Zwitterionic silica copolymer based crosslinked organic–inorganic hybrid polymer electrolyte membranes for fuel cell applications

Abstract

Zwitterionic (ZI) copolymers (consisting of sulfonic acid and amine groups) with plenty of –Si(OCH)₃groups similar to stems, branches and fruits of vines from a bionic aspect, were synthesized as a cross-linking agent. Organic–inorganic hybrid zwitterionic membranes (ZIMs), with high flexibility, charge density and conductivity, was prepared using poly(vinyl alcohol) (PVA). Developed ZIMs with dual acidic and basic functional groups, exhibited high stabilities, water retention ability and cation selectivity. The ZIMs (especially Si–70%) were designed to possess all the required properties (water uptake: 40.6%; ion-exchange capacity: 1.52 equiv. g⁻¹; electro-osmotic flux: (2.34 × 10⁻⁵ cm s⁻¹A⁻¹); and conductivity: 9.67 × 10⁻² S cm⁻¹. ZIMs were designed to possess all of the required properties of a proton-conductive membrane, namely, reasonable swelling, good mechanical, dimensional, and oxidative strength, flexibility, and low methanol permeability along with good proton conductivity due to zwitterionic functionality. Moreover, Si–70% and a Nafion117 membrane exhibited comparable DMFC performance. Also, investigation on a multi-ionic organic–inorganic hybrid ZIM as polymer electrolyte membranes (PEMs) will give rise to a new developing field in materials and membrane science.

---

Introduction

PEMs with high conductivity, low methanol crossover and cost, are greatly desired for direct methanol fuel cells (DMFCs) in order to reduce ohmic losses and enhance their efficiencies during operation.^1–4Perfluorosulfonic acid membrane (Nafion) is a reference membrane for DMFC because of its high electrochemical properties as well as excellent chemical resistance.^2,5Nafion membranes show high methanol crossover, and become dry under conditions of high temperature (above 80 °C) or low humidity rather quickly due to the loss of water from the membrane.^5–13 Thus, widespread efforts were dedicated to develop inorganic–organic composites based on a modified Nafion membrane.^14–17PEMs with high proton conductivity at intermediate temperatures under anhydrous or low-humidity conditions, environmental affability with low methanol crossover, and production cost have attracted much interest recently for problem solving in current technologies.^7,18–22 For the development of cheaper PEMs, fluorine-free materials with properties comparable to those of Nafion, based on sulfonated aromatic polymers, irradiation graft polymers, and cross-linked and blend polymers, were successfully proposed.^23–29

Only a few reports are available for ZI based hybrid nanostructured PEMs.^30–35 The acid–base composite PEMs with high proton conductivity under anhydrous conditions, such as poly(benzimidazole)phosphoric acid or sulfuric acid,^36,37 poly(vinylphosphonic acid) heterocycle,³⁸ and ionic liquids,³⁹ were reported. However, the proton-conductive pathway can be controlled by a suitable molecular assembly between acidic and basic moieties by introducing these functional groups in the same molecule for high proton conductivity, water retention, and low methanol permeability.^40–42

For developing ZIMs, organic–inorganic hybrid materials are a suitable option, because of the synergistic advantages of organic and inorganic segments (e.g., structural durability, dielectric, ductility, processability, thermal and mechanical stability).^43–45 To enhance strength and compatibility, preparation methods for organic–inorganic hybrid functional materials were explored over the last decade.^46,47 Incorporation of an inorganic ion exchange filler into an organic matrix showed the leaching out due to the lack of chemical bonding between two segments.^47–49 In another sol–gel process of alkoxysilane-functionalized polymers followed by cross-linking with conventional agents, (aldehydes and small alkoxysilanes), a low content of organic groups and lack of any functional group are disadvantageous for homogenous and conductive PEMs.⁵⁰ To overcome these problems earlier we reported 3-(3-(triethoxysilyl)propylamino)propane-1-sulfonic acid, a zwitterionomer cross-linking agent, which contained one number of each sulfonic acid and amine groups per molecule.⁵¹ In our continuous effort, we tried to improve numbers of functional groups (sulfonic acid and amine groups) on the zwitterionomer for achieving highly stable and conductive ZIMs with appreciable numbers of ion-exchange sites in the membrane matrix.

Herein, a multi-ionic silicon based copolymer (3,3-dimethoxy-7-(4-sulfonatobutyl)-2-oxa-7,10-diazonia-3-silatetradecane-7,10-diium-14-sulfonate) was prepared as an organic–inorganic hybrid zwitterionomer cross-linking agent using 3-(2-amino ethyl amino)propyl trimethoxysilane (AAPTMS) and 1,4-butanesultonevia a ring opening reaction. Interestingly, the structure of multi-ionic silicon based ZI cross-linking agent may be similar to vines from a bionic aspect. The flexible main chain is comparable with the stem of a vine, functional groups (sulfonic and amine) are like fruit, and –Si(OCH₃)₃groups look like branches. A multi-ionic ZI precursor with plenty of –Si(OCH₃)₃groups has the ability to cross-link with PVA in the presence of TEOS. Structural features of prepared ZIM (long main chain with many branched chains avoids organic–inorganic phase separation and enhances membrane flexibility; plenty of acid and basic groups balance membrane fixed charge concentration and water retention ability) are expected to result in a highly stable and conductive PEM.

Experimental section

Materials

AAPTMS, TEOS, 1,4-butanesultone (distilled under vacuum), and Nafion117 (perfluorinated membrane) were purchased from Sigma Aldrich Chemicals and were used as received. Tetrahydrofuran (THF; Qualigens Fine Chemicals, Mumbai, India) was distilled and kept dry over molecular sieves. PVA (MW, 125 000; degree of polymerization, 1700; degree of hydrolysis, 88%), and all other chemicals, reported in this manuscript were obtained from SD Fine Chemicals (Mumbai, India) of AR grade and used without further purification. In all experiments, double-distilled water was used.

Preparation of zwitterionomer

Zwitterionomer precursor, 3,3-dimethoxy-7(4-sulfonatobutyl)-2-oxa-7,10-diazonia-3-silatetradecane-7,10-diium-14-sulfonate, was synthesized from 1,4-butanesultone and AAPTMS. In a general synthesis method, AAPTMS in THF was stirred under a nitrogen atmosphere, followed by the drop wise addition of 1,4-butanesultone, (dissolved in THF) at 60 °C. The reaction mixture was refluxed at 50 °C under stirring in a nitrogen atmosphere for 1 h, and then the solvent was evaporated. The obtained solid mass was dried in a vacuum oven at 40 °C for 24 h, and yielded a dark yellow coloured solid product.

CHNS: calcd (C, 38.85; H, 7.74; N, 5.66; S, 12.96); obsd (C, 38.81; H, 7.70; N, 5.64; S, 12.93).

Preparation of a cross-linked organic–inorganic hybrid ZIM

PVA (10 wt%) was dissolved in hot deionized water under constant stirring. An appropriate amount of synthesized ZI precursor was dissolved in deionized water at pH∼2, separately, and mixed with PVA solution, in the presence of a fixed amount of TEOS (20 wt% to PVA). The obtained solution was stirred for 8 h at room temperature. The sol–gel process was achieved by acid hydrolysis (pH: 2), and obtained gel was transformed as thin film on a cleaned glass plate. Film was dried in ambient conditions for 24 h and further at 70 °C for 12 h under IR lamp. The obtained transparent membrane was peeled off for further cross-linking with a formal solution (HCHO + H₂SO₄) for 3 h at 60 °C. Prepared membranes were designated as Si–X, where X is the wt% of zwitterionomer in the membrane matrix, and varied between 50–70 wt% of PVA content.

Fourier transform infrared (FTIR) and NMR characterization

FTIR spectra of dried membrane samples were obtained using an attenuated total reflectance (ATR) technique with a Spectrum GX series 49387 spectrometer in the range 4000–600 cm⁻¹. The IR spectrum for a synthesized zwitterionomer was obtained by the KBr pellet method. ¹H NMR spectra was used to characterize the synthesized zwitterionic material recorded by FT NMR (Brüker, 200 MHz) Brucker DPX-200 in a d₆-DMF solvent.

Thermal and mechanical strength analysis

Thermal stabilities of the prepared zwitterionomer membranes were obtained by thermo-gravimetric analysis (Mettler Toledo TGA/SDTA851^e with Star^e software) under nitrogen atmosphere with 10 °C min⁻¹ heating rate between 50 to 600 °C. Differential scanning calorimetry (DSC) measurements were achieved by Mettler Toledo DSC822^e thermal analyzer with Star^e software. The dynamic mechanical strength of zwitterionomer membranes were analysed by Mettler Toledo dynamic mechanical analyzer (DMA) 861^c instrument with Star^e software under nitrogen atmosphere with 10 °C min⁻¹ heating rate between 30 to 300 °C. Details about the estimation of cross-linking density is given in the ESI,† Fig. S1.

Microscopic characterization

Silica distribution in the membrane phase was studied by a JEOL 1200EX transmission electron microscope (TEM). The JEOL 1200EX transmission electron microscope with a tungsten electron source operated at an accelerating voltage up to 120 kV. Scanning electron microscopy (SEM) images were recorded by Leo microscope (Kowloon, Hong Kong) after gold sputter coatings on dried membrane samples (1 and 0.1 Pa). Composition of silica and other elements, were detected by energy-dispersive X-ray (EDX) measurements, using a LEO VP1430 and an Oxford Instruments (Oxfordshire, UK) INCA.

Ion-exchange capacity (IEC)

Detailed methods for estimation of membrane IEC is included in the ESI,† Fig. S2.

Water uptake, state of water, and membrane stabilities

Detail about methodology used for estimation of water uptake and state of water is given in the ESI†(Fig. S3). Methods used for dimensional, oxidative and hydrolytic stabilities studies of developed membranes are included in the ESI† (Fig. S4).

Membrane conductivity studies

Conductivity of ZIMs (at different relative humidity (RH)) was measured by four-probeAc impedance spectroscopy using a potentiostat/galvanostat frequency response analyzer (Auto Lab, Model PGSTAT 30). Prior to measurement, the membrane samples were soaked in deionized (DI) water for 24 h and rinsed repeatedly to remove the last trace of free acid or base. The membranes were mounted between two platinum electrodes (4.0 cm²), which were then placed in DI water to ensure 100% RH. Direct current (DC) and sinusoidal alternating currents (AC) were supplied to the respective electrodes for recording the frequency at a scanning rate of 1 μA s⁻¹ within a frequency range 10⁶ to 1 Hz. Membrane conductivity (κ^m) was estimated from the following equation.

Where d is the distance between electrodes, A is the cross-sectional area of membrane and R is the membrane resistance obtained from AC impedance data (the real part of the impedance at high frequency, i.e. intercept of the impedance semicircle with the Z-axis on the complex plane).

Electro-osmotic permeability measurements

Electro-osmotic permeability for different ZIMs was measured in a two-compartment membrane cell (effective membrane area: 20.0 cm²), in equilibration of 0.01 M HCl solutions. Both the compartments were kept under constant agitation by means of a mechanical stirrer. A known potential was applied across the membrane using Ag/AgCl electrodes, and subsequently volume flux was measured by observing the movement of liquid in a horizontally fixed capillary tube of known radius. The current flowing through the system was also recorded with the help of a digital multimeter. Several experiments were performed to obtain reproducible values.

Methanol permeability

The methanol permeability of the composite membranes was determined in a diaphragm diffusion cell, consisting of two compartments (50 cm³) separated by a vertical membrane with a 20 cm² effective area. The membrane was clamped between both compartments, which were stirred during the experiments. Before the experiment, the membranes were equilibrated in a water–methanol solution for 12 h. Initially, one compartment (A) contained a 30 or 50% (v/v) methanol–water mixture and the other (B) contained double distilled water. Methanol flux arises across the membrane as a result of the concentration difference between the two compartments. The increase in the methanol concentration with time in compartment B was monitored by measuring the refractive index using a digital refractometer (Mettler Toledo RE40D refractometer). The methanol permeability (P) was finally obtained by the equation:where A is the effective membrane area, l the thickness of the membrane, C_B(t) the methanol concentration in compartment B at time t, C_A(t − t₀) the change in the methanol concentration in compartment A between time 0 and t, and V_B the volume of compartment B. All experiments were carried out at room temperature, and the uncertainty of the measured values was less than 2%.

Membrane electrode assembly (MEA) and DMFC performance

Gas diffusion electrodes (three-layer structure) were prepared by: (i) wet proofing of carbon paper (Toray Carbon Paper, thickness: 0.27 mm, wet proofed with 15 wt% PTFE solution by brush painting method); (ii) coating of gas diffusion layer (GDL) on to the carbon paper; (iii) coating of catalyst layer on to the GDL.²⁷ The GDL (25 cm²) was fabricated by coating of slurry (0.95 mg cm⁻²) containing carbon black (Vulcan XC72R) and PTFE dispersion on carbon paper. The anode was made by coating slurry consisting catalyst (20 wt% Pt + 10 wt% Ru on carbon), 5 wt% Nafion ionomer solution, isopropanol, and Millipore water (catalyst ink) on GDL had a loading of 1 mg Pt and 0.5 mg Ru. The cathode was obtained by coating the Pt catalyst ink with the same loading. Electrodes were cold pressed with the membrane and cured at 60 °C for 12 h, followed by hot pressing at 130 °C for 3 min at 1.2 MPa. Obtained MEA was clamped in single cell (FC25-01 DM fuel cell).

The current–voltage polarization curves were recorded with the help of MTS-150 manual fuel cell test station (ElectroChem Inc., USA) with controlled fuel flow, pressure and temperature regulation attached with electronic load control ECL-150 (ElectroChem Inc., USA). The measurements were performed in the air mode of operation at 10 psi pressure with a fixed concentration of methanol fed at the anode side with pressure 7 psi at 70 °C for a representative membrane.

Results and discussion

Synthesis of multi-ionic silicon based copolymer and ZIM

Multi-ionic ZI silica precursor was synthesized by 3-(2-amino ethyl amino) propyl trimethoxysilane (AAPTMS) and 1,4-butanesultonevia a ring opening reaction under heating conditions, Fig. 1. Because of the exothermic reaction, temperature was carefully monitored because of the volatile nature of the solvent (THF). Bands observed at 1039 cm⁻¹ in the FTIR spectrum confirmed sulfonate groups, while hydrated sulfonic acid groups in multi-ionic ZI silica precursor absorbed at 1471–1163 cm⁻¹ (Fig. 2). This indicates that both functional groups were present in the ZI silica precursor.^52,53 A sharp peak observed at 1643 cm⁻¹ and a broad absorbance between 3025–2500 cm⁻¹, also confirmed the presence of a substituted quaternary ammonium group. The ZI silica precursor shows strong intensity bands at 1211 cm⁻¹ which confirmed the presence of –SiOR groups. The ZI silica precursor was zwitterionic in nature confirmed by the presence of two sharp peaks at 604 and 525 cm⁻¹. The ZI silica precursor was characterized by ¹H-NMR in D₂O solution as shown in Fig. 3. A triplet peak observed ca. 4.5 ppm indicates that the butanesultone ring was opened and formed sulphonic acid groups. Without ring opening this peak was not observed. The other proton signals are clearly represented in Fig. 3. The ZI silica precursor was further characterized by ¹³C NMR in D₂O solution and signals were observed at 10.56, 13.37, 21.01, 23.72, 24.23, 26.41, 31.68, 32.34, 48.38, 50.18, 51.63, 55.45, 60.59, 62.07 ppm and shown in (Fig. 4A). X-Ray diffraction pattern data show that the ZI silica precursor is an amorphous powder shown in (Fig. 4B). On the basis of spectral and elemental analysis data the structure of ZI silica precursor was confirmed and is shown in Fig. 1.

Fig. 1 Scheme for the synthesis of ZI silica cross-linking agent and its structural similarity from a bionic aspect.

Fig. 2 FTIR spectra for a multi-ionic ZI silica precursor.

Fig. 3 ¹H NMR of a multi-ionic ZI silica precursor.

Fig. 4 Multi-ionic ZI silica precursor (A) ¹³C NMR in d₆-DMSO (B) XRD.

ZIMs were prepared by condensation polymerization of a ZI silica precursor by an acid catalyzed sol–gel process in aqueous media using PVA (Fig. 1). The obtained gel was transformed into a thin film and cross-linked by HCHO in the presence of acid at 60 °C for 3 h. Cross-linking occurs in two steps: (i) formation of hemiacetal by reacting HCHO with –OH groups of PVA; (ii) further reaction of hemiacetal with the –OH of PVA and formation of acetal. Membranes lose their transparent nature in wet conditions after cross-linking, but retain it in dry conditions. The stable nature of ZIMs was achieved by molecular level tailoring of the sol–gel process, in which organic and inorganic segments were joined covalently.

The silica precursor and water forms a single phase solution in the acid catalyzed sol–gel process, and solution-to-gel transition was responsible for the silica (SiO₂) network attached to the organic matrix. The reaction mechanism proceeds through bimolecular nucleophilic substitution and the nature of the catalyst affects the membrane properties. Because of rapid protonation of OR or OH substituents, directly attached to the Si atom, sufficient numbers of interconnected Si–O–Si bonds formed a three-dimensional cluster or a gel.⁵⁴ Linear and weakly cross-linked polymer clusters formed in acid catalyzed sol–gel reaction because of steric crowding. The obtained membrane showed good mechanical, thermal and electrochemical properties. Presence of –SO₃H groups was confirmed because of a sharp intensity absorption band near 1100–1037 cm⁻¹ and multi headed bands in the region of 1460–1360 cm⁻¹ (Fig. 5). Quaternary ammonium groups in ZIM were confirmed by peaks between 1640–1540 and 3000–2800 cm⁻¹. A medium intensity broad peak between 3400–3300 cm⁻¹ indicated the presence of quaternary ammonium salt in the ZIM phase.⁵³ Formation of a cyclodiether part (–C–O–C–) and cross-linked membrane structure was confirmed by an absorption band between 1170–1030 cm⁻¹. Absorption band for about 1040–1150 cm⁻¹ confirmed Si–O–Si and Si–O–C groups in the membrane matrix, as a result of condensation reaction between hydrolyzed silanol (SiOH) groups.^55,56

Fig. 5 ATR-FTIR spectrum for Si–70% ZIM.

The structural features for multi-ionic ZI silica precursors are similar to a vine from a bionic aspect. The main chains are comparable to the stems of vine, while branched chains are like branches of a vine, and are beneficial for the flexibility of the cross-linked ZIMs. Multi-ionic ZI silica precursors contain plenty of acidic and basic functional groups (–SO₃H and –N⁺(CH₃)₃Cl⁻). Functional groups were comparable to the fruits of vine with controllable charge density, while –Si(OCH₃)₃groups were similar to acetabulas of vines with high cross-linking abilities.

Microscopic characterizations

High contents of organic moiety (PVA) and a relatively larger amino propyl group were responsible for the formation of the worm-like structure. In case of the membrane, 50 nm fine slices were obtained with the help of a microtome cutter. The TEM images show a worm like arrangement in membrane matrix and silica particles are equally distributed in the membrane shown in Fig. 6. SEM images of representative ZIMs (Si–50% and Si–70%) are present in the ESI† (Fig. S1). Excellent compatibility between organic and inorganic phases suggests that the multi ionic ZI silica precursor is an effective cross-linking agent for PVA, because cross-linking occurred through covalent and hydrogen bonds. Some aggregations on the membrane surface were observed because accelerated hydrolysis of silica precursor, with increase in ZI silica precursor content. Absence of phase separation, cracks or holes on the membrane surface suggested a homogeneous and dense nature of ZIM.

Fig. 6 TEM image of a ZI-60 membrane (a) low magnification (b) high magnification.

Thermal and mechanical stabilities

TGA curves for ZIMs in the H⁺ state show similar weight loss patterns (Fig. S2, ESI†). Three weight loss steps attributed to water loss (loose and bound) at 50–160 °C (5 to 8% of the initial weight), defunctionalisation at 260–370 °C (12–18% of initial weight), and membrane matrix degradation beyond 400 °C. Si–70% membrane showed comparatively minimum weight loss, for which a high degree of cross-linking was responsible. Furthermore, high thermal stabilities of developed ZIMs indicate advantages of covalently bonded functional groups and cross-linking.

DSC thermogram for ZIMs (Fig. S3, ESI†) showed 118.35, 117.78, 109.59 and 107.81 °C first endothermic peaks (T_g) and second endothermic peaks at 174.1, 176.2, 179.4 and 231.4 °C for Si–50%, Si–55%, Si–60%, Si–70% membranes, respectively. Incorporation of a multi-ionic ZI silica precursor in the polymer matrix had a profound effect on T_g values. Variation in T_g values may be explained because of a plasticizing effect of the binder (PVA) and alteration in its ordered arrangement with an increase in ZI silica precursor content in the membrane matrix. This observation indicates a highly cross-linked structure for ZIM at elevated temperature. Here, it is interesting to note that T_g for pristine PVA is 78 °C.⁵⁷

Mechanical properties of the membranes (modulus and strength) were analyzed by DMA curves (Fig. S4, ESI†). Elongation of ZIMs increased with ZI silica precursor content, and Si–50% showed maximum stress tolerance which may be explained in terms of the enhanced cross-linked nature of Si–50% membrane. Hydroxyl groups of PVA contributed towards hydrogen bonding and stiffness of the polymer chain. With the decrease in number of hydroxyl groups (because of its involvement in either branching or cross- linking), the hydrogen bonding was attenuated and thus the chain stiffness was reduced. In addition, SiOH groups formed inter and intra molecular cross-linked structures with PVA and a high content of ZI silica precursor in polymer matrix produced agglomerates. Cross-linking density, determined by DMA studies (Fig. 4S, ESI†), and decreased with ZI silica content in the membrane matrix. Variation in cross-linking density may be explained by formation of cohesive domains in the ZIM matrix. This seems to be more predominant than cross-linking with a plasticizer. Thus it is necessary to optimize the ZI silica content in the membrane matrix and cross-linking density for achieving better mechanically stable ZIMs.

Oxidative, hydrolytic and acid–base stabilities

Stability and durability for membranes under stressed conditions are important to explore their practical applications in electrochemical processes. Membrane oxidative stability was evaluated by its treatment in Fenton's reagent (3% aq. H₂O₂ + 3 ppm FeSO₄) at 80 °C for 1 h and recording oxidative weight loss percent (W_ox) (Table 1). The Si–50% membrane exhibited the lowest weight loss, which increased with ZI silica content in the membrane matrix. Because of cross-linking, hydrophilic domains turned compact and ˙OH and ˙OOH radicals (short life time) cannot penetrate inside the siloxane containing domains. Thus, Si–70% membrane with low cross-linking density exhibited high weight loss. Methylene groups in PVA are sensitive to free radical attack and cause chain degradation. ZIMs were also subjected to an accelerated hydrolytic stability test at 120 °C and 100% RH for 24 h. Hydrolytic stability test results showed increased weight loss with a ZI silica precursor (Table 1) and membranes retained their transparency, flexibility and toughness.

Table 1 Water uptake^a (ϕ_w), number of water molecules per ionic site (λ_w), swelling in water (Φ_v), oxidative, hydrolytic weight loss^b (W_OX & W_HS respectively), and water diffusion coefficient (D) values for different ZIMs

Membrane code	ϕ w (wt%)	λ w /SO3^−	Φ v (vol%)	W OX (wt%)	W HS (wt%)	D/10^−6 (cm^2 s^−1)
a Uncertainty in the measurements of 1 : 0.1%. b Uncertainty in the measurements of 2 : 0.01 mg.
Si–50%	28.0	20.9	26.0	5.3	5.9	3.26
Si–55%	31.4	17.8	28.0	7.8	8.4	2.42
Si–60%	35.9	16.5	32.3	8.3	9.5	1.87
Si–70%	40.6	14.9	34.9	9.3	10.9	1.31

---

Acid–base stability for developed ZIMs were tested for 5 h equilibration in acid (HCl) and base (NaOH) solutions of different concentrations and resulting weight loss data are presented in (Fig. 5S, ESI†). Data confirmed the stabilities of ZIMs in 5 M HCl and 7 M NaOH, as weight loss under these conditions was less than 3%. Weight loss for the Si–70% membrane was comparatively high under acidic and basic conditions, and may be because of more ZI content and thus a posses a high level of functionality.

Stability of ZIMs (containing sulfonic acid and amine groups) in strong alkaline solution is highly concerning,⁵⁸ because quaternary ammonium groups are relatively unstable in alkaline solutions due to direct nucleophilic substitution and Hofmann elimination.⁵⁹ Prepared ZIMs were immersed in 5 mol L⁻¹NaOH solution for 100 h to investigate stability of alkaline functional groups. Ratios of IEC_t (IEC of base treated ZIM at time t) and IEC₀ (IEC of untreated ZIM) were recorded, Fig. 7. All lines drop rapidly in 0–25 h (first stage), then attain a level in 0–100 h (second stage). It seems ZIMs contain some quaternary ammonium groups, which undergo Hofmann elimination under the treatment with a strong base. Estimated from IEC₀ and IEC_t at 25 h, the percentage of unstable quaternary ammonium groups was found to be 15.4%, 10.6%, and 2.9% for Si–70%, Si–60%, and Si–50%, respectively. Presence of this group may be attributed to insufficient cross-linking because of high swelling or water uptake (Si–70% membrane). After the degradation of these components, loss in membrane IEC was nearly constant. This study reveals the membrane stability under strong alkaline media against Hofmann degradation, and their suitability for electro-membrane processes. The formation of base in the cathode compartment occurs because of reductive water splitting.

Fig. 7 Alkaline resistance of different ZIMs: ratios IEC_t/IEC₀ over the immersion time in 5 mol L⁻¹NaOH at room temperature.

Water uptake, dimensional changes and water retention capability

Membrane water uptake has a profound effect on its proton conductivity, mechanical and dimensional properties.⁶⁰ Presence of water molecules in the membrane phase facilitates the dissociation of functional groups and is essential for high membrane conductivity. On the other hand, high water volume fraction in the membrane phase reduces its dimensional and thermal stabilities along with ionic concentration in the membrane phase. Water uptake values (ϕ_w) increased with ZI content for ZIMs (Table 1), because of the enhanced hydrophilic nature of the membrane matrix. Water uptake profiles for different ZIM at elevated temperature were also studied, and increased linearly up to 80 °C. In this case, the extent of plasticization and hydrophilic nature of the matrix play an important role for water uptake. At low extent of plasticization (high ZI content) the flexible polymer network permits larger water uptake.⁶¹

Membrane water uptake values significantly depend upon the void porosity of the membrane. Fig. 6S, ESI† shows the effect of void porosity on the water uptake properties. The results revealed that with the increase in membrane void porosity water uptake values increase. Total number of water molecules per ionic site (λ_w) (Table 1) data also support a highly cross-linked and hydrophilic membrane matrix at high ZI content. The lowest number of water molecules per ionic site for Si–70% membrane revealed its highly cross-linked structure with small void volume.

The water vapor sorption and water diffusion properties of ZIMs affect their conductivity and applicability for electrochemical processes. Furthermore, the water retention capability of membranes is helpful to assess their suitability for desired applications. Water retention capability of ZIMs was illustrated in (Fig. 8A) (M_t/M_o)–t (time) curves, and deswelling kinetics of the developed membranes was calculated. The value of k (constant) was derived from (M_t/M_o)–t^1/2 curves (Fig. 8B) based on Higuchi's model for water desorption kinetics:

where M₀ and M_t are the initial amount of water and remaining water in polymer matrix at given time and κ is a constant.

Fig. 8 Water desorption profile for ZIMs: (A) isotherm at 40 °C; (B) Higuchi's model fit of the deswelling behavior.

The obtained straight lines for different ZI contents were fitted to Higuchi's model, and suggested a diffusion controlled water desorption mechanism. The water desorption rate was reduced with ZI content, thus ZI acted as a water binder in the membrane matrix and is responsible for formation of hydrophilic channels and cross-linked chains. Water diffusion coefficients (D) across ZIMs were also evaluated from a best-fit normalized mass change (Table 1) and were found to decrease significantly with ZI content in the membrane matrix. Thus ZI acted as a water release barrier and enhanced the membranewater retention capacity.

State of water

In a membrane matrix, the presence of water may be classified into three types: (i) free water (with the same temperature and enthalpy of melting as bulk water); (ii) freezing bound water (weakly bound with polar or ionic groups of the polymer and shows change in temperature and enthalpy in comparison with bulk water and can be detected by DSC); and (iii) non-freezing bound water (very strong interaction with polar or ionic groups and shows no phase transition). According to Eikerlings theory,⁶² the charged membrane possessed two types of water: bound and bulk water. Bound water is necessary for the solvation of ionic groups, whereas bulk water fills the void volume. Low temperature DSC studies and water uptake values were used to determine the state of water in the membrane phase. DSC thermograms were recorded from −50 to +50 °C (Fig. 3S, ESI†) as a broad endothermic peak of hydrated ZIMs. Enthalpy of melting (ΔH_m), melting temperature (T_m), the full width at half-maximum of the melting peak (ΔT_m), freezing water (λ_f), bound water (λ_b) and bound water degree (χ) were estimated from DSC curves and presented in Table 2. Increase in ΔH_m with ZI content was attributed to low freezing and less bound water percentage of the ZIMs.

Table 2 State of water: enthalpy of melting (ΔH_m), melting temperature (T_m), full width at half-maximum of the melting peak (ΔT_m), glass transition temperature (T_g), total number of water molecules per ionic site (λ_t), number of free water molecules per ionic site (λ_f), number of bound water molecules per ionic site (λ_b), and degree of bound water (χ) for different ZIMs

Membrane	ΔHm (J g^−1)	T m ^a(°C)	ΔTm^b(°C)	T g	λ t	λ f	λ b	χ ^c (%)
a Melting temperature of free and loosely bound water. b Full width at half-maximum of the melting peak. c Bound water degree χ (%) = λb/λt .
Si–50	5.35	−5.3	9.99	118.35	20.9	1.20	19.70	94.26
Si–55	7.81	−4.1	12.33	117.78	17.8	1.32	16.48	92.58
Si–60	10.74	−3.4	14.1	109.59	16.5	1.48	15.02	91.14
Si–70	14.72	−1.1	14.8	107.81	14.9	1.62	13.28	89.73

---

Freezing water (λ_f) for ZIM was obtained by total melting enthalpy by integrating melting curves peak area (Fig. S3, ESI†). Also, bound water percentage (λ_b) was obtained by subtracting freezing water content from total water in the membrane matrix. The degree of bound water in percentage (χ = λ_b/λ_w) was estimated by the ratio of number of bound water molecules and total water. Free water increased with ZI content in the membrane matrix, while bound water reduced because of the enhanced availability of ionic sites for binding. Reduction in bound water with ZI content may be attributed to strong interactions between ionic groups and water molecules which effectively predominates over the siloxane network.

IEC and surface charge concentration

IEC indicates density of dissociable ionic groups in the membrane matrix, which are responsible for ion conduction across the membrane. ZIM contains both types of functional groups: acidic (–SO₃⁻H⁺) and basic (quaternary ammonium), which contribute towards IEC. Acidic and basic IEC values for different ZIMs were estimated by the titration method (Table 3). Both types of IEC increased with ZI content in the membrane matrix, and confirmed the good zwitterionomer nature of the membrane. Here, it is interesting to note that the Si–70% membrane showed 1.512 m equiv./gm IEC for acidic functional groups (–SO₃⁻H⁺), while world-wide used acidic Nafion-117 membrane showed 0.910 m equiv./gm, IEC. Similarly, Si–70% membrane showed 0.827 m equiv./gm alkaline IEC, whereas Neosepta membrane showed 1.20–1.40 m equiv./gm alkaline IEC.⁶²IEC studies confirmed the highly charged and zwitterionic nature of developed ZIMs.

Table 3 Acidic and basic IECs by titration methods (IEC_tit), and their surface charge concentration (X^m) for different ZIMs

Membrane Code	Acidic functional groups		Basic functional groups
	IECtit^a (m equiv./gm)	X ^m (m mol dm^−3)	IECtit^a (m equiv./gm)	X ^m (m mol dm^−3)
a Uncertainty in the measurement was: 0.001 m equiv./g.
Si–50%	0.743	0.931	0.424	0.531
Si–55%	0.983	1.222	0.568	0.706
Si–60%	1.212	1.606	0.734	0.972
Si–70%	1.512	2.004	0.827	1.096

---

Surface charge concentrations (X^m) for different ZIMs were estimated by the following equation.

Where τ is membrane void porosity, ρ_d is the density of the dry membrane, and ϕ_w is the volume fraction of water in the membrane matrix. From acidic and basic IEC values, X^m values for acidic and basic functional groups for different ZIMs were also included in Table 3. Data support the schematic structure of ZIM (Fig. 1) and confirmed the presence of strongly dissociable functional groups (sulfonic acid and quaternary ammonium) in the membrane matrix.

Conductivity of ZIMs

Conductivity of ZIMs (κ^m) was measured under 100% relative humidity at 30 °C (Table 4), increased with ZI content in the membrane matrix and thus has a high concentration of functional groups. With enhanced ZI content in the membrane matrix, high ϕ_w values affect hydrogen bonding and thus membrane conductivity, because of increased surface charge concentration. Results support that charged sites of ZIMs enhanced the hydrophilic nature of the matrix and promoted exchange of ions, which helped for conduction. Fig. 9 shows the variation of κ^m with enthalpy of melting (ΔH_m) in equilibration with 0.5 M NaCl solution. ΔH_m depends on the free water content in the membrane matrix responsible for membrane conductivity. High extent of free water in the membrane phase provides easy conduction of counter-ions. Thus, for highly conductive membrane, designing a precursor with high extent of free water is necessary. The designed ZI precursor contains two acidic and basic functional groups in a molecule and provides a highly charged matrix with appreciable free water content. Knowledge of a membrane's water retention capacity is also helpful to assess the membrane conductivity at high temperature.

Fig. 9 Variation of κ^m with ΔH_m in equilibration with 0.5 M NaCl solution for different ZIMs and (inset) variation in membrane conductivity with different loading of zwitterionomer.

Table 4 Membrane conductivity^a (κ^m), electroosmotic permeability (J_E), and activation energy (E_a), for different ZIMs

Membrane code	κ ^m/10^−2 (S cm^−1)^a	J E/10^−5 (cm s^−1A^−1)^b	E a (kJ/mol)	r (nm)
a Uncertainty in the measurement was: 0.01 × 10^−2 S cm^−1. b Uncertainty in the measurement was: 0.01 × 10^−5 cm s^−1A^−1.
Si–50%	5.97	1.07	3.64	15.33
Si–55%	7.01	1.16	3.79	16.88
Si–60%	8.42	1.34	4.04	18.74
Si–70%	9.67	2.34	4.29	19.23
Nafion 117	9.56	14.01	6.52	—

---

Membrane conductivity values at high temperatures (30 to 100 °C) under 100% RH are presented as an Arrhenius plot (Fig. S5, ESI†) for the estimation of activation energy (E_a), using the following expression,

Where R is the universal gas constant (8.314 J mol⁻¹ K⁻¹), and T is the absolute temperature. E_a values increased with the ZI content in the membrane matrix (Table 4), and varied between 3.64–4.29 kJ mol⁻¹. As a reference Nafion 117 membrane showed 6.52 kJ mol⁻¹activation energy. It seems ion transport across ZIMs followed a conduction mechanism similar to Nafion 117 (Grotthus type conduction mechanism).

Membrane conductivity (Si–50% and Si–70%) depends on its relative humidity (Fig. 10). Conductivity decreased with a decrease in the RH (%) of membranes. At 100% RH, the Si–70% membrane showed 0.0967 S cm⁻¹ conductivity, which was reduced to 0.043 S cm⁻¹ at 25% RH. Furthermore, these ZIMs (especially Si–70%) retained conductivity under relatively low hydrated conditions.

Fig. 10 Conductivity of ZIMs (Si–70% and Si–50%) at different RH.

Electroosmotic permeability studies

Electro-osmotic transport of mass (solvent) across charged membranes revealed their mass electro-driven mass transport properties and equivalent pore radius. Electro-osmotic flux across ion exchange membranes occurred due to: (i) the availability of ionic sites in the membrane matrix and (ii) the existence of an electrical potential at the membrane–solution interface called zeta potential.^63,64 Knowledge of ion and solvent transport rates across the membranes under electro-driven conditions is necessary for intelligently designing a better membrane for desired electro-membrane processes.⁶⁵ Electro-osmotic flux across ZIMs in equilibration with 0.02 M NaCl solution (Fig. S6, ESI†) was used for the estimation of electro-osmotic drag (β) (implies that every coulomb of electricity will exert a drag sufficient to carry β cm³ of water through 1 cm² of the membrane area). Equivalent pore radius (r) for membranes was estimated from the Katchalsky and Curran approach:⁶³where F is the Faraday constant, η denotes the coefficient of viscosity of the permeate, and f⁰_1w is the frictional coefficient between the counter-ion and water in free solution, which can be defined as f⁰_1w = RT/D_i, (where D_i is the diffusion coefficient of the single ion (i) in the free solution, R is the gas constant, and T is absolute temperature). The ionic diffusion coefficient (D_i) at a given electrolyte concentration was obtained from ionic conductivity data.⁵⁷ Equivalent pore radii (r) for ZIMs suggest their dense nature and increased with ZI content in the membrane matrix (Table 4). This information also supports our earlier observations that in ZIMs, hydrophilic, charged and free water content increased with ZI content. Thus it is necessary to optimize the ZI content for desired electro-membrane application for ZIMs.

Methanol permeability and selectivity parameter

ZIMs showed extremely low methanol permeability transmission [(0.95–2.31) × 10⁻⁷ cm² s⁻¹] compared with the Nafion117 membrane (13.10 × 10⁻⁷ cm² s⁻¹) (Fig. 11(A)). The mass-transport behavior for a hydrated ZIM depends on its degree of swelling, water uptake, and bulk microstructure. Methanol permeability values increased with ZI content in the membrane matrix. Incorporation of silica containing ZI precursor into the polymer membranes dramatically altered the membrane transport properties because of alteration in the free void volume and ionic clusters. Thus, understanding the relationship between the polymer structure and membrane performance, in terms of permeability and selectivity, enables the tailoring of the membrane structure for specific purposes. Permeation of liquid/gas molecules through the polymer membrane occurs via the diffusion mechanism, and the permeability of the penetrant (methanol) is the product of its solubility and diffusivity. The penetrant diffusivity is dependent on the free void volume in the membrane, the size of the penetrant molecules, and the segmental mobility of the polymer chain.

Fig. 11 (A) Methanol permeability (P) and (B) selectivity parameter (SP) values as function of ZI content in the membrane matrix for different ZIMs and Nafion 117 membrane.

To directly compare the applicability of ZIMs for DMFC, the ratio of the proton conductivity and methanol permeability (κ^m/P) data was used as the selectivity parameter (SP). SP values for nanocomposite and N117 membranes are also presented in Fig. 11(B). The Si–50 membrane exhibited the highest SP value (6.28 × 10⁵ S cm⁻³ s) among the prepared ZIMs. The SP value decreased with ZI content in the membrane matrix. In similar conditions, the N117 membrane showed an SP value of 0.73 × 10⁵ S cm⁻³ s. It was also noticed that, with an increase in the operating temperature, SP values for ZIMs were also increased.

This observation may be attributed to the relatively low methanol permeability of ZIMs. These results can be explained on the basis of the lack of significant interactions between methanol and functional groups. Ionized groups hydrate strongly and excluded organic solvents (salting-out effect), which is an essential feature of the polyelectrolyte membranes. Furthermore, higher SP values of these membranes indicate a great advantage for DMFC applications.

DMFC performance

DMFC performance for representative PEMs (Si–50% and Si–70%) was tested in a single cell and compared with Nafion117 membrane at 70 °C (Fig. 12). The Si–70% membrane showed 0.95 V open circuit voltage (OCV), while Nafion exhibited 0.497 V, under similar experimental conditions. The current densities at a potential of 0.20 V for Si–70%, and Nafion117 (79.5 and 73.6 mA cm⁻², respectively) indicated a comparable DMFC performance of ZI and Nafion117 membranes. Good performance of the Si–70% membrane was observed due to extremely low methanol permeability in spite of its lower proton conductivity. Si–70% membrane showed 16.1 mW cm⁻² maximum power density, whereas for Nafion117showed 13.7 mW cm⁻². These results indicated suitability of ZIMs for DMFC applications.

Fig. 12 DMFC performance curves for MEAs made with different membranes operated at 343 K for 20% (v/v) methanol used as fuel in air mode with 10 psi pressure.

Conclusions

Multi-ionic silicon copolymervia ring opening reaction was developed as cross-linking agent for preparing stable zwitterionic organic–inorganic hybrid membranes. The prepared ZI silica precursors contain sulfonic acid and amine groups (two number each per molecule), which rendered its ZI nature because of proton transfer. The structure of the ZI silica precursor is similar to a vine from a bionic aspects with a long main chain, and branched chains with large numbers of sulfonic acid, amine and –Si(OCH₃)₃groups. All these are desired properties for designing a stable, homogeneous, and flexible ZIM by a sol–gel process in aqueous media using a suitable water soluble plasticizer (PVA).

Developed membranes showed good stability, flexibility, water uptake, and retention capacity. Incorporation of ZI in the membrane matrix improved the thermal stability of ZIMs, which were thermally stable up to 200 °C in a dry nitrogen atmosphere. Among the developed ZIMs, Si–70 exhibited a higher acidic IEC value (1.512 m equiv g⁻¹), water uptake (40.6%), and proton conductivity (9.67 × 10⁻² S cm⁻¹). Comparable conductivity for Si–70 membrane with Nafion 117 membrane, confirmed the suitability of ZIMs for fuel cell applications. Reported data are promising to design highly conducting acid–base composite systems. Furthermore, relatively lower methanol permeability and SP values of these membranes make them applicable for DMFC. The acid–base composite material may have potential applications not only for the DMFC operated at intermediate temperatures under anhydrous (water-free) or extremely low humidity conditions but also for novel electrochemical devices, where water activity is not required.

Acknowledgements

Financial assistance received from the Department of Science and Technology, New Delhi (Govt. of India), by sponsoring project no. SR/S1/PC/06/2008 is gratefully acknowledged. Instrumental support received from Analytical Science Division, CSMCRI, is also gratefully acknowledged.

References

1. A. Siu, J. Schmeisser and S. Holdcraft, J. Phys. Chem. B, 2006, 110, 6072.
2. F. Pereira, K. Valle, P. Belleville, A. Morin, S. Lambert and C. Sanchez, Chem. Mater., 2008, 20, 1710.
3. J. A. Kerres, J. Membr. Sci., 2001, 185, 3.
4. J. Y. Kim, W. C. Choi, S. I. Woo and W. H. Hong, J. Membr. Sci., 2004, 238, 213.
5. P. Costamagna and S. Srinivasan, J. Power Sources, 2001, 102, 242.
6. M. Khiterer, D. A. Loy, C. J. Cornelius, C. H. Fujimoto, J. H. Small, T. M. McIntire and K. J. Shea, Chem. Mater., 2006, 18, 3665.
7. M. A. Hickner, H. Ghassemi, Y. S. Kim, B. R. Einsla and J. E. McGrath, Chem. Rev., 2004, 104, 4587.
8. T. H Yu, Y. Sha, W. G. Liu, B. V. Merinov, P. Shirvanian and W. A. Goddard, J. Am. Chem. Soc., 2011, 133, 19857.
9. T. Prabhuram, T. S. Zhao, Z. X. Liang, H. Yang and C. W. Wong, J. Electrochem. Soc., 2005, 152, A1390.
10. M. S. Kang, J. H. Kim, J. Won, S. H. Moon and Y. S. Kang, J. Membr. Sci., 2005, 147, 127.
11. Q. Li, R. He, J. O. Jensen and N. Bjerrum, Chem. Mater., 2003, 15, 4896.
12. S. J. Paddison, Annu. Rev. Mater. Res., 2003, 33, 289.
13. B. J. Wang, H. Zhang, X. Yang, S. Jiang, W. Lu, Z. Jiang and S. Z. Qiao, Adv. Funct. Mater., 2011, 21, 971.
14. M. Watanabe, H. Uchida, Y. Seki, M. Emori and P. Stonehart, J. Electrochem. Soc., 1996, 143, 3847.
15. J. A. Asensio, E. M. Sánchez and P. Gómez-Romero, Chem. Soc. Rev., 2010, 39, 3210.
16. H. Ni'mah, W.-F. Chen, Y.-C. Shena and Ping-Lin Kuo, RSC Adv., 2011, 1, 968.
17. J. Yanga, P. K. Shena, J. Varcoeb and Z. Wei, J. Power Sources, 2009, 189, 1016.
18. B. P. Tripathi, M. Schieda, V. K. Shahi and S. P. Nunes, J. Power Sources, 2011, 196, 911.
19. V. K. Shahi, Solid State Ionics, 2007, 177, 3395.
20. B. P. Ladewig, R. B. Knott, A. J. Hill, J. D. Riches, J. W. White, D. J. Martin, J. C. Diniz da costa and G. Q. Lu, Chem. Mater., 2007, 19, 2372.
21. B. P. Tripathi and V. K. Shahi, J. Phys. Chem. B, 2008, 112, 15678.
22. B. Smitha, S. Sridhar and A. A. Khan, J. Power Sources, 2006, 152, 846.
23. L. C. Cogoa, M. V. Batisti, M. A. Pereira-da-Silva, F. C. Nart and F. Huguenin, J. Power Sources, 2006, 158, 160.
24. B. Smitha, S. Sridhar and A. A. Khan, Macromolecules, 2004, 37, 2233.
25. J. Wang, X. Zheng, H. Wu, B. Zheng, Z. Jiang, X. Haob and B. Wang, J. Power Sources, 2008, 178, 9.
26. Y. Wan, Y. Katherine, K. A. M. Creber, B. Peppley and V. Tam Bui, J. Membr. Sci., 2006, 280, 666.
27. A. Saxena, B. P. Tripathi and V. K. Shahi, J. Phys. Chem., 2007, 111, 12454.
28. Chenxi Xu, Yuancheng Cao, Ravi Kumar, Xu Wu, Xu Wang and Keith Scott, J. Mater. Chem., 2011, 21, 11359.
29. Q. Li, J. O. Jensen, R. F. Savinell and N. J. Bjerrum, Prog. Polym. Sci., 2009, 34(5), 449.
30. H. Zou, S. Wu and J. Shen, Chem. Rev., 2008, 108, 3893.
31. J. Liu, T. Xu and Y. Fu, J. Membr. Sci., 2005, 252, 165.
32. J. Liu, T. Xu, M. Gong and Y. Fu, J. Membr. Sci., 2005, 260, 26.
33. A. Narita, W. Shibayama, K. Sakamoto, T. Mizumo, N. Matsumi and H. Ohno, Chem. Commun., 2006, 1926.
34. H. S. Kim, S. J. Park, D. Q. Nguyen, J. Y. Bae, H. W. Bae, H. Lee, S. D. Lee and D. K. Choi, Green Chem., 2007, 9, 599.
35. P. Innocenzi and G. Brusatin, Chem. Mater., 2000, 12, 3726.
36. A. Schechter and R. F. Savinell, Solid State Ionics, 2002, 147, 181.
37. R. Bouchet and E. Siebert, Solid State Ionics, 1999, 118, 287.
38. H. G. Herz, K. D. Kreuer, J. Maier, G. Scharfenberger, M. F. H. Schuster and W. H. Meyer, Electrochim. Acta, 2003, 48, 2165.
39. A. N. Mondal, B. P. Tripathi and V. K. Shahi, J. Mater. Chem., 2011, 21, 4117.
40. M. Yamada and I. Honma, Chem. Phys. Lett., 2005, 402, 324.
41. S. R. Narayanan, S.-P. Yen, L. Liu and S. G. Greenbaum, J. Phys. Chem. B, 2006, 110, 3942.
42. M. Yamada and I. Honma, J. Phys. Chem. B, 2004, 108, 5522.
43. E. Markovic, S. Clarke, J. Matisons and G. P. Simon, Macromolecules, 2008, 41, 1685.
44. K. Tadanaga, H. Yoshida, A. Matsuda, M. Tsutomu and M. Tatsumisago, Chem. Mater., 2003, 15, 1910.
45. Q. G. Zhang, Q. L. Liu, A. M. Zhu, Y. Xiong and X. H. Zhang, J. Phys. Chem. B, 2008, 112, 16559.
46. P. G. Romero, Adv. Mater., 2001, 13, 163.
47. R. K. Nagarale, G. S. Gohil, V. K. Shahi and R. Rangarajan, Macromolecules, 2004, 37, 10023.
48. X. J. Cui, S. L. Zhong and H. Y. Wang, J. Power Sources, 2007, 173, 28.
49. V. V. Binsu, R. K. Nagarale and V. K. Shahi, J. Mater. Chem., 2005, 15, 4823.
50. I. Honma, H. Nakajima, O. Nishikawa, T. Sugimoto and S. Nomura, Solid State Ionics, 2003, 162–163, 237.
51. B. P. Tripathi and V. K. Shahi, ACS Appl. Mater. Interfaces, 2009, 1, 1002.
52. L. J. Bellami, The Infrared Spectrum of Complex Molecules, 2nd ed.; Wiley: New York, 1958.
53. G. Socrates, Infrared Characteristic Group Frequencies; Wiley: New York, 1980.
54. W. G. Klemperer, V. V. Mainz, S. D. Ramamurthi, C. J. Brinker, D. E. Clark, D. R. Ulrich (ed.)Ulrich, Better Ceramics Through Chemistry III, vol. 121, Materials Research Society, Pittsburgh, PA, 198815–25.
55. F. Rubio, J. Rubio and J. L. Oteo, Spectrosc. Lett., 1998, 31, 199.
56. P. Innocenzi, J. Non-Cryst. Solids, 2003, 316, 309.
57. M. A. Vargas, R. A. Vargas and B. E. Mellander, Electrochim. Acta, 2000, 45, 1399.
58. J. R. Varcoe, R. C. T. Slade, E. L. H. Yee, S. D. Poynton, D. J. Driscoll and D. C. Apperely, Chem. Mater., 2007, 19, 2686.
59. J. R. Varcoe and R. C. T. Slade, Fuel Cells, 2005, 5, 187.
60. D. B. Spry, A. Goun, K. Glusac, D. E. Moilanan and M. D. Fayer, J. Am. Chem. Soc., 2007, 129, 8122.
61. P. M. Mangiagli, C. S. Ewing, K. Xu, Q. Wang and M. A. Hickner, Fuel Cells, 2009, 9, 432.
62. M. Eikerling, A. A. Kornyshev and U. Stimming, J. Phys. Chem. B, 1997, 101, 10807.
63. A. Katchalsky, P. F. Curran, Nonequilibrium Thermodynamics in Biophysics, Harvard Univ. Press, Cambridge, 1965.
64. R. Parsons, Handbook of Electrochemical Constants, Butterworths, London, 1959.
65. N. Pismenskaya, E. Laktionov, V. Nikonenko, A. E. Attar, B. Auclair and G. Pourcelly, J. Membr. Sci., 2001, 181, 185.

---

Footnote

† Electronic Supplementary Information (ESI) available. See DOI: 10.1039/c1ra00228g/

---