Structurally Well-Defined Anion-Exchange Membranes Containing Perfluoroalkyl and Ammonium-Functionalized Fluorenyl Groups.

Novel anion-conductive polymers containing perfluoroalkyl and ammonium-functionalized fluorene groups were synthesized and characterized. The quaternized polymers synthesized using a dimethylaminated fluorene monomer had a well-defined chemical structure in which each fluorenyl group was substituted with two ammonium groups at specific positions. The resulting polymers had a high molecular weight (M n = 8.9-13.8 kDa, M w = 13.7-24.5 kDa) to provide bendable thin membranes with the ion-exchange capacity (IEC) ranging from 0.7 to 1.9 mequiv g-1 by solution casting. Both transmission electron microscopy images and small-angle X-ray scattering patterns suggested that the polymer membranes possessed a nanoscale phase-separated morphology based on the hydrophilic/hydrophobic differences in the polymer components. Unlike typical anion-exchange membranes found in the literature, hydroxide ion conductivity of the membranes did not increase with increasing IEC because of their high swelling capability in water. The membrane with IEC = 1.2 mequiv g-1 showed balanced properties of high hydroxide ion conductivity (81 mS cm-1 at 80 °C in water) and mechanical strength (>100% elongation and 14 MPa maximum stress at 80 °C, 60% relative humidity). The polymer main chains were stable in 4 M KOH for 1000 h, whereas the trimethylbenzyl-type ammonium groups degraded under the conditions to cause loss in the hydroxide ion conductivity. An H2/O2 fuel cell with the membrane with IEC = 1.2 mequiv g-1 exhibited a maximum power density of 242 mW cm-2 at 580 mA cm-2 current density.

Introduction

Polymer electrolyte fuel cell is one of the most attractive electrochemical systems for energy conversion. In particular, proton-exchange membrane fuel cells (PEMFCs) have already been commercialized for applications in stationary cogeneration systems and electric vehicles. However, the highly acidic PEMs require the use of precious metals such as Pt as electrocatalysts, which makes the PEMFCs less cost-effective.[1−3] In contrast, anion-exchange membrane fuel cells (AEMFCs) have gained growing attention because of the possible use of abundant non-platinum group metals as electrocatalysts.[4,5] However, the chemical instability and low anion conductivity of the existing AEMs have been major concerns for practical use.[6−8] To address these issues, a wide variety of aromatic polymers such as poly(arylene ether)s,[9,10] polystyrenes,[11,12] and poly(phenylene oxide)s[13,14] functionalized with quaternary ammonium groups have been studied. Among them, cardo structures such as fluorene and phenolphthalein derivatives seem attractive as the scaffold for the ammonium groups because multiple functionalization with ammonium groups is easily achievable. Furthermore, the bulky structures enable the formation of a large free volume in the membrane, which could serve as an ion-conducting pathway.[15−17] Previously, we have synthesized anion-conductive polymers containing ammonium-functionalized fluorene groups via a chloromethylation reaction.[18] The effect of the substitution number and position of ammonium groups on membrane properties was investigated in detail, which showed that a membrane having approximately two ammonium groups per fluorenyl group exhibited optimum membrane properties such as a well-developed phase-separated morphology, high hydroxide ion conductivity, and chemical stability. However, since the harmful chloromethylation reaction often accompanies unfavorable side reactions including the cross-linking reaction, the detailed molecular structure for the optimum membrane properties remain unclear. The objective of the present study was to synthesize a series of partially fluorinated polymers containing ammonium-functionalized fluorenyl groups with a well-defined structure. A preaminated monomer was designed such that each fluorenyl group was substituted with two ammonium groups at specific positions. The effect of the copolymer composition or ion-exchange capacity (IEC) on the membrane properties such as morphology, anion conductivity, mechanical strength, alkaline stability, and fuel cell performance was investigated. The properties were compared with those of our previously reported membranes prepared via the chloromethylation method.

Results and Discussion

Synthesis of Monomer 1 and Polymers 2 and 3

Bis(dimethylaminomethyl)fluorenyl monomer 1 was synthesized according to Scheme . First, p-BCF was chloromethylated with chloromethyl methyl ether (CMME) via Friedel–Crafts reaction using TiCl4 as the Lewis acid catalyst. Under optimized conditions, the reaction proceeded quantitatively and regioselectively at the 2 and 7 positions (or 4 in p-BCF in Scheme ) of fluorene groups as suggested by the 1H NMR spectrum (Figure ). The chloromethylated compound (p-BCF-cm) was aminated with dimethylamine in tetrahydrofuran (THF) solution. Complete amination reaction was confirmed by the 1H NMR spectrum, where the methylene proton (6) appeared at a higher magnetic field (3.4 ppm) compared with that of p-BCF-cm (4.6 ppm). Methyl protons (7) appeared at 2.2 ppm.

Scheme 1

Synthesis of Hydrophilic Monomer (1)

Figure 1

1H NMR spectra of p-BCF, p-BCF-cm, and 1 in CDCl3.

Synthesis of Quaternized Polymer 4

The synthetic procedure for the title quaternized polymer 4 is shown in Scheme . The precursor copolymer (3) was synthesized by Ni-promoted polycondensation reaction of the monomers 1 and 2. The reaction proceeded successfully as confirmed by 1H and 19F NMR spectra. In the 1H NMR spectrum (Figure a), all peaks were well-assigned to the supposed chemical structure. The copolymer composition estimated from the integral peak ratios were in good accordance with those of the feed comonomer ratios (Table ). In the 19F NMR spectrum (Figure S1a), three fluorine peaks were observed to confirm no side reactions in the hydrophobic component. A series of copolymers (3) with different compositions (m:n) were obtained with high molecular weight (Mn = 8.9–13.8 kDa, Mw = 13.7–24.5 kDa) and reasonable polydispersity (1.5–2.1). The quaternization reaction of 3 was carried out using dimethyl sulfate in N,N-dimethylacetamide (DMAc) solution. The progress of the reaction was confirmed by the shift of the methylene and methyl protons to the lower magnetic field in the 1H NMR spectra (Figure b). The 19F NMR spectrum (Figure S1b) did not show evidences of any side reactions. The resulting quaternized copolymer 4 was soluble in polar organic solvents such as dimethyl sulfoxide (DMSO) and DMAc, and provided brown and transparent membranes by casting from DMAc solution (thickness: 40–45 μm for 4-1, 50–70 μm for 4-2, 30–40 μm for 4-3, 30–40 μm for 4-4). The solvent solubility and membrane-forming capability of the copolymers were better than those of our previous copolymer membranes with a similar chemical structure prepared via the chloromethylation method.[18] Molecular weight measurements of 4 were unavailable because of the strong interaction with our gel permeation chromatography (GPC) columns. The IEC values of 4 membranes determined by titration ranged from 0.7 to 1.9 mequiv g–1, which were in fair agreement with those (0.7–1.7 mequiv g–1) calculated from the copolymer compositions obtained from the 1H NMR spectra of 3.

Scheme 2

Synthesis of 4

Figure 2

1H NMR spectra of (a) 3-3 in TCE-d2 and (b) 4-3 in CH3SO4– form in DMSO-d6.

Table 1

Composition and Molecular Weight of 3 and Ion-Exchange Capacity (IEC) and Water Uptake of 4

	composition (m:n)		molecular weighta (kDa)			IEC (mequiv g–1)
no.	feed	obtaineda	Mn	Mw	Mw/Mn	NMRb	titration	water uptakec (%)
3-1	1.00:0.20	1.00:0.19	8.9	13.7	1.5	0.7	0.7	22
3-2	1.00:0.35	1.00:0.35	13.0	20.1	1.6	1.0	1.0	42
3-3	1.00:0.45	1.00:0.47	13.8	23.8	1.7	1.4	1.2	59
3-4	1.00:0.65	1.00:0.62	11.6	24.5	2.1	1.7	1.9	152

Measured for 3.

Calculated from the 1H NMR spectra of 3.

Measured at room temperature (soaked in water for 24 h).

Water Uptake and Hydroxide Ion Conductivity

Water uptake and hydroxide ion conductivity of 4 membranes are shown in Figure . The water uptake increased nearly proportional to IEC. The highest IEC membrane (4-4) exhibited 152 wt % water uptake. The dimensional changes of the 4-3 membrane were 24% (through-plane) and negligibly small (in-plane), respectively. The conductivity also increased with IEC up to IEC = 1.2 mequiv g–1 (47 mS cm–1), and then decreased to 41 mS cm–1 with further increase in IEC. Taking the absorbed water into account, practical IEC values in water were 0.57 mequiv g–1 for 4-1, 0.79 mequiv g–1 for 4-2, 0.87 mequiv g–1 for 4-3, and 0.65 mequiv g–1 for 4-4 membranes, respectively, which explains the dependence of the conductivity on IEC. The 4-3 membrane with IEC = 1.2 mequiv g–1 showed the best balanced properties with the highest conductivity (47 mS cm–1) and a reasonably low water uptake (59%).

Figure 3

IEC dependence of (a) water uptake at room temperature (soaked in water for 24 h) and (b) OH– conductivity of 4 membranes in water at 30 °C.

Temperature dependence of the OH– conductivity in water is shown in Figure . All samples showed an approximate Arrhenius-type temperature dependence of the conductivity up to 80 °C. The apparent activation energies for the ion conduction calculated from the slopes were 7.1 kJ mol–1 for 4-1, 9.9 kJ mol–1 for 4-2, 10 kJ mol–1 for 4-3, and 8.2 kJ mol–1 for 4-4. These values were similar and comparable to those of our previous anion-conductive membranes[18] and typical for the conduction of hydrated hydroxide ions.

Figure 4

Temperature dependence of OH– conductivity of 4 membranes.

Morphology

The morphology of 4 membranes was analyzed by transmission electron microscopy (TEM) images (Figure ). The membranes showed a phase-separated morphology with small hydrophobic (bright areas) and hydrophilic (dark areas) domains. The domain sizes were both ca. 4–5 nm in all membranes. Compared with our previous membrane (5–8 nm) prepared via the chloromethylation method followed by quaternization,[18] the domain sizes and their distribution were slightly smaller. The results suggest that casting from quaternized polymers could provide membranes with more homogeneous morphologies than those of the membranes that were quaternized afterward.

Figure 5

TEM images of (a) 4-1 (b) 4-2 (c) 4-3, and (d) 4-4 membranes in PtCl4– forms.

Then, effect of the humidity on the morphology of 4 membranes was investigated by small-angle X-ray scattering (SAXS) analyses (Figure ). At 30% relative humidity (RH), the membranes showed a clear scattering peak at q = 0.54 nm–1 or d = 11.6 nm for 4-1, q = 0.66 nm–1 or d = 9.5 nm for 4-2, q = 0.69 nm–1 or d = 9.1 nm for 4-3, and q = 0.75 nm–1 or d = 8.4 nm for 4-4. The peak became larger in intensity and the d spacing also became larger with increasing humidity; q = 0.53 nm–1 or d = 11.9 nm for 4-1, q = 0.64 nm–1 or d = 9.8 nm for 4-2, q = 0.63 nm–1 or d = 10.0 nm for 4-3, and q = 0.72 nm–1 or d = 8.7 nm for 4-4 at 90% RH, respectively. Development of the peak on increasing the water uptake suggests that it is associated with periodic hydrophilic domains. The d-spacings observed in SAXS curves were larger than the cluster sizes in the TEM images because of the swelling with the absorbed water. The well-ordered periodic structure of the hydrophilic domains could be responsible for the high hydroxide ion conductivity of the membranes in water. The d-spacing decreased (Figure S2) and the peak became prominent with increasing IEC at both humidities. In the higher IEC membranes, smaller hydrophilic domains with homogeneous size were more likely to form.

Figure 6

SAXS profiles of (a) 4-1 (b) 4-2 (c) 4-3 (d) 4-4 membranes in Cl– forms at 40 °C, 30–90% RH.

Mechanical Properties

Mechanical properties of the membranes were evaluated by tensile strength (Figure ) and dynamic mechanical analyses (Figure ). The 4-1 membrane with a lower molecular weight was not available because of the insufficient strength for the analyses. Other membranes (4-2, 4-3, and 4-4) showed high elongation (>100%) at 80 °C and 60% RH. The 4-2 and 4-3 membranes showed reasonably high maximum stress (15 and 14 MPa, respectively); however, the 4-4 membrane with the highest IEC showed the lowest maximum stress (8 MPa) because of its highest water absorbability. In DMA analyses, three 4 membranes showed similar curves. The storage modulus E′ decreased with increasing temperature, and a broad peak was observed in the loss modulus E″ at ca. 70 °C. The peak would be associated with the glass transition of the polymers. Since the transition temperature was similar among the membranes with different IEC values (and thus, different water contents), it is presumably related to the hydrophobic components. The DMA properties of 4 membranes were similar to those of our previously reported membranes prepared via the chloromethylation method.[18]

Figure 7

Stress–strain curves of 4 membranes at 80 °C and 60% RH.

Figure 8

Temperature dependence of the dynamic mechanical properties of 4 membranes at 60% RH.

Thermal and Alkaline Stability

The thermal stability of the 4-3 membrane (in Cl– form) was measured by thermogravimetric analysis (Figure S3). Two-step weight loss was observed. The initial weight loss from 150 °C was ca. 10% corresponding to the amount of NMe3Cl groups (decomposition of the ammonium groups), whereas the second weight loss above ca. 400 °C would be due to the main chain degradation. The alkaline stability of the 4-3 membrane (in OH– form) was evaluated in 1 and 4 M KOH at 80 °C (Figure ). The hydroxide ion conductivity of the membrane decreased with testing time and the decrease was faster in 4 M KOH than in 1 M KOH. After 1000 h, the conductivity was 15 mS cm–1 (1 M KOH) and 13 mS cm–1 (4 M KOH) (the remaining percentage was 28 and 25%), respectively. The IEC of the postmortem membrane determined by titration also decreased to 0.7 mequiv g–1 under both conditions. Although the postmortem membrane retained bendability, it was insoluble in organic solvents. In the IR spectrum of the postmortem membrane (Figure ), the peak around 1600 cm–1 assigned to the C–H stretching vibration of methyl groups and the peak around 900 cm–1 assigned to the C–N+ stretching vibration were smaller. The changes suggest the chemical degradation of the ammonium groups and are not contradictory to the losses in the ion conductivity and IEC. Since the large peaks at 1100–1200 cm–1 assignable to C–F symmetric and asymmetric stretching vibration did not change, the polymer main chain seemed intact. The postmortem membrane was subjected to morphological analyses via SAXS (Figure ), in which humidity dependence of the scattering peaks was not observed. The results suggest that the decomposition of the ammonium groups (loss of IEC) eventually hampered morphological development in the quaternized membranes, and both of these caused decreased hydroxide ion conductivity.

Figure 9

Time course of OH– ion conductivity of the 4-3 membrane in 1 and 4 M KOH aqueous solutions at 80 °C.

Figure 10

Fourier-transform infrared spectra of the 4-3 membrane after the alkaline stability test.

Figure 11

SAXS profiles of the 4-3 membrane in Cl– form at 40 °C, 30–90% RH after the alkaline stability test.

Fuel Cell Performance

A catalyst-coated membrane was prepared with the 4-3 membrane (37 μm thick), our homemade QPAF-4 ionomer,[20] and the Pt/CB catalyst for the anode and cathode. The fuel cell was operated at 60 °C by supplying fully humidified pure H2 and O2 gases at a flow rate of 100 mL min–1 to the anode and the cathode, respectively. The current density/voltage (I/V) and current density/power density (I/W) curves, and ohmic resistance of the fuel cell are shown in Figure . The fuel cell showed a relatively high open circuit voltage (OCV, 0.96 V) typical for an H2/O2 alkaline fuel cell. The ohmic resistance was ca. 0.15 Ω cm2, which was approximately three times higher than the area-specific resistance of the membrane (0.05 Ω cm2) calculated from the hydroxide ion conductivity (67 mS cm–1 in water at 60 °C, Figure ) and the thickness. The difference in the resistance could be due to the lower hydroxide ion conductivity of the membrane under humidified conditions than in water. The contact resistance between the membrane and the catalyst layers could also be responsible to some extent. The fuel cell achieved high maximum power densities of 242 mW cm–2 at a current density of 580 mA cm–2.

Figure 12

(a) IV and IW performance and (b) ohmic resistance of the fuel cell using the 4-3 membrane at 60 °C.

Conclusions

A series of copolymers containing perfluoroalkyl and ammonium-functionalized fluorenyl groups were synthesized and characterized. Compared with our recent anion-exchange membranes with a similar chemical structure prepared via chloromethylation (which often accompanied unfavorable side reactions), the present copolymer membranes had better solubility in organic solvents and membrane-forming capability. Because of the well-defined hydrophilic structure where each fluorenyl group was substituted with two ammonium groups, the resulting polymer membranes exhibited a well-ordered phase-separated morphology as suggested by TEM images and SAXS analyses. The copolymer membranes exhibited a hydroxide ion conductivity that was less dependent on the gravimetric ion-exchange capacity (IEC) of the dry membranes, which was well-understood taking absorbed water into account. The optimum IEC value of the copolymer membranes was found to be 1.2 mequiv g–1 considering water uptake, hydroxide ion conductivity, and mechanical properties. In the accelerated alkaline stability test, the polymer main chain with no heteroatom linkages was stable; however, the trimethyl benzylammonium groups degraded to cause some losses in the ion conductivity. The hydrogen/oxygen fuel cell was operable under fully humidified conditions to obtain a high OCV and low ohmic resistance, as expected from the membrane properties. Using better-performing catalysts in alkaline media would further enhance the fuel cell’s performance.

Experimental Section

Materials

Titanium(IV) chloride (16–17% as Ti in dil. hydrochloric acid, Wako Chemical), chloromethyl methyl ether (CMME) (>95%, TCI), 40 wt % dimethylamine aqueous solution (Kanto Chemical), tetrahydrofuran (>99.5% Kanto Chemical), hydrochloric acid (35–37%, Kanto Chemical), 2,2′-bipyridine (>99%, TCI), bis(1,5-cyclooctadiene)nickel(0) (Ni(cod)2) (>95%, Kanto Chemical), and dimethyl sulfate (>99%, Kanto Chemical) were used as received. N,N-dimethylacetamide (DMAc) (>99%, Kanto Chemical) was dehydrated over molecular sieve 4A prior to use. 9,9′-Bis(4-chlorophenyl)fluorene (p-BCF) and bis(3-chlorophenyl)perfluorohexane (2) were synthesized according to the literature.[18] 1,6-Diiodoperfluorohexane was kindly supplied by Tosoh Finechem Co.

Synthesis of 2,7-Bis(chloromethyl)-9,9-bis(4-chlorophenyl)fluorene (p-BCF-cm)

A 100 mL one-neck round-bottomed flask equipped with a condenser, a nitrogen inlet/outlet, and a magnetic stirring bar was charged with p-BCF (5.00 g, 12.9 mmol), carbon disulfide (30 mL), and CMME (6.5 mL). To this mixture, titanium(IV) chloride (2.8 mL) in carbon disulfide (6 mL) was added to obtain a brown mixture solution. Stirring for 1 h at room temperature (rt) made the reaction mixture yellow. The reaction was monitored with TLC to confirm complete reaction. The reaction mixture was diluted with dichloromethane and a small amount of methanol, and filtered to remove the precipitate. The filtrate was evaporated and then purified by silica gel column chromatography (eluent: hexane/dichloromethane = 3:1). The obtained solution was evaporated to obtain a white powder (2.66 g, 42% yield). 1H NMR (500 MHz, CDCl3): δ 4.58 (s, 2H), 7.09 (d, J = 8.5 Hz, 1H), 7.22 (d, J = 8.0 Hz, 1H), 7.33 (s, 1H), 7.42 (d, J = 9.0 Hz, 1H), 7.74 (d, J = 7.5 Hz, 1H).

Synthesis of 2,7-Bis(dimethylaminomethyl)-9,9-bis(4-chlorophenyl)fluorene (1)

A 100 mL one-neck round-bottomed flask equipped with a magnetic stirring bar was charged with p-BCF-cm (2.00 g, 4.13 mmol), 40% dimethylamine aqueous solution (5.6 mL), and THF (40 mL). After stirring for 18 h at rt, the reaction mixture was extracted with dichloromethane three times, and then the combined organic layer was washed with sodium hydroxide aqueous solution and water twice. The organic layer was evaporated to obtain a white powder (1.68 g, 81% yield). 1H NMR (500 MHz, CDCl3): δ 2.20 (s, 3H) 3.41 (s, 2H) 7.11–7.7.12 (m, 1H) 7.16–7.19 (m, 1H) 7.24 (s, 1H) 7.30 (d, J = 7.4 Hz, 1H) 7.68 (d, J = 6.6 Hz, 1H).

Synthesis of Precursor Copolymers (3)

A typical synthetic procedure for the precursor copolymers is as follows: a 100 mL three-neck round-bottomed flask equipped with a condenser and a nitrogen inlet/outlet was charged with 1 (0.550 g, 1.1 mmol), 2 (1.80 g, 3.4 mmol), 2,2′-bipyridine (1.42 g, 9.1 mmol), and DMAc (18 mL). The mixture was heated at 80 °C to obtain a homogeneous mixture. To the mixture, Ni(cod)2 (2.50 g, 9.1 mmol) was added and the reaction was continued for 3 h at 80 °C. The reaction mixture was poured into 300 mL of methanol to precipitate a black powder. The crude product was washed with concentrated hydrochloric acid (300 mL) and then treated with saturated potassium carbonate aqueous solution (300 mL) two times. The resulting white powder was washed with deionized water two times, and dried at 50 °C in a vacuum oven overnight to obtain1.6 g of 3 (m:n = 1.00:0.37) in 74% yield.

Quaternization of 3 and Membrane Preparation

A typical procedure is as follows. A 50 mL round-bottomed flask was charged with 3 (1.50 g, 0.91 mmol of dimethylamino groups) and DMAc (10 mL). After dissolution of 3, dimethyl sulfate (3 mL, 32 mmol) was added. The mixture was stirred at 40 °C for 48 h, and then diluted with DMAc (10 mL). The mixture was poured into water (200 mL) to precipitate a white fibrous solid. The crude product was washed with water (200 mL) two times. The white solid was dissolved into 10 mL of DMAc and the solution was filtered with a syringe stuffed with cotton. The filtrate was cast onto a flat glass plate and dried at 50 °C overnight to obtain a pale brown transparent membrane. The resulting membrane was immersed in 1 M KOH aqueous solution for 24 h at 80 °C and then the membrane in OH– form was washed with degassed water for 1 day to remove the excess KOH. The membrane in Cl– form was prepared by soaking the OH– form membrane in 1 M HCl aqueous solution at 40 °C for 48 h.

Measurements

Characterization and property measurements such as NMR, GPC, hydroxide ion conductivity, mechanical properties, alkaline stability, and fuel cell evaluation were carried out according to the methods described in the literature.[19,20]