Turing-type nanochannel membranes with extrinsic ion transport pathways for high-efficiency osmotic energy harvesting Authors: Kehan Zou, Haoyang Ling, Qingchen Wang, Congcong Zhu, Zhehua Zhang, Dehua Huang, Ke Li, Yuge Wu, Weiwen Xin, Xiang-Yu Kong, Lei Jiang, Liping Wen Published: 2024 ABSTRACT Two-dimensional (2D) nanofluidic channels with confined transport pathways and abundant surface functional groups have been extensively investigated to achieve osmotic energy harvesting. However, solely relying on intrinsic interlayer channels results in insufficient permeability, thereby limiting the output power densities, which poses a significant challenge to the widespread application of these materials. Herein, we present a nanoconfined sacrificial template (NST) strategy to create a crafted channel structure, termed as Turing-type nanochannels, within the membrane. Extrinsic interlaced channels are formed between the lamellae using copper hydroxide nanowires as sacrificial templates. These Turing-type nanochannels significantly increase transport pathways and functional areas, resulting in a 23% enhancement in ionic current while maintaining a cation selectivity of 0.91. The output power density of the Turing-type nanochannel membrane increases from 3.9 to 5.9 W m−2 and remains stable for at least 120 hours. This membrane exhibits enhanced applicability in real saltwater environments across China, achieving output power densities of 7.7 W m−2 in natural seawater and 9.8 W m−2 in salt-lake brine. This work demonstrates the promising potential of the Turing-channel design for nanoconfined ionic transport in the energy conversion field. Introduction The global energy crisis is becoming increasingly severe with climate change and population growth. The development and utilization of renewable energy sources, such as solar, wind, and nuclear power, has become an urgent task1,2. Osmotic energy, known as blue energy, derived from salinity gradient difference between different solutions, has attracted much attention for its vast reserves and sustainability3,4. In nature, certain living organisms have evolved superior systems to take advantage of such ion concentration gradient. For example, the electric eel can use electrocytes to generate over 600 volts in a shock, which is based on the rapid exchange of thousands of Na+ and K+ though sodium and potassium ion pumps in electric eel’s body5. Inspired by the electric eel, researchers have constructed numerous nanofluidic channels to harvest osmotic energy. Although some of these channels have shown sizable output power densities, the laborious preparation process and high costs may impede their scalability and broader application6. Over the past decade, 2D nanofluidic membranes have emerged as promising avenues for osmotic energy harvesting7–10. Typically, water molecules and ions are transported in the confined space of 2D nanofluidic channels11. Compared with other nanofluidic channel systems, the 2D nanofluidic membranes show great superiority due to their facile fabrication and easy chemical modification12. Many 2D materials, such as graphene oxide (GO)13, black phosphorus (BP)14, and MXene15, can be fabricated into membranes by the simple self-assembly of nanosheets. Specifically, molybdenum disulfide (MoS2) as a typical metal dichalcogenide (TMD) with a regular laminar structure and abundant negative charges, becomes a good candidate for 2D nanofluidic membranes16. Despite the great potential of 2D nanofluidic membranes in osmotic energy harvesting, further advances in energy conversion performance are still limited by the trade-off effect between permeability and selectivity17. To overcome the shackle, various additives (such as quantum dots18, nanofibers19, and nanosheets14,20,21) have been implanted into 2D nanofluidic channels to expand interlayer spacing or supplement space charge. Unfortunately, these additional components increase the resistance of ion transport, which is detrimental to osmotic energy harvesting22. Consequently, there is a pressing demand for the development of a robust methodology for creating nanofluidic channels that can facilitate large ion throughput and ensure high selectivity. In recent years, Turing structures, particularly for water desalination, have attracted much attention due to their superiority in improving membrane permselectivity23–25. This channel structure could play an important role in overcoming the bottleneck of ion selectivity and flux with regard to membranes for osmotic energy conversion, which, however, has rarely been investigated in this field so far. Herein, we fabricate a MoS2 composite membrane with Turing-type nanochannels between the lamellae. The interlaced nanowire channels are constructed using the nanoconfined sacrificial template (NST) strategy. We initially obtained the MoS2/CHNs/SCNF (MCS) membrane by employing the copper hydroxide nanowires (CHNs) as sacrificial templates and sulfonated cellulose nanofibers (SCNFs) as interlocking agents. After dissolving the CHNs, numerous Turing-type nanochannels were generated in the composite membrane, providing extrinsic channels and functional areas to facilitate high-throughput and selective ion transport. This composite membrane, termed Turing-type nanochannel MoS2 composite membrane (TNC-MCM), broke through the undesirable trade-off between ion flux and selectivity, thereby significantly increasing the output osmotic energy density. Using the TNC-MCM as an osmotic energy generator, the output power density reached 5.9 W m−2 by mixing artificial sea water and river water. Importantly, the TNC-MCM shows excellent performance under real seawater conditions throughout China. The maximum output power density of 9.8 W m−2 was achieved in the salt-lake brine condition, further demonstrating its broad applicability. The design proposed in this work paves an effective road for the construction of 2D nanofluidic channels with high ion throughput and selectivity, which demonstrate a promising potential in wide fields, such as energy storage, water treatment, and biomimetic sensers. Results Fabrication and characterization of TNC-MCM As illustrated in Fig. 1a, the MCS membrane was prepared via simple vacuum filtration of a suspension containing MoS2 nanosheets, CHNs, and SCNFs. To remove the CHNs, the MCS membrane was immersed in an ethylene diamine tetraacetic acid (EDTA) solution. During the process, the EDTA solution changed from colorless to light blue, indicating the successful removal of the CHNs from the MoS2 composite membranes (Supplementary Fig. 1). Subsequently, Turing-type nanochannels with diameters of approximately 6 nm were generated between MoS2 lamellae. Traditionally, the interlamellar spacing of restacked MoS2 nanosheets is around 0.62 nm, which can be easily observed via cross-sectional transmission electron microscopy (TEM), as shown in Fig. 1b. However, the NST strategy enables the creation of numerous Turing-type nanochannels with a 6-nm height between the lamellae while preserving the restacked laminate structure (Fig. 1c). Compared to the conventional laminar MoS2 membrane, the TNC-MCM features additional nanochannels and enhanced space charge, resulting in high ion flux and selectivity (Fig. 1d and Supplementary Fig. 2).Fig. 1Fabrication of the Turing-type nanochannel MoS2 composite membrane (TNC-MCM).a Schematic illustration of the fabricating process and interior structure of TNC-MCM. The size of the Turing-type nanochannel is approximately 6 nm. b Cross-sectional TEM image of the MoS2 membrane. c Cross-sectional TEM image of the TNC-MCM. The dashed areas indicate Turing-type nanochannels. d Schematic comparison of the membrane with and without Turing-type nanochannels. Fabrication of the Turing-type nanochannel MoS2 composite membrane (TNC-MCM). a Schematic illustration of the fabricating process and interior structure of TNC-MCM. The size of the Turing-type nanochannel is approximately 6 nm. b Cross-sectional TEM image of the MoS2 membrane. c Cross-sectional TEM image of the TNC-MCM. The dashed areas indicate Turing-type nanochannels. d Schematic comparison of the membrane with and without Turing-type nanochannels. 1D CHNs were synthesized in a transparent, slightly blue solution containing Cu2+ and ethanolamine (Supplementary Fig. 3). Copper nitrate [Cu(NO3)2] was used as the Cu2+ source, and the shape of the CHNs was controlled by adjusting the experimental conditions26. Scanning electron microscopy (SEM) revealed that the CHNs were nano-sized in diameter and several micrometers in length (Fig. 2a). The diameter distribution of the CHNs was approximately 8 ± 3 nm, excluding the effect of gold spraying (Fig. 2b). The aspect ratio of the CHNs exceeded several thousands. High-resolution transmission electron microscope (HRTEM) showed the lattice of the copper hydroxide (Fig. 2c), with a crystal spacing of 0.24 nm corresponding to the (111) planes of orthorhombic Cu(OH)2. Crystallographic studies suggest that the nanowires are composed of positively charged linear clusters of copper hydroxide27. Atomic force microscopy (AFM) confirmed the diameter of the CHNs (Fig. 2d, e, and Supplementary Fig. 4), which aligned with the size of the Turing-type nanochannels. In addition, few-layer MoS2 nanosheets were prepared using the lithium intercalation method28. During exfoliation, MoS2 transformed from the semiconductor phase (2H-MoS2) to the metallic phase (1T-MoS2) due to sulfur atom rearrangements29. High-resolution X-ray photoelectron spectroscopy (XPS) analysis revealed that the as-prepared MoS2 nanosheets contained approximately 60% 1 T phase (Supplementary Figs. 5 and 6). Raman spectrum further verified the coexistence of 1T-MoS2 and 2H-MoS2 (Supplementary Fig. 7). The mean size of the MoS2 nanosheets was about 150 nm (Fig. 2f, g). TEM image showed the crystal lattice of MoS2 nanosheets, with a Mo atom spacing of 0.3 nm (Fig. 2h). According to the AFM characterization, the thickness of the MoS2 nanosheets was around 2 ~ 3 nm, indicating that the nanosheets consist of two to three layers (Fig. 2i, j). The few-layer MoS2 nanosheets exhibited abundant charge density and dispersed easily in deionized water (Supplementary Fig. 8). However, the pristine MoS2 membrane was found to be fragile and could not be peeled off the filter film after filtration process (Supplementary Fig. 9). Therefore, to enhance the mechanical strength of the MoS2 membrane, sulphonated cellulose nanofibers (SCNFs) were introduced. Fourier transform infrared (FT-IR) spectroscopy demonstrated that the SCNFs have abundant hydroxyl groups and sulfonic acid groups, which can combine with MoS2 nanosheets by hydrogen bonds and electrostatic interactions (Supplementary Fig. 10). Furthermore, these oxygen-containing functional groups provide additional negative charges for enhancing selective ion transport (Supplementary Fig. 11)18.Fig. 2Characterization of the TNC-MCM.a SEM image of the CHNs. b Diameter distribution of CHNs obtained through data statistics from the SEM (the orange line is the fitting curve of the diameter distribution using the “Gauss” model). c TEM image of the CHNs. d, e AFM image of the CHNs and the corresponding height profile. f SEM image of MoS2 nanosheets. g Diameter distribution of MoS2 nanosheets obtained through data statistics from the SEM (the orange line is the fitting curve of the diameter distribution using the “Gauss” model). h TEM image of MoS2 nanosheets. i, j AFM image of the MoS2 nanosheets and the corresponding height profile. k Optical image of the TNC-MCM. l Cross-sectional SEM image of the TNC-MCM. m XRD patterns of pure MoS2 membrane, MCS, and the TNC-MCM. The black, red, and blue dashed lines represent the peak positions of MoS2, MCS, and TNC-MCM, respectively. n Dynamic water contact angle of the MCS and TNC-MCM. o Water contact angle reduction rate of the MCS and TNC-MCM. Error bars give the standard deviation from three independent tests. Characterization of the TNC-MCM. a SEM image of the CHNs. b Diameter distribution of CHNs obtained through data statistics from the SEM (the orange line is the fitting curve of the diameter distribution using the “Gauss” model). c TEM image of the CHNs. d, e AFM image of the CHNs and the corresponding height profile. f SEM image of MoS2 nanosheets. g Diameter distribution of MoS2 nanosheets obtained through data statistics from the SEM (the orange line is the fitting curve of the diameter distribution using the “Gauss” model). h TEM image of MoS2 nanosheets. i, j AFM image of the MoS2 nanosheets and the corresponding height profile. k Optical image of the TNC-MCM. l Cross-sectional SEM image of the TNC-MCM. m XRD patterns of pure MoS2 membrane, MCS, and the TNC-MCM. The black, red, and blue dashed lines represent the peak positions of MoS2, MCS, and TNC-MCM, respectively. n Dynamic water contact angle of the MCS and TNC-MCM. o Water contact angle reduction rate of the MCS and TNC-MCM. Error bars give the standard deviation from three independent tests. As demonstrated in Supplementary Fig. 12, a stable suspension with strong Tyndall effect can be formed by mixing specific amount of MoS2 nanosheets, CHNs, and SCNFs. A freestanding membrane with good flexibility is obtained through simple vacuum filtration and post-treatment (Fig. 2k and Supplementary Fig. 13). The TNC-MCM has a typical laminate structure, with a thickness of approximately 3.5 μm (Fig. 2l and Supplementary Fig. 14). X-ray diffraction (XRD) patterns revealed that the introduction of SCNFs and CHNs effectively extends the interlayer spacing, which remains stable after dissolving the CHNs (Fig. 2m). Moreover, the mechanical properties of the composite membrane are well maintained after the introduction of the Turing structure (Supplementary Fig. 15). In the dynamic water contact angle (WCA) test, the WCA of the TNC-MCM decreased rapidly over time, nearly twice the speed of the MCS. This suggests that water can permeate more rapidly through Turing-type nanochannels, facilitating rapid ion transport through the membrane (Fig. 2n, o). Transmembrane ion transport behavior of TNC-MCM The ion transport properties of the TNC-MCM were investigated by clamping the membrane in a symmetrical electrochemical cell with injected electrolyte solutions (Fig. 3a). Potassium chloride (KCl) was used as the standard electrolyte for better assessment of ion transport behavior through the TNC-MCM. Figure 3b shows a series of current-voltage (I-V) curves for KCl concentrations ranging from 0.05 to 1 M. All curves exhibit linear ohmic properties without ionic current rectification, indicating the symmetric structure of the TNC-MCM30. The ionic conductance of the TNC-MCM shows two distinct characteristic behaviors under different electrolyte concentrations (Fig. 3c). The inserts in Fig. 3c illustrate the simplified electrical potential profile of nanochannel in the two concentration regions. At high concentrations, excess counterions near the surface screen the surface charge. As a result, ionic conductivity follows the bulk rule, exhibiting a linear relationship between concentration and conductance. In contrast, at low concentrations, the thickness of the electrics double layer increases, leading to overlapping of the electric double layers31,32. In this situation, the intrinsic electrical properties of the material determine the ion concentration in the channel33. Consequently, the conductance deviates from the bulk value and even shows a slight increase at low concentrations. Figure 3d illustrates the transport behavior of K+ and Cl− ions under a 1000-fold salt concentration gradient (Chigh/Clow = 1000). Both cations (K+) and anions (Cl−) diffuse from high concentration (0.01 M) to low concentration (0.01 mM). Due to the negative charge of the TNC-MCM, K+ preferentially transport through the membrane, generating a net ionic current. Notably, the I-V curves are centrally symmetric under both forward and reverse concentration gradients, indicating the symmetric structure of the TNC-MCM and the absence of a preferential ion diffusion direction34. Moreover, the concentration gradient significantly affects the ion diffusion process. Both the diffusion potential (Ediff) and short current (Isc) increase as the concentration gradient gradually increases (Fig. 3e). We also tested the current density of the MCS under different concentration gradient solutions (Supplementary Fig. 16). The results show that the current density of the TNC-MCM reaches 58 A m−2, approximately 140% higher than that of the MCS (under a 105-fold KCl concentration gradient). This significant rise in current density indicates increased permeability, which is advantageous for osmotic energy harvesting.Fig. 3Transmembrane ionic transport behaviors of the TNC-MCM.a Schematic illustration of the experimental setup for the transmembrane ionic transport measurement. b I − V curves of the TNC-MCM recorded in neutral KCl electrolyte with different concentrations. c Conductance measurement in neutral KCl electrolyte. In the high concentration region, the transmembrane ionic conductance follows the bulk value (the dashed line is a linear fit of the data points in the high concentration region). In the low concentration region, the transmembrane ionic conductance deviates from the bulk value. Inset: the charged state of nanofluid channels. d I − V curves under the reverse of the 1000-fold concentration gradient. The schematic illustration showed the ion transport behaviors under concentration gradient. e Diffusion voltage and current recorded under a series of concentration gradients. The low concentration side was set to 0.01 mM. The dashed lines are used to guide the eyes. f I − T curves of the TNC-MCM recorded in 0.1 M KCl at different pH values (pH 3, 7, and 11) with an external bias alternating between 0.1 and −0.1 V to investigate the stability of ion transport. g Ion conductance of the MCS and TNC-MCM in various salt solutions (0.1 M). Metal cation species are shown on the top x-axis. The dashed lines are used to guide the eyes. Error bars give the standard deviation from three independent tests. Transmembrane ionic transport behaviors of the TNC-MCM. a Schematic illustration of the experimental setup for the transmembrane ionic transport measurement. b I − V curves of the TNC-MCM recorded in neutral KCl electrolyte with different concentrations. c Conductance measurement in neutral KCl electrolyte. In the high concentration region, the transmembrane ionic conductance follows the bulk value (the dashed line is a linear fit of the data points in the high concentration region). In the low concentration region, the transmembrane ionic conductance deviates from the bulk value. Inset: the charged state of nanofluid channels. d I − V curves under the reverse of the 1000-fold concentration gradient. The schematic illustration showed the ion transport behaviors under concentration gradient. e Diffusion voltage and current recorded under a series of concentration gradients. The low concentration side was set to 0.01 mM. The dashed lines are used to guide the eyes. f I − T curves of the TNC-MCM recorded in 0.1 M KCl at different pH values (pH 3, 7, and 11) with an external bias alternating between 0.1 and −0.1 V to investigate the stability of ion transport. g Ion conductance of the MCS and TNC-MCM in various salt solutions (0.1 M). Metal cation species are shown on the top x-axis. The dashed lines are used to guide the eyes. Error bars give the standard deviation from three independent tests. The stability of ion transport through the TNC-MCM was evaluated under 0.1 M KCl solution. Alternating the external bias between +0.1 V and −0.1 V on the membrane system showed that both the negative and positive currents exhibited minimal fluctuation in each cycle, indicating stable ion transport behavior under the applied voltage. Additionally, the TNC-MCM also demonstrated excellent stability over a wide pH range (Fig. 3f). To further assess the advantages of the TNC-MCM in ion transport, we compared the ion conductance of MCS and TNC-MCM in various salt solutions (Fig. 3g and Supplementary Fig. 17). Both membranes exhibit decreased ion conductance with increasing ionic hydration diameters35. However, the TNC-MCM consistently exhibited higher ionic conductivity than the MCS. The potassium and sodium ion conductivities of the TNC-MCM was four times higher than those of the MCS, highlighting the significant advantage of the TNC-MCM in ion transport. Notably, even for Ca2+ and Mg2+ ions with hydration diameters larger than the interlayer distance, considerable conductance was observed, attributed to the presence of Turing-type nanochannels. Osmotic energy harvesting performance of TNC-MCM The MoS2 membrane with supplementary nanochannels exhibits greater potential for osmotic energy conversion than the conventional laminar MoS2 membrane due to its enhanced ion permeability (Fig. 4a). The artificial sea water (0.5 M) and river water (0.01 M) were used to evaluate osmotic energy harvesting performance of the membrane before and after removing the CHNs. I-V curves revealed that the open-circuit voltage (Voc) and short-circuit current (Isc) of the TNC-MCM were 125.8 mV and 5.8 μA, respectively, both higher than those of the composite membranes without Turing-type nanochannels (Fig. 4b). The osmotic potential (Vos) and osmotic current (Ios) were further corrected by subtracting the contribution of the diffusion potential of the power source (Supplementary Fig. 18). The output power density is calculated as P = I2 × RL, where RL denotes the load resistance on the external circuit. As shown in Fig. 4c, the current density decreases with increasing load resistance, while the output power density reaches a maximum of 5.9 W m−2 when the internal resistance is equal to that of the loading resistance (23 kΩ). The corresponding cation transference number (t+) and energy conversion efficiency (η) are calculated to be 0.91 and 33.62%, respectively, indicating a high-performance energy conversion of the TNC-MCM. In particular, compared with the pure MoS2 and MCS membranes, the TNC-MCM exhibits respectively an approximately 140% and 50% increase in output power density (Supplementary Fig. 19), which is attributed to the additional channels with the enhanced ion selectivity and flux, as well as the stable mechanical property for a long-term use. The influence of the SCNFs weight content was then investigated. As the weight content of SCNFs increased from 5% to 12%, the output power density initially increased and then decreased, reaching a maximum value at 10% (Supplementary Figs. 20 and 21). This phenomenon could be explained by a trade-off effect: a suitable weight content of SCNFs expands the interlayer space and increases the space charge density, favoring ion transport. However, excessive SCNFs increase the spatial resistance, which is unfavorable for osmotic energy conversion. In addition, the content of CHNs has a significant effect on the output power density. Both Voc and Isc reached a maximum value when 5 mL of CHNs suspension was added to the TNC-MCM and decreased with further addition (Fig. 4d). While more CHNs can generate more nanowire channels, excessive positively charged CHNs may combine and agglomerate with negatively charged MoS2 nanosheets, affecting the dispersion of the mixed solution and reducing ion permeate flux and ion selectivity. Therefore, the output power density initially increased from 4.6 to 5.7 W m−2 and then dropped substantially to 3 W m−2 when 10 mL of CHNs suspension was added (Fig. 4e).Fig. 4Osmotic energy harvesting performance of the TNC-MCM.a Comparison of ion transport through traditional stacked laminar MoS2 membrane and TNC-MCM. b I-V curves of the MCS and TNC-MCM under 50-fold NaCl concentration gradient (SCNFs weight content of all composite membranes mentioned below is 10%). c Output power density and current density of the MCS and TNC-MCM under a 50-fold NaCl concentration gradient (0.5/0.01 M). d Influence of the volume of CHNs suspension on Voc and Isc. e Influence of the volume of CHNs suspension on output power density. f I-V curves of the TNC-MCM under different NaCl salinity gradients. The low concentration side was set to 0.01 M. g Output power density of the TNC-MCM under different NaCl salinity gradients. h Surface zeta potential of the MCS and TNC-MCM at different pH values. Error bars give the standard deviation from three independent tests. Osmotic energy harvesting performance of the TNC-MCM. a Comparison of ion transport through traditional stacked laminar MoS2 membrane and TNC-MCM. b I-V curves of the MCS and TNC-MCM under 50-fold NaCl concentration gradient (SCNFs weight content of all composite membranes mentioned below is 10%). c Output power density and current density of the MCS and TNC-MCM under a 50-fold NaCl concentration gradient (0.5/0.01 M). d Influence of the volume of CHNs suspension on Voc and Isc. e Influence of the volume of CHNs suspension on output power density. f I-V curves of the TNC-MCM under different NaCl salinity gradients. The low concentration side was set to 0.01 M. g Output power density of the TNC-MCM under different NaCl salinity gradients. h Surface zeta potential of the MCS and TNC-MCM at different pH values. Error bars give the standard deviation from three independent tests. We further measured the I-V curves under varying salinity gradients (Fig. 4f). Both Voc and Isc increased as the salinity gradient changed from 5 to 500-fold, and the corresponding output power density increased from 0.61 to 14.4 W m−2 (Fig. 4g and Supplementary Table 1). It is found that the internal resistance decreases when the concentration gradient increases from 5 to 500-fold. This can be attributed to the enhanced ion transport with low ion resistance in high concentrations. Additionally, the TNC-MCM is also suitable for different salt solutions (Supplementary Fig. 22 and Table 2), indicating its scalability at various environments. Figure 4h shows the zeta potential of the MCS and TNC-MCM under different pH values. Both membranes exhibited slight fluctuations. Notably, the TNC-MCM exhibited a more negative potential than the MCS, which is advantageous for the osmotic energy harvesting performance. Additionally, the TNC-MCM maintained good output power density over a wide pH range, indicating its great potential for harvesting osmotic energy from acid or alkaline industrial wastewater (Supplementary Figs. 23–25 and Table 3). Application of TNC-MCM in natural water Considering the complexity and variability of natural water resources, the TNC-MCM was placed in a custom electrochemical cell for further evaluation of its osmotic energy harvesting performance (Supplementary Fig. 26)36. By using natural seawater from the four major seas of China as high-concentration electrolytes, the output power density reached 7.0 W m−2, 7.17 W m−2, 7.15 W m−2, and 7.73 W m−2 for the Bohai Sea, Yellow Sea, East China Sea, and South China Sea, respectively. We also tested its performance under salt-lake brine conditions, where the output power density can further increase to 9.7 W m−2 (Fig. 5a). The pH values of natural seawater and salt-lake brine are weakly alkaline and weakly acidic, respectively (Supplementary Fig. 27), which show a negligible effect on the output power density of the TNC-MCM. Furthermore, the contents of different metal ions in natural seawater and salt-lake brine were measured using inductively coupled plasma mass spectrometry (ICP-MS). As shown in Supplementary Table 4, all the used natural seawater contains a high concentration of sodium ions. The South China Sea shows a high metal ion content, leading to the higher output power density compared with other seawater samples. The salt-lake brine contains the highest magnesium concentration of 76.02 g L−1 (Supplementary Table 5). Moreover, the total ion concentration of salt-lake brine is the highest compared with natural seawater samples, thus generating the highest power density. Figure 5b demonstrates the long-term stability of the TNC-MCM under artificial seawater. After testing for 120 hours, the membranes still maintained 95% output power density compared to the initial value, indicating the good stability. In addition, we investigated its long-term structural stability. The Turing structure remained virtually unchanged after immersion in various solutions for 48 hours (Supplementary Fig. 28). Similarly, the morphology of the TNC-MCM remained unchanged after immersion in various solutions for 120 hours and remained intact even after two weeks (Supplementary Figs. 29 and 30). Compared to other 2D lamellar nanofluidic membranes9,14,20,22,37–56, the TNC-MCM exhibited excellent osmotic energy conversion performance (Fig. 5c and Supplementary Table 6), highlighting the exceptional ion transport capacity of our membrane and the critical role of Turing-type nanochannels in osmotic energy harvesting.Fig. 5Application viability of the TNC-MCM.a Output power density of the TNC-MCM under different natural seawater and salt-lake brine. b Long-term stability of the TNC-MCM under 50-fold concentration gradient. c Comparison of output power density between the Turing-type membrane and the reported laminar micron-sized 2D membranes. The dashed line indicates the commercialization benchmark. Application viability of the TNC-MCM. a Output power density of the TNC-MCM under different natural seawater and salt-lake brine. b Long-term stability of the TNC-MCM under 50-fold concentration gradient. c Comparison of output power density between the Turing-type membrane and the reported laminar micron-sized 2D membranes. The dashed line indicates the commercialization benchmark. Discussion In summary, we have developed a paradigm for channel structure, referred to as the NST strategy, for 2D lamellar MoS2 membranes featuring extrinsic nanofluidic channels, which significantly enhances osmotic energy conversion. Using the NST strategy, Turing-type nanochannels were generated between the interlayers of pristine MoS2 membranes following the removal of self-made Cu(OH)2 nanowires. Compared with conventional 2D lamellar membranes, the extrinsic nanochannels in the TNC-MCM increase transmembrane ionic flux while preserving the intrinsic properties of MoS2. The TNC-MCM exhibits an output power density of 5.9 W m−2, approximately 50% higher than that of pristine MoS2 composite membranes. Notably, the TNC-MCM maintains a high output power density across a wide pH range, meeting the requirements for applications in specific environments. Owing to the distinctive channel structure, the TNC-MCM demonstrates exceptional performance in various real seawater environments. The output power densities can achieve 7.7 and 9.8 W m−2, respectively, by using natural seawater and salt-lake brine as electrolytes. This work underscores the critical role of membrane channel structure design in osmotic energy harvesting and serves as an inspiration for structural optimization and innovation of other 2D lamellar nanofluidic membranes. Methods Chemicals MoS2 (1 mg mL−1) was purchased from Nanjing NXNANO Tech. Co., Ltd. Sulphonated cellulose nanofibers (SCNF) were purchased from Tianjin Woodelfbio Cellulose Co., Ltd. Polycarbonate (PC) filter film (pore size ∼0.2 μm) was obtained from Millipore. Copper(II) nitrate (Cu(NO3)2, 99%) was purchased from Thermo Scientific. Ethanolamine (C2H7NO, >99%) was purchased from Sigma. Sodium chloride (NaCl, 99.8%), potassium chloride (KCl, 99.8%), lithium chloride (LiCl, 99%), magnesium chloride (MgCl2, 99%), and calcium chloride (CaCl2, 96%) were purchased from Aladdin. All of the chemicals, including sodium hydroxide (NaOH), potassium hydroxide (KOH), and hydrochloric acid (HCl), were analytically pure. All the experimental solutions were prepared using degassed Milli-Q water (18.2 MΩ cm‒1). Synthesis of copper hydroxide nanowires (CHNs) The CHNs were synthesized by mixing equal volumes 1.4 mM aqueous solution of ethanolamine and 4 mM Cu(NO3)2 under vigorous stirring. Subsequently, the mixture was left at ambient temperature (25°C) for at least 1 day. Fabrication of the Turing-type nanochannel MoS2 composite membrane (TNC-MCM) The MoS2 dispersion (6 mL) was mixed with a certain amount of SCNF suspension. Sonicating the mixing solution for 10 minutes. Then, a certain amount of CHN solution was added into the mixing solution and stirred for at least 15 min to form a uniform suspension. The MoS2/CHNs/SCNF (MCS) membrane was obtained by vacuum filtering the suspension on the PC filter film. To generate the Turing-type nanochannels, the composite membrane was immersed in a 0.15 M EDTA solution for 3 hours, followed by washing the membrane with deionized water repeatedly. After annealing at 60 °C for 24 h, the TNC-MCM can be peeled from the PC filter film. Preparation of the TEM samples The cross-section TEM samples were prepared by ultramicrotomy. First, a small piece was removed from the membrane and embedded in an epoxy resin. Then, the sample was sectioned by using a diamond knife on an ultramicrotome. Subsequently, the ribbons of the slices were collected from the deionizer bath and deposited onto a 300-mesh lacey carbon film copper TEM grid. Characterization The samples’ morphology was captured by field-emission scanning electron microscopy (SEM, Hitachi S-4800) and transmission electron microscopy (TEM, JEM 2100 F). The atomic force microscope (AFM) images were taken by a multifunctional scanning probe microscope (Bruker Multimode 8). X-ray photoelectron spectroscopy (XPS) spectra were collected on an Al Kα source (ESCALAB 250Xi). FT-IR spectrum was collected on an Excalibur 3100 spectrometer. Raman spectra were collected on a Renishaw in Via Raman microscope. The zeta potential of the composite membranes was measured using a Zetasizer Nano (ZS90, Malvern Instruments Ltd., Malvern, England). X-ray diffraction (XRD) patterns were collected on a Bruker D8 Focus diffractometer using an incident wavelength of 0.1540593 nm (Cu Kα radiation) and a Lynx-Eye detector. The contact angles of the membranes were measured at room temperature using an OCA25 contact angle measuring instrument (Dataphysics, Germany) with an adjustable pick-up lens. Tensile test was performed by a tensile-compressive tester (M5-2) with a pulling rate of 0.1 mm min−1. The cation concentrations were determined using the PerkinElmer NexION 1000 G ICP-MS. Electrical measurement The composite membranes were mounted between a customized electrochemical cell for testing. The ion transport properties and osmotic energy conversion tests were carried out on a Keithley 6430 semiconductor picoammeter (Keithley Instruments, Cleveland, OH). The effective testing area of the composite membranes was about 0.03 mm2. The transmembrane electrical potential was applied using a pair of homemade Ag/AgCl electrodes. The testing solutions are all close to neutral except for the pH testing. Diluted HCl, NaOH and KOH were used to regulate the pH values of the electrolyte for specific test environments. The electrochemical cell was repeatedly washed with the testing solutions before each measurement and rinsed with deionized water after the test. During the stability test, the membrane was clamped within the electrochemical cell and remained in the testing solutions at all time. The load resistance was consistent across all tests, and the current was continuously recorded for 300-second intervals. The current measurements were conducted every 24 hours over six consecutive days. The output power density was calculated based on the current and external resistance values. To ensure the accuracy of the test results, the salt solution was replaced every 24 hours to maintain a constant concentration gradient. Calculation of the layer spacing of the composite membranes The XRD patterns show the intense peak at ~14.5° and attributed to the (002) peak of restacked MoS2 nanosheets. According to Bragg’s law18:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2d\sin \theta=n\lambda$$\end{document}2dsinθ=nλwhere d, θ, λ refer to the layer spacing, angle between the incident wave and the scattering plane, wavelength of the incident wave and n = 1. The corresponding layer space of MoS2 and TNC-MCM membrane is about 6 Å and 6.2 Å. Electrode calibration The short circuit current (Isc) and open circuit voltage (Voc) could be obtained by the I-V curve. The measured Voc actually consisted of two parts: the Vos contributed by the membranes and the redox potential on the electrode (Eredox). The value of Vos can be derived by subtracting the contribution of the redox potential (Supplementary Fig. 18). In this work, the value of Eredox was measured using an experimental method.2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{{os}}={V}_{{oc}}-{E}_{{redox}}$$\end{document}Vos=Voc−Eredox Energy conversion efficiency Normally, placing a negatively charged membrane in the solution with a given concentration gradient, the cation transference number (i.e. t+) can be calculated as57:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${t}_{+}=\frac{1}{2}\left(\frac{{V}_{{os}}}{\frac{{RT}}{{zF}}{{{\rm{In}}}}\left(\frac{{\gamma }_{{c}_{H}}{c}_{H}}{{\gamma }_{{c}_{L}}{c}_{L}}\right)}+1\right)$$\end{document}t+=12VosRTzFInγcHcHγcLcL+1where Vos, R, T, F, z, γ, and c refer to the osmotic potential, universal gas constant, absolute temperature, Faraday constant, charge number, activity coefficient of ions and ion concentration, respectively. The energy conversion efficiency can be defined as the ratio of the output energy (electrical energy) to the input energy (Gibbs free energy of mixing). Accordingly, the energy conversion efficiency, corresponding to the maximum power generation can be calculated as58:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\eta }_{\max }=\frac{1}{2}{\left(2{t}_{+}-1\right)}^{2}$$\end{document}ηmax=122t+−12 Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Supplementary Information Reporting Summary Transparent Peer Review file