Nanohybrid membranes with hydroxide ion transport highways constructed from imidazolium-functionalized graphene oxide

Abstract

Graphene oxide (GO) and functionalized GO have been widely employed to design and fabricate polymer–inorganic nanohybrid materials for electrochemical applications. In this study, a series of imidazolium-functionalized graphene oxide (ImGO) nanosheets bearing different types of ligands (namely butyl, decyl, carbethoxy, and benzyl groups) on quaternary ammonium (QA) groups are prepared via distillation–precipitation polymerization and quaternarization, and then embedded into chitosan (CS) to fabricate nanohybrid membranes. The addition of ImGO significantly enhanced the thermal, mechanical, and anti-swelling stabilities of membranes due to the strong electrostatic attractions at CS/ImGO interface. More importantly, hydroxide ion transport highways were constructed at CS/ImGO interface via interfacial interactions. Meanwhile, the influence of the ligands on QA groups on the physicochemical properties, OH⁻ conductivity, and conduction mechanism is systematically elucidated. Due to the optimal hydrophilicity and ion exchange capacity, ImGO with carbethoxy group as the ligand confers the highest OH⁻ conductivity on nanohybrid membrane (up to 1.02 × 10⁻² S cm⁻¹ at 90 °C and 100% RH, about four times of that of CS control membrane). Correspondingly, a fuel cell with such a membrane shows an OCV of 0.71 V and a maximum power density of 75.8 mW cm⁻² at a current density of 298.8 mA cm⁻².

---

1. Introduction

Polymer electrolyte membrane fuel cells, which work at low temperature and efficiently convert the chemical energy stored in the fuel (e.g., H₂ or CH₃OH) directly into electrical energy, are currently being investigated for a variety of applications.¹ Based on the polymer electrolyte membranes, the fuel cells can be classified into two types, that is, proton exchange membrane fuel cells (PEMFCs) and anion exchange membrane fuel cells (AEMFCs), where the membranes conduct proton (H⁺) and hydroxide ion (OH⁻), respectively.² In recent years, extensive efforts have been devoted to developing AEMFCs due to the following advantages over PEMFCs: (i) the fast fuel cell reaction kinetics under alkaline conditions;³ (ii) the improved water management since the electro-osmotic drag transports water away from the cathode, and (iii) the lowered fuel ‘crossover’ due to OH⁻ transportation from cathode to anode.⁴ The anion exchange membranes (AEMs), serving as the carrier of OH⁻ and the barrier of fuel/oxidant simultaneously, determine the properties of AEMFCs to a great extent. Therefore, it is extremely urgent to develop AEMs with favorable properties including high OH⁻ conductivity, low fuel crossover and excellent thermal/chemical stabilities.⁵

Up to now, AEMs are mainly fabricated by quaternary ammonium (QA) functionalized polymers,^6–12 in which QA groups work as OH⁻ carriers. However, there is a trade-off effect between the structural stability and OH⁻ conductivity for this type of AEMs.¹³ Considering this, tremendous efforts have been devoted to developing long-term AEMs with high conductivity and stability, such as cross-linking,¹⁴ blending and hybridizing.¹⁵ Among them, organic–inorganic hybridization¹⁶ by incorporating conductive nanofillers into polymer matrix to fabricate nanohybrid membranes has been employed, in which the swelling of inorganic nanofillers is quite low, and the inorganic nanofillers can suppress the chain mobility in polymer bulk. For instance, Li et al.¹⁶ synthesized carbon nanotubes functionalized with an imidazolium-type ionic liquid, incorporating which has produced an 82.9% increase of tensile strength together with a 95.3% increase in ionic conductivity. Although several studies have been conducted on organic–inorganic nanohybrid AEMs recently, more detailed investigations remain to be conducted.

Generally, the physical structure and chemical structure of inorganic fillers strongly influence their conduction property, which subsequently affect the OH⁻ conductivity of the resultant AEMs. The physical structure mainly determines the transfer pathways created by the surface conducting groups on the fillers; while the chemical structure plays a key role in transporting rate and energy barrier for each hopping of OH⁻. Based on the physical structure, the fillers can be classified as spherical (0D), tubular (1D), and sheet (2D) ones.^17–19 2D sheet filler with larger surface area and higher aspect ratio can donate more wide and long-range pathways under the same filler content, as has been experimentally proved in proton conducting membranes.²⁰ For instance, Kim et al.²¹ employed sulfonated montmorillonite (SMMT) to enhance the proton conductivity of sulfonated poly(arylene ether sulfone) membranes, and noted that 10% SMMT conferred a 35.5% increase in conductivity. Among various 2D materials, graphene oxide (GO), which has unique plane structure, excellent thermal stability, and mechanical stability, has attracted intensive attention.¹⁹ The abundant oxygen-containing groups (carbonyl, hydroxyl, carboxyl, and epoxy groups) on GO surface make it easy to be modified to acquire the needed physicochemical properties.²² More recently, Zarrin et al.²³ developed a fast hydroxide conductor with a chemically modified GO nanocomposite membrane, which exhibited outstanding hydroxide conductivity accompany with high physicochemical strength. In addition to physical structure, the chemical structure of inorganic fillers may also dominate the conduction ability of the AEMs through the following two ways: (i) hydrophilic environment near the conductive site would facilitate the transporting rate of OH⁻ by providing more bound water and attracting OH⁻ ions; (ii) the ligand in conductive group would induce steric hindrance effect on the approaching of OH⁻.²⁴ Recently, varieties of functional groups, such as guanidinium,²⁵ phosphonium,²⁶ imidazolium,²⁷ benzimidazolium²⁸ and metal-cation,²⁹ are being proposed. Among them, the imidazolium functional groups are thought to be environment friendly and to some extent alkaline stable due to the presence of the π-conjugated imidazole ring, as proved in the literature.²⁷ Besides, the ligand in imidazole ring has a profound influence on the conduction ability of functional groups.³⁰ Nevertheless, few efforts have been devoted to the nanohybrid membranes containing imidazolium groups, and the influence of ligand in imidazole ring on the conduction ability of membrane need to be further investigated.

In this study, imidazolium-functionalized graphene oxide (ImGO) nanosheets were synthesized via distillation–precipitation polymerization and subsequent quaternization for the first time. Four chlorinated reagents (chlorobutane (CB), chlorodecane (CD), ethyl chloroformate (EC), and benzyl chloride (BC)) were chosen in the quaternization reaction to synthesize four kinds of QA groups on ImGO surface. CB and CD owned the same electronic inductive effect on imidazole group but with different length of carbon chain, while EC and BC possessed different electron inductive effect. CS was chosen as the nanohybrid membrane matrix on account of its excellent film forming, high mechanical and fuel-barrier properties. The microstructure and physicochemical characteristics of the resulting nanohybrid membranes were investigated in detail. Moreover, the OH⁻ conductivity and transfer mechanism of the nanohybrid membranes were systematically evaluated. By doing so, the relationship between the structure of QA groups in membrane and OH⁻ conductivity was elucidated. For practical applications of the nanohybrid membranes, the single-cell performances were evaluated in this work.

2. Experimental section

2.1 Materials

Natural graphite powders (∼45 μm) were purchased from Sinopharm Chemical Reagent. CS with the degree of deacetylation of 91% was supplied by Golden-Shell Biochemical Co. (Zhejiang, China) and used as received. 3-(Methacryloxy)propyltrimethoxysilan (MPS) was obtained from Aldrich and distilled under vacuum. Ethyleneglycol dimethacrylate (EGDMA) was purchased from Alfa Aesar and used without any further purification. 2,2′-Azobisisobutyronitrile (AIBN) and acetonitrile were purchased from Kewei Chemistry Co., Ltd (Tianjin, China). N-Vinyl imidazole (VI), chlorinated reagents (CB, CD, EC, BC) were purchased from Xiya reagent (Shandong, China). De-ionized water was used throughout the experiment.

2.2 Synthesis of GO and ImGO

GO nanosheets were synthesized by oxidizing natural graphite powders according to the improved technique in the literature.³¹ ImGO was prepared through distillation–precipitation polymerization³² and quaternization method as illustrated in Scheme 1: (i) the graft of MPS onto the surface of GO to introduce reactive vinyl groups; (ii) the formation of polymeric layer (poly(EGDMA-co-VI)) via distillation–precipitation polymerization, in which VI donated imidazole ring while EGDMA worked as cross-linker; and (iii) the synthesis of quaternary ammonium imidazole cationic through Menshutkin reaction.

Scheme 1 Preparation of ImGO and the microstructure of nanohybrid membranes.

Specifically, GO (3.0 g) was dispersed into ethanol (200 mL) by ultrasonic and stirring treatment in a dried flask for more than 2.5 h. Water (12.5 mL) and aqueous solution of ammonium (30%, 19 mL) were added into the above solution and stirred vigorously at 25 °C for 24 h. Afterwards, MPS (1.0 mL) was added into the resultant mixture. After being stirred for another 32 h, the MPS-modified GO (MGO) was purified by centrifugation and followed by drying in a vacuum oven. MGO (0.15 g), VI (0.50 mL), EGDMA (0.30 mL), and AIBN (0.02 g) were dissolved by ultrasonic treatment in acetonitrile (80 mL) in a dried flask. The mixture was heated and kept in boiling state until half acetonitrile was distilled out. Afterwards, the VI modified GO (VI-GO) was purified and dried for quaternization. VI-GO (0.6 g) was dispersed into absolute alcohol (100 mL), and then a certain amount of quaternary aminating reagent was added into the above solution and refluxed under 80 °C for 8 h. The resultant ImGO was purified by centrifugation and washed by ethanol, then dried in a vacuum oven. The synthesized ImGO were named as C₄-ImGO, C₁₀-ImGO, E-ImGO and B-ImGO, corresponding to the four kinds of used quaternary aminating reagents (CB, CD, EC, and BC), respectively.

2.3 Preparation of AEMs

A certain amount of GO or ImGO was dispersed into deionized water (30 mL) with ultrasonic treatment for 24 h. Meanwhile, CS (1.2 g) was dissolved in acetic acid aqueous solution (2%, 30 mL) and stirred for 2 h at room temperature. Afterwards, these solutions were mixed together. Subsequently, a given amount of cross-linker glutaraldehyde (GA, 25%, v/v) aqueous solution was added into the mixture and then stirred vigorously for another 3 h. The resultant homogenous solution was cast onto a clear glass plate and dried at 30 °C for 72 h to obtain a nanohybrid membrane. Afterwards, the membrane was immersed in 1.0 M KOH solution for 48 h to obtain the OH⁻ form, and then washed thoroughly with deionized water to remove the residual KOH until the pH of residual water was neutral. Finally, the membrane was dried at 40 °C until a constant weight. The nanohybrid membranes were designated as CS/GO-X, CS/C₄-ImGO-X, CS/C₁₀-ImGO-X, CS/E-ImGO-X and CS/B-ImGO-X representing GO, C₄-ImGO, C₁₀-ImGO, E-ImGO and B-ImGO as the nanofillers, where X (X = 0.5, 1.0, 1.5, 2.0, and 2.5) referred to the weight percentage of nanofillers to CS. CS control membrane was fabricated in exactly the same way as described above but without incorporating nanosheets. The average thickness of the dry membranes fell in the range of 42–60 μm.

2.4 Characterization of GO, ImGO, and AEMs

The morphology of GO and ImGO was observed by transmission electron microscopy (TEM, TecnaiG220S-TWIN). Fourier transform infrared (FTIR, Nicolet MAGNA-IR560) was measured with a resolution of 4 cm⁻¹ in the range of 4000–400 cm⁻¹ at room temperature. Thermo gravimetric analysis (TGA) was conducted by TGA-50 SHIMADZU from 30 to 800 °C at a heating rate of 10 °C min⁻¹ under N₂ atmosphere. The microstructure of the membrane was observed using scanning electron microscope (SEM, JSM 7500F) after being freeze-fractured in liquid nitrogen and then sputtered with gold. The mechanical property of the membrane (1.0 cm × 4.0 cm) was performed by an instron mechanical tester (Testometric 350 AX) with an elongation rate of 2 mm min⁻¹ at room temperature. RigakuD/max2500v/Pc X-ray diffraction (XRD, CuK 40 kV, 200 mV) was utilized to characterize the structure of the nanosheets and membranes.

2.5 Measurement of water uptake and area swelling

Water uptake of the membrane was determined by measuring the change in weight between the dry and wet membrane. The membrane was first dried at 80 °C for 48 h and then weighed (W_dry). Afterwards, the membrane was immersed in deionized water for 48 h at 25 °C to be completely saturated. Finally, it was taken out and re-weighed (W_wet) after removing the surface liquid droplets. The value of water uptake was the average of three measurements with an error within ±4.0% and calculated from eqn (1):

Water uptake (%) = (W_wet − W_dry)/W_dry × 100 (1)

Area swelling of the membrane was determined by a similar method: soaking the pre-measured membrane (A_dry, cm²) in water for 48 h at 25 °C and then re-measuring the wet membrane area (A_wet, cm²). Area swelling was defined as:

Area swelling (%) = (A_wet − A_dry)/A_dry × 100 (2)

2.6 Evaluation of ion exchange capacities (IEC) and ionic conductivity

IEC values of the AEMs were determined by the back titration method.³³ The obtained anion exchange membranes in OH⁻ form were dried in a vacuum oven at 80 °C for 48 h to reach a constant weight prior to the test. Afterward, the AEMs were immersed into HCl solution (0.1 mol L⁻¹, 20 mL) at room temperature for 24 h. The residual solution was then back titrated with NaOH (0.1 mol L⁻¹) standard solution using phenolphthalein as an indicator. The HCl solution (0.1 mol L⁻¹, 100 mL) was used as the blank sample for the control experiment. The IEC values (mmol g⁻¹) could be calculated from the following expression:

IEC (mmol g⁻¹) = (C_NaOH × V_0,NaOH − C_NaOH × V_X,NaOH)/W_dry (3)

where V_0,NaOH and V_X,NaOH were the volume of the NaOH consumed in the titration without and with membranes, respectively, C_NaOH was the mol concentration of the NaOH, and W_dry was the mass of the dried membrane. Three replicates were conducted for each sample.

The OH⁻ conductivity of the membrane was measured in a conductivity cell by alternating current (AC) impedance spectroscopy method. The membrane impedance was tested by a frequency response analyzer (FRA, Compactstat, Ivium Tech.) with oscillating voltage of 20 mV over a frequency range of 10⁵ Hz to 100 Hz. To test OH⁻ conductivity under hydrated condition, all the samples were immersed in deionized water for 48 h to be fully hydrated prior to the measurement. Then, the sample was put in the cell and heated by water vapor under a certain temperature ranging from 30 °C to 90 °C. The relative humidity (RH) was kept at 100% throughout the test. The sample was kept for some time until the resistance became a constant value. OH⁻ conductivity under anhydrous condition was tested using dry air after the membrane was completely dried at 60 °C for 24 h under high vacuum. The OH⁻ conductivity (σ, S cm⁻¹) of the sample was calculated by the eqn (4):

σ = l/AR (4)

where l, A, and R were the membrane thickness (cm), membrane area (cm²), and membrane resistance (Ω), respectively.

2.7 Membrane electrode assembly (MEA) and fuel cell performance measurements

The catalyst ink was prepared by mixing 20% PtRuO₂/WCNTs with a solution of 5% Nafion (DuPont) and isopropanol, and then sonicated for 4 h to get a homogeneous solution, where the ratio of Pt/C catalyst to Nafion was 3 : 1. The catalyst ink was sprayed onto the carbon paper (Toray TGP-H-090) to deposit a catalyst layer with a Pt loading of 10 mg cm⁻² for both the anode and the cathode. Then four drops of alkaline ionomer, Tokuyama AS-4, were loaded on the catalyst layers.

The MEAs were fabricated by hot-pressing the OH⁻ form CS/E-ImGO-2.0, CS/GO-2.0, and CS control membrane with the catalyst loaded carbon paper at a pressure of 6 MPa at 60 °C for 3 min. The MEAs were evaluated in a single fuel cell with an active area of 3 × 3 cm² using an electronic load. The single cell testing was performed at 60 °C and 100% RH under atmospheric pressure on an electronic load (ZY8714, ZHONGYING Electronic Co., Ltd.) with serpentine flow channels. During the testing, the anode was fed with a 5.0 M methanol solution or 5.0 M ethanol solution at a flow rate of 1.0 mL min⁻¹ and the cathode was fed with air at a flow rate of 400 mL min⁻¹. Polarization curves were obtained using a fuel cell evaluation system (GE/FC1-100).

3. Results and discussion

3.1 Synthesis and characterization of ImGO

The morphological features of GO and ImGO were observed in their TEM images. As shown in Fig. 1a, GO appeared as an exfoliated sheet structure with some wrinkles. These wrinkles were probably attributed to the agglomeration of GO driven by the interlayer attractions. The morphology of the nanosheets was only slightly altered after the surface modification since the imidazolium-functionalization reaction mainly occurred on poly(EGDMA-co-VI) layer, allowing ImGO remaining the sheet structure (Fig. 1b) without obvious destruction. For ImGO, the interlayer interaction was weakened on account of the imidazolium-functionalization, thereby reducing the contact area and avoiding their stacking to some extent.

Fig. 1 TEM images of (a) GO, and (b) ImGO.

To confirm the successful surface modification of GO, FTIR and TGA of GO and ImGO were conducted. The FTIR spectrum of GO (Fig. 2a) showed that all the characteristic bands corresponding to oxygen-containing functional groups, including 1720 cm⁻¹ (stretching vibrations from CO), 1620 cm⁻¹ (broad-coupling from O–H), and 1384 cm⁻¹ (deformation from C–OH) were observed. The band at around 3430 cm⁻¹ could be clearly observed, which was assigned to the stretching of –OH on the surface of GO. Compared with GO, new absorption bands at 752 cm⁻¹ and 2817–3281 cm⁻¹ were observed for ImGO, which were attributed to the bending vibrations of imidazolium ring³⁴ and the QA cation, respectively. By comparison, the band intensity of C₄-ImGO and C₁₀-ImGO at 3046–2890 cm⁻¹ (typical stretching of C–H) became stronger due to the higher content of C–H from butyl and decyl. E-ImGO gave rise to the strongest character bands of carbonyl group at 1720 cm⁻¹ due to the introduction of ester carbonyl group. B-ImGO gave rise to several weak characteristic bands in the range of 1410–1590 cm⁻¹, which could be assigned to the typical vibration of phenyl group. These results indicated that imidazole rings grafted on poly(EGDMA-co-VI) were turned to QA cations successfully through the quaternization.

Fig. 2 (a) FTIR spectra, (b) TGA curves, and (c) XRD patterns of GO and ImGO.

TGA results (Fig. 2b) revealed that GO exhibited three dominant degradation stages. The first stage (30–160 °C) was mainly attributed to the evaporation of moisture (mainly bonded water), and the second stage (180–280 °C) was due to the deoxygenation of GO (mainly oxygen-containing groups). The third stage (280–800 °C) was from the decomposition of nanosheets backbone (mainly C–C bonds).³⁵ By comparison, the imidazolium-functionalized polymeric layer afforded enhanced water holding ability to ImGO, leading to a slightly higher weight reduction during the first stage. During the second stage, the coverage of cross-linked polymeric layer prevented the deoxygenation of ImGO. Therefore, the weight loss was assigned to the degradation of polyelectrolyte polymer on ImGO surface. In addition, according to the char yields of GO (32.2%), C₄-ImGO (20.5%), C₁₀-ImGO (16.3%), E-ImGO (11.3%), and B-ImGO (18.5%), the weight percentages of polyelectrolyte polymer in C₄-ImGO, C₁₀-ImGO, E-ImGO, and B-ImGO could be calculated to be approximately 11.7%, 15.9%, 20.9% and 13.7%, respectively. To sum up, the above analyses suggested that ImGO were successfully synthesized.

Fig. 2c showed the XRD spectra of GO and ImGO. Different from the diffraction peak of graphite at 27° (001) with the interlayer distance of 0.33 nm,³⁶ the (001) peak of GO shifted to 9.5° due to the inserting of oxygen-containing groups and adsorbed water, which made the interlayer distance increase to 0.93 nm.³⁷ After modification, this peak shifted to 10.5° for C₄-ImGO, 11.1° for C₁₀-ImGO, 11.0° for E-ImGO, and 11.1° for B-ImGO, respectively, corresponding to the interlayer distances of 0.84 nm, 0.80 nm, 0.80 nm, and 0.78 nm. The slight decrease of the interlayer distance could be ascribed to the enhanced interlayer interaction due to the entering of QA salts into GO interlayer during the modification.

3.2 Preparation and characterization of AEMs

To investigate the internal morphologies and filler dispersion of AEMs, the SEM images of cross-sectional membranes were observed. As shown in Fig. 3a, all the membranes are relatively dense, uniform and defect-free. Compared with CS control membrane, the cross-section of the nanohybrid membranes became rougher with obvious wrinkles as a result of the presence of nanosheets (Fig. 3b–g). For the preparation of SEM samples, all the membranes were freeze-fractured in liquid nitrogen, during which the nanohybrid membrane would be inclined to break along the surface of nanosheets, making the cross-section rough.³⁸ Besides, the overall morphology of the nanohybrid membranes was uniform without cracks. This observation suggested that the nanosheets (GO and ImGO) were well-dispersed within CS matrix, driven by the interfacial interactions and the subsequent good interfacial compatibility. The presence of imidazolium-functionalized polymer layer would help the dispersion of ImGO, and therefore ImGO-filled membranes possessed more wrinkles than GO-filled membranes. The good dispersion of ImGO would provide more CS-ImGO interfacial domains and construct more hydroxide ion transport highways. Because of the fact that too high GO/ImGO content would lead to serious aggregations, the filler content was kept below 2.5% in this study. Besides, SEM image of the full cross-section area of membrane was also taken to measure membrane thickness. As shown in Fig. 3g, the thickness was measured to be around 50 μm.

Fig. 3 SEM images of the cross-section of (a) CS, (b) CS/GO-1.5, (c) CS/C4-ImGO-1.5, (d) CS/C10-ImGO-1.5, (e) CS/E-ImGO-1.5, (f) CS/B-ImGO-1.5, and (g) CS/E-ImGO-1.5 (full cross-section area).

FTIR spectra were used to characterize the chemical structure of CS control and nanohybrid membranes. As shown in Fig. 4, three characteristic bands around 3358 cm⁻¹, 1648 cm⁻¹ and 1575 cm⁻¹ could be clearly observed for all the membranes. Compared with CS control membrane, the intensities of these characteristic bands in the spectra of nanohybrid membranes became weak, which could be attributed to the generation of hydrogen-bonding interaction (–OH/–NH₂ in CS and oxygenated groups in GO/ImGO). Thus, the number of free –OH/–NH₂ groups on CS chains was reduced (Scheme 1).³⁹ Besides, no new band appeared in the spectra of nanohybrid membranes, indicating that GO and ImGO were physically mixed with CS matrix with no chemical bond. The interferences of GO/ImGO meanwhile disordered packing of CS chains and then reduced the crystalline domains (see XRD in Fig. S1†).

Fig. 4 FTIR spectra of CS control and nanohybrid membranes.

3.3 Thermal and mechanical properties of AEMs

As one of the key components of AEMFCs, AEMs must process excellent thermal and mechanical stabilities for long-term operation. TGA measurements were carried out to investigate the thermal properties of the as-prepared membranes. Fig. 5a suggested that all the membranes underwent the following three-step weight loss: the first stage was the evaporation of water adsorbed in the membrane around 30–150 °C; the second stage was the degradation of CS side-chains around 160–265 °C; and the third stage was the degradation of CS backbones around 280–800 °C. It could be observed that CS/GO-1.5 displayed a similar thermal degradation behavior to that of CS control membrane, inferring that the degradation mechanism of CS might remain unchanged after incorporating GO. However, the TGA curves of CS/ImGO-1.5 were retarded at the second (see Fig. 5b) and third stages when compared with that of CS control membrane. The interference effect from ImGO, which delayed the degradation of CS chains by inhibiting their motion, should contribute to the above change. Besides, elevating the content of ImGO would give slight enhancement of the thermal stabilities for the nanohybrid membranes, as proved in Fig. S2a.† Interestingly, when the content of ImGO reached 2.5%, an obvious enhancement of the thermal stability was observed. This could be attributed to the strong interference of ImGO for CS matrix at this high content. In general, all of the nanohybrid membranes exhibit good thermal stabilities below 200 °C, suggesting their potential applications in AEMFCs (usually 60–80 °C).

Fig. 5 (a and b) TGA curves and (c) stress–strain curves of CS control and nanohybrid membranes.

Fig. 5c presented the mechanical behaviors of the as-prepared membranes in terms of their stress–strain curves. CS control membrane exhibited a Young's modulus of 984.8 MPa, along with the tensile strength of 44.74 MPa and the elongation at break of 17.74%. By comparison, the nanohybrid membranes acquired enhanced mechanical stabilities as shown in Table 1. Such enhancements of mechanical properties were reasonably ascribed to the interfacial interferences, which inhibited the mobility of CS chains and made their stretching more difficult. Specifically, CS/C₄-ImGO-1.5 and CS/C₁₀-ImGO-1.5 exhibited significantly enhanced Young's modulus of 1450.0 and 1509.4 MPa, respectively. By comparison, CS/E-ImGO-1.5 and CS/B-ImGO-1.5 displayed the Young's modulus of 1269.9 and 1236.5 MPa, a little lower than those of the former ones. The reason might be that C₄-ImGO and C₁₀-ImGO possessed stronger interfacial interferences with CS matrix than E-ImGO and B-ImGO. The side chains on imidazole ring of C₄-ImGO and C₁₀-ImGO were straight-chain paraffin, which could help the functional groups on nanosheets form hydrogen bond with the –OH/–NH₂ groups on CS chains.⁴⁰ This would increase the rigidity of CS/C₄-ImGO-1.5 and CS/C₁₀-ImGO-1.5, resulting in higher Young's modulus. Besides, the mechanical strength of nanohybrid membranes roughly increased with the nanosheets content (Fig. S2b†).

Table 1 Mechanical Properties of the AEMs

Membrane	Tensile strength (MPa)	Young's modulus (MPa)	Elongation at break (%)
CS	44.74	984.8	17.74
CS/GO-1.5	54.57	1215.0	5.72
CS/C4-ImGO-1.5	56.17	1450.0	10.15
CS/C10-ImGO-1.5	56.69	1509.4	11.30
CS/E-ImGO-1.5	55.20	1269.9	8.31
CS/B-ImGO-1.5	55.93	1236.5	9.77

---

3.4 Water uptake and area swelling of AEMs

The presence of water in the membranes is a prerequisite for high OH⁻ conductivity since water molecules participate in OH⁻ transfer mechanisms, including Grotthuss mechanism, diffusion and mechanism. Besides, water cluster can work as transport channel for OH⁻ within AEMs.⁴¹ However, excessive water uptake will lead to severe swelling and cause deformation of membrane and in turn increase the fuel crossover. Fig. 6a revealed that CS control membrane showed a high water uptake of 92.0% at 25 °C due to the hydrophilic groups (i.e., –OH and –NH₂). Incorporating nanosheets would inhibit the chain motion of CS matrix, and therefore the nanohybrid membranes exhibited slightly reduced water uptakes. For instance, the uptakes of CS/GO-1.5, CS/C₄-ImGO-1.5, CS/C₁₀-ImGO-1.5, CS/E-ImGO-1.5, and CS/B-ImGO-1.5 were 88.4%, 79.5%, 79.8%, 85.7%, and 83.8%, respectively, following the order of CS/C₄-ImGO-X < CS/C₁₀-ImGO-X < CS/B-ImGO-X < CS/E-ImGO-X, in accordance with the polarity order of the ligand on QA groups (C₄-ImGO < C₁₀-ImGO < B-ImGO < E-ImGO). Accordingly, the reason might be that the hydrophilic ability of the imidazole cations was enhanced with the increase of polarity of the side chains. The increased content of nanosheets would further decrease the water absorption ability of nanohybrid membranes. For instance, when elevating the content of C₁₀-ImGO from 0.5% to 2.5%, the uptake of CS/C₁₀-ImGO-X decreased from 86.1% to 76.4%. Nevertheless, it was noteworthy that the uptakes were still maintained at moderate values, which were adequate for the application of AEMs.⁴¹ In general, the adsorption of water would result in the swelling of the membrane through hydration effects, and thereby the addition of nanosheets reduced the area swellings of nanohybrid membranes as a result of the reduction of water uptakes. The reduction of swelling upon ImGO incorporation inferred the enhanced structural stabilities of nanohybrid membranes.

Fig. 6 (a) Water uptake, (b) area swelling, and (c) IEC values of CS control and nanohybrid membranes at 25 °C.

3.5 IEC and ionic conductivity of AEMs

3.5.1 IEC of membranes. IEC values of ionic-conductive membranes usually reflect the amount of exchangeable groups, and a relatively higher IEC is normally correlative to a higher ionic conductivity.⁴² Fig. 6c revealed that CS control membrane attained a relatively low IEC value of 0.21 mmol g⁻¹. In contrast, incorporating ImGO elevated the IEC values of CS-based membranes, and the IEC values of CS/ImGO-X increased as the nanosheets content increased. For instance, the IEC values of CS/E-ImGO-X increased from 0.34 to 0.45 mmol g⁻¹ as the ImGO content increased from 0.5% to 2.5%. The increased IEC values could be explained by the following two facts: (i) the addition of charge carriers from the imidazolium-functionalized polymer layer on ImGO, which consequently increased the number of exchangeable OH⁻; (ii) the reduced crystalline degree of CS matrix, which promoted the dissociation of OH⁻ from the basic groups. IEC values of ImGO varied with the type of surface functional groups on ImGO and followed the order of CS/E-ImGO > CS/C₄-ImGO > CS/C₁₀-ImGO > CS/B-ImGO under identical conditions. For alkyl substituent groups (butyl and decyl), positive charge was concentrated on the N atoms, which had strong affinity ability for OH⁻. However, excessively long carbon chain would decrease the IEC values as obvious steric hindrance emerged. For CS/B-ImGO, the positive charge was dispersed by the delocalization effect of benzene ring, which weakened their affinity ability to OH⁻. CS/E-ImGO displayed the highest IEC value among the nanohybrid membranes, since ester groups would be partially hydrolyzed to carboxyl group. Such type of imidazolium group possessed the characteristics of zwitterions, which were beneficial to the transfer of OH⁻. The increase of IEC values might allow CS/ImGO-X transporting more OH⁻ by providing more available hydroxide hopping sites.

3.5.2 Ionic conductivity of AEMs under hydrated condition. The ionic conductivity of the AEMs is of particular importance and plays a significant role on AEMFCs performance. Herein, the OH⁻ conductivities of the AEMs were measured under hydrated (100% RH) condition at different temperatures. Fig. 7a presented the OH⁻ conductivities of the membranes at 30 °C and 100% RH. As we could see, CS control membrane attained a conductivity of about 1.00 × 10⁻³ S cm⁻¹, close to the result reported in the literature.⁴³ By comparison, CS/ImGO-X displayed higher OH⁻ conductivities. For instance, CS/E-ImGO-0.5 obtained a conductivity of 3.16 × 10⁻³ S cm⁻¹, with an enhancement of 213%. Considering the reduced water uptake, this increased conductivity might be due to the following few reasons. First, the filled ImGO created additional wide and long-range conducting highways using the sufficient QA groups and the adsorbed water molecules. Second, the interfacial interactions might induce the hydrophilic groups on CS chains to arrange along the ImGO surface, efficiently conducting OH⁻ through a shorter route. Third, the reduced crystallinity in CS matrix might accelerate the OH⁻ migration through the CS phase. It was observed that the OH⁻ conductivities of the nanohybrid membranes increased with the increase of ImGO content (lower than 2.0%) and then decreased with the ImGO content higher than 2.5%. The decreased conductivity might be related to the aggregation of ImGO within membrane, leading to the generation of defective zones in nanohybrid membranes, which reduced the number of transfer pathways.⁴⁴ Similar to the functions of ImGO, the addition of GO enhanced the conduction ability of GO-filled membranes; while the lack of QA groups made their conductivities lower than those of ImGO-filled membranes. Also the conductivity of CS/GO-X first increased and then decreased with the GO content.

Fig. 7 Ionic conduction properties of nanohybrid membranes: (a) OH⁻ conductivity at 30 °C and 100% RH, (b) temperature-dependent conductivity, (c) Arrhenius plots of OH⁻ conductivity under 100% RH, and (d) OH⁻ conductivity at 30 °C and 0% RH.

For CS/ImGO-X, the OH⁻ conductivities differed on account of different transfer sites and obeyed the order of CS/B-ImGO-X < CS/C₁₀-ImGO-X < CS/C₄-ImGO-X < CS/E-ImGO-X. The OH⁻ conductivities of CS/C₄-ImGO-X were higher than those of CS/C₁₀-ImGO-X with the same content. Considering that butyl and decyl owned the similar electronic inductive effect on imidazole group but with different length of carbon chain, this finding suggested that the longer the carbon chain on hopping sites (i.e., imidazole cation groups) was, the stronger the space steric effect for OH⁻ would be. Therefore, OH⁻ transmission through C₁₀-ImGO might have higher –C(O)–O–CH₂CH₃ of CS/E-ImGO-X possessed stronger electron inductive effect. For QA alkaline groups, the promotion of electron cloud density of nitrogen atom would be in favour of the dissociation of OH⁻, resulting in an enhanced alkalinity for QA group. However, the result showed that the OH⁻ conductivities of CS/E-ImGO-X were higher than those of CS/B-ImGO, probably due to the hydrolyzation of –C(O)–O–CH₂CH₃ when being immersed into KOH solution. In such a way, –C(O)–O–CH₂CH₃ would be partially turned to –COO⁻, a strong electron donating group, which had the capability to enhance the electron cloud density of N atom and then lower the energy required for OH⁻ dissociation.

The temperature dependence of ionic conductivities of the nanohybrid membranes was examined under fully hydrated conditions, and the results were shown in Fig. 7b and S3.† The OH⁻ conductivities of the nanohybrid membranes gradually increased with the temperature because of the promotion of chain motion and hence the increase of free volume in membrane matrix.⁴⁵ Several possible dominant transport mechanisms for OH⁻ have been suggested, including Grotthuss mechanism, surface site hopping, diffusion, and convent ion via permeation and osmotic drag.⁴⁶ The Grotthuss mechanism was the usual mechanism given for facilitated proton mobility. However, OH⁻ was also considered to be able to exhibit Grotthuss behavior in aqueous solutions, comparable to protons.² It was noteworthy that the nanohybrid membranes displayed higher OH⁻ conductivities at each testing temperature than CS control membrane. For CS control membrane, the conductivity increased from 1.00 × 10⁻³ (30 °C) to 2.50 × 10⁻³ S cm⁻¹ (90 °C), while for the CS/E-ImGO-1.5, the conductivity increased from 4.85 × 10⁻³ (30 °C) to 9.05 × 10⁻³ S cm⁻¹ (90 °C). This phenomenon demonstrated that the incorporation of ImGO had a great effect on the improvement of the OH⁻ conductivity of the nanohybrid membranes. To further investigate the energy needed for OH⁻ hopping, the OH⁻ conductivities in Arrhenius plot were depicted in Fig. 7c, which could be utilized to calculate the values of activation energy (E_a) for hydroxyl ion transfer. CS control membrane had the E_a values of 13.4 kJ mol⁻¹. By comparison, the E_a values of CS/GO-1.5, CS/C₄-ImGO-1.5, CS/C₁₀-ImGO-1.5, CS/E-ImGO-1.5, and CS/B-ImGO-1.5 decreased to 10.8, 10.4, 11.1, 9.6, and 10.8 kJ mol⁻¹, respectively. These results indicated that, by means of hybridization, additional hydroxide ion transport highways were constructed by ImGO, which could meanwhile connect the conducting groups in polymer phase as “bridges” and then reduce the activation energy during OH⁻ migration.

3.5.3 Ionic conductivity of AEMs under anhydrous condition. To better evaluate the OH⁻ conduction ability of different ligands on QA groups, the OH⁻ conductivities of these membranes under anhydrous condition (0% RH, 30 °C) were tested and shown in Fig. 7d. It was found that CS control membrane exhibited an OH⁻ conductivity of 9.11 × 10⁻⁷ S cm⁻¹, which was much lower than that under hydrated condition (1.00 × 10⁻³ S cm⁻¹), indicating that water was essential to the conduction of OH⁻. By comparison, CS/GO-X displayed higher OH⁻ conduction properties than CS control membrane resulting from the conductive groups on GO. Moreover, ImGO-filled membranes exhibited higher OH⁻ conductivities than CS control membrane and CS/GO-X on account of the construction of continuous transfer highways and the increase in the carriers. The possible mechanism of the construction of transfer pathways was illustrated in Scheme 1. Benefiting from imidazolium-functionalization, ImGO nanosheets could be assembled with CS chains driven by the interfacial interactions. Continuous transfer pathways were therefore constructed along ImGO surface with a method analogous to layer-by-layer assembly.^47–49 Specifically, CS/C₄-ImGO-1.5 attained an OH⁻ conductivity of 7.85 × 10⁻⁶ S cm⁻¹, higher than those of CS control membrane (9.11 × 10⁻⁷ S cm⁻¹) and CS/GO-1.5 (5.48 × 10⁻⁶ S cm⁻¹). For ImGO-filled membranes, the OH⁻ conductivity increased in the order of CS/B-ImGO < CS/C₁₀-ImGO < CS/E-ImGO < CS/C₄-ImGO under identical conditions. Without the presence of water, the OH⁻ transfer through membrane mainly depended on surface site hopping manner. The chemical structure of the ligands on QA groups played a critical role on the transfer of OH⁻. For C₄-ImGO and C₁₀-ImGO, the longer ligand on C₁₀-ImGO generated obvious steric effect to C₁₀-ImGO-filled membrane, unfavorable for the OH⁻ transfer. Therefore, the OH⁻ conductivity of CS/C₄-ImGO-X was higher than that of CS/C₁₀-ImGO. As for E-ImGO, due to the formation of zwitterion, the OH⁻ hopping among the groups was expected to be facilitated on account of the electronegative carboxylate group, endowing a higher OH⁻ conductivity. For B-ImGO, the electron cloud density of N atom was decreased due to the delocalization effect of benzene ring, which was infaust to the conduction of OH⁻. In conclusion, CS/C₄-ImGO-X yielded the highest OH⁻ conductivity of 9.02 × 10⁻⁶ S cm⁻¹ for its advantages on OH⁻ conduction. Collectively, the conductivity data clearly implied that the addition of ImGO nanosheets facilitated the OH⁻ conduction of nanohybrid membranes to a large extent.

3.6 Fuel cell performances

On account of the enhanced transport properties, the superiority of E-ImGO-filled nanohybrid membranes was examined via a fuel cell test. The fuel cell performances assembled with CS control membrane, CS/GO-2.0, and CS/E-ImGO-2.0 as representatives were conducted at 60 °C (Fig. 8). Fig. 8a revealed that all the cells achieve the open circuit voltages (OCVs) of about 0.59–0.71 V operating with 4.0 M KOH + 5.0 M methanol solution. For CS control membrane, the OCV, maximum current density and power density were 0.59 V, 279.5 mA cm⁻², and 49.6 mV cm⁻², respectively. By comparison, the fuel cell performances of the nanohybrid membranes were enhanced after GO/E-ImGO incorporation. It was found that incorporating 2.0% E-ImGO afforded a 53.7% increase in maximum current density (429.7 mA cm⁻²), together with a 52.8% increase in maximum power density (75.8 mW cm⁻²). This enhancement was reasonably ascribed to the enhanced OH⁻ conduction ability of CS/E-ImGO-2.0, which reduced the electrolyte resistance and meanwhile accelerated the cathode reaction. Under identical conditions, GO displayed similar function in elevating the cell performances of CS-based membrane, and CS/GO-2.0 acquired the maximum current density of 419.0 mA cm⁻² and the maximum power density of 72.1 mW cm⁻². For the nanohybrid membranes, GO-filled nanohybrid membrane showed lower cell performances compared with E-ImGO-filled nanohybrid membrane due to its low OH⁻ conduction ability.

Fig. 8 (a) Single-cell performances of CS control membrane, CS/GO-2.0, and CS/E-ImGO-2.0 in 4.0 M KOH + 5.0 M methanol solution operated at 60 °C, and (b) single-cell performances of CS/E-ImGO-2.0 operated at 60 °C in 4.0 M KOH + 5.0 M methanol solution, 4.0 M KOH + 5.0 M ethanol solution, and 1.0 M KOH + 5.0 M methanol solution, respectively.

To further research the influence of the type of fuel and the concentration of KOH electrolyte on the fuel cell performance, methanol and ethanol were chosen as the fuel to test to single-cell performance of CS/E-ImGO-2.0 in 1.0 M and 4.0 M KOH electrolyte, respectively. As shown in Fig. 8b, all the cells achieved the OCVs of about 0.64–0.82 V. Fuel cell with 1.0 M KOH and 5.0 M methanol solution showed a maximum power density of 46.8 mW cm⁻² at a current density of 220.6 mA cm⁻². Increasing the concentration of KOH electrolyte (4.0 M) would enhance the maximum power density and current density to 75.8 mW cm⁻² and 429.7 mA cm⁻², respectively. This phenomenon indicated that the concentration of KOH electrolyte influenced the fuel cell performance. Besides, methanol/O₂ single-cell showed higher maximum power density (75.8 mW cm⁻²) and current density (427.9 mA cm⁻²) than ethanol/O₂ single-cell (72.1 mW cm⁻², 380.0 mA cm⁻²) under identical conditions. This might attributed to the faster oxidation reaction of methanol than ethanol. Collectively, CS/ImGO membrane offered significant promise as an OH⁻ exchange membrane for AEMFC application, and the obtained results were comparable to the data of Nafion and some other membranes in the literature.⁵⁰

4. Conclusions

In this study, highly conductive nanohybrid membranes were prepared by incorporating 2D ImGO nanosheets into CS matrix. The microstructures and physicochemical characteristics of the resultant nanohybrid membranes were investigated systematically. Water uptakes and swellings of the nanohybrid membranes were reduced due to the hybridization, which interfered with the chain motion of CS at the interfacial domains. Meanwhile, the interference allowed the incorporated ImGO enhancing the thermal and mechanical stabilities of nanohybrid membranes. Furthermore, the OH⁻ conductivity results indicated that the incorporation of ImGO significantly improved the OH⁻ transport properties of nanohybrid membranes by continuous low-energy-barrier highways, which were ascribed to the numerous QA groups and the 2D structure of ImGO. To be specific, the introduction of polar ligand would enhance the hydrophilicity of QA group, which in turn facilitated OH⁻ transfer. Besides, the shorter chain ligand grafted on QA group would have lower steric hindrance effect and benefit for the OH⁻ transport. Particularly, CS/E-ImGO-2.0 achieved the highest OH⁻ conductivity of 1.02 × 10⁻² S cm⁻¹ at 90 °C (100% RH), about four times of that of CS control membrane (2.51 × 10⁻³ S cm⁻¹). Besides, a fuel cell with CS/E-ImGO-2.0 showed an OCV of 0.71 V, a maximum power density of 75.8 mW cm⁻² at a current density of 298.8 mA cm⁻². The controllable QA groups and distinct 2D structure guaranteed the as-prepared ImGO to be promising OH⁻ conductors for nanohybrid membrane. Our research in the present work thus might cast a new light on the development of nanohybrid membranes with hydroxide ion transport highway.

Acknowledgements

We gratefully acknowledge the financial supports from National Natural Science Foundation of China (21576244 and U1304215), and China Postdoctoral Science Foundation (2014T70687).

Notes and references

1. J. J. Han, H. Q. Peng, J. Pan, L. Wei, G. W. Li, C. Chen, L. Xiao, J. T. Lu and L. Zhuang, ACS Appl. Mater. Interfaces, 2013, 5, 13405–13411.
2. L.-c. Jheng, S. L.-c. Hsu, B.-y. Lin and Y.-l. Hsu, J. Membr. Sci., 2014, 460, 160–170.
3. N. J. Robertson, H. A. Kostalik IV, T. J. Clark, P. F. Mutolo, H. D. Abruña and G. W. Coates, J. Am. Chem. Soc., 2010, 132, 3400–3404.
4. Y.-J. Wang, J. L. Qiao, R. Baker and J. J. Zhang, Chem. Soc. Rev., 2013, 42, 5768–5787.
5. J. L. Qiao, J. Fu, R. Lin, J. X. Ma and J. S. Liu, Polymer, 2010, 51, 4850–4859.
6. J.-J. Kang, W.-Y. Li, Y. Lin, X.-P. Li, X.-R. Xiao and S.-B. Fang, Polym. Adv. Technol., 2004, 15, 61–64.
7. F. Yi, X. P. Yang, Y. J. Li and S. B. Fang, Polym. Adv. Technol., 1999, 10, 473–475.
8. S. F. Lu, J. Pan, A. B. Huang, L. Zhuang and J. T. Lu, Fuel Cells, 2009, 10, 2–6.
9. L. Li and Y. X. Wang, J. Membr. Sci., 2005, 262, 1–4.
10. J. Fang and P. K. Shen, J. Membr. Sci., 2006, 285, 317–322.
11. G. G. Wang, Y. M. Weng, D. Chu, D. Xie and R. R. Chen, J. Membr. Sci., 2009, 326, 4–8.
12. J. R. Varcoe and R. C. T. Slade, Electrochem. Commun., 2006, 8, 839–843.
13. J. L. Qiao, J. Zhang and J. J. Zhang, J. Power Sources, 2013, 237, 1–4.
14. M. Kumar, S. Singh and V. K. Shahi, J. Phys. Chem. B, 2010, 114, 198–206.
15. C.-C. Yang, S.-J. Chiu, W.-C. Chien and S.-S. Chiu, J. Power Sources, 2010, 195, 2212–2219.
16. Q. Li, L. Liu, S. Liang, Q. B. Dong, B. K. Jin and R. Bai, RSC Adv., 2013, 3, 13477–13485.
17. J. L. Wang and L. L. Wang, Solid State Ionics, 2014, 255, 96–103.
18. R. J. Wang, G. H. Jiang, Y. W. Ding, Y. Wang, X. K. Sun, X. H. Wang and W. X. Chen, ACS Appl. Mater. Interfaces, 2011, 3, 4154–4158.
19. Z. Q. Jiang, X. S. Zhao and Y. Z. Fu, J. Mater. Chem., 2012, 22, 24862–24869.
20. L. P. Zhao, Y. F. Li, H. Q. Zhang, W. J. Wu, J. D. Liu and J. T. Wang, J. Power Sources, 2015, 286, 445–457.
21. D. J. Kim, H. Y. Hwang, S. Y. Nam and Y. T. Hong, Macromol. Res., 2012, 20, 21–29.
22. C. Pak and D. C. Lee, ACS Appl. Mater. Interfaces, 2012, 4, 1021–1029.
23. H. Zarrin, J. Fu, G. P. Jiang, S. Yoo, J. Lenos, M. Fowler and Z. W. Chen, ACS Nano, 2015, 9, 2028–2037.
24. F. Lu, X. P. Gao, X. J. Yan, H. J. Gao, L. J. Shi, H. Jia and L. Q. Zheng, ACS Appl. Mater. Interfaces, 2013, 5, 7626–7632.
25. Q. Zhang, S. H. Lib and S. B. Zhang, Chem. Commun., 2010, 46, 7495–7497.
26. K. J. T. Noonan, K. M. Hugar, H. A. Kostalik IV, E. B. Lobkovsky, H. D. Abruna and G. W. Coates, J. Am. Chem. Soc., 2012, 134, 18161–18164.
27. D. Y. Chen and M. A. Hickner, ACS Appl. Mater. Interfaces, 2012, 4, 5775–5781.
28. O. D. Thomas, K. J. W. Y. Soo, T. J. Peckham, M. P. Kulkarni and S. Holdcroft, J. Am. Chem. Soc., 2012, 134, 10753–10756.
29. Y. P. Zha, M. L. Disabb-Miller, Z. D. Johnson, M. A. Hickner and G. N. Tew, J. Am. Chem. Soc., 2012, 134, 4493–4496.
30. J. Fang, M. Lyu, X. Wang, Y. B. Wu and J. B. Zhao, J. Power Sources, 2015, 284, 517–523.
31. D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitskii, Z. Z. Sun, A. Slesarev, L. B. Alemany, W. Lu and J. M. Tour, ACS Nano, 2010, 4, 4806–4814.
32. J. T. Wang, H. Zhang, X. L. Yang, S. Jiang, W. J. Lv, Z. Y. Jiang and S. Z. Qiao, Adv. Funct. Mater., 2011, 21, 971–978.
33. J. Hu, D. L. Wan, W. G. Zhu, L. H. Huang, S. Z. Tan, X. Cai and X. J. Zhang, ACS Appl. Mater. Interfaces, 2014, 6, 4720–4730.
34. J. Ran, L. Wu, J. R. Varcoe, A. L. Ong, S. D. Poynton and T. W. Xu, J. Membr. Sci., 2012, 415–416, 242–249.
35. M. Yadav, K. Y. Rhee, S. J. Park and D. Hui, Composites, Part B, 2014, 66, 89–96.
36. S. Park, J. An, J. R. Potts, A. Velamakanni, S. Murali and R. S. Ruoff, Carbon, 2011, 49, 3019–3023.
37. F. Ordikhani, M. R. Farani, M. Dehghani, E. Tamjid and A. Simchi, Carbon, 2015, 84, 91–102.
38. Y.-S. Ye, M.-Y. Cheng, X.-L. Xie, J. Rick, Y.-J. Huang, F.-C. Chang and B.-J. Hwang, J. Power Sources, 2013, 239, 424–432.
39. X. M. Yang, Y. F. Tu, L. Li, S. M. Shang and X.-M. Tao, ACS Appl. Mater. Interfaces, 2010, 2, 1707–1713.
40. S. Mallakpour and M. Madani, J. Appl. Polym. Sci., 2015, 132, 1–9.
41. J. Zhang, J. L. Qiao, G. P. Jiang, L. L. Liu and Y. Y. Liu, J. Power Sources, 2013, 240, 359–367.
42. N. W. Li, L. Z. Wang and M. Hickner, Chem. Commun., 2014, 50, 4092–4095.
43. Y. Wan, K. A. M. Creber, B. Peppley and V. T. Bui, J. Appl. Polym. Sci., 2003, 89, 306–317.
44. S. Claes, P. Vandezande, S. Mullens, R. Leysen, K. D. Sitter, A. Andersson, F. H. J. Maurer, H. van den Rul, R. Peeters and M. K. V. Bael, J. Membr. Sci., 2010, 351, 160–167.
45. B. C. Lin, L. H. Qiu, J. M. Lu and F. Yan, Chem. Mater., 2010, 22, 6718–6725.
46. J. R. Varcoe, P. Atanassov, D. R. Dekel, A. M. Herring, M. A. Hickner, P. A. Kohl, A. R. Kucernak, W. E. Mustain, K. Nijmeijer, K. Scott, T. W. Xu and L. Zhuang, Energy Environ. Sci., 2014, 7, 3135–3191.
47. K. Ariga, Y. Yamauchi, G. Rydzek, Q. M. Ji, Y. Yonamine, K. C.-W. Wu and J. P. Hill, Chem. Lett., 2014, 43, 36–68.
48. G. Rydzek, Q. M. Ji, M. Li, P. Schaaf, J. P. Hill, F. Boulmedais and K. Ariga, Nano Today, 2015, 10, 138–167.
49. J. J. Richardson, M. Björnmalm and F. Caruso, Science, 2015, 348, aaa2491.
50. A. N. Lai, L. S. Wang, C. X. Lin, Y. Z. Zhuo, Q. G. Zhang, A. M. Zhu and Q. L. Liu, ACS Appl. Mater. Interfaces, 2015, 7, 8284–8292.

---

Footnote

† Electronic supplementary information (ESI) available: XRD patterns of CS control and nanohybrid membranes, TGA curves of CS/C₁₀-ImGO-X, stress–strain curves of CS/C₄-ImGO-X, temperature-dependent conductivity of (a) CS/GO-X, (b) CS/C₄-ImGO-X, (c) CS/C₁₀-ImGO-X, and (d) CS/B-ImGO-X under 100% RH. See DOI: 10.1039/c5ra18183f

---