A new series of isoreticular copper-based metal-organic frameworks containing non-linear linkers with different group 14 central atoms

Article abstract of DOI:10.1039/c2jm15523k

The synthesis and characterization of two new metal-organic frameworks (MOFs) UHM-2 and UHM-4, which are isoreticular to PCN-12 and UHM-3 (see Inorg. Chem., 2009, 48, 6559), are presented. Both materials comprise dicopper paddle-wheel units and the new tetracarboxylic acid linkers 5,5′-(propane- 2,2-diyl)diisophthalic acid and 5,5′-(dimethylgermanediyl) diisophthalic acid, respectively. In addition, isoreticular mixed-linker MOFs were synthesized with the silicon- and germanium containing linkers. UHM-2 to -4 constitute a homologous series allowing a systematic study on how the incorporation of atoms with higher polarizability in linkers may enhance the overall hydrogen uptake of MOFs. The experimental work is complemented by dispersion-corrected DFT calculations on C_2v conformers of model systems representing the linkers of these MOFs.

Full text of DOI:10.1039/c2jm15523k

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links <
C
Journal of
Materials Chemistry
Cite this: J. Mater. Chem., 2012, 22, 10294
www.rsc.org/materials
PAPER
A new series of isoreticular copper-based metal–organic frameworks
containing non-linear linkers with different group 14 central atoms†
a
Stephanie E. Wenzel, Michael Fischer, Frank Hoffmann and Michael Froba
b
b
b
*
€
Received 28th October 2011, Accepted 27th January 2012
DOI: 10.1039/c2jm15523k
The synthesis and characterization of two new metal–organic frameworks (MOFs) UHM-2 and
UHM-4, which are isoreticular to PCN-12 and UHM-3 (see Inorg. Chem., 2009, 48, 6559), are
presented. Both materials comprise dicopper paddle-wheel units and the new tetracarboxylic acid
linkers 5,50-(propane-2,2-diyl)diisophthalic acid and 5,50-(dimethylgermanediyl) diisophthalic acid,
respectively. In addition, isoreticular mixed-linker MOFs were synthesized with the silicon- and
germanium containing linkers. UHM-2 to -4 constitute a homologous series allowing a systematic
study on how the incorporation of atoms with higher polarizability in linkers may enhance the overall
hydrogen uptake of MOFs. The experimental work is complemented by dispersion-corrected DFT
calculations on C2v conformers of model systems representing the linkers of these MOFs.
They show gravimetric (excess) hydrogen uptakes up to ꢀ10 wt%
1. Introduction
at moderate pressures (40–70 bar) and a temperature of 77 K.1–4
However, in order to be efficient hydrogen storage systems
under non-cryogenic conditions their performance has to
improve considerably as the solid–fluid interactions are much too
weak. Typically, MOFs show heat of adsorption values qst (at
cryogenic conditions and at very low coverage) in the range of 4
to 8 kJ molꢁ1 (ref. 5 and 6) (the highest value reported so far
amounts to 13.5 kJ molꢁ1)7 while for room-temperature storage
an optimum qst value of ꢀ15 kJ molꢁ1 has been proposed by
Bhatia and Myers.8 Ideally, the desired materials should exhibit
qst versus coverage progressions in which the initial qst value at
extremely low coverage is around 15 kJ molꢁ1 and in which this
qst value decreases only very weakly with increasing coverage.
Several strategies have been proposed in order to strengthen the
overall solid–fluid interactions: (i) incorporation of adsorption
sites into MOFs, which exhibit a rather strong and localized
interaction with hydrogen, e.g. coordinatively unsaturated metal
sites (also called ‘‘open metal sites’’, OMS); (ii) lithiation of
MOFs is sometimes beneficial as it was shown in a number of
studies, although such MOFs are reported to have a rather low
stability;9,10 (iii) the utilization of polycyclic aromatic linkers and/
or the replacement of carbon by nitrogen in (poly)aromatic rings
could be a strategy to increase the strength of the interaction with
hydrogen, e.g. due to a stronger dispersive interaction or by
means of the more heterogeneous charge distribution; some
experimental studies have investigated the hydrogen storage
properties of N-doped microporous carbon, with somewhat
inconclusive results;11,12 (iv) tailoring the network topology in
a way that unused pore space is decreased and the interaction
potentials of opposed or adjacent constituents of the pore walls
with hydrogen act in a synergetic manner, i.e. maximizing the
van-der-Waals-like attractive force.
The replacement of fossil fuels is a major challenge of our society:
not only due to their finiteness but also because of the emission of
greenhouse gases, mainly carbon dioxide, which cause the
so-called anthropogenic global warming effect. The transition to
the proposed ‘‘hydrogen-based economy’’ might be a solution.
However, there are several problems to be solved before such
a replacement of fossil fuels with hydrogen can take place; for
instance, the production of hydrogen by renewable energy
sources like solar energy is an absolutely necessary prerequisite.
Concerning the automotive sector, the safe and efficient storage
of hydrogen remains a bottleneck in the replacement of gasoline-
operated engines by fuel cells. In recent years, several classes of
materials with extremely high specific surface areas and pore
volumes have attracted much attention as possible storage
systems, which are based on physisorption of molecular
hydrogen, in particular metal–organic frameworks (MOFs).
aFraunhofer Institute for Material and Beam Technology IWS,
Winterbergstraße 28, D-01277 Dresden, Germany
bInstitute of Inorganic and Applied Chemistry, Department of Chemistry,
University of Hamburg, Martin-Luther-King-Platz 6, D-20146 Hamburg,
Germany. E-mail: Michael.Froeba@chemie.uni-hamburg.de; Fax: +49-
40-42838-6348; Tel: +49-40-42838-3137
† Electronic supplementary information (ESI) available: (1)
EDX-analyses of UHM-3–4; (2) magnified view of the P-XRD patterns
of UHM-2 and UHM-4 in the region 20–35ꢂ 2q; (3) P-XRD patterns of
activated UHM-2, UHM-4, and UHM-3–4; (4) thermal analyses of
UHM-2, UHM-4, and UHM-3–4; (5) determination of the isosteric
heat of adsorption values; (6) illustration of the mode of interactions of
the hydrogen molecule with the model linker systems; (7) CIF files of
UHM-2 and UHM-4; (Please note that these CIF files are based on
a homology modelling study following a Rietveld refinement. They are
not a result of a single-crystal structure analysis; accordingly they are
not submitted to the CCDC database). For ESI and CIF files see DOI:
10.1039/c2jm15523k
10294 | J. Mater. Chem., 2012, 22, 10294–10302
This journal is ª The Royal Society of Chemistry 2012

View Article Online
With respect to (i) and (iv) PCN-12 is worth mentioning, which
shows a hydrogen uptake of 3.05 wt% at 77 K and 1 bar. PCN-12
is one of the MOFs with the highest H2 uptake at these condi-
tions.13 It has been discussed that not only the high density of
open metal sites (adopting a quasi-cubic closest packing) is
responsible for this high uptake value but also that the non-
linearity of the linker (5,50-methylene diisophthalate; mdip),
which assumes two different conformations, is responsible for
the complex topology of the framework, resulting in the presence
of different shaped pores with favourable diameters between 0.7
and 1.3 nm.
dichlorodimethylgermane (ABCR), potassium permanganate
(Fluka, $99.0%), pyridine (Applichem, 99.0%), N,N0-dimethyl-
acetamide (DMA, Sigma-Aldrich, $99.5%), and dimethyl sulf-
oxide (DMSO, Sigma-Aldrich, $99.5%) were used as obtained,
while diethyl ether, tetrahydrofuran, and n-hexane were purified
and dried by standard operations before usage.
2.2. Synthesis of the linkers
1,1-Bis(3,5-dimethylphenyl)ethanol (1). The synthesis of (1) was
carried out under argon atmosphere. Within two hours 40.0 mL
(0.294 mol) 1-bromo-3,5-dimethylbenzene in 20 mL diethyl ether
and 40 mL tetrahydrofuran were added dropwise at ambient
temperature to 8.10 g (0.333 mol) magnesium turnings in 12 mL
diethyl ether and 12 mL tetrahydrofuran, while an exothermic
reaction started, which was accompanied by a colour change to
olive. To complete the reaction it was refluxed for 4 hours. After
cooling to room temperature hydrolysis with demineralized
water followed, which induced a colour change to red, and the
excess magnesium turnings were filtered. The layers were sepa-
rated and the organic one was washed three times with demin-
eralized water, dried with sodium sulfate and concentrated under
reduced pressure. The resulting yellow liquid was purified by
column chromatography (silica gel, 70–230 mesh, 6 n-hexane/1
ethyl acetate) to obtain 12.7 g (0.0499 mol; yield based on ethyl
acetate: 48%) of (1), a pale yellow oil. 1H-NMR (DMSO-d6, 400
MHz): d (ppm) 7.01 (s, 4H), 6.78 (s, 2H), 5.46 (s, 1H), 2.22 (s,
12H), 1.75 (s, 3H). 13C-NMR (DMSO-d6, 100 MHz): d (ppm)
149.3, 136.5, 127.4, 123.4, 74.0, 30.6, 21.0. IR (film, cmꢁ1): 3652,
3449, 2976, 2917, 2864, 1600, 1458, 1372, 1186, 1042, 998,
930, 851.
Subsequent to the publication reporting PCN-12, the iso-
structural MOF PCN-12-Si (UHM-3, University of Hamburg
Material) was synthesized, introducing the new organosilicon
linker 5,50-(dimethylsilanediyl) diisophthalate (dmsdip).14 Here,
we report on the syntheses of two further MOFs (UHM-2 and
UHM-4) being isoreticular to PCN-12 and comprising 5,50-
(propane-2,2-diyl) diisophthalate (dmcdip) and 5,50-(dime-
thylgermanediyl)diisophthalate (dmgdip), respectively, as
linkers, and dicopper paddle-wheel units as connectors.
Furthermore, isoreticular mixed-linker MOFs [UHM-3(x)–4(y)]
were synthesized, containing the two linkers dmsdip and dmgdip
in a ratio of x : y ¼ 75 : 25, 50 : 50, and 25 : 75, respectively.
The idea of the synthesis of these MOFs constituting
a homologous series was the systematic assessment of the central
atom’s (X) influence (X ¼ carbon in UHM-2, silicon in UHM-3,
and germanium in UHM-4) on the hydrogen uptake. In partic-
ular, it could potentially be expected that the increased polariz-
ability of the heavier central atoms leads to a stronger dispersive
interaction with adsorbed molecules. In addition, in comparison
to the linker of PCN-12 the two hydrogen atoms attached to the
central atom are replaced by methyl groups, which protrude into
the pores and may give rise to additional dispersive interaction
possibilities between these methyl groups with the hydrogen
adsorptive, although this effect is expected to be not very
pronounced; this could, in turn, depend on the bond length
X–CH3 (X ¼ C, Si, Ge).
5,50-(Propane-2,2-diyl)bis(1,3-dimethylbenzol) (2). 10.0
g
(0.0393 mol) of (1) in 80 mL toluene were added dropwise at
ambient temperature to a solution of 4.25 g trimethylaluminium
(0.0590 mol) in 200 mL toluene. After the generation of gas
stopped the reaction mixture was refluxed for 16 hours. To
remove excess trimethylaluminium it was hydrolyzed with
150 mL hydrochloric acid (1 M). The layers were separated, and
the aqueous one was washed three times with diethyl ether.
Afterwards the combined organic phases were washed with
saturated sodium hydrogen carbonate solution and dried with
sodium sulfate. The solvent was evaporated under reduced
pressure and a colorless solid was obtained, which was recrys-
tallized from n-hexane to yield 4.20 g (0.0166 mol; yield based on
(1): 42%) of (2). 1H-NMR (THF-d8, 400 MHz): d (ppm) 6.84 (s,
4H), 6.76 (s, 2H), 2.22 (s, 12H); 1.59 (s, 6H). 13C-NMR (THF-d8,
100 MHz): d (ppm) 149.3, 138.5, 129.9, 124.7, 47.9, 30.3, 21.8. IR
(KBr, cmꢁ1): 2961, 2914, 2866, 1598, 1464, 1443, 1373, 1039, 997,
932, 849.
All resulting compounds were characterized by means of
powder X-ray diffraction measurements, thermal gravimetric
analysis (TGA), N2 physisorption, and volumetric low pressure
hydrogen physisorption. Furthermore, SEM/EDX-analyses of
the mixed-linker MOF UHM-3(50)–4(50) are presented. In order
to complement the experimental work, DFT-D calculations were
carried out to evaluate the interaction of hydrogen with the C2v
conformers of three model systems as representatives of the
MOF linkers: dimethyldiphenylmethane (dmdpm), dimethyldi-
phenylsilane (dmdps), and dimethyldiphenylgermane (dmdpg).
The obtained results were compared with the experimental values
and critically discussed.
5,50-(Propane-2,2-diyl)diisophthalic
acid
(3).
4.20
g
2. Experimental section
(0.0166 mol) of (2) were dissolved in pyridine and water and
heated up to 95 ꢂC. Afterwards 50.5 g (0.320 mol) potassium
permanganate were added in several portions. To complete the
oxiꢂdation the reaction mixture was stirred for another 2 hours at
95 C. After cooling to ambient temperature the manganese(IV)
oxide was removed by filtration. The filtrate was treated with
charcoal and acidified with diluted hydrochloric acid to
2.1. Chemicals
Magnesium turnings (Aldrich, $99.5%), 1-bromo-3,5-dime-
thylbenzene
(Sigma-Aldrich,
97.0%),
ethyl
acetate
(Sigma-Aldrich, $99.5%), toluene (Aldrich, $99.7%), and tri-
methylaluminium in toluene (2.0 M, Aldrich) were used without
further
purification.
Lithium
(Fluka,
$99.0%),
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 10294–10302 | 10295

View Article Online
precipitate the raw product (3), which was achieved by filtration.
To purify the acid, it was dissolved in saturated aqueous sodium
carbonate solution, acidified again, filtered and dried in
a vacuum to yield 4.39 g (0.0118 mol, yield: 71% based on (2)) of
lined autoclave and heated up to 85 ꢂC with a heating rate of
1 ꢂC minꢁ1. The temperature was held for 46 hours before cooling
to room temperature with a rate of 5 ꢂC minꢁ1. The resulting blue
powder was collected by filtration, washed several times with
15 mL solvent, and air dried.
1
(3). H-NMR (DMSO-d6, 400 MHz): d (ppm) ¼ 13.35 (s, 4H);
8.34 (t, 4J ¼ 1.49 Hz, 2H); 8.00 (d, 4H, 4J ¼ 1.50 Hz); 1.75 (s, 6H).
13C-NMR (DMSO-d6, 100 MHz): d (ppm) ¼ 166.7; 150.6; 131.5;
131.4; 128.0; 42.9; 30.1. IR (KBr, cmꢁ1): 3450, 2968, 2654, 1700,
1601, 1453, 1409, 1367, 1259, 1153, 931, 828.
2.4. Methods
NMR spectra were acquired using a Bruker AVANCE 400 or
a Varian Gemini-2000BB spectrometer.
Dimethyl(bis-3,5-dimethylphenyl)germane (4). The synthesis of
(4) was carried out under argon atmosphere. In a typical
synthesis 26.7 g (0.144 mol) 1-bromo-3,5-dimethylbenzene were
dissolved in 10 mL diethyl ether and added dropwise to 2.0ꢂ6 g
(0.297 mol) lithium granulate in 50 mL diethyl ether at 10 C.
After stirring for 2.5 hours the reaction mixture was cooled to
0 ꢂC and 10.0 g (0.0576 mol) dichlorodimethylgermane in 10 mL
diethyl ether were added dropwise. Afterwards, the reaction
mixture was stirred for 12 hours at room temperature. To remove
the excess lithium, it was filtered and ice water was added to the
solution. The diethyl ether layer was separated, treated with
water, hydrochloric acid (5%), sodium hydroxide solution (5%),
dried over sodium sulfate, and finally the solvent was evaporated
to give 16.3 g (0.0520 mol, yield: 90% based on dichlor-
odimethylgermane) of (4), a pale yellow solid. 1H-NMR (CDCl3,
400 MHz): d (ppm) ¼ 7.09 (s, 4H); 6.98 (s, 2H); 2.30 (s, 12H); 0.60
(s, 6H). 13C-NMR (CDCl3, 100 MHz): d (ppm) ¼ 140.1; 137.4;
131.4; 130.5; 21.5; ꢁ2.9. IR (KBr, cmꢁ1): 2967, 2905, 2854, 1592,
1576, 1452, 1382, 1266, 1235, 1169, 1150, 1036, 989, 844.
Infrared spectra were acquired using a Bruker Vertex 70 FT-
IR spectrometer.
Powder X-ray diffraction (P-XRD) patterns were recorded at
room temperature with a STOE STADI P transmission powder
diffractometer using Cu Ka radiation (40 kV, 30 mA; counting
time: 70 s; step width: 0.1ꢂ (2q)).
In addition, the mixed-linker MOF UHM-3(50)–4(50) was
characterized by high resolution scanning electron microscopy
(HRSEM, Leo Gemini 982) and by energy-dispersive X-ray
(EDX) analysis at the Justus Liebig University, Gießen.
Thermal analysis (thermogravimetry (TG)/mass spectrometry
(MS)) was conducted under O2/argon (20/80) flow (20 mL minꢁ1
)
with a NETZSCH STA 409 thermobalance coupled by capillary
with a Balzers QMG 421 mass spectrometer (ESI†). The heating
rate was 2 K minꢁ1
.
Prior to physisorption measurements all samples were acti-
vated at elevated temperatures for 24 hours in a vacuum.
Depending on the material the temperature was chosen between
150 and 170 ꢂC. During the activation process, all samples
changed their color from turquoise to blue and purple, respec-
tively. Nitrogen physisorption data were recorded with a Quan-
tachrome QUADRASORB-SI-MP at 77 K. The specific surface
area was calculated from the adsorption branch in the relative
pressure interval from 0.01–0.05 using the BET method. The
total pore volume was estimated from the quantity of gas
adsorbed at a relative pressure of 0.18 or 0.20. Volumetric
hydrogen physisorption data were recorded at 77 K, 87 K, and 97
K (ESI†) on a Quantachrome Autosorb 1-MP (purity of He and
H2: 99.9990%). Isosteric heats of adsorption were calculated
from these isotherms using the software QuadraWin.
5,50-(Dimethylgermanediyl)diisophthalic acid (5). Potassium
permanganate (50.5 g, 0.320 mol) was added in several portions
to a solution of 4.20 g (0.0166 mol) (4) in pyridine and water at
95 ꢂC. To complete the oxidation the reaction mixture was stirred
ꢂ
for another 2 hours at 95 C. Afterwards, manganese(IV) oxide
was removed by filtration. The filtrate was treated with charcoal
and acidified with diluted hydrochloric acid to precipitate the
raw product (5), which was achieved by filtration. To purify the
acid, it was dissolved in saturated aqueous sodium carbonate
solution, acidified again, filtered and dried in vacuum to yield
4.39 g (0.0118 mol, yield: 71% based on (4)) of (5), a colorless
solid. 1H-NMR (DMSO-d6, 200 MHz): d (ppm) ¼ 13.33 (s, 2H;15
8.46 (t, 2H, 4J ¼ 1.67 Hz); 8.22 (d, 4H, 4J ¼ 1.67 Hz); 0.87 (s, 6H).
13C-NMR (DMSO-d6, 100 MHz): d (ppm) ¼ 166.7; 140.6 (C-8,
C-9); 138.0; 131.0; ꢁ3.5; 130.4. IR (KBr, cmꢁ1): 3506, 3428, 3027,
2637, 2538, 1694, 1598, 1431, 1267, 1160, 1124, 808, 749.
2.5. Computational details
All-electron DFT calculations including an empirical (‘‘Grimme-
type’’) dispersion correction were carried out with the ADF
code.16 The geometries of the subsystems (hydrogen, organic
molecule) were optimized using the BLYP-D functional16–18 and
a triple-zeta basis set with polarization functions (TZ2P). To
obtain highly accurate energies, single-point calculations were
carried out for the optimized subsystems using the B3LYP-D
functional18,19 and a quadruple-zeta basis set (QZ4P). After
combining the two fragments in the desired configuration,
a potential energy curve was obtained from single-point calcu-
lations for varying distances. The ADF package directly uses the
results of the two isolated fragments to calculate the binding
energy Eint for the combined system. In a preliminary study, it
was found that the B3LYP-D functional provides for excellent
agreement of the calculated interaction energies with
5,50-(Dimethylsilanediyl)diisophthalic acid (6). The tetra-
carboxylic acid (6) was prepared as described in the literature.14
2.3. Synthesis of the MOFs
In a typical synthesis, 259 mg (1.07 mmol) of copper(II) nitrate
trihydrate and 0.291 mmol of the corresponding tetracarboxylic
acids ((3) for UHM-2; (5) for UHM-4) or a mixture of (5) and (6)
with molar ratios of 75 : 25, 50 : 50, 25 : 75 for
UHM-3(75)–4(25), UHM-3(50)–4(50), UHM-3(25)–4(75) were
dissolved in 15 mL DMSO (for UHM-2) or DMA (for UHM-4
and UHM-3–4). The resulting mixture was placed in a Teflon
10296 | J. Mater. Chem., 2012, 22, 10294–10302
This journal is ª The Royal Society of Chemistry 2012

View Article Online
high-quality ab initio data for the case of hydrogen interacting
with benzene and naphthalene.20
obtained by using molecular modelling techniques in conjunction
with the comparison to experimental powder X-ray diffraction
data. The structures of the compounds UHM-2 and UHM-4
were modelled in analogy to the procedure already described for
UHM-3, i.e. by applying a sort of ‘homology modelling’, using
the structure of the original PCN-12 (CCDC entry no. 662918) as
a starting point. The geometry optimization (including optimi-
zation of the cells) was carried out with the Universal Force Field
(UFF)21 as implemented in the Forcite tool of the Material Studio
v4.4 package.22 This energy minimization procedure resulted in
3. Results and discussion
3.1. Synthesis
The new non-linear linker 5,50-(propane-2,2-diyl)diisophthalic
acid (3) was synthesized in three steps, starting with a Grignard
reaction of 1-bromo-3,5-dimethylbenzene and ethyl acetate,
followed by the conversion of the hydroxy groups by
trimethylaluminium in methyl groups and subsequent oxidation
with potassium permanganate to yield the tetracarboxylic acid as
represented in Scheme 1. The overall yield was 14%.
plausible tetragonal structures with the space group P4/mmm,
3
ꢀ
ꢀ
ꢀ
no. 123 (a ¼ b ¼ 33.0748 A, c ¼ 22.6974 A, V ¼ 24 830 A for
3
ꢀ
ꢀ
ꢀ
UHM-2 and a ¼ b ¼ 33.9424 A, c ¼ 23.4374 A, V ¼ 27 002 A for
UHM-4). These two MOFs, which are isostructural to PCN-12
and UHM-3, respectively, are composed of altogether 24
dicopper paddle wheel units and 24 linker molecules per unit cell,
leading to the composition Cu48C456O240H288 for UHM-2 and
Cu48Ge48C424O240H288 for UHM-4, both with an additional but
undefined amount of solvent molecules (DMSO for UHM-2 and
DMA for UHM-4). In Fig. 1(a) and (b) the experimental and
simulated powder X-ray diffraction (P-XRD) patterns are
compared. Apart from the fact that the experimental patterns are
slightly shifted towards lower angles they show an excellent
agreement. Therefore, it can be confidently assumed that the
structure models (see Fig. 2) are correct. The deviations above 20ꢂ
2q—slightly more pronounced for UHM-4—are mainly related to
the background, the noise and, in particular, limited resolution of
Scheme 1 Synthesis path of the new linker 5,50-(propane-2,2-diyl)di-
isophthalic acid (3).
The synthesis of the new non-linear linker 5,50-(dime-
thylgermanediyl)diisophthalic acid (5) (Scheme 2) could be
accomplished in a two-step reaction. The lithiation of 1-bromo-
3,5-dimethylbenzene and subsequent reaction with the appro-
priate germane were followed by the oxidation of the aromatic
methyl groups. The overall yield was 50%.
Scheme 2 Synthesis path of the new linker 5,50-(dimethylsilanediyl)di-
isophthalic acid (5).
The corresponding MOFs UHM-2, UHM-4, and UHM-3–4,
respectively, could be obtained by a solvothermal synthesis
procedure with copper(II) nitrate trihydrate and the carboxylic
acids (3), (5), and (6) in DMA or DMSO. For all compounds
solvent removal at elevated temperatures led to permanent
porosity.
3.2. Powder X-ray diffraction and structure modelling of
UHM-2 and UHM-4
The crystal sizes of the synthesized compounds UHM-2 and
UHM-4 were too small in order to carry out single-crystal X-ray
diffraction studies. However, plausible structures could be
Fig. 1 Comparison of the experimental (black) and simulated (gray)
P-XRD patterns of UHM-2 (a) and UHM-4 (b).
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 10294–10302 | 10297

View Article Online
UHM-3. Therefore, we decided to investigate the relative
stability and structural features of a series of mixed-linker MOFs
using (5) and (6) as linkers in different ratios (75 : 25, 50 : 50 and
25 : 75). Concerning multicomponent condensation reactions
with components which differ in structure and/or reactivity,
issues of their homogeneous incorporation into the final material
arise. In principle, three different scenarios are conceivable: (i)
the two linkers exhibit strongly different reactivities resulting in
MOFs in which only one of the two linkers is incorporated while
the other remains unreacted in the solution; (ii) the reactivities
are comparable, but homocondensation is preferred giving rise to
a physical mixture of particles which comprise in each case only
one of the linkers; (iii) both linkers will be incorporated in
amounts according to their specified composition of the reaction
mixture, although this could lead to a certain amount of struc-
tural stress due to the different lengths of the two linkers.
In Fig. 3 the P-XRD patterns of the pure (as-synthesized)
phases UHM-3 and UHM-4 from 0 to 20ꢂ 2q are shown as well as
the mixed-linker materials obtained from chemical reaction
mixtures containing the linkers (5) and (6) in the respective molar
ratios of 75 : 25, 50 : 50, and 25 : 75. For comparison reasons the
physical mixtures UHM-3/4 with the same molar ratios of Si/Ge
are also shown. All diffractograms are very similar and support
the assumption that in all cases phases with an isoreticular
structure to PCN-12 were obtained. A small shift of some
reflections (in particular the (001) and (334) reflection) towards
higher d values with increasing germanium content is detectable.
In order to prove that the mixed-linker MOFs are indeed
homogeneous phases EDX analyses were carried out (ESI†). For
each composition five different crystallites were investigated. The
determined Si/Ge ratios were found to be almost equal to the
ratio of the respective linker molecules (5) and (6), which was
specified during the syntheses. The mean values were 75 : 25 for
UHM-3(75)–4(25), 54 : 46 for UHM-3(50)–4(50), and 27 : 73 for
Fig. 2 Crystal structure of UHM-4; 1 ꢃ 1 ꢃ 2 supercell displayed along
the b axis (top), unit cell displayed along the c axis (bottom); gray:
carbon, white: hydrogen, red: oxygen, blue: copper, and green: germa-
nium. UHM-2 is isostructural to UHM-4.
the experimental laboratory X-ray data (for a magnified view of
the XRD patterns between 20 and 35ꢂ 2q, see ESI†).23
Fig. 3 Powder X-ray diffractograms of the pure phases UHM-3
and UHM-4 (black), the mixed-linker MOFs UHM-3(75)–4(25),
UHM-3(50)–4(50), UHM-3(25)–4(75) (green) as well as the correspond-
ing physical mixtures of UHM-3 and UHM-4 with the same molar ratios
of Si : Ge (orange). (The dashed lines are guide to the eyes to facilitate
identification of reflection shifts.)
3.3. Mixed-linker MOFs UHM-3(75)–4(25), UHM-3(50)–
4(50), and UHM-3(25)–4(75)
Activation studies revealed that UHM-4 is not completely stable
upon solvent removal (see below) in contrast to UHM-2 and
10298 | J. Mater. Chem., 2012, 22, 10294–10302
This journal is ª The Royal Society of Chemistry 2012

View Article Online
UHM-3(25)–4(75), verifying the presence of chemically homo-
geneous phases. For the following studies only the mixed-linker
MOF UHM-3(50)–4(50) was considered.
This requires, in turn, higher activation temperatures (170
instead of 150 ꢂC), which probably have a negative impact on the
surface area. The thermal analysis (Fig. S4a†) shows that DMSO
ꢂ
is still detected up to ꢀ340 C indicating a delayed desolvation,
which slightly overlaps with the thermal decomposition of the
material. UHM-4 was synthesized with DMA (as UHM-3).
However, UHM-4 is less thermally stable than UHM-3.
Presumably, the activation procedure leads to considerable
disintegration processes of the material, resulting in the lowest
specific surface area value within this series.
3.4. Activation studies, N2 physisorption and H2 uptake
measurements
In order to get a data set as comprehensive and consistent as
possible the already known MOF UHM-3 (ref. 14) was included
in the following studies.
Volumetric low pressure H2 adsorption measurements at 77 K
reveal values for the hydrogen uptake of 2.28 wt% at 1 bar for
UHM-2, 2.14 and 2.57 wt% for UHM-3 (ref. 24) and 2.62 wt%
for UHM-4 (Fig. 5 (top)). In particular the high value for
UHM-4 is amongst the highest H2 uptakes for MOFs at this
pressure. However, the value of PCN-12 is not reached or even
outperformed. This can be explained by the fact that thermal
activation for all materials in this study did not lead to complete
removal of the solvent as shown by thermal analyses (ESI†). The
heat of adsorption values in dependence on the coverage rate
(ranging from 0.05 to 1.0 wt%) are presented in Fig. 5 (bottom).
All materials show relatively high initial values around
7–8 kJ molꢁ1 at low coverages and only a very moderate, almost
linear decrease with higher loadings, still possessing values
around 6.5 kJ molꢁ1 at 1.0 wt%. This behavior is comparable
with that of Cu3(btc)2.25 In Table 1 all relevant sorption data are
synoptically compared.
Since thermal activation showed the best results regarding
porosity of the compounds, the following physisorption studies
were made after activation in a vacuum at elevated temperatures
(170 ꢂC for UHM-2 and 150 ꢂC for UHM-3 and UHM-4).
Solvent exchange and subsequent activation at moderate
temperatures led to lower values for the specific surface area or to
complete destruction of the structure. In contrast to UHM-2 it
was not possible to activate UHM-4 without the loss of crystal-
linity (ESI, Fig. S3(a) and (b)†). However, complete collapse of
the network is highly improbable since permanent porosity was
observed after solvent removal. Thermal activation at lower
temperatures led neither to preservation of crystallinity nor to
higher values for the specific surface area. Thermal activation of
UHM-3(50)–4(50) was conducted at 150 ꢂC and showed a higher
stability of the material upon solvent removal in comparison to
UHM 4 (ESI, Fig. S3(c)†).
UHM-2, -3, -4 and UHM-3(50)–4(50) exhibit type I N2
isotherms (Fig. 4), which are characteristic of microporous
materials. Analyses of the isotherms reveal values for the specific
Over the past several years, a number of qualitative and semi-
quantitative ‘‘structure–activity’’ relationships for H2 uptake in
MOFs, carbon-based materials, zeolites and even microporous
polymers (at 77 K) have been derived, both from simulations as
well as experiments. It is well documented that there is a roughly
linear correlation between either the specific surface area or total
pore volume and the H2 saturation uptake.5,26–28 Of course, these
relationships are only valid if almost completely activated
samples are considered; sample preparation and activation is
playing a key role and has been identified, for instance, as the
source for different reported H2 uptake values of Cu3(btc)2.29 In
contrast, at low pressures, i.e. low coverage rates, the phys-
isorbed amount of H2 is mainly determined by the isosteric heat
of adsorption.30 The isosteric heat of adsorption in turn is not
only directly correlated with the interaction strength of atoms
showing high affinity towards hydrogen like unsaturated metal
centers but is also indirectly dependent on the pore geometry/
size: the heat of adsorption increases for materials possessing
smaller pores.31,32
surface area of SBET ¼ 1692, 2430, 1360 and 990 m2 gꢁ1
,
respectively (see also Table 1), using the adsorption branch in the
relative pressure interval from 0.01–0.05. The materials exhibit
micropore volumes of Vpore ¼ 0.66 (d < 2.0 nm; calculated at
p/p0 ¼ 0.18), 0.97 (d < 2.1 nm; calculated at p/p0 ¼ 0.20), 0.54
(d < 2.1 nm; calculated at p/p0 ¼ 0.20), and 0.39 cm3 gꢁ1 (d <
2.1 nm; calculated at p/p0 ¼ 0.20), respectively. It is interesting to
note that the values of specific surface areas and pore volumes of
the series UHM-2, -3, and -4 show such significant differences,
although they are isostructural. A possible explanation for the
lower value of UHM-2 compared to UHM-3 could be the fact
that it was synthesized with DMSO instead of DMA.
Inspecting Table 1 and ignoring in the first instance the heat of
adsorption value for UHM-3(50)–4(50) there is a clear trend
showing increasing sorption heats in the order X ¼ C < X ¼ Si <
X ¼ Ge. Considering only the best activated samples the
maximal achievable H2 uptake values confirm this trend (2.28,
2.57, and 2.62 wt%); this trend is even more pronounced when
taking the higher molecular weights of the central atom of the
linkers into account. These results would allow drawing a posi-
tive correlation of the central atom’s polarizability of the linker
with the heat of adsorption, i.e. affinity towards hydrogen, if this
would be the only feature that changed in the series UHM-2–4.
However, one should bear in mind that also some geometrical
features (bond lengths (X–C) and bond angles (C–X–C) are
Fig. 4 Nitrogen physisorption isotherms (,, adsorption; B, desorp-
tion) of activated UHM-2 (blue), UHM-3 (black), UHM-4 (green), and
UHM-3(50)–4(50) (orange) respectively.
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 10294–10302 | 10299

View Article Online
Table 1 Summary of the physisorption properties of UHM-2, -3, -4, and
-3(50)–4(50) concerning the specific inner surface area (SBET), H2 uptake
(at 77 K and 1 bar) and isosteric heat of adsorption at a coverage of
0.05 wt%
for UHM-2, -3, and -4; (ii) as the qst vs. coverage curves do not
decrease significantly after these theoretical values are reached
the materials seem to provide beneficial pore geometries/sizes
leading to overlapping of interaction potentials from two or
more opposed pore walls; (iii) however, the materials differ in
stability and easiness of activation: the comparably lower uptake
capacity at 1 bar of UHM-3(50)–4(50) may probably be justified
with its considerably lesser degree of activation and lower degree
of crystalline order, meaning that probably partial framework
disintegration had already occurred; (iv) this is also reflected in
the values for the specific surface areas, which neither correlate
with the qst values nor the uptake capacities at 1 bar but at least
to a certain amount to the degree of crystallinity as visible in the
P-XRD patterns (ESI†).
SBET
/
wt% H2
(77 K, 1 bar)
qst (0.05 wt%)
(kJ molꢁ1
)
m2 gꢁ1
UHM-2
UHM-3
UHM-4
UHM-3(50)–4(50)
1692
2.28
2.14/2.57 (ref. 24)
2.62
1.88
7.6
7.9
8.1
7.9
2543/2430
1360
990
altered. Whether or not and in which way these changes could be
responsible for the observed order regarding the qst values will be
clarified with the help of DFT calculations (see the following
section).
3.5. DFT-D calculations
The H2 uptake capacity of UHM-3(50)–4(50) (merely 1.88 wt%)
builds an exception in this series: as this sample is composed of Si
and Ge as the central atom of the linker one would expect
capacities which lie between the values of UHM-3 and UHM-4,
respectively. This seems even more astonishing since the qst vs.
coverage plots are almost identical (Fig. 5(b)) as well as
the uptake curves (Fig. 5(a)) up to 1.0 wt% are very similar (p z
0.12 bar). Only at higher pressures the uptake curve of
UHM-3(50)–4(50) lies considerably below the curves of UHM-2,
UHM-3, and UHM-4. This allows the following tentative
assumptions: (i) concerning the ‘‘filling sequence’’, it is likely that
the OMS of the copper paddle-wheel units were occupied first
giving rise to relatively high qst values; theoretically the full
occupation of the OMS (assuming exactly one H2 molecule per
Cu atom) would account for 0.80, 0.78 and 0.72 wt% H2 uptake
Recent computational studies of the adsorption of H2, CO2, and
other small molecules in microporous materials have shown that
DFT methods using an empirical dispersion correction can
provide interesting insights into the interactions between the
framework and adsorbed molecules.33–35 To complement the
experimental work, DFT-D calculations were carried out in
order to evaluate the interaction of hydrogen with the C2v
conformers of three model systems as representatives of the
MOF linkers: dimethyldiphenylmethane (dmdpm, more
correctly termed 2,2-diphenylpropane), dimethyldiphenylsilane
(dmdps), and dimethyldiphenylgermane (dmdpg). Exemplarily,
dmdps is shown in Fig. 6, together with some key geometric
properties of the three molecules. Because a complete screening
of the potential energy surface is beyond the scope of this study,
only an approach of the hydrogen molecule along z (axis of
rotation) from either side of the C2v conformer was considered
(see Fig. S6†). Sets of calculations were carried out for orienta-
tions of the H2 molecule parallel to the three axes of the coor-
dinate system, varying the distance to the central atom X (X ¼ C,
ꢀ
Si, Ge) from 3 to 6 A.
For the approach from the side of the methyl groups, the
resulting potential energy curves are shown in Fig. 7(a) and (b).
An increase of the interaction energy on the increasing atomic
number of X is obvious for all three orientations. In all cases, the
calculated interaction energies are low, ranging between ꢁ0.75
and ꢁ2.2 kJ molꢁ1. The orientation parallel to the z-axis is the
Fig. 5 Hydrogen adsorption isotherms (top) and isosteric heat of
adsorption (bottom) of activated UHM-2 (B, blue), UHM-3 (>, black),
UHM-4 (,, green), and UHM-3(50)–4(50) (O, orange).
Fig. 6 Visualization of C2v conformer of dimethyldiphenylsilane,
together with the reference coordinate system. On the right, some key
geometric properties of the three molecules considered are given.
10300 | J. Mater. Chem., 2012, 22, 10294–10302
This journal is ª The Royal Society of Chemistry 2012

View Article Online
Fig. 7 DFT-D interaction energy curves for H2 interacting with C2v conformers of dimethyl-diphenylmethane/silane/germane, assuming an approach
of the hydrogen molecule along the z-axis from the side of the methyl groups (a and b) and along the z-axis from the side of the phenyl groups (c and d).
An illustration of the mode of interactions can be found in the ESI (Fig. S6†).
energetically favoured orientation, because the steric repulsion
caused by the methyl hydrogens is minimized in this case. Due to
the same steric effects, the orientation parallel to the x-axis is
energetically more favourable than the orientation parallel to the
y-axis. The interaction energy increases with the increasing
atomic number of X, as it can be expected from the increased
contribution of dispersive interactions with the central atom.
However, it is noteworthy that the equilibrium distance decreases
with increasing atomic number, despite the increase of the van
der Waals radius of X. This evolution can be understood when it
is considered that the increase of the X–Cme bond length (Fig. 6)
causes a reduction of the steric repulsion, thereby enhancing the
accessibility of the central atom.
note that the resulting energy curves for the silicon and germa-
nium system are virtually identical. For X ¼ C, the interaction is
considerably stronger and the equilibrium distance is somewhat
shorter. This behaviour is contrary to the expectations based on
the dispersive interaction of the hydrogen molecule with X.
Instead, the geometric arrangement of the moieties around the
central atoms plays a key role: The shorter X–Cph distances, as
well as the slightly smaller Cph–X–Cph angle, cause an increased
overlap of the interactions of the H2 molecule with the two
phenyl rings, and thus lead to a deeper energy minimum.
Although it needs to be emphasized that the results presented
here are only showcase examples that cannot replace a full
screening of the potential energy surface, the two cases consid-
ered permit some conclusions on the impact of the central atom
of the non-linear linker molecule. While the results for the
approach from the side of the methyl groups show the expected
tendency of increasing interaction energy with increasing atomic
number of the central atom, the results for an approach from the
other side exhibit the opposite trend. In this instance, the
geometric differences between the organic molecules, particularly
the varying bond distances and angles around the central atom,
have a more important impact on the total interaction energy
than dispersive contributions stemming from the central atom.
Therefore, the results do not provide any evidence for the
assumption that the introduction of a few atoms with high
polarizability will have a significant impact on the hydrogen
storage properties. On the other hand, their presence could be of
The DFT-D results for the approach from the side of the
phenyl rings are shown in Fig. 7(c) and (d). Here, the orientations
parallel to the y- and z-axis show very similar interaction ener-
gies, ranging near ꢁ5.6 kJ molꢁ1 for dmdpm, and between ꢁ4.1
and ꢁ4.4 kJ molꢁ1 for the silicon and germanium analogues. The
interaction energy for the orientation parallel to the x-axis is
considerably weaker. For the case of simple aromatic systems, it
could be shown by several authors36–38 that the orientation of the
H2 molecular axis perpendicular to the benzene ring is most
favourable due to electrostatic interactions. Similarly, the
orientation parallel to the y-axis is favoured over H2kx in this
case, because the electrostatic contribution is maximized when
the positively polarized ends of the H2 dumbbell protrude into
the negative charge clouds above the rings. It is interesting to
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 10294–10302 | 10301

View Article Online
7 J. G. Vitillo, L. Regli, S. Chavan, G. Ricchiardi, G. Spoto,
P. D. C. Dietzel, S. Bordiga and A. J. Zecchina, J. Am. Chem. Soc.,
2008, 130, 8386.
8 S. K. Bhatia and A. L. Myers, Langmuir, 2006, 22, 1688.
9 S. S. Han and W. A. Goddard, III, J. Am. Chem. Soc., 2007, 129,
8422.
greater use for the storage of more strongly polarizable gases or
for separation applications. For example, it has been shown
recently that the introduction of ‘‘soft’’ cations such as Sb3+ can
enhance the CO2/H2 separation efficiency in aerogels.39
10 D. Himsl, D. Wallacher and M. Hartmann, Angew. Chem., Int. Ed.,
2009, 48, 4639.
4. Conclusions
11 L. Wang and R. T. Yang, J. Phys. Chem. C, 2009, 113, 21883.
12 Y. Xia, G. S. Walker, D. M. Grant and R. J. Mokaya, J. Am. Chem.
Soc., 2009, 131, 16493.
With UHM-2, UHM-4 and UHM-3(50)–4(50) a series of new
isoreticular MOFs is presented, which are composed of dicopper
paddle wheel units and the tetracarboxylic acids 5,50-(propane-
2,2-diyl)diisophthalic acid, dimethyl(bis-3,5-dimethylphenyl)
silane and dimethyl(bis-3,5-dimethylphenyl)germane as linker
molecules. All materials exhibit permanent porosity after acti-
vation. The isoreticular MOF series UHM-2, -3, and -4, in which
the central atom X of the linker is varied (X ¼ C, Si, and Ge)
allowed for a systematic study concerning the influence of the
incorporation of atoms with higher polarizability on the
hydrogen uptake. The hydrogen uptake at 77 K and 1 bar as well
as the isosteric heat of adsorption values slightly increase in the
order X ¼ C < X ¼ Si < X ¼ Ge. However, the effect is not very
pronounced and further investigations by DFT calculations
reveal that the influence of dispersive interactions between the
hydrogen molecule and the central atom is relatively marginal
with respect to the effect of geometric changes, which are
introduced with the interchange of the central atom. The
hypothesis that there is a positive correlation between the
polarizability of linker atoms on the one hand and the hydrogen
uptake on the other could not be verified. Taking the results of
the thermal analyses into account, which indicate incomplete
solvent removal through thermal treatment, the influence of the
activation procedure seems to have a much greater effect than the
chemical nature of the linker. This is also reflected in the values
for the specific surface area, which are strongly dependent on the
overall sample quality. This confirms once again that (a) repro-
ducible syntheses routes—leading to samples of comparable
quality, and (b) efficient activation procedures are key issues for
developing improved hydrogen storage materials based on
MOFs.
13 X.-S. Wang, M. Shengqian, P. M. Forster, Y. Daqiang, J. Eckert,
ꢁ
J. J. Lopez, B. J. Murphy, J. B. Parise and H.-C. Zhou, Angew.
Chem., Int. Ed., 2008, 47, 7263.
€
14 S. E. Wenzel, M. Fischer, F. Hoffmann and M. Froba, Inorg. Chem.,
2009, 48, 6559.
15 Due to H–D-exchange between DMSO-d6 the intensity of this signal
is too less (it should correspond to four protons). Complete oxidation
is proved by the total absence of the signal at 2.22 ppm which
correlates to the methyl groups of the educt.
16 G. te Velde, F. M. Bickelhaupt, E. J. Baerends, C. Fonseca Guerra,
S. J. A. van Gisbergen, J. G. Snijders and T. Ziegler, J. Comput.
Chem., 2001, 22, 931.
17 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B, 1988, 37, 785.
18 S. Grimme, J. Comput. Chem., 2006, 27, 1787.
19 P. J. Stephens, F. J. Devlin, C. F. Chabalowski and M. J. Frisch,
J. Phys. Chem., 1994, 98, 11623.
20 M. Fischer, PhD thesis, University of Hamburg, 2011, http://
ediss.sub.uni-hamburg.de/volltexte/2011/5342/.
ꢁ
21 A. K. Rappe, C. J. Casewit, K. S. Colwell, W. A. Goddard, III and
W. M. Skiff, J. Am. Chem. Soc., 1992, 114, 10024.
22 Materials Studio, v4.4, Accelrys Inc., San Diego, CA, 2010.
23 Restricted Rietveld refinements (scaling factor, zero point shift, lattice
parameters) were carried out, too, for both UHM-2 and UHM-4
structural models. The lattice constants changed only insignificantly
ꢀ
ꢀ
ꢀ
ꢀ
to the following values: a ¼ b ¼ 32.9748 A, c ¼ 22.5110 A, V ¼
3
ꢀ
24 482 A for UHM-2 and a ¼ b ¼ 34.2146 A, c ¼ 23.7709 A, V ¼
3
27 827 A for UHM-4.
ꢀ
24 The value of 2.14 wt% was obtained in this work, while a value of 2.57
wt% was measured in the original work as reported in ref. 14. This
difference reflects some difficulties to gain samples which are
activated to the same amount. However, this study focused mainly
on the systematic evaluation of the interaction of hydrogen with the
linker, therefore the heat of adsorption values, in particular at low
loadings, are of much greater relevance.
25 J. L. C. Rowsell and O. M. Yaghi, J. Am. Chem. Soc., 2006, 128, 1304.
26 A. G. Wong-Foy, A. J. Matzger and O. M. Yaghi, J. Am. Chem. Soc.,
2006, 128, 3494.
27 M. Hirscher and B. Panella, Scr. Mater., 2007, 56, 809.
28 K. M. Thomas, Dalton Trans., 2009, 1487.
Acknowledgements
29 B. Xiao, P. S. Wheatley, X. Zhao, A. J. Fletcher, S. Fox, A. G. Rossi,
I. L. Megson, S. Bordiga, L. Regli, K. M. Thomas and
R. E. J. Morris, J. Am. Chem. Soc., 2007, 129, 1203.
The authors thank Rabea Dippel, Justus Liebig University,
Gießen, for carrying out the EDX analysis and Katharina Pei-
kert for support in the laboratory.
€
30 H. Frost, T. Duren and R. Q. Snurr, J. Phys. Chem. B, 2006, 110,
9565.
31 B. Schmitz, U. Muller, N. Trukhan, M. Schubert, G. Ferey and
€
M. Hirscher, ChemPhysChem, 2008, 9, 2181.
ꢁ
€
€
32 B. Schmitz, I. Krkljus, E. Leung, H. W. Hoffken, U. Muller and
M. Hirscher, ChemSusChem, 2010, 3, 758.
33 K. Sillar, A. Hofmann and J. Sauer, J. Am. Chem. Soc., 2009, 131,
4143.
Notes and references
1 J. Sculley, D. Yuan and H. C. Zhou, Energy Environ. Sci., 2011, 4,
2721.
2 H. Furukawa, N. Ko, Y. B. Go, N. Aratani, S. B. Choi, E. Choi,
34 X. Solans-Monfort, M. Sodupe and J. Eckert, J. Phys. Chem. C, 2010,
114, 13926.
35 L. Valenzano, B. Civalleri, S. Chavan, G. T. Palomino, C. O. Arean
and S. Bordiga, J. Phys. Chem. C, 2010, 114, 11185.
36 T. Heine, L. Zhechkov and G. Seifert, Phys. Chem. Chem. Phys.,
2004, 6, 980.
€
A. O. Yazaydin, R. Q. Snurr, M. O’Keeffe, J. Kim and
O. M. Yaghi, Science, 2010, 329, 424.
ꢁ
€
3 O. K. Farha, A. O. Yazaydin, I. Eryazici, C. D. Malliakas,
B. G. Hauser, M. G. Kanatzidis, S. T. Nguyen, R. Q. Snurr and
J. T. Hupp, Nat. Chem., 2010, 2, 944.
37 A. G. Donchev, J. Chem. Phys., 2007, 126, 124706.
38 S. Figueroa-Gerstenmaier, S. Giudice, L. Cavallo and G. Milano,
Phys. Chem. Chem. Phys., 2009, 11, 3935.
4 M. P. Suh, H. J. Park, T. K. Prasad and D.-W. Lim, Chem. Rev., 2012,
112, 782.
5 D. J. Collins and H.-C. Zhou, J. Mater. Chem., 2007, 17, 3154.
6 D. Zhao, D. Yuan and H. C. Zhou, Energy Environ. Sci., 2008, 1, 222.
39 S. Bag and M. G. Kanatzidis, J. Am. Chem. Soc., 2010, 132, 14951.
10302 | J. Mater. Chem., 2012, 22, 10294–10302
This journal is ª The Royal Society of Chemistry 2012