Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
We studied the permeability of graphene sheets with designed nanopores using the classical molecular
dynamics. To characterize the energy profile for transmission we calculated the potential of the mean
force. A high selectivity for methane + butane mixture with the hydrogen-passivated pore diameter of
0.32 nm was found where the volume exclusion mechanism governs the separation process. In the case
Received 23rd June 2014, of a slightly larger pore diameter of 0.64 nm the same alkane mixture separates completely
Accepted 10th November 2014 unexpectedly: a larger butane molecule permeates much faster than a small methane one. The blocking
DOI: 10.1039/c4cp02745k eect of the permeation path by a larger mixture component when it worked like a cork was also
observed. This is a promising perspective for using graphene to design intelligent membranes which can
www.rsc.org/pccp maintain a constant composition of mixtures in the permeable area.
This journal is the Owner Societies 2014 Phys. Chem. Chem. Phys.
View Article Online
Paper PCCP
Published on 11 November 2014. Downloaded by University of Guelph on 21/11/2014 15:38:39.
Fig. 1 Hydrogenated graphene pores of the computational systems used in this paper. The notation H1 to H6 corresponds to the approximate number
of phenyl groups removed from the graphene sheet. Pore areas and diameters are denoted in the figure.
force field. This force field is used by other authors studying The alkane molecule was inserted in vacuum upright above a
graphene separation properties18,19 or effects occurring on graphene in a distance of 3 nm, vertically aligned with the center of
carbon surfaces.20 As molecular structures and parameters of the pore. Next a series of initial configurations was generated, each
investigated alkanes were also taken from the OPLS-AA to verify corresponding to a dierent alkane center of the mass distance
its reliability we calculated the potential of mean force for (COM) from the COM of graphene sheet along a reaction coordi-
system containing one methane molecule and graphene sheet nate z with a step of 0.05 nm. These configurations were used for
without a pore. The obtained equilibrium distance and binding the umbrella sampling windows. Next separate umbrella simula-
energy for CH4 adsorption on graphene are 0.3568 nm and tions were made by using a harmonic potential (force constant
11.76 kJ mol1 respectively, and are in agreement with experi- k1 = 500 kJ mol1 nm2) and 5 ns period of time (5 106 steps).
mental and theoretical data.21 In some cases such a value of the force constant was too small
Rigid graphene membranes with various nanopore diameters to push a big molecule eq. butane across a small pore eq. H1.
were modeled, ranging from 0.32 to 0.76 nm (Fig. 1). Pore It resulted in some discontinuity of the potential of the mean force
diameters were obtained from the open pore area measurements (PMF) curves close to z = 0.
p
by using the well-known formula d 2 A=p. Pores were passi- We note that our model of molecular deformation is some-
vated with hydrogen atoms and centered on the 4.96 nm what biased since we only allow the rearrangement of CH
4.79 nm graphene sheet. Bonds between carbon and hydrogen bonds, which covers only a fraction of the total deformation.
atoms are flexible to allow possible deformations of the edge of the
nanopores. In the case it is not pointed out, the temperature was 3. Results
maintained at 300 K using velocity rescaling with a stochastic
term22 with a time constant of 0.1 ps. The constant volume of The permeation of methane, ethane, propane, butane, isobutane
the system was imposed; the simulation box sizes are 5.1 nm and their two-component mixtures was simulated using all
4.9 nm 16 nm. The designed pores used during calculations are pore diameters shown in Fig. 1. Fig. 2 shows the snapshots of
shown in Fig. 1. independent pulling simulation (CH4/H1, the rate 0.01 nm s1,
To figure out the changes of potentials connected to pore k1 = 50 000 kJ mol1 nm2). Dierent configurations of hydrogen
gating we calculated the potential of the mean force (PMF) of molecules which passivate a nanopore can be seen. The flexibility
alkane permeation across the designed nanopore. It is crucial of CH bonds allows for the membrane leaflet asymmetry
for the understanding of the selectivity and flux of the designed i.e. different configurations on one side than on the other.
porous graphene. The leaflet eect can be shown more clearly by monitoring
To perform the umbrella sampling calculations the porous the time dependence of the average angle of the CH bonds at
graphene sheet was horizontally located at the position z = 8 nm. the edge of the nanopore and the normal of the graphene sheet
Phys. Chem. Chem. Phys. This journal is the Owner Societies 2014
View Article Online
PCCP Paper
Fig. 3 Time evolution centered on the CH4 passing event across the H1
pore: j is the average angle between the CH bonds and the normal of
the graphene plane, RmetH is the distance between center of mass of CH4
and center of mass of hydrogen atoms passivating the graphene pore.
(see Fig. 3). At the time of methane permeation across the pore
H1, a large-amplitude angular jump is observed, and the
average angle between the graphene plane and CH bonds is
close to the right angle. This effect vanishes with the increasing
difference between the pore and molecule size and in the case
of methane disappears for the pore H5. In the case of small
pores and big molecules the assumption of rigid graphene may
lead to very unrealistic CCH bond angles. Thus the presented
results are rather qualitative.
Fig. 5 PMF profile of methane/H3 nanopore at T = 300 K.
The potential of mean force (PMF) was extracted by using
the Weighted Histogram Analysis Method (WHAM),24 included
in the GROMACS suite.1517 Fig. 4 shows a schematic profile suggests the influence of adsorption eects on the energy
determined during simulations. profile for transmission. Fig. 46 show how the energy profile
We observed the rapid change of PMF with a varying changes if the pore size gets smaller.
distance between an alkane molecule and a graphene sheet. At first, there is no transmission barrier at all (Fig. 4,
The potential well located at z E 0 nm corresponds to the methane at H4 i.e. big pore and small molecule). Simulta-
transition of a molecule across a nanopore and is connected neously the observed asymmetry of potential well is probably
with the strongest attractive interactions between the graphene connected with the leaflet eect of CH bonds at the edge of the
nanopore and the permeating molecule. The negative sign of nanopore. While considering methane transmission through
PMF value at z E 0 nm means that a system requires energy the H3 nanopore i.e. small molecule and medium nanopore
supply to separate the alkane from the graphene sheet and (Fig. 5) the energy barrier corresponding to methane permeation
This journal is the Owner Societies 2014 Phys. Chem. Chem. Phys.
View Article Online
Paper PCCP
Published on 11 November 2014. Downloaded by University of Guelph on 21/11/2014 15:38:39.
Fig. 6 PMF profile of methane/H1 at T = 300 K when the alkane got stuck Fig. 7 Symmetrized PMF profile of methane/H2 at T = 300 K calculated
in the configurations close to z = 0 nm. assuming k1 = 10 000 kJ mol1 nm2. Still the complete energy profile
cannot be derived (summit of the peak, configurations close to z = 0).
appears but it is still compensated through adsorption. As a
result, the double well is observed. Their asymmetry still testifies
the occurrence of leaflet effect. Finally, Fig. 6 shows the case of determine complete PMFs the energy barrier accompanying
very small pore and small molecule (methane and H1 nanopore). molecule transmission is too high (e.g. Fig. 6).
In this case the energy barrier is huge and the complete profile To estimate the importance of neglected pore flexibility
cannot be derived anymore from the umbrella sampling. Trans- (rigid graphene and CH bonds deformation allowed),
mission of methane is practically impossible with this pore size. we separately calculated the PMF curve for CH4/H2 assuming
Undoubtedly, except pore size the kind of atoms used to k1 = 10 000 kJ mol1 nm2 (Fig. 7). Such extremely high values
passivate the graphene pore as well as the length and flexibility of the force constant makes it possible to calculate the energy
of bonds between carbons and passivating groups have an profile almost in the whole range of methane configurations.
influence on the energy profile for transmission. Here we used It allows us to estimate the value of energetic barrier for CH4
hydrogen but probably in other cases such an eect will be transmission through the H2 nanopore, it is equal to approx.
greater. According to the leaflet eect, if the flexibility of 70 kJ mol1 (E17 kcal mol1). On the other hand, ab initio
passivating groups does not aect the energy profile of molecule barrier energies for the same system calculated by Hauser and
transmission, it has a symmetrical shape. Schwerdtfeger14 are, in the case of fully frozen geometry, equal
While the dierence between the pore diameter and alkane to 90.24 kJ mol1 (21.553 kcal mol1) and, allow full relaxation
size decreases, the simulation system comes to the point where, of nanopore, 34.69 kJ mol1 (8.283 kcal mol1). It follows that
for a given k1, it is not possible to push the molecule across the the pore relaxation reduces the barrier heights and our model
pore. It concerns the alkane position close to z = 0. Such a case of molecular deformation is visibly biased. It seems that a valid
can be recognized by analyzing the histograms of the configura- estimate of the energy can only be assumed in the cases where
tions within the umbrella sampling windows. When they do not the difference between the actual pore diameter and the
overlap, the alkane molecule gets stuck in a given umbrella molecular size is larger than about 0.2 nm. In such cases a
window and the potential force is too small to move the molecule relatively low value of the force constant k1 = 500 kJ mol1 nm2
(see Fig. 6 as an example). is sufficient to determine the complete energy profile for alkane
The performed calculations indicate that for k1 = transmission.
500 kJ mol1 nm2 the complete energy profile for trans- To study the permeability and selectivity of graphene sheets
mission can be derived from sampling in the following cases: with the designed nanopores we simulated the separation of
methane/H3H5 nanopores gaseous mixture composed of 500 + 500 alkane molecules.
ethane/H3H5 nanopores We used the same box size with two graphene sheets located
propane/H4H6 nanopores at z = 4 nm and z = 12 nm and an imposed constant volume of the
butane/H5H6 nanopores simulation box. The gaseous mixture was located below and above
In the other studied systems the dierences between the graphene (retentate area) and the permeate area is situated in
alkane molecular size and the pore diameter are too small to the range 4 nm o z o 12 nm (initially there is a vacuum).
Phys. Chem. Chem. Phys. This journal is the Owner Societies 2014
View Article Online
PCCP Paper
Published on 11 November 2014. Downloaded by University of Guelph on 21/11/2014 15:38:39.
Fig. 8 Separation of the mixture CH4 (yellow) and C2H6 (orange) across the H4 graphene sheet. The image is rotated to 90 degrees.
Fig. 10 Temperature influence on the permeation rate of methane and butane across the H4 nanopore.
This journal is the Owner Societies 2014 Phys. Chem. Chem. Phys.
View Article Online
Paper PCCP
4. Conclusion
The application of graphene as the membrane for gas separa-
tion is much more puzzling than it seems. The shape of PMFs
shows that the leaflet asymmetry also exists in the case of
one-atom-thin porous membrane passivated with hydrogen.
Fig. 11 The influence of temperature on the average time spent by the
While considering the separation of methane + butane by
permeating butane molecule in the center of the nanopore H4 during porous graphene we showed that for small pore diameters the
separation of the CH4 + C4H10 mixture. sieving mechanism works as expected. But in the case of larger
Phys. Chem. Chem. Phys. This journal is the Owner Societies 2014
View Article Online
PCCP Paper
pores butane transmission is preferred. Thus, besides the 10 M. Bieri, M. Treier, J. Cai, K. At-Mansour, P. Rueux,
simple volume exclusion mechanism, the separation by porous O. Groning, P. Groning, M. Kastler, R. Rieger, X. Feng,
graphene can be also controlled by blocking the access to the K. Mullen and R. Fasel, Porous Graphenes: Two-Dimensional
nanopore of a given component or the cork eect. Due to Polymer Synthesis With Atomic Precision, Chem. Commun.,
adsorption the permeating molecule can get stuck at the center 2009, 69196921.
of nanopore and cause a decrease of flux of the second mixture 11 D. Jiang, V. R. Cooper and S. Dai, Porous Graphene As The
component. Although such an eect is reported in other Ultimate Membrane For Gas Separation, Nano Lett., 2009, 9,
studies of molecules moving through pores and channels, it 40194024.
deserves special attention. Hence adsorption accompanying 12 S. Blankenburg, M. Bieri, R. Fasel, K. Muellen, C. A.
Published on 11 November 2014. Downloaded by University of Guelph on 21/11/2014 15:38:39.
separation by nanoporous graphene plays a very important role Pignedoli and D. Passerone, Porous Graphene As an Atmo-
which should not be underestimated. spheric Nanofilter, Small, 2010, 6, 22662271.
Our paper represents an eort to demonstrate the possibility 13 Y. Li, Z. Zhou, P. Shen and Z. Chen, Two-Dimensional
of using porous graphene as an intelligent membrane for Polyphenylene: Experimentally Available Porous Graphene
mixture separation. The accurate design of the pore diameter as a Hydrogen Purification Membrane, Chem. Commun.,
and the adequate choice of the method of pore functionaliza- 2010, 46, 36723674.
tion as well as temperature and pressure can create a great 14 A. W. Hauser and P. Schwerdtfeger, Methane-selective
perspective. It can be used as a membrane in fuel cells, gas nanoporous graphene membranes for gas purification,
sensors or medical equipment which automatically doses the Phys. Chem. Chem. Phys., 2012, 14, 1329213298.
amount of drug into the bloodstream. Of course, such outlooks 15 H. J. C. Berendsen, D. van der Spoel and R. van Drunen,
can be treated more or less realistically, and the obtained GROMACS: A Message-Passing Parallel Molecular Dyna-
promising results deserve further studies. mics Implementation, Comput. Phys. Commun., 1995, 91,
4356.
16 E. Lindahl, B. Hess and D. van der Spoel, GROMACS 3.0:
References
A Package for Molecular Simulation And Trajectory Analy-
1 H. W. Kroto, J. R. Heath, S. C. OBrien, R. F. Curl and sis, J. Mol. Model., 2001, 7, 306317.
R. E. Smalley, C60: Buckminsterfullerene, Nature, 1985, 318, 17 B. Hess, C. Kutzner, D. van der Spoel and E. Lindahl,
162163. GROMACS 4: Algorithms for Highly Ecient, Load-
2 S. Iijima, Helical microtubules of graphitic carbon, Nature, Balanced, and Scalable Molecular Simulation, J. Chem. Theory
1991, 354, 5658. Comput., 2008, 4, 435447.
3 R. P. Wesoowski and A. P. Terzyk, Pillared Graphene As a 18 H. Liu, S. Dai and D. Jiang, Insights into CO2/N2 Separation
Gas Separation Membrane, Phys. Chem. Chem. Phys., 2011, Through Nanoporous Graphene from Molecular Dynamics,
13, 1702717029. Nanoscale, 2013, 5, 99849987.
4 K. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, 19 P. Lazar, F. Karlicky, P. Jurecka, M. Kocman, E. Otyepkova,
S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Electric Field K. Safarova and M. Otyepka, Adsorption of Small Organic
Eect In Atomically Thin Carbon Films, Science, 2004, 306, Molecules on Graphene, J. Am. Chem. Soc., 2013, 135,
666669. 63726377.
5 A. H. C. Neto, F. Gwinea, N. M. R. Peres, K. S. Novoselov and 20 G. Gianese, V. Rosato, F. Cleri, M. Celino and P. Morales,
A. K. Geim, The Electronic Properties of Graphene, Atomic-Scale Modeling of the Interaction Between Short
Rev. Mod. Phys., 2009, 81, 109162. Polypeptides and Carbon Surfaces, J. Phys. Chem. B, 2009,
6 S. T. Oyama, D. Lee, P. Hacarlioglu and R. F. Paraf, Theory of 113, 1210512112.
Hydrogen Permeability in Nonporous Silica Membranes, 21 O. O. Adisa, B. J. Cox and J. M. Hill, Modeling the Surface
J. Membr. Sci., 2004, 244, 4553. Adsorption of Methane on Carbon Nanostructures, Carbon,
7 J. S. Bunch, S. S. Verbridge, J. S. Alden, A. M. van der Zande, 2011, 49, 32123218.
J. M. Parpia, H. G. Craighead and P. L. McEuen, Imperme- 22 G. Bussi, D. Donadio and M. Parrinello, Canonical Sampling
able Atomic Membranes From Graphene Sheets, Nano Lett., Through Velocity Rescaling, J. Chem. Phys., 2007, 126,
2008, 8, 24582462. 014101.
8 M. D. Fischbein and M. Drndic, Electron Beam Nanosculpting 23 W. Humphrey, A. Dalke and K. Schulten, VMD: Visual
of Suspended Graphene Sheets, Appl. Phys. Lett., 2008, 93, Molecular Dynamics, J. Mol. Graphics, 1996, 14, 3338.
113107113110. 24 J. S. Hub, B. L. de Groot and D. van der Spoel, G_wham-A
9 D. C. Bell, M. C. Lemme, L. A. Stern, J. R. Williams and Free Weighted Histogram Analysis Implementation Including
C. M. Marcus, Precision Cutting And Patterning of Graphene Robust Error and Autocorrelation Estimates, J. Chem. Theory
With Helium Ions, Nanotechnology, 2009, 20, 455301455305. Comput., 2010, 6, 37133720.
This journal is the Owner Societies 2014 Phys. Chem. Chem. Phys.