Tunable C2N Membrane for High Efficient Water Desalination

Abstract

Water scarcity represents one of the most serious global problems of our time and challenges the advancements in desalination techniques. Although water-filtering architectures based on graphene have greatly advanced the approach to high performance desalination membranes, the controlled-generation of nanopores with particular diameter is tricky and has stunted its wide applications. Here, through molecular dynamic simulations and first-principles calculations, we propose that the recently reported graphene-like carbon nitride (g-C₂ N) monolayer can serve as high efficient filters for water desalination. Taking the advantages of the intrisic nanoporous structure and excellent mechanical properties of g-C₂ N, high water transparency and strong salt filtering capability have been demonstrated in our simulations. More importantly, the "open" and "closed" states of the g-C₂ N filter can be precisely regulated by tensile strain. It is found that the water permeability of g-C₂ N is significantly higher than that reported for graphene filters by almost one order of magnitude. In the light of the abundant family of graphene-like carbon nitride monolayered materials, our results thus offer a promising approach to the design of high efficient filteration architectures.

Tunable CN Membrane for High

OPEN

YanmeiYang Li Zhou, Xiaoming Zhang & Mingwen Zhao

N

Because of the surging growth of population, rapid urban development and industrialization progress, the augmentation in available freshwater resources is dwindling in many water-stressed countries of the world1. Seawater represents a plentiful resource to compensate the stress of potable water supply through desalination. As one representative desalination methods, reverse osmosis (RO) technique has been commonly adopted2. In this strategy, a semi-permeable membrane is placed at the interface between seawater and pure water. Pressure applied at the seawater side forces water ow towards the pure water side, while ions are blocked. The pure water production rate of the RO method is typically low, which is ~0.1 Lcm2day1MPa1 for commercially used RO membranes3,4. The sustainability of water resource through desalination, therefore, highly depends on the development of new desalination membranes.

A promising approach is graphene lter which is rst proposed from computer simulations36 and eventually realized in recent experiments7,8. Thanks to the ultrathin thickness (only one atomic layer) of graphene, the energy consumption in seawater desalination is greatly reduced because water ux scales inversely with the lter thickness. This approach may also works for other two-dimensional (2D) materials, such as such as hexagonal boron nitride9, silicene10,11, phosphorene1215 and molybdenum disulde (MoS2)16. Recently, the MoS2-based lter was reported to have high performance given that nanopores are introduced to the MoS2 monolayer through bombardment or chemical treatment17,18. However, to be used as ltering membranes, these materials face the same limitation: they do not have inherent nanopores for water ow and thus post-treatment is essential. The shape and size of the nanopores generated during nanoengineering process are decisive factors for water transparency and ion selectivity. However, the successful experimental realization of proper nanopores still remains challenging at present, which becomes to be a major obstacle to the design of advanced lteration architectures based on these 2D materials.

It is noteworthy that there are various types of graphinic carbon nitrides with inherent nanopores in different shapes, which are potential filteration material candidates1924. In particular, the nanopores in the recently-synthesized graphene-like carbon nitride25 (referred to as g-C2N, as shown in Fig.1) has diameter of ~4.1 (determined by the N atoms and minus the diameter of a N, ~1.5 ); the size is very close to that of the water molecule (~4). The C-C and C-N covalent bond-based framework result in excellent mechanical properties of the g-C2N which is almost comparable to graphene. The stability of g-C2N monolayer at high temperature

School for Radiological and Interdisciplinary Sciences (RAD-X) and Collaborative Innovation Center of Radiation School of Physics

R

A

P

SCIENTIFIC REPORTS

1

www.nature.com/scientificreports/

Figure 1. (a) Illustration of the simulation model; (b) Top view of the nanoporous g-C2N layer; (c) Local structure of one nanopore in g-C2N.

has also been demonstrated experimentally25. Thus, the intrisic porous structure implies high prospect of using porous g-C2N monolayer as desalination lter without post-treatment to generate nanopores.

In this work, we report the seawater desalination performance of the g-C2N membrane through molecular dynamics (MD) simulations and rst-principles calculations. It is found that tensile strain can eectively modulate the permeation of water through the g-C2N lter continuously from closed to open states. The water permeability is about two orders of magnitude higher than the commercial RO membranes, while ion transmission is totally blocked. The high water transparency and vigorous salt ltering capability are attributed to the steric hindrance and electrostatic interactions on the translocations of water molecules and ions through nanoscaled connements. Our results thus highlight the signicance of a new 2D material family for high performance sea-water desalination lters.

As illustrated in Fig.1a, the MD simulation model is composed of a seawater region and a pure water region, separated by a g-C2N lter. A graphene sheet mimicing the piston is placed on the top of the seawater and force is loaded on it (at 20, 40, 60, 80 and 100MPa pressure equivalents) in the z direction to push water ow towards the pure water region. The g-C2N lter (Fig.1b) containing 30 nanopores has a dimension of roughly 4.16 4.33nm2 at the strain-free state. The structure of nanopore is illustrated in Fig.1c, for which the framework is composed of sp2 hybridized C and pore edges are terminated by N atoms. Strain on the g-C2N lter is modulated by increasing and xing the size of the cross sectional area.

All the simulations were performed with the GROMACS package26. The AMBER03 force eld27 was used in the simulations. The seawater region contains 50 Na+ and Cl ions, and 7000 TIP3P water molecules28, corresponding to a salt concentration of 27g L1, slightly lower than the salinity of seawater (~35g L1). A lower salinity was chosen for the consideration that water owing towards the pure water region will eectively increase the salinity in seawater side during the simulation. For the g-C2N lter, the atom types of CA and NB were assigned to C and N atoms. The RESP point charges were calculated using the Gaussian 09 code29 at the HF/631G* level, yielding a value of 0.24|e| and 0.48|e| for C and N atoms, respectively. SHAKE constraints30 were applied to all bonds involving hydrogen atoms. The long-range electrostatic interactions were treated with the Particle Mesh Ewald method31,32, and atypical distance cuto of 12 was adopted for the van der Waals (vdW) interactions. The non-bonded interaction pair list was updated every 10fs. The cross section in the x-y plane of the simulation box was xed to a certain value in order to mimic the strained lter. The box was coupled to a constant at 1.0 atm only

SCIENTIFIC REPORTS

2

www.nature.com/scientificreports/

Figure 2. (a) Variation of energy in response to tensile strain. (b,c) Phonon spectra of g-C2N under the biaxial tensile strains of 12% and 12.1%. The negative frequencies correspond to the imaginary frequency modes which are dynamically unstable.

along the z direction. Canonical sampling was performed through velocity rescaling method at constant temperature 300K33. A movement integration step of 1 fs was used in the simulations. Each system was rst equilibrated for 10ns, followed by 300ns productive simulation for the data collection.

First-principles calculations were further performed to check the structure stability using the Vienna ab initio simulation package (VASP)3437. The electron-electron interactions are treated using a generalized gradient approximation (GGA) in the form of Perdew-Burke-Ernzerhof (PBE) for the exchange-correlation functional38. The energy cuto of the plane waves was set to 520 eV with an energy precision of 108 eV. Vacuum space larger than 15 was used to avoid the interaction between adjacent images. The Monkhorst-Pack meshes of 99 1 were used in sampling the Brillouin zone for the g-C2N lattice. Tensile strain was applied by xing the lattice constants to dierent values. Atomic coordinates were optimized using the conjugate gradient (CG) scheme until the maximum force on each atom was less than 0.01 eV1. Phonon spectra were calculated using a supercell approach within the PHONON code39.

The structural stability of g-C2N membrane under tensile strain is an important issue. The strain energy (Es) of g-C2N under tensile strain () was rstly calculated using rst-principles calculations. As shown in Fig.2a, with the increase of tensile strain, Es increases monotonously as < 20%. However, it is found that the derivate of strain energy (dEs/d) reaches the maximum at tensile strain of around 13%, suggesting that the g-C2N lattice becomes soer beyond this point. This is futrher conrmed by the features of phonon spectra. When the tensile strain is smaller than 12%, the phonon spectra are free from imaginary frequency modes (Fig.2b), which indicates the stability of the lter at this critical point. When the tensile strain exceeds 12%, imaginary frequency branch appears as demonstrated in Fig.2c, which reveals the fact that the g-C2N lattice becomes unstable, as external disturbance may destroy the g-C2N framework due to the imaginary frequency mode. Both the energy evolution and phonon spectra conrm that the g-C2N has excellent mechanical stability which can bear tensile strain of up to 12%. The high structural stability of the g-C2N sheets fullls the requirement of lters working at high pressure.

We rst tested the water transparency of the g-C2N lter under the pressure of 100MPa. The cumulative numbers of water molecules transferred through the g-C2N lter are summarized in Fig.3a. For g-C2N lter at the equilibrium state (without tensile strain) and under weakly strained ( 1%) conditions, no event of water passage is observed during the entire simulations, indicating that the lter is at closed state. The transition point from closed to open appears at a strain level near 2%, for which 16 water molecules were found to pass through the lter during the 300 ns simulation. As the strain is further increased, the g-C2N lter becomes more transparent. For 6%, each curve in Fig.3abegins with a linear region and eventually reaches a saturation point (around 6500 water molecules) where the entire reservoir of water molecules on the seawater region is completely depleted. This indicates that tensile strain can eectively modulate the permeation of water through the g-C2N lter.

Notably, for all the simulations at dierent tensile strain, the water ow (the slope of curves in Fig.3a) is constant in time, conrming that well-converged statistics is obtained for all the simulations. The water ow with respect to external pressure for the 12%-strained lter is summarized in Fig.3b. It is seen that the water ow is proportional to the strength of applied pressure. This allows us to evaluate the water permeability of g-C2N lter by extrapolating the dynamic quantities derived here down to the operating condition that is more typical of RO plants (usually several MPa). We expressed the water permeability in liters of output per square centimeter of the lter per day and per unit of applied pressure. As shown in Fig.3c, it varies from zero (for the unstrained g-C2N) to 35.1 Lcm2day1MPa1 (for the extremely strained case). The high performance benets from the densely packed nanopores in g-C2N lter (with density of 1.61014cm2). For comparison, the highest

SCIENTIFIC REPORTS

3

www.nature.com/scientificreports/

Figure 3. (a) Number of water ltered by g-C2N membrane as a function of simulation time under piston pressure of 100MPa, (b) Water ow at various pressure through the 12%-stretched g-C2N, (c) Pore diameter and water permeability with respect to tensile strain and (d) Water permeability with respect to the charges of nitrogen of the g-C2N lter under tensile strain of 6% and 11%.

performance that has been achieved for graphene lter is around 6 Lcm2day1MPa1 which is only 1/6 of that of g-C2N lter reported here7. It is also noteworthy that the performance of commercial RO is only in the order of 0.1Lcm2day1MPa15,40. This solidly highlights the importance of the unique porous structure of g-C2N as a lter material. More interestingly, there exists a strain window (611%) where the water permeability scales almost linearly with strain strength. This property allows a precise control of the g-C2N lter which is quite crucial for the design of tunable devices for ltration and other applications.

As illustrated in Fig.3c, for the 6%-strained lter, the diamater of the permeable pores reaches 4.6, compared to a value of 4.1 for strain-free g-C2N. For even larger strain, the pore diameter increases almost linearly with respect to the strain strength. The tunable water permeation through the g-C2N lter is mainly attributed to the expanded nanopores upon stretching which eectively decrease the steric connement eect for water passage.

For graphene based lter, it has been evidenced that the edge morphology of the nanopores, especially the hydrophobicity, would signicantly regulate the water ow5. In view that stretching of chemical bonds may cause slight electron re-distribution between adjacent elements, we examined the eect from atomic charges of the g-C2N lter on the water ux. For N, we have considered values from 0.4 to 0.6 |e| in our simulations (the atomic charge of C was changed accordingly). The water permeabilities under tensile strain of 6% and 11% are summarized in Fig.3d. It is clear that the nanopores with less-charged edge atoms have larger water permeability. On the contrary, accumulation of electrons at the nanopores will decrease the water permeability. In general, the change of water permeability in response to the N charge is only 1.0Lcm2day1MPa1 per 0.1|e|, which is much lower than the strain eect. Therefore, it is safe to propose that the regulation of the water ux by strain is mainly determined by the changes of steric hindrance upon modulations of the nanopore size.

The desalination efficiency is determined by the trade-o between water permeability and salt ion selectivity. Beside high water permeability, a desalination lter should eectively hinder the passage of salt ions. For graphene, the nanopores generated during pre-treatment always have a broad range of diameter, resulting in lower salt rejection performance. For the g-C2N, it is exciting that no events of ion passage through the lter have been observed throughout all the simulations with serious of strain and pressure. It is worth noticing that during the simulation, the eective salinity in seawater region keeps increasing because the piston pressure depletes the water molecules in seawater side. This indicates that the salt rejection performance of g-C2N lter is robust, which is quite crucial for desalination performance improvement. This should be attributed to the considerable small size of the nanopores, which can efficiently block the ion transmission, corresponding to the maximal water desalination efficiency.

SCIENTIFIC REPORTS

4

www.nature.com/scientificreports/

Figure 4. Potential of mean force (PMF) of water, Na+ and Cl passing through the 12%-stretched g-C2N lter from seawater region to pure water region. The lter is placed at 0nm.

The physical origins of the high performance of g-C2N filter can be explained by the potential of mean force (PMF) analysis through umbrella sampling. The PMF curves for Na+,

Cl, and water to across the g-C2N lter were obtained by sampling the force experienced by the salt ions or water molecules when passing through the nanopores. Without losing the validity, we have only calculated the PMF curves for the 12%-strained lter. As shown in Fig.4, the PMF for water molecules is most shallow, with no energy barrier (peak) exceeding 3.32 kBT. Typically, an energy barrier of around 5 kBT is considered to be low enough for water permeation to happen4143. Hence water molecules can pass through the nanopores easily, leading to high water transparency. Na+ is blocked because of the high energy barrier of 12.27kBT near the nanopore center.

The probability (Kr) for certain solution component to overcome an energy barrier (Eb), Kr exp((Eb)/kBT), indicates that the water molecules have approximately 8103 times higher chance to pass through the nanopores compared to Na+ ions at room temperature. For Cl to approach the nanopores from the seawater side, an energy barrier of larger than 35kBT is predicted which is almost inaccessible for transmission to happen.

In experiments, there are several approaches available for applying tensile strain to the 2D monolayer. For instance, g-C2N membrane can be deposited on exible substrate like polymethyl methacrylate. Strain can be directly applied on the substrate to induce deformation of the substrate and accordingly the stretching of the g-C2N. Such approach has been adopted for the tuning of electronic structure of MoS2 monolayer by strain44.

Other than direct mechanical stretching, novel substrates which have large thermal expansion coefficient would also serve as the necessary support and introduce strain on the g-C2N lter45.

In summary, through molecular dynamics simulations we demonstrate that the recently reported g-C2N monolayer can serve as an efficient lter material for seawater desalination. Under tensile strain, the inherent nano-pores of g-C2N sheets can conduct water in a high transparency manner, while salt ions are completely rejected.

The high water transparency and vigorous salt ltering capability are attributed to the steric hindrance and electrostatic interactions on the translocations of water molecules and ions through nanoscaled connements. The highest water permeability is about two orders of magnitude higher than the commercially-used RO membranes and six times higher than that reported for graphene-based lters. More importantly, the open and closed states for water ow can be accurately regulated by applying tensile strain. The advantage of the g-C2N lter over graphene lter is the inherent porous framework with permeable pores. The easy regulation of the lter with tensile strain and the precise pressure responsive behavior make the g-C2N on the horizon to advance the development of desalination lter. Our results also support the design and proliferation of tunable devices for ltration and other applications working at nanoscale.

1. Elimelech, M. & Phillip, W. A. The future of seawater desalination: energy, technology, and the environment. Science 333, 712717 (2011).

2. Sidney, L. & Srinivasa, S. In Saline Water Conversion - II Vol. 38 Advances in Chemistry Ch. 9, 117132 (American Chemical Society, 1963).

3. Sint, K., Wang, B. & Krl, P. Selective ion passage through functionalized graphene nanopores. J. Am. Chem. Soc. 130, 1644816449 (2008).

4. Suk, M. E. & Aluru, N. Water transport through ultrathin graphene. J. Phys. Chem. Lett. 1, 15901594 (2010).5. Cohen-Tanugi, D. & Grossman, J. C. Water desalination across nanoporous graphene. Nano Lett. 12, 36023608 (2012).6. Sun, C. et al. Mechanisms of molecular permeation through nanoporous graphene membranes. Langmuir 30, 675682 (2014).7. Surwade, S. P. et al. Water desalination using nanoporous single-layer graphene. Nat. Nanotechnol. 10, 459464 (2015).8. OHern, S. C. et al. Selective ionic transport through tunable subnanometer pores in single-layer graphene membranes. Nano Lett. 14, 12341241 (2014).

9. Watanabe, K., Taniguchi, T. & Kanda, H. Direct-bandgap properties and evidence for ultraviolet lasing of hexagonal boron nitride single crystal. Nat. Mater. 3, 404409 (2004).

SCIENTIFIC REPORTS

5

www.nature.com/scientificreports/

10. Lalmi, B. et al. Epitaxial growth of a silicene sheet. Appl. Phys. Lett. 97, 223109 (2010).11. Vogt, P. et al. Silicene: compelling experimental evidence for graphenelike two-dimensional silicon. Phys. Rev. Lett. 108, 155501 (2012).

12. Liu, H. et al. Phosphorene: an unexplored 2D semiconductor with a high hole mobility. ACS Nano 8, 40334041 (2014).13. Reich, E. S. Phosphorene excites materials scientists. Nature 506, 1919 (2014).14. Li, W., Zhang, G. & Zhang, Y.-W. Electronic properties of edge-hydrogenated phosphorene nanoribbons: a rst-principles study. J. Phys. Chem. C 118, 2236822372 (2014).

15. Li, W., Yang, Y., Zhang, G. & Zhang, Y.-W. Ultrafast and directional diusion of lithium in phosphorene for high-performance lithium-ion battery. Nano Lett. 15, 16911697 (2015).

16. Coleman, J. N. et al. Two-dimensional nanosheets produced by liquid exfoliation of layered materials. Science 331, 568571 (2011).17. Li, W., Yang, Y., Weber, J. K., Zhang, G. & Zhou, R. Tunable, strain-controlled nanoporous MoS2 lter for water desalination. ACS Nano 10, 18291835 (2016).

18. Heiranian, M., Farimani, A. B. & Aluru, N. R. Water desalination with a single-layer MoS2 nanopore. Nat. Commun. 6 (2015).19. Wang, A. & Zhao, M. Intrinsic half-metallicity in fractal carbon nitride honeycomb lattices. Phys. Chem. Chem. Phys. 17,

2183721844 (2015).

20. Zhang, X. & Zhao, M. Prediction of quantum anomalous Hall eect on graphene nanomesh. RSC Adv. 5, 98759880 (2015).21. Zhang, X., Wang, A. & Zhao, M. Spin-gapless semiconducting graphitic carbon nitrides: A theoretical design from rst principles. Carbon 84, 18 (2015).

22. Li, H. et al. Tensile strain induced half-metallicity in graphene-like carbon nitride. Phys. Chem. Chem. Phys. 17, 60286035 (2015).23. Wang, A., Zhang, X. & Zhao, M. Topological insulator states in a honeycomb lattice of s-triazines. Nanoscale 6, 1115711162 (2014).24. Li, F., Qu, Y. & Zhao, M. Efficient helium separation of graphitic carbon nitride membrane. Carbon 95, 5157 (2015).25. Mahmood, J. et al. Nitrogenated holey two-dimensional structures. Nat. Commun. 6 (2015).26. David Van Der, S. et al. GROMACS: Fast, exible, and free. J. Comput. Chem. 26, 17011718 (2005).27. Duan, Y. et al. A point-charge force eld for molecular mechanics simulations of proteins based on condensed-phase quantum mechanical calculations. J. Comput. Chem. 24, 19992012 (2003).

28. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926935 (1983).

29. Frisch, M. et al. Gaussian 09, Revision A. 02, Gaussian. Inc., Wallingford, CT 200 (2009).30. Ryckaert, J. P., Ciccotti, G. & Berendsen, H. J. C. Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J. Comput. Phys. 23, 327341 (1977).

31. Essmann, U. et al. A smooth particle mesh Ewald method. J. Chem. Phys. 103, 85778593 (1995).32. Darden, T., Perera, L., Li, L. & Pedersen, L. New tricks for modelers from the crystallography toolkit: the particle mesh Ewald algorithm and its use in nucleic acid simulations. Structure 7, R55R60 (1999).

33. Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through velocity rescaling. J. Chem. Phys. 126, 014101 (2007).34. Kresse, G. & Hafner, J. Ab initio molecular dynamics for open-shell transition metals. Phys. Rev. B 48, 1311513118 (1993).35. Kresse, G. & Furthmuller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 1116911186 (1996).

36. Kresse, G. & Furthmuller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 6, 1550 (1996).

37. Kresse, G. & Joubert, D. From ultraso pseudopotentials to the projector augmented-wave method. Phys. Rev. B 59, 17581775 (1999).

38. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 38653868 (1996).39. Parlinski, K., Li, Z. Q. & Kawazoe, Y. First-principles determination of the so mode in cubic ZrO2. Phys. Rev. Lett. 78, 40634066 (1997).

40. Pendergast, M. M. & Hoek, E. M. V. A review of water treatment membrane nanotechnologies. Energ Environ. Sci. 4, 19461971 (2011).

41. de Groot, B. L. & Grubmller, H. The dynamics and energetics of water permeation and proton exclusion in aquaporins. Curr. Opin. Struct. Biol. 15, 176183 (2005).

42. Borgnia, M. J. & Agre, P. Reconstitution and functional comparison of puried GlpF and AqpZ, the glycerol and water channels from Escherichia coli. Proc. Natl. Acad. Sci. 98, 28882893 (2001).

43. Won, C. Y. & Aluru, N. Water permeation through a subnanometer boron nitride nanotube. J. Am. Chem. Soc. 129, 27482749 (2007).

44. He, K., Poole, C., Mak, K. F. & Shan, J. Experimental demonstration of continuous electronic structure tuning via strain in atomically thin MoS2. Nano Lett. 13, 29312936 (2013).

45. Plechinger, G. et al. Control of biaxial strain in single-layer molybdenite using local thermal expansion of the substrate. 2D Mater. 2, 015006 (2015).

This work was supported by the National Basic Research Program of China (No. 2012CB932302), the National Natural Science Foundation of China (grant nos 11304214, 21405108, and 21433006). A Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD) and Jiangsu Provincial Key Laboratory of Radiation Medicine and Protection.

Y.Y., W.L., H.Z. and X.Z. carried out the simulations and analysis; W.L. and M.Z. initiated the idea and checked the results. All the authors wrote and reviewed the manuscript text.

Competing nancial interests: The authors declare no competing nancial interests.

How to cite this article: Yang, Y. et al. Tunable C2N Membrane for High Efficient Water Desalination. Sci. Rep. 6, 29218; doi: 10.1038/srep29218 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS

6