pubs.acs.

org/Langmuir
2010 American Chemical Society

Metal-Organic Framework MIL-101 for Adsorption and Effect of Terminal

Water Molecules: From Quantum Mechanics to Molecular Simulation
Y. F. Chen, R. Babarao, S. I. Sandler, and J. W. Jiang*,

Department of Chemical and Biomolecular Engineering, National University of Singapore, 117576, Singapore,
and Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716

Received November 29, 2009. Revised Manuscript Received January 6, 2010

MIL-101 is a chromium terephthalate-based mesoscopic metal-organic framework and one of the most porous
materials reported to date. In this study, we investigate the adsorption of CO2 and CH4 in dehydrated and hydrated
MIL-101 and the effect of terminal water molecules on adsorption. The atomistic structures of MIL-101 are constructed
from experimental crystallographic data, energy minimization, and quantum mechanical optimization. The adsorption
isotherm of CO2 predicted from molecular simulation agrees well with experiment and is relatively insensitive to the
method (Merz-Kollman or Mulliken) used to estimate the framework charges. Both the united-atom and five-site
models of CH4 predict the isotherm fairly well, though the former overestimates and the latter underestimates.
Adsorption first occurs in the microporous supertetrahedra at low pressures and then in the mesoscopic cages with
increasing pressure. In the dehydrated MIL-101, more adsorbate molecules are located near the exposed Cr2 sites than
the fluorine saturated Cr1 sites. The terminal water molecules in the hydrated MIL-101 act as additional interaction sites
and enhance adsorption at low pressures. This enhancement is more pronounced for CO2 than for CH4, because CO2 is
quadrapolar and interacts more strongly with the terminal water molecules. At high pressures, however, the reverse is
observed, as the presence of terminal water molecules reduces free volume and adsorption. For the adsorption of CO2/
CH4 mixture, a higher selectivity is found in the hydrated MIL-101.

1. Introduction quantitatively elucidating the underlying physics of adsorption

Metal-organic frameworks (MOFs) are a new family of hybrid
porous materials. By judiciously tuning the organic linkers and
metal oxides, MOFs can be synthesized with a wide variety of pore
sizes, volumes, and functionalities.1 Therefore, MOFs are regarded
as ideal candidates for gas storage, separation, and catalysis. In the
past few years, numerous experimental studies have been reported
for gas adsorption in MOFs. For example, isoreticular MOFs with
different organic groups (-Br, -NH2, -OC3H7, -OC5H11, -C2H4,
and -C4H4) were examined for CH4 storage.2 CO2 adsorption was
determined in a series of MOFs (IRMOF-1, IRMOF-3, IRMOF-6,
IRMOF-11, MOF-2, MOF-74, MOF-177, MOF-505, and Cu3-
(BTC)2) and MOF-177 exhibits a capacity of 33 mmol/g.3 Also,
the extent of H2 storage was measured in MOFs (IRMOF-1,
IRMOF-6, IRMOF-11, IRMOF-20, MOF-177, MOF-74, and
HKUST-1), and good correlation was found between surface
area and saturation capacity.4
Since a very large number of MOFs have been synthesized to
date, and many more are possible, experimental screening of
appropriate MOFs for gas storage is not economically feasible.
As an alternative, molecular simulations have been conducted on
gas adsorption in MOFs. With the ever-increasing computational
power, molecular simulations have become increasingly impor-
tant in the chemical and material sciences. In particular, the
microscopic insights gained from simulations are indispensable in
*Corresponding author. Tel: 65-65165083, Fax: 65-67791936, E-mail:
chejj@nus.edu.sg.
(1) Yaghi, O. M.; OKeeffe, M.; Ockwig, N. W.; Chae, H. K.; Eddaoudi, M.;
Kim, J. Nature 2003, 423, 705.
(2) Eddaoudi, M.; Kim, J.; Rosi, N.; Vodak, D.; Wachter, J.; OKeefe, M.;
Yaghi, O. M. Science 2002, 295, 469.
(3) Millward, A. R.; Yaghi, O. M. J. Am. Chem. Soc. 2005, 127, 17998.
(4) Wong-Foy, A. G.; Matzger, A. J.; Yaghi, O. M. J. Am. Chem. Soc. 2006, 128,
3493.
processes and subsequently to provide guidelines for the intelli-
gent design of new materials. CH4 adsorption in MOFs has been
simulated to identify storage materials for environmentally
friendly energy sources.5 Adsorption of various light gases in
MOFs has also been simulated and compared to available
experimental data to test the validity of model potentials.6,7 H2
uptake in IRMOFs obtained from simulations was found to
correlate well with the heat of adsorption at low pressures, with
the surface area at moderate pressures, and with the free volume
at high pressuress.8 Among the different types of nanoporous
materials (IRMOF-1, silicalite, and a nanoporous carbon), IR-
MOF-1 was found to have the largest capacity for CO2 and CH4.9
MOFs with various linkers, pore sizes, and topologies were
investigated for CO2 storage, and the capacity was found to be
a complex interplay of framework density, surface area, free
volume, and porosity.10,11 One-, two-, and three-dimensional
covalent-organic frameworks (COFs) were examined for CO2
storage, and exceptionally high capacity of up to 100 mmol/g was
predicted in COF-105 and COF-108.12 CO2 adsorption was also
simulated in zeolitic imidazolate frameworks, and the small pores
formed by the organic linkers were found to be the preferential
adsorption sites.13 A simulation tool was developed to quickly
calculate the unit cell shape and size of MOF structures loaded
(5) Duren, T.; Sarkisov, L.; Yaghi, O. M.; Snurr, R. Q. Langmuir 2004, 20, 2683.
(6) Garberoglio, G.; Skoulidas, A. I.; Johnson, J. K. J. Phys. Chem. B 2005, 109,
13094.
(7) Skoulidas, A. I.; Sholl, D. S. J. Phys. Chem. B 2005, 109, 15760.
(8) Frost, H.; Duren, T.; Snurr, R. Q. J. Phys. Chem. B 2006, 110, 9565.
(9) Babarao, R.; Hu, Z. Q.; Jiang, J. W.; Chempath, S.; Sandler, S. I. Langmuir
2007, 23, 659.
(10) Babarao, R.; Jiang, J. W. Langmuir 2008, 24, 6270.
(11) Yang, Q. Y.; Zhong, C. L.; Chen, J. F. J. Phys. Chem. C 2008, 112, 1562.
(12) Babarao, R.; Jiang, J. W. Energy Environ. Sci 2008, 1, 139.
(13) Liu, D. H.; Zheng, C. C.; Yang, Q. Y.; Zhong, C. L. J. Phys. Chem. C 2009,
113, 5004.

Langmuir 2010, 26(11), 87438750 Published on Web 01/27/2010 DOI: 10.1021/la904502h 8743
Article
Figure 1. Unit cell of dehydrated MIL-101 constructed from experimental crystallographic data, energy minimization, and density
functional theory calculation (see the text for details). The pentagonal and hexagonal windows are enlarged for clarity. Color code: Cr,
orange polyhedra; F, cyan; C, blue; O, red; H, white.
with adsorbates.14 Further, gas cluster formation for adsorption
in MOFs was observed at subcritical temperatures.15 Several
recent reviews have highlighted the central importance of mole-
cular simulations in characterizing MOFs for adsorption
applications.16-18
There has been considerable interest in gas adsorption in the
MIL (Material Institute Lavoisier) series of MOFs. Upon CO2
adsorption in MIL-53, a structural transformation was observed
between the large- and narrow-pore forms.19 The favorable sites in
MIL-53 (Al, Cr) and MIL-47 (V) were identified for CO2 adsorp-
tion to elucidate the adsorption mechanism and breathing effect.20
A record high uptake of 40 mmol/g at 303 K and 50 bar was found
for CO2 in MIL-101.21 A related study reported the adsorp-
tion properties of CO2 and CH4 along with C3H8, SF6, and Ar
in MIL-101 using a standard gravimetric method.22 MIL-101 is a
chromium terephthalate-based mesoscopic MOF and one of
the most porous materials reported to date. It is stable in air or
boiling water, and its structure is not altered in organic solvents
or solvothermal conditions. These remarkable properties and the
highly porous structure have resulted in great interest in MIL-101
for gas storage, catalysis, drug delivery, and other versatile applica-
tions.23 Despite a number of experimental studies in this unique
MOF,24 we are not aware of any molecular simulation study
reported for gas adsorption in MIL-101. However, such molecular
information is important to understand the microscopic adsorp-
tion behavior and facilitate the development of new superadsor-
bents.
(14) Dubbeldam, D.; Krishna, R.; Snurr, R. Q. J. Phys. Chem. C 2009, 113, 19317.
(15) Krishna, R.; van Baten, J. M. Langmuir 2009, in press.
(16) Duren, T.; Bae, S. Y.; Snurr, R. Q. Chem. Soc. Rev. 2009, 38, 1237.
(17) Keskin, S.; Liu, J.; Rankin, R. B.; Johnson, J. K.; Sholl, D. S. Ind. Eng.
Chem. Res. 2009, 48, 2355.
(18) Krishna, R. J. Phys. Chem. C 2009, 113, 19756.
(19) Ramsahye, N. A.; Maurin, G.; Bourrelly, S.; Llewellyn, P. L.; Devic, T.;
Serre, C.; Loiseau, T.; Ferey, G. Adsorption 2007, 13, 461.
(20) Ramsahye, N. A.; Maurin, G.; Bourrelly, S.; Llewellyn, P. L.; Serre, C.;
Loiseau, T.; Devic, T.; Ferey, G. J. Phys. Chem. C 2008, 112, 514.
(21) Llewellyn, P. L.; Bourrelly, S.; Serre, C.; Vimont, A.; Daturi, M.; Hamon,
L.; De Weireld, G.; Chang, J. S.; Hong, D. Y.; Hwang, Y. K.; Jhung, S. H.; Ferey,
G. Langmuir 2008, 24, 7245.
(22) Chowdhury, P.; Bikkina, C.; Gumma, S. J. Phys. Chem. C 2009, 113, 6616.
(23) Ferey, G.; Mellot-Draznieks, C.; Serre, C.; Millange, F.; Dutour, J.; Surble,
S.; Margiolaki, I. Science 2005, 309, 2040.
(24) Hong, D. Y.; Hwang, Y. K.; Serre, C.; Ferey, G.; Chang, J. S. Adv. Funct.
Mater. 2009, 19, 1.
8744 DOI: 10.1021/la904502h
Chen et al.
In this study, we investigate the adsorption of CO2 and CH4 in
MIL-101 at 303 K combining quantum mechanics and molecular
simulation. To examine the role of terminal water molecules
coordinated to metal sites, the adsorption is examined in both
dehydrated and hydrated structures. We first introduce the
molecular models in section 2, and in section 3, the simulation
methods are briefly described. In section 4, the adsorption
isotherms, simulation snapshots, and radial distribution functions
are presented and discussed. The adsorption of the CO2/CH4
mixture is also examined. The concluding remarks are summar-
ized in section 5.
2. Models
MIL-101 is assembled by corner-sharing supertetrahedra,
which consist of Cr3O trimers and 1,4-benzenedicarboxylic
acids.23 The supertetrahedra are microporous with an 8.6 A free
aperture. The 4 vertices and 6 edges of the supertetrahedra are
occupied by the trimers and organic linkers, respectively. MIL-101
has an augmented three-dimensional MTN zeotype structure
with a very large cell volume (702 000 A3). Two types of
mesoporous quasi-spherical cages exist in MIL-101: a small cage
of 20 supertetrahedra and a free diameter of 29 A accessible
through a pentagonal window with a 12 A aperture, and a large
cage of 28 supertetrahedra and a free diameter of 34 A accessible
through both hexagonal and pentagonal windows with a 14.7 
16 A2 aperture. MIL-101 is highly porous with BET and Lang-
muir surface areas of about 4100 ( 200 and 5900 ( 300 m2/g,
respectively. Each octahedral Cr is bonded to 4 oxygen atoms
from carboxylates, 1 3O oxygen atom, and 1 terminal site. The
terminal site could be occupied by a fluorine atom or a terminal
water molecule. As a consequence, there are 3 terminal sites in
each Cr3O trimer and the ratio of fluorine to water is 1:2. The
terminal water molecules can be removed after dehydration, thus
providing exposed metallic Lewis acid sites in MIL-101.23
Figure 1 shows a unit cell of dehydrated MIL-101, in which the
pentagonal and hexagonal windows are enlarged for clarity. We
constructed the unit cell in a three-step process based on experi-
mental crystallographic data and computational methods. Each
Cr3O trimer in the crystallographic data was coordinated with
three terminal water molecules. In the first step, all disordered and
terminal water molecules in the crystallographic data were
Langmuir 2010, 26(11), 87438750
Chen et al.
Figure 2. Optimized geometry and Merz-Kollman (MK) charges of the Cr3O trimer with terminal fluorine and water molecules in (a)
dehydrated and (b) hydrated MIL-101. The cleaved bonds of Cr3O (indicated by the circles) were saturated by methyl group. Color code: Cr,
orange; F, cyan; C, blue; O, red; H, white.
Figure 3. Electrostatic potential maps around the Cr3O trimer in (a) dehydrated and (b) hydrated MIL-101.
removed (actually only the oxygen atoms of water molecules were
removed as the hydrogen atoms were not in the data). To
construct hydrated MIL-101, one terminal water molecule in
each Cr3O trimer was removed. In the second step, hydrogen and
fluorine atoms were added to phenyl rings and exposed Cr sites
(one in each Cr3O trimer). For hydrated MIL-101, hydrogen
atoms were also added to the oxygen atoms of terminal water
molecules. The unit cell was then energetically minimized using
Materials Studio,25 in which the framework atoms were repre-
sented by the universal force field (UFF).26 The phenyl rings in
experimental crystallographic data were slightly nonplanar and
transformed to planar after energy minimization. In the third
step, geometry optimization was conducted by density-functional
theory (DFT) in DMol326 to precisely determine the positions of
the added hydrogen and fluorine atoms. Because of the very large
number of atoms in a unit cell, the DFT optimization was
performed on a Cr3O trimer as shown in Figure 2. To maintain
the original hybridization, the cleaved bonds of the trimer were
saturated by methyl groups. The Becke exchange plus Lee-
Yang-Parr correlation functional and all-electron core potentials
were used. The double- numerical polarization (DNP) basis set
was adopted, which is comparable to the 6-31G(d,p) Gaussian-
type basis set. The DNP basis set incorporates d-type polarization
(25) Materials Studio, 4.3 ed.; Accelrys, San Diego, 2008.
(26) Rappe, A. K.; Casewit, C. J.; Colwell, K. S.; Goddard, W. A.; Skiff, W. M.
J. Am. Chem. Soc. 1992, 114, 10024.
Langmuir 2010, 26(11), 87438750
Article
into heavier atoms and p-type polarization into hydrogen atoms.
In both dehydrated and hydrated MIL-101, the optimized dis-
tance from F to Cr1 was found to be approximately 1.86 A. In the
hydrated MIL-101, the oxygen atom (OW) of terminal water to
Cr2 is about 2.26 A.
The DFT calculations also gave electrostatic potential maps
around the trimer. Figure 3 illustrates that the fluorine atom
possesses large electronegativity in both dehydrated and hydrated
MIL-101. The significant difference is the electric fields generated
by the terminal H2O molecules in the hydrated MIL-101. As we
shall see, this has an intriguing effect on adsorption. The atomic
charges of MIL-101 framework atoms were fitted to the electro-
static potentials using the Merz-Kollman (MK) scheme.27,28 It is
worthwhile to note that the concept of atomic charges is solely an
approximation, and no unique straightforward method is cur-
rently available to determine them on a rigorous level. To evaluate
the sensitivity of framework charges, the Mulliken population
analysis29 was also used to estimate the charges.29 The MK and
Mulliken charges are listed in Figure 2 and Supporting Informa-
tion Figure S1, respectively.
The free volumes of the dehydrated and hydrated MIL-101
structures were estimated by Monte Carlo method.30 A single
(27) Singh, U. C.; Kollman, P. A. J. Comput. Chem. 1984, 5, 129.
(28) Besler, B. H.; Merz, K. M., Jr.; Kollman, P. A. J. Comput. Chem. 1990, 11,
431.
(29) Mulliken, R. S. J. Chem. Phys. 1955, 23, 1833.
(30) Myers, A. L.; Monson, P. A. Langmuir 2002, 18, 10261.
DOI: 10.1021/la904502h 8745
Article
Table 1. Potential Parameters and Atomic Charges of CO2, CH4, and
Terminal H2O
site (adsorbate) site (A) /kB (K)
CO2 C 2.789 29.66
O 3.011 82.96
CH4 (united-atom) 3.73 148.0
CH4 (five-site) C 3.40 55.055
H 2.65 7.901
H2O OW 3.151 76.42
HW 0 0
R
helium atom was inserted into the framework and V exp[-uHe ad -
(r)/kBT] dr gave the free volume, in which uHead is the interaction
between helium and adsorbent. It should be noted that the free
volume estimated by helium adsorption is slightly greater than the
actual geometrical volume.31 The calculated free volumes in the
dehydrated and hydrated MIL-101 were 1.89 and 1.85 cm3/g,
respectively; and the porosities were 0.83 and 0.81. In our recent
study on the energetic and dynamic behavior of a drug molecule,
ibuprofen,32 the MIL-101 model was adopted from experimental
crystallographic data. With the new model constructed in the
current study, the loading and mobility of ibuprofen were found
to remain largely unchanged.
For the adsorption study, CO2 was represented as a three-site
rigid molecule and its intrinsic quadrupole moment was described
by a partial charge model. The partial charges on C and O atoms
were qC = 0.576e and qO = -0.288e (e = 1.6022  10-19),
respectively. The C-O bond length used was 1.18 A and the bond
angle OCO was 180. CO2-CO2 interactions were modeled as
a combination of Lennard-Jones (LJ) and Coulombic potentials.
A united-atom model was used for CH4 with the LJ potential
parameters from the TraPPE force field that was developed to
reproduce the critical parameters and saturated liquid densities of
alkanes.33 The united-atom model has been commonly used for
the adsorption and transport studies of CH4. For comparison, a
fully atomistic five-site model34 was also adopted here for CH4
adsorption, in which the octopole moment was incorporated. The
LJ potential parameters of the terminal water molecules used
were taken from TIP3P (three-point transferable interaction
potential) model.35 It has been shown that the TIP3P model gives
a reasonably good interaction potential compared to experimen-
tal value.36 Nevertheless, the effect of terminal water molecules
from the current study would not depend on the model of water.
Table 1 lists the potential parameters and atomic charges of CO2,
CH4, and terminal water molecule. The dispersion interactions of
the framework atoms in MIL-101 were modeled using the UFF.26
The Lorentz-Berthelot combining rules were used to calculate
the cross LJ interaction parameters. A number of simulation
studies have shown that UFF can accurately predict gas adsorp-
tion in various MOFs.6,7,9,10
3. Methods
Adsorption of CO2 and CH4 in both dehydrated and hydrated
MIL-101 was simulated at 303 K using grand canonical Monte
Carlo (GCMC) method37 at fixed adsorbate chemical potential,
volume, and temperature. As the chemical potentials of adsorbate
(31) Do, D. D.; Herrera, L. F.; Do, H. D. J. Colloid Interface Sci. 2008, 328, 110.
(32) Babarao, R.; Jiang, J. W. J. Phys. Chem. C 2009, 113, 18287.
(33) Martin, M. G.; Siepmann, J. I. J. Phys. Chem. B 1998, 102, 2569.
(34) Sun, Y.; Spellmeyer, D.; Pearlman, D. A.; Kollman, P. A. J. Am. Chem.
Soc. 1992, 114, 6798.
(35) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein,
M. L. J. Chem. Phys. 1983, 79, 926.
(36) Mark, P.; Nilsson, L. J. Phys. Chem. A 2001, 105, 9954.
(37) Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids; Oxford
University Press: Oxford, 1987.
8746 DOI: 10.1021/la904502h
Chen et al.
Z (e)
0.576
-0.288
0
-0.66
0.165
-0.401
0.758
Figure 4. Adsorption isotherms of CO2 and CH4 on a gravimetric
basis. The squares, diamonds, and circles are the experimental data
in MIL-101a (as-synthesized), MIL-101b (activated by hot
ethanol), and MIL-101c (activated by hot ethanol and KF),
respectively.21
in adsorbed and bulk phases are identical at thermodynamic
equilibrium, GCMC allows one to directly relate the chemical
potentials of the adsorbate in both phases and has been widely
used to simulate adsorption. The simulation box in this study
contained a unit cell of MIL-101 and the periodic boundary
conditions were used in all three dimensions. The framework
was assumed to be rigid and the framework atoms were fixed
during simulation. In the hydrated MIL-101, the terminal water
molecules were strongly coordinated to metal sites. In reality, the
water molecules may rotate and vibrate slightly. However, as the
temperature in our study is much lower than the activation
temperature required in experiment, the terminal water molecules
were also assumed to be rigid. The vibration of framework is
expected to have a marginal effect for adsorption of small gases. A
recent simulation study on the adsorption of noble gases in
IRMOF-1 demonstrated that rigid and semiflexible frameworks
gave very similar results at both low and room temperatures.38
The simulation box was divided into three-dimensional grids with
the potential energy of each adsorbate atom tabulated in advance
and then used by interpolation during simulation. This treatment
accelerated the simulation by approximately 1 to 2 orders of
magnitude.
The LJ interactions were evaluated with a spherical cutoff of 15
A with long-range corrections added; the Coulombic interactions
were calculated using the Ewald sum method.37 The real/recipro-
cal space partition parameter and the cutoff for reciprocal lattice
vectors were chosen to be 0.2 A-1 and 10, respectively, to ensure
the convergence of Ewald sum. The number of trial moves in a
typical GCMC simulation was 2  107, though additional trial
moves were used at high pressures. The first 107 moves were used
for equilibration and the remaining 107 moves for computing the
ensemble averages. Five types of trial moves were randomly
attempted in GCMC simulations: displacement, rotation, and
partial regrowth at a neighboring position; complete regrowth at a
new position; and swap with reservoir including creation and
deletion with equal probability. In the simulation of CO2/CH4
mixture, an additional trial move, identity exchange, was also
performed. The simulation uncertainties were smaller than the
symbol sizes presented below.
4. Results and Discussion
4.1. Sensitivity of Framework Charges. CO2 adsorption in
the dehydrated MIL-101 was simulated with the MK and
Mulliken charges on framework atoms, and the results are shown
in Supporting Information Figure S2. For comparison, the
(38) Greathouse, J. A.; Kinnibrugh, T. L.; Allendorf, M. D. Ind. Eng. Chem.
Res. 2009, 48, 3425.
Langmuir 2010, 26(11), 87438750
Chen et al.
Figure 5. Adsorption isotherms of CO2 and CH4 based on the number of adsorbed molecules per unit cell.
experimental data are included from the study by Llewellyn et al.21
They measured the adsorption of CO2 and CH4 in three MIL-101
samples, namely, MIL-101a as-synthesized, MIL-101b activated
by ethanol, and MIL-101c activated by ethanol and KF.21 These
samples differ primarily in the amount of residual terephthalic
acid and in the density of Lewis acid Cr sites (500, 700, and
1000 mol/g, respectively). CH4 adsorption was found to be
essentially the same in all three samples. However, CO2 adsor-
ption was affected by the activation method, and the extent of
adsorption increased in the order of MIL-101a < MIL-101b <
MIL-101c.21 Among the three samples, MIL-101b and MIL-101c
were activated more thoroughly and contained less residual
terephthalic acid and more open Cr sites; consequently, they
had stronger interactions with CO2 and larger free volume than
MIL-101a. The dehydrated MIL-101 model in our study is similar
to experimental MIL-101b and MIL-101c rather than MIL-101a.
As seen in Supporting Information Figure S4, the predicted CO2
adsorption from the MK and Mulliken charges match well with
the experimental data of MIL-101b and MIL-101c. This validates
the models and force fields used in our study and suggests that
CO2 adsorption is relatively insensitive to the method, either MK
or Mulliken, for the framework charges. Unless otherwise stated,
the results for CO4 presented below are based on the MK charges.
4.2. United-Atom and Five-Site Models of CH4. To
examine the performance of the united-atom and five-site models
of CH4, Supporting Information Figure S3 shows the adsorption
isotherms of CH4 in the dehydrated MIL-101 predicted from both
models. Because CH4 adsorption was found experimentally to be
insensitive to the activation method, only the experimental data in
MIL-101c are included in Supporting Information Figure S3. The
predictions of both united-atom and five-site models are in fairly
good agreement with the experimental data, especially at low
pressures. Nevertheless, the united-atom model overestimates,
whereas the five-site model underestimates the extent of adsorp-
tion. Such a trend was also observed in CH4 adsorption on graphite
and in graphitic slit pores and attributed to the better packing of
the united-atom model in confined space.39 Unless otherwise
stated, the results for CH4 are based on the united-atom model.
4.3. Adsorption of Pure CO2 and CH4. Figure 4 shows the
adsorption isotherms of CO2 and CH4 in both dehydrated and
hydrated MIL-101 on a gravimetric basis. CH4 uptake increases
almost linearly with increasing pressure and is still not saturated
at 50 bar because of the very porous structure of MIL-101. This
implies that the interaction between CH4 and the MIL-101 is
weak and a higher pressure is needed to attain saturation. Despite
a slight overestimation at high pressures, the simulated isotherms
(39) Do, D. D.; Do, H. D. J. Phys. Chem. B 2005, 109, 19288.
Langmuir 2010, 26(11), 87438750
Article
of CH4 in both dehydrated and hydrated MIL-101 are in good
agreement with the experimental data. The dehydrated MIL-101
exhibits greater adsorption than its hydrated counterpart at high
pressures. As will be seen, however, the opposite is observed at
low pressures due to the presence of terminal water molecules.
CO2 adsorption is apparently greater than CH4 because CO2
has a quadrupole moment and thus interacts more strongly with
the adsorbent. The simulated isotherms of CO2 in dehydrated and
hydrated MIL-101 compare fairly well with the experimental data
in MIL-101b and MIL-101c samples. The deviations can be
attributed primarily to the difference between theoretical model
and real sample. In our study, the model does not contain any
residual terephthalic acid, though some probably remained in the
sample. CO2 uptake in MIL-101 at 50 bar ranges from 30 to 40
mmol/g depending on the activation method, which is higher than
that in IRMOF-1 (22 mmol/g)3 and Cu-BTC (11 mmol/g),3 and
comparable to that in MOF-177 (33.5 mmol/g at 42 bar)3 and
UMCM-1, IRMOF-10, and IRMOF-14 (40-45 mmol/g).10
Similar to that observed in CH4 adsorption, the dehydrated
MIL-101 exhibits greater adsorption for CO2 at high pressures,
but the reverse is true at low pressures, which will be further
discussed below.
Figure 5 shows the adsorption isotherms of CO2 and CH4 on
the basis of the number of adsorbed molecules per unit cell. At low
pressures, the adsorption particularly of CO2 in the hydrated
MIL-101 is greater than in the dehydrated framework. Interest-
ingly, the opposite is observed at high pressures. Such a phenom-
enon was recently also found for the adsorption of CO2 and CH4
in microporous Cu-BTC40 and of CO2/H2 mixture in charged soc-
MOF.41 At low pressures, the enhanced adsorption in the
hydrated MIL-101 is attributed to the terminal water molecules
that act as additional adsorption sites; thus, adsorbates experience
stronger interactions. As seen in Figure 5a, the enhancement is
more pronounced for CO2 than CH4. This is because CO2 is a
quadrapolar molecule and interacts strongly with the charged
hydrogen and oxygen atoms of the terminal water molecules.
However, the united-atom model of CH4 can only interact with
water molecules through a weak dispersion potential. It can be
expected that the enhancement would be larger if using the five-
site model of CH4. At high pressures, the pores in MIL-101 are
largely filled by adsorbate, and the free volume is a dominant
factor in determining adsorption capacity. Apparently, the term-
inal water molecules occupy part of the volume and prevent the
framework from accommodating more adsorbate molecules.
(40) Yazaydin, A. O.; Benin, A. I.; Faheem, S. A.; Jakubczak, P.; Low, J. J.;
Willis, R. R.; Snurr, R. Q. Chem. Mater. 2009, 21, 1425.
(41) Jiang, J. W. AIChE J. 2009, 55, 2422.
DOI: 10.1021/la904502h 8747
Article
Figure 6. Snapshots of CO2 and CH4 in a pentagonal window in the dehydrated MIL-101 at 10, 100, and 1000 kPa. Color code: Cr, orange; F,
cyan; C, blue; O, red; H, white; CO2, green; CH4, pink.
Figure 7. Radial distribution functions of CO2 and CH4 around
Cr1 and Cr2 atoms in the dehydrated MIL-101 at 10, 100, 1000, and
5000 kPa.
Consequently, the hydrated MIL-101 exhibits lower adsorption
capacity than the dehydrated form at high pressures.
Figure 6 contains the simulation snapshots of CO2 and CH4 in
a pentagonal window in the dehydrated MIL-101 that can be used
to identify the favorable adsorption sties. At 10 kPa, the adsor-
bate molecules are located exclusively in the supertetrahedra.
Attributed to the strong overlap of surface potentials, the micro-
porous supertetrahedra are the most favorable adsorption sites.
At 100 kPa with increased adsorption, adsorbate molecules
appear in the pentagonal window in addition to the supertetrahe-
dra. Compared to CH4, CO2 molecules also occupy the edge of
the pentagonal window and are close to the exposed Cr2 sites.
Note that, as shown in Supporting Information Figure S4, the
exposed metal sites are outside the supertetrahedra. At 1000 kPa,
8748 DOI: 10.1021/la904502h
Chen et al.
Figure 8. Schematic locations of CO2 and CH4 near Cr3O trimer
in the dehydrated MIL-101. Color code: Cr, orange; F, cyan; C,
blue; O, red; H, white; CO2, green; CH4, pink.
Figure 9. Radial distribution functions of CO2 and CH4 around
the oxygen (OW) atoms of terminal water molecules in the hy-
drated MIL-101 at 10, 100, 1000, and 5000 kPa.
more adsorbate molecules populate the pentagonal window.
Though not shown here, at very high pressures the mesoporous
cages are gradually filled.
The simulation snapshots in Figure 6 only provide qualitative
information. To quantitatively characterize the adsorption sites,
the radial distribution functions of adsorbate molecules around
framework atoms were calculated using
Nij r, r rV
gij r 1
4r2 rNi Nj
where r is the distance between species i and j, Nij(r,r r) is
the ensemble averaged number of j species around an i species
Langmuir 2010, 26(11), 87438750
Chen et al.
Figure 10. Adsorption of equimolar CO2/CH4 mixture: (a) isotherm and (b) selectivity.
within a shell from r to r r, V is the system volume, and Ni and
Nj are the numbers of the i and j species. Figure 7 shows the g(r) of
CO2 and CH4 around the Cr1 and Cr2 atoms in the dehydrated
MIL-101 at various pressures, respectively. For all four cases
shown, pronounced peaks are observed at r = 5-9 A, revealing
that the supertetrahedra are preferential adsorption sites.
Furthermore, two additional peaks are observed in the g(r) of
CO2-Cr1. This is attributed to the attractions between fluorine
atoms (bonded to Cr1 sites) and CO2 molecules in two neighbor-
ing supertetrahedra. As illustrated by the electrostatic potential
maps in Figure 3, fluorine atoms possess large electronegativity
and, therefore, interact strongly with CO2 molecules.
There is a distinct difference in the short region of g(r) around
Cr1 and Cr2. In the dehydrated MIL-101, the Cr1 sites are bonded
with fluorine atoms, whereas the Cr2 sites are exposed. Small
peaks are observed in the g(r) of CO2-Cr2 at r = 3.5 A; though
there are no peaks in the g(r) of CH4-Cr2, shoulders exist in the
region r = 3-4 A. In marked contrast, such peaks and shoulders
are not seen around the Cr1 sites. That is, the adsorbate molecules
are closer to the Cr2 rather than the Cr1 sites. Because of the
quadrapole moment, CO2 interacts more strongly with the Cr2
sites than CH4, leading to the peaks in g(r) of CO2-Cr2. Figure 8
schematically demonstrates the locations of CO2 and CH4 mole-
cules around the Cr2 sites. We should note that the region near the
Cr2 sites is substantially smaller than the size of supertetrahedra;
therefore, the peaks in g(r) at 3.5 A are much lower and narrower
than those at 5-9 A.
With increasing pressure, the number of adsorbed molecules
increases. As defined by eq 1, g(r) is the local density relative to the
overall density. The peaks in g(r) at 5-9 A decrease as pressure
increases, since the local density increases less rapidly than the
overall density. Similar behavior was also observed in our recent
studies for the adsorption of gas and water in charged rho zeolite-
like MOF.42,43 Interestingly, different behavior is found for the
peaks at 3.5 A close to the exposed Cr2 sites. Specifically, the peak
of CO2-Cr2 increases with increasing pressure from 10 to 100 and
then to 1000 kPa, but decreases from 1000 to 5000 kPa. This
implies that the number of CO2 molecules near the exposed Cr2
sites increases up to 1000 kPa; however, these regions are small
and get saturated beyond 1000 kPa. Unlike CO2-Cr2, the
shoulder of CH4-Cr2 still increases marginally when pressure
rises from 1000 to 5000 kPa. This is because the interaction of
CH4-Cr2 is weaker than that of CO2-Cr2; thus, the CH4
saturation on the Cr2 sites occurs at higher pressure. The radial
(42) Babarao, R.; Jiang, J. W. J. Am. Chem. Soc. 2009, 131, 11417.
(43) Nalaparaju, A.; Babarao, R.; Zhao, X. S.; Jiang, J. W. ACS Nano 2009,
2563.
Langmuir 2010, 26(11), 87438750
Article
distribution functions discussed here provide useful insights into
the favorable interaction sites in the dehydrated MIL-101.
We next consider the hydrated MIL-101, in which both Cr1 and
Cr2 sites are coordinated and thus exhibit similar g(r). As shown in
Supporting Information Figure S5, the g(r) are almost the same for
CO2-Cr1 and CO2-Cr2, and also for CH4-Cr1 and CH4-Cr2.
While there are peaks at 5-9 A, no peaks or shoulders are observed
at 3.5 A, which are significantly different from Figure 7. The
pressure dependence of the g(r) is similar to that in the hydrated
framework, that is, the peak decreases with rising pressure.
Figure 9 shows the radial distribution functions of CO2 and
CH4 around the oxygen atoms (OW) of the terminal water
molecules in the hydrated MIL-101 at 10, 100, 1000, and 5000
kPa, respectively. Similar to the g(r) of CO2-Cr2 and CH4-Cr2 in
Figure 7, peaks and shoulders are observed in the g(r) of
CO2-OW and CH4-OW at r = 3.8 A. The reason is that the
terminal water molecules act as additional adsorption sites,
especially for CO2. Comparing Figures 7 and 9, we find that the
peaks and shoulders are broader in the latter, suggesting that the
terminal water molecules result in stronger interactions with
adsorbates than do the exposed Cr2 sites. As a consequence,
adsorption at low pressures is greater for both CO2 and CH4 in the
hydrated MIL-101.
4.4. Adsorption of CO2/CH4 Mixture. Adsorption of an
equimolar mixture of CO2 and CH4 was also simulated in the
dehydrated and hydrated MIL-101. The isotherm in Figure 10a
shows that the extent of adsorption of each component is not
substantially affected by hydration. With a closer look, however,
we see that CO2 adsorption at low pressures is marginally higher in
the hydrated MIL-101 than in the dehydrated counterpart. At high
pressures, there is no discernible difference in adsorption between
the two structures. In contrast, CH4 adsorption in dehydrated
MIL-101 is slightly higher at moderate and high pressures. The
separation factor of the mixture is quantified by selectivity Si/j =
(xi/xj)(yj/yi), where xi and yi are the mole fractions of component i
in adsorbed and bulk phases, respectively. As shown in Figure 10b,
with increasing pressure the selectivity of CO2/CH4 first increases,
then drops rapidly, and finally increases slightly over the pressure
range under study. The initial increase is due to the preferential
adsorption of CO2 in the multiple supertetrahedra. Upon adsorp-
tion occurring in the less attractive windows and mesoporous cage,
the selectivity drops. At high pressures, the selectivity increases
because of the cooperative attractions between the dominantly
adsorbed CO2 molecules. Interestingly, the selectivity of CO2/CH4
is higher in the hydrated MIL-101, implying that the separation
efficacy could be enhanced by introducing terminal H2O mole-
cules. This was also observed in the adsorption of CO2/CH4 in
Cu-BTC40 and of CO2/H2 in charged soc-MOF.41
DOI: 10.1021/la904502h 8749
Article
5. Conclusions
We have investigated the adsorption of CO2 and CH4 in
mesoporous MIL-101. The dehydrated and hydrated MIL-101
structures were constructed on the basis of experimental crystal-
lographic data, molecular mechanical minimization, and first-
principles optimization. Adsorption occurs exclusively in the
microporous supertetrahedra at low pressures, and then in the
mesoporous cages at high pressures. The simulated adsorption
isotherm of CO2 matches well with measurement, despite
variations in experimental data depending on the activation
method used. CH4 adsorption increases almost linearly with
increasing pressure and good agreement was found between
simulation and experimental results. The terminal water mole-
cules in the hydrated MIL-101 have an interesting effect on
adsorption by acting as additional adsorption sites that enhance
adsorption at low pressures. The enhancement is greater for
CO2, as it is a quadrapolar molecule and interacts strongly with
the charged hydrogen and oxygen atoms of the water molecules.
However, at high pressures the terminal water molecules reduce
free volume and lead to less adsorption compared to the de-
hydrated MIL-101.
8750 DOI: 10.1021/la904502h
Chen et al.
In the dehydrated MIL-101, peaks and shoulders exist in the
short region of the radial distribution functions around the exposed
Cr2 sites, but not the fluorine saturated Cr1. In the hydrated MIL-
101, the radial distribution functions are similar for all the Cr sites.
Compared to those in the dehydrated MIL-101, the peaks and
shoulders near the terminal water molecules are broader. This
indicates that the terminal water molecules interact more strongly
with adsorbates than do the exposed Cr2 sites and the hydrated
MIL-101 has greater adsorption at low pressures. The adsorption
selectivity of CO2 over CH4 is slightly higher in the hydrated MIL-
101. The current study provides useful insights into the microscopic
adsorption behavior in unique MIL-101 and underlines the inter-
esting effect of the terminal water molecules on adsorption.
Acknowledgment. The authors gratefully acknowledge the
support from the National University of Singapore and the
Singapore National Research Foundation.
Supporting Information Available: Additional figures as
described in the text. This material is available free of charge
via the Internet at http://pubs.acs.org.
Langmuir 2010, 26(11), 87438750