Biomass-Based Anion Exchange Membranes Using Poly (Ionic Liquid) Filled Bacterial Cellulose with Superior Ionic Conductivity and Significantly Improved Strength

ABSTRACT How to simultaneously improve the ionic conductivity and mechanical properties is a key problem facing currently used anion-exchange membranes (AEMs). Here, novel AEMs were prepared using quaternized bacterial cellulose (QBC) as a dual-functional substrate and then filled with a polymeric ionic liquid (poly(vinylbenzyl) trimethylammonium chloride, PVD) with high ion-exchange capacity through in situ polymerization and crosslinking. The dense quaternary ammonium groups grafted on the surface of BC nanofibers greatly increased the ionic conductivity, while the special three-dimensional network structure of BC significantly enhanced the tensile strength and chemical stability of the obtained PVD filled quaternized BC (QBC/PVD) membranes. The ionic conductivity of QBC/PVD membrane reached as high as 111 mS cm−1 at 80°C, which was 109% higher than that of the pure BC/PVD membrane (only 53 mS cm−1). Moreover, the QBC/PVD membrane exhibited extremely high dry strength of 72.3 MPa and satisfactory wet strength and flexibility, this membrane can hang a container containing 500 g of water when at fully hydrated state. The alkaline direct methanol fuel cell equipped with QBC/PVD output a peak power density of 64 mW cm−2, showing its great application potential as an AEM.


Introduction
The current global pollution and the increasing use of electricity have attracted a lot of attention to clean or renewable energy sources (An and Zhao 2017;Dekel 2018;Pan et al. 2017). Alkaline anion exchange membrane fuel cells (AAEMFCs) have gradually become a research hotspot in recent years due to their advantages of high efficiency, environmental friendliness and wide fuel selection, while avoiding the disadvantages such as high cost and high methanol permeability of proton exchange membrane fuel cells (Chang et al. 2018;Vijayakumar et al. 2019;You, Noonan, and Coates 2020). Anion exchange membrane (AEM) is the core component of an AAEMFC, which plays the role of separating cathode and anode and transferring hydroxide ions during the operation of fuel cells (Dong et al. 2015). Therefore, AEMs should have high ionic conductivity, good mechanical properties, sufficient chemical stability under alkaline conditions, and low production cost. Many polymers, especially aromatic polymers such as polystyrenes, polyether ether ketones, and polyether sulfones, have been widely used to prepare AEMs by chloromethylation of these polymers followed by cation functionalization (e.g. quaternizaiton or imidazolization) (You, Noonan, and Coates 2020). However, the complex chloromethylation process and strongly carcinogenic chloromethylation reagents (e.g., chloromethyl ethers) limit the widespread application of these polymers as AEMs. Moreover, there are still challenges to achieve a balance between ionic conductivity and mechanical properties and alkaline stability in preparing AEMs (Deng et al. 2015;Noh et al. 2019;Omasta et al. 2018;Wong et al. 2020).
Polymeric ionic liquids (PILs), a class of macromolecules deriving by the polymerization of ionic liquid (IL) monomers, have shown great potentials as solid electrolyte candidates because they combine the advantages of ILs (high ionic conductivity, chemical and electro-chemical stability, and low flammability) and polymers (good mechanical properties) (Li et al. 2019;Qian, Texter, and Yan 2017;Vilela et al. 2017). Although PILs possess a shape of soft solids and are more mechanically robust than ILs, PILs are difficult to form self-supporting membranes due to their relatively low glass transition temperature (generally below room temperature) (Evans et al. 2016). The commonly used methods to improve the mechanical stability of PILs are to increase their molecular weight (Nakamura et al. 2011) or copolymerize with other monomers to form block copolymers or graft copolymers (Mecerreyes 2011;Vilela et al. 2017). However, such improvement of mechanical reliability through increasing the molecular weight of a PIL usually means the increased difficulty in its processability due to the increased polymer viscosity. In the latter method, the molecular weight and each component can be changed and designed to obtain PIL copolymers with balanced mechanical properties and conductivity. For instance, Elabd and coworkers synthesized a PIL diblock copolymer comprising an IL monomer (1-[(2-methacryloyloxy)undecyl]-3-butylimidazolium bicarbonate) and a nonionic monomer (methyl methacrylate) through the reverse addition−fragmentation chain-transfer (RAFT) polymerization. The bicarbonate conductivity of the copolymer was 14.8 mS cm −1 at 60°C and 95% RH, but no mechanical property data were reported (Nykaza et al. 2016). Ouadah et al. (2017) synthesized copolymers of butylvinylimidazolium and para-methyl styrene via free radical polymerization and then coupled with poly(4,4′-diphenylether-5,5′-bibenzimidazole) (DPEBI) at a high temperature reaching 210°C. The resulting new PILs materials exhibited a high conductivity of 73.1 mS cm −1 at 100°C. Nevertheless, the synthesis procedures of these PIL copolymers are usually complicated and time-consuming. Moreover, the confinement impact of other components in PIL copolymers as well as the complex interaction between solvents and copolymers during the process of membrane formation often affect the final morphology and thus ionic conductivity of the membranes (Hoarfrost et al. 2012).
The development of composite membranes consisting of PILs reinforced by porous substrates is a simple and effective strategy to obtain nanostructured AEMs possessing high ionic conductivity together with good mechanical durability. Commonly used porous substrates are polytetrafluoroethylene (PTFE) (Lin et al. 2005;Yamaguchi, Miyata, and Nakao 2003;Zhang et al. 2010;Zhao et al. 2012), polyvinylidene fluoride (PVDF) (Liu et al. 2020;Nasef et al. 2006), which generally have good thermal properties and excellent mechanical strength. For this reason, Xie et al. (2019) developed the pore-filled membranes based on PILs with quaternary ammonium and tertiary amine head groups that filled with PE for AEMFCs. The resulting membranes exhibited a high ionic conductivity of 32.8 mS cm −1 at 60°C and good chemical stability up to 350 h in a 1 M KOH solution. And the tensile strength at break was 80 MPa. 2015 recently used an electrospun PVDF with hexagonal platelets of Mg-Al layered double hydroxide (LDH) as a porous substrate and then filled with vinylbenzyltrimethylammonium chloride (VBTAC) to prepare composite AEMs. The VBTAC-filled PVDF-LDH membranes had a remarkable conductivity of 87 mS cm −1 at 70°C and restricted swelling. However, its tensile strength was only 15 MPa, and PVDF is very easy to remove hydrogen fluoride (HF) to form C=C structure in strong alkaline environment, which resulted in the decreased alkaline stability. In addition, these traditional synthetic porous substrates are usually highly hydrophobic, which brings difficulties for subsequent wetting and filling of hydrophilic ILs or PILs.
Bacterial cellulose (BC), a type of nano-cellulose synthesized by bacteria, is a versatile and attractive material with excellent mechanical properties, low density, high surface area, and environmentally friendly nature (Bae and Shoda 2004). Importantly, the large three-dimensional space between the fibers along with the huge number of hydroxyl groups on the fiber surface is very favorable for the infiltration and filling with subsequent polar quaternized polymers or cationic monomers. In our previous work (Ni et al. 2022), BC served as the template to anchor lamellar inorganic hydroxide ion conductor-LDH, and then quaternized chitosan was impregnated into the as-prepared LDH coated on BC (LDH@BC) bifunctional porous substrate to prepare composite membranes. On the one hand, the three-dimensional space of BC nanofibers can effectively limit the stacking of LDH layers, thus fully exposed the anion transport sites of LDH. On the other hand, LDH@BC can also effectively enhance the mechanical strength of quaternized chitosan that could not even withstand the mechanical pressure during the fuel cell assembly.
In this work, a continuous effort was made to combine a functionalized biomass-based BC porous substrate and a PIL (poly(vinyl benzyl) trimethylammonium chloride, PVD) to obtain AEMs with simultaneously enhanced ionic conductivity and mechanical properties via a simple and environment-friendly route. First, in order to obtain a dual-functional nanofiber porous substrate that could synchronously serve as a fast ion transport medium and an ultra-strong supporting substrate, a quaternized porous substrate (QBC) was synthesized by grafting a silane coupling agent (TPC) containing quaternary ammonium groups onto the BC fiber surface through a hydrolytic condensation reaction. Then, (vinyl benzyl) trimethylammonium chloride (VBTAC) IL monomer was filled in the pores of the QBC porous substrates, followed by in situ crosslinking and polymerization via the "impregnation polymerization" technique to prepare high-performance PVD impregnated QBC (QBC/PVD) AEMs. The modified QBC porous substrates together with the filled PIL lead to an overall improved and optimized membrane material with ultra-high ionic conductivity, remarkable mechanical properties, and stable chemical properties.

Purification of bacterial cellulose (BC)
The commercial BC was cut into squares of 10 × 10 cm, followed by treatment with 0.01 M sodium hydroxide solution for 3 h at 90°C to remove impurities on the surface of BC. After that, the BC pellicles were rinsed with deionized water until pH value was neutral; thereafter, they were stored in deionized water at room temperature.

Synthesis of the quaternized bacterial cellulose (QBC)
BC was quaternized by a hydrolytic condensation reaction between the hydroxyl group of BC and the silanophilic group of TPC. The detailed experimental method is as follows: a purified BC membrane was immersed in a solution containing deionized water (100 mL), different additive amounts of TPC (5, 7.5, or 10 mL), and ammonia (3 mL) with continuous magnetic stirring at 70°C for 24 h. Then, the TPC treated BC substrate was washed with deionized water and ethanol to remove the unreacted TPC. The obtained quaternized BC porous substrates treated with different TPC concentrations were noted as QBC-1, QBC-2, and QBC-3, respectively.

Preparation of QBC/PVD membranes
QBC/PVD anion exchange membranes were prepared by filling VBTAC monomer into QBC porous substrates through an in situ impregnation polymerization method (as shown in Figure 1). The typical preparation procedure is as follows (using the optimal QBC/PVD AEM as an example). Approximately 1.5 g of VBTAC monomer was added in a flask with 19 mL of ethanol, 0.075 g of V 50 (used as an initiator), and 0.246 mL of DVB (used as a cross-linker, 15 wt.% based on the weight of VBTAC). After mixing, the monomer mixture solution was poured onto a Teflon sheet, and then the QBC porous substrate was immersed in the monomer solution at 4°C for 2 h. The monomerimpregnated membrane was then kept at 60°C for 16 h to polymerize and crosslink with VBTAC and DVB. The as-prepared membrane was washed repeatedly using deionized water to remove unreacted monomers and then dried on a glass plate at 50°C. Finally, the membrane (Cl− form) was converted to its OH− form by treating with KOH aq (1 mol L −1 ) for 24 h. The thickness of the prepared membranes was about 50 ± 5 μm.

Structure and morphology characterization of QBC and QBC/PVD membranes
Scanning electron microscopy (SEM) analysis of the sample surfaces and cross-sections were obtained on JSM-7500F scanning electron microscope (Hitachi Co., Japan), and the cross-sections were obtained by quenching the section with liquid nitrogen. Moreover, the equipped X-Max 80 Electro-Cooled X-Ray Spectrometer (EDS) was utilized for elemental analysis. The Attenuated Total Reflection-Fourier transform infrared spectroscopy (ATR-FTIR) of the samples was tested on a Nicolet iS50 spectrometer (Thermo Fisher, USA) within a spectral range of 4000-400 cm −1 . The phase structure of the QBC and QBC/PVD AEMs were tested on an X-ray diffractometer (XRD) (D8 Advance, Bruker, Germany) within the 2θ range of 5-50° at a scanning speed of 5° min −1 . The samples were cut into 1 × 4 cm (width × length) to perform tensile tests on a tensile testing machine (Model XLW-EC-A, Labthink, China) using a tensile rate of 10 mm min −1 . Before testing, the samples were dried at 60°C to a constant weight in an oven. Thermogravimetric analysis (TGA) with STA 6000 (PerkinElmer, USA) was used to study the thermal stabilities of the samples. The samples were first preheated at 70°C for 5 min to remove the absorbed water and then heated to 700°C at a heating rate of 10°C min −1 under the protection of nitrogen atmosphere. All the above test environment was 25°C and about 40% relative humidity.

Water uptake and swelling ratio
Water absorption and swelling changes of the as-prepared membranes under the dry and fully hydrated states were measured, respectively. After immersing the membranes in deionized water for 24 h at different temperatures (20 and 80°C), their mass (M w ), area (A w ), and thickness (T w ) were measured after quickly wiping off the surface water. After that, the mass (M d ), area (A d ), and thickness (T d ) were measured after drying at 60°C overnight. The water uptake and swelling ratio can be calculated using the following equations:

Ionic conductivity and alkaline stability
The in-plane conductivities of the fully hydrated membranes were tested using the two-electrode electrochemical impedance spectroscopy (EIS) method on an electrochemical station (Autolab PGSTAT 302 N, Netherlands) from 1 to 1 MHz at different temperatures. The samples were cut into 2 × 3 cm and immersed in 1 M KOH aq for 24 h. Then, the samples were washed to neutral for test. The ion conductivity (σ, mS cm −1 ) is calculated using Equation (4): where l (cm), A (cm 2 ) and R (Ω) are the distance between two electrodes, the effective cross-sectional area of the membrane, and the membrane resistance, respectively. The alkaline stability was determined by monitoring the residual conductivity of the membrane in a solution of 1 M KOH aq and 5 M KOH aq at 60°C for a certain time.

Ion-Exchange Capacity (IEC)
Ion-exchange capacity was measured by a conventional acid-base titration method. The membrane in Cl-form was immersed in a KOH solution (1 mol L −1 ) for 24 h to be ion exchanged into its OH-form. The excess KOH solution on the surface of the membrane was washed thoroughly and then dried at 60°C. The obtained alkaline membrane was immersed in a sealed HCl solution (0.01 mol L −1 ) for 48 h. The membrane was titrated with a 0.01 mol L −1 NaOH solution, and phenolphthalein was used as indicator. The IEC (mmol g −1 ) is calculated using Equation (5): where V HCl (L) is the volume of HCl solution, V NaOH (L) is the volume of NaOH solution consumed during the titration process, C HCl (mol L −1 ) is the concentration of HCl solution, C NaOH (mol L −1 ) is the concentration of NaOH solution, M (g) is the mass of the dry membrane.

Single cell performance measurements
The electrode is a traditional GDE (gas diffusion electrode) customized by Sunlaite Co. Ltd (China). The catalyst layers were purchased by Shanghai Hesen Electric Appliance Co., Ltd. The catalyst loadings were 4 mg cm −2 for the anode (Pt-Ru/C) and 2 mg cm −2 for the cathode (Pt/C), respectively. The membrane electrode assembly (MEA) was prepared by sandwiching a membrane with dimensions of 3 × 3 cm (length × width) between the anode and cathode. MEA was assembled with anode GDE/AEM/cathode GDE in a single cell. The alkaline direct methanol fuel cell (DMFC) performance was evaluated by single-cell test at 80°C on an electrochemical workstation (Autolab PGSTAT302N, Switzerland). The fuel (2, 3, 4, or 5 M KOH in a 2 M methanol solution) was injected into the anode at a flow rate of 1 mL min −1 , and oxygen was injected into the cathode at a flow rate of 100 mL min −1 .

Characterization of QBC porous substrate
In order to solve the trade-off among the hydroxideconductivity, mechanical properties, and alkaline stability, a nano-structured BC, which possesses interconnected cellulose nanofibers, ultra-high mechanical strength, and surface hydroxyl groups, was utilized as the original porous substrate to perform the further surface cationic functionalization. Benefitting from the high concentration of anionic-conductive sites on the surface of BC nanofibers, the functionalized BC porous substrate, QBC, can not only serve as a strong mechanical support for the subsequent filling polyionic liquid, but also increase the hydroxide conductivity of the final membranes. ATR-FTIR was used to determine the successful grafting of quaternary amine groups in the modified BC and the obtained ATR-FTIR spectra are shown in Figure 2a. In the spectra of BC and QBC, the absorption peak at 3349 cm −1 is the stretching vibration O-H, and the peaks at about 2979 and 2898 cm −1 are assigned to the asymmetric stretching (υ as ) and symmetric stretching (υ s ) vibrations of C-H, respectively. In the low-wavenumber region, the weak peak located at 1649 cm −1 is related to the bending vibration of -OH of the absorbed water. Besides, a strong absorption band at 1060 cm −1 is attributed to the vibration of C-O-C in the pyranose ring skeleton (Castro et al. 2011). When compared with the spectrum of BC, a new peak appeared at 1478 cm −1 in QBC comes from the bending vibration of the -CH 3 and -CH 2 in the grafted silane coupling agent. Moreover, the bending vibration of the C-N bond appears at 952 cm −1 (Liu et al. 2020;Xiong et al. 2008), verifying that TPC has been successfully grafted on the surface of BC nanofibers.
In order to quantify the grafting degree of quaternization of the modified bacterial cellulose, the TGA test was performed, and the obtained curves are shown in Figure 2b. The thermal weight loss values of BC, TPC, QBC-1, QBC-2, and QBC-3 at 700°C were 87.3, 64.1, 75.3, 69.3, and 64.0 wt.%, respectively. Therefore, the grafting content of the modified bacterial cellulose with TPC can be calculated by the following equation: where W 0 , W m , and W p are the original substrate (BC), the surface modified material (TPC) and the prepared modified material (QBC), respectively. w is the grafting amount of silane coupling agent. According to Equation 6, in this study, the calculated TPC contents on BC nanofibers (QBC-1, QBC-2, and QBC-3) modified with different amounts of silane coupling agents were 51.7 wt.%, 77.6 wt.%, and 100 wt.% on the substrates, respectively.

Morphology of QBC porous substrate
SEM scans were performed to better observe the internal microscopic morphology of BC and QBC. As shown in the SEM images (Figure 3), it can be clearly observed that the pure BC substrate showed a loose and porous structure, while the fibrous structure of the quaternized modified  bacterial cellulose membrane showed a great change. It can be seen that an envelope layer was formed on the fiber surface due to the grafting of silane coupling agent on the fibers of the modified bacterial cellulose. The SEM image of Figure 3b shows that QBC-1 porous substrate presented a dense structure with small pores. The SEM image of Figure 3c shows that the QBC-2 porous substrate exhibited a large number of small pore structures, which are more favorable for the formation of ion channels to promote the transport of OH-ions (Zou et al. 2021). However, as shown in Figure 3d, the number of small pores in the QBC-3 porous substrate decreased as the degree of quaternization increased, which led to a decrease in ion channels that could cause the conductivity to decrease. As shown in SEM-EDX mapping images (Figures 3e and 3f), Si and N were found in the QBC-2 porous substrate, which further indicated the successful grafting of TPC on the BC nanofibers.

Characterization of QBC/PVD composite membranes
To verify the successful in situ polymerization of VBTAC inside the QBC porous substrates, we performed FTIR-ATR test. As for QBC/PVD (Figure 4a), the peaks appearing at 1682, 1412, and 896 cm −1 are due to the C=C double bond in VBTAC, the C-N bond in the benzene ring, and the disubstituted benzene, respectively. In addition, the intensity of peak at 1478 cm −1 is enhanced when compared to that of BC/PVD membrane due to the superposition of trimethyl groups on VBTAC and TPC. Figure 4b compares the XRD patterns of BC, BC/PVD, and QBC/PVD. In general, the broad, blunt, and low-intensity XRD peaks reflect the low crystallinity of polymer electrolyte membranes, which facilitates the ion conduction because the transport process in membranes is dominated by amorphous phase rather than crystalline phase (Ma and Sahai 2013). As shown in Figure 4b, pure BC had three diffraction peaks at 2θ = 14.1°, 16.4°, and 22.5° attributed to the (1-10), (110) and (020) crystalline crystal planes of BC, respectively, indicating that BC has a typical high crystallinity cellulose I isomeric structure (Wu et al. 2018). When the BC nanofiber substrate was surface modified and filled with PVD, the peak at 2θ = 17.5° was significantly weakened, which means that the proportion of amorphous state is increased (Dai et al. 2019). This is ascribed to the hydrolytic condensation reaction of -OH on the BC membranes and Si-OH on the silane coupling agent, thereby reducing the hydrogen bonding within the membranes.

Morphology of QBC/PVD composite membranes
To study the microstructure of the prepared QBC/PVD AEMs, we performed SEM observations on the cross-sections of all membrane samples. As shown in Figure 5, after the PVD polyelectrolyte was introduced inside the QBC porous substrates, a few nanofiber fractures of the BC can be seen in the cross-sectional view. The presence of VBTAC in the pores of porous substrates was confirmed that the in-situ "impregnation polymerization" method allowed to obtain structurally dense QBC/PVD AEMs. There was no fiber pulled out at the membrane fracture, no obvious pore and phase separation interface appeared in Figure 5, we believed that the pores of QBC nanofibers were completely filled by PVD via adjusting the concentration and dosage of cross-linked PVD. Such dense structure facilitates the improvement of mechanical properties and the reduction of methanol penetration. From the SEM-EDX mapping images of Figures 5e-5h, it can be clearly seen that Si and N elements are scattered in a large number of cross-sections in addition to C and O elements, which further confirms the homogeneity of PVD filling.

Mechanical properties of QBC/PVD composite membranes
The AEM prepared by pure quaternary ammonium-type vinyl benzyl chloride polymer has poor chemical stability and mechanical properties. The use of BC with high strength and modulus as a substrate can effectively improve the mechanical stability of PVD. As shown in Figure 6e, the tensile strength of QBC/PVD AEM first increased and then decreased with the increase in the grafting content of silane coupling agent. Among all the membrane samples, the QBC/PVD-2 composite membrane showed the most remarkable improvement in the strength (72.3 MPa), which was 142% higher than that of the pure BC/PVD AEM (only 29.9 MPa). The specific data of BC/PVD and QBC/ PVD AEMs can be seen in Table 1. Moreover, under the fully hydrate state, the QBC/PVD AEM showed excellent flexibility and could be easily folded into complicated shapes such as the kelp junction (Figure 6b). In addition, the highly flexible QBC/PVD AEM with a width of 8 mm can withstand a container containing 500 g of water (Figure 6c), while the pure PVD membrane alone could not support itself. The ultra-high strength together with good flexibility of the QBC/PVD AEM can fully meet the mechanical property requirements of fuel cells.
From the TGA curves in Figure 7, it was found that the QBC/PVD AEMs started to decompose from 220°C, verifying the enough thermal stability of these membranes for the application in AAEMFCs. The overall weight loss trend of different silane coupling agent grafting amounts with the increasing temperature is consistent. The TGA curve is roughly divided into three parts: (1) there is almost no weight loss of QBC/PVD AEMs between 70 and 220°C demonstrated the thermal stability is improved at this stage, which is mainly attributed by the three-dimensional network structure of the BC membrane and the filling of PVD in the voids; (2) the weight reduction of QBC/PVD AEMs from 220 to 450°C is due to the decomposition of quaternary ammonium groups in TPC and the degradation of pure BC film and PVD side chains; (3) the weight loss of QBC/PVD AEMs in 450-700°C is mainly due to the degradation of pure BC membrane and PVD main chain. The TGA curves demonstrate that the QBC/PVD AEMs possess good thermal stability, and they are well suited for use in fuel cells. Figure 8 presents the water uptake, swelling ratio, and thickness swelling of the QBC/PVD membrane samples. In AEM fuel cells, water molecules not only provide ion transport carriers but also help form hydrogen bonding networks, which are critical for OH-to transfer through the formation and fracture of them, so water molecules have an important impact on ion conductivity (Peckham and Holdcroft 2010). Most conventional alkaline AEMs are homogeneous films with isotropic swelling behavior, which will undergo excessive swelling in both the through-plane direction and in-plane direction, resulting in poor mechanical stability. In comparison, the prepared QBC/PVD AEMs exhibit

Sample
Tensile strength (MPa) Elongation at break (%) BC/PVD 29.9 ± 8.1 5.5 ± 0.2 QBC/PVD-1 59.3 ± 2.3 4.3 ± 0.7 QBC/PVD-2 72.3 ± 5.9 6.6 ± 0.5 QBC/PVD-3 60.0 ± 3.1 5.7 ± 0.3 anisotropic swelling in both directions: the water absorption of the fully hydrated QBC/PVD-2 AEM is as high as 314.7% at 80°C, with a nearly threefold increase in thickness of about 298.1%, but the inplane swelling is negligible at about 2.5%. Here, the use of this highly absorbent and high-strength uniquely structured BC porous substrate allows the composite membrane to swell effectively along the through-plane direction and in-plane direction, so that the prepared QBC/PVD AEMs have high mechanical stability even in high water absorption state. Interestingly, the bulk thickness swelling  occurs only in the free-state QBC/PVD AEMs and is significantly limited when assembled to MEA. Benefit from the BC substrate of the QBC/PVD AEMs, the dimensional stability does not significantly reduce even under high water absorption, which can provide good mechanical properties for membranes under the operating environment.

Ionic conductivity and IEC of QBC/PVD composite membranes
Ionic conductivity is a very important parameter for anion exchange membranes, and IEC is essential for the construction of continuous ion transport channels (Lee et al. 2017). In general, excessive water uptake not only brings about excessive swelling but also dilutes the ionic concentration and thus leads to a decrease in ionic conductivity . In comparison, the as-prepared QBC/PVD AEMs with higher water uptake demonstrated higher IEC and higher ionic conductivity. Figure 9a presents the ionic conductivity values of QBC/PVD AEMs at different temperatures (from 20 to 80°C). Clearly, the ionic conductivities increased significantly for all the samples with increasing temperature. The increase in temperature accelerates the internal OH-transport rate within the QBC/PVD AEMs, thus increasing the conductivity. In particular, the ionic conductivity of the QBC/PVD-2 membrane exceeded those of all the other membranes we studied over the entire temperature range. For instance, the ionic conductivity value of the as-prepared QBC/PVD-2 membrane was as high as 111 mS cm −1 at 80°C, which was 109% higher than that of the pure BC/PVD membrane (only 53 mS cm −1 ) under the same condition. Besides, as shown in Figure 9b, the QBC/PVD-2 membrane showed the lowest activation energy (7.6 kJ mol −1 ) among all the samples, confirming the fast ion transportation in the membrane. For IEC (as shown in Figure 9c), these values first increased and then decreased with increasing graft content of silane coupling agent, which was consistent with the trend of conductivity.
The IEC values of these three QBC/PVD AEMs all exceeded 2.5 mmol g −1 and far beyond that of BC/ PVD (only 1.93 mmol g −1 ), and the highest IEC was as high as 2.77 mmol g −1 (QBC/PVD-2 AEM). The possible reasons for the high conductivity and IEC values of QBC/PVD AEMs are as follows: (1) the grafting of quaternary ammonium-terminated silane coupling agent on the surface on BC nanofibers can provide additional OH-transport sites, thus increasing the IEC values; (2) the interaction between QBC substrate and the filled PVD facilitates to form long-range ionic conduction channels in the membranes by means of continuous BC fibers; (3) the reduced crystallinity of the composite matrix promotes the ionic dissociation of basic functional groups (García-Cruz et al. 2016). As can be seen from the results of conductivity, Arrhenius plot, and IEC, the modified QBC porous substrates can not only act as an excellent mechanical support for PVD but also significantly improved its ionic conductivity.

Alkaline stability of QBC/PVD composite membranes
Quaternary ammonium (QA) groups have been reported to be susceptible to degradation by βhydrogen elimination and direct OH nucleophilic attack. Therefore, sufficient alkaline stability is essential for the long-term operation of AEMs in AAEMFCs. To evaluate the alkaline stability of the prepared composite membranes, the ionic conductivity values of all the membranes were monitored after immersion in KOH solution (1 M) at 60°C for a certain time. As shown in Figure 9d, the ionic conductivity of the as-prepared QBC/PVD-2 membrane reached to 107 mS cm −1 at 80°C, and its residual conductivity retained more than 95% after immersing in a 1 M KOH solution at 60°C for 100 h. In particular, the QBC/PVD-2 membrane was still stable after 650 h immersion in a KOH solution with a rather satisfied residual conductivity ratio of 79%. It can be seen from the visual graph that the membrane still maintained its integrity after 650 h alkaline stability test. In addition, the QBC/PVD-2 membrane exhibited the most excellent alkaline stability even after immersion in a concentrated KOH solution (5 M) at 60°C for 100 h. Its conductivity at 80°C was 98 mS cm −1 and the residual conductivity ratio was still as high as 88%. Table 2 shows the comparison of mechanical properties, ionic conductivity, and alkaline stability of this study and other AEMs. The results showed that QBC/ PVD AEMs were highly stable in strong alkaline solutions.

Effect of KOH concentration on cell performance
Generally, AEMs with high water absorption will always lead to poor mechanical strength and cannot even be used in the devices. In addition, the large planar swelling of AEMs will greatly reduce the dimensional stability and even destroy the structure of MEA. Fortunately, in our designed QBC/PVD system, although the QBC/PVD AEMs showed a huge water swelling behavior in the through-plane direction (>300% of original thickness), they possessed excellent dimensional stability in the in-plane direction (not exceeded 6% of original area). Such huge difference in swelling behavior in the throughplane and in-plane directions can help AEMs to adaptively fill the micro gaps between the electrolyte and catalyst layer, thus greatly alleviating the slow interfacial mass transfer (Xu et al. 2021). Moreover, the excellent size stability of the QBC/PVD membranes in the in-plane direction will greatly limit their through-plane swelling when assembled into MEAs, which can effectively maintain their original properties. Considering the overall performance, QBC/PVD-2 AEM with the optimal conductivity, excellent mechanical properties, and good alkaline stability was selected as the electrolyte to assembly DMFC to determine its applicability in real electrochemical devices. The cell performance was investigated at different temperatures as well as at different alkali concentrations. Figure 10a shows the DMFC power density-current density-open circuit voltage curves of the QBC/PVD-2 AEM fueled with 2 M methanol and different concentrations of KOH solution at 80°C. It can be clearly seen from the figure that the power density of the fuel cell increased from 42 mW cm −2 to 64 mW cm −2 when the KOH concentration increased from 2 M to 5 M. In general, increasing the KOH concentration means that more OH-can participate in the electrode reaction and thus accelerates electrode reaction rates. However, further increasing the KOH concentration to 6 M did not obtain continuously increased fuel cell performance, but sharply decreased OCV and even could not record the cell performance curves. This decline might be ascribed to the fact that very high KOH concentration results in a high viscosity of the anode fuel, thus bringing the adverse effect on the free water diffusion and ion hydration. Figure 10b represents the power density-current density-open circuit voltage curves of the DMFC equipped with the QBC/PVD-2 membrane as the electrolyte membrane and using 2 M methanol and 5 M KOH solution as the anode fuel at different working temperatures. The open-circuit voltage (OCV) values at 40, 60, and 80°C were 0.77, 0.80, and 0.81 V, respectively. As can be seen that the temperature increased from 40°C to 60°C, the DMFC open-circuit voltage got a certain boost of 0.04 V, which can be ascribed to the fact that OCV is closely related to the electrode reaction and methanol permeation. At 40°C, the lower temperature leads to a slow catalytic reaction and it is difficult for methanol to diffuse into the catalyst layer to participate in the electrode reaction, thus exhibiting a lower open-circuit voltage at low temperatures (Varcoe and Slade 2005). When the temperature reached to 60°C, close to the boiling point of methanol (64.7°C), partially vaporized methanol can reach the catalytic layer through the diffusion layer extremely fast to participate in the electrolysis reaction. With the increase in temperature, the electrolysis reaction rate was greatly enhanced, and the power density was also remarkably increased. The power density at 80°C reached 64 mW cm −2 , which was 161% higher than that of the cell at 40°C (25 mW cm −2 ). In summary, the high ionic conductivity of QBC/PVD-2 membrane is responsible for the enhanced cell performance.

Conclusions
In conclusion, a dual-functional porous substrate QBC was prepared through a hydrolytic condensation reaction between the -OH of BC and the silanophilic group of TPC. Then, ionic liquid monomer VBTAC with high ion-exchange capacity was filled into QBC substrate to prepare QBC/PVD membranes through in situ impregnation polymerization and crosslinking process. The highly surface functionalized QBC can not only serve as a powerful reinforcing porous substrate for PVD but also provide additional OHtransport sites as well as form long-range ionic conduction channels in the membranes, thus significantly increasing the ionic conductivity to 111 mS cm −1 (80°C). The resulting QBC/PVD membrane showed extremely high mechanical strength both at dry state and at fully hydrate state. Consequently, the DMFC assembled with the QBC/PVD electrolyte membrane output an open-circuit voltage of 0.81 V and a peak power density of 64 mW cm −2 when using 2 M methanol and 5 M KOH as the anode fuel at 80°C. The results demonstrated that the QBC/PVD membrane with excellent comprehensive performance is suitable for AEM.