MOF Crystal Growth
DOI: 10.1002/anie.200905627
Time-Resolved In Situ Diffraction Study of the Solvothermal
Crystallization of Some Prototypical Metal–Organic Frameworks**
Franck Millange,* Manuela I. Medina, Nathalie Guillou, Grard Frey, Kathryn M. Golden, and
Richard I. Walton*
The crystallization of metal–organic framework (MOF)
materials
[1]
is an extremely attractive way in which to produce
functional solid materials with complex three-dimensional
structures, since an element of “design” is considered possible
in their synthesis: the idea being that by using chosen metal
polyhedral units linked by polydentate organic ligands with a
known coordination preference, a network structure of
desired connectivity may be formed.
[2,3]
This synthetic
approach is presently the focus of some considerable atten-
tion and is yielding many novel hybrid inorganic–organic
solids, often that possess some porosity on the nanoscale
suitable for applications, such as gas separation, molecular
sieving, and shape-selective catalysis.
[4]
Although these uses
are already well-established in silicate zeolite chemistry, there
are distinct advantages offered by the MOFs for uses under
mild conditions. For example, the choice of framework metal
may offer desirable binding sites for gases (currently the focus
of much attention in the topical areas of storage of hydro-
gen,
[5]
methane, or carbon dioxide
[6]
), the functionalization of
organic linkers (either pre- or post-synthesis) may allow
tuning of the porosity, reactivity, and the selectivity towards
binding of guest molecules,
[7–9]
and the use of chiral ligands
may result in chiral framework materials.
[10,11]
In addition,
MOF structures also often show great flexibility in the solid
state, giving them properties distinct from the traditional
inorganic zeotype materials.
[12–14]
To explore the extent to which new MOF materials may
be “designed” it is now important to elucidate the funda-
mental physico-chemical details of the crystallization of MOF
materials: knowledge of how complex extended network
structures are assembled from simple chemical precursors in
solution could ultimately permit some fine tuning of synthesis
conditions to test and realize the ideas of design in syn-
thesis.
[15]
Only a few such studies have been reported to date.
These include extended X-ray absorption fine structure
(EXAFS) spectroscopy studies of reactive solutions to
examine the presence of structural building units in solution,
through the amorphous intermediate to final crystalline
product;
[16]
light scattering from clear solutions to observe
the formation of colloidal nanocrystals;
[17,18]
and mass spec-
trometry to examine the interaction of Mg
2+
ions with (
+
)-
camphoric acid to identify possible building units for the
construction of a MOF.
[19]
Shoaee et al. recently used atomic
force microscopy (AFM) to examine a growing face of a
copper MOF after injection of a reactive solution and
suggested that the growth unit from solution was actually
smaller than the paddle-wheel-shaped building unit identified
in its crystal structure.
[20]
The study of MOF crystallization
mechanism has so far been concerned with the local structure
of solution species prior to the appearance of crystal order,
but it is important to examine crystal growth over all length
scales to build up a complete picture of crystallization.
Herein, we describe observations of the emergence of the
crystal order of MOFs from reactive solutions, above room
temperature, by using the time-resolved energy-dispersive X-
ray diffraction (EDXRD) method for two established tran-
sition-metal carboxylate MOFs. The technique has been used
successfully for the in situ study of the crystallization of a
variety of inorganic materials,
[21–26]
although to date it has not
been applied to the study of hybrid MOF materials. Its
advantage lies in the use of high intensity white beam X-rays,
which allows the non-invasive penetration of laboratory-scale
reaction vessels under elevated temperature and autogeneous
pressure. Thus the evolution of Bragg peaks as a function of
reaction conditions and time can be monitored with a time
resolution of less than 1 min.
Time-resolved EDXRD measurements were made on
Beamline F3 of the HASYLAB facility (DESY, Hamburg,
Germany). The first system we studied was the copper(II)
benzene tricarboxylate HKUST-1, [Cu
3
(BTC)
2
·solvent]
(where BTC = 1,3,5-benzene-tri-carboxylate, and the frame-
work contains occluded solvent in its as-made form).
[27]
This
phase crystallizes under solvothermal conditions from a clear
solution prepared by dissolving copper(II) nitrate hemi-
(pentahydrate) and trimesic acid in a mixture of water,
ethanol, and DMF, within a sealed quartz vessel. Figure 1
shows a surface plot of diffraction data during the crystal-
lization of the material at 125
8
C. A series of Bragg peaks, all
[*] Dr. F. Millange, Dr. M. I. Medina, Dr. N. Guillou, Prof. G. Frey
Institut Lavoisier, Universit de Versailles
UMR 8180, 78035 Versailles (France)
E-mail: franck.millange@uvsq.fr
K. M. Golden, Dr. R. I. Walton
Department of Chemistry, University of Warwick
Coventry, CV4 7AL (UK)
E-mail: r.i.walton@warwick.ac.uk
[**] We thank DESY for provision of beamtime at HASYLAB, and we are
grateful to the group of Prof. Dr. W. Bensch of the Christian-
Albrechts-Universitt zu Kiel, in particular Beatrix Seidlhofer and
Elena Antonova, for their assistance with use of Beamline F3 and
the loan of their heating device. The ESRF provided beamtime on
ID31 and we thank Dr. I. Margiolaki and Dr. A. Fitch for their
assistance with measuring data there. We are grateful to Nikos
Kourkoumelis, University of Ioannina, for modifying his PowDLL
program to allow analysis of the EDXRD data. This work was
supported by CNRS and French ANR “CONDMOFs” funding.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200905627.
Angewandte
Chemie
763Angew. Chem. Int. Ed. 2010, 49, 763 –766 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
indexed on the expected Fm
33m cubic unit cell
[27]
(with a =
26.34 ), appear simultaneously after no detectable induction
period and grow to their maximum intensity in around 30 min.
The in situ study was repeated at five temperatures to obtain
information about the kinetics of the crystallization: Figure 2
shows normalized crystallization curves produced by integra-
tion of the most intense (222) Bragg peak at each temper-
ature. (Integration of all the Bragg peaks was performed and
the crystallization curves from each were superimposable; see
Supporting Information.) Analysis of the crystallization
curves was performed using the Avrami–Erofe’ev model by
the method of Sharp and Hancock
[28]
(Figure 2). Although
this kinetic model is empirical in nature, it has been widely
applied to situations where a nucleation-growth process is
expected from a homogeneous medium, such as crystalliza-
tion from a solution or gel, and it allows rate constants to be
extracted and compared when reactions conditions are
varied.
[29]
The value of n, the Avrami exponent, is believed
to give some indication of the mechanism of crystallization, in
particular the balance of the rates of nucleation versus crystal
growth. We observe values of n close to 1.5 (Supporting
Information) suggesting that crystallization is controlled
largely by the formation of nucleation sites, rather than
diffusion of reactive species to the sites or crystal growth itself
at the sites. Zacher et al. studied the earliest stage of HKUST-
1 formation using light scattering
[18]
and also concluded that
crystallization was dominated by homogeneous nucleation
and growth, with nucleation extending over the time-scale of
their experiment (up to 30 min). It is also noteworthy that the
initial formation of nucleation sites is effectively instanta-
neous at all the temperatures studied, with no observation of
an induction time, yet it is the continued formation of
nucleation sites that dominates the crystallization rate.
Similar Avrami exponents have been found in the solution
crystallization of other inorganic materials, for example, the
liquid-phase reconstruction of layered hydroxides from
amorphous oxides,
[30]
the hydrothermal crystallization of
barium titanate,
[31]
and the formation of mixed tungsten-
molybdenum oxides.
[32]
In the case of silicate zeolites, values
of n > 2 have typically been reported, but the difference is
that zeolites often crystallize from viscous gels rather than
from clear solutions.
[24,33]
An Arrhenius plot (Supporting
Information), gave an activation energy of 73.3 kJ mol
1
. This
value is larger, yet comparable, to the nucleation-controlled
crystal growth of other types of inorganic solids from hydro-
thermal solutions, such as barium titanate (55 kJmol
1
)
[31]
and
Mg-Al hydroxide (41 kJmol
1
).
[30]
The second system we studied was the iron(III) tereph-
thalate with the MIL-53 structure, [Fe
III
(OH,F){O
2
C-C
6
H
4
-
CO
2
}·H
2
O].
[34]
This material has a one-dimensional channel
system made up of trans linked octahedral Fe units, cross-
linked by the dicarboxylate. Crystallization of MIL-53 takes
place from a clear solution produced by dissolving iron(III)
chloride hexahydrate, terephthalic acid, and hydrofluoric acid
in DMF heated at atmospheric pressure under reflux. Figure 3
shows 3D plots of time-resolved EDXRD data measured at
150
8
C: the ultimate product is the expected MIL-53 but its
crystallization is preceded by the transient appearance of
another crystalline phase. If the temperature is lowered to
125
8
C the intermediate phase has a longer lifetime and at
100
8
C it is the sole product after 6 h, although on extended
heating the expected MIL-53 product is eventually found.
Quenching experiments allowed the intermediate to be
isolated as a highly crystalline powder suitable for structure
elucidation by high-resolution powder diffraction (see Sup-
porting Information). This analysis reveals that the inter-
mediate material is related to a previously reported phase,
MOF-235,
[35]
in which trimers of iron(III) oxy-octahedra are
linked by terephthalate ligands, and trapped within the
porous three-dimensional network are not only DMF solvent
Figure 1. Time-resolved in situ EDXRD data measured during the
crystallization of the copper carboxylate HKUST-1 at 125
8
C. The Bragg
peaks are indexed on an F-centered cubic unit cell with a = 26.34 .
Inset: view of the structure of HKUST-1 with five-coordinate Cu as
pink polyhedra, blue C, red O.
Figure 2. Kinetic analysis of the crystallization of HUKST-1: plot of
extent of crystallization (a) curves against time (t) obtained by
integration of the 222 Bragg peak in the EDXRD data. Inset: analysis
by the method of Sharp and Hancock to test fitting to the
Avrami–Erofe’ev nucleation-growth crystallization model
a = 1exp{(k(tt
0
))
n
} with lines that are the result of linear
regression analysis.
Communications
764 www.angewandte.org 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2010, 49, 763 –766
molecules but also FeCl
4
anions that balance the positive
charge of the MOF framework (Figure 4). The topology of
this intermediate is unrelated to the final MIL-53 product,
indeed the building units of the structure are completely
different, which suggests that the conversion of one into
another is most unlikely to occur by a solid-state rearrange-
ment: it is probable that the first phase dissolves, releasing
reactive solution species for crystallization of the final
product.
In summary, for two transition-metal carboxylate MOFs
we see two distinct solvothermal crystallization scenarios:
classical nucleation-growth kinetics from solution, or crystal-
lization via a metastable precursor. This illustrates the
complexity of MOF crystallizations and shows how in situ
measurements are crucial for a further understanding of their
synthesis. These measurements would be of great benefit in
the study of more complex MOF systems; for example, in
understanding the competitive formation of several phases,
studying the effect of ligand modification, and for efficiently
assessing reaction conditions (solvent, pH, temperature etc.)
for the discovery of new materials, as well as allowing some
deeper physical understanding of how the complex solids are
formed. Future work must also include the development of
better kinetic models, which relate to the chemical trans-
formations actually taking place during crystallization,
[29]
and
also the use of combined techniques for following crystal-
lization over several length scales simultaneously.
[26,32]
Received: October 7, 2009
Published online: December 16, 2009
.
Keywords: crystal growth · metal–organic frameworks ·
structure elucidation · X-ray diffraction · zeolite analogues
[1] For an introduction to the fast-developing area of MOF
chemistry see a special issue of Chemical Society Reviews
published in 2009 (Volume 38, Issue 5): J. R. Long, O. M. Yaghi,
Chem. Soc. Rev. 2009, 38, 1213.
[2] D. J. Tranchemontagne, J. L. Mendoza-Corts, M. OKeeffe,
O. M. Yaghi, Chem. Soc. Rev. 2009, 38, 1257.
[3] J. J. Perry IV, J. A. Perman, M. J. Zaworotko, Chem. Soc. Rev.
2009, 38, 1400.
[4] A. U. Czaja, N. Trukhan, U. Mller, Chem. Soc. Rev. 2009, 38,
1284.
[5] L. J. Murray, M. Dinca
ˇ
, J. R. Long, Chem. Soc. Rev. 2009, 38,
1294.
[6] J.-R. Li, R. J. Kuppler, H.-C. Zhou, Chem. Soc. Rev. 2009, 38,
1477.
[7] R. A. Fischer, C. Woll, Angew. Chem. 2008, 120, 8285; Angew.
Chem. Int. Ed. 2008, 47, 8164.
[8] K. K. Tanabe, S. M. Cohen, Angew. Chem. 2009, 121, 7560;
Angew. Chem. Int. Ed. 2009, 48, 7424.
[9] D. Farrusseng, S. Aguado, C. Pinel, Angew. Chem. 2009, 121,
7638; Angew. Chem. Int. Ed. 2009, 48, 7502.
[10] D. Bradshaw, J. B. Claridge, E. J. Cussen, T. J. Prior, M. J.
Rosseinsky, Acc. Chem. Res. 2005, 38, 273.
[11] C. Livage, N. Guillou, P. Rabu, P. Pattison, J. Marrot, G. Frey,
Chem. Commun. 2009, 4551.
[12] C. Serre, F. Millange, C. Thouvenot, M. Nogues, G. Marsolier, D.
Louer, G. Frey, J. Am. Chem. Soc. 2002, 124, 13519.
[13] S. Kitagawa, K. Uemura, Chem. Soc. Rev. 2005
, 34, 109.
[14] F. Millange, C. Serre, N. Guillou, G. Frey, R. I. Walton, Angew.
Chem. 2008, 120, 4168; Angew. Chem. Int. Ed. 2008, 47, 4100.
[15] R. E. Morris, ChemPhysChem 2009, 10, 327.
[16] S. Surbl, F. Millange, C. Serre, G. Frey, R. I. Walton, Chem.
Commun. 2006, 1518.
[17] S. Hermes, T. Witte, T. Hikov, D. Zacher, S. Bahnmuller, G.
Langstein, K. Huber, R. A. Fischer, J. Am. Chem. Soc. 2007, 129,
5324.
[18] D. Zacher, J. N. Liu, K. Huber, R. A. Fischer, Chem. Commun.
2009, 1031.
[19] J. A. Rood, W. C. Boggess, B. C. Noll, K. W. Henderson, J. Am.
Chem. Soc. 2007, 129, 13675.
[20] M. Shoaee, M. W. Anderson, M. R. Attfield, Angew. Chem.
2008, 120, 8653; Angew. Chem. Int. Ed. 2008, 47, 8525.
[21] J. Munn, P. Barnes, D. Hausermann, S. A. Axon, J. Klinowski,
Phase Transitions 1992, 39, 129.
[22] R. J. Francis, S. J. Price, J. S. O. Evans, S. OBrien, D. OHare,
S. M. Clark, Chem. Mater. 1996, 8, 2102.
Figure 3. Time-resolved EDXRD measured during the crystallization of
MIL-53 at 150
8
C, with a crystalline transient phase seen at short
reaction times.
Figure 4. A view of the structure of transient intermediate phase seen
during MIL-53 crystallization, isolated by quenching. The MOF frame-
work is constructed from trimers of iron(III) octahedra linked by
terephthalate linkers (ball-and-stick model) and contains occluded
FeCl
4
tetrahedral units. Extra-framework solvent molecules have been
omitted for clarity. This is a projection viewed along the c axis.
Angewandte
Chemie
765Angew. Chem. Int. Ed. 2010, 49, 763 –766 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org
[23] R. I. Walton, T. Loiseau, D. OHare, G. Frey, Chem. Mater.
1999, 11, 3201.
[24] R. I. Walton, F. Millange, D. OHare, A. T. Davies, G. Sankar,
C. R. A. Catlow, J. Phys. Chem. B 2001, 105, 83.
[25] R. Kiebach, N. Pienack, M. E. Ordolff, F. Studt, W. Bensch,
Chem. Mater. 2006, 18, 1196.
[26] G. Sankar, T. Okubo, W. Fan, F. Meneau, Faraday Discuss. 2007,
136, 157.
[27] S. S. Y. Chui, S. M. F. Lo, J. P. H. Charmant, A. G. Orpen, I. D.
Williams, Science 1999, 283, 1148.
[28] J. D. Sharp, J. H. Hancock, J. Am. Ceram. Soc. 1972, 55, 74.
[29] E. E. Finney, R. G. Finke, Chem. Mater. 2009, 21, 4692.
[30] F. Millange, R. I. Walton, D. OHare, J. Mater. Chem. 2000, 10,
1713.
[31] R. I. Walton, F. Millange, R. I. Smith, T. C. Hansen, D. OHare, J.
Am. Chem. Soc. 2001, 123, 12547.
[32] R. Kiebach, N. Pienack, W. Bensch, J. D. Grunwaldt, A.
Michailovski, A. Baiker, T. Fox, Y. Zhou, G. R. Patzke, Chem.
Mater. 2008, 20, 3022.
[33] A. T. Davies, G. Sankar, C. R. A. Catlow, S. M. Clark, J. Phys.
Chem. B 1997, 101, 10115.
[34] F. Millange, N. Guillou, R. I. Walton, J. Grenche, I. Margiolaki,
G. Frey, Chem. Commun. 2008, 4732.
[35] A. C. Sudik, A. P. Cote, O. M. Yaghi, Inorg. Chem. 2005, 44,
2998.
Communications
766 www.angewandte.org 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2010, 49, 763 –766
