Ambient humidity cycle-induced transmembrane ion fluxIonic liquids (ILs), often labeled as ‘non-volatile’ solvents due to their resistance to evaporation, are highly valuable in various industries, including flexible electronics and energy conversion31,32,33. ILs demonstrate a spectrum of hygroscopic properties, ranging from highly hygroscopic types like imidazolium halide to less hygroscopic varieties such as imidazolium bistriflimide. For our device, we specifically chose protic ILs with chloride anions for two main reasons. Firstly, the effectiveness of our device depends on the interaction between air moisture and the IL, making the highly hygroscopic protic ILs an ideal choice. Secondly, we selected chloride as the ion to facilitate the conversion of ion current into detectable electrical output because of its effective interaction with the electrodes. The dynamics of protic ILs with moisture, involving both absorption and desorption processes, are crucially influenced by the surrounding humidity levels. During high humidity conditions, the IL tends to absorb water molecules from the air more vigorously. Conversely, in low humidity settings, the IL more readily releases water molecules back into the environment.Our strategy is based on the daily humidity cycle, which refers to the consistent changes in humidity levels experienced over a 24-h period. This change is a continuous cycle that happens every day. When ILs are exposed to varying atmospheric humidity, they will experience a constant disequilibrium with the ambient moisture levels, indicating a perpetual shift in water content. Our device comprises a polycationic membrane sandwiched between two layers of ILs, each with a thickness of 2 mm (Fig. 1a). One of these IL layers is air-isolated (isolated layer) while the other is exposed to the atmospheric conditions (open layer). With increasing or decreasing humidity, the water content in the open layer varies, deviating from the state of the isolated layer. As the isolated layer’s water content lags behind that of the open layer. This persistent discrepancy in water content across the membrane prompts ion flux, which is selectively regulated by the polycationic membrane.Fig. 1: Conceptualization of transmembrane ion flux induced by humidity cycle.a Schematic representation of transmembrane concentration imbalance caused by humidity cycle triggered moisture absorption/desorption of IL. This setup comprises two IL layers enveloping a polycationic membrane, with one layer insulated from the air, while the other is left exposed for direct atmospheric interaction. As the ambient humidity is regularly fluctuating, there will be a continuous concentration imbalance. Inset represents schematic of concentration imbalance induced selective transmembrane ion flux. b Three-day continuous monitoring of ion current (Isc) from the aforementioned system during outdoor tests, coupled with the tracking of the corresponding ambient humidity (φRH). c, d Tracking of Isc during periods of increasing and decreasing ambient humidity, illustrating a strong correlation between humidity cycles and ion current.In our preliminary investigations, we positioned our devices in an open-air environment while keeping them shielded from direct sunlight. We meticulously tracked and depicted the variations in humidity using data collected by hygrometer. In order to validate our hypothesis that continuous changing in humidity could consistently stimulate a transmembrane ion current, we observed the short-circuit current (Isc) across our device over a period of 3 days. As presented in Fig. 1b, the setup could maintain an uninterrupted ion current with Isc fluctuating from ~−14.6 to 17.9 μA. These variations corresponded to changes in relative humidity (RH) between 51.3% and 95.5%. The ongoing current output can be credited to the constant shifts in water content saturation spurred by fluctuating humidity levels, which assures that the exposed IL layer remains either unsaturated or oversaturated. We further advanced our testing by subjecting our devices to practical air conditions with either increasing or decreasing humidity levels (Fig. 1c, d). The outcomes demonstrated a reversal in ion current direction corresponding to the change in humidity trend. Above evidence collectively suggests that the ambient humidity cycle facilitates the achievement of selective transmembrane ion flux.The diurnal humidity cycle is a universal process that occurs in the same consistent fashion throughout the world. The studies above use this daily fluctuation of environmental humidity to create a continuous and sustainable driving force for ion flux and energy conversion. While photovoltaic cells and wind turbines are limited by specific weather conditions and geography, this approach of ours does not and, therefore, can give an even flow of energy every time, independent of the day and the weather. This makes it particularly suitable for areas with consistent low-power requirements. This is well exemplified by solar panels, which perform best in high-irradiance environments and variable-free stable weather conditions (mainly constant clear weather); their output, thus, becomes quite variable. This sharply contrasts with the diurnal humidity cycle’s source of energy, which is versatile, reliable, and can be applied easily to a large number of environments with utter disregard for either local weather patterns or geographic constraints.Constructing pyridine-cluster channelsThe membrane with a dense polycationic cluster configuration was developed by using a bottom-up technique, as shown in Supplementary Fig. 1. Initially, a super-thin, high-density membrane was formed through self-assembly of a block copolymer (BCP). This BCP comprised a crosslinking polyisoprene (PI) segment and an ion transport poly-4-vinylpyridine (P4VP) segment. Following that, S2Cl2 was used to crosslink the assembled BCP membrane, enhancing its mechanical robustness34. The BCP membrane, now sulfur-crosslinked (s-BCP), underwent an acidification process. Concurrently, chloride ions were attached to the P4VP segment due to their electrostatic attraction. This process culminated in the creation of an ultrathin, dense, and durable ion channel membrane (p-BCP).The membrane thickness was measured using a cross-sectional AFM height image (Fig. 2a), revealing an average thickness of ~74 nm. Supplementary Figs. 2, 3 showed the nanostructural details of the BCP nanochannel membrane. The changes in chemical composition have been verified through XPS and FT-IR tests. Following the crosslinking with S2Cl2, the XPS spectra display signals for both Cl and S elements (Supplementary Figs. 4, 5). To validate the protonation of pyridine, detailed XPS spectra for N and Cl elements were taken. As depicted in Fig. 2b, N 1s peak emerges at a higher binding energy of 401.2 eV, signifying the chemical linkage between N and Cl. The detailed XPS spectra for Cl elements validate the presence of chloride post-acidification, thus confirming the creation of the Cl clusters. In the FT-IR spectra, the intensity of the double bond peak diminishes post-crosslinking, and the pyridine peak transitions from 1598 to 1608 cm−1 due to its protonation (Fig. 2c)35.Fig. 2: Construction of the dense polycation-cluster channels.a Height profile for the p-BCP membrane, showcasing an approximate thickness of 74 nm. Inset: Sectional AFM height visualization of the p-BCP membrane placed on a silicon wafer. b Detailed XPS readings for N and Cl elements in both BCP and p-BCP membranes. c FT-IR readings for both the BCP and p-BCP configurations. d 1D GI-SAXS interpretation derived from the 2D GI-SAXS layout, highlighting the periodic design. e SEM depiction coupled with EDS traces for the p-BCP membrane. f TEM visualization of the p-BCP membrane, revealing the hexagonal alignment of the polycation clusters. Inset: AFM height depiction of the p-BCP structure.The dense and regularly arranged polycation-cluster architecture is examined using GI-SAXS, SEM, AFM, and TEM methods. The 2D GI-SAXS pattern reveals a pronounced scattering void in the equatorial direction, signaling a regular vertical arrangement (Supplementary Fig. 6). Additionally, the 1D curve suggests a periodic distance of ~30.6 nm (Fig. 2d and Supplementary Fig. 7). The transmembrane ion-cluster channel structure is confirmed by the cross-sectional SEM image. The EDS patterns on the p-BCP membrane surface provide insights into its elemental makeup (Fig. 2e), aligning well with the XPS findings. The distinct transmembrane cluster arrangement is further validated by the AFM and TEM visuals (Fig. 2f). Post-acidification, the P4VP framework turns more water-attracting and swells due to electrostatic pushback (Supplementary Fig. 8). As a result, the super-thin, crosslinked membrane boasting dense pyridine clusters constructed nanochannels for selective transport of Cl− is successfully crafted.Humidity-driven selective anion flux across membranes with pyridine clustersWe exposed the IL layer, which had a specific water content, to various humidity levels at 25 °C and measured the ion current. This was done to uncover the mechanisms behind the humidity cycle-induced ion current (Fig. 1). In Fig. 3a, we sealed the membrane in conjunction with two IL layer, which both have a water content of about 14.6%—this is balanced with an ambient humidity of roughly 50%. When the device was subjected to RH of 73% and 41%, the equilibrium Isc was found to be ~10.1 μA and −5.4 μA, respectively. This suggests that the ion flux direction can be changed by adjusting the surrounding RH.Fig. 3: Mechanism behind the transmembrane ion flux driven by humidity variation.a The change in experimentally measured Isc when subjected to surrounding RH values of ~73% and 41%. The IL layers’ water content is initially balanced with an ambient RH of about 50%. b Measured Isc mapped against the surrounding RH, ranging from as low as ~0% to as high as ~100%. c Equilibrium water content (\({w}_{{{{{\rm{H}}}}}_{2}{{{\rm{O}}}}}\)) in relation to different RH values from experimental measurement. d Simulated distribution of cation, Cl−, and water number density at the membrane-liquid interface, with the liquid being a mixture of ILs and water at varying concentrations. e Simulated comparative number density of Cl− ion, and cation with respect to water content, with an inset of the simulation model. The error bars represent standard deviations and nd = 5 for each data point.Extending our investigation, we exposed the device to a spectrum of RH values, spanning from ~0% to ~100% (Fig. 3b). We found that increasing the difference in RH from the base 50% correspondingly increased the ion current. For instance, by raising the surrounding RH to 80%, the Isc climbed to 14.8 μA. Conversely, by reducing the surrounding RH to 32%, the Isc reduced to −9.9 μA. When the surrounding RH is set at 50%, the Isc gravitates towards ~0 μA. These findings emphasize that the ion current has a strong association with the surrounding RH. A greater deviation from the 50% RH mark augments the difference in water content across the membrane (Fig. 3c), which in turn influences the ion current. Notably, this device works well under extreme humidity conditions, such as RH of ~0% and ~100% (Supplementary Fig. 9).To better understand how the design of our device impacts the conversion of humidity changes into electrical current, we explored the effect of varying the thickness of the IL layer. We designed devices with the open IL layer thickness ranging from 2 to 1 mm, while maintaining the isolated layer at 2.0 mm. These devices were exposed to a relative humidity of ~73% at 25 °C to evaluate their responsiveness to humidity changes. The findings, detailed in Supplementary Fig. 10, demonstrate that a reduction in the thickness of the open layer leads to a more sensitive response, as evidenced by the output current. Specifically, when the open layer thickness is decreased from 2.0 to 1.0 mm, the time required for the current to stabilize at an equilibrium value is shortened from ~35 to ~10 min. Additionally, the equilibrium current increases from 10.1 to 12.9 μA. This improvement is attributed to the decreased diffusion distance and enhanced rate of moisture absorption. A thinner IL layer means water molecules have a shorter path to travel from the air to the membrane interface, speeding up permeation and thereby boosting water absorption. Furthermore, this reduction in thickness increases the surface area-to-volume ratio at the interface where the air contacts the IL, allowing a greater portion of the IL to interact directly with ambient air, which accelerates moisture exchange.To elucidate the origin of ion current, we examined the distribution of IL species at the liquid-membrane interface on a molecular scale, using full-atomistic molecular dynamics simulations. These simulations revealed that chloride ions Cl− readily penetrate the membrane’s inner regions, while cations predominantly remain in the liquid phase (as illustrated in Fig. 3d). Additionally, we explored the impact of water content on ion distribution, observing in Fig. 3e that the average number density of Cl− and cations varies with water content. Notably, the number density of Cl− increases with rising water content and significantly exceeds that of the cations. We also investigated the diffusive behavior and properties of ILs through mean square displacement (MSD) analysis (Fig. 4a, b). As water content increases, the self-diffusion coefficient (SDC) of Cl− rises more rapidly than that of the cations, indicating that Cl− diffuses faster than cations. The aggregation and rapid diffusion of Cl− within the membrane support our theory that Cl− plays a dominant role in the entire ion current generation process.Fig. 4: Simulated ion diffusion characteristics at the membrane interface.a Analysis of the mean square displacement (MSD) for Cl− and cations in systems with varying water content. b Evaluation of the self-diffusion coefficients for chloride ions and cations relative to water content. The inset shows the variance in self-diffusion coefficients between the two types of ions.Hydrophobic interactions enhanced selective flux across pyridine clustersBeyond our investigation of the membrane, we have also explored how the IL layers affect ion flux and ion current generation. The particular IL under scrutiny possesses a hydrophobic chain that is terminated by ammonium at both ends. These hydrophobic chains can cluster together to reduce exposure to polar or charged entities. Such a gathering of ions in the IL, termed ion clustering, results in them grouping together rather than distributing uniformly. This creates regions within the IL where ions are more densely concentrated.To elucidate the impact of hydrophobic chain length on transmembrane ion movement, we synthesized three polyether amine fatty acid ILs: [PEA230]Cl2, [PEA400]Cl2, and [PEA2000]Cl2. Protic ILs have good ionic conductivity due to mobile ions36, making them ideal for electrochemical devices like batteries and supercapacitors. So we use PEA ILs to reduce internal electrical resistance of the IL layers. In addition, PEA ILs is highly hygroscopic, absorbing moisture from the environment37, which is crucial for our device that generates ion flux through humidity cycles. Their polyether backbone also provides flexibility for varying ion transport, ensuring high ion selectivity. Additionally, they offer good thermal stability38,39. In contrast, common cations in aprotic ILs, such as imidazolium and pyridinium, cannot dissociate protons and lack these advantages.The structures of these ILs were confirmed using 1H NMR and FTIR spectroscopy, as illustrated in Fig. 5a, b, and Supplementary Fig. 11. The 1H NMR spectra revealed the disappearance of the –NH2 peak at 1.48 ppm in PEA, replaced by a new NH3+ peak at 8.13 ppm. Similarly, the FTIR spectra showed the characteristic –CH2 and –CH3 stretching vibration peaks at 2900 cm−1, along with new peaks at 1460 and 1380 cm−1 corresponding to –CH2 and –CH3 groups. The peak at 1115 cm−1 was identified as the stretching vibration of the C–O–C bond. These findings collectively confirm the successful synthesis of the three PEA ILs.Fig. 5: IL clustering enhanced selective ion flux.a, b 1H NMR and FTIR spectra of the ILs with n = 2, 5, and 32. c Power density chart for ILs with n values of 2, 5, and 32, plotted against external resistances spanning from 1 to 106 Ω. d, e Overview of the peak power density and the associated Isc. f Equilibrium water molar ratio at an ambient humidity of ~73% at 25 °C. g The size distribution of cluster in ILs-water mixture system for n = 2, 5, and 32. The error bars represent standard deviations and nd = 5 for each data point.Power density (Pd) is widely recognized as the criterion for assessing energy harvesting potential. We integrated the aforementioned ILs into our apparatus and determined the output power density when subject to a constant ambient RH of ~73%. Initially, the water content in the isolated and open IL layers were both balanced with an air humidity of about 50%. We used the generator (with an effective area, A) to power external resistive loads (Re) and recorded the resultant current (Ie). As depicted in Fig. 5c, the power density, defined as Pd = Ie2Re/A, peaks with an Re of about 10 kΩ for IL with n value of 5. We would like to highlight that our device demonstrates a significant advancement in both power density and current density compared to other devices that generate electricity triggered by air moisture (Supplementary Fig. 12). Stability is another crucial factor affecting the practical application of our device. To evaluate its stability, we continuously monitored the Isc over a period of ~4 weeks (under the same condition with Fig. 3a, RH of ~73%). The results, presented in Supplementary Fig. 13, demonstrate that our device can operate continuously for 4 weeks without obvious degradation in performance.Additionally, we analyzed the variation in power density with respect to n, as illustrated in Fig. 5c–e demonstrating an optimal n value. The length of the side chains significantly impacts power generation: increasing n to 32 or reducing it to two both result in a simultaneous decrease in power density and Isc. Based on Fig. 5f, altering the length of the side chain, either by increasing or decreasing, leads to a reduction in the equilibrium water molar ratio. For n = 5, the water-to-IL ratio stands at around 6:1. This is more than the ratios for n = 2 (5:1) and n = 32 (3:1). The capability to take in more water molecules contributes to a pronounced difference in water content across the membrane, bolstering selective ion flux and subsequently, a greater power output. In addition, as the IL absorbs more water, it enhances ion dissociation, thus increasing the number of free ions within the system. These liberated ions can move more freely throughout the IL and play a crucial role in increasing current generation, as a greater number of ions are available to engage in the transmembrane ion flux. This aligns with the operational mechanism depicted in Fig. 3b. Nonetheless, it’s worth pointing out that even though IL with n = 2 exhibits a substantially higher water molar ratio than IL with n = 32, their power densities and corresponding Isc values are contrary (Fig. 5d, e).We attribute the observed discrepancy to ion clustering, as depicted in Fig. 5g. Ion clustering is the phenomenon where ions in a solution come together to create bigger entities known as clusters. In our study, these clusters of cations arise due to the hydrophobic interactions linked to the length of the side chains. Specifically, [PEA2000]2+ shows a more pronounced propensity to aggregate compared to [PEA230]2+, leading to the formation of larger cation clusters. The larger clusters with stronger charges experience significant electrostatic repulsion from the similarly charged polycationic membrane. This repulsion prevents the clusters from easily penetrating the membrane, thereby enhancing the overall current generation in the device.Scalability of the polycation membrane for energy conversion and storageFinally, we demonstrate the practicality and scalability of using humidity cycle-driven selective ion flux as a renewable energy source, by connecting this device in series or parallel. Initially, we constructed this device on microfluidic platforms (Supplementary Fig. 14a). The design of this microfluidic chip includes two arrays of pores, separated by a central membrane. IL (n = 5) was infused into these pores, and a sealing membrane was used to isolate one side from the air, leaving the opposite side exposed. We introduced Ag/AgCl to measure the ion current. As shown in Supplementary Fig. 14b, we observed that when the pores (each 300 μm in diameter) are connected in parallel, they produce a short-circuit current of ~2700 μA. This result indicates that our device can be effectively scaled in a parallel arrangement. Additionally, we evidence that a single unit can efficiently charge capacitors ranging from 1 to 1000 μF in a brief period (around 2 min), as depicted in Supplementary Fig. 14c. Then, we engineered a circuit, shown in Supplementary Fig. 15, which consists of 16 capacitors connected in series, designed for energy storage. This setup successfully generated a voltage up to ~2.21 V (Supplementary Fig. 14d), sufficient to power a green LED requiring about 1.8 V. Collectively, these findings establish the potential of humidity cycles as a viable and eco-friendly energy resource, which is not heavily dependent on specific geographic or climatic conditions.
