Elucidating gating effects for hydrogen sorption in MFU-4-type triazolate-based metal-organic frameworks featuring different pore sizes

Article abstract of DOI:10.1002/chem.201001872

A highly porous member of isoreticular MFU-4-type frameworks, [Zn ₅Cl₄(BTDD)₃] (MFU-4l(arge)) (H ₂-BTDD=bis(1H-1,2,3-triazolo[4,5-b],[4′,5′-i])dibenzo[1, 4]dioxin), has been synthesized using ZnCl₂ and H₂-BTDD in N,N-dimethylformamide as a solvent. MFU-4l represents the first example of MFU-4-type frameworks featuring large pore apertures of 9.1a A Here, MFU-4l serves as a reference compound to evaluate the origin of unique and specific gas-sorption properties of MFU-4, reported previously. The latter framework features narrow-sized pores of 2.5a A that allow passage of sufficiently small molecules only (such as hydrogen or water), whereas molecules with larger kinetic diameters (e.g., argon or nitrogen) are excluded from uptake. The crystal structure of MFU-4l has been solved ab initio by direct methods from 3D electron-diffraction data acquired from a single nanosized crystal through automated electron diffraction tomography (ADT) in combination with electron-beam precession. Independently, it has been solved using powder X-ray diffraction. Thermogravimetric analysis (TGA) and variable-temperature X-ray powder diffraction (XRPD) experiments carried out on MFU-4l indicate that it is stable up to 500°C (N₂ atmosphere) and up to 350°C in air. The framework adsorbs 4a wt% hydrogen at 20a bar and 77a K, which is twice the amount compared to MFU-4. The isosteric heat of adsorption starts for low surface coverage at 5a kJmol^-1 and decreases to 3.5a kJmol ^-1 at higher H₂ uptake. In contrast, MFU-4 possesses a nearly constant isosteric heat of adsorption of ca. 7a kJmol^-1 over a wide range of surface coverage. Moreover, MFU-4 exhibits a H₂ desorption maximum at 71a K, which is the highest temperature ever measured for hydrogen physisorbed on metal-organic frameworks (MOFs).

Full text of DOI:10.1002/chem.201001872

FULL PAPER
DOI: 10.1002/chem.201001872
Elucidating Gating Effects for Hydrogen Sorption in MFU-4-Type Triazolate-
Based Metal–Organic Frameworks Featuring Different Pore Sizes
Dmytro Denysenko,[a] Maciej Grzywa,[a] Markus Tonigold,[b] Barbara Streppel,[c]
Ivana Krkljus,[c] Michael Hirscher,[c] Enrico Mugnaioli,[d] Ute Kolb,[d] Jan Hanss,[a] and
Dirk Volkmer*[a, b]
Abstract: A highly porous member of
isoreticular MFU-4-type frameworks,
(e.g., argon or nitrogen) are excluded
from uptake. The crystal structure of
MFU-4l has been solved ab initio by
direct methods from 3D electron-dif-
fraction data acquired from a single
nanosized crystal through automated
ments carried out on MFU-4l indicate
that it is stable up to 5008C (N2 atmos-
phere) and up to 3508C in air. The
framework adsorbs 4 wt% hydrogen at
20 bar and 77 K, which is twice the
amount compared to MFU-4. The iso-
steric heat of adsorption starts for low
surface coverage at 5 kJmolÀ1 and de-
creases to 3.5 kJmolÀ1 at higher H2
uptake. In contrast, MFU-4 possesses a
nearly constant isosteric heat of ad-
sorption of ca. 7 kJmolÀ1 over a wide
range of surface coverage. Moreover,
MFU-4 exhibits a H2 desorption maxi-
mum at 71 K, which is the highest tem-
perature ever measured for hydrogen
physisorbed on metal–organic frame-
works (MOFs).
[Zn5Cl
BTDD=bis(1H-1,2,3-triazolo
[4’,5’-i])dibenzo[1,4]dioxin), has been
4ACHTUNGTRENNUNG(BTDD)3] (MFU-4lCATHUNGTRENNNUG
AHCTUNGTRENNUNG
A
ACHTUNGTRENNUNG
synthesized using ZnCl2 and H2-BTDD
in N,N-dimethylformamide as a sol-
vent. MFU-4l represents the first ex-
ample of MFU-4-type frameworks fea-
turing large pore apertures of 9.1 ꢀ.
Here, MFU-4l serves as a reference
compound to evaluate the origin of
unique and specific gas-sorption prop-
erties of MFU-4, reported previously.
The latter framework features narrow-
sized pores of 2.5 ꢀ that allow passage
of sufficiently small molecules only
(such as hydrogen or water), whereas
molecules with larger kinetic diameters
electron
diffraction
tomography
(ADT) in combination with electron-
beam precession. Independently, it has
been solved using powder X-ray dif-
fraction. Thermogravimetric analysis
(TGA) and variable-temperature X-ray
powder diffraction (XRPD) experi-
Keywords: adsorption · hydrogen ·
metal–organic frameworks · ther-
mal desorption spectroscopy
triazolates
·
Introduction
[a] D. Denysenko, Dr. M. Grzywa, Dr. J. Hanss, Prof. Dr. D. Volkmer
Institute of Physics
Hydrogen is an attractive energy carrier that could replace
petroleum in the future. However, hydrogen storage is a dif-
ficult problem that still has to be solved. Investigation of
metal–organic frameworks (MOFs) as porous materials for
hydrogen storage is currently an important research field.[1]
Due to a recent requirement of the United States Depart-
ment of Energy, a hydrogen storage tank must contain
6 wt% of hydrogen. Hydrogen uptake depends on several
properties of porous material. At low pressure, hydrogen
uptake correlates with the heat of adsorption. At intermedi-
ate pressure (30 bar), uptake correlates with the surface
area, and at high pressure (100 bar and more), uptake corre-
lates with the free volume.[2] Uptake values up to 10 wt%
H2 at 100 bar and 77 K were reported for MOF-5.[3] Howev-
er, a material with a high surface area or free volume alone
is not necessarily a good candidate for hydrogen storage.
Chair of Solid State and Material Science
Augsburg University
Universitꢁtsstrasse 1, 86135 Augsburg (Germany)
Fax : (+49)821-598-5955
E-mail: dirk.volkmer@physik.uni-augsburg.de
[b] M. Tonigold, Prof. Dr. D. Volkmer
Institute of Inorganic Chemistry II—Materials and Catalysis
Ulm University
Albert-Einstein-Allee 11, 89081 Ulm (Germany)
[c] B. Streppel, I. Krkljus, Dr. M. Hirscher
Max Planck Institute for Metals Research
Heisenbergstrasse 3, 70569 Stuttgart (Germany)
[d] Dr. E. Mugnaioli, Dr. U. Kolb
Institute of Physical Chemistry
Johannes Gutenberg-University Mainz
Welderweg 11, 55099 Mainz (Germany)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201001872.
Chem. Eur. J. 2011, 17, 1837 – 1848
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1837

Relatively high adsorption enthalpy is required for success-
ful application. Calculations[4] show that the adsorption en-
thalpy of 15 kJmolÀ1 would be optimal for a reversible ad-
sorption–desorption cycle at room temperature. However,
due to very weak van der Waals interactions of H2 mole-
cules, the heat of hydrogen physisorption typically ranges
from 4 to 7 kJmolÀ1. Metal–organic frameworks with unsa-
turated metal sites can reach an adsorption enthalpy of over
10 kJmolÀ1.[5] The pore-size distribution of a MOF strongly
influences the heat of adsorption. Thus, frameworks with
smaller pore sizes have higher heat of adsorption due to a
stronger interaction between adsorbed hydrogen molecules
and cavities.[6] The desorption temperature of hydrogen like-
wise increases with decreasing pore size.[7] Monte Carlo cal-
culations have been used to show that pores with a diameter
of 7 ꢀ should be optimal for hydrogen adsorption.[8] Howev-
er, these calculations did not consider the important gating
effects of very small apertures or narrow channels intercon-
necting the internal voids of the storage material. Finally,
the stability of a metal–organic framework plays a very im-
portant role for its potential use as a hydrogen-storage ma-
terial as well as for other applications. Thus, the perfor-
mance of MOF-5 in hydrogen adsorption depends strongly
on the preparation and handling conditions since this frame-
work is not stable under ambient conditions owing to its
sensitivity towards hydrolytic decomposition.[3]
on Kuratowski-type pentanuclear SBUs of the general for-
mula [MIIZn4X4(L6)] (Figure 1, right). These feature a cen-
tral metal ion coordinated to six triazolate ligands (L) that
span a Cartesian system and can either be assembled into
discrete coordination compounds[12] or into porous cubic
frameworks.[11] The central metal ion in the octahedral coor-
dination environment can be varied[12] giving the opportuni-
ty of potentially obtaining redox-active SBUs that are also
Lewis acidic, with chemical properties that can be fine-
tuned by selecting appropriate metal ions. Changing the
Zn2+ ions in tetrahedral positions to other transition-metal
ions could improve hydrogen-sorption properties.[5]
The graph theoretical analysis proves that [MIIZn4X4(L6)]
units contain the nonplanar K3,3 graph. According to a theo-
rem of C. Kuratowski,[13] a finite graph is planar only if it
does not contain a subgraph that is a subdivision of K5 (the
complete graph on five vertices) or K3,3 (a complete bipar-
tite graph on six vertices, three of which connect to each of
the other three). As can be seen in Figure 2, the molecular
In a general sense, the success of inventing functional
MOFs will rely on a systematic development of suitable sec-
ondary building units (SBUs) from which a structurally and
functionally diverse class of novel materials might evol-
ve.[9a,b] Herein we present a unique “solid-state construction
kit” that is based on novel pentanuclear SBUs, for which
the term “Kuratowski-type” SBUs is proposed. We have re-
cently developed two major lines of functional MOF com-
pounds, termed MFU-1 and MFU-4, respectively. MFU-1 is
a pyrazolate-based redox-active analogue of the famous
MOF-5[9c] featuring four tetrahedrally coordinated CoII cen-
ters that are connected by a central m4-bridging oxide anion
(Figure 1, left). This compound shows enhanced stability
against hydrolytic and oxidative decomposition and it can
be employed as catalyst in a range of radical-centered oxida-
tion reactions.[10] Since the particular SBU of MFU-1 seems
to be limited to the presence of a {Co4O} core, we have de-
veloped a modular MOF family (MFU-4),[11] which is based
Figure 2. Formal derivation of the K3,3 graph, which can be used to repre-
sent the connectivity scheme in Kuratowski-type SBUs.
graph of [MIIZn4X4(L6)] units in fact contains a subgraph of
K3,3. Accordingly, there is no way to draw [MIIZn4X4(L6)] as
a planar graph and thus we propose a pseudoperspective
skeletal formula as derived in Figure 3 to represent Kura-
towski-type coordination compounds in this paper and in
the future.
Figure 3. Derivation of a skeletal formula representing the connectivity
of [MIIZn4X4(L6)] units (X: terminal ligand; L: 1,2,3-triazolate-type
ligand).
The first triazolate-based MOF featuring Kuratowsky-
type secondary building units, MFU-4, was constructed from
benzobistriazolate linkers and {Zn5Cl4}6+ cores featuring a
very high thermal and hydrolytic stability. Due to its small
pore apertures (2.5 ꢀ) it is highly selective for the adsorp-
tion of atoms or small molecules such as H2 and it can there-
fore be applied in molecular sieving applications, some of
which are hard to achieve with other kinds of porous mate-
rials. However, to separate mixtures of larger molecular ad-
sorbates or for catalytic transformations a porous frame-
Figure 1. Structural features of SBUs found in MFU-1 and MFU-4.
1838
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 1837 – 1848

Hydrogen Sorption in Metal–Organic Frameworks
FULL PAPER
work featuring large pore apertures is required. Herein we
describe MFU-4l
ACHTUNGTRENNUNG
triazolo[4,5-b],[4’,5’-i])dibenzoACTHUNGTRENNUNG
A
U
2), which is a novel member of isoreticular MFU-4-type
frameworks (Scheme 1).
Scheme 1. Triazolate linker 1 for MFU-4 and 2 for MFU-4l.
Comparing hydrogen-sorption properties of MFU-4 and
MFU-4l featuring small and large pores, respectively, might
help to improve present concepts about the influence of
pore size on the potential of a porous material for hydrogen
storage by means of physisorption. In the following we will
demonstrate that small pore apertures are responsible for
selective hydrogen adsorption in MFU-4. Its small pore di-
ameters result in a higher heat of adsorption due to a stron-
ger interaction between adsorbed molecules and the MOF.
This desirable property of MFU-4, however, is compromised
by a reduced void volume if compared with MFU-4l, which
features larger pores and a higher surface area, which allow
for uptake of a larger total amount of hydrogen at high
pressure. Only a few studies with accurate determination of
the isosteric heat of adsorption from isotherms have been
reported so far.[6] A powerful tool that can help us to under-
stand the mechanism of adsorption and diffusion of hydro-
gen in MOFs is thermal desorption spectroscopy (TDS).[7]
Herein we present hydrogen-adsorption isotherms for
MFU-4 and MFU-4l measured over a wide temperature
range (77–298 K) as well as TDS studies at low tempera-
tures between 20 and 120 K.
Scheme 2. Synthesis of ligand 2.
Scheme 3. Synthesis of MFU-4l.
byproducts that have been observed when using low metal/
ligand ratios. The ratio of 20:1 was found to be optimal.
MFU-4l was obtained as a microcrystalline powder with typ-
ical crystal sizes ranging from 1 to 5 mm (Figure 4). Crystals
of this size are not suitable for single-crystal X-ray structure
analysis, thus powder X-ray diffraction and single-crystal
electron diffraction have been used for structural characteri-
zation.
Results and Discussion
Synthesis and characterization: The bis(1H-1,2,3-triazolo-
A
ACHTUNGTRENNUNG[4’,5’-i])dibenzoACTHUNGTRENNNUG
AHCTUNGTRENNUNG
Figure 4. SEM image of MFU-4l prepared by the solvothermal method
(scale bar: 20 mm).
AHCTUNGTRENNUNG
scribed in the literature[14] gives a very low yield (17%) of
the desired product and requires several purification steps.
Therefore, we have developed an improved and simple pro-
cedure that allowed us to obtain 2,3,7,8-tetranitrodibenzo-
The use of N-methylpyrrolidone (NMP) as solvent offers
the advantage of a higher solubility of the linker. Also,
some larger crystals, up to 40 mm, could be obtained in NMP
under solvothermal conditions. However, the yield was very
poor and the product contains a considerable amount of
amorphous impurities.
ACHTUNGTRENNUNG[1,4]dioxin (3) in good yield. The reduction of the nitro com-
pound was carried out according to the literature proce-
dure.[15]
MFU-4l can be synthesized in high yield using solvother-
mal or microwave methods and N,N-dimethylformamide
(DMF) as a solvent (see Scheme 3).
Different ratios of ZnCl2/2 (from 2:1 to 40:1) can be used
for MOF synthesis. However, a large excess amount of zinc
chloride is preferable to avoid the formation of amorphous
Structure solution by electron diffraction: Automated elec-
tron diffraction tomography (ADT)[16–18] is a new technique
in which the reciprocal space is sampled by tilting a nano-
sized single crystal in small steps over the full tilt range ac-
Chem. Eur. J. 2011, 17, 1837 – 1848
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1839

D. Volkmer et al.
cessible in the transmission electron microscope. We used
nanobeam electron diffraction to create a low-dose, small,
semiparallel electron beam and scanning TEM to image and
track the crystal; this technique is particularly suitable for
beam-sensitive materials that are normally not accessible by
conventional electron-diffraction techniques.[19] After recon-
struction of the 3D reciprocal space, cell parameter determi-
nation with a maximum error of 5% is possible. The intensi-
ties that can be observed after indexing are already quasi-
kinematical and cover most of the symmetrically independ-
ent reciprocal space. ADT can be combined with precession
electron diffraction (PED), thus integrating the space be-
tween the tilts to enhance intensity quality further. Based on
ADT data sets, the full structure can be found ab initio by
direct methods in a pure kinematical approach. This tech-
nique was able to solve complex nanoporous structures such
as charoite.[20]A cubic face-centered cell (a=32.0(4) ꢀ) was
automatically determined. No further extinction was detect-
ed in the reconstructed three-dimensional reciprocal space
(Figure 5). All independent reflections up to a resolution of
Figure 6. Ball-and-stick representation of the MFU-4l framework along
the crystallographic a direction (octahedrally coordinated Zn2+: dark-
gray octahedra; tetrahedrally coordinated Zn2+: pale-gray tetrahedra).
All hydrogen atoms are omitted for clarity. The larger spheres represent
the cavities of B cells, whereas the smaller spheres represent the cavities
of A cells.
can be described as follows: the nodes (vertices) are repre-
sented by cationic pentanuclear {Zn5Cl4}6+ clusters and the
links (edges) of the net are represented by finite rods of
BTDD2À anions. There are two types of Zn ions: first, a tet-
rahedrally coordinated zinc atom (Zn1) with 3m site sym-
metry is surrounded by three N atoms from BTDD ligands
(forming the base of the tetrahedron) and one Cl atom (con-
stituting the apex of the tetrahedron); second, an octahe-
Figure 5. Projections of three-dimensional reconstructed reciprocal space
by ADT data: a) [111] projection; b) [001] projection.
¯
1.3 ꢀ were sampled and integrated. Intensities collected in
such a way were quasi-kinematical because they were inte-
grated from nonoriented diffraction patterns. The amount
and quality of the data obtained by ADT is remarkably su-
perior to any previous electron-diffraction experiment per-
formed on MOFs, for which cell parameters were deter-
mined by single-oriented projections.[21] Ab initio structure
solution delivered the complete structure directly except for
one carbon atom. The structure was refined by imposing a
rigid benzene ring and soft bond restraints. The final residu-
al R value was 32.1%, which is in an acceptable range for
electron-diffraction data.
drally coordinated zinc atom (Zn2), with 43m site symmetry
surrounded by six N atoms from six hexadentate-coordinat-
ed BTDD ligands. All BTDD ligands are twisted around
their coordinative bonds to the central Zn2 ions. In the solu-
tion from the powder X-ray diffraction data, the distances
À
Zn N for a tetrahedrally coordinated Zn atom are smaller
(2.010(5) ꢀ) than those between octahedrally coordinated
Zn atoms and nitrogen donors (2.0414(9) ꢀ). In the ADT
À
solution, the distances Zn N for a tetrahedrally coordinated
Zn atom are 2.09(3) ꢀ and for a octahedrally coordinated
Zn atom are 2.12(4) ꢀ, respectively. These values are in
good agreement with other Zn–triazolate complexes (1.98–
2.05 ꢀ for tetrahedral coordination and 2.14–2.22 ꢀ for oc-
tahedral coordination).[11]
Crystal structure description: The crystal structure of MFU-
4l has been independently solved ab initio from powder X-
ray diffraction data by direct methods. Both structure solu-
tions—from powder X-ray and ADT diffraction data, re-
spectively—deliver almost identical structures, which show
small differences in bond lengths only. The maximum devia-
tion in atom positions from the ADT solution is 0.2 ꢀ. The
crystal structure of MFU-4l is similar to that of MFU-4
(Figure 6) in that it possesses a cubic six-connected net. This
According to the results obtained by using the PLATON/
SQUEEZE[22] program, the total potentially accessible void
volume is 23563.2 ꢀ3, which is 78.7% of the unit-cell
volume.
Analogously to MFU-4, the framework of MFU-4l has
two different types of cavities (the smaller and the larger
ones are referred to as A and B cells, respectively) arranged
in an alternating fashion (Figure 6). The A cells are repre-
1840
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 1837 – 1848

Hydrogen Sorption in Metal–Organic Frameworks
FULL PAPER
sented by the cubic arrangement of eight chlorine atoms
that have a minimum nonbonding distance of 8.9320(2) ꢀ
from each other. Taking the van der Waals radii of Cl atoms
(1.75 ꢀ) into account, imaginary spheres with a diameter of
11.97 ꢀ could fit into the A cells. Each of the larger B cells,
on the other hand, is surrounded by twelve dioxin rings and
an imaginary sphere with a diameter of 18.56 ꢀ would fit
into it, taking the van der Waals radii of the C atoms into
account. The aperture between A and B cells would admit
the passage of an imaginary sphere with a diameter of
9.13 ꢀ (taking the van der Waals radii of Cl atoms into ac-
count.
Large pore apertures in MFU-4l allow the adsorption and
free diffusion of larger molecules. Table 1 shows selected
structural parameters of MFU-4 and MFU-4l for compari-
son.
Figure 8. TGA/MS data of MFU-4l.
the temperature range between 200 and 3008C corresponds
to CO2, which was also identified by TGA/IR spectroscopy.
The third peak is DMF (m/z 73 for the molecular ion and
m/z 44 for (CH3)2N+).
The presence of DMF was further confirmed by using a
thermodesorption GC-MS trace of volatile products formed
after heating MFU-4l at 2608C.
Table 1. Selected structural features of MFU-4 and MFU-4l.
MFU-4
MFU-4l
¯
¯
space group
Fm3m (225)
Fm3m (225)
cell length a=b=c [ꢀ]
21.6265(9)
10114(9)
1.23
31.057(1)
29955(4)
0.56
V [ꢀ3]
1calcd [gcmÀ3
]
void volume [%]
diameter A cells [ꢀ]
diameter B cells [ꢀ]
aperture [ꢀ]
53.1
3.88
11.94
2.52
78.7
11.97
18.56
9.13
Variable-temperature
X-ray
powder
diffraction
(VTXRPD) studies (Figure 9a) showed that MFU-4l is
stable up to 3508C in air, which is similar to the stability of
MFU-4. Small differences in the intensities of the reflections
are observed at higher temperatures due to the removal of
residual solvent molecules. At 4008C, zinc oxide (PDF no.
36-1451) appears as a new crystal phase that predominates
Thermal analyses: Due to a larger pore size, MFU-4l loses
solvent molecules much easier then MFU-4. Thus, MFU-4l
heated in a vacuum at 1808C contains no solvent as ob-
served by thermogravimetric analysis (TGA) and shows no
weight loss up to 5008C (Figure 7). In contrast, MFU-4 must
be heated at 2508C under vacuum to remove DMF mole-
cules, which cannot pass through small apertures easily.
Figure 7. Temperature-dependent weight-loss of MFU-4l (sample ex-
posed to flowing nitrogen gas).
By using more accurate TGA/MS measurements on the
sample stored in air, traces of adsorbed molecules were de-
tected (Figure 8). Weight loss below 2008C corresponds to
water (m/z 18). The second peak with m/z 44 observed in
Figure 9. VTXRPD plots of MFU-4l in the range of a) 50–4508C under
air and b) 30–6008C under a nitrogen atmosphere.
Chem. Eur. J. 2011, 17, 1837 – 1848
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1841

D. Volkmer et al.
above 4508C, at which the framework is completely decom-
posed. Under a nitrogen atmosphere it is stable even up to
5008C (Figure 9b), which is in agreement with TGA meas-
urements.
To evaluate the pore-size distribution of MFU-4l, the
argon-sorption isotherms sampled at 77 K were analyzed
using nonlocal density functional theory (NLDFT)[25] imple-
menting a carbon equilibrium transition kernel for argon ad-
sorption at 77 K based on a slit-pore model.[26] The distribu-
tions calculated by fitting the adsorption data (Figure 11)
Physisorption results: MFU-4l exhibits permanent porosity,
which was confirmed by argon gas sorption. Prior to mea-
surement, the sample was suspended multiple times in di-
chloromethane, which led to solvent exchange of less-vola-
tile DMF molecules. The sorption isotherm of MFU-4l ob-
tained with Ar gas reveals a type-I sorption behavior, which
is characteristic of microporous solids (Figure 10). The pore
Figure 11. Pore-size distribution for MFU-4l calculated by fitting NLDFT
models to the argon-adsorption data.
are indicative of micropores with average aperture diame-
ters between 1.1–2.0 nm for MFU-4l. These values are in
good agreement with the average pore diameters calculated
from crystallographic data (1.2 nm for the pores in the A
cell and 1.86 nm for pores in the B cell).[27]
However, both the results of the pore geometries derived
from NLDFT calculations and the Dubinin–Radushkevich
(DR) equation have to be regarded with caution and should
not be over-interpreted, since the available slit-pore model
in the case of NLDFT is slightly erroneous due to wrong
pore geometry assumptions, and DR suffers from the fact
that is does not give a realistic description of micropore fill-
ing because it is derived from classical, macroscopic theo-
ries.[23]
Figure 10. Representative argon adsorption isotherm at 77 K for a desol-
*
vated sample of MFU-4l ( : adsorption, &: desorption).
volume obtained from the sorption isotherm at P/P0 =0.9 is
1.26 cm3 gÀ1 (1.15 cm3 gÀ1 pore volume and 1.93 nm pore di-
ameter determined by the Dubinin–Radushkevich equa-
tion),[23] which is close to the value expected from the crys-
tallographic data (1.42 cm3 gÀ1 for the solvent-free crystal
and 1.86 nm pore diameter). The adsorption data was fitted
to the BET equation to give a surface area of 2750 m2 gÀ1
for MFU-4l,[24] which is close to the theoretical values of the
specific surface area of 2987 m2 gÀ1 as derived from the crys-
tal structure by a Monte Carlo integration technique (in
which a probe molecule, here argon, is “rolled” over the sur-
face).[24] Experimental values of MFU-4 and MFU-4l are
compared in Table 2.
Hydrogen adsorption: Excess hydrogen-adsorption iso-
therms of MFU-4l up to 20 bar for temperatures between
77 K and room temperature are shown in Figure 12. At
77 K, a typical type-I isotherm is observed but saturation is
not reached up to 20 bar. At 20 bar the excess hydrogen
uptake is 4 wt% for 77 K and decreases with rising tempera-
ture. At room temperature the hydrogen uptake increases
linearly with the pressure up to 0.1 wt%. The temperature
and pressure dependence of the hydrogen uptake for MFU-
4l is very similar to MOF-5, which shows an uptake of
4.5 wt% at 77 K and also no saturation up to 20 bar.[6] The
investigated MOF-5 reference sample had a specific surface
area of 2360 m2 gÀ1 (nitrogen BET), which is comparable to
the surface area of MFU-4l of 2750 m2 gÀ1 (argon BET).
In Figure 13a the excess hydrogen uptake of MFU-4l and
MFU-4 (all isotherms measured for MFU-4 are given in the
Table 2. Measured and calculated[24] specific surface areas.
Framework
Adsorbate
Specific surface area [m2 gÀ1
]
Calculated
Measured
MFU-4
MFU-4
MFU-4l
MFU-4l
H2
Ar
H2
Ar
1736
1350
3095
2987
ca. 0
2750
1842
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 1837 – 1848

Hydrogen Sorption in Metal–Organic Frameworks
FULL PAPER
is different for MFU-4 and MFU-4l. This leads to a shift of
the crossover to higher pressures at higher temperatures.
For example, at 117 K the crossover is shifted from 0.24 to
1.2 MPa (Figure 13b). At 117 K and 20 bar, MFU-4 stores
1.3 wt% and MFU-4l just slightly more with 1.8 wt%.
The difference in the shape of the isotherm is reflected in
the heat of adsorption (Figure 14). The isosteric heat of ad-
sorption is calculated from a variant of the Clausius–Cla-
Figure 12. Excess hydrogen adsorption isotherms for MFU-4l at 77 K
~
&
(liquid nitrogen, ), 87 K (liquid argon, ), 87 K (cooling system (CS),
!
*
^
), 97 K (CS, ), 107 K (CS, ), 117 K (CS, 3), and 298 K (").
Figure 14. Isosteric heat of adsorption for hydrogen in MFU-4 (triangles)
and MFU-4l (squares) normalized to the hydrogen uptake at 20 bar and
77 K.
peyron equation from high pressures (0–2 MPa) isotherms
for intermediate surface coverage. For MFU-4 the isosteric
heat of adsorption is constant within the experimental un-
certainty at 7 kJmolÀ1, which is one of the highest heat of
adsorption ever observed over such a wide range of surface
coverage for hydrogen physisorption in porous materials.[6]
For MFU-4l, the isosteric heat of adsorption at low surface
coverage is approximately 5 kJmolÀ1 and strongly decreases
with hydrogen uptake. Above 30% surface coverage the
heat of adsorption remains constant at 3.5 kJmolÀ1. This dif-
ference in the isosteric heat of adsorption is caused by the
pore structure, as smaller pores lead to higher overlap of the
van der Waals potentials of the wall and therefore higher
heat of adsorption.[6]
Thermal desorption spectroscopy: Figure 15 shows thermal
desorption spectra of hydrogen for MFU-4 and MFU-4l for
two different heating rates, 0.1 and 0.01 KsÀ1 in a tempera-
ture range between 20 and 120 K. At higher temperatures
no hydrogen was desorbed. The TDS spectra show distinct
differences in the temperature profile and in the magnitude
of the signal between MFU-4 and MFU-4l. Furthermore,
the desorption spectra measured with a slower heating rate
are slightly shifted toward lower temperatures, which indi-
cates that the hydrogen release is thermally activated. For
MFU-4l, the majority of the adsorbed hydrogen molecules
were desorbed below 60 K, whereas for MFU-4 the major
hydrogen desorption just began at this temperature. The
Figure 13. Excess hydrogen uptake at a) 77 K and b) 117 K for MFU-4
(triangles) and MFU-4l (squares).
Supporting Information) at 77 K are shown. The two iso-
therms cross each other at 0.24 MPa. MFU-4 stores more
hydrogen at pressures below 0.24 MPa, while at higher pres-
sures MFU-4l does not reach saturation and stores up to
twice as much hydrogen as MFU-4 (4 wt% at 20 bar). The
temperature dependence of the maximum hydrogen uptake
Chem. Eur. J. 2011, 17, 1837 – 1848
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1843

D. Volkmer et al.
the hydrogen is desorbing at temperatures above 60 K,
which is much higher than that of the other MOFs.[9] Indeed
MFU-4 exhibits a desorption maximum with the highest
temperature ever measured for physisorbed hydrogen on
MOFs. The MFU-4 framework is constructed from smaller
and larger cavities, arranged in an alternate fashion.[11]
Owing to this structure, hydrogen molecules moving from
one larger cavity to another must pass through a smaller
cavity. Therefore, the diffusion of the hydrogen molecule
through the aperture of the small cavities, with a diameter
of approximately 2.52 ꢀ, is the limiting factor for the de-
gassing of hydrogen and gives rise to the high desorption
temperature.
The total hydrogen uptake was calculated to about 1 wt%
for MFU-4, in comparison to 2.5 wt% for MFU-4l. These
values are lower than the maximum excess adsorption at
high pressures and 77 K, which typically, for MOFs are cor-
related to the specific surface area and are independent of
the compound.[7,29] The results obtained from TDS are
sometimes lower, and especially if the heat of adsorption is
low, part of the adsorbed hydrogen will be pumped away
during evacuation already at 20 K. The lower value for
MFU-4 may be caused by a partial filling of the cavities at
700 mbar and cooling down to 20 K, because the kinetics of
the filling is limited by the diffusion through the small
window of the smaller cavity.
Figure 15. Hydrogen thermal desorption spectra of MFU-4 (a) and
MFU-4l (c) recorded with
a
heating rate of 0.1 KsÀ1 (top) and
Conclusion
0.01 KsÀ1 (bottom).
We have successfully prepared and characterized a novel
member of isoreticular MFU-4-type cubic frameworks,
MFU-4l, constructed from BTDD2À dianions and {Zn5Cl4}6+
coordination units. The linker can be easily synthesized in
three steps. Compared to solvothermal synthesis, the prepa-
ration of MFU-4l by microwave irradiation leads to a large
reduction in reaction time. Thermogravimetric and
VTXRPD analyses indicate that MFU-4l possesses very
high thermal stability (5008C under nitrogen). Large pore
apertures of 9.1 ꢀ allow adsorption and free diffusion of dif-
ferent molecules. In contrast, MFU-4, with small pore aper-
tures of 2.5 ꢀ, is highly selective for the adsorption of atoms
or small molecules such as He or H2 and it can therefore be
applied in molecular sieving applications, some of which are
hard to achieve with any other kinds of porous materials.
MFU-4l, which has a much higher surface area, generally is
able to adsorb more hydrogen than MFU-4. However, at
higher temperatures and lower pressures MFU-4 adsorbs
more hydrogen than MFU-4l. Moreover, the values of iso-
steric heat of adsorption, which is constant over a wide
range of surface coverage and of desorption temperature,
belong to the highest ever observed for hydrogen physisorp-
tion on porous materials. This makes MFU-4 more suitable
for hydrogen adsorption than MFU-4l. These observations
demonstrate the importance of pore size in designing new
materials for hydrogen storage. Figure 16 shows a compari-
son of MFU-4 and MFU-4l frameworks. The diffusion path
total amount of desorbed gas, corresponding to the area
under the desorption curve, was about 2.5 and 1 wt% for
MFU-4l and MFU-4, respectively.
MFU-4l exhibits two hydrogen-desorption maxima, one
broad and large maximum centered at approximately 32 K
and a smaller maximum centered at 51 K (for a heating rate
of 0.1 KsÀ1). During cooling under a hydrogen atmosphere a
certain amount of gas is liquefied or adsorbed in multilayers
giving rise to additional desorption in this low-temperature
region.[7,28] For MFU-4l this effect gives rise to the low-tem-
perature shoulder and the first large desorption peak for a
heating rate of 0.1 and 0.01 KsÀ1, respectively. Nevertheless,
the presence of two distinct desorption maxima for MFU-4l
indicates clearly the presence of two hydrogen adsorption
sites possessing different heats of adsorption. The larger first
desorption peak at about 32 K can be assigned to hydrogen
adsorbed in the cavities. The smaller maximum at about
51 K corresponds to stronger binding sites as already indi-
cated by the higher heat of adsorption for low surface cover-
age in the pressure–capacity–temperature (PCT) measure-
ments (Figure 14).
In contrast, for MFU-4 one dominant desorption maxi-
mum is observed at 71 K and a smaller signal around 30 K.
The desorption signal in the low-temperature region may be
caused by gas liquefied or adsorbed in multilayers. Most of
1844
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 1837 – 1848

Hydrogen Sorption in Metal–Organic Frameworks
FULL PAPER
Figure 16. Frameworks of MFU-4l (top), MFU-4 (bottom left and middle), and the diffusion path of hydrogen molecules in MFU-4 through the aperture
between A and B cells (bottom right).
range 4000–400 cmÀ1 on a Bruker IFS FTIR spectrometer. The following
indications are used to characterize absorption bands: very strong (vs),
strong (s), medium (m), weak (w), shoulder (sh), and broad (br). Ele-
of hydrogen molecules in MFU-4 through the aperture be-
tween the A and B cells is shown on the bottom right.
In conclusion, we have demonstrated that MFU-4-type
frameworks have a huge potential for the development of
functional materials for hydrogen storage, gas separation
(MFU-4), and possible catalytic applications (MFU-4l). Fur-
ther investigations of these frameworks are currently under-
way in our laboratories. For structure solution and refine-
ment of MFU-4l, three-dimensional electron-diffraction
data acquired by using the ADT technique was used suc-
cessfully. This is the first ab initio structure solution of an
unknown MOF performed with electron-diffraction data.
The new ADT method delivers intensity data superior in
quality and quantity and has proven to be quick, reliable,
and promising for future structural investigations on MOF
compounds, in which single crystals of sufficient sizes are
notoriously hard to obtain.
mental analyses (C, H, N) were carried out on a Perkin–Elmer 2400 ele-
mental analyzer. TGA was performed with a TGA/SDTA851 Mettler
Toledo analyzer in the temperature range of 25–11008C under flowing ni-
trogen at a heating rate of 10 KminÀ1. TGA/MS analysis was carried out
using Netsch thermoanalyzer STA 409 C connected to a Balzers QMG
mass spectrometer by a Skimmer coupling system in the temperature
range of 20–7008C under a N2 flow with a heating rate of 10 KminÀ1
.
Thermodesorption GC–MS analysis was carried out using Perkin–Elmer
ATD 400 inlet and a Varian 3400 gas chromatograph coupled with a Fin-
nigan Mat ITS40 mass spectrometer. The sample was heated in a glass
tube at 2608C for 15 min before measurement. SEM images were record-
ed using a Zeiss DSM 962 scanning electron microscope. Argon-gas sorp-
tion isotherms were measured with a Quantachrome Autosorb-I ASI-CP-
8 instrument. Prior to measurements, the samples of dichloromethane-ex-
changed MFU-4l were heated at 1808C for 24 h under high vacuum to
remove the occluded solvent molecules. Argon-sorption experiments
were performed at 77.3 K in the range of 5.00ꢃ10À5 ꢀP/P0 ꢀ1.00 with
Ar.
2,3,7,8-Tetranitrodibenzo
was added to trifluoroacetic acid anhydride (16 mL) under cooling in an
ice/water bath. Dibenzo[1,4]dioxin (5 g, 27.2 mmol) was added in small
portions to a well-stirred nitration mixture while keeping the temperature
below 108C. The reaction mixture was stirred for 1.5 h at 60–708C at
reflux and then poured into ice/water (400 mL) with stirring. The precipi-
tate was removed by filtration, washed well with water, and dried under
vacuum over P4O10. Yield: 7.2 g (73%); 1H NMR (400 MHz,
[1,4]dioxin (3):[14] Fuming nitric acid (24 mL)
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
Experimental Section
Materials and general methods: All starting materials were of reagent
grade and used as received from the commercial supplier. Fourier trans-
form infrared (FTIR) spectra were recorded from KBr pellets in the
Chem. Eur. J. 2011, 17, 1837 – 1848
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1845

D. Volkmer et al.
[D6]DMSO): d=8.03 ppm (s, 4H); 13C NMR (100 MHz, [D6]DMSO):
d=113.8, 138.7, 143.6 ppm; IR (KBr): n˜ =3104 (m), 3055 (m), 1641 (w),
1605 (m), 1558 (s), 1495 (s), 1431 (m), 1376 (s), 1348 (s), 1306 (s), 895 (s),
818 cmÀ1 (s); elemental analysis calcd (%) for C12H4N4O10: C 39.58, H
1.11, N 15.38; found: C 39.08, H 1.17, N 15.18.
tential to a ghost close to a Zn atom. The missing carbon atom (potential
11) was placed in a wrong special position.
The structure was refined by SHELXL.[32] The ghost disappeared during
the refinement and the missing carbon atom was placed imposing a rigid
benzene group. The final refinement, performed with soft restraints on
bond lengths, resulted in a final residual R of 32.1%.
H2-BTDD·0.5H2O (2): A well-stirred mixture of compound
4 (7 g,
19.8 mmol), acetic acid (70 mL), and water (10 mL) was cooled in an ice/
water bath and a solution of sodium nitrite (2.9 g, 42 mmol) in water
(10 mL) was added slowly while keeping the temperature below 108C.
The mixture was diluted with water (100 mL), the precipitate was re-
moved by filtration, washed well with water and methanol, and dried
under vacuum over P4O10. Yield 3.72 g (69%); 1H NMR (400 MHz,
CF3COOD): d=7.85 ppm (s, 4H); 13C NMR (100 MHz, CF3COOD): d=
99.8, 131.1, 144.7 ppm; IR (KBr): n˜ =3448 (br), 3129 (s), 2899 (s), 2803
(s), 1711 (w), 1595 (m), 1480 (s), 1416 (m), 1355 (s), 1218 (s), 1076 (m),
1004 (m), 919 (m), 860 (s), 633 (w), 431 cmÀ1 (w); elemental analysis
calcd (%) for C12H7N6O2.5: C 52.37, H 2.56, N 30.54; found: C 52.45, H
2.52, N 29.91.
Crystal structure determination by PXRD: For the PXRD study, a por-
tion of the sample was powdered and placed between two sheets of foil.
Intensity data were collected using a STOE STADI P powder diffractom-
eter with germanium monochromator, operated at 40 kV, 40 mA, Cu
target; transmission geometry, fixed divergence slit 1/48. The PXRD pat-
tern was taken at room temperature in the 2q range from 2.8 to 708, step
size 0.01, and time per step 596.6 s.
Variable-temperature X-ray powder diffraction (VTXRPD) measure-
ments were performed under air or nitrogen with a PANalytical X’Pert
PRO diffractometer with a X’Celerator detector operated at 45 kV,
1
40 mA, with CuKa radiation, fixed divergence slit = 8, equipped with an
2
Anton Paar HTK 1200N reaction chamber. Measurements were per-
formed at a temperature range from 30 to 6008C, by employing 2q
ranges from 3.0 to 80.08, step size 0.0338 2q, time 98 s per step. The heat-
ing rate was 58CminÀ1. The sample was heated before measurement at
each temperature for 15 min.
ACHTUNGTRENNUNG[Zn5Cl4ACHTUNGTRENNUNG(BTDD)3] (MFU-4l)
Solvothermal method: H2-BTDD (760 mg, 2.77 mmol) was dissolved in
DMF (760 mL) under stirring and heating at 1458C for 30 min. Anhy-
drous zinc chloride (7.78 g, 57.2 mmol) was added to a cooled (ca. 508C)
solution of linker and the mixture was stirred until the zinc chloride was
completely dissolved. The resulting solution was heated with stirring
under reflux at 1458C for 18 h and then cooled down to room tempera-
ture. The precipitate was removed by filtration, washed slowly with DMF
(3ꢃ50 mL), methanol (3ꢃ50 mL), and dichloromethane (3ꢃ50 mL), and
dried for 24 h at 1808C under vacuum (ca. 0.2 mbar). Yield 940 mg (90%
based on ligand) of an almost white microcrystalline powder; IR (KBr):
n˜ =3420 (br), 3076 (w), 2924 (w), 2854 (w), 1731 (w), 1576 (w), 1460 (s),
1346 (m), 1171 (s), 915 (m), 802 (m), 732 (w), 601 (w), 500 cmÀ1 (m); ele-
mental analysis calcd (%) for C36H12Cl4N18O6Zn5: C 34.28, H 0.96, N
19.99; found: C 33.98, H 1.21, N 19.55.
Extractions of the peak positions, pattern indexing, and determination of
the lattice parameters for MFU-4l were carried out with the PROSZKI
package.[33] Independently, the indexing process was performed by the N-
TREOR09 program implemented in the EXPO2009 package.[34] Space
group determination by probabilistic approach was performed by using
EXPO2009. The set of the most probable space groups was found: ex-
¯
¯
tinction group F; space groups F23 (196), Fm3 (202), F432 (209), F43m
¯
¯
(216), Fm3m (225). The Fm3m (225) space group was chosen for further
structure determination procedures. During pattern decomposition the
lattice parameters were not refined. The positions of heavy atoms Zn
and Cl were found by direct methods; missing light atoms O, C, and N
were localized on difference Fourier maps. Hydrogen atoms were placed
in idealized position in the SHELXL program.[32]
Microwave irradiation method: H2-BTDD (5 mg, 0.0188 mmol) was dis-
solved in DMF (5 mL) under stirring and heating at 1458C for 10 min. A
1m solution of anhydrous zinc chloride in DMF (0.4 mL, 0.4 mmol) was
added to a cooled (ca. 508C) solution and the mixture was placed in a
Pyrex sample tube (10 mL). The tube was sealed and placed in a micro-
wave synthesizer (CEM, Discover S). The resulting mixture was heated
to 1558C at 300 W, kept under these conditions for 30 min, and then
cooled down to room temperature. The precipitate was removed by fil-
tration, washed slowly with DMF (5 mL), methanol (5 mL), and dichloro-
methane (3ꢃ10 mL), and dried for 24 h at 1808C under vacuum (ca.
0.2 mbar). Yield 5.2 mg (73%) of an almost white microcrystalline
powder. This material exhibited the same analytical results as those ob-
tained by the solvothermal method and is phase-pure according to
XRPD measurement.
The Rietveld refinement was carried out using the Jana2006 program.[35]
Weak geometric restraints on bond lengths were used during the refine-
ment process. No preferred orientation has been observed. Experimental
details and crystal data for MFU-4l are listed in Table 3. The final Riet-
veld refinement plots are presented in Figure 17.
CCDC-776578 contains the supplementary crystallographic data for this
paper. These data can be obtained free of charge from The Cambridge
Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
Hydrogen adsorption measurements: Hydrogen-adsorption measure-
ments were performed with an automated Sievertsꢄ apparatus
Crystal structure determination by ADT: ADT[16–18] was performed in a
FEI F30 TEM. The sample was deposited as a dry powder on a carbon
grid and cooled to À1608C inside the microscope. An area of around
300ꢃ300 nm, on the edge of a crystal that was 1000ꢃ600 nm large, was
selected for data acquisition. To have a quasi-parallel electron beam of
70 nm in diameter and a low electron dose on the sample, a C2 condens-
er aperture of 10 mm was inserted and a high gun lens current and spot
size were used (respectively 8 and 8 for that microscope). During the tilt,
the crystal position was tracked in STEM microprobe mode and the elec-
tron-diffraction patterns were collected every 18 of tilt, with an exposure
time of 5 s. Two tilt series were collected with and without precession of
the beam.[30] The tilt without precession, sampling a range of 608, was
used for cell-parameter determination. The tilt with precession, sampling
a range of 668, was used for intensity extraction.
Table 3. Crystal and experimental data for MFU-4l.
chemical formula
formula weight
T [K]
2q range [8], step size [8]
X-ray source, wavelength [ꢀ]
C36Cl4N18H12O6Zn5
1261.32
293(2)
2.80–70, 0.01
CuKa, l=1.54178
cubic
crystal system
space group
a [ꢀ]
¯
Fm3m (225)
31.0569(6)
29955.2(5)
75.01 (0.00002, 30)
184.77 (0.00416, 30)
8, 0.5592
5
6720
315
4.63
7.16
5.88
5.82
V [ꢀ3]
M30
F30
Z, 1calcd [gcmÀ3
]
no. of atoms
no. of observations
unique reflections
Rp
Rwp
Robs
Rwobs
All the possible 412 independent reflections up to a resolution of 1.3 ꢀ
were integrated (coverage of 100% of the reciprocal space). The internal
Rsym was 28.63%. The isotropic thermal factor determined by a Wilson
plot was 0.048 ꢀ2. Ab initio structure solution was performed by direct
methods implemented in SIR2008[31] with a fully kinematic approach (I=
F2). The almost complete structure was delivered in one run. Inside the
first 10 potentials, 9 corresponded to 9 atoms of the structure and 1 po-
1846
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 1837 – 1848

Hydrogen Sorption in Metal–Organic Frameworks
FULL PAPER
zelnen Molekꢅlen zu nanoskopisch strukturierten Materialien”) is grate-
fully acknowledged.
˘
[1] L. J. Murray, M. Dinca, J. R. Long, Chem. Soc. Rev. 2009, 38, 1294–
1314.
[2] H. Frost, T. Dꢅren, R. Q. Snurr, J. Phys. Chem. B 2006, 110, 9565–
9570.
[3] S. S. Kaye, A. Dailly, O. M. Yaghi, J. R. Long, J. Am. Chem. Soc.
2007, 129, 14176–14177.
[4] S. K. Bhatia, A. L. Myers, Langmuir 2006, 22, 1688–1700.
[5] M. Dinca, J. R. Long, Angew. Chem. 2008, 120, 6870–6884; Angew.
Chem. Int. Ed. 2008, 47, 6766–6779.
[6] B. Schmitz, U. Mꢅller, N. Trukhan, M. Schubert, G. Fꢆrey, M.
Hirscher, ChemPhysChem 2008, 9, 2181–2184.
[7] B. Panella, K. Hçnes, U. Mꢅller, N. Trukhan, M. Schubert, H.
Pꢅtter, M. Hirscher, Angew. Chem. 2008, 120, 2169–2173; Angew.
Chem. Int. Ed. 2008, 47, 2138–2142.
[8] M. Rzepka, P. Lamp, M. A. de La Casa-Lillo, J. Phys. Chem. B
1998, 102, 10894–10898.
Figure 17. The Rietveld refinement plots for MFU-4l. Dotted and solid
lines represent observed and calculated patterns, respectively, with peak
markers and the difference plot shown at the bottom. For clarity, the
inset shows an expanded view in the range 7–708 2q.
[9] a) M. Eddaoudi, D. B. Moler, H. Li, B. Chen, T. M. Reineke, M.
OꢄKeeffe, O. M. Yaghi, Acc. Chem. Res. 2001, 34, 319–330; b) M.
Eddaoudi, J. Kim, N. Rosi, D. Vodak, J. Wachter, M. OꢄKeeffe,
O. M. Yaghi, Science 2002, 295, 469–472; c) H. Li, M. Eddaoudi, M.
OꢄKeeffe, O. M. Yaghi, Nature 1999, 402, 276–279.
[10] M. Tonigold, Y. Lu, B. Bredenkçtter, B. Rieger, S. Bahnmꢅller, J.
Hitzbleck, G. Langstein, D. Volkmer, Angew. Chem. 2009, 121,
7682–7687; Angew. Chem. Int. Ed. 2009, 48, 7546–7550.
[11] S. Biswas, M. Grzywa, H. P. Nayek, S. Dehnen, I. Senkovska, S.
Kaskel, D. Volkmer, Dalton Trans. 2009, 6487–6495.
[12] S. Biswas, M. Tonigold, D. Volkmer, Z. Anorg. Allg. Chem. 2008,
634, 2532–2538.
[13] K. Kuratowski, Fund. Math. 1930, 15, 271–283.
[14] H. Gilman, J. J. Dietrich, J. Am. Chem. Soc. 1958, 80, 366–368.
[15] K. Chichak, U. Jacquemard, N. R. Branda, Eur. J. Inorg. Chem.
2002, 357–368.
[16] U. Kolb, T. Gorelik, C. Kꢅbel, M. T. Otten, D. Hubert, Ultramicro-
scopy 2007, 107, 507–513.
(PCTPro2000, HY-Energy LLC, Setaram Inc.) and are described in detail
elsewhere.[6] MFU-4l (115 mg) was evacuated at 1808C overnight prior to
the measurements and between each isotherm for 2 h at 508C. Equilibri-
um was reached for all pressures below 1 min. For MFU-4, two measure-
ments were performed, first at temperatures between 77 and 117 K on a
180 mg sample and later at temperatures between 127 and 177 K on a
216 mg sample. In both experiments the sample was out-gassed overnight
prior to the measurement at 250 (first sample) and 1808C (second
sample). Between individual measurements of isotherms the sample was
out-gassed at temperatures above 508C for more than 1 h. For further
analysis, results from both samples were combined.
In porous solids hydrogen is mainly adsorbed as a monolayer on inner
surfaces. The amount of hydrogen that is in this adsorbed layer is called
absolute adsorbed hydrogen.[36] Whereas, the excess adsorption, typically
measured by taking into account the skeletal density of the material, rep-
resents the amount of hydrogen that is stored in addition to the amount
of hydrogen gas that would be present in the pore volume due to exter-
nal pressure. The excess adsorption is therefore smaller than the absolute
adsorption by the amount of hydrogen gas that would be in the adsorbed
layer if no adsorption occurred. Hence the absolute adsorption is calcu-
lated from the excess adsorption with the assumption that the adsorbed
layer exhibits the density of liquid hydrogen (see the Supporting Infor-
mation). The isosteric heat of adsorption is calculated from the absolute
adsorbed hydrogen according to a variant of the Clausius–Clapeyron
equation.[6]
[17] U. Kolb, T. Gorelik, M. T. Otten, Ultramicroscopy 2008, 108, 763–
772.
[18] E. Mugnaioli, T. Gorelik, U. Kolb, Ultramicroscopy 2009, 109, 758–
765.
[19] U. Kolb, T. E. Gorelik, E. Mugnaioli, A. Stewart, Polymer Reviews
2010, 50, 385–409.
[20] I. Rozhdestvenskaya, E. Mugnaioli, M. Czank, W. Depmeier, U.
Kolb, A. Reinholdt, T. Weirich, Mineral. Mag. 2010, 74, 159–177.
[21] a) O. I. Lebedev, F. Millange, C. Serre, G. Van Tendeloo, G. Fꢆrey,
Chem. Mater. 2005, 17, 6525–6527; b) D. Chandra, M. W. Kasture,
A. Bhaumik, Microporous Mesoporous Mater. 2008, 116, 204–209;
c) D. Jiang, T. Mallat, F. Krumeich, A. Baiker, J. Catal. 2008, 257,
390–395; d) S. Turner, O. I. Lebedev, F. Schrçder, D. Esken, R. A.
Fischer, G. Van Tendeloo, Chem. Mater. 2008, 20, 5622–5627; e) F.
Schrçder, D. Esken, M. Cokoja, M. W. E. van den Berg, O. I. Lebe-
dev, G. Van Tendeloo, B. Walaszek, G. Buntkowsky, H.-H. Limbach,
B. Chaudret, R. A. Fischer, J. Am. Chem. Soc. 2008, 130, 6119–
6130; f) S. J. Yang, J. Y. Choi, H. K. Chae, J. H. Cho, K. S. Nahm,
C. R. Park, Chem. Mater. 2009, 21, 1893–1897.
[22] A. L. Spek, J. Appl. Crystallogr. 2003, 36, 7–13.
[23] F. Rouquerol, J. Rouquerol, K. Sing, Adsorption by Powders &
Porous Solids, Academic Press, San Diego, 1999.
[24] a) T. Dꢅren, F. Millange, G. Fꢆrey, K. S. Walton, R. Q. Snurr, J.
Phys. Chem. C 2007, 111, 15350–15356; b) K. S. Walton, R. Q.
Snurr, J. Am. Chem. Soc. 2007, 129, 8552–8556.
Thermal desorption spectroscopy: The degassing of hydrogen was studied
by low-temperature thermal desorption spectroscopy (TDS) using a spe-
cial apparatus described in detail elsewhere.[28] The typical mass used is
about 3 mg and the samples were out-gassed at a temperature of 440 K
under high vacuum for approximately 10 h prior to the first measurement
and between the measurements for at least 1 h. Then the samples were
exposed to a hydrogen atmosphere (purity 99.999%) of 700 mbar at
room temperature and then cooled down to 20 K. After about 30 min at
20 K, the sample cell was evacuated. Following this, the temperature was
increased with a linear heating rate (0.1 and 0.01 KsÀ1) from 20 to 370 K
and the desorbed gases were analyzed by using a quadrupole mass spec-
trometer. The total number of desorbed hydrogen atoms was derived for
each TDS spectrum, since the spectrometer was previously calibrated
with Pd.[28]
[25] P. I. Ravikovitch, A. V. Neimark, Colloids Surf. A 2001, 187–188,
11–21.
Acknowledgements
[26] J. Jagiello, M. Thommes, Carbon 2004, 42, 1227–1232.
[27] The van der Waals radii of C (0.17 nm) and Cl (0.175 nm) were used
to calculate the width of the apertures: A. Bondi, J. Phys. Chem.
1964, 68, 441–451.
Financial Support by the DFG (Priority Program SPP 1362 “Porous
Metal-organic Frameworks” and Sonderforschungsbereich 625 “Von ein-
Chem. Eur. J. 2011, 17, 1837 – 1848
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1847

D. Volkmer et al.
[28] B. Panella, M. Hirscher, B. Ludescher, Microporous Mesoporous
Mater. 2007, 103, 230–234.
[29] B. Panella, M. Hirscher, H. Puetter, U. Mueller, Adv. Funct. Mater.
2006, 16, 520–524.
[30] R. Vincent, P. A. Midgley, Ultramicroscopy 1994, 53, 271–282.
[31] M. C. Burla, R. Caliandro, M. Camalli, B. Carrozzini, G. L. Cascara-
no, L. De Caro, C. Giacovazzo, G. Polidori, S. Diligi, R. Spagna, J.
Appl. Crystallogr. 2007, 40, 609–613.
[33] W. Lasocha, K. Lewinski, J. Appl. Crystallogr. 1994, 27, 437–438.
[34] A. Altomare, M. Camalli, C. Cuocci, C. Giacovazzo, A. Moliterni,
R. Rizzi, J. Appl. Crystallogr. 2009, 42, 1197–1202.
[35] V. Petricek, M. Dusek, L. Palatinus, Jana2006. The crystallographic
computing system. Institute of Physics, Praha, Czech Republic, 2006.
[36] F. O. Mertens, Surf. Sci. 2009, 603, 1979–1984.
Received: May 10, 2010
[32] G. Sheldrick, Acta Crystallogr. Sect. A 2008, 64, 112–122.
Published online: January 12, 2011
1848
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 1837 – 1848