Desalination 422 (2017) 83–90

Contents lists available at ScienceDirect

Desalination
journal homepage: [Link]/locate/desal

On the design of graphene oxide nanosheets membranes for water MARK

desalination
Sina Safaei, Rouhollah Tavakoli⁎
Department of Materials Science and Engineering, Sharif University of Technology, Tehran, Iran

G RA P H I C A L AB S T R A C T

A R T I C L E I N F O A B S T R A C T

Keywords: According to current researches, graphene oxide (GO) membranes show promising desalination properties due to
Graphene oxide ease of synthesis, low production cost, and high efficiency. There are several experimental works to study ionic
Desalination sieving properties of GO membranes. However, it is difficult to characterize atomistic mechanism of water
Water permeation and ion rejection by experimental approaches. On the other hand, there exist a few reports in which
Membrane
the atomistic picture of water permeation across GO membranes is investigated by means of molecular dynamics
Molecular dynamics
(MD) simulation. In the present work, in addition to water desalination, the atomic scale mechanism of ion
rejection is studied using large scale MD simulation. For this purpose, surface color maps based on the potential
of mean force (PMF) are computed between GO nanosheets to indicate interaction between functional groups
and existing species in saline water. The radial distribution function (RDF) between water molecules and
functional groups are measured to study the disordering of water molecules between GO nanosheets.
Furthermore, the effect of different layers separation value and applied pressure are examined to explore the
optimal design of GO membranes. According to our simulation results, the oxygen atoms in hydroxyl and ep-
oxide groups play an important role in rejection of the Cl ions and attraction of the Na ions. The hydroxyl groups
have the most impact on disordering of the water molecules between GO membranes. In addition, our designed
GO membrane, showed a water permeability of one to three orders of magnitude higher than commercial reverse
osmosis membranes.

⁎
Corresponding author.
E-mail address: rtavakoli@[Link] (R. Tavakoli).

[Link]
Received 21 May 2017; Received in revised form 27 July 2017; Accepted 19 August 2017
Available online 31 August 2017
0011-9164/ © 2017 Elsevier B.V. All rights reserved.
S. Safaei, R. Tavakoli Desalination 422 (2017) 83–90

1. Introduction nanosheets orientation and distribution, also the water, flows in per-
pendicular direction to the membrane as it occurs in reality. The water
“By 2030 nearly half the global population could be facing water pressure is considered close to realistic pressures that is used in com-
scarcity, with demand outstripping supply by 40 percent”, said United mercial RO membranes. In addition to water molecules, we studied the
Nations Secretary General Ban Ki-Moon [1]. Over 97% of the water on interaction of salt ions with functional groups of GO nanosheets. To
the Earth is saline water, and two thirds of remaining 3%, is frozen [2]. explore the atomistic mechanism of desalination, we introduced a new
It seems that in the near future, desalination of seawater will be the way to study the interaction between salt ions and functional groups by
only way to provide fresh water. The methods based on distillation and using the surface color maps for indicating potential of mean forces,
reverse osmosis (RO) are common approaches to desalinate seawater corresponding to each species. The information from surface color maps
[3,4]. In the RO method, an applied pressure is used to overcome can help us to decide which functional groups give us better perfor-
natural osmotic pressure such that water passes through a semi- mance for satisfactory water flux and excellent ion rejection ability si-
permeable membrane leaving salt behind. In the distillation methods, multaneously. The radial distribution function (RDF) between water
seawater is evaporated and then condensed to produce freshwater. Both molecules and functional groups are measured to study disordering of
of these methods waste a lot of energy and are very expensive [5]. water molecules between GO nanosheets. In addition, we show that the
Recently, nanotube-based and graphene-based membranes have membrane antifouling property can be seen in limited timescales of
attracted many interests for their potential in water desalination due to molecular dynamics simulation. Finally, we investigate the effect of
high permeability and excellent ion rejection. In spite of these ad- layers separation value and applied water pressure on desalination
vantages, the synthesis and fabrication are remained as major chal- properties.
lenges to produce cost effective membranes [6–15].
Graphene oxide (GO) is a chemical derivative of graphene with
2. Material and methods
several functional groups such as epoxide and hydroxyl that is produced
from graphite by the Hummer's method [16]. Recently, it is synthesized
We performed Classical MD simulations using the large-scale atomic
and fabricated in the forms of papers and films at the industrial-scale.
molecular massively parallel simulator (LAMMPS) [36]. The visual
Functional groups and layers separation of GO membranes can easily
molecular dynamics (VMD) and the open visualization tool (OVITO) are
optimized during the synthesis process to achieve best performance for
used for analysis and visualization of results [37,38]. All simulations
desalination [17–19].
are carried out in NVT ensemble with a Nosé-Hoover thermostat and a
Dry GO membranes have a layers separation of ~ 5 Å which permits
damping constant of 10 femtoseconds. The velocity-verlet algorithm is
only water vapor molecules permeate through the membrane. When a
used to solve motion equations with a time step of 1 femtosecond. The
GO membrane is immersed in water, it is swelled so the layers se-
periodic boundary conditions (PBC) are considered in all three direc-
paration is increased to ~10 Å (excluding the thickness of carbon
tions.
atoms). The Na ions are the smallest in the saline water which has a
The all-atom optimized potential for liquid simulations (OPLS-AA)
hydrated diameter of ~9 Å. Therefore, after swelling of the membrane,
is used for GO and salt ions [39]. This potential contains many-body
small ions like Na+ can permeate easily that leads to reduction of ion
terms, including bond coefficients, angle coefficients, van der Waals
rejection property. Several methods such as physical confinement and
parameters, and electrostatic interactions. In addition, OPLS uses a
crosslinking of nanosheets are employed to prevent swelling of GO
geometric combining rule for the Lennard-Jones coefficients. The ex-
membranes [20–26].
tended simple point charge model (SPC/E) is applied for water mole-
In the GO membranes, water molecules permeate through the na-
cules, following previous studies on similar systems [26,28–34]. The
nochannels between oxidized regions (within pristine regions) provided
corresponding parameters of these force fields are given in the Table S1
by the hydrophobicity of functional groups. Particles that have a
to Table S4 (see supporting information).
smaller size than the GO nanochannels can permeate in the membrane
The shake algorithm is used to fix the bond distance between
with the speed of orders of magnitude greater than conventional
oxygen and hydrogen atoms in the water molecules due to high fre-
membranes [22,26–29].
quency vibrations of hydrogen atoms, so that we can use longer time
Since the nanochannels are created by pristine areas of GO na-
steps. The interaction between water and GO includes van der Waals
nosheets, some molecular dynamics (MD) studies, explored the beha-
and electrostatic terms. The van der Waals forces cut-off is 1.0 nm, and
vior of water molecules across graphene sheets [21,30]. Other MD re-
the long-range Coulomb interactions are computed by using the par-
searches, examined The velocity profiles for water flow across GO
ticle-particle particle-mesh (PPPM) algorithm.
sheets [28,31], density profile of water molecules between GO sheets
As it is illustrated in the Fig. 1 in our model of GO, hydroxyl and
[28,31–33], the effect of GO surface oxidization degree on water flow
epoxide groups are considered on both sides and at the edges of GO
and ion rejection [31], the breaking and forming of hydrogen bonds
nanosheets, following the Lerf–Klinowski model that is the most
during the passage of water through the membrane [33], potential of
common GO model [40]. The infrared spectroscopic data for a stack of
mean force profiles for water molecules [31,33], and Alignment of
large-area GO sheets, indicated negligible quantities of carboxylic acid
water molecules in graphene and GO nanochannels [31,34,35].
groups [41]. Therefore, we used just hydroxyl groups at the periphery
In the present work, we tried to bring molecular dynamics simula-
of nanosheets due to limited edges of nanoscale GO model.
tion of desalination systems one step closer to the realistic situation that
The structure of a GO single nanosheet contains 18 epoxide and 25
is occurred in practice. For this purpose, we designed a membrane in
hydroxyl groups, with dimensions of 1.8 × 3 nm2. The functional
agreement with experimental reports considering graphene oxide
groups are distributed on both sides of GO nanosheets. Furthermore,

Fig. 1. (a) Lerf–Klinowski GO model [40], (b) our designed model for GO nanosheets.

84
S. Safaei, R. Tavakoli Desalination 422 (2017) 83–90

Fig. 2. Designed membrane with 13 graphene oxide nanosheets. Blue arrows show the trajectory of water molecules during filtration across membrane. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)

the single nanosheet of GO contains 206 carbon atoms and 43 oxygen change.
atoms. Therefore, the ratio of C/O is about 4.8 which is in agreement In our simulations, saltwater is considered on the feed side of the
with the Lerf–Klinowski model. The size of simulation box are 17, 37 membrane which is consisted of 4800 water molecules and 52 Na+/Cl−
and 11 nm in the x, y and z directions, respectively. To avoid distortion pairs, corresponding to a salt concentration of 35.5 g/L. This con-
or movement of the membrane during simulation, carbon atoms in the centration is close to the normal salinity of seawater (~ 35 g/L).
edges of the nanosheets are fixed.
In all of our simulations, the membrane is designed with 13 GO
3. Results and discussion
nanosheets and two layers, inspiring from GO membranes structure
proposed in previous practical studies [19,30]. In this study, the effect
Fig. 4a is a color map showing potential of mean force (PMF) for a
of edges is not examined, therefore the distance between the nanosheets
Cl ion passing through two nanosheets of GO. The PMF is computed
edges is considered 2 nm which is two times bigger than the potential
based on steered molecular dynamics (SMD). For this purpose, the
cut-off. Fig. 2 shows the designed membrane in the present work.
harmonic potential U = K(x − x0)2/2 with K equal to 20 kcal/mol-
Simulations are carried out for multiple values of layers separation
angstrom2 is applied. The end of spring moves with velocity of
from 7 to 8.5 Å with increment of 0.5 Å. According to our work, layers
0.00005 Å/femtosecond that is enough for reversible pulling. To create
separation values lower than 7 Å and higher than 8.5 Å resulted in
each color map of PMF, 30 simulations are performed to cover all of the
extremely low water flux and unfavorable ion rejection, respectively.
GO nanosheets width. As shown in Fig. 4b, every particle follows 30
For each choice of layers separation, three simulations are performed
different trajectories during passage throughout nanosheets.
for different water pressures of 500, 1000, and 2000 atm. Water pres-
In Fig. 4a the functional groups existing between two nanosheets are
sure on the feed side of the membrane is enforced by applying specified
shown with color map simultaneously, to indicate how hydroxyl and
and uniform forces in the z-direction to the piston atoms, thus the water
epoxide groups affect salt ions (Cl ion in this figure). The functional
pressure is kept constant during the simulation.
groups at the bottom and top nanosheets are illustrated with black and
Fig. 3 shows the desalination system and its components, i.e. the
white color, respectively. The PMF shows the energetically favorable
membrane, the water molecules, the salt ions, and the piston. The layers
and unfavorable regions for specific species. As we expected due to
separation is 8.5 Å and water pressure is 2000 atm. In the Fig. 3a after
negative charge of oxygen atoms in epoxide and hydroxyl groups, the
0.1 ns, the water molecules entered into the membrane region and in
regions including these type of atoms are unfavorable areas, so the Cl
the Fig. 3b after 14 ns, more than half of the water molecules purified
ions are repelled by these atoms. In contrast, the charge of hydrogen
and we have 94% salt rejection. Salt ions are shown in green color and
atom in hydroxyl group is positive, so it can attract the Cl ions. In
water molecules are shown in red-white color. The red rectangle around
hydroxyl groups, the hydrogen atom can rotate around oxygen easily in
the system is the simulation box and the blue arrows on the wall, show
radius of about 1 Å. On the other hand, the difference between energy
the pressure on the piston and direction of its moving. The movement of
of the regions can make movement harder for species. For example, a Cl
wall is about 2 nm after 14 ns. After this time the wall position and
ion can easily trap in a favorable region but for continuing to its
concentration of salt in different sections of simulation box will not
movement, it needs to overcome the energy difference between its

Fig. 3. Desalination system (a) after 0.1 ns equilibration, (b) after 14 ns filtration with 94% salt rejection.

85
S. Safaei, R. Tavakoli Desalination 422 (2017) 83–90

Fig. 4. (a) Color map of PMF for a particle

passing through two nanosheets of GO, (b)
every particle (water, Na ion, and Cl ion)
follows 30 different trajectories to cover all
of the GO nanosheets width for calculating
PMF color maps.

current region and the next region. It is important to note that the different layers separation. Considering Fig. 5a, b, and c, as we dis-
existence of the repulsive atoms at both of top and bottom nanosheets cussed earlier, in each trajectory a Cl ion faces many barriers that either
in same area is necessary to make a region unfavorable. Therefore, reduce its mobility or prevent its movement. Furthermore, it may stuck
functional groups on just one nanosheet cannot block ions, even the within the valleys of energy and repelled by functional groups later.
layers separation value is small, e.g. about 7 Å. With attention to Fig. 5d, e, and f, the Na ions have smaller radius and
We computed the PMF for three different layers separation. As a can be repelled by just hydrogen atoms in hydroxyl groups. Therefore,
result, we extracted PMF surface color maps that show barriers and compared to the Cl ions, for the Na ions the corresponding barriers are
valleys of energy between the GO nanosheets. In Fig. 5 the PMF surface smaller and valleys are shallower, such that the Na ions which stuck in
color maps of Cl ions, Na ions, and water molecules are shown for the valleys, can move out with lower energy. Moreover, the Na ions can

Fig. 5. Surface color maps of PMF for, (a),

(b), (c) Cl ion passing through GO layers with
different separation values, (d), (e), (f) Na ion
passing through GO layers with different se-
paration values, (g), (h), (i) water molecule
passing through GO layers with different se-
paration.

86
S. Safaei, R. Tavakoli Desalination 422 (2017) 83–90

move easily in continuous valleys (little energy difference) without

need for overcoming any energy barriers. For example, a Na ion can
attracted by an epoxide group and move freely to next epoxide group
with the water flow. According to surface color maps Fig. 5g, h, and i,
flat surfaces of PMF indicate easy movement of water molecules be-
tween GO nanosheets without encountering significant barriers and/or
valleys. Since the water molecules have a neutral charge, it is expected
that functional groups would not affect them too much. The only effect
is the physical constraint due to their atomic radius.
Increasing layers separation makes functional groups (at top and
bottom nanosheets) farther away from each other. As we discussed
earlier, functional groups on just one nanosheet cannot block ions even
in lowest layers separation value. Therefore, increasing the layers se-
paration reduces the height of barriers and depth of valleys, which
would consequently facilitates the transportation of ions and water
molecules. In the case of the Na ions, it seems that the layers separation
has not a significant effect on depth of valleys. Furthermore, we found
that the minimum energy of the valleys are decreased slightly, but the
more important impact of increasing layers separation is the expanding
of favorable regions.
Fig. 6 illustrates the radial distribution function (RDF) for water and
functional groups. In this figure we can see the difference of RDFs for
oxygen and hydrogen atoms in water molecules between feed and
membrane region. Furthermore, the effect of functional groups on
water molecules in membrane region are examined.
Fig. 6a compares the oxygen-oxygen and oxygen-hydrogen RDFs for
water molecules in feed and membrane region. For the water in feed
region, the first peak for O-H and O-O are seen at 1.75 and 2.72 Å, Fig. 7. (a) Effect of layers separation and water pressure on the number of water mole-
respectively. The result is in agreement with reported bulk water RDFs cules in the membrane vs. time, (b) effect of layers separation and water pressure on the
in previous studies [35,42]. The RDFs for water in membrane show that number of filtered water molecules vs. time.
in compare to bulk water, the position of the peaks are not changed, but
the intensities are different due to limitation imposed by GO na-
epoxides just can create one coordination shell with low intensity. In
nosheets. In contrast to bulk water, in membrane region the first and
contrast to epoxide groups, the hydroxyl groups have a significant effect
second coordination shells (second shell at 3.22 Å) for O-H RDFs have
on disordering of water molecules between GO nanosheets. As it is seen
the same intensities which can be happened because of water molecules
in the Fig. 6c and d the hydrogen and oxygen atoms in hydroxyl groups,
disordering between GO nanosheets [31,34,35].
can create two or three coordination shells of water around themselves.
In Fig. 6b, c, and d the effect of epoxide and hydroxyl groups on
Since the hydrogen and oxygen in hydroxyl groups are bonded to-
water molecules are investigated. It is obvious that the epoxide groups
gether, position of some peaks in the Fig. 6c and d can be identical with
have the lowest impact on disordering between nanosheets. The

Fig. 6. Radial distribution function (RDF) for, (a) water-water in membrane and feed region, (b) water-epoxide, (c) water-oxygen of hydroxyl, (d) water-hydrogen of hydroxyl.

87
S. Safaei, R. Tavakoli Desalination 422 (2017) 83–90

little shift in distance. The position of the functional groups determine In the filtered region there are a few peaks which are corresponding
the size of nanochannels between GO nanosheets in membrane. In other to ions passed through membrane suddenly. However, the salt con-
words, modifying the number and distribution of functional groups centration is reduced as a result of water molecules passage through
specify the pristine area of GO nanosheets which enable us to control membrane into filtered region, before reaching to the next peak. This
the water flux in membrane [43]. scenario is repeated after the each peak again, resulting the observed
Fig. 7a shows the number of water molecules versus time in the trend in this figure. The salt concentration in the membrane region
membrane region. For each value of separation there is a limit for the fluctuates around the mean value of 17 g/L in the course of simulation.
number of water molecules that can exist within the membrane region. These fluctuations are enough to ensure that the membrane fouling
In the case of higher pressures, the membrane gets filled faster. The does not occur, even when the salt concentrations will be higher than
trend of number of water molecules in membrane region reaches to a conventional values, e.g. 90 g/L.
plateau at longer times (for instance t > 5 ns), so just the separation In Fig. 8b the layers separation and water pressure are equal to
value controls it in this region. Therefore, without regard to the water 8.5 Å and 2000 atm respectively. As it is clear the trends are the same as
pressure, the membrane region will be filled with water completely. Fig. 8a, except the number of peaks are increased in the filtered region
Fig. 7b indicates the number of filtered water molecules against as a result of larger layers separation.
time. This Figure shows effect of layers separation and feed pressure on Fig. 9a shows the flux of water (volume per unit of time per area)
rate of desalination. These graphs are plotted until the time that the half passing through the membrane as a function of applied pressure and
of the water molecules are desalinated (50% desalination). It is clear layers separation. In our simulations, we had to use higher pressures
that for the majority of samples, after about 5 ns the membrane is filled compared to the typical pressures that is required for desalination,
approximately and we reach to a time-independent flow, as a result of because of intrinsic time scale limitation of classic MD simulation. We
linear relation between filtered molecules and time. Therefore, it takes can solve this problem by computing permeability (volume per unit of
a few nanoseconds to reach the steady state condition [28,31–33]. As time per area per pressure) of membrane that is shown in Fig. 9b.
well, the feed pressure and layers separation can affect this time con- Another possible method is extrapolating the graphs in Fig. 9a to lower
siderably. pressures like 100 atm, consequently, we can approach to practical si-
Fig. 8 illustrates salt concentration in the three regions of system, tuation, due to existence of linear relation between flux and pressure.
i.e. feed, membrane and filtered against time. In Fig. 8a the layers se- Considering Fig. 9b, it is obvious that by increasing the layers separa-
paration of 8 Å and pressure of 2000 atm are considered. At start of all tion, the membrane permeability increases linearly (R2 = 0.99). As it is
simulations, the salt concentration is 35.5 g/L in the feed region which expected permeability of membrane is in agreement with former work
is identical to the seawater salinity. [31]. Furthermore, according to previous studies [44,45], the reported
The salt concentration in the feed region is gradually increased as permeability for RO membranes is about 0.2 to 6 L/h·bar·m2, so the GO
time proceeds, for instance it reaches to 90 g/L when the half of the membranes have water permeability of one to three orders of magni-
water is desalinated. tude higher that commercial RO membranes.
Fig. 9c illustrates the salt rejection for the membranes as a function
of layers separation and water pressures. The salt passage is calculated
as ((filtered salt ions number at time t) / (initial salt ions number in the
feed side)) and t is the time that half of the water molecules passed from
membrane. Therefore, the salt rejection percent is equal to
100 × (1 − salt passage). According to Fig. 9c, the salt rejection is
reduced by increasing the feed pressure or layers separation. It appears
that the complete ion rejection can be reached by reducing the pressure
to conventional RO pressure values. As we discussed earlier (Fig. 4),
increasing layers separation makes functional groups (at top and
bottom nanosheets) farther away from each other so the van der Waals
and electrostatic forces from functional groups are reduced which leads
to facilitate permeation of water molecules and salt ions. Therefore, as
we can see in Fig. 9, increasing the layers separation result in higher
water flux, membrane permeability, and lower ion rejection. Con-
sidering Fig. 5, the energy that the water molecules and salt ions need
for passing through the barriers and valleys between GO nanosheets, is
provided from feed pressure. The lower feed pressure, cause slower flux
but the salt ions cannot pass through membrane.

4. Conclusions

The water permeation in the GO membranes are examined in nu-

merous studies. However, the mechanism of ion rejection and ion
passage through nanosheets are not investigated in depth. In the pre-
sent work, the atomic scale mechanism of water desalination and ion
rejection is studied using surface color maps of PMF and RDF results.
The PMF results indicate that the spaces between the GO nanosheets are
divided into the favorable and unfavorable regions. The oxygen atoms
in hydroxyl and epoxide groups play an important role in rejection of
the Cl ions and attraction of the Na ions. We found that during desa-
lination, the Cl ions, need a significant external force for overcoming
Fig. 8. Salt concentration in the feed, membrane, and filtered regions against time for, (a) repulsive forces from functional groups. In contrast, the Na ions need
layers separation of 8 Å and pressure of 2000 atm, (b) layers separation of 8.5 Å and this external force for releasing from attractive forces of functional
pressure of 2000 atm.
groups. Furthermore, the Na ions can pass through the membrane

88
S. Safaei, R. Tavakoli Desalination 422 (2017) 83–90

Fig. 9. (a) Effect of different layers separation and water pressure on water flux, (b) effect of different layers separation on membrane permeability, (c) effect of different layers separation
and water pressure on salt rejection percentage.

easier than Cl ions, due to smaller size and fewer energy barriers on Conference on Land and Poverty, World Bank, 2015.
[2] E.P. Glenn, J. Jed Brown, J.W. O'Leary, Irrigating Crops With Seawater, Scientific
their path. The RDF between water molecules and functional groups are American-American Edition, 279 (1998), pp. 76–81.
measured to study the most important reason for disordering of water [3] M. Elimelech, W.A. Phillip, The future of seawater desalination: energy, technology,
molecules between GO nanosheets. The water molecules are ordered and the environment, Science 333 (2011) 712–717.
[4] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Marinas, A.M. Mayes,
within pristine areas of the GO nanosheets but around each hydroxyl Science and technology for water purification in the coming decades, Nature 452
group, a disordered structure in radius of two or three coordination (2008) 301–310.
shell of water are created. Furthermore, the effect of different layers [5] B. Van der Bruggen, C. Vandecasteele, Distillation vs. membrane filtration: over-
view of process evolutions in seawater desalination, Desalination 143 (2002)
separation value and applied pressure are examined to explore the 207–218.
optimal design of GO nanosheets. Our designed GO membrane, shows a [6] S.P. Surwade, S.N. Smirnov, I.V. Vlassiouk, R.R. Unocic, G.M. Veith, S. Dai,
water permeability of one to three orders of magnitude higher that S.M. Mahurin, Water desalination using nanoporous single-layer graphene, Nat.
Nanotechnol. 10 (2015) 459–464.
commercial RO membranes. In addition, in our simulation the anti-
[7] G. Liu, W. Jin, N. Xu, Graphene-based membranes, Chem. Soc. Rev. 44 (2015)
fouling property of the GO membranes can be seen clearly, even in 5016–5030.
higher salinities compare to seawater. Therefore, according to our re- [8] Y. Han, Z. Xu, C. Gao, Ultrathin graphene nanofiltration membrane for water
sults and previous studies, GO membranes exhibit much better desali- purification, Adv. Funct. Mater. 23 (2013) 3693–3700.
[9] P.S. Goh, A.F. Ismail, B.C. Ng, Carbon nanotubes for desalination: performance
nation performance in compare to commercial RO membranes such as evaluation and current hurdles, Desalination 308 (2013) 2–14.
high water permeability and excellent resistance to fouling. [10] R. Das, M.E. Ali, S.B.A. Hamid, S. Ramakrishna, Z.Z. Chowdhury, Carbon nanotube
membranes for water purification: a bright future in water desalination,
Desalination 336 (2014) 97–109.
Acknowledgments [11] C.H. Ahn, Y. Baek, C. Lee, S.O. Kim, S. Kim, S. Lee, S.-H. Kim, S.S. Bae, J. Park,
J. Yoon, Carbon nanotube-based membranes: fabrication and application to desa-
The authors wish to thank Dept. of material science and en- lination, J. Ind. Eng. Chem. 18 (2012) 1551–1559.
[12] A. Aghigh, V. Alizadeh, H.Y. Wong, M.S. Islam, N. Amin, M. Zaman, Recent ad-
gineering, Sharif University of Technology, for providing instrumental vances in utilization of graphene for filtration and desalination of water: a review,
facilities to carry out this work. Desalination 365 (2015) 389–397.
[13] L. Wang, R.S. Dumont, J.M. Dickson, Nonequilibrium molecular dynamics simula-
tion of water transport through carbon nanotube membranes at low pressure a, J.
Appendix A. Supplementary data Chem. Phys. 137 (2012) 044102.
[14] W.D. Nicholls, M.K. Borg, D.A. Lockerby, J.M. Reese, Water transport through (7, 7)
Non-bonded parameters for OPLS-AA and SPC/E force fields, bond carbon nanotubes of different lengths using molecular dynamics, Microfluid.
Nanofluid. 12 (2012) 257–264.
parameters for OPLS-AA and SPC/E force fields, angle parameters for
[15] J. Goldsmith, C.C. Martens, Molecular dynamics simulation of salt rejection in
OPLS-AA force field, dihedral parameters for OPLS-AA force field, two model surface-modified nanopores, J. Phys. Chem. Lett. 1 (2009) 528–535.
videos showing desalination and ion rejection of the membrane. [16] W.S. Hummers Jr., R.E. Offeman, Preparation of graphitic oxide, J. Am. Chem. Soc.
Supplementary data associated with this article can be found in the 80 (1958) (1339-1339).
[17] Z. Luo, Y. Lu, L.A. Somers, A.C. Johnson, High yield preparation of macroscopic
online version, at [Link] graphene oxide membranes, J. Am. Chem. Soc. 131 (2009) 898–899.
[18] N. Huang, H. Lim, C.H. Chia, M.A. Yarmo, M. Muhamad, Simple room-temperature
References preparation of high-yield large-area graphene oxide, Int. J. Nanomedicine 6 (2011)
3443–3448.
[19] M. Hu, B. Mi, Enabling graphene oxide nanosheets as water separation membranes,
[1] M. McDermott, C. Selebalo, S. Boydell, Towards the Valuation of Unregistered Land, Environ. Sci. Technol. 47 (2013) 3715–3723.
in: Linking Land Tenure and Use for Shared Prosperity: Annual World Bank [20] J. Wang, P. Zhang, B. Liang, Y. Liu, T. Xu, L. Wang, B. Cao, K. Pan, Graphene oxide

89
S. Safaei, R. Tavakoli Desalination 422 (2017) 83–90

as an effective barrier on a porous nanofibrous membrane for water treatment, ACS [33] J. Muscatello, F. Jaeger, O.K. Matar, E.A. Müller, Optimizing water transport
Appl. Mater. Interfaces 8 (2016) 6211–6218. through graphene-based membranes: insights from nonequilibrium molecular dy-
[21] R. Joshi, P. Carbone, F.-C. Wang, V.G. Kravets, Y. Su, I.V. Grigorieva, H. Wu, namics, ACS Appl. Mater. Interfaces 8 (2016) 12330–12336.
A.K. Geim, R.R. Nair, Precise and ultrafast molecular sieving through graphene [34] S. Zheng, Q. Tu, J.J. Urban, S. Li, B. Mi, Swelling of graphene oxide membranes in
oxide membranes, Science 343 (2014) 752–754. aqueous solution: characterization of interlayer spacing and insight into water
[22] B. Mi, Graphene oxide membranes for ionic and molecular sieving, Science 343 transport mechanisms, ACS Nano 11 (6) (2017) 6440–6450.
(2014) 740–742. [35] M. Zokaie, M. Foroutan, Comparative study on confinement effects of graphene and
[23] A. Nicolaï, B.G. Sumpter, V. Meunier, Tunable water desalination across graphene graphene oxide on structure and dynamics of water, RSC Adv. 5 (2015)
oxide framework membranes, Phys. Chem. Chem. Phys. 16 (2014) 8646–8654. 39330–39341.
[24] A. Nicolaï, P. Zhu, B.G. Sumpter, V. Meunier, Molecular dynamics simulations of [36] S. Plimpton, Fast parallel algorithms for short-range molecular dynamics, J.
graphene oxide frameworks, J. Chem. Theory Comput. 9 (2013) 4890–4900. Comput. Phys. 117 (1995) 1–19.
[25] J.W. Burress, S. Gadipelli, J. Ford, J.M. Simmons, W. Zhou, T. Yildirim, Graphene [37] A. Stukowski, Visualization and analysis of atomistic simulation data with
oxide framework materials: theoretical predictions and experimental results, OVITO–the open visualization tool, Model. Simul. Mater. Sci. Eng. 18 (2009)
Angew. Chem. Int. Ed. 49 (2010) 8902–8904. 015012.
[26] J. Abraham, K.S. Vasu, C.D. Williams, K. Gopinadhan, Y. Su, C.T. Cherian, J. Dix, [38] W. Humphrey, A. Dalke, K. Schulten, VMD: visual molecular dynamics, J. Mol.
E. Prestat, S.J. Haigh, I.V. Grigorieva, Tunable sieving of ions using graphene oxide Graph. 14 (1996) 33–38.
membranes, Nat. Nanotechnol. 12 (6) (2017) 546–550. [39] W.L. Jorgensen, D.S. Maxwell, J. Tirado-Rives, Development and testing of the
[27] M. Zhang, C. Hou, A. Halder, H. Wang, Q. Chi, Graphene papers: smart architecture OPLS all-atom force field on conformational energetics and properties of organic
and specific functionalization for biomimetics, electrocatalytic sensing and energy liquids, J. Am. Chem. Soc. 118 (1996) 11225–11236.
storage, Mater. Chem. Front. 1 (2017) 37–60. [40] H. He, J. Klinowski, M. Forster, A. Lerf, A new structural model for graphite oxide,
[28] N. Wei, X. Peng, Z. Xu, Understanding water permeation in graphene oxide mem- Chem. Phys. Lett. 287 (1998) 53–56.
branes, ACS Appl. Mater. Interfaces 6 (2014) 5877–5883. [41] D.R. Dreyer, S. Park, C.W. Bielawski, R.S. Ruoff, The chemistry of graphene oxide,
[29] N. Wei, C. Lv, Z. Xu, Wetting of graphene oxide: a molecular dynamics study, Chem. Soc. Rev. 39 (2010) 228–240.
Langmuir 30 (2014) 3572–3578. [42] E.K. Goharshadi, G. Akhlamadi, S.J. Mahdizadeh, Investigation of graphene oxide
[30] R. Nair, H. Wu, P. Jayaram, I. Grigorieva, A. Geim, Unimpeded permeation of water nanosheets dispersion in water based on solubility parameters: a molecular dy-
through helium-leak–tight graphene-based membranes, Science 335 (2012) namics simulation study, RSC Adv. 5 (2015) 106421–106430.
442–444. [43] B. Corry, Water and ion transport through functionalised carbon nanotubes: im-
[31] H. Dai, Z. Xu, X. Yang, Water permeation and ion rejection in layer-by-layer stacked plications for desalination technology, Energy Environ. Sci. 4 (2011) 751–759.
graphene oxide nanochannels: a molecular dynamics simulation, J. Phys. Chem. C [44] T. Humplik, J. Lee, S. O'hern, B. Fellman, M. Baig, S. Hassan, M. Atieh, F. Rahman,
120 (2016) 22585–22596. T. Laoui, R. Karnik, Nanostructured materials for water desalination,
[32] J.A. Willcox, H.J. Kim, Molecular dynamics study of water flow across multiple Nanotechnology 22 (2011) 292001.
layers of pristine, oxidized, and mixed regions of graphene oxide, ACS Nano 11 [45] D. Cohen-Tanugi, J.C. Grossman, Water desalination across nanoporous graphene,
(2017) 2187–2193. Nano Lett. 12 (2012) 3602–3608.

90