Synthesis and characterization of novel imidazolium-functionalized polyimides for high temperature proton exchange membrane fuel cells

Abstract

Novel imidazolium-functionalized polyimides (ImPI-x) were successfully synthesized from polyimides containing trifluoromethyl groups, ether linkages and four phenyl substituents (4PhODA/PI) via chloromethylation followed by quaternization with 1-methylimidazole. The cleavage of the ether linkage and crosslinking during chloromethylation can be circumvented by carrying out this reaction at 60 °C with suitable concentrations of chloromethylation reagents, catalyst and polyimides. The degree of substitution (DS) ranging from 0.15 to 2.17 per repeating unit can be achieved without polymer degradation and crosslinking. Phosphoric acid (PA) uptakes of ImPI-x ranged from 34 to 159% and increased with the increased DS values. The ImPI-x membranes also exhibited good thermal stability and mechanical properties in both their dry and PA doped states. The proton conductivity of the ImPI-x membranes with PA uptakes of 84–159% were from 0.008 to 0.057 S cm⁻¹ at 160 °C under anhydrous conditions. ImPI-1.51 had a higher proton conductivity of 0.057 S cm⁻¹ than m-PBI (0.046 S cm⁻¹) even though it had a lower PA uptake (159%). A single fuel cell based on the ImPI-1.51 membrane with a PA uptake of 159% exhibited the peak power density of 551 mW cm⁻² with H₂/O₂ under anhydrous conditions at 160 °C, which was higher than that of m-PBI (419 mW cm⁻²) with a PA uptake of 216%. From AFM phase images of ImPI-x, microphase separation which might have resulted from hydrophobic trifluoromethyl groups and hydrophilic imidazolium groups can be observed. The microphase separation might facilitate the formation of ionic channels and the transport of protons.

---

Introduction

In the search for alternative energy for fossil fuels, fuel cells, converting chemical energy from fuels such as hydrogen and methanol directly into electrical energy, have thus drawn extensive research efforts in the last several decades. Among various fuel cells, proton exchange membrane fuel cells (PEMFCs) are especially important because they can be used as portable, vehicle, and stationary energy sources. One of the key components for the PEMFC is a proton exchange membrane (PEM), which provides an ionic pathway for proton transfer while preventing the mixing of reactant gases. The current state-of-the-art PEM materials are perfluorinated polymers such as Nafion® due to their superior chemical and thermal stability, outstanding proton conductivity under high relative humidity.¹ However, the applications of Nafion® are hindered by its high cost and severe methanol crossover for direct methanol fuel cells (DMFCs).² In addition, they exhibit strong humidity dependence of proton conductivity because of using water as proton transfer medium. The operating temperature is limited to less than 80 °C.³ In comparison with low temperature PEMFCs, an elevated operating temperature (140–160 °C) has the additional advantages such as reducing CO poisoning on platinum catalyst, increasing the reaction kinetics and simplifying the water management.^4–6 Hence, the design and synthesis of novel anhydrous polymer electrolytes become important to the applications of high temperature PEMFCs.

In comparison with water, phosphoric acid (PA) exhibits excellent thermal stability and low vapor pressure at elevated temperature. Moreover, it can be both proton donor (acidic) and proton acceptor (basic) even in an anhydrous condition to form dynamic hydrogen bond networks, which facilitate protons to migrate readily by hydrogen bond breaking and forming (Grotthuss mechanism).^7,8 It is regarded as the good medium for proton transfer in high temperature PEMFCs. Numerous thermally stable polymers containing basic heterocyclic groups were synthesized and acted as proton acceptors to absorb phosphorous acid for high PEMFC applications.^9–17 Among them, commercially available poly[2,2′-(m-phenylene)-5,5′-bibenzimidazole] (m-PBIs) as well as its derivatives have attracted considerable attention due to their excellent thermal stability, good mechanical property and high proton conductivity (>0.01 S cm⁻¹) at 160 °C after PA doping.¹⁸ However, m-PBI shows limited solubility in common organic solvents when its inherent viscosity is higher than 1.14 dL g⁻¹ due to the structure rigidity and strong hydrogen bonding. Wainright et al. reported that heating a 10 wt% m-PBI solution in DMAc at 94 °C for 5 h would leave 69% of PBI (1.14 dL g⁻¹) undissolved.^19,20 The molecular weight of PBI is thus limited in order to assure solution processability. This might has an adverse effect on the mechanical property of PA-doped membranes. In addition, the poor solubility of PBIs might also lead to the premature precipitation or gelation during polymerization. Some reports attributed the gelation or precipitation to the possible side acylation reaction.²¹ Therefore, the synthesis of high molecular-weight and film-formable PBIs still remains a challenge even today.

As the alternative materials for high temperature PEMFCs, a new type of PEMs based on poly(phthalazinone ether sulfone ketone) containing quaternary ammonium groups was first reported by Li.²² Quaternary ammonium groups have good bonding ability with PA. Some approaches based on quaternized poly(ether sulfone),^23–25 quaternized poly(arylene ether ketone)^26–28 and quaternized PBI²⁹ have been reported. They all exhibited high proton conductivity after PA doping. Therefore, polymers containing quaternary ammonium groups after PA doping could be good candidates as PEMs for high temperature PEMFC applications. On the other hand, the preparation of polyimides is relatively easy, compared with that of PBIs. Polyimides can be synthesized directly by high temperature polycondensation from diamines and dianhydrides if their solubility allows. They are well known for the good thermal stability, mechanical property, and film-forming ability. The applications of polyimides on PEMFCs have been limited to sulfonated polyimides (SPI) in their hydrated state as proton exchange membranes for low temperature PEMFC applications. Even though the synthesis of polyimides containing quaternary pyridine moieties on polymer backbones has been reported,³⁰ the applications of polyimides bearing quaternary ammonium groups as alternatives to PBIs on high temperature PEMFC applications have not been reported yet.

In this work, novel polyimides containing pendent imidazolium groups (ImPI-x) as PEMs are reported for the first time. The pendent imidazolium groups are expected to absorb phosphoric acid via acid–base interaction. The properties of the PEMs, such as thermal stability, oxidative stability, PA uptake, mechanical property and proton conductivity, are investigated in detail. The single cell performance of high temperature PEMFCs and long-term stability of PA doped ImPI-x membranes are also demonstrated.

Experiment

Materials

Paraformaldehyde was purchased from Acros and dried at 80 °C for 12 h under reduced pressure prior to use. Chlorotrimethylsilane (Me₃SiCl), anhydrous stannic chloride (SnCl₄) and dry chloroform with 4 Å molecular sieves were purchased from Aldrich Chemical Co. and used as received. Dimethylacetamide (DMAc) was purified by stirring over CaH₂ for 24 h, then distilled under reduced pressure, and stored over 4 Å molecular sieves. 1-Methylimidazole was obtained from Kriskev Co. and used as received. m-PBI membrane (30 μm) was obtained from Danish Power System Corporation. Other chemicals and solvents were used as received.

Measurements

Proton nuclear magnetic resonance (¹H NMR) spectra were measured at 600 MHz on a Bruker Avance-600 spectrometer. Molecular weights were measured on a JASCO Gel Permeation Chromatography (GPC) system (PU-980) equipped with an RI detector (RI-930), a Jordi Gel DVB Mixed Bed column (250 mm × 10 mm), using DMAc as the eluent and calibrated with polystyrene standards. Thermal gravimetric analyses (TGA) were performed in nitrogen with a TA TGA Q500 thermogravimetric analyzer using a heating rate of 10 °C min⁻¹. The tensile properties of the membranes were measured on a Universal Testing Machine (Testometric M500-25AT) at a cross-head speed of 5 mm min⁻¹ at 25 °C. A scanning electron microscope (SEM S360, Cambridge Instrument) equipped with energy dispersive X-ray (EDX) spectroscopy was used to investigate the surface and cross sectional morphology of the membranes. The images were taken at an accelerating voltage of 15 kV. Before imaging, all samples were cut into small pieces (5 × 10 mm²) then sputtered with gold. Tapping mode atomic force microscopy (AFM) was performed in air with Bruker-ICON2-SYS, using micro-fabricated cantilevers with a force constant of approximately 30 N m⁻¹. Each scan line contains 512 pixels, and a whole image is composed of 256 scan lines. Dry samples were dried at 80 °C under reduced pressure for 12 h and PA-doped samples were immersed into 85% H₃PO₄ for 48 h before testing. All samples were imaged immediately in ambient conditions.

Synthesis of chloromethylated polyimide (CMPI-x)

Chloromethylated polyimide (CMPI-x) was synthesized by a Friedel–Crafts like reaction, as developed by Avram and coworkers.³¹ A typical synthesis procedure of CMPI-0.74, where 0.74 refers to the degree of substitution (the number of chloromethyl groups/repeat unit), was described as follows. Paraformaldehyde (1.98 g, 66 mmol), chlorotrimethylsilane (8.4 mL, 66 mmol) and then anhydrous stannic chloride (0.125 mL, 1.10 mmol) were added into chloroform (40 mL) in a 100 mL flask with a magnetic stirrer. The reaction mixture was stirred at 60 °C to allow paraformaldehyde completely dissolved, then was cooled to room temperature. 4PhODA/PI (0.20 g, 0.22 mmol), synthesized by the method described in our previous report,³² was added and the mixture was further stirred at 60 °C for 24 h. The reaction mixture was then poured into rigorously stirred methanol. The fibrous product was collected by filtration and dried in reduced pressure at 50 °C for 12 h (0.20 g, 96% yield).

Synthesis of imidazolium-functionalized polyimide (ImPI-x)

Imidazolium-functionalized polyimide (ImPI-x) was synthesized by Menshutkin reaction between CMPI-x and 1-methylimidazole. A typical synthesis procedure of ImPI-0.74 was described as follows. CMPI-0.74 (0.40 g, 0.312 mmol chloromethyl group) was dissolved in DMAc (5.7 mL, 7 w/v%) at room temperature in a 50 mL one-necked, round-bottomed flash with a magnetic stirrer. 1-Methylimidazole (75 μL, 0.936 mmol) was added and the reaction mixture was stirred at room temperature in nitrogen atmosphere for 48 h. The reaction mixture was then poured into rigorously stirred acetone. The fibrous product was collected by filtration and dried in reduced pressure at 80 °C for 12 h (0.38 g, 90% yield).

Preparation of phosphorous acid (PA) doped membrane (ImPI-x/PA) and dimensional stability

ImPI-x membranes were obtained by dissolving ImPI-x in DMAc (5 w/v%) at room temperature and casting the solution on a clean glass plate at 60 °C for 24 h. The as-cast membranes was then dried in a vacuum oven at 150 °C for 12 h. For the phosphorous acid (PA) uptake test, ImPI-x membranes with thickness of 40–50 μm were cut into pieces of 2 cm × 2 cm and dried at 100 °C under reduced pressure for 12 h to obtain their dry weights (W_D). Then, the membranes were immersed in 85% PA solution at room temperature for different periods of time. The membranes were taken out from PA solutions, wiped dry and dried at 80 °C under reduced pressure for 12 h to remove water. Their weights were recorded (W_w). The PA uptake, defined as the weight changes of the membrane before and after doping, was calculated by using eqn (1)

PA uptake (%) = (W_w − W_D)/W_D × 100% (1)

Dimensional stability was measured by area and volume changes which were calculated by eqn (2). D_D and D_w are the dimensions of dry and doped membranes, respectively.

Dimensional change (%) = (D_w − D_D)/D_D × 100% (2)

Proton conductivity

Proton conductivity was measured using a two-electrode in-plane method by electrochemical impedance spectroscopy (EIS) with a Zahner potentiostat–galvanostat electrochemical workstation model PGSTAT over a frequency range of 1 Hz to 100 kHz with the oscillating voltage of 10 mV. Proton conductivity was calculated from the impedance data according to the eqn (3),

σ = d/(Rtw) (3)

where σ is the proton conductivity (S cm⁻¹), d is the distance between the electrodes (cm), t and w are the thickness (cm) and width of the sample (cm), respectively. R is the resistance (Ω) associated with the ionic conductivity of the sample from the impedance data.

Membrane electrode assembly (MEA) fabrication and polarization test

Catalyst inks were prepared by blending the calculated amount of Pt/C (from Johnson Matthey, 40 wt% Pt) with an m-PBI/DMAc solution (5 wt%). Gas diffusion layers (from E-Tek) were used to deposit the catalyst layers so that Pt loadings were 0.8 and 0.6 mg cm⁻² for cathode and anode, respectively. The amount of m-PBI in the catalyst layers was 1/20 weight ratio to Pt/C suggested in the report by Lobato and coworkers.³³ After DMAc was evaporated by heating at 70 °C, the gas diffusion electrodes were then impregnated with 20% phosphoric acid for two days prior to use. Membrane electrode assemblies (MEAs) with a test area of 5 cm² were prepared by sandwiching a PA doped ImPI-x membrane between the anode and cathode electrodes.

For polarization test, highly pure hydrogen and oxygen without humidification at a flow rate of 200 and 500 standard cubic centimeter per minute (sccm) were fed to anode and cathode, respectively. A single cell was tested in a PEMFC test station (FCED-DD50, Asia Pacific Fuel Cell Technologies Ltd.) and the back pressure gauges of both electrodes read 1 atm. The polarization curves (cell voltage vs. current density) were recorded in a steady state.

Results and discussion

Synthesis of chloromethylated polyimides (CMPI-x)

The starting material (4PhODA/PI) was synthesized from 2,2′,6,6′-tetraphenyl-4,4′-oxydianiline (4PhODA) and bis(3,4-dicarboxyphenyl)hexafluoropropane dianhydride (6FDA) by one-step, high-temperature polycondensation in m-cresol. The detailed synthetic procedure has been reported in our previous publication.³² In order to prepare the PEMs, three chemical reactions including chloromethylation, quaternization and PA doping should be performed as shown in Scheme 1. Among them, chloromethylation is the most important reaction that determines the degree of subsequent quaternization and the proton conductivity. The proposed mechanism is shown in Scheme 2.^31,34 Chloromethylation normally involves the formation of electrophile (Scheme 2a), followed by the electrophilic substitution on active phenyl rings (Scheme 2b). However, it has been also reported that the active phenyl ring would attack the chloromethyl group attached on another phenyl ring in a Friedel–Crafts alkylation fashion (Scheme 2c).³⁴ This would form methylene bridges between macromolecular chains and result in crosslinking or gel formation. The crosslinking reaction not only leads to the decreased number of chloromethyl groups available for quaternization, but also hinders polymers from further membrane preparation. To avoid such undesired crosslinking reaction, it is necessary to study the effects of reaction parameters on the chloromethylation of polyimides. The ¹H NMR measurement was used to investigate these effects.

Scheme 1 Synthetic route for ImPI-x/PA.

Scheme 2 The mechanism of chloromethylation: (a) formation of electrophile, (b) chloromethylation and (c) crosslinking reaction.

The carcinogenic chloromethyl methyl ether commonly used as the chloromethylation reagent³⁵ was replaced by paraformaldehyde and chlorotrimethylsilane with SnCl₄ as catalyst. The chloromethylation was carried out in anhydrous chloroform (60 °C). As an electrophilic substitution reaction, chloromethylation would take place on the phenyl side groups (para positions) of 4PhODA/PI due to the high electron-density. The chloromethylated polyimides are named as CMPI-x, where x is the number of chloromethyl groups per repeating unit, defined as the degree of substitution (DS), and can be controlled by reaction parameters. Fig. 1a and b show the ¹H NMR spectra of 4PhODA/PI and the formed chloromethylated polyimide (CMPI-1.50, entry 9 in Table 1). The peak appeared at 4.66 ppm (H_e) was assigned to the protons of the attached chloromethyl groups. The additional peaks (H_a′) appeared nearby the peak H_a (δ = 6.95 ppm) were resulted from random chloromethylation. It indicated that different number of chloromethyl groups were attached to the repeating unit of polyimide. Moreover, the proton peaks of side phenyl groups (H_b and H_c) shifted slightly to the deshielding region (H_b′ and H_c′) due to the electron-withdrawing chloromethyl groups. The degree of substitution (DS) of CMPI-x was calculated from the integration (DS = 2H_e/(H_a + H_a′)).

Fig. 1 ¹H NMR spectra of (a) 4PhODA/PI, (b) CMPI-1.50 and (c) ImPI-1.50.

Table 1 Effect of the reaction parameters on the chloromethylation of 4PhODA/PI^a

Entry	(CH2O)n^b (mol mol^−1)	Me3SiCl^b (mol mol^−1)	SnCl4^b (mol mol^−1)	Polymer concentration (w/v%)	Time (h)	DS^c
a Reaction temperature was 60 °C, unless specified otherwise.b mol mol^−1: moles per PI repeating unit.c Number of substituted chloromethyl groups per PI repeating unit, calculated by ^1H NMR spectra. DS = 2He/(Ha + Ha′).d PI was dissolved before chloromethylation reagents and SnCl4 were added.e Reaction temperature was 40 °C.f Gelation occurred; DS was determined from the soluble part.
1^d	300	300	—	0.50%	24	0
2^d	—	300	—	0.50%	24	0
3^d	—	—	5	0.50%	24	0
4^e	300	300	5	0.50%	24	0.05
5^d	300	300	5	0.50%	24	0.68
6	300	300	5	0.50%	12	0.22
7	300	300	5	0.50%	24	0.74
8	300	300	5	0.50%	48	1.04
9	300	300	5	0.50%	72	1.50
10	300	300	5	0.50%	96	1.75
11	300	300	5	0.50%	120	2.17
12	300	300	5	0.50%	336	2.75^f
13	50	50	5	0.50%	24	0.19
14	150	150	5	0.50%	24	0.30
15	400	400	5	0.50%	24	0.31
16	300	300	5	0.25%	24	0.32
17	300	300	5	0.67%	24	1.07
18	300	300	5	1.00%	24	1.48^f
19	300	300	1	0.50%	24	0.15
20	300	300	10	0.50%	24	1.07
21	300	300	20	0.50%	24	1.69^f
22	300	300	30	0.50%	24	1.81^f
23	300	300	30	0.50%	72	3.89^f

---

The effect of reaction parameters on DS of CMPI-x

The effects of reaction temperature, reaction time and concentration of chloromethylation reagents, polymer and catalyst on the degree of substitution (DS) were investigated. The results are summarized in Table 1. It can be observed clearly both chloromethylation reagents and catalyst play the very important role in the reaction. No chloromethylated polyimides could be obtained in the absence of SnCl₄ or chloromethylation reagents (entry 1–3). Besides, only very few chloromethyl groups (DS = 0.05) were attached when the reaction was carried out at 40 °C for 24 h (entry 4). This might be also resulted from the poor solubility of paraformaldehyde in chloroform at this temperature. DS was increased to 0.74 as reaction temperature was raised to 60 °C (entry 7). Further increase in reaction temperature was hindered by the boiling point of reaction solvent (chloroform). Therefore, reaction temperature was set as 60 °C, unless specified otherwise. DS also increased from 0.22 to 2.75 when reaction time increased from 12 to 336 h (entry 6–12). Fig. 2 presents the ¹H NMR spectra of chloromethylated polyimides with DS from 0.22 to 2.17 (entry 6–11), indicating the effect of reaction time on DS. However, long reaction time (entry 12, 336 h) also led to partial crosslinking or gelation as the mechanism shown in Scheme 2c.

Fig. 2 ¹H NMR spectra of CMPIs with the different degrees of chloromethylation: (a) 0.22, (b) 0.74, (c) 1.04, (d) 1.50, (e) 1.75 and (f) 2.17.

The influence of chloromethylation reagent concentration on DS was also investigated. With an increase in the concentration of chloromethylation reagents from 50 to 300 mol mol⁻¹ repeating unit, DS also increased from 0.19 to 0.74 (entry 7, 13, 14). However, the impact of chloromethylation reagent concentration on increasing DS was not as significant as that of the reaction time. Attempts to add more chloromethyl groups were unsuccessful by merely increasing the concentration of chloromethylation reagents to 400 mol mol⁻¹ repeating unit (entry 15). When polymer concentration was increased from 0.25 to 1.00 w/v%, DS increased from 0.32 to 1.48 (entry 7, 16–18). Unfortunately, gelation occurred when polymer concentration was 1.00 w/v% (entry 18).

The amount of catalyst used is another important parameter for the reaction. For example, DS increased from 0.15 to 1.81 as the amount of SnCl₄ was increased from 1 to 30 mol mol⁻¹ repeating unit when the reaction was carried out for 24 h at 60 °C (entry 7, 19–22). Gelation was also observed when the amount of catalyst reached 20 and 30 mol mol⁻¹ repeating unit (entry 21, 22).

Theoretically, the maximum number of attached chloromethyl groups should be 4, because it has four reactive phenyl groups on the repeating unit of 4PhODA/PI. From the results shown in Table 1, the maximum number of attached chloromethyl groups could only reach 3.89 at a vigorous reaction condition (entry 23). Under this reaction condition, severe gelation occurred, which is undesirable for the following quaternization and membrane casting.

In summary, chloromethylation on electron-rich phenyl rings can be controlled by properly adjusting the reaction parameters such as reaction temperature and time, concentrations of chloromethylation reagents, catalyst and polymer. Chloromethylation can be better controlled by reaction time under mild reaction conditions (entry 7–11) with DS in the range of 0.74 to 2.17. Attempts to increase DS by conducting the reaction in more vigorous conditions (higher concentration of catalyst and polymer) were unsuccessful due to gelation.

The effect of reaction parameters on molecular weight of CMPI-x

The effect of reaction parameters on the molecular weights and polydispersity of CMPI-x were also investigated. The results are summarized in Table 2. The reaction parameters of each entry in Table 2 were the same as those shown in Table 1. Chloromethylation was initially performed by dissolving 4PhODA/PI in chloroform before the chloromethylation reagents and SnCl₄ were added (entry 1–3, 5). It can be observed that the molecular weights decreased, indicating the degradation of polyimides (entry 1–3, 5). For example, the number-average molecular weight (M_n) decreased from 45 000 to 22 000 g mol⁻¹ after chloromethylation (entry 5). It is interesting that degradation also occurred when 4PhODA/PI was treated only with chlorotrimethylsilane or SnCl₄, respectively, at 60 °C for 24 h (entry 2, 3). It has been reported that the ether group could be cleaved with Lewis acids. For example, Seeberger et al. reported the aminated benzyl ether was quantitatively cleaved with TiCl₄ or SnCl₄ in only 5 min at room temperature.³⁶ Sharma et al. also reported an efficient ZrCl₄-catalyzed protocol for the ether cleavage of p-methoxybenzyl (PMB) ethers and esters in short reaction time.³⁷ The mechanism of ether cleavage was proposed that the lone pair of ether group attacks the electron-deficient Lewis acid to form a complex, followed by the departure of the OR by either an SN₁ or SN₂ reaction.³⁸ Dialkyl and alkyl aryl ethers can also be cleaved with iodotrimethylsilane or a mixture of chlorotrimethylsilane and NaI.^39,40 It is, therefore, very likely that 4PhODA/PI would degrade by the proposed mechanisms when contact with chlorotrimethylsilane and SnCl₄ separately.

Table 2 Molecular weight and polydispersity after chloromethylation

Entry	DS	Mn (g mol^−1)	PDI^b
a PI was dissolved before chloromethylation reagents and SnCl4 were added.b PDI = Mw/Mn.
Pristine	—	45 000	1.29
1^a	0	30 000	1.36
2^a	0	30 000	1.36
3^a	0	26 000	1.30
5^a	0.68	22 000	1.41
7	0.74	46 000	1.61
9	1.50	44 000	2.19
10	1.75	48 000	2.33
17	1.07	42 000	2.21

---

Even in the presence of paraformaldehyde (entry 5), polymer degradation still occurred with a decreased M_n of 26 000 g mol⁻¹. According to the mechanism shown in Scheme 2a, chlorotrimethylsilane is supposed to react with paraformaldehyde in the presence of Lewis acid catalyst to form electrophilic species that should attach onto the active phenyl rings. Unfortunately, degradation is still inevitable due to the poor solubility of paraformaldehyde in reaction solvent (chloroform). Chlorotrimethylsilane and catalyst can still be able to react with the ether linkages of 4PhODA/PI before they are fully consumed by paraformaldehyde.

In order to minimize the degradation, the procedure of chloromethylation was changed so that 4PhODA/PI was added after paraformaldehyde, chlorotrimethylsilane and SnCl₄ were completely dissolved in chloroform. As shown in Table 2 (entry 7), CMPI-0.74 had an almost unchanged M_n of 46 000 g mol⁻¹. It indicates that polymer degradation can be circumvented if paraformaldehyde was completely dissolved and reacted with chlorotrimethylsilane and SnCl₄ before polyimides were added. Moreover, chloromethylated polyimides with larger polydispersity (PDI) were obtained when they were prepared in more vigorous reaction conditions (entry 9, 10 and 17), even though they had similar number-average molecular weights. The larger PDI indicates that polymers containing higher and lower molecular weights were formed. As mentioned earlier, the crosslinking reaction (or methylene linkage formation intermolecularly) could happen if chloromethylation was carried out in more vigorous reaction conditions, leading to the increase of molecular weights. At the same time, excessive chlorotrimethylsilane and SnCl₄ might have better chances to cause the cleavage of ether linkages, leading to the formation of low molecular weight polymers. The combination of these two effects leads to larger PDIs.

Synthesis of imidazolium-functionalized polyimides (ImPI-x)

Quaternization was carried out by Menshutkin reaction between CMPI-x and 1-methylimidazole. The ¹H NMR spectrum of the obtained ImPI-1.50 is shown in Fig. 1c. It can be observed that three characteristic peaks corresponding to imidazolium groups appeared at 7.71 ppm (H₇), 7.76 ppm (H₈) and 9.58 ppm (H₆). The peak assignment of the protons on imidazolium is consistent with that reported in the literature.²⁵ In addition, the proton peaks of methylene (H₅) and methyl (H₉) were observed at 5.54 and 3.87 ppm, respectively. The integration ratio of H₅ to H₉ was close to 2 : 3. It indicated the complete conversion from the chloromethyl group to the imidazolium group.

Thermal stability

The thermal stability of polymers were investigated by thermo-gravimetric analysis (TGA). Fig. 3a shows TGA curves of 4PhODA/PI and CMPI-x (x: 0.84 and 1.99). 4PhODA/PI is thermally stable up to 520 °C. In contrast, two decomposition stages were observed from the TGA curves of CMPI-x. The first stage started initially at 240 °C, which was attributed to the loss of chloromethyl groups. The second stage occurred at around 475 °C, which was ascribed to the main chain decomposition. Furthermore, the weight losses at the first stage were measured to be 4.35% and 9.23% for CMPI-0.84 and CMPI-1.99, respectively. These results are in a good agreement with the theoretical contents of chloromethyl groups in these polymers. The theoretical contents were 4.23% and 10.03% for the above two samples when calculated from their DS values determined by ¹H NMR spectra.

Fig. 3 TGA curves: (a) 4PhODA/PI and CMPIs, (b) ImPI-0.84 with Cl⁻ and H₂PO₄⁻ anions.

Fig. 3b shows the TGA curves of ImPI-0.84 in Cl⁻ and H₂PO₄⁻ forms. ImPI-0.84 in H₂PO₄⁻ form was prepared by immersing the Cl⁻ form into 85% PA solution for 24 h at room temperature. Cl⁻ anion was replaced by H₂PO₄⁻ and the excess PA was removed through continuous washing by water then dried at 80 °C under vacuum for 24 h. Similar to its chloromethyl precursor, ImPI-0.84 in Cl⁻ form exhibited two stages of decomposition at 180 and 470 °C. These two stages of weight loss corresponded to the loss of imidazolium groups and main chain decomposition, respectively. ImPI-0.84 in H₂PO₄⁻ form also exhibited two stages of weight loss at 240 and 465 °C for the same reason. It should be noted that ImPI-0.84 in H₂PO₄⁻ form had higher initial decomposition temperature than in Cl⁻ form. It was attributed to the protecting effect of dihydrogen phosphate ions around the imidazolium groups.²⁶ No weight loss of ImPI-0.84 in H₂PO₄⁻ form can be observed at temperature below 230 °C, revealing the potential for high temperature PEMFC applications.

Phosphorous acid (PA) uptakes

Fig. 4 shows the PA uptakes of ImPI-x (40–50 μm) and m-PBI membranes (30 μm) as a function of immersion time in 85% phosphorous acid solution at room temperature. All polymer membranes showed a similar PA doping behavior. The PA uptake increased rapidly in 12 h and then levelled off after 48 h. m-PBI absorbed more phosphoric acid than ImPI-x. For example, the maximum PA uptakes of ImPI-1.51 and m-PBI at room temperature were 159% and 216%, respectively. An increase in PA uptake with the increase in DS values was also observed for ImPI-x. The PA uptakes of ImPI-x membranes with DS values from 0.51 to 1.51 was in the range of 35–165%. It can be attributed to the higher degree of chloromethylation that led to more hydrophilic imidazolium groups by quaternization, and thus more phosphoric acid absorbed by acid–base interaction. With saturated PA uptakes, all membranes were mechanically strong enough for the following proton conductivity and fuel cell performance tests.

Fig. 4 PA uptakes of ImPI-x and m-PBI membranes as a function of immersion time in 85% phosphoric acid at room temperature.

Dimensional stability and mechanical properties

The dimensional stability of PA doped membranes was estimated by measuring the variations in surface area and volume after doping process. The results are listed in Table 3. The dimensional changes in surface area and volume of ImPI-x increased as PA uptakes increased. For example, the surface area and volume change increased from 3.9% to 12.2% and from 13.8% to 42.1%, respectively when the PA uptakes of ImPI-x increased from 34% to 159%. m-PBI exhibited more surface area and volume changes (29.1% and 44.0%) which was resulted from its higher PA uptake (216%). In addition, the isotropic area changes were also calculated from the measured volume changes and compared with the measured area changes (Table 3). ImPI-x membranes with low PA uptakes (34 and 39%) showed anisotropic dimensional change. They swelled more in thickness direction than in plane direction. ImPI-x membranes with higher PA uptakes (higher than 63%) exhibited more isotropic swelling. In contrast, commercial m-PBI membranes showed anisotropic dimensional change with larger swelling in plane direction than in thickness direction.

Table 3 PA uptakes, dimension stability and mechanical properties of ImPI-x and m-PBI membranes before and after PA doping

Polymer	PA uptake (%)	Area^a (%)	Volume^b (%)	Tensile strength (MPa)	Elongation at break (%)	Young modulus (GPa)
a Calculated by (ΔA/A) × 100%; A = initial area; ΔA = area changes.b Calculated by (ΔV/V) × 100%; V = initial volume; ΔV = volume changes.c The values in the bracket mean isotropic area changes calculated from volume changes.d The values in the bracket mean mechanical properties of dry membranes.
ImPI-0.51	34	3.9 (5.7)^c	13.8	9.8 (32.9)^d	22.5 (3.0)^d	0.22 (1.09)^d
ImPI-0.63	39	4.7 (6.0)	14.9	9.4 (37.6)	22.9 (2.9)	0.19 (1.29)
ImPI-0.75	63	9.2 (9.6)	30.1	7.9 (41.3)	24.3 (3.7)	0.16 (1.11)
ImPI-0.84	84	10.4 (10.2)	32.4	7.1 (45.8)	23.8 (3.4)	0.15 (1.34)
ImPI-1.09	100	10.6 (10.5)	34.2	6.2 (31.1)	28.4 (2.5)	0.14 (1.24)
ImPI-1.20	141	11.8 (11.5)	39.3	5.9 (32.8)	25.5 (3.0)	0.14 (1.08)
ImPI-1.51	159	12.2 (12.1)	42.1	5.5 (33.4)	27.7 (2.4)	0.12 (1.39)
m-PBI	216	29.1 (12.5)	44.0	10.8 (86.7)	59.5 (10.7)	0.16 (2.73)

---

The tensile strength, elongation at break and Young's modulus of ImPI-x and m-PBI membranes are also listed in Table 3. Dry ImPI-x membranes had tensile strength of 31.1–45.8 MPa, elongation of 2.4–3.7% and Young's modulus of 1.08–1.39 GPa. After PA doping, ImPI-x membranes, with PA uptakes from 34 to 159%, had tensile strength of 5.5–9.8 MPa, elongation of 22.5–27.7% and Young's modulus of 0.12–0.22 GPa. The lower mechanical strength and longer elongation at break as PA uptakes increased were resulted from the plasticization effect of phosphoric acid.⁴¹ Compared with commercial m-PBI with a PA uptake of 216%, PA doped ImPI-x membranes showed acceptable mechanical properties for MEA fabrication.

Proton conductivity

The proton conductivity of PA doped ImPI-x membranes was measured from 60 to 160 °C under anhydrous conditions and compared with that of commercial m-PBI. The results are shown in Fig. 5. A stable increase in the proton conductivity with temperature increasing up to 160 °C was observed for all membranes. ImPI-x membranes with higher PA uptakes resulted from more imidazolium groups exhibited higher proton conductivity. ImPI-1.20 and ImPI-1.51 with PA uptakes of 141% and 159%, had proton conductivity of 0.040 and 0.057 S cm⁻¹ at 160 °C, respectively. Their proton conductivity in the temperature range of 60 to 160 °C was all higher than 0.01 S cm⁻¹, which was required for PEMFC application.⁴² The proton conductivity of m-PBI depends primarily on PA uptakes.^43–45 With a PA uptake of 216%, m-PBI membrane had a proton conductivity of 0.046 S cm⁻¹ at 160 °C. In contrast, ImPI-1.20 exhibited a similar proton conductivity of 0.041 S cm⁻¹ even though it had a much lower PA uptake (141%). The high proton conductivity of ImPI-1.20 and ImPI-1.51 with lower PA uptakes might be resulted from the hydrophobic trifluoromethyl groups of 4PhODA/PI. As we reported previously,⁴⁶ these hydrophobic groups might facilitate the phase separation in nanometer scale that would lead to the formation of ionic channels at lower PA uptakes, and thus promote the proton conduction.

Fig. 5 Proton conductivity of PA doped ImPI-x and m-PBI membranes as a function of temperature. The values in the bracket represent PA uptakes.

Fuel cell performance

Fig. 6 shows the polarization and power density curves of fuel cells using PA doped ImPI-x membranes as PEMs, with H₂/O₂ at 160 °C under anhydrous conditions. PEMFCs based on PA doped commercial m-PBI membranes were also tested under the same condition for comparison. ImPI-x membranes with more imidazolium groups would absorb more phosphoric acid. When the catalyst loadings, gas flow, and test temperature were fixed, PEMFCs based on ImPI-x membranes with higher PA uptakes exhibited higher power density. For example, the peak power density of fuel cell based on ImPI-0.84 with a PA uptake of 84% was found to be 247 mW cm⁻², whereas ImPI-1.09 based fuel cell exhibited higher power density of 352 mW cm⁻² due to its higher PA uptake (100%). PEMFC based on ImPI-1.51 with a PA uptake of 159% showed the highest power density of 551 mW cm⁻². This can be attributed to the more free acids in PEMs that can enhance the proton conduction and accelerate the electro-chemical reaction. However, PA doped m-PBI had the lower peak power density (419 mW cm⁻²) than ImPI-1.51 even though m-PBI had a higher PA uptake (216%). It could be attributed to the supposedly formed ionic channels by both hydrophobic trifluoromethyl groups and hydrophilic imidazolium groups of ImPI-x as discussed above. Moreover, PEMFCs based on ImPI-x membranes had open circuit voltage in the range of 0.91 to 0.94 V at 160 °C, indicating that there is no obvious increase in the gas permeability of ImPI-x membranes when PA uptakes increased from 84% to 159%. These values are higher than that of commercial m-PBI based PEMFC (0.88 V).

Fig. 6 Polarization and power density curves of fuel cells based on PA doped ImPI-x membranes using H₂/O₂ at 160 °C.

SEM, EDX and AFM morphology

The surface and cross sectional morphology of dry and PA doped ImPI-1.20 membranes were investigated using scanning electron microscopy (SEM) and energy dispersive X-ray analysis (EDX). As shown in Fig. 7a–d, the SEM images of surface and cross section of ImPI-1.20 membrane revealed a dense and non-porous morphology even after PA doping. The dense membranes could minimize fuel gas crossover during fuel cell operation, and thus lead to the high open circuit voltages as mentioned earlier. In Fig. 7e, the EDX analysis of dry ImPI-1.20 shows C, N, O and F element peaks at 0.25, 0.35, 0.51 and 0.73 keV, respectively. These were contributed by polyimide chains. In addition, Cl peaks were also found at 0.26 and 2.68 keV which were contributed by the imidazolium groups in Cl⁻ form. In Fig. 7f, an additional P peak at 2.10 keV suggested a conversion from Cl⁻ to H₂PO₄⁻ form after PA doping. Fig. 7g and h show the chlorine and phosphorus mapping on the cross section of dry and PA doped ImPI-1.20 membranes. It could be found that the chlorine atoms were not randomly distributed in the dry membrane. It implies that some ionic aggregations might be formed due to the hydrophilic imidazolium groups. After doping, PA molecules would be absorbed more in the hydrophilic region thus the non-random distribution of phosphorus atoms was also observed.

Fig. 7 SEM images of ImPI-1.20 membranes: surface images: (a) dry and (b) PA doped; cross sectional images: (c) dry and (d) PA doped; EDX analysis: (e) dry and (f) PA doped; (g) EDX chlorine mapping on the cross section of dry ImPI-1.20 membrane; (h) EDX phosphorus mapping on the cross section of PA doped ImPI-1.20 membrane.

The morphology of 4PhODA/PI, dry and PA doped ImPI-x membranes was also investigated by AFM. Fig. 8 shows their tapping mode phase images under ambient conditions on a 700 nm × 700 nm scale. As shown in Fig. 8a, 4PhODA/PI exhibited featureless phase morphology. On the other hand, all the ImPI-x membranes exhibited microphase separation morphology as illustrated by the bright and dark region. The bright region was assigned to hydrophobic trifluoromethyl groups and polyimide chains rich domain, while the dark region was assigned to the softer region formed by hydrophilic imidazolium groups containing small amount of absorbed water (dry membranes) and/or the absorbed phosphoric acid (in case of PA doping). For ImPI-0.84 (Fig. 8b), an isolated ionic cluster region was found with a diameter of 15–20 nm. For ImPI-1.09 (Fig. 8d), the phase contrast of the hydrophilic ionic domains increased and became more easily distinguished, but the ionic domains were still segregated with diameters of approximately 30 nm. For ImPI-1.20 and ImPI-1.51 (Fig. 8e and f), significant changes in phase images were observed. The ionic domains became continuous to form the channels. Similar continuous ionic channel structure was also observed in the case of Nafion.⁴⁷ Moreover, in order to study the morphological change of ImPI-x after PA doping, ImPI-0.84 membrane was doped in 85% PA solution for 48 h to reach a saturated PA uptake of 84%. Fig. 8c shows the AFM phase imaging of PA doped ImPI-0.84 membrane. The absorbed PA molecules resulted in the swelling of the hydrophilic ionic domains and the more distinct phase separations were also observed. In this research, the hydrophobic trifluoromethyl group and hydrophilic imidazolium group greatly affected the microphase separation of ImPI-x membranes. The connectivity of ionic regions increased as the increased degree of substitution. The well-connected ionic channels facilitated the transport of protons, which explained that PA doped ImPI-1.51 membrane with a lower PA uptake exhibited the higher proton conductivity and peak power density than m-PBI.

Fig. 8 AFM tapping-mode phase images of (a) 4PhODA/PI, (b) ImPI-0.84, (c) PA doped ImPI-0.84, (d) ImPI-1.09, (e) ImPI-1.20 and (f) ImPI-1.51 membranes.

Long-term stability test on proton conductivity and molecular weight

The long-term stability of proton conductivity and weight-average molecular weight (M_w) of ImPI-1.20 were evaluated by placing the membranes in an air-circulating oven at 160 °C. The proton conductivity and M_w were then measured at different time intervals. The results are shown in Fig. 9. ImPI-1.20 with a PA uptake of 141% exhibited high proton conductivity of 0.040 S cm⁻¹ at 160 °C before thermal aging. The proton conductivity was continuously decreased from 0.040 to 0.031 S cm⁻¹ within the period of 480 h. No further discernible decrease was observed up to 800 h. The deterioration in proton conductivity was attributed to the PA leaching and the conversion of PA to polyphosphoric acid.^46,48,49 If the membranes after thermal aging and conductivity measurements were again immersed in 85% PA solution for 30 min at room temperature, the proton conductivity would restore almost to the initial values. The molecular weight (M_w) of PA doped ImPI-1.20 membrane was measured by GPC after thermal aging at different time intervals. The results are also shown in Fig. 9. It has been reported that polyimides would undergo hydrolysis degradation in basic or acidic conditions at high temperature, and leading to lower M_w and worse mechanical property.⁵⁰ The M_w of PA doped ImPI-1.20 decreased quickly from 51 000 to 35 800 g mol⁻¹ within 288 h at 160 °C. The degradation slowed down after 300 h. When PA doped ImPI-1.20 was thermal aged for 800 h, the M_w decreased to 33 800 g mol⁻¹. The hydrolytic stability might be improved by crosslinking the chloromethyl groups with bis-imidazole compounds. The use of six-membered imide rings might also be helpful to solve this issue.

Fig. 9 The proton conductivity and M_w of PA doped ImPI-1.20 membranes after thermal aging at 160 °C for different time intervals.

Conclusions

Novel imidazolium-functionalized polyimides (ImPI-x) were successfully synthesized from polyimides containing trifluoromethyl groups, ether linkages and four phenyl substituents (4PhODA/PI) via chloromethylation followed by quaternization with 1-methylimidazole. Different degrees of substitution (DS) can be achieved by controlling the reaction parameters during the chloromethylation and confirmed by ¹H NMR spectra. The cleavage of ether linkages and crosslinking during chloromethylation can be avoided by carrying out the reaction at 60 °C with suitable concentrations of chloromethylation reagents, catalyst and polymer. ImPI-x membranes also showed the good thermal stability and mechanical properties in both their dry and PA doped states. Higher proton conductivity can be achieved as ImPI-x membranes had higher DS values and thus absorbed more phosphoric acid. The proton conductivity of ImPI-1.51 (0.057 S cm⁻¹) was higher than that of commercial m-PBI (0.046 S cm⁻¹) at 160 °C. In the fuel cell test performed with H₂/O₂ under anhydrous conditions at 160 °C, the peak power density of ImPI-1.51 based PEMFC with a PA uptake of 156% was 551 mW cm⁻², which was higher than that of m-PBI based PEMFC (419 mW cm⁻²) with a PA uptake of 216%. The higher proton conductivity and peak power density of ImPI-1.51 were attributed to microphase separation that might be resulted from the hydrophobic trifluoromethyl groups and hydrophilic imidazolium groups. The microphase separation suggested by AFM phase images might facilitate the formation of ionic channels at lower PA uptakes. The combination of excellent thermal stability, good mechanical properties, high proton conductivity and high peak power density makes ImPI-x membranes promising candidates as electrolyte membranes for high temperature fuel cell applications.

Notes and references

1. Q. Li, E. He, J. O. Jensen and N. J. Bjerrum, Chem. Mater., 2003, 15, 4896.
2. A. Kuver, I. Vogel and W. Vielstich, J. Power Sources, 1994, 52, 77.
3. P. Costamagna and S. Srinivasan, J. Power Sources, 2001, 102, 242.
4. S. Bose, T. Kuila, T. X. Hien Nguyen, N. H. Kim, K. Lau and J. H. Lee, Prog. Polym. Sci., 2011, 36, 813.
5. Q. Li, J. O. Jensen, R. F. Savinell and N. J. Bjerrum, Prog. Polym. Sci., 2009, 34, 449.
6. J. A. Asensio, E. M. Sánchez and P. Gómez-Romero, Chem. Soc. Rev., 2010, 39, 3210.
7. M. Schuster, T. Rager, A. Noda, K. D. Kreuer and J. Maier, Fuel Cells, 2005, 5, 355.
8. S. J. Paddison, K. D. Kreuer and J. Maier, Phys. Chem. Chem. Phys., 2006, 8, 4530.
9. M. Geormezi, V. Deimede, N. Gourdoupi, N. Triantafyllopoulos, S. Neophytides and J. K. Kallitsis, Macromolecules, 2008, 41, 9051.
10. M. Geormezi, C. L. Chochos, N. Gourdoupi, S. G. Neophytides and J. K. Kallitsis, J. Power Sources, 2011, 196, 9382.
11. C. H. Shen, S. L. C. Hsu, E. Bulycheva and N. Belomoina, J. Mater. Chem., 2012, 22, 19269.
12. K. D. Papadimitriou, M. Geormezi, S. G. Neophytides and J. K. Kallitsis, J. Membr. Sci., 2013, 433, 1.
13. D. W. Seo, Y. D. Lim, S. H. Lee, H. H. Jang, S. Y. Choi, Y. T. Jeon, H. Ju and W. G. Kim, Int. J. Hydrogen Energy, 2013, 38, 6261.
14. S. Yuan, X. Guo, D. Aili, C. Pan, Q. Li and J. Fang, J. Membr. Sci., 2014, 454, 351.
15. N. Gourdoupi, A. K. Andreopoulou, V. Deimede and J. K. Kallitsis, Chem. Mater., 2003, 15, 5044.
16. E. K. Pefkianakis, V. Deimede, M. K. Daletou, N. Gourdoupi and J. K. Kallitsis, Macromol. Rapid Commun., 2005, 26, 1724.
17. J. K. Kallitsis, M. Geormezi and S. G. Neophytides, Polym. Int., 2009, 58, 1226.
18. J. S. Wainright, J. T. Wang, D. Weng, R. F. Savinell and M. Litt, J. Electrochem. Soc., 1995, 142, L121.
19. M. Litt, R. Ameri, Y. Wang, R. Savinell and J. Wainright, Mater. Res. Soc. Symp. Proc., 1999, 548, 313.
20. J. S. Wainright, M. Litt and R. F. Savinell, in Handbook of Fuel Cells: Fundamentals, Technology, and Applications, ed. W. Vielstich, A. Lamm and H. A. Gasteiger, John Wiley and Sons, New York, 2003, vol. 3, p. 438.
21. A. Y. Leykin, A. I. Fomekov, E. G. Galpern, I. V. Stankevich and A. L. Rusanov, Polymer, 2010, 51, 4053.
22. M. Li, H. Zhang and Z. G. Shao, Electrochem. Solid-State Lett., 2006, 9, A60.
23. M. Li, K. Scott and X. Wu, J. Power Sources, 2009, 194, 811.
24. X. Wang, C. Xu, B. T. Golding, M. Sadeghi, Y. Cao and K. Scott, Int. J. Hydrogen Energy, 2011, 36, 8550.
25. J. Yang, Q. Li, J. O. Jensen, C. Pan, L. N. Cleemann, N. J. Bjerrum and R. He, J. Power Sources, 2012, 205, 114.
26. W. Ma, C. Zhao, H. Lin, G. Zhang, J. Ni, J. Wang, S. Wang and H. Na, J. Power Sources, 2011, 196, 9331.
27. W. Ma, C. Zhao, J. Yang, J. Ni, S. Wang, N. Zhang, H. Lin, J. Wang, G. Zhang, Q. Li and H. Na, Energy Environ. Sci., 2012, 5, 7617.
28. N. Zhang, B. Wang, C. Zhao, S. Wang, Y. Zhang, F. Bu, Y. Cui, X. Li and H. Na, J. Mater. Chem. A, 2014, 2, 13996.
29. C. Xu, K. Scott, Q. Li, J. Yang and X. Wu, Fuel Cells, 2013, 13, 118.
30. X. Sun, Y. K. Yang and F. Lu, Macromol. Chem. Phys., 1997, 198, 833.
31. E. Avram, E. Butuc and C. Luca, J. Macromol. Sci., Part A: Pure Appl.Chem., 1997, 34, 1701.
32. J. C. Chen, J. A. Wu, S. W. Li and S. C. Chou, React. Funct. Polym., 2014, 78, 23.
33. J. Lobato, P. Cañizares, M. A. Rodrigo, J. J. Linares and F. J. Pinar, Int. J. Hydrogen Energy, 2010, 35, 1347.
34. C. Lü, B. Gao, Q. Liu and C. Qi, Colloid Polym. Sci., 2008, 286, 553.
35. J. Wang, S. Li and S. Zhang, Macromolecules, 2010, 43, 3890.
36. O. J. Plante, S. L. Buchwald and P. H. Seeberger, J. Am. Chem. Soc., 2000, 122, 7148.
37. G. V. Madhava Sharma, C. G. Reddy and P. R. Krishna, J. Org. Chem., 2003, 68, 4574.
38. R. L. Burwell Jr, Chem. Rev., 1954, 54, 615.
39. G. A. Olah, S. C. Narang, B. G. B. Gupta and R. Malhotra, J. Org. Chem., 1979, 44, 1247.
40. M. E. Jung and M. A. Lyster, J. Org. Chem., 1977, 42, 1977.
41. T. H. Kim, S. K. Kim, T. W. Lim and J. C. Lee, J. Membr. Sci., 2008, 323, 362.
42. S. J. Hamrock and M. A. Yandrasits, Polym. Rev., 2006, 46, 219.
43. A. Schechter and R. F. Savinell, Solid State Ionics, 2002, 147, 181.
44. C. E. Hughes, S. Haufe, B. Angerstein, R. Kalim, A. Reiche and M. Baldus, J. Phys. Chem. B, 2004, 108, 13626.
45. Y. L. Ma, J. S. Wainright, M. H. Litt and R. F. Savinell, J. Electrochem. Soc., 2004, 151, A8.
46. J. C. Chen, J. A. Wu, C. Y. Lee, M. C. Tsai and K. H. Chen, J. Membr. Sci., 2015, 483, 144.
47. R. S. McLean, M. Doyle and B. B. Sauer, Macromolecules, 2000, 33, 6541.
48. S. Yu, L. Xiao and B. C. Benicewicz, Fuel Cells, 2008, 8, 165.
49. M. A. Molleo, X. Chen, H. J. Ploehn and B. C. Benicewicz, Fuel Cells, 2014, 14, 16.
50. R. Deiasi and J. Russell, J. Appl. Polym. Sci., 1971, 15, 2965.

---