A subnano-confinement in robust MoS₂-based membranes for high-performance osmotic energy conversion

Abstract

Osmotic energy harvesting from salinity gradients shows great potential for sustainable electricity generation, which can be fulfilled using two-dimensional ion-selective nanofluidic devices. Metal dichalcogenide membranes like MoS₂ exhibit good anti-swelling properties in aqueous solution and can be applied in nanofluidic device development. However, conventional MoS₂-based membranes encounter the major issue of low ion selectivity, reducing the electricity generation efficiency. In this paper, we propose the strategy of subnano-confinement using the environmentally benign hydrophilic bacterial nanocellulose (BNC) with negative charges to create high ion-selectivity channels in robust MoS₂-based membranes. The developed membrane exhibited an interlayer spacing of 9.8 Å with desirable negativity in nanochannels, thus generating a favorable confinement for enhancing Na⁺ transport but blocking Cl⁻. The tested membrane provided an area of 0.78 mm², exceeding those of other reported macroscopic-scale membranes. The electrochemical device delivered the power densities of 73 and 233 W m⁻² at ambient temperature and 343 K, respectively, under a 50-fold concentration gradient, outperforming previously reported 2D nanofluidic membranes by a factor of up to 70. Furthermore, the membrane exhibited exceptional long-term stability up to 40 days without performance decay. The current work makes a breakthrough in developing 2D nanofluidic membranes for harvesting osmotic energy.

---

Broader context

There is a long-held dream of using reverse electrodialysis (RED) to harvest osmotic (blue) energy at an estuary where salinity gradients naturally exist as fresh river water meets the ocean. Herein, we propose to use the environmentally benign bacterial nanocellulose as the subnano-confinement in the nanochannel space of an MoS₂ membrane with good anti-swelling properties. The subnano-confinement strategy has dual functions: (1) adds negative charges to create high sodium ion selectivity and (2) enlarges the nanochannel space to enhance ion flux, thus significantly improving the electricity generation efficiency. Furthermore, the membrane exhibited exceptional long-term stability for at least 40 days without performance decay. The current work made a big advancement in developing 2D nanofluidic membranes for harvesting osmotic energy.

Introduction

Osmotic energy, a kind of blue energy, can be generated from abundant and renewable sources by the salinity gradients between low-salinity river water and high-salinity saltwater due to the release of Gibbs free energy during mixing.^1–6 Motivated by the biomimetic discoveries emerging from the study of electric eels, high electric current can be produced by co-ion transport in a particular direction, which converts the energy of a salinity gradient into electrical power.^7,8 The presence of highly charge-selective and ion-conducting channels is a prerequisite for the devices to be applied in osmotic energy conversion.

Traditional reverse electrodialysis (RED) systems based on ion-exchange membranes can be used to harvest the osmotic energy; however, there is still a large gap with the commercial benchmark of 5 W m⁻² due to the inferior selectivity and high internal resistance.^9–11 Alternatively, two-dimensional (2D) nanofluidic channels exhibit superior ion selectivity in an electrically dispersive layer.^12–14 For example, the device with the atomic-layer thick molybdenum disulfide (MoS₂) having a single nanopore achieved a power density of 10⁶ W m⁻²,¹⁵ which can be regarded as the theoretical maximum value. It is noteworthy that the power density of the real membrane should be greatly reduced due to the existence of pore defects (lowering the ion selectivity), strong ion concentration polarization resulting from high pore density and large membrane thickness (slowing down the ion transport rate). In this regard, 2D nanofluidic membranes, including graphene oxide (GO), titanium carbide (MXene), covalent organic frameworks (COFs), vermiculite and transition metal dichalcogenides (TMDs), are developed for osmotic energy conversion due to the scalability of membrane fabrication and high ion transport rates.^10,16–23 In contrast, there are no other functional groups on the surface of MoS₂ nanosheets and hence they are not prone to swelling and folding in the presence of solvents.^24,25 This endows MoS₂ with better stability and antifouling properties than GO or MXenes in aqueous solution. However, MoS₂ monolayers with an interlayer spacing distance of 0.62 nm can largely facilitate water (molecular size: 0.276 nm) permeation, but it is difficult to selectively transport hydrated ions, including Cl⁻ (0.664 nm) and Na⁺ (0.716 nm), resulting in low energy conversion.^26,27 Therefore, manipulation of ion transportation is an essential consideration for laminar 2D stacked MoS₂-based membranes consisting of parallel nanosheets due to their tunable channel size. Zhu et al. used cellulose nanofibers (CNF) as a reinforcing agent with a chemically exfoliated high-concentration metal phase to modify MoS₂ nanosheets to prepare stable and robust two-dimensional metal–MoS₂ composite membranes as high-performance osmotic power generation devices.¹⁸ However, in spite of their achievement of a power density of 6.7 W m⁻², higher than the commercial benchmark, there is still too much space for improvement due to undesirable low ion selectivity and poor stability for practical applications.

Based on the prevailing understanding of 2D nanochannels in laminar membranes, the synergistic effect between the intrinsic properties of the membrane material itself and the external functions can be reasonably applied to regulate ion selective transportation in osmotic energy conversion.²⁸ For electrical double layer (EDL) controlled nanochannels of a 2D MoS₂-based membrane, the increasing negativity in nanochannels as one of the external functions is favorable for the cation selectivity.^29–31 Bacterial nanocellulose (BNC),³² consisting of β-D-glucopyranose units linked by β-1,4-glycosidic bonds, is a linear, unbranched exopolysaccharide synthesized by some bacteria, possessing high negativity. The use of BNC and MoS₂ nanosheets to design a composite membrane is a solution for not only improving mechanical strength and but also intensifying ion selective transport. Herein, we report MoS₂/BNC composite membranes as an effective platform for osmotic energy conversion (Fig. 1). MoS₂ nanosheets can tightly integrate with BNC through S–H bonds, and BNC is confined inside ion transport channels on the sub-nanoscale. The improved space negativity effectively intensifies the ion selectivity for permeation, resulting in a high power density of 73 W m⁻² and stability of 40 days using the developed robust membrane with the area exceeding those of the previously reported 2D nanofluidic membranes. The successful preparation of MoS₂/BNC composite membranes provided a promising strategy to assemble a high-performance osmotic energy conversion device.

Fig. 1 Schematic of osmotic energy conversion. Insets: Cation transmembrane transport through a confined channel and the interaction between MoS₂ nanosheets and one-dimensional bacterial cellulose (BNC).

Results and discussion

Microstructure and basic characterization of MoS₂/BNC nanosheets

The crystal structure of the prepared MoS₂ nanosheets consists of a metallic-phase (1T MoS₂) or a semiconductor phase (2H MoS₂). In general, the metallic-phase MoS₂ (1T MoS₂) tends to achieve a more desirable cation selectivity than the other one.³³ Because of its thermodynamic sensitivity, the metastable 1T MoS₂ would be more likely to transform into a more stable 2H MoS₂ at high temperatures, which is not favorable for the growth of 1T MoS₂.^34,35 Hence, a milder hydrothermal method using a lower temperature was selected for the synthesis of 1T MoS₂.³⁶ The initial starting material was the octahedral structured MoO₃, and urea was used as a weak reducing agent to preserve the original crystal structure of MoO₃ during its conversion to the MoS₂ matrix (Fig. S1 and Experimental section, ESI†). The XRD pattern of the prepared MoS₂ (Fig. 2a) shows the characteristic diffraction peaks at 14.0°, 32.6°, and 57.3°, which correspond to the (002), (100), and (110) crystalline planes of MoS₂ (JCPDS card, 73-1508).^27,37 Fig. S2 (ESI†) displays the Raman spectra of MoS₂, with the peaks at 148, 224, and 322 cm⁻¹ corresponding to the J₁, J₂, and J₃ phonon modes of 1T MoS₂, respectively. Vibrational patterns of 2H MoS₂ were also detected at 402 (A_1g) and 379 cm⁻¹ (E¹_2g). This demonstrates that the prepared MoS₂ exists as a mixture of metallic and semiconductor phases.^38–40 Additionally, the content of the 1T phase in the prepared MoS₂ was analyzed by XPS (Fig. S3, ESI†). The Mo 3d spectra, shown in Fig. S3a (ESI†), reveal the 1T MoS₂ peaks at 232.2 and 229.0 eV, which are attributed to Mo⁴⁺ 3d_3/2 and Mo⁴⁺ 3d_5/2, respectively, while the characteristic peaks of the 2H phase at the binding energies of 223.4 and 230.1 eV are also observed. The peak at 226.3 eV was attributed to the 2s S. In a similar vein, the S 2p spectra (Fig. S3b, ESI†) display 2H peaks at 164.2 and 162.1 eV, and 1T peaks at binding energies of 163.1 and 161.9 eV for S 2p_5/2 and S 2p_3/2, respectively.^41–44 Comparing the respective peak areas, the content of 1T MoS₂ is significantly higher than that of 2H MoS₂, indicating the successful preparation of a higher amount 1T MoS₂ under milder conditions. TEM images clearly show the morphology (Fig. S4a, ESI†) and the lattice planes (Fig. 2b) of the prepared MoS₂ nanosheets. In line with the previous Raman and XPS data, the TEM image also shows that the triangular lattice of 1T MoS₂ and the honeycomb lattice of the 2H phase coexist (Fig. 2b).^45,46 It is also clear that the interatomic distance for the 2H phase is 0.34 nm, while the lattice spacing for the 1T phase is 0.27 nm. However, both 2H and 1T MoS₂ are more favorable for water permeation but not conducive to the selective transmission of Na⁺, which limits the application of MoS₂ nanosheets for osmotic power generation. AFM (Fig. 2c and d), SEM images (Fig. S4b, ESI†) and particle distribution (Fig. S5b, ESI†) show the prepared 2D MoS₂ and MoS₂/BNC nanosheets with the thickness of 1.86 nm and 5.35 nm, respectively, and a size of 0.5–1 μm, consistent with the previous report.¹⁸ Due to a layer of adsorbed water molecules under each flake or folding, the thickness may be higher than the ideal 2D nanosheets. Therefore, the appropriate interlayer spacing is a prerequisite for ion permeation.

Fig. 2 Characterization of MoS₂ and MoS₂/BNC membranes. (a) XRD patterns of as-prepared MoS₂. (b) TEM images of as-prepared MoS₂. AFM of (c) MoS₂ and (d)MoS₂/BNC nanosheets (note that there is always a layer of adsorbed water molecules under each flake). (e) Digital photograph (left) (diameter 40 mm and thickness 0.1 mm), cross-sectional SEM images (middle and right) of the MoS₂/BNC₃ composite membrane. (f) FT-IR spectra of BNC, MoS₂ and MoS₂/BNC₃ powder. (g) XRD images of the MoS₂ and MoS₂/BNC₃ membranes. (h) TGA analyses of MoS₂, BNC and MoS₂/BNC_X membranes.

Rich hydroxyl groups of BNC with a three-dimensional network structure can tightly combine with MoS₂ nanosheets by S–H bonding. As a result, the confined interlayer spacing of the membrane is efficiently adjusted and the aggregation of MoS₂ nanosheets is successfully inhibited. This also effectively contributes to the stability of MoS₂/BNC composite colloids, which was verified by the obvious Tyndall effect after more than 90 days (Fig. S5, ESI†) and guarantees the successful preparation of the composite membrane. The resultant MoS₂/BNC composite membrane via vacuum filtration displayed a representative layered structure (Fig. 2e, the magnified parts show the tight combination between the membrane and the substrate from Fig. S6, ESI†), consistent with the reported two-dimensional membranes, e.g. GO and MXenes.^47,48 A homogenous distribution of Mo, S, O and C can be observed from the EDS mapping in Fig. S7 (ESI†). This shows that the BNC was successfully intercalated into the MoS₂ membrane, consistent with the analysis in Fig. 1. As shown in Fig. S6b (ESI†), the critical load is about 2.1 mN, echoing a strong combination of the functional membrane and substrate. The FT-IR spectra of the prepared MoS₂/BNC show the successful combination of MoS₂ and BNC characteristic bands (Fig. 2f, Fig. S6(c) and Supplementary Note 1, ESI†) due to the detection of S–H at ∼2552 cm⁻¹. The XRD patterns of MoS₂ membranes with BNC incorporation are displayed in Fig. 2g and Fig. S8 (ESI†). The layer spacing of the membrane can be obtained from the diffraction peak based on the Scherrer formula, and the addition of BNC shifts the peak to a lower angle.⁴⁹ This reflects the enlarged interlayer spacing of the membrane with increasing content of BNC, helpful to improve the ion permeation. From TGA testing, the BNC contents in the series of composite membranes were determined to be about 0, 1, 3, 8, 9 and 20 wt%, respectively; and the corresponding interlayer spacings are 8.83 ± 0.17, 9.68 ± 0.21, 9.79 ± 0.22, 9.86 ± 0.22, 9.92 ± 0.23 and 10.25 ± 0.23 Å (Fig. 2h and Fig. S8, ESI†), respectively. Furthermore, with the increase of BNC content, the hydrophilicity of the composite membrane gradually enhances (Fig. S9, ESI†), beneficial for water permeation.

Superior mechanical properties of composite membranes are also important for practical applications. Hence, mechanical strengths of composite membranes with the different BNC contents were evaluated. As shown in Fig. S10 and Supplementary Note 2 (ESI†), the maximum tensile force, tensile strength, yield strength, and failure stress all increased with BNC content. The mechanical strength of the composite membrane is obviously higher than that of the nylon substrate. The maximum bearing capacity, tensile strength, failure stress, and yield strengths of the composite membranes increased from 11.53 to 27.41 N, 25.6 to 60.9 MPa, 25 to 56.5 MPa, and 19.3 to 60.8 MPa, respectively, with the corresponding BNC contents from 0, 1, 3, 8, 9 to 20 wt%. The increase of mechanical strength is attributed to the addition of flexible BNC and the formation of network structure through hydrogen or S–H bonds between the BNC and MoS₂.

Transmembrane ion transport

To investigate the ion transport behaviors of the subnano-channels due to the interlayered BNC, the resultant MoS₂/BNC composite membrane was sandwiched between two electrochemical cells (Fig. 3a). Potassium chloride (KCl) was selected as the standard electrolyte solution due to its similar anion and cation diffusion and migration rates. The current–voltage (I–V) curves at varying KCl concentrations from 0.05 to 1 M are shown in Fig. 3b. The composite membrane exhibits symmetric structure and ion conducting properties due to all current responses showing linear ohmic characteristics.^9,50 To investigate ion transport behavior, the relationship between ion conduction and ion concentration is displayed in Fig. 3c. Over a broad range of KCl concentrations, the distinctive behaviors revealed the different ionic conductivities of composite membranes. In the region of high salty concentration (>1 M), the ionic conductivity of the composite membrane was linearly related to the KCl concentration, which was comparable to the ionic conductivity of the electrolyte solution. In the low concentration region (<1 M), the ionic conductivity of the composite membrane gradually reached a plateau and deviated from the linear relationship. An electrical double layer (EDL) is formed when the surface charge of membranes draws the counter ions in solution. When the nanofluidic channel size and the EDL thickness in the composite membrane are similar, excellent ion selectivity can be obtained. According to previous reports,^14,51 EDL thickness decreases as electrolyte concentration rises. This indicates that, at low concentrations, surface charge mostly regulates ionic transport. Furthermore, for the MoS₂/BNC membrane, due to increasing space negativity of nano-channels (Fig. S11, ESI†) induced by the confined interlayered BNC, the surface-charge-governed ionic transport behaviors would be intensified compared with the pristine MoS₂ membrane. In addition, the interlayered BNC results in a larger cation flux due to enlarged interlayer spacing and inhibits restacking of nanosheets.

Fig. 3 Cationic K⁺ transport behavior of the MoS₂/BNC composite membrane with 3 wt% BNC content. (a) Schematic of the experimental setup to measure transmembrane ionic transport; (b) I–V curves of composite membranes in neutral KCl electrolyte of different concentrations; and (c) conductance measurement in KCl electrolyte of different concentrations.

To optimize the contribution of space charge for ion transport, the numerical model based on Poisson–Nernst–Planck and Navier–Stokes (PNP–NS) equations was built (Fig. S12 and Supplementary Note 3, ESI†).^52,53 The effects of surface charge and space charge on ion transport under a 50-fold NaCl concentration gradient (0.5 M/0.01 M) are analyzed. The simulation results in Fig. 4 reveal that the nanochannel with coexistence of surface charge and space charge displays a stronger Na⁺ selectivity than that with either surface charge or space charge. As shown in Fig. S13 (ESI†), the ion selectivity of the nanochannels firstly improved with the increase of the space charge density in the nanochannel from −2 to −5 mC m⁻², indicating that the increase of BNC can contribute to energy conversion. However, excessively high space charge densities (>−10 mC m⁻²) cause the phenomenon of concentration polarization, which decreases the effective concentration gradient on both sides of the nanochannels resulting in the reduced driving force for ion transport. Therefore, the BNC confinement in the interlayer space should be controlled at a suitable content.

Fig. 4 Schematic showing numerical simulation of the ion transport performance under a 50-fold NaCl concentration gradient with the charge distribution of nanochannels (left column) and the concentration distribution of Na⁺ cations (middle column) and Cl⁻ anions (right column) in the 2D nanochannels. (a) Charge-free 2D nanochannels; (b) only surface charge in 2D nanochannels; (c) only space charge and (d) synergistic effect of surface and space charges in 2D nanochannels.

To better understand the effect of interlayer charge on ion transport, the migration energy barriers of Na⁺ and Cl⁻ in different structures were investigated based on Density Functional Theory (DFT, Supplementary Note 4, ESI†), as shown in Fig. S14 (ESI†). Simulation results (Table 1) indicate that the interlayer energy barrier for Na⁺ through pure MoS₂ is 0.093 eV. When BNC was inserted into the interlayer of MoS₂ nanosheets, resulting in the interlayer spacing of 9.79 Å, the migration energy barrier of Na⁺ decreased to 0.079 eV, indicating the easier transport of Na⁺. Due to the limiting effect of the surface charge and spacing charge on the Cl⁻ migration process, its adsorption energy at the outer edge of MoS₂ is about 5 times that of Na⁺, showing that it is difficult for Cl⁻ to migrate into the interlayer of the resultant membrane. In spite of increasing interlayer spacing by the addition of BNC resulting in the decrease of the migration energy barrier of Cl⁻ from 0.476 to 0.404 eV, it is still about 5 times that of Na⁺, further showing the difficult transport of Cl⁻ in subnano-channels of the composite membrane due to the confinement with a negative charge. In addition, the enlarged interlayer spacing for the MoS₂/BNC composite membrane can effectively reduce the migration energy barriers of both Na⁺ and Cl⁻. However, the difference between Na⁺ and Cl⁻ migration energy barriers is not sufficient to maintain high Na⁺ selectivity, and the synergetic effect of the space charge provided by BNC and the surface charge of MoS₂ is responsible for the ion transport of Na⁺, consistent with the results of numerical simulation.

Table 1 The migration energy barrier of Na⁺ and Cl⁻ in the MoS₂ membrane and the MoS₂/BNC composite membrane

Materials	Interlayer spacing (Å)	Ion	Activation energy (eV)
MoS2	8.83 ± 0.17	Na^+	0.093
MoS2/BNC3 (3 wt%)	9.79 ± 0.22	Na^+	0.079
MoS2	8.83 ± 0.17	Cl^−	0.476
MoS2/BNC3 (3 wt%)	9.79 ± 0.22	Cl^−	0.404

---

Osmotic power conversion

The osmotic power generation of the MoS₂/BNC composite membrane was evaluated in a homemade flow cell module that contained a mixed solution (0.01 M/0.5 M NaCl) (Experimental section 4 and Fig. S15, ESI†). Fig. S16 (ESI†) provides an equivalent circuit diagram of the osmotic energy conversion system. Permeation current (I_os) and permeation potential (V_os), which were 1.1 V and 52.28 μA, respectively, can be easily derived from the I–V curve measured at the transmembrane concentration gradient based on the MoS₂/BNC composite membrane with 3 wt% BNC (Fig. 5a and Fig. S17, ESI†). The power density gradually increased with BNC content up to 3 wt% to achieve the maximum power density of 73 W m⁻² (Fig. 5b). The addition of BNC to the composite membranes enhanced their electronegativity, which in turn led to an improvement in ion selectivity and an increase in power density. At the same time, the enlarged interlayer space by intercalated BNC provides a more active cation transport pathway. However, with a further increase of BNC content (>3 wt%), the enhanced spatial potential resistance of the membrane led to a decline in power density. Another possible reason is the enlarged interlayer spacing, lowering the ion selectivity. Hence, the MoS₂/BNC composite membrane with 3 wt% BNC content (referred to as MoS₂/BNC₃) was employed in the subsequent testing due to its exceptional power generation, consistent with results of numerical modeling. The influence of composite membrane thickness on osmotic energy conversion is also considered. When the thickness of the MoS₂/BNC₃ membrane increased from 302 to 486 nm (Fig. S18, ESI†), the osmotic power density increased from 21 to 70 W m⁻² (Fig. 5c). This is attributed to the decreasing intrinsic defects and the reduction of ion concentration polarization of the resultant membrane due to high ion selectivity in confined channels. However, with a further increase of membrane thickness, the power density decreased. This is due to the resistance increment for ion transport, consistent with the results of the diffusion-osmotic transport model in a previous report.⁵⁴ At an external resistance of about 40 kΩ, the peak output power density reaches 85 W m⁻², corresponding to the internal resistance of the membrane with 3 wt% BNC (Fig. 5d and Fig. S19, ESI†).

Fig. 5 The osmotic energy conversion of the MoS₂/BNC composite membrane. (a) I–V curve of the MoS₂/BNC₃ composite membrane under a transmembrane salinity gradient (0.01 M/0.5 M NaCl) before (blue line) and after (red line) the subtraction of the contribution of the redox potential. (b) Influence of the weight content of BNC on the output power density. (c) Influence of membrane thickness on the output power density. (d) Current density and output power density of the MoS₂/BNC₃ composite membrane depend on the external load resistance in 0.01 M/0.5 M NaCl. (e) Output density measured under different concentration gradients. Note that the low concentration side was fixed at 0.01 M. (f) Dependence of the output power density on the pH value. (g) Power density of the MoS₂/BNC₃ composite membrane at different temperatures. (h) Long-term stability of the MoS₂/BNC₃ composite membrane-based device.

To explore the practical application of MoS₂/BNC composite membranes in osmotic energy conversion, we investigated the relationship between the osmotic energy generation of composite membranes and the electrolyte concentration gradient. The concentration of NaCl was fixed at 0.01 M on the low-concentration side. The obtained power density gradually increased with increasing salinity gradient from 5-fold to 100-fold, as shown in Fig. 5e. Nonetheless, the captured power density exhibited a declining tendency when the salinity gradient reached 500-fold. The driving force for ion migration increases with the increase of concentration gradient and therefore the osmotic conversion capacity of the membrane is enhanced. However, the extremely high concentration of the electrolyte solution caused the concentration polarization, which might reduce the effective concentration gradient and impede the capture of osmotic energy.^55,56 This is consistent with the modeling results in Fig. S20 (ESI†).

The osmotic energy conversion of composite membranes can also be improved by pH changes. The power densities of up to 94 W m⁻² were harvested at a pH of ∼8 (Fig. 5f). Under acidic conditions, the S atoms on the MoS₂ surface strongly interact with protons to form S–H bonds. The adsorbed protons can exchange with the cations enriched in the nano-channels of the MoS₂/BNC composite membrane. As a result, the interlayer charge is reduced or even disappears. Therefore, this effect prevents more Na⁺ permeation into the nanochannels of the membrane, lowering the power density of the osmotic energy. Under alkaline conditions without the formation of the S–H bond, more cations can be embedded in the interlayers in terms of surface charge control, reflected by the enlarged interlayer spacing of the composite membrane. This led to reduction in the ion selectivity of the membrane as a result of decreased power density.⁵⁷ In addition, we also investigated the impact of temperature on power generation from artificial seawater and river water with a 50-fold salinity gradient. As seen in Fig. 5g, with the increase of the test temperature from 283 to 343 K, the power density of the MoS₂/BNC membrane increased from 45 to 233 W m⁻². This increase is attributed to the change in surface charge and chemical properties of the resultant membrane. Furthermore, as the temperature rises, the viscosity of the electrolyte decreases, one of the favorable factors for the rapid transport of ions are governed by surface charge.

The operational stability of MoS₂ membranes is greatly enhanced by the addition of BNC. As shown in Fig. S21 (ESI†), the continuous energy conversion had a stable duration longer than 18 hours versus less than 4 hours for the pristine MoS₂ membrane, which highlights the high stability of the composite membrane due to selective Na⁺ ion transfer in the confined channels. When fresh electrolyte was maintained every day, the system could remain stable for more than 40 days (Fig. 5h). This further indicates that the resultant composite membrane can effectively promote continuous osmotic energy harvesting with stable concentration difference across the membrane. In other words, the obtained membrane would exhibit excellent stability in practical applications due to constant salinity in natural sea water and river water. The electrochemical device in this work delivered the power densities of 73 and 233 W m⁻² at ambient temperature and 343 K, respectively, under a 50-fold concentration gradient (Fig. 5g and b). Comparing with the state-of-the-art membranes listed in Table S1 (ESI†), the device fabricated using the developed composite membrane in this work improved the power generation efficiency by a factor from 2 to 70. Furthermore, its excellent stability demonstrates the promising application of the resultant membrane. It is noteworthy that a decrease of power density to 20 W m⁻² was observed for enlarging the testing membrane area to 3.14 mm² (Fig. S22, ESI†). With a further increase in the membrane area, the power density was reduced quickly. Such a rapid decay can be attributed to the joint effects of the increasing number of co-ions to hinder counter-ion diffusion, intensified concentration polarization and membrane defects as discussed previously.⁵⁸ Moreover, the electrolyte categories also have an essential effect on the osmotic energy conversion of the membrane. To study the impact of electrolytes on output power density, the common monovalent (Na⁺, K⁺, and Li⁺) and divalent (Ca²⁺, Mg²⁺) cations in seawater were chosen. As shown in Fig. S23 (ESI†), the monovalent cations exhibited high energy conversion because monovalent cations have a higher ionic conductivity resulting from a larger ionic diffusion coefficient and a smaller hydration diameter.⁵⁹ The above results indicate that the osmotic energy generator fabricated using MoS₂/BNC composite membranes has great potential for practical applications.

To further show the potential for practical applications, the power output of the device was tested using real seawater and river water. A high power density of 114.2 W m⁻² was achieved based on the salinity gradient between low-salinity river water from the Jixia Lake in the university campus and high-salinity seawater from the Yellow Sea near the Tsingtao city (Fig. S24, ESI†). The relative high energy output from sea water is possibly due to the mixed transport of other monovalent ions (like Li⁺ and K⁺) and divalent ions (Ca²⁺ and Mg²⁺). It is noteworthy that the osmotic power density remained stable for 6 hours and then gradually decreased to 90 W m⁻² at about 18 hours (Fig. S25, ESI†). The effects of the presence of multivalent ions and fouling resulting from organics or microbes in the real sea water are still under exploration.

Conclusions

In conclusion, we have successfully prepared a MoS₂/BNC composite membrane with high Na⁺ selectivity via a vacuum filtration method. The introduction of negatively charged BNC in the composite membrane causes the electrostatic rejection of the negatively charged anions like Cl⁻; on the contrary, it facilitates Na⁺ diffusion. The BNC incorporation also enhances the mechanical strength and structural stability of the composite membrane due to the formation of hydrogen bonds and S–H bonds. Therefore, the developed membrane can improve osmotic power generation efficiency due to the synergetic effect of the space charge provided by BNC and the surface charge carried by MoS₂. The developed membrane showed an osmotic power density of 73 W m⁻² in simulated sea water and river water under a 50-fold salinity gradient based on the tested sample of 0.78 mm² area, greatly exceeding the performance of all reported membranes. The power density of 114.2 W m⁻² was achieved under the salinity gradients between real high-salinity seawater and low-salinity river water, further showing the possibility for practical applications. Our membrane exhibits excellent long-term stability compared with state-of-the-art membranes. With the increase of temperature to 343 K, a power density of 233 W m⁻² for the MoS₂/BNC composite membrane can be obtained. Numerical modeling reveals preferred Na⁺ transport through the designed membrane via the nano-confinement. These results recommend the MoS₂/BNC composite membrane for osmotic energy conversion. Future studies will solve the challenge of scalable preparation of large area membranes for practical interest.

Author contributions

X. Wang wrote the initial draft. Z. Wang performed the experiments. Z. Xue provided useful help for the computational technique. Y. Fan, J. Yang, Q. Zhang, and Y. Jin provided useful help for collecting study data. X. Meng and N. Yang supervised the research activity planning. N. Yang also provided the funding. S. Liu helped in writing and reviewing this paper. All authors discussed the results, commented on the manuscript, and contributed to the writing of the paper.

Data availability

We confirm that the data that support the findings of this study are available from the authors upon request.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work is financially supported by the National Natural Science Foundation of China (21978157, 22179073 and 22178015) and Natural Science Foundation in Shandong Province (ZR2022QB147 and ZR2023MB093). The authors greatly acknowledge the comments from Professor Haihui Wang in Tsinghua University to improve the quality of this paper. The authors also acknowledge the Analytical Testing Center of Shandong University of Technology.

Notes and references

1. P. Liu, T. Zhou, L. Yang, C. Zhu, Y. Teng, X.-Y. Kong and L. Wen, Energy Environ. Sci., 2021, 14, 4400–4409.
2. M. Rastgar, K. Moradi, C. Burroughs, A. Hemmati, E. Hoek and M. Sadrzadeh, Chem. Rev., 2023, 123, 10156–10205.
3. Z. Zhang, L. P. Wen and L. Jiang, Nat. Rev. Mater., 2021, 6, 622–639.
4. J. Xiao, M. Cong, M. Li, X. Zhang, Y. Zhang, X. Zhao, W. Lu, Z. Guo, X. Liang and G. Qing, Adv. Funct. Mater., 2023, 34, 2307996.
5. Z. Man, J. Safaei, Z. Zhang, Y. Wang, D. Zhou, P. Li, X. Zhang, L. Jiang and G. Wang, J. Am. Chem. Soc., 2021, 143, 16206–16216.
6. W. W. Xin, L. Jiang and L. P. Wen, Acc. Chem. Res., 2021, 54, 4154–4165.
7. X. T. Xiao, Y. Mei, W. T. Deng, G. Q. Zou, H. S. Hou and X. B. Ji, Small Methods, 2023, 2201435.
8. Y. J. Zhang, J. Riexinger, X. Y. Yang, E. Mikhailova, Y. C. Jin, L. N. Zhou and H. Bayley, Nature, 2023, 620, 1001–1006.
9. J. Rao, Z. Lv, X. Yan, J. Pan, G. Chen, B. Lü and F. Peng, Adv. Funct. Mater., 2023, 34, 2309869.
10. X. Zhang, M. Li, F. Zhang, Q. Li, J. Xiao, Q. Lin and G. Qing, Small, 2023, 19, 2304603.
11. B. Zhou, J. Zou, Z. Lin, Z. Yuan, X. Qin, P. Chen and D. Ye, Chem. Eng. J., 2023, 457, 141167.
12. Y. D. Wu, L. Jiang and L. P. Wen, Mater. Today, 2023, 65, 189–206.
13. S. Li, J. Wang, Y. Lv, Z. Cui and L. Wang, Adv. Funct. Mater., 2023, 34, 2308176.
14. J. Wang, D. Wang, Z. Y. Song, N. Jiang, S. Z. Li, Y. F. Zhang, B. Huang, H. J. Zhou, Z. Cui and L. Wang, Energy Environ. Sci., 2023, 16, 3146–3157.
15. J. Feng, M. Graf, K. Liu, D. Ovchinnikov, D. Dumcenco, M. Heiranian, V. Nandigana, N. R. Aluru, A. Kis and A. Radenovic, Nature, 2016, 536, 197–200.
16. F. K. Wang, Z. Y. Wang, X. X. Meng, X. Y. Wang, Y. Y. Fan, Y. Jin, W. M. Zhang and N. T. Yang, J. Membr. Sci., 2023, 686, 121975.
17. Z. Gao, J. Zhang, M. Ahmad, B. Jiang, Z. Sun, S. Wang and Y. Jin, Carbohydr. Polym., 2022, 296, 119847.
18. C. Zhu, P. Liu, B. Niu, Y. Liu, W. Xin, W. Chen, X.-Y. Kong, Z. Zhang, L. Jiang and L. Wen, J. Am. Chem. Soc., 2021, 143, 1932–1940.
19. Z. W. Huang, M. A. Fang, B. Tu, J. L. Yang, Z. Yan, H. G. Alemayehu, Z. Y. Tang and L. S. Li, ACS Nano, 2022, 16, 17149–17156.
20. J. Wang, Z. Cui, S. Li, Z. Song, M. He, D. Huang, Y. Feng, Y. Liu, K. Zhou, X. Wang and L. Wang, Nat. Commun., 2024, 15, 608.
21. L. Chang and X. Xiao, Diamond Relat. Mater., 2023, 136, 109971.
22. C. Chen, D. Liu, L. He, S. Qin, J. Wang, J. M. Razal, N. A. Kotov and W. Lei, Joule, 2020, 4, 247–261.
23. M. Graf, M. Lihter, D. Unuchek, A. Sarathy, J.-P. Leburton, A. Kis and A. Radenovic, Joule, 2019, 3, 1549–1564.
24. M. Q. Ma, C. Zhang, C. Y. Zhu, S. Huang, J. Yang and Z. K. Xu, J. Membr. Sci., 2019, 59, 1117316.
25. Z. Y. Wang and B. X. Mi, Environ. Sci. Technol., 2017, 51, 8229–8244.
26. Y. J. Qian, J. Shang, D. Liu, G. L. Yang, X. G. Wang, C. Chen, L. Z. Kou and W. W. Lei, J. Am. Chem. Soc., 2021, 143, 5080–5090.
27. X. Wang, W. Ding, H. Li, H. Li, S. J. Zhu, X. G. Zhu, J. M. Dai, Z. G. Sheng, H. Wang, X. B. Zhu, Y. P. Sun and S. X. Dou, J. Mater. Chem. A, 2019, 7, 19152–19160.
28. Q. Ma, Y. Li, R. S. Wang, H. Q. Xu, Q. J. Du, P. C. Gao and F. Xia, Nat. Commun., 2021, 12, 1573.
29. J. Li, S. Y. Chen, Z. T. Wu, Z. L. Han, X. Y. Qu, M. T. Jin, Y. H. Jia, Z. Zhou and H. P. Wang, Nano Energy, 2023, 112, 108482.
30. H. R. Geng, X. Zhang, W. N. Xie, P. F. Zhao, G. Z. Wang, J. H. Liao and L. J. Dong, J. Colloid Interface Sci., 2022, 609, 33–42.
31. Y. Fei, M. Liang, T. Zhou, Y. Chen and H. W. Zou, Carbon, 2020, 167, 575–584.
32. K. Ludwicka, M. Kaczmarek and A. Białkowska, Polymers, 2020, 12, 2209.
33. M. Acerce, D. Voiry and M. Chhowalla, Nat. Nanotechnol., 2015, 10, 313–318.
34. S. Presolski and M. Pumera, Mater. Today, 2016, 19, 140–145.
35. A. D. Marinov, L. B. Priegue, A. R. Shah, T. S. Miller, C. A. Howard, G. Hinds, P. R. Shearing, P. L. Cullen and D. J. L. Brett, ACS Nano, 2023, 17, 5163–5186.
36. X. M. Geng, W. W. Sun, W. Wu, B. Chen, A. Al-Hilo, M. Benamara, H. L. Zhu, F. Watanabe, J. B. Cui and T. P. Chen, Nat. Commun., 2016, 7, 10672.
37. Q. Yuan, L. Z. Huang, P. L. Wang, T. Mai and M. G. Ma, J. Colloid Interface Sci., 2022, 624, 70–78.
38. E. Son, S. J. Lee, J. Seo, U. Kim, S. H. Kim, J. M. Baik, Y. K. Han and H. Park, ACS Nano, 2023, 17, 10817–10826.
39. G. H. Nam, Q. He, X. Wang, Y. Yu, J. Chen, K. Zhang, Z. Yang, D. Hu, Z. Lai, B. Li, Q. Xiong, Q. Zhang, L. Gu and H. Zhang, Adv. Mater., 2019, 31, 1807764.
40. J. Yang, M. K. Choi, Y. Sheng, J. Jung, K. Bustillo, T. Chen, S.-W. Lee, P. Ercius, J. H. Kim, J. H. Warner, E. M. Chan and H. Zheng, Nano Lett., 2019, 19, 1788–1795.
41. D. Sun, D. Huang, H. Wang, G.-L. Xu, X. Zhang, R. Zhang, Y. Tang, D. Abd Ei-Hady, W. Alshitari, A. Saad Al-Bogami, K. Amine and M. Shao, Nano Energy, 2019, 61, 361–369.
42. J. Wei, P. He, J. Wu, N. Chen, T. Xu, E. Shi, C. Pan, X. Zhao and Y. Zhang, Chem. Eng. J., 2022, 428, 131014.
43. G. Liu, Q. Zeng, X. Sui, S. Tian, X. Sun, Q. Wu, X. Li, Y. Zhang, K. Tao, E. Xie and Z. Zhang, Small, 2023, 19, 2301085.
44. Z. Bo, X. Cheng, H. Yang, X. Guo, J. Yan, K. Cen, Z. Han and L. Dai, Adv. Energy Mater., 2022, 12, 2103394.
45. J. Wu, J. Liu, J. Cui, S. Yao, M. Ihsan-Ul-Haq, N. Mubarak, E. Quattrocchi, F. Ciucci and J.-K. Kim, J. Mater. Chem. A, 2020, 8, 2114–2122.
46. Z. P. Liu, Z. C. Gao, Y. H. Liu, M. S. Xia, R. W. Wang and N. Li, ACS Appl. Mater. Interfaces, 2017, 9, 25291–25297.
47. S. Wang, Z. Wang, Y. Fan, X. Meng, F. Wang and N. Yang, J. Membr. Sci., 2023, 668, 121203.
48. D. Yu, L. Sun, Y. Zhang, Y. Song, C. Jia, Y. Wang, Y. Wang, M. J. Kipper, J. Tang and L. Huang, Chem. Eng. J., 2024, 480, 148009.
49. K. Nakagawa, T. Ueno, Z. Wang, T. Yoshioka, J. Kulhavy, K. Taniya, A. Matsuoka, E. Kamio, S. C. E. Tsang and H. Matsuyama, Chem. Eng. J., 2023, 474, 145671.
50. F. Zhang, J. Yu, Y. Si and B. Ding, Adv. Mater., 2023, 35, 2302511.
51. Z. Zhang, S. Yang, P. P. Zhang, J. Zhang, G. B. Chen and X. L. Feng, Nat. Commun., 2019, 10, 2920.
52. Y.-S. Su, A. R. Fauziah, C.-Y. Wong, T.-Y. Huang and L.-H. Yeh, Small Sci., 2023, 2300167.
53. Y. C. Liu, S. P. Zhang, R. Y. Song, H. O. Zeng and L. D. Wang, Nano Lett., 2023, 24, 26–34.
54. Z. Zhang, L. Wen and L. Jiang, Nat. Rev. Mater., 2021, 6, 622–639.
55. G. Laucirica, M. E. Toimil-Molares, C. Trautmann, W. Marmisollé and O. Azzaroni, Chem. Sci., 2021, 12, 12874–12910.
56. L. Cao, F. Xiao, Y. Feng, W. Zhu, W. Geng, J. Yang, X. Zhang, N. Li, W. Guo and L. Jiang, Adv. Funct. Mater., 2017, 27, 1604302.
57. C. Y. Hu, A. Achari, P. Rowe, H. Xiao, S. Suran, Z. Li, K. Huang, C. Chi, C. T. Cherian, V. Sreepal, P. D. Bentley, A. Pratt, N. Zhang, K. S. Novoselov, A. Michaelides and R. R. Nair, Nature, 2023, 616, 719–723.
58. Z. Ren, Q. Zhang, J. Yin, P. Jia, W. Lu, Q. Yao, M. Deng, Y. Gao and N. Liu, Adv. Funct. Mater., 2024, 2404410.
59. X. J. Lin, Y. J. Dong, S. M. Tao, X. Feng, X. J. Wang, T. Song, J. Liu, Z. H. Zhong, Y. C. Wang and H. S. Qi, Nano Energy, 2023, 107, 108156.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ee01381f

‡ These authors contributed equally to this work.

---