Ordered Ionic‐Liquid Channels Enable Fast Anhydrous Proton Conduction at up to 240°C for Fuel Cells

ABSTRACT

Intermediate‐temperature electrochemical technologies offer a range of important advantages but require high‐performance and robust anhydrous proton exchange membranes. Herein, we demonstrate that confinement of phosphonium‐, imidazolium‐, and sulphonium‐based ionic liquids (ILs), within channels formed by stacked monolayers of boron nitride and graphene nanosheets functionalised with polyethyleneimine, provides robust and effective proton conductivity while suppressing bulk diffusion of the IL ions. We propose that the positively charged nanosheets might induce spatial ionic ordering, anchoring anions near the channel walls while concentrating protons and cations along the central region to establish parallel, continuous pathways for rapid proton transfer under anhydrous conditions. The resulting solid‐state composite membranes exhibit stable anhydrous proton conductivities of up to 10⁻² S cm^{− 1} at 240 ± 1°C, outperforming many existing IL‐based and polymer‐based systems. When integrated into hydrogen‐oxygen fuel cells, the [EMIM][NTf₂]‐based membrane enables a peak power density of 82 ± 5 mW cm⁻² under anhydrous conditions at 240 ± 1°C and atmospheric pressure. In a broader context, this work presents a strategy for the design of membranes with properties tailored to a specific application, which can be achieved by tuning chemical structures of the confined IL and/or functional layers within the 2D channels.

Infiltration of ionic liquids into a composite scaffold comprised of graphene and boron nitride nanosheets functionalised with positively charged polyethyleneimine produces robust proton conducting membranes for the intermediate temperature electrochemical devices. This work showcases application of such membranes for proton exchange membrane fuel cells within the temperature range of 160°C–240°C.

1. Introduction

Proton exchange membranes (PEMs) are a cornerstone component in a number of electrochemical technologies such as water electrolysis [1], CO₂ reduction [2], and fuel cells (FCs). PEMFCs represent a key technology in the pursuit of clean energy transition, delivering high energy efficiency of the fuel‐to‐electricity conversion and proven durability on a practical timescale [3]. Among PEMFC systems, the most mature and most extensively researched are hydrogen‐fed devices based on Nafion and similar perfluorinated polymer membranes [4, 5, 6], which provide high conductivity at low temperatures, low levels of gas crossover, and excellent mechanical and chemical stability [7]. However, Nafion‐based PEMFCs are limited to operation at temperatures typically below 120°C, necessitating the use of highly pure hydrogen fuel devoid of CO and other impurities, which easily poison the platinum catalyst at these low temperatures [8, 9]. Higher operating temperatures cause significant degradation of Nafion and similar electrolytes [10, 11, 12].

Operation of PEMFCs at elevated temperatures offers a range of technological advantages, including improved reaction kinetics and hence reduced loadings of expensive Pt catalysts [13], improved water management [14], synchronous generation and utilisation of both electricity and high‐value heat [13], and flexible use of fuels with relatively high levels of admixtures such as CO, the catalyst poisoning effect of which is notably suppressed at > 100°C [8, 15, 16]. For example, molten carbonate fuel cells offer high efficiency, reduced system cost, and the ability to use CO directly as a fuel [17, 18]. On the negative side, these devices are prone to corrosion and electrolyte degradation [17, 18] and also require a continuous CO₂ supply at the cathode [17]. Solid oxide fuel cells (SOFCs) do not use precious metals as catalysts [19] while providing excellent combined heat and power efficiency [19, 20] and long‐term stability on the scale of > 10 years [20]. On the downside, SOFCs require very high operating temperatures [19, 21] and have slow start‐up times [19, 20].

An alternative and more flexible solution are intermediate‐temperature PEMFCs, which operate between 100°C and 300°C. With the inability of the Nafion‐type and solid‐oxide electrolytes to perform robustly and effectively at such temperatures, a significant body of research has focused on exploring alternative anhydrous intermediate‐temperature membranes. Some of the prominent examples include materials based on polyimides (PI) [22, 23], sulfonated polyether ether ketone (SPEEK) [24, 25], and polybenzimidazole (PBI) [26]. Among those, PBI‐based membranes impregnated with inorganic acids, such as H₂SO₄ and H₃PO₄, have delivered highly promising results, exhibiting high thermal stability up to 200°C and operation at < 1% relative humidity (RH) [27, 28, 29, 30, 31]. Bjerrum and colleagues reported on the operation of an H₂‐ and O₂‐fed (1 atm) fuel cell equipped with an H₃PO₄/PBI membrane at 190°C with a maximum power density of ca. 0.55 W cm⁻² at ca. 1.2 A cm⁻² [32]. Using a similar membrane, the same group demonstrated a PEMFC with a maximum power density of ca. 1.3 W cm⁻² at ca. 2.5 A cm⁻², 200°C and 3 atm pressure of H₂ and O₂ [33]. Importantly, this cell was operated with no gas humidification and with a significant, 1 vol%, amount of CO present in the hydrogen feed.

These and other reports highlight the promise of PBI‐based PEMs for the intermediate‐temperature hydrogen fuel cells. However, these membranes still present several technological limitations, majorly associated with long‐term durability. Phosphoric acid is subject to leaching and condensation, thus leading to power density decay in the long run [34, 35, 36]. High H₃PO₄ contents in the membrane, while enhancing proton conductivity, facilitate corrosion within the device [37] and enhance gas crossover during extended tests [38]. Moreover, the H₃PO₄/PBI membranes have been observed to be thermally unstable above 200°C [9].

To address the limitations of the H₃PO₄/PBI systems, current research explores alternative membrane scaffold materials and electrolytes. One promising strategy is the use of ionic liquids (ILs) [8, 39, 40] as proton‐conducting media within the state‐of‐the‐art PBI frameworks. The well‐known properties of the ILs, such as extremely low vapour pressures, broad electrochemical stability windows, and often high proton conductivity, render these compounds as highly promising electrolytes for the FC applications [35, 41, 42, 43]. Adibi and co‐workers presented composite membranes combining H₃PO₄/PBI with the [P(MIM)₂][NTf₂]₂ (1,3‐di(3‐methylimidazolium) propane bis(trifluoromethanesulfonyl)imide) and [HMIM][NTf₂] (1‐hexyl‐3‐methylimidazolium bis(trifluoromethanesulfonyl)imide) ILs. The authors tested these materials in a H₂|O₂ fuel cell (1 atm) operating at 190°C and achieved maximal power output of ca. 0.44 W cm⁻² at ca. 0.88 A cm⁻² current density [44]. Xu and colleagues reported on the H₃PO₄‐free PBI membranes impregnated with a [N_1,2,2][OTf] (diethylmethylammonium trifluoromethanesulfonate) electrolyte, which enabled a maximal power density of ca. 62 mW cm⁻² for a H₂|O₂ fuel cell (1 atm) operating at 0.025 A cm⁻² and 250°C [35].

Thus, the IL‐based proton‐conducting membranes are yet to achieve levels of performance such as those provided by the H₃PO₄/PBI systems. Besides, the use of ILs in combination with PBI and other polymer‐based frameworks does not resolve the problem of the electrolyte leaching in operation [45]. This can be addressed by developing alternative electrolyte scaffolds.

Monolayer graphene (Gr) and hexagonal boron nitride (BN) exhibit high permeability to thermal protons while effectively blocking larger molecules and gases. This exceptional selectivity makes them attractive building blocks for proton exchange membranes, enabling efficient proton transport with minimal fuel crossover [46, 47, 48]. However, the use of such monolayer membranes is challenging in practice, and these materials are more commonly used as multilayered scaffolds with a typical thickness of ∼100 µm and highly ordered nanoscale channels that can accommodate different electrolytes, including H₃PO₄ [49]. To improve the proton conductivity of such composites, the BN and Gr surfaces can be functionalised with positively charged polyethyleneimine (PEI), which was proposed to enable electrostatic ordering of the cations and anions of the electrolytes confined within these nanosheets [50]. Such ion‐conducting channels can facilitate rapid proton conduction through nanopores (Figure 1a). In contrast, the nonpolar PBI backbone does not provide such ionic ordering but random ion distributions, which increase electrostatic repulsion between like‐charged species and restrict proton mobility (Figure 1a).

FIGURE 1.

(a) Schematic illustration of the proton transport mechanism through the IL/BN‐Gr and IL/PBI membranes. (b, c) Structures of the (b) cations and (c) anions of the ILs investigated in this work.

To the best of our knowledge, the PEI‐functionalised BN‐Gr composites have not been explored in combination with IL electrolytes as proton‐conducting membranes for the intermediate‐temperature hydrogen‐oxygen fuel cells before. The present work investigated this new class of intermediate‐temperature PEMs using [P_6,6,6,14][NTf₂] ([P_6,6,6,14]⁺ = trihexyl(tetradecyl)phosphonium), [EMIM][NTf₂] ([EMIM]⁺ = 1‐ethyl‐3‐methylimidazolium), [S_1,2,2][NTf₂] ([S_1,2,2]⁺ = diethylmethylsulfonium), and [EMIM][BF₄] IL electrolytes (Figure 1b,c). We investigated the conductivity of the new membranes and of the PBI‐based analogues by hydrogen pumping and tested their performance in the H₂|O₂ fuel cells operating within the 160°C–240°C temperature range.

2. Results and Discussion

2.1. Physical Characterisation and Thermal Stability of the Membranes

The nanosheet membranes were fabricated via a vacuum‐assisted filtration to form a bilayer structure, where the BN nanosheets provide electrical insulation while the Gr nanosheets form the major body of the membrane (Figure 2a). Both layers exhibit a developed network of channels, which can accommodate the IL electrolyte. Cross‐sectional scanning electron microscopy (SEM) images of the membranes displayed a bilayer architecture (Figure 2b and Figures S4–S6). The Gr layer appeared denser and more compact because of the high aspect ratio of the graphene nanosheets, which enhances mechanical strength and provides robust structural support. The BN layer exhibited a more porous and loosely packed morphology. The overall membrane thicknesses ranged within ∼86–94  µm. Energy dispersive X‐ray spectroscopy (EDS) elemental mapping further highlighted the bilayer structure of the scaffold and demonstrated the presence of all key elements throughout the membrane (Figure 2c and Figures S4–S6). Some variations in the spatial distribution of the fluorine and sulphur for the membranes impregnated with the [NTf₂]⁻‐based salts and of phosphorus in [P_6,6,6,14][NTf₂]/BN‐Gr were noted between the BN and Gr layers. The slightly higher concentrations of the ILs in BN might be associated with its less dense structure as compared to Gr. The relatively weak N signal in the BN‐rich region may be due to geometric effects in the cross‐section. Because the BN layer is slightly inclined relative to the electron beam, the low‐energy N K_α emission is strongly absorbed along its escape path, leading to underestimation of the nitrogen intensity compared to the higher‐energy F K_α and S K_α signals.

FIGURE 2.

(a) A schematic of a bilayer structure and a photograph, (b) cross‐sectional SEM image, and (c) EDS elemental maps of a [EMIM][NTf₂]/BN‐Gr membrane. Yellow dashed lines in panels (b, c) highlight boundaries of the BN layer.

Fourier transform infrared (FTIR) spectra of the IL/BN‐Gr membranes exhibited characteristic vibrational features corresponding to their respective IL components, further confirming successful infiltration of the electrolytes (Figure 3a). Specifically, the membranes containing the [NTf₂]⁻ anion produced intense absorption bands in the 1350–1100 cm^{− 1} region, attributed to the stretching of the SO₂ and CF₃ groups [51, 52]. As expected, [EMIM][BF₄]/BN‐Gr lacked these features but exhibited distinct B–F stretching vibrations near 1050 cm^{− 1} [53]. The B‐N stretching region was apparent in all the samples in its characteristic 1375 cm^{− 1} region [54]. Additionally, the C –H stretching region (2800–3000 cm^{− 1}) [52] varied among samples, with [P_6,6,6,14][NTf₂]/BN‐Gr showing the strongest absorption due to the long alkyl chains of its cation. Characteristic N‐H stretching regions due to the presence of PEI within the BN‐Gr framework were present in the 2900–3000 cm⁻¹ region for all membranes [55].

FIGURE 3.

(a) FTIR spectra, (b) XRD patterns, and (c) TGA in N₂ of the [EMIM][NTf₂]/BN‐Gr (purple), [EMIM][BF₄]/BN‐Gr (green), [S_1,2,2][NTf₂]/BN‐Gr (blue), [P_6,6,6,14][NTf₂]/BN‐Gr (red) membranes, and of the BN‐Gr scaffold (black).

X‐ray diffraction (XRD) analysis revealed that the interlayer spacing of the BN–Gr layers increased significantly upon incorporation of the ILs, as concluded from a negative shift of the major (002) diffraction signal from 27° for BN‐Gr to 10°–15° for the IL/BN‐Gr samples (Figure 3b). These shifts correspond to the interlayer spacing expansion from ca. 0.34 nm to approximately 0.65, 0.71, 0.74, and 0.78 nm for the [EMIM][BF₄], [EMIM][NTf₂], [S_1,2,2][NTf₂], and [P_6,6,6,14][NTf₂]‐infused membranes, respectively. The observed expansion correlates with the increasing molecular size of the IL ions. The resulting sub‐nanometre channels provide a confined environment that promotes dense IL packing and enhance ionic interactions, which can be expected to promote the proton transfer. The IL‐based materials also exhibited a low‐intensity signal at 26.7°, reflecting a small portion on non‐exfoliated nanosheets that remained unimpregnated with the electrolyte, which is not unexpected. These results confirm that a significant fraction of the interlayer space within the BN‐Gr scaffold was effectively filled with the examined ILs. It was expected that the resulting network of (sub)nanochannels and micron‐scale channels between the BN‐Gr nanosheets infiltrated with the electrolyte would collectively provide the effective proton conduction pathways (Figure 2b).

Small‐angle X‐ray scattering (SAXS) measurements were performed in an attempt to probe the ordering of the IL within the examined membranes, but no relevant scattering signals could be detected (Figure S7). This is likely due to the small interlayer spacing (< 1.2 nm), which approaches the lower detection limit of SAXS and limits its sensitivity to short‐range or interfacial ionic organisation. The absence of detectable SAXS features suggests that long‐range ionic ordering does not occur within the membrane, and that any structural organisation of the ions is likely confined to sub‐nanometre scales.

Thermogravimetric analysis (TGA) of the composite membranes in N₂ demonstrated their high thermal stability, with no significant mass loss detected until ca. 250°C (Figure 3c). BN is known to be stable at temperatures above 1300°C in air [56], whereas graphene has been reported to be stable up to 575°C in air [57] and up to 800°C in N₂ [58]. Hence, the processes observed in the TGA curves for the examined membranes within the examined temperature range were likely associated with the IL components. The least stable were the [S_1,2,2][NTf₂]/BN membranes, the bulk degradation of which commenced at ca. 210°C, eventuating in a ∼80% mass loss by ∼400°C. The [P_6,6,6,14][NTf₂]/BN‐Gr materials demonstrated a ∼10% mass loss across the 200°C–400°C range, which might be attributed to the release of tightly bound H₂O within the salt prior to commencement of the bulk degradation at ∼400°C. The latter value is satisfactorily consistent with the previous report by Arkhipova et al. [59].

The [EMIM][BF₄]/BN‐Gr membrane demonstrated three stages of mass loss, starting with ∼3% within the 100°C–120°C range, which was likely associated with the release of non‐bound H₂O, and subsequent slow loss of another ∼2% up to 380°C (Figure 3c). The latter might be attributed to the presence of more tightly bound H₂O [60]. Finally, a sharp mass loss at 390°C–410°C is consistent with thermal decomposition of both [EMIM]⁺ and [BF₄]⁻ [60]. The most thermally stable among the investigated membranes were those containing the [EMIM][NTf₂] electrolyte. These materials did not lose any mass until ∼310°C, which marked the onset of bulk degradation ending at ∼470°C and resulting in a ∼58% mass loss, consistent with the previous studies on the high thermal stability of [EMIM][NTf₂] [61].

2.2. Proton Conductivity

The capability of the membranes to conduct protons was assessed using hydrogen pumping, that is, by oxidising H₂ at a 40 wt.% Pt/C/CFP anode and evolving it at an identical cathode with a membrane of interest compressed between the two electrodes (Figure 4a). The H⁺ conductivity measured in this manner (σ _H+) is not the same, and often lower than that derived from the electrochemical impedance spectroscopy (EIS; σ _EIS), as summarised in Tables S1–S4. This highlights the non‐ion‐selective nature of the EIS technique, which measures the impedance associated with the movement of all ions within the system. In contrast, the hydrogen pumping probes the intrinsic proton conduction through the membrane. Herein, the EIS and hydrogen pumping data produced consistent results for the membranes based on the bulkiest and most viscous [P_6,6,6,14][NTf₂] IL (Table S1). For other electrolytes containing smaller and more mobile cations, the EIS consistently produced ionic conductivity values that were significantly, sometimes by an order of magnitude, higher than the H⁺ conductivity derived from hydrogen pumping (Tables S2–S4). Note that poisoning of the Pt/C catalyst with the IL components was unlikely to be a significant factor in these experiments, as the current density versus potential dependencies always exhibited close‐to‐perfect linear ohmic, rather than exponential Butler‐Volmer, behaviour (Figure 4b).

FIGURE 4.

(a) Staircase chronoamperometry at −0.01, −0.02, −0.04, −0.08, −0.1, −0.08, −0.04, −0.02, and −0.01 V recorded at 240°C under 1 atm H₂ (50 mL min⁻¹), (b) corresponding current density versus potential plots (symbols show experimental data; lines are linear fits used to calculate proton conductivity, σ _H+), (c) Arrhenius plot of the temperature dependence of σ _H+ (symbols are experimental data; lines are linear fits used to calculate activation energies within the 23°C–160°C range where apparent linearity was observed), and (d) stability tests at 200°C for the [EMIM][NTf₂]/BN‐Gr (purple), [EMIM][BF₄]/BN‐Gr (green), [S_1,2,2][NTf₂]/BN‐Gr (blue), [P_6,6,6,14][NTf₂]/BN‐Gr (red). Data in panels a and b are shown as mean (curves, symbols) ± standard deviation (shading, error bars) derived from testing n = 3 independent membrane samples of each type. Currents are normalised to the geometric surface area of the 40 wt.% Pt/C/CFP electrodes (1 cm²). Thickness of the membranes was 90 ± 4 µm. Extended data are provided in Figures S8–S15 and Tables S1–S4.

The j‐E plots derived from the hydrogen pumping measurements demonstrated a pure ohmic behaviour (Figure 4b) and were used to calculate proton conductivities of the tested membranes (note that the thicknesses of the membranes were the same within ±4 µm). For the [EMIM][BF₄]/BN‐Gr membrane, the ambient temperature σ _H+ was 0.09 ± 0.01 mS cm⁻¹, rising to a maximum of 0.43 ± 0.04 mS cm⁻¹ at 160°C, after which a decrease in performance was observed with σ _H+ declining to 0.35 ± 0.04 mS cm⁻¹ at 240°C (Figure 4c). Much better performance was provided by [EMIM][NTf₂]/BN‐Gr, which produced the σ _H+ values of 0.38 ± 0.06, 5.5 ± 0.9, and 5.2 ± 0.9 mS cm⁻¹ at ambient temperature, 200°C and 240°C, respectively. Thus, the [NTf₂]⁻ anion substantially improves the proton conductivity as compared to [BF₄]⁻. This might be interpreted in terms of the formation of HNTf₂—a super Brønsted acid enabling high proton conductivity in PEMFC electrolytes [62, 63], as well as possible decomposition of HBF₄ into BF₃ and HF.

Replacing the [EMIM]⁺ cation with [S_1,2,2]⁺ while using the same [NTf₂]⁻ anion resulted in approximately 2‐fold deterioration in σ _H+ at temperatures above 100°C. On the contrary, the [P_6,6,6,14][NTf₂]/BN‐Gr membrane exhibited the highest performance corresponding to the H⁺ conductivity of 1.1 ± 0.2, 8.7 ± 1.3, and 9.5 ± 1.4 mS cm⁻¹ at ambient temperature, 200°C and 240°C, respectively (Figure 4c). The latter result is particularly remarkable in the light of the previous reports on the ionic conductivity of the poly(vinylidene fluoride‐co‐hexafluoropropylene) (PVDF‐HFP) membranes impregnated with [P_6,6,6,14][NTf₂]. For example, Tafur and Romero [64] reported that the ionic conductivity of ([P_6,6,6,14][NTf₂] + Zn(OTf)₂)/PVDF‐HFP at ambient temperature, measured by EIS, was 2 orders of magnitude lower than σ _H+ of [P_6,6,6,14][NTf₂]/BN‐Gr. The advantageous properties of this and other membranes presented herein might be associated with the BN‐Gr scaffold, as will be discussed later in the article.

The stability of the IL/BN‐Gr membranes was examined over 24 h hydrogen pumping experiments (−0.1 V) at 160°C and 200°C (Figures S16 and 4d). At 200°C, the best‐performing [P_6,6,6,14][NTf₂]/BN‐Gr demonstrated ca. 7% decrease in the current density over the initial 4 h but then showed progressive improvement, almost completely restoring initial performance after 24 h. Other membranes improved the H₂ pumping current densities throughout the tests. As alluded to above, the proton conductivity of the examined composites improved up to ca. 200°C, while the Arrhenius‐type dependence was observed up to ca. 160°C (Figure 4c). The apparent activation energies of H⁺ transfer for the best‐performing [P_6,6,6,14][NTf₂]/BN‐Gr and [EMIM][NTf₂]/BN‐Gr membranes were the same within error. Comparable, within ±5 kJ mol⁻¹, values were reported for the [C₆C₁Imi, HSO₃]/PVA membrane [65], while the apparent activation energies of the proton transfer for the PBI‐based membranes impregnated with [BMIM]‐based ILs were significantly higher, around 65–84 kJ mol⁻¹ [34]. The deviation of the σ _H+ versus temperature dependence from the Arrhenius behaviour at temperatures above 160°C observed herein might be associated with partial rearrangement of the ILs at higher temperatures leading to disruption in the proton transport pathways. Nonetheless, the σ _H+ decrease for [EMIM][NTf₂]/BN‐Gr between 200 and 240°C was only 6%.

2.3. H₂|O₂ Fuel Cell Performance

Further, the IL/BN‐Gr membranes were integrated into intermediate temperature H₂|O₂ fuel cells. The high‐frequency resistance of the devices (R _HF) expectedly decreased with temperature (exemplified for [EMIM][NTf₂]/BN‐Gr in Figure 5a). The trends in the R _HF were consistent with the data obtained during the EIS data collected under the hydrogen pumping conditions, with the best results provided by [EMIM][NTf₂]/BN‐Gr within the temperature range of 80°C–240°C (Figure 5b, Figures S17 and S18, Tables S5–S8).

FIGURE 5.

(a) Nyquist plots of the EIS data for a FC equipped with [EMIM][NTf₂]/BN‐Gr at 160 ± 1 (purple), 200 ± 1 (dark purple) and 240 ± 1°C (magenta); inset shows the high‐frequency region with corresponding R _HF values. (b) Temperature dependence of R _HF (EIS are exemplified in Figures S15 and S16 and data are summarised in Tables S5–S8), and (c, d) performance of the H₂|O₂ FCs at (c) 160 ± 1°C and at (d) 160 ± 1°C (purple), 200 ± 1°C (dark purple), and 240 ± 1°C (magenta) equipped with [EMIM][NTf₂]/BN‐Gr (purple, dark purple, magenta), [EMIM][BF₄]/BN‐Gr (green), [S_1,2,2][NTf₂]/BN‐Gr (blue), [P_6,6,6,14][NTf₂]/BN‐Gr (red). (e) Stability of the H₂|O₂ fuel cell equipped with [EMIM][NTf₂]/BN‐Gr at 200°C and 100 mA cm⁻². (f) Comparison of H₂|O₂ fuel cell performance of the [EMIM][NTf₂]/BN‐Gr membrane with the IL‐based membranes reported in the literature [8, 35, 67, 68, 69, 70, 71, 72].

The enhanced ionic conductivity of [EMIM][NTf₂]/BN‐Gr discussed in the previous section was mirrored in the highest fuel cell performance provided by this membrane, as exemplified for tests undertaken in 160°C in Figure 5c. It is worth mentioning that the [P_6,6,6,14][NTf₂]/BN‐Gr membrane, while exhibiting the highest current densities under applied hydrogen flux (Figure 4), delivered a reduced performance of the H₂|O₂ fuel cell (Figures 5a and S19). This can be attributed to the low oxygen solubility in [P_6,6,6,14][NTf₂], blocking the oxygen reduction reaction (ORR) and hence causing a significant increase in the polarisation resistance [66].

Increasing the operating temperature of the FC equipped with [EMIM][NTf₂]/BN‐Gr significantly improved the power output, with the highest peak power density of 82 ± 5 mW cm^{− 2} achieved at 240 °C (Figure 5d). Importantly, the open‐circuit cell voltage of the devices was always close to 1 V, indicating minimal gas crossover through the membrane. The extended performance tests of the fuel cell at 200°C and 100 mA cm^{− 2} demonstrated satisfactory durability of the device with ca. 14% power density loss over 30 h (Figure 5e). EIS data recorded before and after the stability test demonstrated only a slight increase in the high‐frequency resistance but a pronounced rise in the charge‐transfer resistance of ca. 17%, which was comparable to the observed voltage loss (Figure S20). Hence, the FC performance deterioration primarily originated from the electrode‐related processes, possibly associated with structural changes to the nanoparticulate platinum catalysts at elevated temperatures. At the same time, the membrane maintained its conductivity under operating conditions due to the effective confinement of the IL within the laminar channels of the BN‐Gr scaffold.

Comparing the results for [EMIM][NTf₂]/BN‐Gr to alternative IL‐based materials reported in the literature further demonstrates the advantages of the composite membranes presented herein, especially at elevated temperatures (Figure 5f and Table S13). These improvements are attributed to the combined effects of the thermally stable [EMIM][NTf₂] electrolyte and the layered BN‐Gr nanoarchitecture, which provides aligned, confined pathways for the efficient anhydrous proton transport (Figure 1a). The structural integrity of the composite membrane further helps mitigate leaching of the IL and gas crossover, contributing to both high power output and operational durability.

2.4. Proton Transport Mechanism

The superior conductivity of the [EMIM][NTf₂]‐based membranes can be attributed to the high thermal stability and low viscosity of the IL electrolyte, combined with the hydrophobic and anhydrous laminar (sub)nanometre and micron‐scale channels of the BN‐Gr scaffold. Importantly, the in‐plane conductivity of [EMIM][NTf₂]/BN‐Gr was found to be consistently higher than the through‐plane conductivity across the whole temperature examined herein (Figure S21). Coupled with the laminar structure of the membranes visualised by SEM (Figure 2b) and the interlayer expansion induced by the infiltration of ILs into the BN‐Gr discussed above (Figure 3b), these results strongly point towards the channels between nanosheets as major proton transport pathways within the investigated materials [73, 74].

To further test this hypothesis, we compared the ionic conductivities and fuel cell performance of the IL/PBI and IL/BN‐Gr membranes at 160°C (Figure 6a,b; Figure S23; Tables S9–S12). Replacing BN‐Gr with the conventional PBI scaffold resulted in significant, approximately 60‐fold, suppression of the conductivities for all ILs examined (Figure 6a). Correspondingly, the power density of the device equipped with [EMIM][NTf₂]/BN‐Gr was ca. 40 times higher than that provided by [EMIM][NTf₂]/PBI at 160°C (Figure 6b).

FIGURE 6.

(a) Ionic conductivity at 160°C for the IL/BN‐Gr and IL/PBI membranes under H₂|O₂ fuel cell conditions at 160 ± 1°C. (b) Performance of the H₂|O₂ fuel cells equipped with [EMIM][NTf₂]/BN‐Gr (purple) and [EMIM][NTf₂]/PBI (grey) membranes at 160 ± 1°C. (c) Zeta potential of the BN‐Gr (PEI‐functionalised) and thermally treated PEI‐free BN‐Gr scaffolds. (d) Temperature dependence of the ionic conductivity of the PEI‐functionalised (purple) and PEI‐free (grey) [EMIM][NTf₂]/BN‐Gr membranes.

In bulk, imidazolium‐based ILs exhibit supramolecular organisation, including hydrogen‐bond networks and nanoscale segregation into polar and non‐polar domains [75, 76]. Under sub‐nanometre confinement (≈0.5–0.8 nm), this bulk structuring can be significantly altered. Confinement and strong interfacial interactions can suppress isotropic nano‐domain formation and instead promote directional ionic organisation. This is expected to favour the formation of aligned polar regions along the laminar channels, which act as continuous pathways for proton transport [75]. Additionally, the positively changed PEI might promote electrostatic ordering of the IL by attracting the anions toward the channel walls. At the (sub)nanometre scale, such ordering can further support rapid in‐plane proton migration (Figure 1a). π–π stacking and weak C–H···π interactions can further promote formation of nanostructured domains and anion‐rich channels that can act as preferential proton transport pathways [76]. Similar enhancements in conductivity due to nanoconfinement‐induced ordering were reported previously [73, 74], and could be among the factors defining the high performance of the membranes presented herein.

The NTf₂ ⁻ anion, known for its weak coordination and chemical inertness, also facilitates efficient proton transport while suppressing ion pair formation. Transient hydrogen‐bond networks between the imidazolium cation and [NTf₂]⁻ can enable Grotthuss‐type hopping of protons [77] involving H⁺ exchange between Brønsted acid–base sites of the imidazolium cation, specifically the slightly acidic C₂ proton [78], and the NTf₂ ⁻ anion within these domains:

$$\mathrm{EMIM}-{\mathrm{H}}^{+}+{\mathrm{NTf}}_2^{-}\rightleftharpoons \mathrm{EMIM}+{\mathrm{HNTf}}_2\rightleftharpoons \mathrm{EMIM}-{\mathrm{H}}^{+}+{\mathrm{NTf}}_2^{-}$$

This dynamic proton‐exchange relay can enable long‐range proton transport without the need for physical diffusion of bulky ionic species and might explain the high anhydrous conductivity observed in the [EMIM][NTf₂]/BN–Gr membranes where the IL is confined within and between the positively charged nanosheets. In contrast, the non‐polar and electronically inert surface of the PBI framework lacks both interfacial charge and nanoscale confinement to induce comparable ionic ordering, resulting in a random distribution of ions and increased electrostatic repulsion between protons and cations, thereby limiting proton mobility (Figure 6a,b).

The importance of the positive‐charge modification of the membrane scaffold was demonstrated through tests of the PEI‐free control materials synthesised by thermal treatment of BN‐Gr at 600°C under nitrogen to remove the amine functionalities. XPS analysis confirmed the absence of the N 1s signal (Figure S22), while zeta potential measurements showed a dramatic decrease from 4.94 mV for PEI‐BN‐Gr to 0.38 mV after the PEI removal, revealing a near‐neutral surface with minimal electrostatic interaction with the IL (Figure 6c). Correspondingly, the ionic conductivity at 220°C decreased from 16.2 mS cm^{− 1} for the PEI‐modified [EMIM][NTf₂]/BN‐Gr membrane to 8.3 mS cm^{− 1} for the PEI‐free material (Figure 6d). This pronounced decline underscores the important role of the PEI functionalisation for the creation of continuous, low‐barrier proton transport pathways through the laminar (sub)nanometre and micron scale channels, which can be provided by the electrostatically induced ordering of the confined IL (Figure 1). It is also possible that the PEI‐modification of BN‐Gr improves wetting of the scaffold with the IL, including at the sheet boundary regions, further supporting the formation of effective ion‐conducting channels.

3. Conclusion

The present work highlights the promise of combining 2D nanosheet scaffolds functionalised with positively charged PEI and ILs for the design of intermediate‐temperature proton exchange membranes that might find their use in the sustainable power conversion technologies. Herein, this was demonstrated for the hydrogen‐oxygen fuel cells, which achieved the power density of 82 ± 5 mW cm⁻² at 240 ± 1°C when using the [EMIM][NTf₂]/BN‐Gr membrane. The key role of the PEI functionalisation and the BN‐Gr scaffold in enabling high proton conductivity under practical conditions was shown, along with a pronounced effect of the chemical nature of the IL on the performance. The improved performance of the presented membranes was collectively provided by the combination of the proton‐conducting micron‐scale laminar channels between the BN‐Gr nanosheets and sub‐nanometre channels formed within the nanosheets. The important role of the latter was demonstrated through comparisons to the conventional PBI matrix. The robustness and composite nature of the presented membranes open up numerous opportunities for the development of materials for various applications beyond hydrogen fuel cells.

Author Contributions

Mazin Al‐alawi: investigation, conceptualisation, writing – original draft. Kaiqiang He: conceptualisation, investigation, writing – original draft. Mattia Belotti: investigation, validation. Fanmengjing Wang: investigation. Jiffin Oommen: investigation. Douglas R. Macfarlane: supervision, conceptualisation, writing – review and editing. Hoang‐Long Du: conceptualisation, supervision, writing – review and editing. Alexandr N. Simonov: conceptualisation, funding acquisition, writing – review and editing, supervision, resources, project administration. Huanting Wang: conceptualisation, writing – review and editing, project administration, supervision, resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie72707‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.