well-fitted by the free-volume model in the case of water, the curves for methanol deviate qualitatively from the predictions of the free-volume model. The deviation is particularly evident for the monovalent Li+ ion (cf. Figure 2), which exhibits a pronounced sigmoidal dependence that cannot be reproduced by the simple exponential function in eq 5. This observation suggests that the underlying assumptions of the free-volume model do not apply to methanol-containing membranes. The assumption that is most likely to be violated in this case is the approximation that only two distinct phases are present in the membrane, consistent with the idea that methanol might occupy a third, interfacial phase of the types proposed by Yeager et al.13 or Mauritz and Moore.14 Membrane Morphology. In their studies of water and methanol content in Nafion membranes, Riven et al.45 suggested that the difference in the trends of the solvent translational diffusion coefficient for increasing water and increasing metha-
10684
J. Phys. Chem. B, Vol. 113, No. 31, 2009
nol is due to the alcohol’s tendency to swell the fluoroether side groups whereas water swells the ion end groups. Steep slopes in the sigmoidal shapes for ethanol and propanol were attributed to the decreased ability of those solvents to solvate the ion groups when interion group interactions are strong. Methanol, on the other hand, did not have such a steep initial slope, possibly because of preferential interactions with the ion groups at low methanol contents. Both the solvent uptake and probe rotation measurements presented here indicate that more regions of the membrane are accessible to methanol than to water, particularly in the presence of monovalent ions. This result is consistent with the model that water interacts most strongly with the ionic end groups of the polymer and swells the hydrophilic porous regions of the membrane, whereas methanol interacts with both the hydrophobic and hydrophilic regions of the membrane, so that swelling is partitioned between these regions. Based on their molecular modeling studies, Neimark and Vishnyakov4 suggested that trivalent aluminum ions do not dissociate from the sulfonate anion in water as readily as monovalent potassium ions. They also observed a decrease in polymer flexibility in the presence of Al3+ but found little difference in the effects of monovalent or divalent ions,4 consistent with our measurements of spin-probe rotation and solvent uptake in water-containing Nafion. The spin-probe measurements presented here for monovalent and divalent ions show water uptake corresponding to the order of ionic radius as observed in other measurements; however, they suggest that the stronger ionic interactions of the trivalent Al3+ ion rigidify the backbone so that it can accommodate less water. Membranes swollen with methanol exhibit a much stronger dependence on the exchange ion than water-swollen membranes. In particular, both of the multivalent ions studied were found to curtail the relatively high methanol uptake that has been observed in H+ Nafion and Li+-exchanged Nafion. Previous studies of Al3+-exchanged membranes4,64 concluded that Al3+ increases backbone stiffness and decreases the extent of ion dissociation compared with H+ Nafion. The higher rigidity of the backbone in the presence of Ca2+ and Al3+ also appears to exclude the methanol from the side-chain regions of the membrane. In summary, our results extend the validity of the Fujita freevolume model for describing the rotational diffusion of a spinprobe molecule within a polymer phase. The results support a model in which multivalent ions interact with multiple ionic groups of the polymer, introducing distortions that increase the free volume fraction in the polymer region of the membrane but slightly contract the fluid phases of the membrane. Although one might expect such interactions to affect mainly the side chains of the polymer, the results show a significant increase in the free volume of the polymer phase in the presence of higher-valence ions that suggests a more general distortion of the polymer structure, including the backbone. The rigidity in the side-chain and backbone regions promoted by these interactions appears to restrict access of the less polar methanol solvent to the hydrophobic side-chain region of the polymer. The results indicate that exchange with multivalent ions can afford useful control over the free volume fractions available to diffusants in both the polymer and fluid phases of the membrane in order to tailor its permselectivity for various applications. Conclusions The rotational mobility of spin probes in membranes parallels the behavior of translational diffusion of solvent molecules as
Lawton and Budil a function of solvent composition and content, leading to a unified explanation for the different diffusive processes in terms of the Fujita free-volume theory. The accuracy with which the rotational diffusion rate of a spin probe can be measured by ESR spectroscopy makes the method very sensitive to even small changes in free volume within a membrane. Solvent uptake is significantly altered by ion exchange in both water- and methanol-containing membranes, but measurements for the small Al3+ ion are not consistent with previous conclusions that solvent uptake is determined mainly by ionic radius. The Al3+ ion produces a rigid structure that imposes a strict upper limit on the absorption of both water and methanol. Measurements of the rotational rate and local polarity of the spin probe versus the solvent content identify a critical volume below which the probe appears to occupy a clathrate structure in which it interacts strongly with the exchange ion and above which the solvent regions of the membrane are more homogeneous and exhibit some of the properties of a bulk solvent phase. The results suggest a model in which multivalent ions interact with multiple ionic groups of the polymer, contracting the solvent phase and distorting the polymer chains so as to increase the fraction of free volume in the polymer phase. Acknowledgment. We thank Dr. Nathan Schneider for helpful discussions. This work was supported by National Science Foundation Grant CHE 0443616. Supporting Information Available: Examples of ESR spectra with corresponding nonlinear least-squares fits, and energy-dispersive X-ray (EDS) spectra. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Prakash, P.; Hoskins, D.; SenGupta, A. K. J. Membr. Sci. 2004, 237, 131–144. (2) Schneider, N. S.; Rivin, D. Polymer 2004, 45, 6309–6320. (3) Schneider, N. S.; Rivin, D. Polymer 2006, 47, 3119–3131. (4) Vishnyakov, A.; Neimark, A. V. J. Chem. Phys. 2008, 128, 164902. (5) Vishnyakov, A.; Neimark, A. V. J. Phys. Chem. B 2008, 112, 14905–14910. (6) Chou, T. J.; Tanioka, A. J. Colloid Interface Sci. 1999, 212, 576– 584. (7) Hamann, C. H.; Theile, V.; Koter, S. J. Membr. Sci. 1993, 78, 147– 153. (8) Scott, K.; Taama, W. M.; Argyropoulos, P.; Sundmacher, K. J. Power Sources 1999, 83, 204–216. (9) Piela, P.; Eickes, C.; Brosha, E.; Garzon, F.; Zelenaya, P. J. Electrochem. Soc. 2004, 151, A2053-A2059. (10) James, P. J.; Elliott, J. A.; McMaster, T. J.; Newton, J. M.; Elliott, A. M. S.; Hanna, S.; Miles, M. J. J. Mater. Sci. 2000, 35, 5111–5119. (11) Gebel, G. Polymer 2000, 41, 5829–5838. (12) Gierke, T. D.; Munn, G. E.; Wilson, F. C. J. Polym. Sci. B: Polym. Phys. 1981, 19, 1687–704. (13) Yeager, H. L.; Steck, A. J. Electrochem. Soc. 1981, 128, 1880– 1884. (14) Mauritz, K. A. M.; Robert, B. Chem. ReV. 2004, 104, 4535–4585. (15) Evans, C. E.; Noble, R. D.; Nazeri-Thompson, S.; Nazeri, B.; Koval, C. A. J. Membr. Sci. 2006, 279, 521–528. (16) Nandan, D.; Mohan, H.; Iyer, R. M. J. Membr. Sci. 1992, 71, 69– 80. (17) Villaluenga, J. P. G.; Barragan, V. M.; Seoane, B.; Ruiz-Bauza, C. Electrochim. Acta 2006, 51, 6297–6303. (18) Lee, K.; Nam, J.-D. J. Power Sources 2006, 157, 201–206. (19) Ramkumar, J.; Maiti, B.; Mathur, P. K. Sep. Sci. Technol. 1998, 33, 2423–2429. (20) Young, S. K.; Trevino, S. F.; Tan, N. C. B.; Paul, R. L. Determining Extent of Ion-Exchange in Various Counterion Nafion Membranes Using Prompt Gamma Neutron ActiVation Analysis (PGAA); Army Research Laboratory Report ARL-TR-2679; Army Research Laboratory: Aberdeen Proving Ground, MD, Mar 2002. (21) Alonso-Amigo, M. G.; Schlick, S. J. Phys. Chem. 1989, 93, 7526– 7528.
Investigation of Water and Methanol Sorption (22) Elliott, J. A.; Hanna, S.; Elliott, A. M. S.; Cooley, G. E. Macromolecules 2000, 33, 4161–4171. (23) Jeon, J.-D.; Kwak, S.-Y. J. Phys. Chem. B 2007, 111, 9437–9443. (24) Devanathan, R.; Venkatnathan, A.; Dupuis, M. J. Phys. Chem. B 2007, 111, 13006–13013. (25) Venkatnathan, A.; Devanathan, R.; Dupuis, M. J. Phys. Chem. B 2007, 111, 7234–7244. (26) Jayakody, J. R. P.; Stallworth, P. E.; Mananga, E. S.; FarringtonZapata, J.; Greenbaum, S. G. J. Phys. Chem. B 2004, 108, 4260–4262. (27) AdVanced ESR Methods in Polymer Research; Schlick, S., Ed.; John Wiley & Sons: New York, 2006. (28) Schlick, S.; Alonso-Amigo, M. G.; Eaton, S. S. J. Phys. Chem. 1989, 93, 7906–12. (29) Schlick, S.; Gebel, G.; Pineri, M.; Rius, G.; Volino, F. Colloids Surf. A: Physicochem. Eng. Aspects 1993, 78, 181–8. (30) Lawton, J.; Smotkin, E.; Budil, D. J. Phys. Chem. B 2008, 112, 8549–8557. (31) Rivera, H.; Lawton, J. S.; Budil, D. E.; Smotkin, E. S. J. Phys. Chem. B 2008, 112, 8542–8548. (32) Manley, D. S.; Williamson, D. L.; Noble, R. D.; Koval, C. A. Chem. Mater. 1996, 8, 2595–2600. (33) Zawodzinski, T. A.; Neeman, M.; Sillerud, L. O.; Gottesfeld, S. J. Phys. Chem. 1991, 95, 6040–6044. (34) Pushpa, K. K.; Nandan, D.; Iyer, R. M. J. Chem. Soc., Faraday Trans. 1988, 84, 2047–2056. (35) Onishi, L. M.; Prausnitz, J. M.; Newman, J. J. Phys. Chem. B 2007, 111, 10166–10173. (36) Schneider, D. J.; Freed, J. H. Calculating Slow-Motional Magnetic Resonance Spectra: A User’s Guide. In Biological Magnetic Resonance, Vol. 8; Berliner, L. J., Reuben, J., Eds.; Springer: New York, 1989; Chapter 1. (37) Budil, D. E.; Sale, K. L.; Khairy, K. A.; Fajer, P. G. J. Phys. Chem. A 2006, 110, 3703–13. (38) Earle, K. A.; Moscicki, J. K.; Ge, M.; Budil, D. E.; Freed, J. H. Biophys. J. 1994, 66, 1213–21. (39) Kawamura, T.; Matsunami, S.; Yonezawa, T. Bull. Chem. Soc. Jpn. 1967, 40, 1110–1115. (40) Plato, M.; Steinhoff, H.-J.; Wegener, C.; To¨rring, J. T.; Savitsky, A.; Mo¨bius, K. Mol. Phys. 2002, 100, 3711–3721. (41) Smirnova, T. I.; Chadwick, T. G.; Voinov, M. A.; Poluektov, O.; van Tol, J.; Ozarowski, A.; Schaaf, G.; Ryan, M. M.; Bankaitis, V. A. Biophys. J. 2007, 92, 3686–3695. (42) Ondar, M. A.; Grinberg, O. Y.; Dubinskii, A. A.; Lebedev, Y. S. Khim. Fiz. 1984, 3, 527–36.
J. Phys. Chem. B, Vol. 113, No. 31, 2009 10685 (43) Lebedev, Y. S. In Modern Pulsed and Continuous-WaVe Electron Spin Resonance; Kevan, L., Bowman, M. K., Eds.; John Wiley & Sons, Inc.: New York, 1990. (44) Skou, E.; Kauranen, P. S.; Hentschel, J. Solid State Ionics 1997, 97, 333–337. (45) Rivin, D.; Kendrick, C. E.; Gibson, P. W.; Schneider, N. S. Polymer 2001, 42, 623–635. (46) Ferry, J. D. Viscoelastic Properties of Polymers; 3rd ed.; John Wiley and Sons: New York, 1980. (47) Bueche, F. J. Chem. Phys. 1952, 20, 1959–64. (48) Fujita, H. Fortschr. Hochpolym. Forsch. 1961, 3, 1–47. (49) Turnbull, D. J.; Cohen, M. H. J. Chem. Phys. 1970, 31, 1164–9. (50) Goulay, A. M.; Vincent-Geisse, J. Chem. Phys. 1985, 100, 281–9. (51) McGregor, R.; Iijima, T.; Sakai, T.; Gilbert, R. D.; Hamada, K. J. Membr. Sci. 1984, 18, 129–60. (52) Fujita, H. Organic Vapors above the Glass Transition Temperature. In Diffusion in Polymers; Crank, J., Park, G. S., Eds.; Academic Press: New York, 1968; Chapter 3. (53) Kulkarni, S. S.; Stern, S. A. J. Polym. Sci. B: Polym. Phys. 1983, 21, 441–465. (54) Schneider, N. S.; Moseman, J. A.; Sung, N.-H. J. Polym. Sci. B: Polym. Phys. 1994, 32, 491–499. (55) Stern, S. A.; Sampat, S. R.; Kulkarni, S. S. J. Polym. Sci. B: Polym. Phys. 1986, 24, 2149–2166. (56) Sun, Y.-M.; Wu, T.-C.; Lee, H.-C.; Jung, G.-B.; Guiver, M. D.; Gaoc, Y.; Liu, Y.-L.; Lai, J.-Y. J. Membr. Sci. 2005, 265, 108–114. (57) Lide, D. R., Ed. CRC Handbook of Chemistry and Physics; 89th ed.; CRC Press: Boca Raton, FL, 2008. (58) Martini, G.; Ottaviani, M. F.; Romanelli, M. J. Colloid Interface Sci. 1983, 94, 105–113. (59) Damian, G.; Cozar, O.; Miclaˇus¸, V.; Paizs, C.; Znamirovschi, V.; Chis, V.; David, L. Colloids Surf. A: Physicochem. Eng. Aspects 1998, 137, 1–6. (60) MacMillan, B.; Sharp, A. R.; Armstrong, R. L. Polymer 1999, 40, 2471–2480. (61) Thimmegowda, M. C.; Sathyanarayana, P. M.; Shariff, G.; Ashalatha, M. B.; Ramani, R.; Ranganathaiah, C. J. Biomater. Sci. Polym. Ed. 2002, 13, 1295–1311. (62) Boyd, R. H.; Pant, P. V. K. Macromolecules 1991, 24, 6325. (63) Pant, P. V. K.; Boyd, R. H. Macromolecules 1992, 25, 494. (64) Vishnyakov, A.; Neimark, A. V. J. Phys. Chem. B 2008, 112, 14905–14910.
JP902750J
