Single-layer graphyne membranes for super-excellent brine separation in forward osmosis

Abstract

Forward osmosis (FO) technology has shown great promise in sea water desalinization and in power generation from the mixing of fresh water and seawater in estuaries. However, the desalination efficiency and the power density level of present FO systems are very low owing to the low water flux of commercial FO membranes. In this study, we report the brine separation performance of single-layer graphyne-N (N = 3, 4, 5, 6) membranes in a forward osmosis system by a molecular dynamics simulation. Our calculations show that graphyne-3 not only can achieve a high water flux of up to about 39.15 L cm⁻² h⁻¹, which is three to four orders of magnitude higher than that of conventional osmotic membranes, but also can achieve a perfect salt rejection rate. The results indicate that structural characteristics and charge properties of these membranes, as well as the distribution pattern of both water molecules and salt ions in each system, are the main factors that affect the brine separation performance of graphyne membranes. Besides, we found that the formation of hydrated salt ions is the basic reason for brine separation in salt solutions. Graphyne membranes have good prospects in brine separation, regeneration-free applications and high salinity applications in FO.

---

1 Introduction

Water crisis is spreading to more and more countries and regions, owing to population growth, changes in rainfall patterns, environmental pollution,^1–4 etc. A report of the United Nations predicted that two-thirds of the population would face water crisis by 2025,^5,6 which has stimulated many countries to target brine separation to reduce the increasingly severe water crisis. The methods of brine separation gradually developed from traditional heating, which demands for high energy consumption and high investment in equipment such as those involving multi-effect evaporation and multi-stage flash, to methods using membranes that are easy to operate and require low energy consumption.⁷

Currently, reverse osmosis (RO) occupies an important position in brine separation due to its relatively low energy consumption, flexible installation, lower susceptibility to corrosion and scaling, as well as being a low one-time investment. Nevertheless, there are still some problems with the RO process such as the short life of the membranes and a low water recovery rate,^8,9 which makes forward osmosis (FO) a new hot research area.

In FO system, two sides of a semipermeable membrane are respectively filled with feed solution and draw solution. Water molecules on the feed solution side spontaneously diffuse to the draw solution side depending on the osmotic pressure difference. Subsequently, the separation of the draw solution can obtain pure water.^9–11 FO needs no external pressure; thus, membrane fouling is less prevalent compared to that of RO. The water recovery rate of FO is more than 75%, while it is 35–50% for RO.¹¹ FO has a great value not only in the field of brine separation but also in other fields such as the conservation and reuse of osmosis energy.^10,12

Studies on FO membranes, one of the most important parts of FO systems,^13–16 are attracting significant attention. An ideal FO membrane must ensure a high water flux and a high salt rejection rate.¹³ The brine separation performance of FO membranes can be improved by grafting different types of hydrophilic groups onto the active layer or support layer to ameliorate the hydrophilicity of the membranes.^14–17 Some researchers add a pore-forming agent^17–20 into the film-forming solution to increase the porosity of the support layer and adjust the bending factor of the holes, whereas others try to reduce the thickness of the support layer by directly embedding polyester mesh into the polymer membranes.^20,21 However, these membranes still cannot satisfy the desired requirements for brine separation.

The concentration polarization phenomenon (CP)^22,23 is an important factor hindering the achievement of high separation performance of FO membranes. In particular, the inner concentration polarization (ICP) occurring in the support layer leads water flux to decrease by up to 80%.²³ Considering that water flux is inversely proportional to the thickness of the membrane, reducing the thickness of membranes can greatly increase the water flux. However, if the support layer is removed or the membrane thickness is reduced significantly, the mechanical strength of the membranes cannot meet practical requirements. Therefore, an entry point in solving the problem is to explore new porous membranes with both a sufficiently high mechanical strength and ultrathin thickness to maximally minimize the mass transfer resistance of water molecules and the influence of CP on the membrane performance.

Relevant studies show that carbon nanotubes (CNTs) and boron nitride nanotubes, which possess hollow structures, can achieve ultra-fast water transport performance and good salt rejection,^24–30 but it is difficult to form nanotube membranes with vertically arranged nanotubes. In some studies, CNTs were embedded into polymer membranes, but the CP impact has not been reduced or eliminated on account of the existence of the polymer membranes.^26–29 To minimize the CP phenomenon, an ultrathin porous membrane with high mechanical strength is undoubtedly an excellent choice.

In some reports, the mechanical properties and permeability of single-layer graphene or modified graphene porous membranes, such as graphene nitride, graphene fluoride, and graphene oxide, were studied using molecular simulations.^30–37 Molecular simulation results show that these porous membranes not only have high mechanical properties but can also achieve a high water flux and a high salt rejection rate in FO. With the rapid development of graphene materials, there have been many inspiring breakthroughs in the fabrication of single-layer graphene membranes^38–42 since they have been discovered in 2004 in the laboratory. More excitingly, a high-quality graphene film that predominantly consists of single-layer graphene⁴² has been successfully fabricated by roll-to-roll chemical vapor deposition (CVD) synthesis on a suspended copper foil under the conditions of 1000 °C and low pressure. The length can reach up to 100 m and the width can reach about 210 mm. Several researchers have experimentally shown the feasibility of using single-layer porous graphene in brine separation or desalination.^43–45 For example, in one of the studies,⁴³ water transport properties of various plasma-etched single-layer graphene membranes were investigated using four configurations, and the resulting membranes exhibited a salt rejection rate of nearly 100% and a fast water transport property. Nevertheless, a single-layer graphene membrane is non-porous and needs drilling, and it is difficult to precisely control the size and distribution of the pores in practice.^35–37

Graphyne, which is a carbon allotrope, has the same hexagonal symmetry as graphene. Graphyne-N (N = 1, 2, 3,…) are obtained by connecting N-acetylenic bonds among the adjacent six-membered rings of graphene,^46–56 and the effective van der Waals pore diameter of graphyne-N increases with the increase of N.^48,49 Studies by molecular simulation showed that a single-layer graphyne-N (N = 3, 4, 5, and 6) membrane has high mechanical strength, its Young's modulus reaches up to 365–700 GPa, the uniaxially stretched final strength reaches up to 25–100 GPa and the ultimate uniaxial tensile strain is over 6%, which are sufficient for them to withstand the stress and deformation in a clamping process or external pressure when used in RO. Graphyne-N (N = 3, 4, 5, and 6) membranes also hold excellent chemical and thermal stability.^48–56

There have been some inspiring reports that show the successful synthesis of flake-like graphyne structures.^57–59 Most strikingly, Li et al.⁶⁰ reported that graphdiyne films with an area of 3.61 cm² were successfully fabricated on the surface of copper via a cross-coupling reaction using hexaethynylbenzene. The synthesizing and fabrication of graphyne has attracted increasing attention due to its exceptional structure and excellent mechanical and electrical properties. Furthermore, compared with graphene membranes, the pores of graphyne membranes are naturally formed; thus, the complicated process of making pores can be avoided and the fabrication of single-layer graphyne membranes is more energy-saving. The natural regularity and the uniform distribution of those holes creates an excellent condition for graphyne-N (N = 3, 4, 5, and 6) membranes to achieve high separation performance. Because of rapid development in the synthesis and fabrication of large-area graphene membranes, thanks to the hard work of researchers, the successful fabrication of larger, high-quality single-layer graphyne membranes with lower energy consumption is promising. Considering that using single-layer porous graphene in brine separation or desalination is feasible, it is also promising to use single-layer graphyne membranes in these fields. Although scaling up of these membranes for use in practical application remains a significant challenge, it promises to be realized in the near future with the sustained efforts of researchers.

At present, only a few researchers have studied the performance of graphyne membranes in water treatment by molecular simulation.^46–50 Kou et al.⁴⁶ reported that the net water flux through the monolayer graphyne-3 membrane was considerable higher than that through the CNT membrane with similar nanopore diameter, and the remarkable hydraulic permeability was attributed to hydrogen bond formation. Their group, in another study,⁴⁹ also showed that the water permeability and salt rejection of graphyne are closely related to the type of graphyne membrane, the salt concentration of solution and hydrostatic pressure. Zhu et al.⁴⁷ showed that water permeability through graphyne exhibited nonlinear dependence on the pore size, and they attributed this counter-intuitive behavior to the quantized nature of water flow at the nanoscale. Lin et al.⁴⁸ showed that hydrophobic graphyne membranes exhibited higher rejection rates for hydrophilic contaminants compared to the hydrophobic ones and that the size exclusion effect resulted in different rejection rates. Xue et al.⁵⁰ reported that pristine graphyne could achieve complete rejection of nearly all ions from seawater and this resulted from the higher energy barriers for ions than for water. Nevertheless, all these studies are aimed at the separation performance of graphyne membranes in the RO process, and the effect of the charge properties of graphyne membranes on water permeability and on salt rejection have not been considered.

In this study, brine separation performance of single-layer graphyne-N (N = 3, 4, 5, and 6) membranes in FO have been studied by molecular dynamics (MD) simulations. The results show that single-layer graphyne membranes can not only achieve high water flux but also high salt rejection rates. We studied the reasons for their high brine separation performance by analyzing the laws of diffusion and distribution of water molecules and salt ions in these systems as well as the effect of the charge properties of graphyne-N membranes on water permeability and salt rejection.

2 Simulation details

In this study, MD simulations were used to obtain information about the brine separation performance of the single-layer graphyne-N (N = 3, 4, 5, and 6) membranes in FO systems. Pure water was chosen as the feed solution and sodium chloride solution with the mass percent concentration of 5% was adopted as the draw solution; the structure of graphyne-N (N = 3, 4, 5, and 6) membranes and the system for simulation are shown in Fig. 1.

Fig. 1 Four graphyne membranes used in this simulation: (A) graphyne-3, (B) graphyne-4, (C) graphyne-5, and (D) graphyne-6. (E) A snapshot of the simulation framework. The gray spheres are the carbon atoms of the graphyne membrane. Sodium chloride is shown in sphere representation and in CPK style with Na⁺ in purple and Cl⁻ in green. Water molecules are represented as spheres with oxygen in red and hydrogen in white.

The MD simulations were performed on unit cells containing graphyne-N (N = 3, 4, 5, and 6) membranes. The modeling details are described as follows: first, graphyne-N models were built as a single-layer, the corresponding cubic unit cell box was obtained and the length of c axis was set at 50 Å. The single-layer graphyne membrane was always located in the middle and was perpendicular to the c axis. Second, energy minimization was performed to optimize the cell structure. Finally, pure water and salt solutions with the mass percent concentration of 5% were respectively filled into the two spaces on both sides of each membrane.

The COMPASS force field was used and the charge of each atom was determined by the forcefield assigned methods, and Ewald and atom based methods were used to calculate the electrostatic and van der Waals effects, respectively. Subsequently, smart minimizer method was used to perform energy minimization for 5000 steps to obtain the lowest-energy conformation of every system.

MD simulations were run for 5 ns with a time step of 1 fs, and they were performed on the initial configurations, which had been subjected to energy minimization. To try to meet the actual FO operating conditions, the NPT ensemble was chosen and a thermostat was used to maintain the temperature of the system constant at 298 K. The nose mode was adopted to perform the fixed temperature calculations. This mode is more stable than the Andersen method for temperature control and it can be used to adjust the temperature for the system that is not balanced, while the Andersen method is more adaptable to the systems under thermal equilibrium. The Berendsen method was adopted to maintain the external pressure constant at 1 atm. The force field adopted for the molecular dynamics calculations was COMPASS; the charges were distributed using the forcefield assigned methods, and both the electrostatic and the van der Waals effects were calculated by the Ewald method.

3 Results and discussion

3.1 Water flux and salt rejection

Simulations were operated using the unit cell of single-layer graphyne-N (N = 3, 4, 5, and 6) membranes. Both the water flux and rejection rate were calculated using the final statistical results of 5 ns, and each system was simulated three times. Rejection rates were calculated using the following formula:where R is rejection coefficient, which is a measure of the ability of the membrane to inversely reject salt ions from the draw solution in our systems, c_jl is the permeate concentration and c_jo is the initial concentration of the draw solution. For a perfectly selective membrane, c_jl = 0 and R = 100%, whereas for a completely unselective membrane, c_jl = c_jo and R = 0. We calculated the average values of the three simulation results for the water flux and rejection rate of each system and the results are shown in Fig. 2; the fluctuation of the results displayed by the error bars are also showed in these water flux and rejection rate curves. We can observe that in the FO process, graphyne-3 has an average water flux of up to about 39.15 L cm⁻² h⁻¹, which is three to four orders of magnitude higher than that of the industrial RO or FO membranes,^{7–16,18–21} and the salt rejection rate is maintained at 100% throughout the whole simulation process. Graphyne-4, graphyne-5 and graphyne-6 membranes have fluxes that are on the same order of magnitude as graphyne-3, but they cannot reach the ideal salt rejection as the graphyne-3 membrane does. They can be used in brine separation of lower salt rejection requirements or wastewater treatment with larger pollutant molecules.

Fig. 2 Water flux (triangle), rejection of Na⁺ (square) and Cl⁻ (pentagram) vs. type of graphyne membrane, namely, graphyne-N (N = 3, 4, 5, and 6).

3.2 Water transport in the FO systems of single-layer graphyne membranes

According to Richard W. Baker,⁶¹ the solution-diffusion model or the pore-flow model can be used to describe the mechanism of permeation. In the solution-diffusion model, permeants dissolve in the membrane and then diffuse through the membrane due to a concentration gradient. Solution-diffusion model usually applies to polymeric membranes, because of the free-volume elements of polymers, namely, the inter-chain voids appear and disappear around the same timescale as the motions of the permeants traversing the membrane; permeants can dissolve into these inter-chain voids and are separated by their different solubilities and diffusion rates. In the pore-flow model, permeants transport through the tiny pores depending on the pressure-driven convective flow and they are separated by molecular filtration. The pore-flow model is usually applied to microporous membranes of which the pores do not fluctuate in position or volume on the timescale of permeant motion; these pores are relatively large and fixed, and are connected to one another.

Graphyne-N (N = 3, 4, 5, and 6) membranes are rigid and the pores of these membranes are relatively large and fixed, and the pore-flow model is more reasonable to be applied here to describe the water transport mechanism of graphyne-N membranes in FO. In these four systems, the temperature and the pressure as well as the initial osmotic pressure difference are the same. Three parameters can be used to characterize the structure complexity of the graphyne-N membranes according to the theory of the pore flow model, namely, membrane porosity, membrane tortuosity and the pore diameter. It is obvious that the porosity and pore diameter of graphyne-N membranes gradually increase with the increase of N, whereas the tortuosities of the graphyne-N membranes are same. Fig. 2 shows that the water fluxes of graphyne-N (N = 3, 4, 5, and 6) membranes are in the same order of magnitude. There is no linear correlation between the water flux and the structure parameters, including pore diameter and membrane porosity, for graphyne-N membrane systems in FO.

To further study the effect of the charge property of each graphyne-N (N = 3, 4, 5, and 6) membrane on the water transport, we conducted molecular dynamics simulations by changing the charges on all the carbon atoms of graphyne-N (N = 3, 4, 5, and 6) membranes to be zero with other settings consistent with that of the previous four graphyne-N membrane systems. The simulation results are shown in Fig. 3. We can observe from Fig. 3 that the water flux of graphyne-N membranes is higher than that of the corresponding uncharged graphyne-N membrane, especially for N = 3 and N = 6. It can be concluded that the charge properties of graphyne-N (N = 3, 4, 5, and 6) membranes are beneficial for the water transport in these FO systems.

Fig. 3 Water flux of graphyne membrane and uncharged graphyne membrane vs. membrane type, namely, graphyne-N (N = 3, 4, 5, and 6).

Fig. 4(a) and (b) show the mean square displacement (MSD) curves of water molecules with time varying in these four systems of graphyne-N (N = 3, 4, 5, and 6) membranes and uncharged graphyne-N (N = 3, 4, 5, and 6) membranes respectively. Diffusion coefficient is defined as one-sixth of the slope of the linear fitting for each MSD curve. The calculated diffusion coefficients are listed in Fig. 4(a) and (b). We can observe from Fig. 4(a) that the diffusion coefficient of the water molecules in graphyne-N membrane systems gradually decreases with the increase of N (N = 3, 4, 5, and 6). In these systems, as the temperature and external pressure, as well as the initial osmotic pressure difference, are same, the diffusion of water molecules is influenced by hydrogen bonding interactions and electrostatic forces as well as van der Waals forces caused by the surrounding water molecules, membranes and salt ions. Fig. 4(a) and (b) show that the charge properties of the graphyne-3 membrane and graphyne-4 membrane can promote the diffusion of water molecules in their FO systems, while the promoting effects of the charge properties of the graphyne-5 and graphyne-6 membranes for the diffusion of water molecules are not obvious due to their lower charge densities.

Fig. 4 The mean square distance (MSD) curves of water molecules and corresponding fitting lines by the least squares method vs. simulation time and calculated values of the diffusion coefficient in each system: (a) graphyne-N (N = 3, 4, 5, and 6); (b) uncharged graphyne-N (N = 3, 4, 5, and 6).

Fig. 5(1) shows the distribution density of water molecules on two sides of a graphyne-N membrane in each system. We can see that water molecules are densely packed in the vicinity of the graphyne-N (N = 3, 4, 5, and 6) membranes; the distance from each membrane to the densely packed region is about 3.5 to 5.3 Å and the density gradually increases with N. Water is a structured liquid composed of a hydrogen bonding network. Recent computer calculations using molecular dynamic computations employing SPC/E water have shown that the number of hydrogen bonds per water molecule at 298–300 K in water bulk is between 3.2 and 3.4.⁶² The four radial distribution function (RDF) curves in Fig. 5(2) show the density of hydrogen atoms surrounding every oxygen atom in these four systems. The calculation results show that the number of hydrogen bonds per water molecule in these four systems increases slowly with the increase of N; the values are 2.00, 2.09, 2.10 and 2.10 from N = 3 to N = 6, respectively. We infer that the water structures in these four systems have been disturbed compared with the pure water bulk and that the number of hydrogen bonds per water molecule has decreased by about 1.1 to 1.4.

Fig. 5 (1) The radial distribution function (RDF) curves of water molecules on both sides of the membrane in these four systems. The position where the value of the distance is 0 is the position where graphyne-N (N = 3, 4, 5, and 6) membranes are located, and the inset depicts a magnified image of the marked peak area. (2) The RDF curves that reflect the distribution of water molecules relative to other water molecules in these systems and the two peaks show the density of hydrogen atoms surrounding each water molecule, which acts as a benchmark.

From Fig. 5(1) and (2), we can infer that the water structure in each graphyne-N (N = 3, 4, 5, and 6) membrane system has been disturbed because the number of hydrogen bonds per water molecule is smaller than that of the pure water bulk. Several researchers have demonstrated that water molecules transport in a single file when they permeate through the pores of the graphyne membranes; the formation of two hydrogen bonds in the system is beneficial for the realization of that transmission mode.^37,47,48 The number of hydrogen bonds per water molecule move gradually away from the value of two from graphyne-3 membrane system to graphyne-6 membrane system, which leads to the quantized transmission of water molecules becoming more difficult and the slowing of the water transport with the increase of N. With the increase of N, although the pore diameter and porosity of graphyne-N membranes gradually increases, the quantized transmission of water molecules becomes more difficult. Moreover, the charge properties of graphyne-N (N = 3, 4, 5, and 6) membranes have different effects on the diffusion of water molecules. These comprehensive effects result in the water flux curve trend shown in Fig. 2.

3.3 Salt rejection by single-layer graphyne membranes in FO process

The van der Waals diameters of sodium cations (Na⁺) and chloride anions (Cl⁻) are 3.7 and 5.0 Å, respectively, the diameter of water molecule is 4.0 Å, and the effective van der Waals diameters of the pores in graphyne-N (N = 3, 4, 5, and 6) membranes are about 3.8, 5.4, 7.0 and 8.6 Å, respectively.⁴⁸ Considering the structure characteristics of the water molecule and the diameter of each matter, we can infer that a water molecule and Na⁺ are able to permeate through these four membranes, and Cl⁻ can permeate through graphyne-4, graphyne-5 and graphyne-6 membranes. However, Fig. 2 shows that graphyne-3 can simultaneously achieve perfect rejection for Na⁺ and Cl⁻, and graphyne-4, graphyne-5 as well as graphyne-6 can achieve a certain degree of rejection for Na⁺ and Cl⁻. We can speculate from the abovementioned phenomenon that there may be some other factors that hinder Na⁺ and Cl⁻ from permeating through the membrane pores.

Fig. 6 shows the RDF curves that reflect the distribution of water molecules around Na⁺ and Cl⁻; the element (hydrogen atom or oxygen atom) corresponding to each peak and the values of r (the distance from ion center to the center of a hydrogen atom or oxygen atom) have been indicated. The two blue-dotted lines indicate the locations of concentrated oxygen atoms within these two hydration shells around chloride anions. This shows that both Na⁺ and Cl⁻ are surrounded by two hydration shells, and the densities of the water molecules are different. We calculated the coordination numbers of Na⁺ and Cl⁻ using the following formula:

where N_ij(r) is coordination number, ρ_j is the density of j in system, r is the distance between i and j, g_ij(r) is the RDF of j with respect to i. The coordinate number in the first hydration shell of Na⁺ and Cl⁻ are about 5.2 and 6.2, respectively, and the corresponding snapshots are shown in Fig. 6. This figure also shows that the diameters of the hydrated sodium cations and hydrated chloride anions are larger than the diameters of the graphyne-N (N = 3, 4, 5, 6) membrane pores, indicating that only some water molecules in the hydration shells are stripped off and the hydrated ions can permeate through the membrane pores. The formation of these hydrated ions leads to the separation of salt and water. With the increase of N from 3 to 6, the pore diameter and porosity gradually increase, and the numbers of water molecules that should be stripped-off as hydrated ions gradually reduce and the barriers triggered by the size exclusion effect also gradually decrease; thus, both the rejection rate of Na⁺ and Cl⁻ gradually decrease with the increase of N.

Fig. 6 The RDF curves of water molecules outside sodium ions and chloride ions. There are two hydration shells outside the sodium and chloride ions. The calculated coordinate numbers for the first hydration shells of the hydrated sodium and chloride ions are 5.2 and 6.2, respectively.

Furthermore, we can infer from Fig. 7 that the rejection rate of both Na⁺ and Cl⁻ of the uncharged graphyne-N membrane are higher than or equal to that of the graphyne-N membrane when N = 3, 4, and 5, while the rejection rate of both Na⁺ and Cl⁻ of uncharged graphyne-N membrane are lower than that of the graphyne-N membrane when N = 6. When N = 3, 4, and 5, the rejection rate of Na⁺ is equal to that of the Cl⁻ for uncharged graphyne-N membrane, while the rejection rate of Na⁺ is lower than that of the Cl⁻ for the graphyne-N membrane. Thus, we can infer that the charge properties of graphyne-N membranes also have certain effects on the rejection rate of salts ions besides the size exclusion effect. These effects are relevant to electrostatic interactions between the carbon atoms of graphyne membranes and the salt ions or water molecules.

Fig. 7 Rejection rates of the graphyne membrane and the uncharged graphyne membrane vs. membrane type, namely graphyne-N (N = 3, 4, 5, and 6).

3.4 Prospect for using graphyne membranes in water treatment

Experimental results show that the magnitude of the water flux of laboratory and commercial permeable membranes in RO or FO process are generally 10⁻³ when the units are L cm⁻² h⁻¹.^16,20,21 Simulation results in this study show that the magnitude of the water flux of single-layer graphyne membranes in FO process are 10 when the same units are used. The simulation conditions are comparable to those in the experiments. This suggests that the difference in the water flux is about four orders of magnitude. As for laboratory and commercial permeable membranes, the total thickness of the dense selective layer and the support layer are tens to hundreds of microns and the thickness of the dense selective layer are tens to hundreds of nanometers.

When these polymeric membranes are used as selectively semipermeable membranes in FO processes, the concentration polarization phenomenon (both external concentration polarization (ECP) and internal concentration polarization (ICP)) will occur and the osmotic pressure difference across the membrane will be considerably lower than the bulk osmotic pressure difference, which will lead to a great reduction of the water flux. In contrast, the van der Waals thicknesses of the single-layer graphyne membranes are only one carbon atom in diameter, namely, 3.4 Å.⁶³ When single-layer graphyne membranes are used in FO, the ICP that can markedly reduce water flux can be neglected;³⁷ the adverse effect of the ECP, which plays a minor role in osmotic-driven membrane processes, can be minimized by increasing the flow velocity and turbulence at the membrane surface.⁶⁴ Because graphyne membranes possess excellent chemical stability, membrane life will be greatly improved, which will save the expenditure for replacing osmosis membranes. For desalination, even considering the optimization of the draw solution and the benefits of reduced fouling during regeneration in FO, the energy efficiency of RO is likely to be superior to FO on account of the draw-dilution step in FO. However, in regeneration-free applications or high salinity applications, where the osmotic pressures of feeds are excessively great for existing reverse osmosis technologies, using graphyne membranes in FO is greatly superior compared to its use in RO.⁶⁵

4 Conclusion

We have studied the brine separation performance of graphyne-N (N = 3, 4, 5, and 6) membranes in the FO process by a molecular dynamics simulation study. The results show that single-layer graphyne-3 membrane can simultaneously achieve a high water flux and a rejection rate of 100%, which are also accompanied by a very high utilization of osmosis energy. There is no linear correlation between the water flux and the structure parameters, including pore diameter and membrane porosity, for graphyne-N membrane systems in FO. The water structures in graphyne-N membrane systems are disturbed for the numbers of hydrogen bonds per water molecule, which are smaller than that of the pure water bulk. The formation of hydrated salt ions leads to brine separation. The charge properties of graphyne-N membranes have certain effects on water transport and salt rejection. Owing to their naturally formed pores, high mechanical properties and excellent chemical and thermal stability, single-layer graphyne-N membranes have good prospects in brine separation, wastewater treatment, conservation and reuse of osmosis energy or regeneration-free applications for FO.

Acknowledgements

This study was supported by the National Natural Science Foundation of China (51473097), (51003067), the Opening Project of the State Key Laboratory of Polymer Materials Engineering (Sichuan University) (sklpme2014-3-14), and the Fundamental Research Funds for the Central Universities (2012SCU04A03).

References

1. J. Eliasson, Nature, 2015, 517, 6.
2. M. Alamaro, Nature, 2014, 514, 7.
3. C. Packer, R. Hilborn and A. Mosser, et al., Science, 2005, 307, 390–393.
4. C. J. Vörösmarty, P. Green and J. Salisbury, et al., Science, 2000, 289, 284–288.
5. S. Kar, R. C. Bindal and P. K. Tewari, Nano Today, 2012, 7, 385–389.
6. M. Elimelech, J. Water Supply: Res. Technol.--AQUA, 2006, 55, 3–10.
7. M. Park, J. J. Lee and S. Lee, et al., J. Membr. Sci., 2011, 375, 241–248.
8. J. R. McCutcheon, R. L. McGinnis and M. Elimelech, J. Membr. Sci., 2006, 278, 114–123.
9. J. R. McCutcheon and M. Elimelech, J. Membr. Sci., 2006, 284, 237–247.
10. S. Zhang, K. Y. Wang and T. S. Chunga, et al., J. Membr. Sci., 2010, 360, 522–535.
11. J. R. McCutcheon, R. L. McGinnis and M. Elimelech, Desalination, 2005, 174, 1–11.
12. T. Y. Cath, A. E. Childress and M. Elimelech, J. Membr. Sci., 2006, 281, 70–87.
13. L. Shi, S. R. Chou and R. Wang, et al., J. Membr. Sci., 2011, 382, 116–123.
14. J. R. McCutcheon and M. Elimelech, J. Membr. Sci., 2008, 318, 458–466.
15. G. Han, T. S. Chung and M. Toriida, et al., J. Membr. Sci., 2012, 423–424, 543–555.
16. L. Huang, N. N. Bui and M. T. Meyering, et al., J. Membr. Sci., 2013, 437, 141–149.
17. K. Y. Wang, T. S. Chung and R. Rajagopalan, Ind. Eng. Chem. Res., 2007, 46, 1572–1577.
18. K. Y. Wanga, T. S. Chung and J. J. Qin, J. Membr. Sci., 2007, 300, 6–12.
19. Q. Yang, K. Y. Wang and T. S. Chung, Environ. Sci. Technol., 2009, 43, 2800–2805.
20. C. Q. Qiu, L. Setiawan and R. Wang, et al., Desalination, 2012, 287, 266–270.
21. K. Y. Wang, R. C. Ong and T. S. Chung, Ind. Eng. Chem. Res., 2010, 49, 4824–4831.
22. J. R. McCutcheon and M. Elimelech, AIChE J., 2007, 53, 1736–1744.
23. G. T. Gray, J. R. McCutcheon and M. Elimelech, Desalination, 2006, 197, 1–8.
24. S. Joseph and N. R. Aluru, Nano Lett., 2008, 8, 2.
25. Y. X. Jia, H. L. Li and M. Wang, et al., Sep. Purif. Technol., 2010, 75, 55–60.
26. B. Corry, J. Phys. Chem. B, 2008, 112, 1427–1434.
27. S. Kar, R. C. Bindal and P. K. Tewari, Nano Today, 2012, 7, 385–389.
28. R. Das, M. E. Ali and S. B. A. Hamid, et al., Desalination, 2014, 336, 97–109.
29. M. S. Mauter and M. Elimelech, Environ. Sci. Technol., 2008, 42, 16.
30. M. E. Suk, N. R. Aluru and J. Phys, Chem. Lett., 2010, 1, 1590–1594.
31. K. Sint, B. Wang and P. Kral, J. Am. Chem. Soc., 2008, 130, 16448–16449.
32. P. Z. Sun, M. Zhu and K. L. Wang, et al., ACS Nano, 2013, 7(1), 428–437.
33. K. Celebi, J. Buchheim and R. M. Wyss, et al., Science, 2014, 344, 18.
34. S. C. O. Hern, M. S. H. Boutilier and J. C. Idrobo, et al., Nano Lett., 2014, 14, 1234–1241.
35. C. T. David and J. C. Grossman, Nano Lett., 2012, 12, 3602–3608.
36. J. G. Gai, X. L. Gong and W. W. Wang, et al., J. Mater. Chem. A, 2014, 2, 4023–4028.
37. J. G. Gai and X. L. Gong, J. Mater. Chem. A, 2014, 2, 425–429.
38. S. Bae, H. Kim and Y. Lee, et al., Nat. Nanotechnol., 2010, 5, 574.
39. X. Li, W. Cai and J. An, et al., Science, 2009, 324, 1312.
40. T. Hesjedal, Appl. Phys. Lett., 2011, 98(13), 3106.
41. T. Yamada, M. Ishihara and J. Kim, et al., Carbon, 2012, 50(3), 2615.
42. T. Kobayashi, M. Bando and N. Kimura, et al., Appl. Phys. Lett., 2013, 102, 023112.
43. S. P. Surwade, S. N. Smirnov, I. V. Vlassiouk, et al., Nat. Nanotechnol., 2015, 10, 459–464.
44. M. S. H. Boutilier, C. Z. Sun and S. C. O'Hern, et al., ACS Nano, 2014, 8, 841.
45. F. Perreault, A. F. de Faria and M. Elimelech, Chem. Soc. Rev., 2015, 44, 5861–5896.
46. J. L. Kou, X. Y. Zhou and Y. Y. Chen, et al., J. Chem. Phys., 2013, 139, 064705.
47. C. Q. Zhu, H. Li and X. C. Zeng, et al., Sci. Rep., 2013, 11, 1–7.
48. S. C. Lin and M. J. Buehler, Nanoscale, 2013, 5, 11801.
49. J. L. Kou, X. Y. Zhou and H. J. Lu, et al., Nanoscale, 2014, 6, 1865.
50. M. M. Xue, H. Qiu and W. L. Guo, Nanotechnology, 2013, 50, 5720.
51. Y. L. Yang and X. M. Xu, Comput. Mater. Sci., 2012, 61, 83–88.
52. S. W. Cranford, D. B. Brommer and M. J. Buehler, Nanoscale, 2012, 4, 7797.
53. Q. Yue, S. L. Chang and J. Kang, et al., J. Phys. Chem. C, 2013, 117, 14804–14811.
54. S. W. Cranford and M. J. Buehler, Carbon, 2011, 49, 4111–4121.
55. A. L. Ivanovskii, Prog. Solid State Chem., 2013, 41, 1–19.
56. Y. J. Li, L. Xu and H. B. Liu, et al., Chem. Soc. Rev., 2014, 43, 2572–2586.
57. Y. Tobe, K. Kubota and K. Naemura, J. Org. Chem., 1997, 62, 3430–3431.
58. H. Nishide, M. Takahashi and J. Takashima, et al., J. Org. Chem., 1999, 64, 7375–7380.
59. Y. Tobe, I. Ohki and M. Sonoda, et al., J. Am. Chem. Soc., 2003, 125, 5614–5615.
60. G. Li, Y. Li and H. Liu, et al., Chem. Commun., 2010, 46, 3256.
61. M. G. Hill, Membrane Technology And Applications, ed. R. W. Baker, 2nd edn, 2004, pp. 1–534.
62. M. Yizhak, Chem. Rev., 2009, 109, 1346–1370.
63. J. M. Ducéré, C. Lepetit and R. Chauvin, J. Phys. Chem. C, 2013, 117, 21671–21681.
64. T. Y. Cath, A. E. Childress and M. Elimelech, J. Membr. Sci., 2006, 281, 70–87.
65. R. K. McGovernn and J. H. Lienhard V, J. Membr. Sci., 2014, 469, 245–250.

---