Journal of Membrane Science 256 (2005) 122–133 Polymer blends based on sulfonated poly(ether ketone ketone) and poly(ether sulfone) as proton exchange membranes for fuel cells Steven Swier a,∗ , V. Ramani b , J.M. Fenton b , H.R. Kunz b , M.T. Shaw a,b , R.A. Weiss a,b a b Institute of Materials Science, University of Connecticut, Storrs, CT 06269, USA Department of Chemical Engineering, University of Connecticut, Storrs, CT 06269, USA Received 17 December 2004; received in revised form 30 January 2005; accepted 6 February 2005 Available online 13 April 2005 Abstract The importance of the blend microstructure and its effect on conductivity and structural integrity of proton exchange membranes (PEM) were investigated. Sulfonated poly(ether ketone ketone) (SPEKK) was selected as the proton-conducting component in a blend with either poly(ether sulfone) (PES) or SPEKK with a different sulfonation level. The second component was added to improve the mechanical stability in the fuel cell environment. Membranes were cast from solution using N-methyl-2-pyrrolidone (NMP) and dimethylacetamide (DMAc). Special attention was paid to the ternary solution behavior. Solution cast SPEKK/PES membranes are homogeneous for all studied compositions, 8/2 through 5/5 (w/w), and sulfonation levels, 1.7–3.5 mequiv./g. Although this polymer pair does not show evidence for intrinsic compatibility, the excellent solvent quality results in a frozen-in structure during solution casting. The morphology of SPEKK/SPEKK blends can be tailor-made by finding the right balance between composition, casting solvent and temperature. Co-continuous morphologies can be devised for an SPEKK blend with sulfonation levels of 1.2 and 2 mequiv./g. Both blends show lower swelling than the parent SPEKK. This results in better stability of PEMs during fuel cell testing. © 2005 Elsevier B.V. All rights reserved. Keywords: Poly(ether ketone ketone); Poly(ether sulfone); Polymer blend; Proton-exchange membrane 1. Introduction A proton exchange membrane (PEM) serves as a solid electrolyte in a fuel cell, separating the anode (hydrogen or methanol) from the cathode (oxygen/air) compartments. Proton conduction occurs by a combination of both ion-hopping and convection mechanisms, and for sulfonate ionomer membranes the presence of water is required [1,2]. Polymers used for PEMs must exhibit high proton conductivities and a good resistance to the high-temperature, humid and oxidative environment of a fuel cell. ∗ Corresponding author. E-mail address: [email protected] (S. Swier). 0376-7388/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2005.02.013 Perfluorinated polymer electrolyte membranes such as NafionTM have been the principal choice of PEM in the past. Such ionomers consist of a hydrophobic backbone with sulfonic acid groups placed at the ends of short perfluoro-ether side chains [3]. The acid groups aggregate to form nano-phase separated hydrophilic domains. When swollen with water, a percolation pathway is formed with water-filled channels connecting the ionic domains, which results in the high proton conductivity [3,4]. The hydrophobic phase formed from the perfluorinated backbone provides mechanical integrity and limits the swelling of the PEM. However, perfluorinated polymers are expensive, have relatively poor resistance to methanol transport, which is important in direct methanol fuel cells (DMFC) and have relatively poor mechanical properties when highly swollen by water. S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 Contemporary PEM research involves the development of new polymer electrolytes based on hydrocarbon polymers [5]. Typically, aromatic polymers are functionalized with proton conducting groups, e.g. by sulfonating the phenylene groups in the backbone. One polymer that has been considered a candidate PEM is sulfonated poly(ether ether ketone) (SPEEK), which has a reasonably good proton conductivity [6–8] and has shown promising performance data for both hydrogen/oxygen [9] and direct methanol fuel cells [10]. An advantage of the sulfonated hydrocarbon polymers versus the perfluorinated ionomers is the ability to achieve higher ion-exchange capacities, IEC (i.e. concentration of sulfonate groups in equivalents per mass), but since the perfluorosulfonic acid is a stronger acid, for comparable IEC and hydration [4,11], the perfluorinated ionomers have higher proton conductivities [12]. Polymer blends wherein a sulfonated polymer with high proton conductivity is combined with a non-conductive engineering thermoplastic chosen to maintain mechanical integrity has become a popular contemporary approach to the design of improved PEM materials. A variety of different polymer pairs have been considered, in particular and of relevance to the current paper, SPEEK in combination with poly(ether imide) (PEI) [10,13–18], nylon 6 (PA6) [16], poly(ether sulfone) (PES) [4,19,20] and poly(2,6-dimethyl phenylene oxide) (PPO) [10]. The improvement in mechanical stability can be attributed to the entanglement of these polymers and to possible (partial) mixing due to specific interactions, e.g., ion–dipole, dipole–dipole and proton transfer [10,13,16,21–24]. Such interactions also are reported to reduce significantly methanol cross-over in DMFCs [21]. Intermolecular interactions may also weaken the sulfonateion pair, thus reducing the amount of water needed to promote proton transport. Other poly(aryl ether ketone)s like poly(ether ketone) (PEK) and poly(ether ether ketone ketone) (PEEKK) and poly(ether ketone ketone) (PEKK) have also been investigated in combination with PBI, PEI, [22] poly(phenylene oxide) (PPO) and PES [4,10]. Although key fuel cell performance properties such as proton conductivity, swelling, reactant crossover and voltage–current performance have been studied for most polymer-blend PEMs, the role and control of the blend microstructure has been mostly overlooked. When PEI was used as the minority component in SPEEK/PEI blends, TEM showed a dispersed droplet morphology with PEI particles dispersed in the SPEEK matrix [14]. Miscible blends were found for polymers exhibiting strong acid–base interactions, while weaker bases showed microphase separation [23,24]. The added engineering thermoplastic was usually present as the minority component. The overriding hypothesis of the work described in this paper is that deliberate control of the membrane microstructure is required to optimize proton conductivity and membrane integrity. The extra degree of freedom afforded by using a polymer blend allows for the separation of the transport and mechanical properties, which is not possible in ho- 123 mopolymer ionomer membranes. This paper discusses the control of the blend microstructure by using different parent polymers and mixing ratios and varying the solution casting procedure. Knowledge of the solution behavior of the constituent polymers is essential. In case of phase separation in a metastable region of the phase diagram, nucleation and growth is most probable, resulting in dispersed droplet morphologies, while a co-continuous microstructure is expected for a quench within the unstable, spinodal region. The droplet morphology might be advantageous when the protonconducting phase is the matrix, assuring percolation, and reinforcement of the ionomer by the engineering thermoplastic can improve the mechanical properties of the membrane. A co-continuous, interconnected microstructure is another potentially interesting morphology, where both the conductive polymer and the engineering thermoplastic constitute the matrix. In this way, percolation of the former polymer assures high proton conductivities, while the continuous phase of the engineering thermoplastic should also provide more efficient mechanical reinforcement. Strong interpolymer interactions and/or vitrification from the miscible region of the phase diagram may produce homogeneous microstructures where the ionic microphases are dispersed randomly and entanglements provide membrane integrity. In this case, percolation of the acidic groups has to be ensured by carefully selecting the content of the non-conductive polymer. The work described herein considers blends of sulfonated poly(ether ketone ketone) (SPEKK) ionomers with poly(ether sulfone) (PES). This blend can be compared to the SPEEK/PES system [4,19,20,25]. The thermal properties, water uptake and performance of SPEKK as a PEM material, have been described previously [26]. One of the reasons to select poly(aryl ether ketone)s with higher K/E ratios is the concomitant increase in oxidative stability [27] and the lower degree of swelling upon hydration of the sulfonated polymer [28]. Finally, a promising, novel approach to the design of a blend PEM will be discussed, in which two different SPEKKs are blended, one with high and one with low sulfonation. 2. Experimental 2.1. Materials PEKK with a T/I ratio of 6/4 (OXPEKK SPb, Tg = 154 ◦ C, Tm = 300 ◦ C) was obtained from Oxford Performance Materials, Enfield, CT. PES was obtained from BASF (Ultrason® E6020P, Tg = 225 ◦ C, Mw = 51 kg mol−1 , Mw /Mn = 3.5) and was used as received. Sulfonation of PEKK was performed by using a 5% (w/v) mixture of 53/47 (v/v) concentrated sulfuric acid to fuming sulfuric acid as detailed previously [29]. The sulfonated polymer, SPEKK, was precipitated by drop-wise addition of the solution into six volumes of rapidly stirred de-ionized ice water. The SPEKK was filtered, washed repeatedly with de-ionized water to remove excess acid and dried at 60 ◦ C overnight and then under vacuum at 120 ◦ C for 3 days. The nomenclature used for the ionomer in this pa- 124 S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 per is SPEKKx, where x is the sulfonation level expressed as ion-exchange capacity (IEC), defined as the concentration of sulfonate groups in equivalents per total mass as determined by titration [26]. The IEC and the degree of sulfonation XS , defined as the average number of sulfonate groups per repeat unit, are related as follows: XS IEC (meq/g) = 1000 × 300 + 81XS For the sulfonation of PEKK with fuming sulfuric acid, sulfonation can occur on either of the aromatic rings that have one ether and one ketone substituent; the aromatic ring attached to two ketone groups is not susceptible to an electrophilic substitution reaction [28]. Therefore, the maximum degree of sulfonation of PEKK is two sulfonate groups per (EKK)-repeat unit. That corresponds to a maximum ionexchange capacity (IEC) of 4.4 mequiv./g. Membranes based on blends of SPEKK/PES and SPEKK/SPEKK were prepared by casting 5% (w/v) solutions in N-methyl-2-pyrrolidone (NMP, b.p. = 202 ◦ C) or dimethylacetamide (DMAc, b.p. = 165 ◦ C) onto glass plates. The solution volume and the area of the glass plate were fixed to reproduce the evaporation rate and the final membrane thickness. Typically, the solvent was allowed to evaporate for about 1 day at 60 ◦ C, after which the films were washed at 25 ◦ C with de-ionized water to leach out residual solvent. The residual solvent present after this procedure as obtained from thermogravimetric analysis was below 3 wt.% for both casting solvents. The effect of casting temperature was also investigated by using evaporation temperatures ranging from 40 to 120 ◦ C. Membrane-electrode assemblies (MEA) were prepared for selected systems. The catalysts used were 46.5 wt.% Pt/C on the cathode and 30.1% Pt–23.4% Ru/C on the anode. Both anode and cathode catalysts were provided by Tanaka Kikinzoku Kogyo K. K., Japan. The catalyst powders were dispersed by adding methanol and Nafion® 1100 (5% solution) equivalent to a 25 wt.% ionomer loading. The catalyst was applied to a 5 cm2 area of membrane using an airbrush to yield the MEA. Successive layers of ink were applied to the membrane, with a 5–10 min interval between each application to ensure solvent evaporation, which was facilitated by an infrared heating lamp placed directly behind the membrane. The electrode noble metal loading was in the range 0.3–0.45 mg/cm2 . 2.2. Materials characterization Blend morphology was studied with a Philips 300 transmission electron microscope (TEM) operating at 80kV and a Topometrix’ Explorer atomic force microscope (AFM). The sample preparation for both techniques consisted of embedding membranes in an EponTM -Araldite® embedding mixture followed by ultra-microtomy with a diamond knife to obtain thin sections. Sections were placed on copper grids for TEM analysis or pre-cleaned glass cover slips for AFM. The TEM specimens were stained with a 1 M Pb(NO3 )2 aque- ous solution, which preferentially darkens the SPEKK phase in the TEM images. For SPEKK/SPEKK blends, the higher IEC phase was more noticeably darkened, which provided the sample contrast. For AFM, drying of the sections on glass resulted in shrinkage of the swollen ionomer phase (microtomed sections were initially floated onto water), and the higher IEC SPEKK phase, which shrank the most, corresponded to areas that were lower in topography. The thermal properties of membranes were measured using a TA Instruments Q100 differential scanning calorimeter (DSC). The glass transition of SPEKK as a function of water content was obtained by exposing DSC pans containing the SPEKK membranes to controlled relative humidity environments. Equilibrium was attained after 12 h as confirmed from transient water uptake experiments. Sample pans were crimped quickly after about 48 h exposure. To measure the glass transition temperatures in the presence of water beyond 100 ◦ C, high pressure stainless steel pans were used. Proton conductivity measurements of hydrated membranes were made at room temperature along the membrane direction over a frequency range of 10 mHz to 106 Hz and an ac voltage amplitude of 50 mV using a frequency response analyzer (Solartron SI 1260, impedance/gain-phase analyzer) combined with a potentiostat (Solartron SI 1287, electrochemical interface). Membranes were equilibrated in a 98% relative humidity, R.H., environment using a conductivity cell based on previous work [5]. Turbidity measurements on ternary solutions were obtained from visual observations during evaporation of the solvent at 60 ◦ C. Solutions with increasing polymer concentration were allowed to equilibrate at 25 ◦ C for about 1 h. When a cloudy mixture was obtained, a small amount of solvent was added to clear the solution and obtain a better estimate of the cloud point. Subsequent to anode and cathode catalyst application, MEAs were placed between two thin Teflon® sheets, which were then placed between two rubber sheets. The entire assembly was introduced between two stainless steel plates and compression molded in a Carver hot press at 120 ◦ C and 207 kPa. The MEAs were assembled in a 5 cm2 hardware with single serpentine flowfields (Electrochem Inc.; model no. FC05–01SP). Commercial gas-diffusion layers (GDLs) (SGL Carbon; model no. 10BB) were used. The resistance of the membrane was obtained using the current-interrupt technique built into the cell testing system. The conductivity was estimated from the measured value of resistance using recorded values of the active area (5 cm2 ) and thickness of the membrane. The performance of the cell was evaluated by obtaining polarization curves at 80, 100 and 120 ◦ C. For the experiments at 80 ◦ C both anode and cathode gases were saturated at 80 ◦ C and 73 ◦ C for 100% and 75% R.H. operation, respectively. Fuel cell tests at 100 and 120 ◦ C were performed at 70% R.H. and 35% R.H., respectively. H2 was used at the anode and air or O2 at the cathode. All data were obtained at a pressure of 101 kPa (1 atm). Further details of the system used and testing conditions adopted are provided elsewhere [30]. S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 Fig. 1. Water content for SPEKK as a function of sulfonation level: hydration number λ (䊉), wt.% absorption () at 25 ◦ C and 98% R.H. The filled square () is the hydration number measured for NafionTM 112. 3. Results and discussion 125 Fig. 2. Glass transition temperature of SPEKK1.9 as a function of hydration number λ. The membranes were equilibrated in different humidity environments at 25 ◦ C (䊉); after immersing in water at 50 ◦ C (). The dashed line denotes λ from which melting of water crystals were detected, and the solid line is the Tg prediction from the Fox mixing rule with Tg,H2 O = −139 ◦ C [33]. 3.1. The role of water in SPEKK Promising proton conductivities were previously reported for SPEKK with relatively high sulfonation levels; e.g., 0.06 S/cm for SPEKK2.0 at 25 ◦ C and 98% R.H. [26]. That value is comparable to that reported for NafionTM 112, which has an IEC of only 0.9 mequiv./g. The higher intrinsic acidity of the perfluorosulfonic acid sites, combined with a larger hydrophobic/hydrophilic separation, has been stated to result in the higher intrinsic conductivity of perfluorinated ionomers [4,11]. As indicated earlier, water is needed to assure high proton mobilities in ionomers. Fig. 1 shows that although the water uptake of SPEKK increases with sulfonation level, the number of water molecules per sulfonic acid site or the hydration number λ is almost constant at 10 for 25 ◦ C and 98% R.H., which is also comparable to the value for NafionTM [1]. Of the 10 water molecules associated with each sulfonic acid group, three are strongly bound to the acid sites and the other seven are loosely bound or free water within the microphaseseparated hydrophilic domains [5,31]. The glass transition temperature (Tg ) of the hydrated membranes was very sensitive to the amount of plasticizing water that was retained, Fig. 2. A dramatic drop in Tg of SPEKK1.9 from 200 to 85 ◦ C occurred when about three water molecules per sulfonic acid group were introduced into the ionomer. That amount corresponds to the solvation of the proton and the sulfonate ion, and λ = 3 also corresponds to the percolation threshold and the insulator-to-conduction transition [3]. Additional water is not strongly bound to the sulfonic acid group, as is evidenced by the presence of a melting peak around 0 ◦ C in DSC (not shown). Melting peaks occur for values of λ to the right of the dashed line in Fig. 2. In that region, water swells the polymer, but does not significantly affect the Tg . This can be attributed to phase separation of a water-rich phase, which results in a deviation of the polymer Tg from the Fox mixing rule [32]. The strong effect of water on the Tg as the membranes were swollen is further revealed by the effect of sulfonation level. Fig. 3 shows that increasing the sulfonation level of dry SPEKK membranes led to a small increase in Tg , probably due to hydrogen bonding interactions. That increase, however, is completely offset by the presence of water. For example, for the water-swollen membranes, Tg decreased by 90 ◦ C per mequiv./g, which corresponded to ∼7 ◦ C/wt.% H2 O. The drop in Tg of SPEKK below 100 ◦ C in the presence of water is important in view of the temperature window for fuel cell operation, ∼80–120 ◦ C. During operation, the dry SPEKK membrane undergoes a change from a glassy to an elastic, gel-like material. Fig. 4 shows the equilibrated water content for SPEKK membranes with varying IEC as a function of temperature. For SPEKK0.8 and SPEKK1.2, the water content did not increase significantly with temperature, resulting in a behavior similar to that of NafionTM 112. The low Fig. 3. Glass transition temperature of SPEKK as a function of IEC: membranes vacuum dried at 120 ◦ C (, slope: 14 ± 2 ◦ C/(mequiv./g), P < 0.0001); membranes immersed in water at 25 ◦ C for 48 h (䊉, slope: −90 ± 14 ◦ C/(mequiv./g), P = 0.003); P is the probability of being wrong by rejecting the hypothesis of slope = 0, i.e., no trend. 126 S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 Fig. 4. Hydration number, λ, as a function of the water temperature for SPEKK (T/I 6/4); IEC = 2.1 (), 1.7 (), 1.4 (), 1.2 (), 0.8 (䊉), NafionTM 112 (). The equilibration time at each temperature was 48 h. amount of hydrophilic, water-containing sites produced only a moderate Tg depression. Moreover, the sulfonation level of SPEKK0.8 was sufficiently low that the polymer retained some residual crystallinity [26]. However, excessive swelling of the membranes occurred for SPEKK1.4 above 60 ◦ C and at even lower temperatures for the higher IEC membranes. Similar water absorption trends were also reported for sulfonated poly(ether ether ketone ketone) (SPEEKK) membranes [4]. Unfortunately, the high sulfonation levels of SPEKK needed to attain competitive proton conductivities resulted in poor membrane integrity and premature failure in actual fuel cell operation. SPEKK2.0, for example, attained a cell voltage of 0.5 V at a current loading of 1 A/cm2 for about 4 h, after which fracture occurred around the gasket edge. These results provided the incentive to alternative membrane designs, namely composite membranes of SPEKK with a non-conductive engineering thermoplastics, in this case, poly(ether sulfone), PES. 3.2. SPEKK/PES composite membranes with a homogeneous morphology Homogeneous SPEKK/PES blends were obtained by solution casting from DMAc, as indicated by the TEM micrograph in Fig. 5a for an 85/15 (w/w) blend. All blend compositions used, 80/20 through 50/50, and SPEKK sulfonation levels, 1.7–3.5 mequiv./g, resulted in similar homogeneous morphologies for solution-cast membranes. The single-phase morphology could arise from thermodynamic miscibility of the two polymers or from the solution casting procedure, which could produce a metastable morphology for the blend by freezing-in a single-phase state of the ternary solution during solvent evaporation. Turbidity measurements were performed on SPEKK/ PES/DMAc solutions, and sulfonation levels from 1.7 to 2.5 mequiv./g, with increasing total polymer concentration. All mixtures studied in this way were transparent, including the fully dried SPEKK/PES binary film. Similarly, no cloud Fig. 5. TEM micrograph (cross-sectional view) of (a) 85/15 (w/w) SPEKK1.7/PES blend and (b) 85/15 (w/w) SPEKK1.7/PEI blend cast at 60 ◦ C from a 5% (w/v) DMAc solution. point was detected for SPEKK/DMAc and PES/DMAc binary solutions. This is in contrast to the SPEKK/PEI system, where the limited solubility of PEI in DMAc typically produced phase separation of the polymers at a total polymer concentration of 15% (w/v) and resulted in heterogeneous morphologies, see Fig. 5b [22]. Infrared spectroscopy was performed on cast SPEKK/PES membranes in an attempt to identify a chemical interaction responsible for the miscibility. The focus was on the specific S–O absorption of the aromatic sulfonic acid group at 926 cm−1 . No shift in that absorption from that in the neat SPEKK was observed, which is consistent with the absence of specific hydrogen bonding interactions between the two polymers. However, an electron donor-acceptor complex or a similarity in structure cannot be excluded as a possible explanation for the miscibility. The electron donor-acceptor complex has been proposed as the origin of miscibility in a similar blend, SPEEK/PES [17]. A different approach to determine SPEKK/PES miscibility is to bring thin SPEKK and PES films in contact, fol- S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 127 Fig. 7. Polarization curves at 80 ◦ C and 75% R.H. for SPEKK1.7 (, failure occurred after 10 h; conductivity σ from current interrupt = 0.02 S/cm, thickness: 24 m) and a 85/15 (w/w) SPEKK1.7/PES blend (♦, no failure before 50 h; σ = 0.01 S/cm, thickness: 38 m). Fig. 6. Proton conductivity (σ at 25 ◦ C, 98% R.H., 䊉) and hydration number (λ at 90 ◦ C, immersed in water, ♦) for SPEKK1.7/PES membranes as a function of (A) wt.% PES and (B) for 5/5 (w/w) blends as a function of SPEKK IEC. lowed by annealing in a temperature range where miscibility is expected. The annealing temperature has to be selected above the glass transition of at least one of the components (>200 ◦ C) and lower than the desulfonation temperature of SPEKK (<240 ◦ C). TEM micrographs showed sharp interfaces on a scale of 500 nm in cross-sectional views, indicating poor interfacial mixing. Since SPEKK and PES do not seem to be a miscible pair, the excellent solvent quality of DMAc for SPEKK and PES seems to be responsible for the homogeneous morphologies found for solution cast blends. The temperature/solvent history results in a quench from solution, interfering with the formation of an equilibrium, phase-separated blend. This is further confirmed by substituting PES for the similar polysulfone (PSU), resulting in macroscopically phase separated SPEKK/PSU blends. Ternary SPEKK/PSU/DMAc solutions were indeed cloudy at relatively low polymer loadings (<20%). Homogeneous SPEKK/PES blends offer significant advantages in regard to membrane swelling, as illustrated for a range of sulfonation levels and blend compositions in Fig. 6. The addition of 15 wt.% PES to SPEKK1.7 decreased the water concentration at 90 ◦ C to below 20 H2 O/SO3 H, even though the neat SPEKK1.7 exhibited excessive swelling (>80 H2 O/SO3 H). A similar amount of PEI added to SPEKK1.7 still resulted in a very high water content, ∼60 H2 O/SO3 H, which was a consequence of poor interaction between the SPEKK and PEI [22] Water-soluble SPEKKs (IEC > 2.4 mequiv./g) were rendered insoluble by the addition of 50% PES. While the addition of PES to the SPEKK lowers the inherent conductivity of the blend, that can be compensated by increasing the IEC of the SPEKK, especially if it is no longer water extractable from the membrane. Thus, the design of the SPEKK/PES membranes offers several degrees of freedom, including the SPEKK IEC and blend composition that could provide a membrane with high conductivity, yet reasonably low water swelling and good mechanical properties. The fuel cell performance of SPEKK1.7 is compared to an 85/15 SPEKK1.7/PES blend in Fig. 7. Although a lower performance was found for the latter PEM, the blend was stable in the fuel cell environment over 50 h in comparison to 10 h for the parent SPEKK. Note that our current MEA set-up does not allow for long-term stability tests. An alternative to a single-phase structure for the blend is to develop a well-defined heterogeneous, phase-separated morphology. This may have advantages in light of the increased interfacial region between the two phases that might concentrate the ionic groups due to the specific interaction between the two polymers. As indicated before, SPEKK and PES have limited compatibility. Thermally induced phase separation by annealing the cast blend PEM’s above their glass transition temperature (200–240 ◦ C) was attempted, but this was not successful largely due to the restriction of the temperature to below the desulfonation temperature of SPEKK, ∼240 ◦ C. Preliminary results, however, indicated that phase separated morphologies could be obtained by neutralizing the SPEKK to its Zn2+ or Mn2+ -salt prior to solution casting. The additional parameters of a metal counter-ion and the extent of neutralization provide even more flexibility with regard to the design and control of morphology of these blend PEMs. Once the desired morphology is achieved using the metal salt ionomers, one can re-acidify the ionomer by exposure to a 128 S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 strong acid and retain the morphology in the sulfonic acid form of the blend. 3.3. Co-continuous morphology of SPEKK/SPEKK blend PEMs As noted in the discussion of Fig. 4, low IEC SPEKK produces fairly robust membranes, due primarily to relatively low swelling by water. That ionomer may, therefore be a viable blend component for improving the performance of a high IEC SPEKK. Although the proton conductivity of the low IEC ionomer is limited [26], it still has an advantage over a non-conductive component in that it will contribute to the conductivity of the system. Several intriguing morphologies were obtained by solution casting high IEC/low IEC, SPEKK/SPEKK membranes, as illustrated by the micrographs in Fig. 8. For both TEM and AFM the darker regions in Fig. 8 represent the higher IEC phase (see Section 2). The morphologies found with both techniques corresponded very well. When the sulfonation levels of both SPEKK’s were similar, e.g., the SPEKK2.0/SPEKK1.5 blend, a homogeneous morphology was formed by solution casting from NMP. When the low IEC component was reduced to SPEKK1.2, a dispersed-particle morphology was achieved, and for an SPEKK2.0/SPEKK0.8 blend, the dispersed phase became larger. DSC thermograms of the blends and the pure components are compared in Fig. 9. Following solution casting, all membranes were converted to the sodium salt to increase the Tg difference between the constituents. This is based on the finding that the Tg of the acid derivative of SPEKK increased about 14 ◦ C/(mequiv./g), while the stronger intermolecular ionic interactions present in the salt form result in an increase of 90 ◦ C/(mequiv./g) [26]. The neutralization procedure did Fig. 8. TEM and AFM micrographs of 5/5 (w/w) blends of (a) SPEKK2.0/SPEKK1.5, (b) SPEKK2.0/SPEKK1.2 and (c) SPEKK2.0/SPEKK0.8 cast from NMP at 60 ◦ C; plotting the size of the minority phase particles as a function of IEC ratio results in a positive slope of 5 m with P = 0.07. S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 129 Fig. 9. DSC thermograms (scan rate = 10 ◦ C/min) of SPEKKs and 50/50 (w/w) SPEKKx/SPEKKy (x/y) blends cast from NMP at 60 ◦ C. not affect the blend morphology as confirmed from TEM (not shown). The thermal behavior of the blends corroborates the morphologies shown in Fig. 8. The high IEC combination, SPEKK2.0/SPEKK1.5, exhibited a single Tg , indicating a single-phase morphology, while the other two blends showed two distinct Tg ’s, consistent with the phase-separation of the two. A combination of factors determines the occurrence of phase separation in SPEKK/SPEKK blends. It is well known that differences in monomer content [34] and sequence distribution [35] can induce phase separation in copolymer/copolymer blends (copolymer effect). In addition, the effect of the casting solvent has to be understood. One factor that might be crucial in this respect is that the solubility of SPEKK decreased with decreasing sulfonation level, so that mixing the two components became increasingly more difficult as the difference in IEC increased. For example, SPEKK0.8 was only partially soluble in NMP and DMAc, while SPEKK with IEC > 1.4 mequiv./g was soluble in those solvents over the entire concentration range [26]. For SPEKK with 0.8 < IEC < 1.4 mequiv./g, SPEKK was initially soluble, but exhibited solvent-induced crystallization after about 20 h, as evidenced from the resulting opaque solutions and from X-ray analysis on the dried membranes. This can also account for the SPEKK/SPEKK phase separation when the difference in sulfonation level is increased. Note that extensive crystallization from solution embrittles the resulting PEMs. The SPEKK/SPEKK membranes studied here were therefore quickly mixed and cast from transparent solutions. Further morphology control can be obtained by changing the casting conditions, e.g., temperature and solvent. The effect of changing to a lower boiling casting solvent (DMAc) while keeping the casting temperature at 60 ◦ C is shown by the AFM micrographs in Fig. 10 for SPEKK2.0/SPEKK1.2 blends with compositions from 4/6 to 6/4 (w/w). Note that the darker phase corresponds to the high-IEC component. The corresponding TEM images were similar. The difference in morphology between the 5/5 (w/w) SPEKK2.0/SPEKK1.2 Fig. 10. AFM micrographs of SPEKK2.0/SPEKK1.2 blends: (a) 6/4, (b) 5/5 and (c) 4/6 (w/w) blends cast from DMAc at 60 ◦ C. blend cast from NMP (Fig. 8) and DMAc (Fig. 10) is quite striking; the latter membrane appears to possess a co-continuous morphology. The faster evaporation rate of DMAc probably results in a fast quench in the unstable 130 S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 Fig. 11. Proton conductivity (σ at 25 ◦ C, 98% R.H.) and hydration number (λ at 90 ◦ C, immersed) for SPEKK2.0/SPEKK1.2 blends as a function of blend composition. The membranes were cast from DMAc at 60 ◦ C. region of the phase diagram associated with spinodal-type behavior. In addition, SPEKK1.2/DMAc solutions show a stronger tendency to solvent-induced crystallization than SPEKK1.2/NMP solutions. Phase inversion occurred close to the 4/6 SPEKK2.0/SPEKK1.2 composition, and disperseddroplet morphologies were observed for larger composition deviations from 5/5 blends. Fig. 11 shows the proton conductivities (25 ◦ C for 98% R.H.) and water absorption at 90 ◦ C for SPEKK2.0/ SPEKK1.2 blends cast from DMAc as a function of blend composition. The conductivities exceeded 0.01 S/cm for blends containing up to 60 wt.% SPEKK1.2. The co-continuous morphology found at a 4/6 SPEKK2.0/ SPEKK1.2 composition (Fig. 10) assured a continuous path for the conductive SPEKK2.0 component in the blend. On the other hand, having a continuous SPEKK1.2 phase explains the striking decrease in water content accompanying the transition to a co-continuous morphology around a 5/5 composition. For higher SPEKK1.2 concentrations, the low IEC component became the only continuous phase and the conductivity dropped accordingly. A PEM with optimum properties will, therefore, most likely be found around the 4/6 SPEKK2.0/SPEKK1.2 composition. Morphological changes can also be achieved by changing the casting temperature, as illustrated for the 4/6 (w/w) SPEKK2.1/SPEKK1.15 blend in Fig. 12. Co-continuous morphologies were attained for the lower casting temperatures. Note that a slightly larger difference in sulfonation level was used in comparison to the blends in Fig. 10. The proton conductivity did not change appreciably with casting temperatures from 40 to 80 ◦ C, see Table 1. The highest casting temperature used, 120 ◦ C, produced a grossly phaseseparated blend and a membrane with a rough surface. The incompatibility of the two polymers became more distinct as the casting temperature increased, due presumably to a higher rate of interdiffusion in the initial stages of solvent evaporation. Ultimately, a decreasing solvent content in the membrane will increase its Tg above the casting temperature, resulting in a frozen-in microstructure. Fig. 12. TEM micrographs of 4/6 (w/w) SPEKK2.1/SPEKK1.15 blends cast from DMAc using different casting temperatures. A measure of the merit of the different membranes can be obtained by comparing the proton conductivities and water contents of pure SPEKK with those of SPEKKbased blends as a function of the average IEC (Fig. 13). The proton conductivities of the blend membranes are slightly higher than those for pure SPEKK. The most significant improvement was found in the membrane integrity S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 131 Table 1 Proton conductivity ( at 25 ◦ C, 98% R.H.) and hydration number ( at 90 ◦ C, immersed) for 4/6 (w/w) SPEKK2.1/SPEKK1.15 blends cast from DMAc using different casting temperatures (x indicates excessively swollen membranes) Tcasting (◦ C) σ (S/cm) λ at 90 ◦ C 40 50 60 80 120 0.019 0.020 0.023 0.020 0.012 69 80 x x x as evidenced by much lower water contents at 90 ◦ C for SPEKK-based blends. An SPEKK/PES blend membrane with an average IEC of 1.75 mequiv./g, for example, takes up less than 20 H2 O/SO3 H, while pure SPEKK membranes fail at this IEC. Plotting proton conductivity (25 ◦ C) as a function of water content (90 ◦ C) in Fig. 14 further confirms the advantage of using blend membranes. Although these properties were not obtained at the same temperature, the proton conductivity follows an Arrhenius-type dependency with similar activation energies for different sulfonation levels [36]. This was confirmed by measurement of E ∼ 20 kJ/mol for SPEKK1.0 and SPEKK2.0. Fig. 14 again shows that SPEKK/SPEKK and SPEKK/PES blends exhibit higher proton conductivities with less water content than do neat SPEKKs, and the two blend PEMs also exhib- Fig. 13. Proton conductivity (σ at 25 ◦ C, 98% R.H.) and water content (λ at 90 ◦ C, immersed) as a function of the average IEC for different SPEKKbased membranes; neat SPEKK (䊉), SPEKK/SPEKK (), SPEKK/PES (). Fig. 14. Proton conductivity (σ at 25 ◦ C, 98% R.H.) as a function of water content (λ at 90 ◦ C, immersed) for different SPEKK-based membranes; neat SPEKK (IEC range: 0.8–1.4 mequiv./g) (䊉), SPEKK/SPEKK (IEC range 0.8–2.1 mequiv./g) (), SPEKK/PES (IEC range 1.7–3.5 mequiv./g) (). ited better mechanical stability and durability in a fuel cell environment. The MEA performance data at 80 ◦ C of some selected SPEKK/SPEKK membranes with different conditions of relative humidity and reactant gases are compared with that of a neat SPEKK1.4 membrane in Fig. 15. SPEKK1.4 constitutes the highest sulfonation level for which a PEM was stable for up to 2 days in a fuel cell at 80 ◦ C and 75% R.H. All blendbased PEM’s performed better than pure SPEKK, due in large part to the higher proton conductivity that was obtained for a similar water content (Figs. 13 and 14) and the higher average IEC that could be used for the blend membranes (1.45 and 1.53 mequiv./g for the 32/68 and 4/6 membranes, respectively). Figs. 15 and 16 show the MEA performance curves for SPEKK2.1/SPEKK1.15 composite membranes at Fig. 15. Polarization curves of selected membranes measured at 80 ◦ C: (䊉) SPEKK1.4 (failure occurred after 36 h, σ = 0.003 S/cm, thickness: 30 m, 75%R.H. H2 /O2 ); () SPEKK2.1/SPEKK1.15 (32/68 w/w) cast at 60 ◦ C (σ = 0.013 S/cm, thickness: 50 m, 75%R.H. H2 /O2 ); SPEKK2.1/SPEKK1.15 (4/6 w/w) cast at 50 ◦ C (thickness: 51 m): () 75%R.H. H2 /Air (σ = 0.015 S/cm), () 100%R.H. H2 /Air (σ = 0.046 S/cm), () 100%R.H. H2 /O2 (σ = 0.038 S/cm). 132 S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 Fig. 16. Polarization curves (H2 /Air) for an SPEKK2.1/SPEKK1.15 (4/6 w/w) cast at 40 ◦ C from DMAc: (䊉) 80 ◦ C and 100% R.H. (σ = 0.036 S/cm, thickness: 38 m); () 100 ◦ C and 70% R.H. (σ = 0.005 S/cm, thickness: 38 m); () 120 ◦ C and 35% R.H. (thickness: 38 m). 80–120 ◦ C. The performance suffers significantly at the higher temperatures, as a result of the low humidity conditions. As indicated before, this can be attributed to the crucial role of water for promoting high proton mobility [4,11]. The literature suggests several ways to rectify this problem, including using acid–base intermolecular interactions between polymer pairs to induce ionic crosslinking, improve membrane integrity, and to weaken the proton-sulfonate interaction, making proton conductivity less dependent on water content [10,13,16,21,23–25]. The best improvements in high temperature, low humidity performance involved complexation of basic polymers using oxo-acids like phosphoric acid [4,37–40]. Introducing a basic polymer as the second component in SPEKK-based blends, possibly doped with phosphoric acid, is a logical next step in the research described in this paper. SPEKK and PES are intrinsically incompatible as indicated by the absence of specific hydrogen bonding interactions and the lack of intermixing between stacked SPEKK/PES films. On the other hand, obtaining heterogeneous morphologies by high-temperature annealing of cast films is hampered by the onset of desulfonation of SPEKK. Preliminary experiments show that neutralizing SPEKK prior to casting does allow heterogeneous morphology development. In this way, the drop in proton conductivity corresponding to adding the non-conductive PES might be limited. A novel strategy was devised based on blends of SPEKK with different sulfonation levels. Controlled morphologies can be attained by blending a high-IEC SPEKK, having high proton conductivity, with a low-IEC SPEKK, exhibiting good resistance to swelling. The best compromise of these two properties can be obtained in case of a cocontinuous microstructure, characterized by interconnection between the two SPEKKs. A relatively high IEC for both components results in homogeneous solution cast membranes. Heterogeneous structures are formed for a larger difference in IEC. This can probably be attributed mostly to the decrease in solubility of the low IEC SPEKK in the casting solvent. Morphology control can be obtained not only by the IEC difference and composition of the ionomers, but also by changing the casting temperature and solvent. Polarization data on SPEKK/SPEKK and SPEKK/PES blends at 80 ◦ C and high relative humidities indicate an important improvement in performance and structural stability over the pure SPEKK membranes. However, since SPEKK relies strongly on the presence of water to ensure high proton mobilities, high temperature/low humidity fuel cell operation is difficult to achieve. Employing a basic polymer as the second component in combination with acid doping will be attempted to reach this technical goal. Having a tool to control the morphology of these blends will be very useful in this respect. 4. Conclusions Although competitive proton conductivities are found for SPEKK with a high sulfonation level, the presence of water causes a drop in Tg concomitant with excessive membrane swelling. This jeopardizes the long-term stability of SPEKKbased PEMs in the fuel cell environment. The added flexibility of polymer blends can resolve this issue to some extent. As a first strategy, a miscible blend of SPEKK with the non-conductive engineering thermoplastic PES can be obtained by solution casting. Both components are soluble in DMAc over the entire concentration range, eventually leading to a frozen-in homogeneous microstructure after solvent evaporation. Intimate entanglement of the polymer pair results in excellent PEM integrity in the fuel cell environment. Highly sulfonated, water soluble SPEKK can be turned into a robust membrane by adding PES, further evidence for the strong interpenetration. Acknowledgement The authors thank the Department of Energy (Grant 69797-001-033D) and Connecticut Innovations, Inc (Yankee Ingenuity Grant 01Y09) for support of this research, Oxford Performance Materials for supplying the PEKK samples and BASF for supplying PES. References [1] T.A. Zawodzinski, M. Neeman, L.O. Sillerud, S. Gottesfeld, Determination of water diffusion coefficients in perfluorosulfonate ionomeric membranes, J. Phys. Chem. 95 (1991) 6040. [2] M. Eikerling, A.A. Kornyshev, A.M. Kuznetsov, J. Ulstrup, S. Walbran, Mechanisms of proton conductance in polymer electrolyte membranes, J. Phys. Chem. 105 (2001) 3646. S. Swier et al. / Journal of Membrane Science 256 (2005) 122–133 [3] M. Matsuyama, E. Kokufuta, T. Kusumi, K. Harada, Ion percolation and insulator-to-conductor transition in Nafion perfluorosulfonic acid membranes, Macromolecules 13 (1980) 198. [4] K.D. Kreuer, On the development of proton conducting polymer membranes for hydrogen and methanol fuel cells, J. Membr. Sci. 185 (2001) 29. [5] M. Rikukawa, K. Sanui, Proton-conducting polymer electrolyte membranes based on hydrocarbon polymers, Prog. Polym. Sci. 25 (2000) 1463. [6] S. Kaliaguine, S.D. Mikhailenko, K.P. Wang, P. Xing, G. Robertson, M. Guiver, Properties of SPEEK based PEMs for fuel cell application, Catal. Today 82 (2003) 213. [7] T. Kobayashi, M. Rikukawa, K. Sanui, N. Ogata, Proton-conducting polymers derived from poly(ether-etherketone) and poly(4phenoxybenzoyl-1,4-phenylene), Solid-State Ionics 106 (1998) 219. [8] P. Xing, G.P. Robertson, M.D. Guiver, S.D. Mikhailenko, K. Wang, S. Kaliaguine, Synthesis and charaterization of sulfonated poly(ether ether ketone) for proton exchange membranes, J. Membr. Sci. 229 (2004) 95. [9] B. Bauer, D.J. Jones, J. Roziere, L. Tchicaya, G. Alberti, M. Casciola, L. Massinelli, A. Peraio, S. Besse, E. Ramunni, Electrochemical characterisation of sulfonated polyetherketone membranes, J. New Mater. Electrochem. Syst. 3 (2000) 93. [10] L. Jorissen, V. Gogel, J. Kerres, J. Garche, New membranes for direct methanol fuel cells, J. Power Sources 105 (2002) 267. [11] K.D. Kreuer, On the complexity of proton conduction phenomena, Solid-State Ionics 136–137 (2000) 149. [12] S. Koter, P. Piotrowski, J. Kerres, Comparative investigations of ionexchange membranes, J. Membr. Sci. 153 (1999) 83. [13] J. Kerres, A. Ullrich, F. Meier, T. Haring, Synthesis and characterization of novel acid–base polymer blends for application in membrane fuel cells, Solid-State Ionics 125 (1999) 243. [14] S.D. Mikhailenko, S.M.J. Zaidi, S. Kaliaguine, Electrical properties of sulfonated polyether ether ketone/polyetherimide blend membranes doped with inorganic acids, J. Polym. Sci. Pol. Phys. 38 (2000) 1386. [15] S.M.J. Zaidi, S.D. Mikhailenko, G.P. Robertson, M.D. Guiver, S. Kaliaguine, Proton conducting composite membranes from polyether ether ketone and heteropolyacids for fuel cell applications, J. Membr. Sci. 173 (2000) 17. [16] W. Cui, J. Kerres, G. Eigenberger, Development and characterization of ion-exchange polymer blend membranes, Sep. Purif. Technol. 14 (1998) 145. [17] R.J. Karcha, R.S. Porter, Miscible blends of modified poly(aryl ether ketones) with aromatic polyimides, J. Polym. Sci. Pol. Phys. 31 (1993) 821. [18] R.J. Karcha, R.S. Porter, Miscible blends of a sulfonated poly(aryl ether ketone) and aromatic polyimides, J. Polym. Sci. Pol. Phys. 27 (1989) 2153. [19] C. Manea, M. Mulder, New polymeric electrolyte membranes based on proton donor–proton acceptor properties for direct methanol fuel cells, Desalination 147 (2002) 179. [20] C. Manea, M. Mulder, Characterization of polymer blends of polyether sulfone/sulfonated polysulfone and polyether sulfone/sulfonated polyetherether ketone for direct methanol fuel cell applications, J. Membr. Sci. 206 (2002) 443. [21] M. Walker, K.M. Baumgartner, M. Kaiser, J. Kerres, A. Ullrich, E. Rauchle, Proton-conducting polymers with reduced methanol permeation, J. Appl. Polym. Sci. 74 (1999) 67. 133 [22] S. Swier, M.T. Shaw, R.A. Weiss, Morphology control of sulfonated poly(ether ketone ketone) poly(ether imide) blends and their use in proton-exchange membranes, J. Membr. Sci., Submitted for publication. [23] J.A. Kerres, Development of ionomer membranes for fuel cells, J. Membr. Sci. 185 (2001) 3. [24] J. Kerres, A. Ullrich, Synthesis of novel engineering polymers containing basic side groups and their application in acid–base polymer blend membranes, Sep. Purif. Technol. 22–23 (2001) 1. [25] J. Kerres, A. Ullrich, T. Haring, M. Baldauf, U. Gebhardt, W. Preidel, Preparation, characterization and fuel cell application of new acid–base blend membranes, J. New Mater. Electrochem. Syst. 3 (2000) 229. [26] S. Swier, Y.S. Chun, J. Gasa, M.T. Shaw, R.A. Weiss, Sulfonated poly(ether ketone ketone) ionomers as proton exchange membranes, Polym. Eng. Sci., Submitted for publication. [27] C.A. Linkous, Development of solid polymer electrolytes for water electrolysis at intermediate temperatures, Int. J. Hydrogen Energy 18 (1993) 641. [28] H.H. Ulrich, G. Rafler, Sulfonated poly(aryl ether ketone)s, Angew. Makromol. Chem. 263 (1998) 71. [29] S. Swier, J. Gasa, M.T. Shaw, R.A. Weiss, Sulfonation reaction kinetics of poly(ether ketone ketone) (PEKK) using a mixture of concentrated and fuming sulfuric acid, Ind. Eng. Chem. Res. 43 (2004) 6948. [30] V. Ramani, H.R. Kunz and J.M. Fenton, Stabilized heteropolyacid/Nafion, composite membranes for elevated temperature/low relative humidity PEFC operation, Electrochim. Acta, in press. [31] Y.S. Kim, L.M. Dong, M.A. Hickner, T.E. Glass, V. Webb, J.E. McGrath, State of water in disulfonated poly(arylene ether sulfone) copolymers and a perfluorosulfonic acid copolymer (Nafion) and its effect on physical and electrochemical properties, Macromolecules 36 (2003) 6281. [32] T.G. Fox, P.J. Flory, J. Appl. Phys. 21 (1950) 581. [33] A. Hallbrucker, E. Mayer, G.P. Johari, J. Phys. Chem. 93 (1989) 4986. [34] G. ten Brinke, F.E. Karatz, W.J. MacKnight, Phase behavior in copolymer blends: poly(2,6-dimethyl-1,4-phenylene oxide) and halogen-substituted styrene copolymers, Macromolecules 16 (1983) 1827. [35] A.C. Balazs, F.E. Karasz, W.J. MacKnight, H. Ueda, I.C. Sanchez, Copolymer/copolymer blends: effect of sequence distribution on miscibility, Macromolecules 18 (1985) 2784. [36] L. Li, J. Zhang, Y. Wang, Sulfonated poly(ether ether ketone) membranes for direct methanol fuel cell, J. Membr. Sci. 226 (2003) 159. [37] A. Bozkurt, M. Ise, K.D. Kreuer, W.H. Meyer, G. Wegner, Protonconducting polymer electrolytes based on phosphoric acid, SolidState Ionics 125 (1999) 225. [38] J.T. Wang, R.F. Savinell, J. Wainright, M. Litt, H. Yu, A H2 /O2 fuel cell using acid doped polybenzimidazole as polymer electrolyte, Electrochim. Acta 41 (1996) 193. [39] O. Savadogo, Emerging membranes for electrochemical systems. I. solid polymer electrolyte membranes for fuel cell systems, J. New Mater. Electrochem. Syst. 1 (1998) 47. [40] O. Savadogo, Emerging membranes for electrochemical systems. Part II. High temperature composite membranes for polymer electrolyte fuel cell (PEFC) applications, J. Power Sources 127 (2004) 135.
© Copyright 2026 Paperzz