Influence of Blend Miscibility on the Proton Conductivity and Methanol Permeability of Polymer Electrolyte Blends JEFFREY V. GASA,1 R. A. WEISS,1,2 MONTGOMERY T. SHAW1,2 1 Polymer Program, Institute of Materials Science, University of Connecticut, Storrs, Connecticut 06269-3136 2 Department of Chemical Engineering, Institute of Materials Science, University of Connecticut, Storrs, Connecticut 06269-3136 Received 30 November 2005; revised 4 April 2006; accepted 14 April 2006 DOI: 10.1002/polb.20865 Published online in Wiley InterScience (www.interscience.wiley.com). The influence of miscibility on the transport properties of polymer electrolyte blends composed of a proton conductor and an insulator was investigated. The protonconductive component in the blends was sulfonated poly(ether ketone ketone) (SPEKK), while the nonconductive component was either poly(ether imide) (PEI) or poly(ether sulfone) (PES). The phase behavior of PEI-SPEKK blends was strongly influenced by the sulfonation level of the SPEKK. At low sulfonation levels (ion-exchange capacity (IEC) ¼ 0.8 meq/g), the blends were miscible, while at a slightly higher level (IEC ¼ 1.1 meq/g), they were only partially miscible and for IEC 1.4 meq/g they were effectively immiscible over the entire composition range. The PES-SPEKK blends were miscible over the entire range of SPEKK IEC considered in this study (0.8–2.2 meq/g). At high IEC (2.2 meq/g) and at low mass fractions of SPEKK (<0.5), the miscible blends (PES-SPEKK) had higher proton conductivities and methanol permeabilities than the immiscible ones (PEI-SPEKK). The opposite relationship was observed for high mass fractions of SPEKK (>0.5). This behavior was explained by the differences in morphology between these two blend systems. At low IEC of SPEKK (0.8 meq/g), where both PEI-SPEKK and PESSPEKK blend systems exhibited miscibility, the transport properties were not significantly C 2006 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 44: 2253–2266, 2006 different. V Keywords: miscible blend; PEI; PEKK; PES; proton-exchange membrane ABSTRACT: INTRODUCTION Recent advances in direct methanol fuel cell technology are quite dramatic and have placed this technology on the brink of commercialization. While there are several bottlenecks in the current art, many regard the membrane, which must separate the methanol fuel and oxidant yet transport hydrogen ions, as the key polymer challenge because of the need for greater chemi- Correspondence to: M. T. Shaw (E-mail: montgomery.shaw@ uconn.edu) Journal of Polymer Science: Part B: Polymer Physics, Vol. 44, 2253–2266 (2006) C 2006 Wiley Periodicals, Inc. V cal resistance for operation with methanol as the fuel.1 These proton-exchange membranes (PEMs), which are mostly polymers that contain sulfonic acid groups, tend to have high methanol permeability, which lead to high crossover of methanol fuel from the anode to the cathode of the fuel cell. Methanol crossover is undesirable because it reduces fuel utilization efficiency and adversely affects the performance of the cell because of a mixed-potential effect.1 Methanol crossover must be low, especially for portable applications wherein the current densities are relatively low. At low current densities, the rate of methanol oxidation is low and therefore unreacted methanol concentrates at the anode, promoting diffusion of methanol through the membrane. 2253 2254 GASA, WEISS, AND SHAW Perfluorosulfonic acid (PFSA) membranes, such as Nafion1, have been historically the standard material for PEM in DMFC. However, the limitations of PFSA membranes in DMFC applications, such as high methanol permeability and poor mechanical stability under swollen conditions, and the high cost of its manufacture have motivated researchers to develop PEMs based on aromatic hydrocarbons.2,3 These types of polymers are known to have excellent mechanical and barrier properties, and therefore are usually called engineering thermoplastics. Several studies have shown that PEMs based on sulfonated aromatic hydrocarbons exhibit proton conductivities comparable to Nafion1.2,3 However, high swelling in methanol remains a problem for both types of PEMs. Blending of a hydrophobic (nonsulfonated) polymer with an acidic polymer has become a widely used approach for the design of PEMs with improved resistance to methanol. A variety of polymer pairs have been considered, in particular, blends of Nafion1 with fluoropolymers4–7 and hydrocarbons.8,9 More relevant to this work are binary blends of sulfonated poly(ether ether ketone) (SPEEK) with poly(ether imide) (PEI),10–16 poly (aryl ether sulfone) (PES),17–19 and poly(2,6-dimethyl phenylene oxide).10 Obviously, incorporation of a nonconductive material in a conductor may cause a reduction in the proton conductivity of the membrane. However, Mikhailenko et al. reported that it is possible to improve the conductivity of SPEEK by blending it with relatively small amounts of PEI, a nonconductive engineering thermoplastic.12 They found that blending with PEI first results in an increase and then a decrease in proton conductivity at PEI concentrations greater than 5%. PEEK and PEI are known to be miscible,20,21 but sulfonated PEEK and PEI are immiscible, according to SEM studies done by Mikhailenko et al.12 However, sev- eral others have reported that sulfonated PEEK and PEI are miscible, based on thermal analysis.14–16,22 The objective of our study was to investigate the influence of blend miscibility on the transport properties of polymer electrolyte membranes composed of an acidic polymer and a nonsulfonated polymer. The acid polymer used in this study was sulfonated poly(ether ketone ketone) (SPEKK). SPEKK is an ionomer similar to SPEEK, but with higher transition temperatures and thermo-oxidative stability.11,16,23 The preparation of SPEKK and its potential use as a PEM in fuel cells were described previously.23–26 Poly(ether ketone ketone) (PEKK) itself is a relatively new engineering thermoplastic that has high temperature stability, excellent chemical and solvent resistance, and excellent mechanical properties. PEKK is actually a family of copolymers with different ratios of terephthaloyl (T) and isophthaloyl (I) moieties (Fig. 1). In this study, the polymer of principal focus was the acid form of SPEKK with a T/I ratio of 6/4. PEKK with this T/I ratio is miscible with PEI.20 However, SPEKK with relatively high levels of sulfonation is immiscible with PEI.25 Therefore, we expect different levels of miscibility when different sulfonation levels of SPEKK are blended with PEI. Another blend system that we have investigated in this work comprised SPEKK and PES, which is also an engineering thermoplastic. Swier et al. reported that SPEKK/PES blends are miscible.26 The purpose of this work was to compare the transport properties (proton conductivity and methanol permeability) of immiscible blends (SPEKK/PEI) with that of miscible ones (SPEKK/ PES). Knowledge of the conductivity–miscibility relationships should be useful for designing ionomer blend systems for direct methanol fuel cell applications. Figure 1. Chemical structure of PEKK. Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb INFLUENCE OF BLEND MISCIBILITY AND CONDUCTIVITY EXPERIMENTAL Similarly, Sulfonation of PEKK Weq ¼ The PEKK used was OXPEKK-SP (Oxford Performance Materials, New Britain, CT), which has a terephthaloyl (T) to isophthaloyl (I) ratio of 6:4 (Fig. 1). The effect of increasing the isophthaloyl content is to reduce crystallinity and lower the Tg. Sulfonation was achieved by dissolving the polymer (5% w/v) in a mixture of 53/47 (v/v) concentrated sulfuric acid (96.3%) and fuming sulfuric acid (30% free SO3).23,24 The sulfonation time and temperature were varied to achieve different levels of sulfonation.24 The resulting sulfonated polymer (SPEKK) was precipitated by drop-wise addition of the solution (total volume of solution was about 200 mL) into 1.5 L of rapidly stirred de-ionized water. The SPEKK was filtered and washed repeatedly with de-ionized water to remove excess acid. After washing, it was soaked in copious amounts (5 g of polymer to 2 L of water) of de-ionized water for 24 h to remove residual acid. The washing process was followed by measuring the pH of the soaking solution; washing continued until the reading was neutral. The washed SPEKK was air dried in a fume hood overnight at room temperature and then under vacuum at 120 8C for 3 days. The sulfonation level was determined from the carbon/sulfur ratio obtained from elemental analysis and/or by titration of the sulfonic acid groups. Previous results showed good agreement (discrepancies were no greater than 0.04 meq/g) between the two methods.24 For the titration method, SPEKK was ion-exchanged with excess saturated aqueous sodium chloride solution overnight. The HCl product was then titrated with a normalized sodium hydroxide solution using phenolphthalein as an acid–base indicator. The sulfonation level may be expressed as: (i) a degree of sulfonation (Xs) defined as the average number of sulfonate groups per repeat unit; (ii) equivalent weight (Weq), which is the mass of polymer per sulfonate group; or (iii) as a concentration of sulfonate groups, expressed as ion-exchange capacity (IEC, equivalents per mass). For SPEKK, the relationship between degree of sulfonation (Xs) and IEC in terms of milliequivalents of SO3H per gram of SPEKK (meq/g) is: 300 ðIECÞ Xs ¼ 1000 81 ðIECÞ 2255 ð1Þ Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb 1000 300 81Xs ¼ IEC Xs ð2Þ The numbers 300, 81, and 1000 refer to the molecular weight of the PEKK repeat unit, molecular weight of sulfonic acid, and conversion factor, respectively. The units of Weq are grams of polymer per equivalent of sulfonate groups. The maximum IEC of SPEKK is 4.35 meq/g (i.e., two SO3H groups per repeat unit). Either Weq or IEC are most useful for comparing blends that have complex structures. Blend Preparation Blends of SPEKK with either PEI or poly(ether suflone) (PES) were prepared by codissolving the two polymers (at various mass ratios) in N-methyl pyrrolidone (NMP) to form 5 wt % polymer solutions. These solutions were cast onto clean glass plates at 60 8C and air dried in a fume hood until most of the solvent had evaporated. After drying, the membranes were soaked in de-ionized water at 20 8C for 48 h to remove residual solvent. Blends of unmodified PEKK with PEI and PES proved difficult to make because of the crystallinity of PEKK reduced its solubility in reasonable solvents at low temperatures. Impedance Spectroscopy Impedance spectroscopy has been widely used to measure the proton conductivity of PEMs.11,27–35 The conductivity of the membranes was measured using a Hewlett-Packard Agilent1 4284A LCR meter covering a frequency range of 20–106 Hz. The applied voltage was 50 mV. Membrane conductivities were measured using a cell described by Zawodzinski et al.27 This cell measures the conductivity along the plane of the membrane, which generally is easier than through-the-plane measurements because the membrane resistance can be made arbitrarily high by spacing the electrodes far apart. If the membrane is isotropic, as ours appeared to be, the results should be independent of the current direction. Measurements were made at 98% relative humidity and room temperature (23 8C). Water Sorption The water sorption of the membranes was measured by placing films in a controlled humidity 2256 GASA, WEISS, AND SHAW environment and weighing them every 24 h until the mass of the membranes reached equilibrium, as judged by the point where the changes in mass were merely random weighing errors. The membranes were then dried in vacuum at 100 8C for 24 h and re-weighed. Water sorption is defined here as the mass of the absorbed water divided by the mass of the wet membrane, i.e., the mass fraction of water. Microscopy The morphology of the composite membranes was studied using a Nikon Labophot optical microscope in transmission mode and an environmental scanning electron microscope (ESEM; Philips ESEM 2020). Refractive index contrast was very high, eliminating the need for staining or interference optics. For ESEM, the membranes were cryo-fractured in liquid nitrogen and images were made of the section thickness. Methanol Permeability The methanol permeabilities of the membranes were measured using the cell shown in Figure 2. The design of this cell is a variation of the device used by Walker et al. to measure methanol permeation through membranes based on sulfonated aromatic hydrocarbons and also PFSA membranes, such as Nafion1 117.36 The lower part of the cell, which was a 20-mL glass vial, was filled with 10 mL methanol. Methanol vapor in equilibrium with the liquid diffused along the concentration gradient through the membrane, which was clamped between the mouth of Figure 2. Sectional view of the permeation-measuring cell. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] the vial (2 cm in diameter) and the cap. The cap had a 1-cm hole so that the methanol that diffused through the membrane could escape. The cell was placed inside a drying oven, which provided temperature control (30 8C) and an air draft that maintained the methanol concentration above the surface of the membrane at a minimum. The humidity was not controlled, but was measured and found to be relatively constant at (14 6 2)%. The mass of the methanol inside the cell was measured as a function of time to find the molar flux. For the results reported here, the methanol permeability (P) was calculated by applying Fick’s first law: J ¼ D dCm Cb Cb ¼ DK ¼ P dx L L ð3Þ where J is the molar flux of methanol, D is the methanol diffusivity, K is the partition coefficient or the solubility of methanol in the membrane, Cm is the molar concentration of methanol in the membrane, Cb is the methanol concentration in the gas phase, and L is the thickness of the membrane. Use of the driving force based on concentrations or partial pressures in the gas phase lead to a host of units for permeability. It is often convenient to refer to the driving force in terms of a molar concentration, that is, mol/m3, which yields units of m2/s for permeability. Assuming equilibrium and ideal-gas mixtures, the molar concentration of methanol in the vapor phase will not vary with the total pressure, but will be given by the simple expression aiP0i /RT, where ai is the activity in the condensed phase and P0i is the pure-component vapor pressure at temperature T. For condensable vapors, there is the obvious issue of liquid versus vapor molar concentration; with pure methanol, the liquid Cb is about 0.025 mol/cm3 (25 kg mol/ m3), while for the equilibrium vapor at room temperature Cb is about 6.7 106 mol/cm3 (0.0067 kg mol/m3). In this simple device (Fig. 2), which superficially resembles the setup in ASTM E 96-95, it is difficult to quantify accurately the value of DCb. Although Cb adjacent to the inner surface of the membrane should be close to the equilibrium concentration (e.g., mol/m3) of MeOH, it was difficult to estimate Cb adjacent to the outer surface of the membrane because of the unknown factors controlling the mass transfer away from the film. To circumvent this problem, a reference sample with known methanol permeability (here, Nafion 1121) was also tested and the methanol permeabilities of Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb INFLUENCE OF BLEND MISCIBILITY AND CONDUCTIVITY 2257 the samples were reported relative to the methanol permeability of the reference sample. In this way, the driving force term (DC) in eq 3 cancels out because all the samples were measured in one batch, and therefore were exposed to the same conditions (same driving force). Thus Jsample Lsample Psample ¼ Jreference Lreference Preference ð4Þ At high SPEKK, the PEI-SPEKK membranes swelled and distorted excessively; thus only membranes with SPEKK concentrations below 40% were included in the comparison of the two blend systems. Fortunately, it was in this range that the major differences due to miscibility were found. Figure 4. Optical micrographs of PEI-SPEKK blends wherein the IEC of SPEKK is 0.8 meq/g for both blends and the SPEKK weight fractions are (a) 0.7 and (b) 0.5. The images were taken at room conditions (23 8C, 48% relative humidity). RESULTS AND DISCUSSION Phase Behavior of PEI-SPEKK Blends Figure 3. Optical micrographs of the neat SPEKK membrane with an IEC of 0.8 meq/g (a). Under a cross polarizer, a micrograph of the same membrane is shown in (b). The images were taken at room conditions (23 8C, 48% relative humidity). Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb It has been reported in the literature that PEKK and PEI are miscible.20 However, we found that upon sulfonation of PEKK, these two polymers become immiscible at a certain level of sulfonation. The miscibility behavior of PEI-SPEKK blends was studied by microscopic techniques but the phase behavior could not be supported by thermal analysis because the glass transitions of the two components were too close to each other (Tg of PEI ¼ 215 8C, Tg of SPEKK ¼ 180–200 8C for IEC ¼ 0.8–2.2 meq/g, respectively).23,25 An optical micrograph of the neat SPEKK membrane with an IEC of 0.8 meq/g is shown in Figure 3(a). The pure SPEKK membrane exhibited birefringence under cross-polarized light as 2258 GASA, WEISS, AND SHAW Figure 5. Optical micrographs of PEI-SPEKK blends wherein the IEC of SPEKK is 1.1 meq/g for all the blends and the SPEKK weight fractions are (a) 0.3, (b) 0.5, (c) 0.7, and (d) 0.9. The images were taken at room conditions (23 8C, 48% relative humidity). shown in Figure 3(b), which is an indication of crystallinity. It was previously reported that wide-angle X-ray scattering of SPEKK at this sulfonation level (0.8 meq/g) showed diffraction peaks due to crystallinity.23 Almost certainly, the rough texture in the micrograph of the pure SPEKK membrane was also due to crystallinity. The blends (Fig. 4) showed no evidence of phase separation and the morphology was more homogeneous (smoother and more transparent) than the pure SPEKK. These are indications of miscibility. No phase separation was observed over the entire composition range at an SPEEK sulfonation level of 0.8 meq/g. At a slightly higher IEC for SPEKK (1.1 meq/g), evidence of a miscibility gap appeared (Fig. 5). At SPEKK mass fractions less than 0.3, the blends appeared to be miscible (no droplets formed). At a 0.3 mass fraction of SPEKK, small droplets of SPEKK appeared, which was confirmed by sulfur microanalysis of the dispersed phase using energy dispersive X-ray spectroscopy (EDS). Between 30 and 70% SPEKK, the blends were phase-separated, with an indication of a spinodal decomposition process at 0.5 mass fraction. At and above 70% SPEKK, the blends were miscible again. At an IEC of SPEKK 1.5 meq/g, the blends were phase-separated (see Fig. 6) over the entire composition range considered in this study (SPEKK mass fractions from 0.1 to 0.9). ESEM of the cryo-fractured sections of the blends (Fig. 7) confirmed that the droplets were indeed solid polymeric phase and not artifacts such as water droplets or air bubbles. The findings from the optical microscopy of the PEI-SPEKK blends at room temperature (23 8C) are summarized in an isothermal phase diagram shown in Figure 8. There seems to be a lower critical IEC value. Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb INFLUENCE OF BLEND MISCIBILITY AND CONDUCTIVITY 2259 et al. examined the PES blend, which because of its miscibility might be most sensitive to water, using AFM and TEM, and could find no evidence of nano-phase separation.26 One hypothesis is that the water-insensitive PES retards the kinetics of phase separation substantially. In this case, rapid phase separation might occur at higher temperatures. Figure 6. Optical micrographs of PEI-SPEKK blends wherein the IEC of SPEKK is 2.2 meq/g for both blends and the SPEKK weight fractions are (a) 0.4 and (b) 0.5. The images were taken at room conditions (23 8C, 48% relative humidity). Phase Behavior of PES-SPEKK Blends In contrast to the phase behavior of the PEISPEKK blends, the PES-SPEKK blends were miscible over the entire range of IEC values and mass fractions of the SPEKK component considered in this work (0.8–2.2 meq/g). Figure 9(a) shows an optical micrograph of a 50/50 PES-SPEKK blend wherein the IEC of SPEKK was 0.8 meq/g and one wherein the IEC of SPEKK was 2.2 meq/g [Fig. 9(b)]. Neither blend showed any evidence of phase separation. Other compositions and IEC values between 0.8 and 2.2 meq/g exhibited similar miscibility behavior. For any of these blends, as well as other blends of this nature, a concern is the influence of water absorption on the phase behavior. Swier Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb Figure 7. Electron micrographs (ESEM) of a neat SPEKK membrane with an IEC of 2.0 meq/g (a) and a blend with 20% PEI (b). The images were taken at room conditions (23 8C, 48% relative humidity). 2260 GASA, WEISS, AND SHAW Figure 8. Isothermal phase diagram of PEI-SPEKK showing miscible and immiscible blends at various IEC versus concentration. The temperature was 20 8C. in either the dispersed phase or the matrix. Because the discontinuity in the conductivity of the blends was observed at SPEKK weight fractions between 0.4 and 0.5, the percolation threshold is expected to be close to these values, which was quite evident from the morphology of these blends. To explore more quantitatively the effect of mixing on the conductivity of these blends, a theoretical model and a scaling equation were compared with the data. The theoretical model used was based on an improved effective-medium theory (EMT) developed by Nakamura.37 The limitation of the simple EMT is that it can only be applied to composites with small volume fractions of the dispersed phase (0.1 or less), where local Proton Conductivity The proton conductivity of the PEI-SPEKK blends as a function of the concentration of the ionic conductor (SPEKK) is shown in Figure 10. At 15% SPEKK, the conductivity was about four orders of magnitude lower than for pure SPEKK. As the SPEKK content increased, the conductivity increased monotonically with decreasing slope in a semilog scale [Fig. 10(a)]. A step increase in conductivity was observed between SPEKK mass fractions of 0.4–0.5, which can be explained by comparing the morphology of these two blends. At a SPEKK mass fraction of 0.4, the SPEKK droplets in this two-phase blend are separated from each other [Fig. 6(a)]. Surrounding these ion-conducting droplets is a relatively insulating matrix (PEI) that hinders the migration of protons from one droplet (conductor) to the other (the solubility of SPEKK at this IEC is small, but finite, so the conductivity is not expected to be 0). At a SPEKK mass fraction of 0.5, interconnectivity of some SPEKK droplets was observed [Fig. 6(b)], which is a reasonable explanation for its higher conductivity relative to the blend containing 40 wt % SPEKK. At SPEKK mass fractions greater than 0.5, the conductivity increased monotonically with increasing SPEKK concentration. It is in this concentration range also that the morphology changed from an interconnected but bi-continuous structure to one featuring fully dispersed droplets of PEI in a continuous SPEKK phase. Sulfur microanalysis using EDS showed that the blends were highly phase-separated over the entire composition range; only one component (either pure SPEKK or pure PEI) could be detected Figure 9. Optical micrographs of 50/50 PES-SPEKK blends wherein the IEC of SPEKK are (a) 0.8 meq/g and (b) 2.2 meq/g. The images were taken at room conditions (23 8C, 48% relative humidity). Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb INFLUENCE OF BLEND MISCIBILITY AND CONDUCTIVITY 2261 when compared with the other component, which is true for the SPEKK/PEI system. In such a case, the model simplifies to r ¼ r1 Figure 10. Proton conductivity (23 8C and 98% relative humidity) of PEI-SPEKK blends wherein the IEC of SPEKK is 2.2 meq/g. (a) The solid line is based on the improved EMT developed by Nakamura,37 eq 5. (b) Solid line comprises two scaling laws connected by an exponential sigmoid. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.] field effects are unimportant.38 In this work, local field effects could not be ignored because the blend compositions covered almost the entire range (0.15–0.9). Therefore, the improved EMT by Nakamura was more suitable to this work because this theory extends the validity of the EMT to the entire composition range (0–1). According to Nakamura’s theory, a two-phase composite can be considered as a random mixture of both series and parallel connectivity patterns of the two phases. The equation he developed is quite cumbersome but it becomes relatively simple if the conductivity of one of the phases is negligible Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb VðV Vc Þ 1 Vc V > Vc ð5Þ where V is the volume fraction, r1 is the conductivity, and Vc is the percolation volume fraction threshold of the conductive phase (SPEKK). The density of SPEKK was assumed to be close to that of PEI such that the SPEKK mass fraction on the abscissa in Figure 10(a) can be considered equivalent to the volume fraction of SPEKK. The specific gravity of PEI (Ultem 1000TM) is about 1.27, whereas PEKK is about 1.31. It was reported by Zaidi et al. that sulfonation reduced the density of PEEK because of reduction in crystallinity.13 A similar reduction in density of PEKK upon sulfonation was expected; thus, the near equivalence of the densities is probably a very reasonable assumption. We tried to estimate the specific gravity of SPEKK (IEC ¼ 2.2 meq/g) by measuring the dimensions of the membrane using a micrometer. The mass of the membrane was measured by using an analytical balance. The approximate density was 1.24 6 0.4 g/cm3. On the basis of the modeling results, the percolation volume fraction threshold (Vc) and the conductivity of the SPEKK-rich phase (r1) were about 0.33 6 0.12 and 0.16 6 0.02 S/cm, respectively. The regressed Vc (0.33) is higher than the theoretical value of 0.16 for an irregular array of the conductive particles.39 According to a description provided by Lux, the viscosity of the polymer blend has an inhibiting influence on the equilibrium phase separation process such that Vc can be higher than the ‘‘true’’ equilibrium value.39 This is especially true for the SPEKK-PEI blend where phase separation in the ternary (NMPSPEKK-PEI) solution only occurred when the total polymer concentration reached about 20 wt % (starting from 5 wt %) during solvent casting. At 20 wt % polymer concentration, the ternary solution was already quite viscous. Equation 5 does not apply at low SPEKK weight fractions (<0.4). Theoretically, the conductivity of the composite should be 0 when the volume fraction of the conductive phase is lower than the percolation threshold. However, thermodynamics dictates that the miscibility of SPEKK in the nonconductive matrix (PEI) will be finite, which could cause leakage of ions from one conductive droplet to another. If the conductivity data are extrapolated to 100% PEI content, the assumption that 2262 GASA, WEISS, AND SHAW the conductivity of one phase (PEI) is negligible when compared with the other phase (SPEKK) is still valid. The missing impedance data at V ¼ 0, 0.05, and 0.1 were obtained, but the conductivities could not be calculated because the impedance plots for the three composites exhibited capacitive behavior (phase angle ¼ 908; no ionic conduction) over the entire frequency range of 1 Hz to 1MHz. Another possible reason for the finite conductivity at low SPEKK levels is the seemingly bimodal size distribution of the droplets [Fig. 7(b)]. Swier et al. reported that in SPEKK-PEI blends, the droplet size increased with increasing weight fraction of SPEKK.25 At low weight fractions of SPEKK, the small droplets might have percolated at around 0.17 volume fraction. As the volume fraction of SPEKK increases, larger droplets appear, but the smaller droplets are still present [Fig. 7(b)]. At a total SPEKK volume fraction of around 0.4 (sum of smaller and larger droplets), the larger droplets conceivably reached their percolation threshold, which caused the sudden increase in conductivity. A two-stage scaling equation is considered in Figure 10(b). This log–log plot shows very clearly the transition between two conductivity regions: one below and one above the percolation threshold of a SPEKK-rich phase. Above the percolation limit of about x ¼ 0.45, the scaling exponent is about 3, while below the percolation limit, the slope is significantly higher (P < 0.001) at about 4. In the latter region for the blend, which has low concentrations of SPEKK, the proton concentration is increasing in a linear fashion in the single phase, while at concentrations above the solubility limit of SPEKK, highly conductive but discrete regions are increasing in number and sized within a fixed-conductivity matrix. This combination of changes could be responsible for the apparently higher power found at low concentrations. It should be noted that the Nakamura equation (eq 5) has no explicit scaling exponent, but tends to be quadratic at high concentrations. It is developed assuming a uniform droplet size. Figure 11(a) compares the proton conductivities of PES-SPEKK blends (miscible) with those of PEI-SPEKK blends (immiscible) wherein the IEC of SPEKK was 2.2 meq/g for both blend systems. There is an apparent crossover of the two conductivity curves. At low weight fractions of SPEKK (<0.5), the conductivities of the miscible blends were higher than those of the immiscible ones. However, this relationship reverses at higher Figure 11. (a) Proton conductivity (23 8C and 98% relative humidity) of PEI-SPEKK and PES-SPEKK blends wherein the IEC of SPEKK is 2.2 meq/g; and (b) log-log plot of PES-SPEKK conductivity results. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] weight fractions of SPEKK; the immiscible blends became more conductive than the miscible ones. (Chance probability of two runs comprising four positive and six negative numbers is 1%). This crossover in conductivity curves can be explained by the morphology of these blends. At low weight fractions of SPEKK, the sulfonic acid groups in the immiscible blends (droplet-wise morphology) are contained within the droplets and percolation of the acid groups is hindered by the insulating matrix (PEI). In the miscible blend, the sulfonic acid groups are dispersed homogeneously throughout the membrane and percolation can still be achieved, although the presence of the nonconductJournal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb INFLUENCE OF BLEND MISCIBILITY AND CONDUCTIVITY ing component (PES) dilutes the concentration of protons and reduces the conductivity. In contrast, for the immiscible blend at high mass fractions of SPEKK (>0.5), the conductive SPEKK matrix can swell and conduct protons almost independently from the nonconductive PEI droplets because the concentration of dissolved PEI in the SPEKK is relatively low. The clustering of sulfonic acid groups within the conductive phase is somewhat similar to the pure conductive component (SPEKK). The effect of the insulating droplets is to reduce the overall concentration of charge carriers in the blend and increase the tortuosity of the proton pathways, and thereby reduce the conductivity. In the miscible blends, the PES component can affect the clustering of the sulfonic acid groups in SPEKK because PES chains are present locally in between SPEKK chains and can hinder both intrachain and interchain clustering of sulfonic acid groups. Figure 11(b) depicts the conductivity in the same fashion for PES-SPEKK blends as is done in Figure 10(b) for the PEI-SPEKK mixtures. Assuming a single scaling law for the entire composition range for this miscible blend gives a power of 4.2 6 0.6, which is indistinguishable from that (3.9 6 0.3) for the low-concentration region of the PEISPEKK blends, suggesting similar effects. The sinuous behavior of the data in Figure 10(b) is statistically significant according to the test for runs of residuals; e.g., the probability of finding three runs with 10 points is only about 4%. However, there is no obvious reason why the conductivity should behave in the fashion shown if the system is indeed miscible. This is discussed in more detail later. The crossover in the conductivity curves can also be associated with the water absorption of these blends [Fig. 12(a)]. At high SPEKK mass fractions (>0.5), the water uptakes of the immiscible blends were higher than those of the miscible ones, but this relation flips at low SPEKK mass fractions (<0.5). Figure 12(b) shows the hydration number (k) or the number of water per sulfonic acid groups in the blends. At high SPEKK mass fractions (>0.5), the hydration number is almost constant for both blend systems and the immiscible blends have higher hydration number than the miscible ones. From SPEKK mass fraction of 0.5–0.4, the hydration number of the immiscible blends drops from about 9 to 7.5 water molecules per sulfonic acid, whereas the miscible blends remained almost constant at about 8. The abrupt increase in hydration number at low SPEKK concentrations is Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb 2263 Figure 12. Equilibrium water sorption (a) and hydration number (b) of PEI-SPEKK and PES-SPEKK blends wherein the IEC of SPEKK is 2.2 meq/g. Membranes were equilibrated at 98% relative humidity and room temperature (23 8C). [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.] trivially due to the contribution of the water uptake of the matrix that becomes prominent at low SPEKK weight fractions. As discussed earlier, both blend systems (PEISPEKK and PES-SPEKK) were miscible when the IEC of SPEKK was low enough, i.e., 0.8 meq/g. Figure 13 shows the proton conductivity of the 2264 GASA, WEISS, AND SHAW Figure 13. Proton conductivity (23 8C and 98% relative humidity) of PEI-SPEKK and PES-SPEKK blends wherein the IEC of SPEKK is 0.8 meq/g. two blend systems wherein the IEC of SPEKK was 0.8 meq/g. There are no apparent differences in the conductivities between the two blend systems, which mean that the influence of PES and PEI on the conductivity of these miscible blends was similar. This also supports the claim that the crossover in the conductivity curves of the two blend systems at high IEC of SPEKK was due to morphology differences and not due to inherent differences between the two secondary components (PEI and PES). Attempts to reach a more global description of the conductivity in terms of the Equivalent Box Model,40 or similar parallel-series models, were frustrated by the large number of parameters in these model, even when simplified, relative to the number of conductivity observations. Methanol Permeability Figure 14 is a plot of the mass of methanol that diffused through the blend membranes as a function of time. The molar flux (J) is obtained by dividing the slope of this plot converted to g/s by the molecular weight of methanol (32 g/mol) and Figure 14. Steady-state mass flow of methanol through (a) PEI-SPEKK and (b) PES-SPEKK blends. Measurements were taken at 30 8C. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.] Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb INFLUENCE OF BLEND MISCIBILITY AND CONDUCTIVITY the area normal to the flux (cm2). The diameter of the hole in the cap was used to calculate the area. The thickness-normalized molar flux (J multiplied by L) for Nafion 1151 was 1.08 108 mol/ cm s, which is close to what Walker et al. reported (1.39 108 mol/cm s) using pure methanol on one side of the membrane while sweeping the other side with nitrogen gas.36 A possible reason for their higher value is that they did the measurements at a higher temperature (40 8C). In their paper, they did not report the calculated value for the methanol permeability. By turning the vial upside down, we measured with Nafion 1151 a permeability of 4 106 cm2/s based on the molar concentration in the pure liquid, which compares with 1.7 106 measured by Einsla et al. using Nafion 1171.41 Figure 15 shows the methanol permeability of the blend membranes relative to that of Nafion1. As expected, the methanol permeability increased with increasing concentration of SPEKK for both blend systems. The IEC of SPEKK in these blends was 2.2 meq/g. At low mass fractions of SPEKK (<0.5), the immiscible PEI-SPEKK blends exhibited lower methanol permeabilities than the miscible PES-SPEKK blends. This behavior is again Figure 15. Methanol permeability of PEI-SPEKK and PES-SPEKK blends relative to Nafion 1151. Measurements were taken at 30 8C. [Color figure can be viewed in the online issue, which is available at www. interscience.wiley.com.] Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb 2265 due to the respective morphologies of these blends and the explanation is analogous to that of the conductivity behavior. In the immiscible blends, the PEI matrix impedes the permeation of methanol from one SPEKK droplet to another, whereas in the miscible blends, wherein the SPEKK chains are dispersed throughout the blend, there is a continuous and uniform path for methanol transport. The methanol permeabilities of the insulating components (PEI and PES) were both on the low end and are almost the same in magnitude. Therefore, the observed difference between the permeabilities of the immiscible blends and those of the miscible blends were not due to an intrinsic difference between the permeabilities of the two nonconducting components. CONCLUSIONS The influence of blend miscibility on the transport properties of polymer electrolyte blends was investigated. Miscible blends (PES-SPEKK) can have remarkably different proton conductivities and methanol permeabilities when compared with immiscible blends that have the same overall IEC. These differences in the transport properties can be explained by the differences in the morphology between these two blend systems. Blends of PEI and SPEKK exhibited different miscibility behavior depending on the sulfonation level of the SPEKK component. Miscible blends were obtained at IEC values of SPEKK of up to 0.8 meq/g. Above 0.8 meq/g, PEI-SPEKK blends were immiscible and formed dispersed droplet morphologies. On the other hand, blends of SPEKK and PES were miscible over the entire investigated range of blend compositions and of SPEKK IEC values of 0.8–2.2 meq/g. At high IEC of the SPEKK component (2.2 meq/g), the proton conductivities and methanol permeabilities of the miscible blends (PES-SPEKK) were significantly higher than those of the immiscible blends (PEI-SPEKK) at low SPEKK mass fractions. At high mass fractions, the relation was reversed. At low IEC of SPEKK (0.8 meq/g), where both PEI-SPEKK and PES-SPEKK blend systems exhibited miscibility, the transport properties were not significantly different. This indicates that the contributions of PES and PEI to the transport properties of the blends are the same. This also supports the claim that the crossover in the conductivity versus blend composition curves of the two blend systems at high IEC of SPEKK is due to 2266 GASA, WEISS, AND SHAW morphology differences and not due to inherent differences in the transport properties between the two nonconductive components (PEI and PES). The authors gratefully acknowledge financial support from the Army through the Connecticut Global Fuel Cell Center and the Department of Energy (Grant 10761-001-05). We also thank S. Boggs for use of the impedance spectroscopy equipment. REFERENCES AND NOTES 1. Wasmus, S.; Kuver, A. J Electroanal Chem 1999, 461, 14. 2. Savadogo, O. J New Mater Electrochem Syst 1998, 1, 47. 3. Rikukawa, M.; Sanui, K. Prog Polym Sci 2000, 25, 1463. 4. Yu, H.; Yi, B. Fuel Cell 2004, 4, 96. 5. Kim, H. J.; Kim, H. J.; Shul, Y. G.; Han, H. S. J Power Sources 2004, 135, 66. 6. Song, M.; Kim, Y. T.; Fenton, J. M.; Kunz, H. R.; Rhee, H. J Power Sources 2003, 117, 14. 7. Cho, K.; Jung, H.; Choi, N.; Sung, S.; Park, J.; Choi, J; Sung, Y. Solid State Ionics 2005, 176, 3027. 8. Sauk, J.; Byun, J.; Kim, H. J Power Sources 2005, 143, 136. 9. Park, H.; Kim, Y.; Hong, W.; Choi, Y.; Lee, H. Macromolecules 2005, 38, 2289. 10. Jorissen, L.; Gogel, V.; Kerres, J.; Garche, J. J Power Sources 2002, 105, 267. 11. Kerres, J.; Ullrich, A.; Meier, F.; Haring, T. Solid State Ionics 1999, 125, 243. 12. Mikhailenko, S. D.; Zaidi, S. M. J.; Kaliaguine, S. J Polym Sci Part B: Polym Phys 2000, 38, 1386. 13. Zaidi, S. M. J.; Mikhailenko, S. D.; Robertson, G. P.; Guiver, M. D.; Kaliaguine, S. J Membr Sci 2000, 173, 17. 14. Cui, W.; Kerres, J.; Eigenberger, G. Sep Purif Technol 1998, 14, 145. 15. Karcha, R. J.; Porter, R. S. J Polym Sci Part B: Polym Phys 1993, 31, 821. 16. Karcha, R. J.; Porter, R. S. J Polym Sci Part B: Polym Phys 1989, 27, 2153. 17. Kreuer, K. D. J Membr Sci 2001, 185, 29. 18. Manea, C.; Mulder, M. Desalination 2002, 147, 179. 19. Manea, C.; Mulder, M. J Membr Sci 2002, 206, 443. 20. Sauer, B. B.; Hsiao, B. S.; Faron, K. L. Polymer 1996, 37, 445. 21. Tseng, Y. C.; Woo, E. M. Macromol Rapid Commun 1998, 19, 215. 22. Chun, Y. S.; Kwon, H. S.; Kim, W. N.; Yoon, H. G. J Appl Polym Sci 2000, 78, 2488. 23. Swier, S.; Chun, Y. S.; Gasa, J.; Shaw, M. T.; Weiss, R. A. Polym Eng Sci 2005, 45, 1081. 24. Swier, S.; Gasa, J.; Shaw, M. T.; Weiss, R. A. Ind Eng Chem Res 2004, 43, 6948. 25. Swier, S.; Shaw, M. T.; Weiss, R. A. J Membr Sci 2006, 270, 22. 26. Swier, S.; Ramani, V.; Fenton, J. M.; Kunz, R. H.; Shaw, M. T.; Weiss, R. A. J Membr Sci 2005, 256, 122. 27. Zawodzinski, T. A.; Neeman, M., Jr.; Sillerud, L. O.; Gottesfeld, S. J Phys Chem 1991, 95, 6040. 28. Beattie, P. D.; Orfino, F. P.; Basura, V. I.; Zychowska, K.; Ding, J.; Chuy, J.; Schmeisser, J.; Holdcroft, S. J Electroanal Chem 2001, 503, 45. 29. Wan, Y.; Creber, K. A. M.; Peppley, B.; Bui, V. T. Polymer 2003, 44, 1057. 30. Nunes, S. P.; Ruffman, B.; Rikowski, E.; Vetter, S.; Richau, K. J Membr Sci 2002, 203, 215. 31. Chilcott, T. C.; Chan, M.; Gaedt, L.; Nantawisarakul, T.; Fane, A. G.; Coster, H. G. L. J Membr Sci 2002, 195, 153. 32. Lehmani, A.; Turq, P.; Perie, M.; Perie, J.; Simonin, J. J Electroanal Chem 1997, 428, 81. 33. Cappadonia, M.; Wilhelm, E.; Saberi, J.; Niaki, S. M.; Stimming, U. Solid State Ionics 1995, 77, 65. 34. Zaluski, C.; Xu, G. J Phys Chem Solids 1994, 55, 1507. 35. Cahan, B. D.; Wainright, J. S. J Electroanal Chem 1993, 140, L185. 36. Walker, M.; Baumgartner, K. M.; Feichtinger, J.; Kaiser, M.; Rauchle, E.; Kerres, J. Surf Coat Technol 1999, 116, 996. 37. Nakamura, M. Phys Rev B 1984, 29, 3691. 38. Nan, C. W.; Smith, D. Mater Sci Eng B 1991, 10, 99. 39. Lux, F. J Mater Sci 1993, 28, 285. 40. Kolařı́k, J.; Kruliš, Z.; Šlouf, M.; Fambri, L. Polym Eng Sci 2006, 46, 817. 41. Einsla, B. R.; Kim, Y.-S.; Hickner, M. A.; Hong, Y.-T.; Hill, M. L.; Pivovar, B. S.; McGrath, J. E. J Membr Sci 2005, 255, 141. Journal of Polymer Science: Part B: Polymer Physics DOI 10.1002/polb
© Copyright 2026 Paperzz