Upload
q12werty
View
12
Download
0
Embed Size (px)
DESCRIPTION
This critical review focusses on a strange behaviour of crystallized solid matter: its reversibleswelling with large magnitude.
Citation preview
This article was published as part of the
2009 Metal–organic frameworks issueReviewing the latest developments across the interdisciplinary area of
metal–organic frameworks from an academic and industrial perspective Guest Editors Jeffrey Long and Omar Yaghi
Please take a look at the issue 5 table of contents to access the other reviews.
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online / Journal Homepage / Table of Contents for this issue
Large breathing effects in three-dimensional porous hybrid matter:
facts, analyses, rules and consequencesw
Gerard Ferey*ab and Christian Serreb
Received 2nd October 2008
First published as an Advance Article on the web 26th February 2009
DOI: 10.1039/b804302g
This critical review focusses on a strange behaviour of crystallized solid matter: its reversible
swelling with large magnitude. This will be of interest for experts in porous solids but also for
solid state chemists and physicists. Some examples, classified according to the dimensionality of
the inorganic subnetwork, present the general requirements and the structural rules which govern
the existence of this phenomenon. Its consequences concern specific applications related to
sensors, energy savings, sustainable development and health (100 references).
1. Introduction
Breathing is associated with life, at the macroscopic level and
at the molecular scale as well. Biological matter is known for a
long time to imply molecular movements with large magnitude
during biochemical reactions, and can be mimicked by clever
organic chemists.1,2 On the contrary, inorganic matter is often
associated with rigid bodies and a quasi-immobility except
during phase transitions in the solid state.
However, some recent papers run counter to these pre-
conceived ideas and describe solids which, under an external
stimulus (temperature, pressure, chemical inclusion. . .),
exhibit a very important flexibility, while retaining the same
or similar topologies. The atomic displacements reach some-
times several A. This paper will analyze some of these solids
and try to explain the different origins of such flexibility. We
shall call this phenomenon ‘breathing’3–6which, according to
other authors, is also labelled as ‘dynamic frameworks’,7
‘springlike’,8 ‘sponge like’9,10 or ‘accordion’11 effects.
Breathing is associated with reversible movements between
two states corresponding to expansion and contraction,
respectively. It is obvious for humans who inhale oxygen
and breathe out carbon dioxide. However, as seen below,
two situations are possible: many times, expansion is
associated with an input and contraction with an output but
the inverse phenomenon can occur: expansion 3 output and
contraction3 input. Whatever the sense of the breathing, the
external stimulus generates the movement.
An important point of the breathing is also its magnitude.
Generally, in solid state matter, the movements are very small
(o0.8 A) and often occur during displacive (without bond
breaking) phase transitions. A good example is provided by
the distortions of the perovskites12 or the FeF3 structure
(Fig. 1) with the ReO3 type structure.13 The rhombohedral
room temperature variety (contracted form) corresponds to
the tilting of octahedra represented in Fig. 1(a). It transforms
reversibly by heating at 400 1C into the ideal cubic ReO3 type,
a Institut universitaire de France, France.E-mail: [email protected]. E-mail: [email protected]
b Institut Lavoisier, (CNRS 8180), University of Versailles, 45,Avenue des Etats-Unis, F-78035 Versailles Cedex, France
w Part of the metal–organic frameworks themed issue.
Gerard Ferey
Gerard Ferey received his PhDfrom Paris VI University in1977. He was Professor ofInorganic Chemistry firstin Le Mans University(1981–1996) and then inVersailles University wherehe created the InstitutLavoisier (1996–. . .). He isnow Professor at Institutuniversitaire de France andmember of the French Acad-emy des Sciences. He has re-ceived many internationalAwards. After working on themagnetism of transition metal
inorganic fluorides, his current interests concern the structuralchemistry of inorganic and hybrid porous solids, their mechan-isms of formation and their applications in gas storage, energy,drug delivery and nanosciences.
Christian Serre
Christian Serre, 38 years old,is an engineer from the EcoleSuperieure de Physique et deChimie Industrielles de Paris.He obtained a PhD inInorganic Chemistry in 1999.After a post-doctoral fellow-ship in the USA, he moved toa permanent CNRS researchposition in 2001 in the group ofProf. Ferey in Versailles. Heis currently working on thesynthesis, structure determina-tion and applications of poroushybrid solids. He received theCNRS bronze medal in 2006
and a European research council young researchers’ grantin 2008.
1380 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
CRITICAL REVIEW www.rsc.org/csr | Chemical Society Reviews
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
showing no tilting, after a cooperative movement of all the
fluoride ions of 0.56 A (Fig. 1(b)).
This phenomenon, despite being general in inorganic solid
state chemistry and in zeolites,14 implies however too small
atomic displacements for being further considered in this
paper, devoted exclusively to large breathing effects in three-
dimensional organic–inorganic hybrid solids containing only
strong bonds. For this reason, the purely organic or bio-
organic compounds, which exhibit similar displacements15,16
and are governed by hydrogen bond interactions between
molecular species, will not be considered here. In a similar
way, two-dimensional hybrids are also excluded. Only the
layers are hybrid. Guests are incorporated between the sheets,
leading to a rich guest exchange chemistry, but the guest
extraction can render (reversibly or not9) the solid amorphous,
due a disordered arrangement of the layers to a (turbostratic
disorder).
Large breathing effects represent a new aspect of non-
organic matter. From the description of some flexible solids
and of some counter examples, this paper will analyze the
structural reasons of breathing and try to generate some
building rules for obtaining such swelling phases. In a final
part, the consequences of this breathing effect will be examined
in several domains such as energy, energy savings, sensors,
molecular recognition, and gas storage for sustainable devel-
opment and health.
2. General requirements for breathing in
crystalline hybrid solids
The need of a chemical or physical stimulus for generating
reversible movements was already noticed. From the
structural point of view, the flexibility inherent to breathing
implies also a contrast in the strength, the directivity and the
nature of the different bonds existing in the solid. Such effects
are particularly evident in hybrid solids. In their skeleton,
these compounds generally associate ionic or ionocovalent
bonds in the inorganic part, and covalent and sometimes
p–p ones in the organic moieties. Hydrogen or van der Waals
interactions18,19 occur between the framework and the
occluded species.
The spatial occupancy of the different parts of the hybrid
solid plays a role. Indeed, the inorganic part is often associated
with a rigid behaviour, whereas the organic moieties corres-
pond to flexibility. The amount of each component in the
final solid will influence the degree of breathing of the solid.
Therefore, one must consider not only the whole dimension-
ality of the hybrid solid, but also the dimensionality of each
subnetwork. In a first step, if one excepts the case of molecular
solids, two groups of three situations occur with non-
interpenetrated solids.
� for three-dimensional frameworks, the inorganic subnet-
work can be either two-, one ot zero-dimensional and linked
by organic moieties in one, two or three directions.
� the same situation occurs for the inorganic contribution in
two-dimensional solids, but this time, the organic linkage
occurs only within the layers, the space between the latter
being occupied by the occluded species (solvents,
templates. . .). Hydrogen and p–p bonds are involved, and
have been discussed in papers of Kitagawa.17
Fig. 1 Perspective view of (a) rhombohedral and (b) cubic FeF3
which correspond to the contracted and expanded forms of this solid
(see text).
Fig. 2 The six classes of Kitagawa. In all the classes, G is the symbol for Guest. In the 1D class, the voids between the chains are occupied by small
molecules and can exhibit ion exchange. In the first case of 2D class, the manner of stacking of the layers (superimposed or shifted) is strongly
dependent on the nature of the guest and the weak interactions they have with the layers. In the second case, the interdigitated layers are
superimposed and form 1D channels. Closed without guests, they open with some of them, resulting in an elongation of the stacking parameter. In
the 3D cases, three situations occur. When pillared layers are concerned, the reversible phenomenon of interlayer elongation and shortening is
realized by non-rigid pillars. The expanding and shrinking frameworks act as sponges. Keeping the same topology, the drastic volume change is
induced by strong host–guest interactions. Depending on the structure, the volume increase is associated with either the evacuation or the inclusion
of the guests. Finally, in the case of interpenetrated grids, they are densely packed in the absence of guests and the introduction of molecules
generates a sliding of one network.
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1381
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
A special situation concerns the interpenetrated or inter-
woven networks17,18,20,21 which must take also into account
the interactions between the two (or more) framework contri-
butions. All these cases led Kitagawa17 to propose six classes for
the breathing of dynamic coordination polymers (Fig. 2)
A third requirement obviously concerns the need of a free
space in the structure (which is often realized for hybrid
porous solids) in order to accommodate the modifications of
the steric hindrance of the flexible part during the movements.
Finally, and related to the contrast between the nature of
the bonds (ionic, covalent. . .), non-rigid areas (hereafter
denoted as ‘weak points’) must exist in the solid to allow the
breathing modes. Even if it is not obvious, their existence is
necessary but not sufficient for generating breathing, as seen
later. The topology of some elements of the structure also
plays a crucial role. All these requirements will be illustrated in
the following examples.
3. Structural analysis of some illustrative examples
of three-dimensional networks
3.1 Two-dimensional inorganic subnetworks. The case of
pillared c-zirconium phosphates
The interesting two forms, a and g, of hydrated zirconium
phosphates (often labelled a- and b-ZrP) have been the subject
of many studies and applications since their discovery thirty
years ago (see for instance the references contained in refs. 11
and 22–25) They are lamellar solids with guest water molecules
within the layers, and correspond to type a-1 in the Kitagawa’s
classification.17 The layers are built from the dense corner-
sharing connection of zirconium octahedra and phosphorus
tetrahedra. For our purpose, g-ZrP or Zr2(HPO4)4�4H2O
(Fig. 3(a) and (b)) is of particular interest. Indeed, Alberti11
and Clearfield25 proved that the phosphate groups of the
layers could be substituted, totally or partially, by mono- or
diphosphonate groups (linking two layers) with variable
length of aliphatic chains. Alberti11 alone looked at
the thermal behaviour of these phases which were, to
our knowledge, the first to exhibit such a drastic breathing
(B8 A in the best case).
The study of the system Zr/phosphorous acid/
alkanediphosphonic acids/acetone/water at 350 K lead
to a series of phases of general formula ZrPO4[O2P(OH)2]1�x-
[O2P(OH)–(CH2)n–(HO)PO2]x/2�mH2O with 4 r n r 16
corresponding to the number of carbons in the alkyl chain;
x was allowed to vary from 0 to 1, x = 0 corresponding to
g-ZrP, x = 1 to the fully pillared Zr diphosphonate, and
intermediate values between 0 and 1 to an increasing degree of
pillaring between the layers of g-ZrP. Two examples of
complete pillaring are shown in Fig. 3(c) (for n = 4) and
Fig. 3 (a) Polyhedral projection of the structure of g-ZrP along [010] and (b) along [001]; (c) and (d) [010] projections of the completely pillared Zr
1,4- and 1,10-alkane diphosphonates, (e) a schematic view of the fully hydrated phosphate/diphosphonate and (f) of the corresponding dehydrated
sample showing the contraction of the carbon chain. Zr octahedra are in pale blue, phosphate groups in yellow, water molecules in dark blue,
OH groups in red and white and the carbons in black.
1382 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
Fig. 3(d) (for n = 10) and the distance between the layers
increases linearly following the law d001 = 0.93 + 0.127n sin a(a being the angle �52.71 in this family—between the alkyl
chain and the plane of the layer). In the case of full pillaring,
three features are noteworthy: (i) the configuration of the
chains corresponds to the most stable arrangement (maximum
elongation), (ii) the distance between the chains is fixed by the
distance separating the phosphorus atoms within the same side
of the layer (ca. 5.4 A) and therefore (iii) the steric hindrance
of the chains authorizes neither degrees of freedom for the
conformation of the chains nor the insertion of additional
water molecules, since m, which represents the number of H2O
molecules per Zr remains 1. As a consequence, the fully
pillared series do not exhibit any flexibility.
Breathing occurs (Fig. 3(e) and (f)) for low values of x. They
correspond to situations where two different alkanediphos-
phonic chains are separated, in a direction parallel to the
layer, by [O2P(OH)2] groups. For a given n, either when the
solids are wet or exposed to an atmosphere with 100% of
relative humidity (r.h.) (which represents the stimulus at
300 K)), the distance is, whatever x, the same as for the
completely pillared material. In this case, additional water
molecules take the place of the missing chains, which explains
why m is important for low x (for n= 10, m= 3.0 for x= 0.1
instead of m = 1 for x = 1). Their role implies, at this stage,
only space filling up to an amount which preserves the
maximum stability of the carbon chain.
However, a thorough study of the term n = 10 proves that,
varying the r.h. conditions and/or temperature, efficient
breathing occurs mainly for low values of x (Fig. 4(a) and
(b)). Compared to the fully hydrated sample for r.h. = 100%,
the decrease of r.h. to 75% leads, for instance for x= 0.1, to a
small (2.6 ) 2.4) but significant decrease of m and, correla-
tively, an important decrease of the d001 distance (19.2 A )14.1 A). For the same x, the completely dehydrated sample
(which can be considered as either the 0% r.h. or the result of
the stimulus of temperature), corresponds to the following
changes: 2.6 ) 1.0 for m and 19.2 A ) 11.2 A (decrease
of 8 A!) for d001. Rehydration of the dehydrated solid
regenerates the initial fully hydrated sample with all its
specifications. Fig. 4(a) and (b) show the evolution of the
different characteristics of breathing as a function of the
pillaring percentage x in this series.
This example provides a first analysis of the reasons for such
a large breathing, whereas the structural topology remains the
same. Not only the stimulus is necessary but also its intensity
(in this case, water vapor pressure). Keeping the dense and
rigid arrangement of the layers, breathing (here uniaxial)
indeed occurs only for low values of the pillaring percentage x.
It is noteworthy that a threshold around 66%maximum seems
to be necessary for such an effect. Below, this corresponds
to the creation of enough space for the mobility of species and,
therefore, a flexibility up to a certain amount, depending on
the intensity of the stimulus.
Two features are responsible of these movements: (i) as soon
as the carbon chains have sufficient space between them, they
can develop their extraordinary plasticity due to the covalent
bonding and the free rotation around the axis of the bonds.
The torsion of these chains by increasing the cis-positions can
lead to variations in length of more than 50% even if,
thermodynamically, they are unstable as such. In the case of
the Alberti’s phosphonates, and even if no structural determi-
nation was given in their paper, the evolution of the d001distance between the fully hydrated and the dehydrated phase
can be explained by an helicoidal arrangement of the chain. It
is indeed easy to calculate26 that, for n = 10, such a helix coils
around a cylinder with a radius of 1.08 A, the pitch of the helix
being 0.668 A. The length between the two terminal carbons of
the chains is then 6.012 A (instead of 11.2 A for the elongated
chain) and explains the 19.2 A ) 11.2 A decrease for d001.
The ‘weak point’ is then the carbon chain which acts as a
spring, which is compressed in the anhydrous form and
elongated for the fully hydrated one. This spring effect (ii) is
allowed because the water molecules are only loosely bound to
the skeleton and can move easily in the structure and give
enough space to the chain for its contraction.
3.2 One-dimensional inorganic subnetworks
3.2.1 The case of trivalent metal terephthalates (MIL-47
and -53)3–6. This situation was recently discovered in the
laboratory3,4 during the search of functionalized hybrid
nanoporous solids with large pores.27,28 It concerns metal
1,4-benzenedicarboxylates (BDC) [M(X)[BDC]�xG (M is
VIV, with X = O for MIL-47, and trivalent Al, Cr, Fe, Ga
and X = OH, F for MIL–53). The templates (G) are
terephthalic acid H2BDC, water or solvents such as DMF.
Fig. 4 (a) For n = 10, evolution of the number m of occluded water
molecules per Zr atom and of the decrease of the d001 parameter
between the layers vs. the pillaring percentage x for different values of
r.h. (b) Evolution of the d001 parameter vs. x and different situations of
conditioning of the samples (n = 10).
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1383
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
At variance to the others, MIL-47 does not breathe. Indeed,
there are no OH groups in the skeleton in MIL-47, and it will
be shown below that these hydroxyl groups play a major role
in MIL-53 for the creation of symmetrical hydrogen bonds
with the oxygens of the occluded water.
At variance to MIL-47, MIL-53 breathes. According to the
different treatments, three variants of MIL-53 (hereafter
labelled -as, -ht and -lt) appear. They have in common a
three-dimensional network composed of corner-shared trans
chains of metallic octahedra (the OH groups being the shared
species) linked in the two other directions by terephthalate
ions. This determines large unidimensional lozenge-based
tunnels (Fig. 5(a)). In terms of nets, the topology of MIL-53
corresponds to a 44 lattice.
In the as-synthesized form MIL-53-as (V = 1440 A3) the
tunnels are occupied by structurally disordered molecules of
neutral terephthalic acid (Fig. 5(b)). They leave the structure
by heating at 573 K, and create the structure MIL-53-ht, with
empty tunnels (V = 1486 A3). By cooling in air, the structure
reabsorbs water. This leads to the third form MIL-53-lt
(V = 1012 A3). This transition is fully reversible when water
is involved, but the structure shows a very high selectivity
toward other adsorbable species. Indeed, whereas MIL-53-ht
is strictly inactive for the readsorption of acetone and ethanol
even after 4 days, it immediately adsorbs DMF in a reversible
way. Moreover, in MIL-53-lt, water is easily exchanged by
DMF, showing the great affinity of DMF for this type of
structure. The resulting product has cell parameters inter-
mediate between MIL-53-as and MIL-53-lt, proving the
adaptability of the breathing network to the shape of the
chemical stimulus. Note that the introduction of large
amounts of some gases leads to the largest cell (see section 4.2).
The structures of the three forms, solved from X-ray powder
diffraction, allow to quantify the large breathing effects during
the transformations, characterized, at constant topology, by
movements of large amplitude each time, created when either
adsorbed species or temperature act as stimuli.
For MIL-53(Cr) [X: O,F], the disappearance of the template
H2BDC during the transition MIL-53-as ) MIL-53-ht
corresponds to an increase by 3.16% of the volume of the
cell (Fig. 5(c)) with significant variations of two of the cell
parameters (17.34 A ) 16.73 A and 12.18 A ) 13.04 A). The
third, which corresponds to the direction of the metallic
octahedral chains, remains invariant (B6.8 A). The breathing
out paradoxically relates to an expansion of the cell instead of
the expected contraction. During the transition MIL-53-ht )MIL-53-lt, the adsorption of water (the inhaling step) corres-
ponds (Fig. 5(d)) to a drastic decrease of the cell volume
(�32%) associated this time with very large variations of the
two parameters involved above (16.7 A ) 19.7 A (Dd: 3 A)
and 13.0 A ) 7.85A (Dd: 5.15 A)).
The reasons for such changes are not the same as for the
diphosphonates. Indeed, now, if the rigidity of the inorganic
moieties remains almost the same, the sp2 hybridization in the
phenyl rings renders the carbon skeleton rigid. The area of
flexibility of the structure, the ‘weak point’, is restricted to the
connection between the inorganic chain and the carboxylic
function of the terephthalate (Fig. 5(e) and (f)). The two
oxygens of the latter are covalently bound to carbon but are
linked to chromium via a ionocovalent bond. This means
a possible rotation around the O–O axis (which acts as a
‘kneecap’) of the two planes O–Cr–Cr–O and O–C–O
(Fig. 5(f)). The corresponding dihedral angle a between these
two planes (177.51 for -as, 1801 for -ht and 1391 for -lt)
characterizes mechanically the extent of breathing better than
the angles between two chains, depicted in Fig. 5(b)–(d). It is
of note that the most expanded structure corresponds to a
dihedral angle of 1801.
This potential flexibility is activated by the presence of
inserted species in the tunnels and the creation of weak bonds
between the guest and the skeleton. Here, the oxygens of the
chromium octahedra are involved. The onset of these addi-
tional bonds decreases the distance between the chains along
the short diagonal of the lozenge. They are separated by 13 A
in the expanded empty MIL-53-ht (Fig. 5(c)). The distance is
lowered by 0.9 A in MIL-53-as when the template H2BDC is
present and the shrinkage reaches 5.2 A in MIL-53-lt when
fixed water molecules are involved and create a ‘lock-in’
between the chains, assimilating inhalation with contraction.
The strong rotation around the O� � �O axis of the carboxylate
functions simultaneously induces some small changes for the
tilting angle M–(OH)–M in the chains, as proved by NMR
and IR measurements. Therefore, the rigidity of the inorganic
moiety is less strict for a 1D subnetwork than it was for a 2D
Fig. 5 (a) Perspective view of MIL-53-ht; (b)–(d) projection along the
direction of the tunnels of (b) MIL-53-as (with some atoms of the
disordered terephthalic acid in blue), (c) MIL-53-ht, and (d) MIL-53-lt
with variable parameters; (e), (f) perspective views of the connection
between the chromium chains and the terephthalate ions. Chromium
octahedra are in green, water molecules in dark blue, OH groups in
pale blue, oxygens in red and carbons in black.
1384 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
one. This allows more degrees of freedom for this type of
breathing which is, at variance to the first example, essentially
two-dimensional.
Moreover, in situ solid-state 13C, 1H and 27Al NMR
experiments of the Al-homologue5 during rehydration have
shown that the water molecules show three types of hydrogen
bonds (Fig. 6). The first corresponds to a guest–guest inter-
action of the water molecules along the axis of the tunnels. The
two others relate to host–guest interactions between (i) the
oxygen of the guest with the OH group of the host alterna-
tively with one chain and the other and (ii), between one
hydrogen of the water molecule and one oxygen of the
carboxylate linked to Al. The latter is clearly evidenced
by 13C NMR experiments. The COO contribution of the
dehydrated sample progressively disappears during hydration
and is replaced by the COO–H2O signal, which exists alone for
the fully hydrated sample. 27Al NMR also confirms the
structural results and the small distortions and tiltings of the
Al(OH)2O4 octahedra occuring during hydration.
The above interactions seemed to be sufficient for explaining
the breathing phenomenon but, very recently, the study of the
Fe(III) analog, with the same topology, provided surprising
results.29 Indeed, whereas its hydrated form is identical to the
other ones, the full dehydration of MIL-53(Fe) increases
the shrinkage of the structure in two subtle steps, instead of
the expansion of the others. This means that, beside the above
interactions, host–host interactions must also be considered.
The comparison of the various distances observed in all the
hydrated forms does not allow to find a geometrical criterion
able to explain the above difference.
Anyhow, two possible reasons could be assumed for the
moment: (i) the influence of the nature of the metal and
(ii) strong p–p interactions. In most cases, the cation is a 3d
transition metal and magnetic dipolar interactions between the
chains may be taken into account. These interactions are
proportional to the number of unpaired d electrons. While
they can be considered as very weak with chromium(III) (d3)
and vanadium (d2), they become significant with iron (d5) and
may participate in the stability of the structure. In the case of
MIL-53(Fe), this could explain that they are strong enough for
keeping the structure closed in the absence of water within the
tunnels. However, even if it is possible, the argument is
unlikely due to the long distance between the closest Fe(III)
along the short diagonal of the tunnel (7.64 A (hydrated) and
6.87 A (anhydrous)).
The distances between the phenyl rings of the terephthalates
can be used to estimate the strength of p–p interactions.
Usually, they are considered as strong when the smallest
distances between the rings lie in the range 2.9–3.5 A. Here,
they are in the range 3.7 (Fe)–4,1 A (Al, Cr) for the whole
family of MIL-53 hydrated solids. The significantly smaller
distance in hydrated MIL-53(Fe) (Dd E0.4 A) could explain
why the anhydrous MIL-53(Fe) remains closed. In such a case,
this would imply that it corresponds to the strongest host–host
interaction in this topology. The evolution of the short ring–
ring distances in MIL-53(Fe) [3.73 A (hydrated form) and
3.41 A when anhydrous], indicates an enhancement of the p–pinteraction when the steric influence of the occluded water
disappears. It is probably sufficient for keeping the structure
closed. On the contrary, the larger distances observed for
the Al and Cr solids (4–4.1 A), allow the opening of the
structure upon dehydration. The p–p interactions obviously
play a role and a threshold seems to exist in the series of
values, below which opening is forbidden. This is confirmed
by MIL-69,30 an aluminium naphthalene-dicarboxylate
AlIII(OH)[O2C–C10H6–CO2], with the same topology, with
naphthalene rings replacing phenyl ones. It also remains
closed after dehydration (shorter ring–ring distances 3.78 A).
3.2.2 The structural counter example of vanadium
terephthalate MIL-68. and the influence of the topology. Besides
MIL-53, another three-dimensional vanadium(III) phase
(MIL-686) appears in the same system, when using a non-
aqueous solvent. It has the same chemical formula as MIL-53:
MIII(OH)[BTC]�xG but its density is ca. 75% of that of
MIL-53. It represents a lacunar variant (25% vacancies)
of MIL-53 (Fig. 7). In terms of nets, MIL-68 is 6.3.6.3. instead
of 44 for MIL-53. According to O’Keeffe and Hyde,31 no direct
transition can occur between these two nets. However, both
structures are formed with the same octahedral chains and
linked in the same way by the same ligands. Only the topology
is different: MIL-68 contains hexagonal and triangular tunnels
parallel to the metallic chains, with angles being strictly 120 and
601, respectively, instead of lozenge sections in MIL-53.
The thermodiffractometry of MIL-68 shows that the cell
remains unchanged, whatever the temperature, up to 350 1C,
which is the temperature of collapse of the structure. This
time, the absence of breathing has only one reason: the
existence of triangular cycles, already known to be strictly
rigid. They forbid any flexibility at the weak points of the
previous type of structure and MIL-68 does not breathe. This
example illustrates the role of the whole topology on the
possibility of breathing.
As a first conclusion of this part, breathing effects result
from the combination of several factors: (i) the existence of
only even cycles in the structure and of ‘weak points’, which
allow the structure to distort under the action of a guest
stimulus, and (ii) the occurrence of strong guest–guest and
host–guest interactions, stronger than host–host ones. The
Fig. 6 (a) Pespective view of the disposition of the water molecules
within the tunnel of MIL-53-lt with the hydrogen bonds.
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1385
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
latter, if the strongest, prevent the opening of the structure.
This point will be later developed in section 3.4.
3.3 Zero-dimensional inorganic subnetworks. Examples and
counter examples
For such a type of dimensionality, it is now common to speak
about porous coordination polymers (PCP), a continuously
expanding domain. In such solids, the inorganic part has
finite dimensions and corresponds either to single metallic
polyhedra or to oligomeric clusters. The dynamic effects have
already been documented by the groups of Kitagawa
and Rosseinsky from the experimental point of view, with
numerous examples,7,32–36 and the reader is recommended to
return to the reviews and the references they contain.
Without going into detail, one can distinguish two types of
flexibility resulting from (i) displacive phase transitions in
which the skeleton remains the same, while distorted or
(ii) reconstructive transitions with reversible cleavage/
formation of a metal–carboxylate bond and a scissoring
motion induced by the size of the guest.36
The first case is illustrated by the ‘jungle gym’ solid
[Zn2(1,4-BDC)2(DABCO)]�G.37–41 It is based on Zn2 dimers
which, by linkage with four 1,4-BDC molecules, provide 2D
square grids. The pillaring of the latter by the nitrogens of
DABCO ensures the 3D structure (Fig. 8).
For example, during the adsorption of propane 2-ol
(IPA),41 a monoclinic distortion of the primitive tetragonal
unit cell first occurs after the introduction of up to three IPA
molecules per cell. The volume decreases by 21% with a
significant (0.3 and 0.9 A) unexpected decrease of the length
of the edges, since it corresponds to the length of both rigid
linkers. The distortion corresponds to a shift of the square
grids from one to the other by 161. By increasing the amount
of IPA up to 4.5 IPA regenerates the dimensions of the
original cell. This structural behaviour is obviously similar to
that observed with MIL-53(Cr) (section 4.2). The sequence
dimer–DABCO–dimer corresponds to a rigid moiety, as also
for the chains in MIL-53. The O� � �O axis of the BDC plays the
same role of ‘kneecap’ in both cases. Unfortunately, the high
disorder of isopropanol molecules prevents an accurate struc-
tural determination and therefore, any discussion about the
nature of interactions, at least for three IPA in the cell. The
same behaviour also occurs in a solid recently discovered by
Long et al.42 with a cobalt (1,4-benzenedipyrazolate).
Displacive transitions are also observed in interpenetrating
and interdigitated 3D frameworks. This time, it is only the
slipping motion of the rigid interpenetrated layers which
occurs under the action of the stimulus. This behaviour is
one of the aspects of the ‘gate effect’.43 The original phase can
Fig. 7 Perspective view of MIL-68, with hexagonal and triangular
tunnels.
Fig. 8 Structure of the ‘jungle gym’ [Zn2(1,4-BDC)2(DABCO)]�xIPA, and a scheme of its intermediary distortion.
1386 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
be non-porous although, under a stimulus, it opens.
For instance, the (ZnI2)3(tpt)2 framework44 undergoes
considerable motion of the two interpenetrating (10,3b)
networks during nitrobenzene guest removal and reintroduc-
tion. In the same way, Kitagawa described45 a copper(II)
coordination polymer [Cu(dhbc)2(4-40-bipy]�H2O which is
non-porous up to 50 bar but opens above, when the pressure
of nitrogen increases.
The reconstructive phase transitions are by far more
rare. Up to now, only [Cu2(pzdc)2(dpyg)]�8H2O46 and
[Ni2(4,40bipy)3(NO3)4]
35,47,48 exhibit this behaviour. The first
solid is a 3D pillared structure. It undergoes a transformation
during the adsorption process, attributed to cleavage/
formation of the Cu–carboxylate bond. During adsorption
of methanol and water, it shrinks with a 27,9% contraction
and then expands, but not with methane.
These few examples (and many others) exhibit significant
changes for the cell volume which reached, until recently,
ca. 40%. They also identify the effect of the stimulus on the
structural changes in the skeleton but, for several reasons (high
disorder of the guests, partial structure determination. . .) in
most of the cases, attention was not paid enough to the
localization of the guests in order to determine what part of
the skeleton is affected by host–guest interactions, and to
evaluate the nature and the intensity of the latter, either
structurally or using computer simulations.49,50 Moreover,
in situ X-ray powder diffraction studies in real time are not
enough developed in order to understand the dynamics of the
evolution during adsorption. This will be discussed later
(section 4.2) in detail from recent examples.
This was recently performed in our group on a new series of
isoreticular solids (families MIL-8851,52 and -8953,54) which
represents the largest expansions ever evidenced: the ratio
between the volumes of the open and closed forms can
reach more than 300%. These solids are iron(III) or
chromium(III) carboxylates with a general formula
[MIII3O(X)3[
�OOC–(CH)n–COO�]3]�xG (X = CH3OH, H2O, F).
The organic moieties, always rigid, can be either an alkyl chain
with double bonds (fumarate [�OOC–(CH)2–COO�] (MIL-
88A) and muconate [�OOC–(CH)4–COO�]) (MIL-89) or ben-
zyl moieties (1,4-benzene dicarboxylate (1,4-BDC) [MIL-88B],
naphthalene dicarboxylate (NDC) [MIL-88C], or diphenyl
dicarboxylate (DPDC) [MIL-88D]. They all have an hexago-
nal symmetry (P�62c).
The inorganic cluster is a trimer of MIII3O octahedra
sharing a m3-oxygen (Fig. 9). Each of the six carboxylate
functions links two octahedra of the trimer, with three above
the trimer and three below. The second COO function of the
dicarboxylate is grafted on another trimer of the structure,
leading to a three-dimensional network.
In the as-synthesized products, the corresponding arrange-
ment gives rise to two types of cavities: large tunnels and large
cages having a triangular bipyramid (TBP) shape, within
which solvent molecules are located (Fig. 9). The vertices of
the bipyramid are occupied by the trimers. The distance
between two trimers of the TBP corresponds to the ahex cell
parameter, whereas the distance between the apical trimers of
the TBP refers to the chex parameter. The linking dicarboxylates
are on the oblique edges of the bipyramid; there is no link in
the equatorial plane of the TBP. This absence of a rigid
linker in this case is an essential point for breathing because
the distance between the three equatorial trimers can vary
without constraints under the influence of a stimulus (in this
case, the hydrogen and van der Waals bonds created by the
occluded species). As in MIL-53, the ‘kneecap’ role of the
O� � �O axis of the COO group is preserved. For a given length
of the linker, its flexibility allows the bipyramid to breathe
because it has two degrees of freedom: (i) the absence of
linkage between equatorial trimers (which avoids any
constraint in the equatorial plane of the bipyramid), and
(ii) the possibility of rotation of the COO group around its
O� � �O axis (Fig. 10).
When the solvent molecules leave the structure, the original
space group is preserved (P�62c), but one observes a drastic
increase of the length between the two non-equatorial trimers
(chex cell parameter), correlated with a decrease of the distance
between the equatorial ones (ahex cell parameter). On the other
hand, when the solvent is incorporated in the structure, the
bipyramid flattens and its volume increases (Fig. 9). In the
MIL-88D structure using diphenyl dicarboxylate as a linker,
the passage from the empty to the filled form corresponds to
an increase of ahex from 10.1 (as) to 20.04 A (Dd = 10 A)
whereas chex decreases from 27.8 to 22.9 A (Dd = 4.9 A).
Correlatively, the cell volume increases from 2362 to 7968 A3
Fig. 9 Views of the cages (top) and of the (001) projections (down) of
the different forms: dry, as (as synthesized) and open of MIL-88D. The
inorganic trimers are clearly shown on the projection.
Fig. 10 Local movements of the inorganic trimer and of the linker
during the swelling.
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1387
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
which means a final volume more than three times the initial
one and reversible atomic displacements of ca. 10A. . .
The most surprising feature concerns the fact that this
breathing occurs without any apparent bond breaking. It is
difficult to imagine that such variations could occur without a
phase transition. However, if the latter occurred, both phases
would coexist, at least during a few minutes, in the X-ray
powder pattern. This is definitively not the case. During the
dehydration–rehydration process of MIL-88D a continuous
evolution of the pattern, with drastic changes in the position of
the Bragg peaks is observed.
The evolution of the volume can be calculated analytically.
Indeed, a flexible bipyramid is a polyhedron in which the
length a of the oblique edges is fixed, whereas the length xa
((ahex) of the edges of the equatorial plane can vary. In these
conditions, it is easy to show that chex can be expressed as h =
[aO( 3 � x2)]/O3 and the volume as V = (1/6)x2a3O(3 � x2) ,
maximum for x = O2. The value x = O3, for which V = 0,
corresponds to the complete flattening of the bipyramid
(Fig. 11). The value of the experimental cell parameters for
MIL-88A-D shows that their open form always corresponds
to a complete filling.
Another interesting point must be mentioned. The pore
opening is strongly dependent of the nature of the solvent,
indicating some selectivity in adsorption. The soaking with
various polar and non-polar liquids shows three degrees of
pore opening: (i) small polar molecules (water and methanol)
and polar solvents induce a small swelling magnitude (6–8% in
volume); (ii) more hydrophobic but still polar liquids (butanol,
DMF, dimethyl carbonate) generate larger pore openings
(15–60%) whereas (iii) pyridine and diethylformamide
completely open the structure. The reason of such differences
in the magnitude of opening can be found in the characteristics
of the tunnels of the structure and in the nature of the guest
and its associated interactions. Indeed, along the tunnels,
hydrophilic inorganic parts of constant thickness (2.9 A) and
aperture of 3.7 A alternate with hydrophobic organic parts
(free aperture 3.2 A), the thickness of these depending on the
length of the linker. Therefore, the small polar liquids will
interact with the inorganic trimer, non-polar liquids interact
only with the linker via van der Waals, p–p or C–H–pinteractions but the presence of water on the trimer does not
favour their adsorption.52 Finally, these experiments shed
some light on the kinetics of adsorption, an aspect rarely
taken into account in the literature. It seems that the kinetics
of breathing is distinctive for each MIL-88. For example,
MIL-88A and MIL-88B breathe in a few seconds in the
presence of ethanol, but it takes several days to MIL-88B to
open its pores completely when water and nitrobenzene are
used. In the same way, MIL-88C opens rapidly with pyridine
(o1 min) but slowly with DEF (hours).
The above examples could indicate that breathing is a
general phenomenon, but, on the whole, it is rather scarce
and many MOFs do not swell. This is the case for the well
known zinc (1,4-benzenedicarboxylate) MOF-555 and the
series of IRMOF-n56 which correspond to the replacement
of 1,4-BDC by longer organic ligands. They do not breathe,
despite the existence of large windows (8 A for MOF-5) which
allow an easy access to different molecules and the possibility
of breathing. However, a structure determination at 350 1C
shows21 that, even without guests, the structure remains
unaltered and therefore insensible to the temperature stimulus.
Why MOF-5 does not breathe? Why the jungle-gym do? The
origin of such an inertness must be found not only from the
topology, as before, but from the structure of the zinc cluster
itself. Indeed, if the symmetry of the cell was taken alone into
account, the presence of the guest species could at least lead to
a rhombohedral cell distortion (elongation or compression
along one diagonal of the cube), which corresponds in both
cases to a decrease of its volume (the volume of a rhombo-
hedron V = a3(1 � 3cos2a + 2cos3a)1/2 is maximum for
a = 901) in agreement with the disappearance of the guests.
The terephthalate ions could accommodate such a distortion
owing to the ‘kneecap’ role of the O–C–O moieties, which
would allow to privilege one of the four threefold axes of the
original cubic cell if the cluster had a different symmetry.
However, considering the tetrahedron of tetrahedra forming
the cluster (Fig. 12) a cooperative movement of the ligands is
impossible during a rhombohedral elongation (for example).
If the ligands of the upper part of Fig. 12 satisfy this
movement because the O� � �O axes of the carboxylates are
perpendicular to the elongation axis, those of the lower part
are almost parallel to this axis, which renders impossible a
rotation of the ligands in the same way owing to the rigid sp2
configuration of the COO group. MOF-5 illustrates the case
when the configuration of the cluster prohibits breathing
because the positions of some of the ligands do not allow a
cooperative and symmetrical rotation of the linkers. This is
probably due to the octahedral disposition of the carbons of
the carboxylates around Zn4, which creates odd cycles between
these carbons, a condition which does not favour breathing, as
seen for MIL-68. Therefore, a new condition for swellingFig. 11 The geometry of the bipyramid (left) and the analytical
evolution of the volume vs. x.
1388 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
arises from this example: not all inorganic bricks (SBU) can
allow dynamic effects, which are possible only if the carboxylates
have mirror positions around the organic brick.
In conclusion, one can ask a question: why does the jungle
gym [Zn2(1,4-BDC)2(DABCO)], described above, swell
despite the fact that it is cubic like MOF-5, has the same
topology as MOF-5, and has six ligands in octahedral disposi-
tion around the cluster? The ‘brick’ is not the same, however:
instead of a tetrahedron of tetrahedra, it is a dimer, but also
there are two types of ligands: four 1,4-BDC and two DABCO
in trans position. The latter are intrinsically monodentate, with
strong and rigid metal–nitrogen bonds which do not allow
rotations around N. Therefore, the distortion can only occur
in the square grids formed by four carboxylates which can
present cooperative and symmetrical rotation.
A question arises from all these observations: can empirical
rules be extracted for predicting possible breathings of the
skeleton of hybrid porous solids?
3.4 Some empirical structural rules for possible breathing
(a) The first rule concerns the inorganic brick (SBU). Inde-
pendently of the nature of the ligand, it seems clear that the
only bricks which permit swelling are those which possess a
mirror plane with the carboxylates in symmetrical position
towards it. This is the case of the dimer of the jungle gym
[Zn2(1,4-BDC)2(DABCO)] and of the trimer of MIL-88
(Fig. 13(a) and (b)). Otherwise, the situation of MOF-5 occurs
(Fig. 13(c)). This will be an a priori indication for colleagues
discovering new clusters before trying any experiment related
to dynamic effects. This also incites chemists to search for new
inorganic bricks, the currently existing ones being rather
rare.57
(b) It seems that the ratio C/M (C: number of carbons of the
carboxylate surrounding the cluster; M: number of metallic
atoms within the cluster) is a good indicator of the possibility
of the brick to potentially allow swelling. From the few results
in the literature, it seems that C/M must be Z 2 for such an
effect. This is the case for [Zn2(1,4-BDC)2(DABCO)] and of
MIL-88 (4/2 and 6/3, respectively), but not for MOF-5 and
IRMOF-n (6/4) but this rule needs to be validated in many
other examples.
(c) Breathing effects can only occur with ditopic carboxylates.
Between two clusters, the kneecaps provided by the
O� � �O axes of the different carboxylates ought to be parallel
for possible phase rotations. This condition obviously prohi-
bits the use of tri- or tetratopic ligands for expecting swelling.
This is typically the case of MOF-77,58 MIL-10359 which
contains tritopic 1,3,5-benzene trisbenzoate and MIL-10260
formed with the naphthalene 1,4,5,8-tetracarboxylate.
(d) The existence of odd cycles in the structure, both at the
level of the cluster and/or at the level of the topology of the
skeleton, is unfavourable for dynamic effects, owing to their
rigidity (MIL-68, MOF-5 for instance). A particularly illus-
trative example is provided by MIL-10161 which, despite
favorable conditions (trimeric brick, ditopic ligand) does not
breathe only because most of the windows of the cages are
pentagonal. But, at variance to the other rules, it cannot be
predicted because the topology is known a posteriori.
(e) Breathing occurs only if all the conditions above are
satisfied simultaneously.
(f) As it will be shown below, as soon as the above
geometrical conditions are fulfilled, adsorption–desorption
phenomena can exhibit hysteretic effects (gate effect, multi-
ple-step adsorption. . .) depending on the nature of the guest
and on the relative energies of host–guest and guest–guest
interactions.
Fig. 12 Explanation of the absence of breathing in MOF-5. The stars
indicate the impossibility of rotation of the linker in the direction of
the elongation axis.
Fig. 13
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1389
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
4. Consequences of breathing for academic
knowledge and possible applications
4.1 Sensors and adsorbate–adsorbant interactions
The extreme sensitivity of breathing solids toward the action
of a stimulus suggests using them as sensors for evaluating
host–guest interactions for a better understanding of their role
in the dynamics of the framework.
This was realized very recently in our group using the
MIL-53 type.62 Starting from the hydrated form of
MIL-53(Fe) immersed in water, and placed in the beam of
the synchrotron radiation source, the solvent molecules were
introduced dropwise (an impregnation of MIL-53 in another
solvent leads to an immediate and full exchange with water,
preventing a kinetic study) and the X-ray powder patterns
were collected after each droplet addition in order to follow
structurally the evolution of the cell volume during the
exchange.
Looking first at the volumes of the fully exchanged samples,
whatever the evolution of the space groups (C2/c - Pnam -
Imcm, when the volume increases), it appears that:
(i) The structural changes occur even for the very smallest
added amounts of guests.
(ii) the topology of the structure of MIL-53 remains
invariant and can be described in all the cases from a reduced
unit cell with a lozenge-based prism. The ratio d/D of the two
diagonals d (smallest distances between chains) and D
therefore characterizes the extent of breathing. Small ratios
relate to the shrunk form of MIL-53 and large ones to the
expanded variety; d is the signature of the strength of the
host–guest interactions and the strong shrinkage of hydrated
MIL-53 shows the strength of the hydrogen bonds which link
the inserted water molecules to both the OH groups and the
oxygens of the carboxylates of the inorganic octahedral
chains.5 All the experimental data related to the different guest
molecules fit nicely with the theoretical evolution of the cell
volume vs. d/D (Fig. 14). This proves that all the molecules
involved in this study have each a selective response in terms of
swelling, and the evolution of the cell volumes provides a
quantitative classification of the strength of the host–guest
interactions: the smaller the expansion, the stronger the
interaction.
(iii) The size of the molecules does not play any steric role
during the swelling. Only host–guest (through the OH groups
of the chains and/or the phenyl rings) and guest–guest inter-
actions are involved in the process.
(iv) Except for the transient phase with methanol and light
alcohols, only the Bragg reflections of the starting and final
phases are observed without any evolution of their position, at
least on the time-scale of the experiment, implying a dramatic
stepwise expansion of the material, even in the presence of
minute amounts of guests which however open up the struc-
ture (Fig. 15). This is what we called the ‘forceps’ effect during
the adsorption, indicating that there is no solid solution during
the exchange which corresponds to a biphasic system.
(v) The time-resolved study of the exchange with pyridine,
lutidine and m-xylene also shows that the kinetics of adsorp-
tion is strongly dependent on the nature of the guest. This is
fast for pyridine (o10 mn) while full exchange occurs in one
hour for lutidine and in 90 mn for m-xylene.
Fig. 14 Comparison of the theoretical evolution of the volume vs.
d/D (black squares) with the fully exchanged experimental ones
(red, green and blue filled circles) associated with the changes of
symmetry and space groups during swelling. These discontinuities
illustrate the influence of host–guest interactions on the phenomenon.
The orange stars correspond to the increase of the swelling of the
transient water–methanol phase at t = 0, 25 and 40 mn.62
Fig. 15 Evolution of the powder patterns of MIL-53(Fe)�H2O during
the dropwise addition of lutidine. The Miller indices on the left refer to
the original MIL-53(Fe)�H2O, and those on the right to MIL-53(Fe)�lutidine. It is clearly seen that the positions of the peaks do not change
during the exchange and that, in this biphasic system, the lutidine
phase appears quasi-immediately after the introduction of the first
drops, even if the contribution is not very visible on the graph during
the first minutes (see text).
1390 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
(vi) For alcohols (methanol and ethanol), the same ‘forceps’
effect occurs when introducing minute amounts of guests,
before the appearance of intermediary phases during the
exchange. The parameters of these transient phases follow
the normal evolution of volumes. This implies that the
methanol–host interactions are rather strong although weaker
than the water–host ones. Moreover, and, at variance to the
observation with pyridine and lutidine, once the transient
phase formed, further addition of methanol provides a slight
increase of the cell volume (and therefore a decrease of
host–alcohol interactions). Within a certain range of composi-
tions (corresponding to a solid solution behaviour) which
stabilise the intermediate phase, the water–alcohol mixture
within the tunnels plays a specific structural role. In terms of
guest–guest interactions, strong hydrogen bonding may be
evoked for explaining this behaviour.
A few structures of the MIL-53(Fe) type with different
guests have been solved recently.63–68 They shed more light
on the different interactions which are responsible of the
swelling of the tunnels. Three types of interactions are
currently evidenced: (i) guest–guest ones (hereafter denoted
G–G), (ii) host–guest ones through p–p and CH–p interactions
between the guests and the terephthalate of the skeleton
(H–G-p), and (iii) for N containing guests, interactions
between them and the OH of the structure (H–GN).
Depending on their relative importance, three situations
occur (Fig. 16), illustrated by the cases of the room- and
high-temperature forms of MIL-53(Fe)�quinone68 and of
MIL-53(Fe), 0.85pyridine:63
(a) In the room-temperature form of MIL-53(Fe)�quinone(V = 1454 A3), whatever the tunnel, all the guest molecules
have the same disposition, parallel to one half of the
terephthalates (BDC) of the skeleton, and creating CH–pbonds with the other half. Between two rows of the same
tunnel, the O� � �O axes of the quinone are orthogonal. This
indicates three types of interactions: (i) strong p–p ones
between the centers of the phenyl rings of the BDC and of
quinone (d: 4.06 A), (ii) medium CH–p ones between the H of
quinone and the BDC and (iii) weaker guest–guest interactions
(4.99 A). However, despite this ranking, all the interactions are
relatively weak since the volume is very high as seen in the
curve of Fig. 15.
(b) In the high-temperature form of MIL-53(Fe)�quinone,the interactions are a little weaker than above (V = 1494 A3).
The main difference with (a) concerns the disposition of the
guests between two adjacent tunnels. Whereas the disposition
of quinones within one tunnel remains roughly the same apart
of a small deviation, it becomes orthogonal from one tunnel to
the other instead of parallel in case (a) but the characteristic
distances are very close. This corresponds to a thermally
induced transition, showing also that this host–guest inter-
action drastically depends on weak variations of the tempera-
ture. The same situation is encountered with molecules such
as m-xylene66 in MIL-53(Fe), and aniline and thiophene in
MIL-47(V).64,65
(c) When N donors exist in the inserted molecule, the
situation is completely different. This is the case for pyridine
(py), DMF63 and lutidine in MIL-53(Fe).68 This time, the
molecules are stacked head to tail with a small shift along the
axis of the tunnel. The predominant interaction exists between
the N atom and the OH groups linking two octahedra of the
skeleton chains (d = 2.68 A for py). Two facts prove that this
interaction is rather strong: (i) the lower volume (V= 1393 A3),
(ii) the comparison with what is observed in solid pyridine
(4.7 A). This enhancement of the strength relates to confine-
ment effects. Otherwise, the G–G interaction between two
consecutive py is rather strong since the distance between
two rings is 3.47 A. The other interactions are weaker.
Some remarks can be made around these situations, parti-
cularly when N donors are involved. Is has been noted that
MIL-53(Fe), as well as MIL-63(Ga)67 accepts a maximum of
0.85 py per Fe, with no means of introducing more. This
implies some disorder within the tunnels. This is suppressed
with Al, with the 1 : 1 stoichiometry being reached. This time,
the organization of pyridine molecules corresponds to the
situation described in (a).67 Two conclusions emerge from this
fact: (i) within a structure type, the nature of the metal plays a
Fig. 16 The three different dispositions of the guests within the
tunnels of MIL-53(Fe): (a) quinone at 300 K; (b) quinone at 333 K;
pyridine at 300 K.
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1391
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
role on the relative strength of all the interactions and (ii) small
variations in the amount of the guest in the pore induce drastic
modifications of the relative strengths of all the interactions
which occur within the tunnels. However, by exposure to air,
the stoichiometric Al terephthalate loses 0.15 py to give the
classical MIL-53(Al)�0.85py with the same characteristics as
the Fe and Ga solids.
A second remark concerns the comparison of the
behaviours of pyridine and lutidine (or dimethylpyridine) in
MIL-53(Fe). The cell volume with lutidine (V = 1575 A3) is
considerably larger than for pyridine (V = 1393 A3)
suggesting weaker interactions. A preliminary structural
study68 indicates that while the relative disposition of the
guests remains the same in both solids, the characteristic
distances change. If the lutidine–lutidine interactions, even if
slightly longer (3.66 A), remain almost the same, this is not the
case for the N–OH interactions which increase significantly
because the nitrogen atom is masked by the two methyl groups
surrounding it, weakening the interaction for steric reasons.
The increase of the volume decreases correlatively the
confinement effects of the tunnel and the positions of lutidine
are not as constrained as those of pyridine. It explains why the
local environments of lutidine in its solid state or included in
the tunnels are very close. In pyridine, the two states are
different.
The third remark concerns the existence of another type
of interaction encountered once: acido-basic one. This was
demonstrated in the solid [Cu2(pzdc)2(pyz)] (pzdc = pyrazine-
2,3-dicarboxylate, pyz = pyrazine) with permanent 1D
channels with cross-section 4 � 6 A.70 After adsorption of
C2H2, these molecules reside in the channels in such a way that
the acidic hydrogens of acetylene interact with the terminal
oxygens of the skeleton and electron density calculations69
proved that there is an overlap between the electronic clouds
of the two atoms, indicative of a weak acidobasic reaction.
The final remark is more general and concerns a beginning
of justifications of the conditions of synthesis in chemistry.
Indeed, looking at the recipes leading to hybrid solids (and not
only in this domain), it can be seen that the solvent is just
mentioned by the authors, with of course, no reason for its
choice. ‘It is the art of the chemist’ is replied to those who ask
for the question. . . Art can become science too! And probably,
this study of the breathing effect can provide some elements of
an answer. Indeed, it proves indirectly that, even during the
reaction, and depending on the nature of the solvents and of
the species (inorganic bricks (also called Secondary Building
Units (SBU)) and linkers as well) in the solution, the different
interactions involved in the process can change as a function
of their strength and orientate the crystallization of one phase
or another.
To conclude this section, the reversible sponge-like
properties of MIL-53(Fe) applies to a variety of simple organic
molecules (including both protic and non-protic species and
polar and non-polar molecules). Its extreme sensitivity and
selectivity toward organic molecules have provided interesting
information about the host–guest and guest–guest interactions
which govern the adsorption and separation phenomena, as
well as their different strengths. This could find applications in
the field of sensors for their detection in small amounts.
4.2 Adsorption and diffusion
These phenomena are obviously the first application of
breathing solids since the adsorption of gases or liquids is
one of the possible stimuli for the swelling effect. We shall
restrict our discussions to physisorption, whereas breathing
effects involving chemisorption can arise, particularly with
molecular solids.71 This review does not aim at presenting a
large amount of data relative to adsorption; such data are
found in the reviews of the Kitagawa and Rosseinsky groups.
On the contrary, this review extracts from some selected
examples, the specificity of adsorption by breathing matter
in its different macroscopic behaviours, in relation with the
structures when data are available. As yet this is not general
and most authors simply mention the adsorption curves
without following at least in situ the evolution of the
X-ray diffraction pattern to evaluate the cell parameters. Full
structural determinations of the filled samples are currently
extremely rare.
The isotherms of adsorption–desorption will first depend
simultaneously on the nature of the adsorbate and of the
surface of the adsorbant, on the strength of host–guest
interaction, but also on the state of the starting material,
either filled by guests/closed by previous activation, or opened
by the same treatment.
If the starting structure is closed or filled by initial guests, it
is extremely difficult to force entry into the structure. In
doing so, high pressures are necessary for opening.
[Cu(4,40-bipy)(dhbc)2]�2H2O72 represents a textbook example
for illustrating the ‘gate effect’ in this case (Fig. 17(a)). No
Fig. 17 (a) Adsorption isotherms of N2, O2, CH4 and CO2 in
[Cu(4,40-bipy)(dhbc)2]�2H2O72 showing the gate effect as a function
of the nature of the gas. (b) Adsorption isotherm of N2 in [Co(BDP)�2DEF�H2O] with the strange five steps.74
1392 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
adsorption is observed at low pressures except for CO2 and the
pressure threshold for adsorption strongly depends on the
nature of the gas. The pressure of opening is an indirect
signature of the energy of adsorbate/adsorbant interaction
(or the exothermic enthalpy of adsorption, unfortunately
scarcely measured). The larger the enthalpy, the lower the
threshold. This behaviour was also observed during the
adsorption of methanol by [Cu2(pzdc)2(dpyg)]�8H2O.73 This
effect opens large possibilities in separation processes. This will
be developed in section 4.3.
Until very recently, it was thought that this behaviour
corresponded to a one-step transition, but a recent paper by
Long74 with the study of the adsorption of N2 by cobalt
(1,4-benzenepyrazolate (BDP)) [Co(BDP)�2DEF�H2O] shows
that the problem is more complex. This structure is topo-
logically equivalent to the jungle gym [Zn2(1,4-BDC)2(DABCO)],
with dimeric corner-sharing Co tetrahedra replacing the Zn
dimers. The activation of the tetragonal solid distorts the
structure in a way not solved for the moment, but probably
of the same type as MIL-53. During the first stages of the N2
adsorption at low pressure and 77 K, the isotherm exhibits
an unprecedended five steps (Fig. 17(b)), the last one corres-
ponding to the characteristics of the original unit cell.
Increasing slightly the temperature of adsorption (87 K)
makes the steps less and less visible. The reason of this strange
behaviour is currently unknown but in situ measurements of
the evolution of the diffraction patterns vs. pressure is
currently under way. Anyhow, this example is interesting
because it may be supposed that the successive ‘plateaus’
correspond to the fixation of N2 on different sites of the
distorted structure. The resolution of the different structures
during adsorption will be of paramount importance for the
understanding of this phenomenon. Note that the isotherm of
adsorption of H2 exhibits only a single step.
At variance to the above cases, the structure of
MIL-53(Cr,Al) opens under activation. For a better under-
standing of the adsorption of gases, a detailed study, both
experimental and theoretical, was undertaken when H2, CO2
and CH4 and homologous alkanes are concerned, both on
MIL-53(Cr,Al) solids and their counter example MIL-47 (V),
which remains opened and does not breathe. The main struc-
tural difference between the two types results from the presence
or not of m2-OH groups linking the octahedra of the chains.
Experimentally, two situations occur. In the first, the
adsorption curves exhibit a classical type I behaviour; no
unusual behavior is observed in the curve of amount adsorbed
vs. pressure (Fig. 18). This situation is encountered for all the
solids with hydrogen and methane and only with MIL-47
adsorbing CO2. The second, encountered only a few
times,47,75–77 belongs to a new case of adsorption in the field
of hybrid porous solids. It is a two-step behaviour. After a fast
adsorption at very low pressure, the curve reaches a ‘plateau’
in a certain range of pressure before increasing once more.
Even at 70 kbar, saturation is not reached. This phenomenon
occurs only with MIL-53(Cr,Al) during CO2 adsorption
(Fig. 19). As mentioned in ref. 77 this two-step behaviour is
‘still difficult to understand’.
For this reason, we undertook a series of experiments
combining for each gas in situ powder diffraction in real time
at various pressures, with microcalorimetry and molecular
simulations in order to provide complementary elements of
explanations for such an unusual feature. Powder diffraction
was performed at the ESRF Grenoble using a new set-up
placed in the beam. The gas is introduced at increasing
pressures in the tube containing the activated solid and the
powder patterns are collected for each pressure. This gives
the evolution of the cell parameters for each pressure with the
possibility, if the product is sufficiently well crystallized, to solve
the complete structure in order to provide structural informa-
tion about the location of the guests. Quasi-elastic neutron
scattering (QENS) permits access to the dynamics of diffusion.
Microcalorimetry provides experimentally the thermodynamic
elements, in particular the adsorption enthalpy, which indicates
the strength of the guest–host interactions. Finally, molecular
simulations (DFT, Monte Carlo, molecular dynamics) combine
experiment and theory for the fit of the isotherms and give an
image of the movements of the guests within the tunnels and
their jumps from one adsorption site to another.
Our recent series of papers79–87 shed some light on the
explanation of adsorption phenomena.
Concerning hydrogen,78,79 the enthalpies of adsorption are
very weak (3–7 kJ mol�1) and the solids do not breathe. QENS
shows79–87 however that MIL-47(V) and MIL-53(Cr) have
some points of difference in their behaviours in terms of
self-diffusivity. In both cases, diffusivity coefficients D indicate
an orientationally averaged motion, but MIL-47(V) exhibits a
higher diffusivity than MIL-53(Cr) , which is consistent with a
lower activation energy (experimentally 0.6 kJ mol�1 for
MIL-47; 1.6 kJ mol�1 for MIL-53; calculated 0.6 and
1.3 kJ mol�1, respectively). In particular, the self-diffusivity
coefficient of MIL-47(V) (2 � 10�6 m2 s�1 at 300 K) is much
higher than those observed in liquid hydrogen (10�8 m2 s�1)
and in zeolites (10�8–10�10 m2 s�1). In these conditions, it is
illusory to try to localize H2 molecules in the tunnels.
However, in order to more deeply understand the diffusion
mechanism of H2 in the tunnels, molecular dynamics simula-
tions were performed. They show that the diffusion of H2 in
MIL-53(Cr) is mainly governed by the interaction between H2
and the m2-OH groups of the chains, leading to a 1D diffusion
Fig. 18 Adsorption isotherms of H278 in MOF-5, MIL-53(Al) and
HKUST-1 (a copper 1,3,5-benzene tricarboxylate).
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1393
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
along the tunnel via a jump sequence involving these OH
groups. In contrast, in MIL-47(V), the simulation shows a
three-dimensional diffusion process with random motions
within the pore of the material.
More attention has been paid to CH4 and CO2. In terms of
breathing, CH4, like H2 does not lead to swelling as proved by
its Type I isotherms.75 In MIL-47(V), five molecules of
methane per tunnel are present at 30 bar. The experimental
adsorption enthalpies increase from �15 to �19.5 kJ mol�1
between 0 and 20 bar, and suggest that methane probes
MIL-47(V) as a homogeneous energetic surface with no pre-
ferential adsorption sites within the pore. QENS experiments
indicate strong similarities with those obtained for H2:
unidirectional diffusion, high self-diffusivity coefficient
[4.2 � 10�7 m2 s�1 at 300 K], but lower than for hydrogen.
The comparison of the probability density in both solids
MIL-47(V) and MIL-53(Cr) is fruitful (Fig. 20). Except for
adsorption enthalpies which remain constant (�17 kJ mol�1),
it once more evidences the influence of the presence of m2-OH
groups on the unidimensional process. It shows that the lower-
energy regions are centered around the m2-OH as well as at the
middle of the tunnel. In MIL-47(V), this latter region is much
more favourable for the methane molecules. In MIL-53(Cr),
one can imagine a sequence of jumps between two consecutive
OH groups. According to QENS data, it is possible to estimate
average jump lengths of 5.6 A (MIL-53) and 8.1 A (MIL-47)
(with a large distribution) with characteristic intervals of
7.3 and 6.0 ps, respectively.
With CO2, the situation is completely different. It breathes
with the m2-OH–containing MIL-53(Al) and (Cr) but not with
MIL-47(V) which does not possess OH groups. Our in situ
structural study of MIL-53(Cr) during adsorption and
desorption of CO281 proves first that, when activated the
structure is open with a cell volume of 1486 A3. Second,
during the first steps of adsorption (Fig. 19), a large amount
of CO2 is adsorbed (two molecules per unit cell) and leads to a
plateau in the adsorption isotherm) between 1 and 4 bar.
In situ diffraction shows that, at that moment, the structure is
contracted by ca. 32% (V = 1072 A3 instead of 1013 when
water is adsorbed). A further adsorption of CO2 at higher
pressures reopens the framework while accepting additional
CO2 molecules into a newly formed tunnel (ca. 9 molecules per
unit cell). From the structure determination at 2 bar, (Fig. 21),
it appears that the CO2 molecules interact with the two
opposite chains corresponding to the small diagonal of the
rhomblike section of the tunnel. The linear molecules are lined
up along the axis of the latter, with an intermolecular distance
of 3.4 A, in the same order of magnitude as distances existing
in solid CO2 (3.24–3.57 A) and indicates strong guest–guest
interactions within the tunnel. Regarding the interactions
between CO2 molecules and the framework, the interatomic
distances suggest two scenarios: (i) interaction between the
carbon atom of CO2 with the oxygen atom of the OH group of
the chain (dC–O = 2.8 A) and (ii) as the CO2 molecule
has a strong quadrupolar moment (�1.4 � 10�35 C m), a
quadrupolar–dipolar interaction between CO2 and OH
(the distance between the two oxygens is 3.06 A).
To improve the knowledge of these interactions, in situ IR
spectroscopic experiments were performed in the same
conditions as for the diffraction studies. They provide
an elegant signature of the dynamics of swelling during
Fig. 19 Adsorption isotherms of methane and carbon dioxide75 in MIL-53(Al) (left) and the corresponding enthalpies of adsorption.
Fig. 20 2D free-energy maps of methane in MIL-53(Cr) (left) and
MIL-47(V) (right) at 250 K obtained through the xz plane of a given
tunnel for a loading of five CH4 molecules per unit cell. White
corresponds to regions of lower free energy, and black to regions of
higher free energy. The dashed lines are guides to the eye.
1394 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
adsorption–desorption.82,83 The increasing amount of
adsorbed CO2 is estimated from the evolution of the integrated
intensity of the n2 bands between 645 and 665 cm�1. The shift
from 1022 to 1017 cm�1of the n18a ring mode of the terephthalate
entities corresponds to the passage from the open to
the closed forms of the skeleton. At lower wavenumbers
(550–600 cm�1), the evolution of the spectra characterizes
the distortions of the framework under the action of the
CO2 molecules. The split of the n2 band into two components
at 653 and 662 cm�1 also reveals that the main interaction
involves the formation of electron donor–acceptor complexes
between the C atom of the CO2 molecule and the electron
donor center of the framework. Clearly, CO2 acts as an
electron acceptor while the OH group plays the role of donor.
Incidently, this is the first time that such an interaction
between CO2 and OH is characterized in a solid.
Computer simulation experiments provide more details
about the phenomenon. For MIL-53(Al),84,85 the calculated
enthalpy of adsorption is �37 kJ mol�1 (exptl. �35 kJ mol�1)
and the distances between the oxygens of CO2 molecules and
the hydrogens of the OH groups are calculated as 1.94 and
2.02 A. In conformity with the structural data at 2 bar
(2 molecules/u.c.), a double interaction between the oxygen
atoms of the adsorbate and the two hydrogen atoms on the
hydroxyl groups on opposite pore walls is the preferential
adsorption arrangement and is responsible of the shrinkage of
the tunnel. At increasing pressures, the favourable geometry is
broken, due to the interaction between the additional CO2
molecules. The resulting arrangement corresponds to a less
energetic single interaction between the adsorbate and the OH
group, and therefore, the structure reopens. The same study
with MIL-53(Cr)86 confirms the results obtained with MIL-
53(Al), with an enthalpy of adsorption is �35 kJ mol�1. It
provides also an important new information which could not
be obtained experimentally due to the extremely fast CO2
uptake: the situation when only one CO2 molecule is in the
tunnel instead of two at the plateau (Fig. 22). For one
molecule, calculations prove that the structure remains open
and that the contraction occurs only above two CO2 molecules
per unit cell (in agreement with the experimental results). The
structure begins to re-expand between 5 and 6 guests cell�1.
This result is very important for the explanation of the
phenomenon. It shows that, even for molecules having a high
enthalpy of adsorption, breathing occurs only for a minimum
number of guests in the cell, a minimum which corresponds to
the establishment (for reasons of proximity) of strong
guest–guest interactions within the tunnel These CO2–CO2
interactions act as a backbone for the adsorbed species and
allow the double interaction between the oxygen atoms of the
adsorbate and two hydrogen atoms on the hydroxyl groups on
opposite chains. This symmetry and a threshold intensity
of the interactions is essential for the occurrence of the
contraction. This last remark regarding intensity is crucial
and explains why MIL-53 does not breathe with methane. Its
enthalpy of adsorption is too low and, mainly due to the
spherical and apolar characters of methane, rules out the
development of strong guest–guest interactions which could
lead to a contraction. In a recent paper,87 we prove that with
superior linear alkanes, both the larger enthalpies (which
increase roughly by 10 kJ per additional carbon in the chain)
and their stronger interactions make that the corresponding
solids breathe once more.
In other words, explaining the swelling phenomenon is
indeed difficult because many parameters occur in order to
allow the dynamic movements. However the general picture is
becoming more clear now. Beside this important improve-
ment, other aspects are of course of potential interest in
relation with the breathing phenomenon. They concern
dedicated properties and applications which are detailed
below. They have not received the same level of attention as
above but future studies should be forthcoming.
4.3 Adsorption and separation
Separation is based on the selective times of adsorption of
different molecules on a surface or by the different boiling
Fig. 21 The CO2 molecules in the shrunk tunnel. The OH groups of
the skeleton appear as large grey spheres in the chains.
Fig. 22 Top: Calculation of the evolution with time of the cell volume
of MIL-53(Cr) for different values of the number of CO2 molecules
within one unit cell; Bottom: the evolution of the cell volume vs. the
number of CO2 molecules within the cell. Lines are simply a guide for
the eye.
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1395
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
temperatures of the constituents of a mixture. It is a strategic
problem for industry which, most of the time, requires the use
of energy to perform these separations, and an evaluation of
the induced costs. Hybrid solids can provide in some cases a
solution at room temperature in order to circumvent the
problem. We shall present two examples in this area, using
the same family of solids, one breathing (MIL-53(Cr)), the
other non-breathing (MIL-47).
It has been seen above that depending on their nature and
that of the skeleton, molecules are oriented in a specific
manner within the pores, which governs their time of residence
on the surface. It is therefore assumed that isomers, despite
similar orientations, will present different energies of surface
adsorption in relation with the positions of the substituents in
the molecule.
With this in mind, the De Vos group66 has recently proved
that the non-breathing MIL-47, which remains open whatever
the guest (see above) is a very good material for the separation
of the isomers of xylene. Their separation, in a mixture with
ethylbenzene is quantitative and effective in less than 20 min at
room temperature.
The second example uses MIL-53(Cr). It was previously
seen that, when it is activated (open), the CO2 adsorption
corresponds to a two-step behaviour whereas that of methane
shows a type I isotherm. When adsorption measurements are
performed on the hydrated form of this solid, the situation
becomes completely different (Fig. 23). The adsorption of
methane becomes close to zero while, for CO2, the curve is
characteristic of a gate effect, and therefore of a one-step
behaviour.88
All the examples above (threshold of the gate effect depend-
ing on the nature of the gas for the same structure type, two-
step behaviour depending on the activation of the solid,
different temperatures of adsorption. . .) make breathing solids
excellent candidates for separation at room temperature,
playing on the different parameters cited above. They are
highly selective towards mixtures of guests and could represent
a cheap alternative for current processes.
4.4 Conducting materials
This title seems surprising as hybrid solids are well known for
being insulators. This relates to a general feature of hybrid
solids. Except in a few cases with copper-based solids, the
metals of the skeleton are in a single valence state. The idea
was therefore to introduce mixed-valency for metals using a
post-treatment procedure. This was performed electro-
chemically by introduction of lithium in MIL-53(Fe).89 It
must be remembered that this solid, in its dehydrated form,
is shrunk (V = 900 A3) at variance to its Cr rand Al
analogues. The process was followed by in situ methods
(diffraction, Mossbauer spectroscopy) in real time. These
prove that (i) both Fe2+ and Fe3+ coexist in the final solid
with a maximum ratio Fe2+ : Fe3+ of 3 : 2; (ii) the trans-
formed solid becomes a conductor and exhibits both electronic
and ionic conductivity; (iii) the structure swells during the
charging process with a complete opening of the structure
(V = 1562 A3) and the process is reversible during discharge.
The breathing occurs not only from the introduction of
lithium but also from the simultaneous insertion of the
electrolyte. The extent of the swelling depends on the nature
of the latter. This presence of electrolyte ensures better
contacts in the half-cell during the conductivity measurements
and opens possibilities of using breathing solids as new
electrode materials for fuel cells.
Incidently, and beside its aspect, it must be noted that, after
this work, the introduction of lithium in MOFs has become
topical for hydrogen storage purposes. From ideas of
Goddard90 and Blomquist,91 some attempts were successfully
realized 92,93 for introducing Li chemically in the structures of
MOFs. This indeed improves the performances of H2 storage
of these MOFs.
4.5 Drug storage and delivery
The synthesis of new bioactive compounds of very high
molecular weight with therapeutic activity and/or with a low
aqueous solubility becomes more and more complex and the
processes of their commercialization very slow. This requires
the use of carrier systems to improve the activity of known
molecules. These systems usually allow a better control of the
drug plasmatic effects, increasing the efficiency and decreasing
the toxicity, as well as an increase in the drug stability by
protection of the biodegradation. Until recently, polymeric
and mixed systems have been proposed for a better controlled
release of drugs.94,95 In particular, mesoporous inorganic
solids such as the ordered mesoporous silicas (pure or func-
tionalized by a post-synthesis modification) are now used.96
However, this process leads to a decrease of the drug storage
capacity.97,98
Recently, our group proposed an alternative route
(the hybrid route), using for the first time porous MOFs as
new controlled delivery systems99 because of their high pore
volume, regular porosity and the presence of tunable organic
groups within the framework which allow an easy modulation
of the size of the pores. The first example concerned the
adsorption of the model molecule ibuprofen in the rigidMOFs
MIL-100 and MIL-101 with mesopores. They exhibit a very
high drug storage capacity, up to an unprecedented 1.4 gram
Fig. 23 Comparison of the performances of anhydrous and hydrated
forms of MIL-53(Cr) in terms of CO2 and CH4 adsorptions, indicating
a high selectivity for the latter.
1396 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
of drug per gram of porous solid, and a complete drug
controlled release under physiological conditions from
3 to 6 days (Fig. 24).
For studying the effect of this breathing phenomenon on the
drug adsorption and delivery properties of pharmacological
molecules, Ibuprofen (hereafter denoted Ibu) was once more
chosen as a test molecule for looking at two points (i) the
influence of breathing on drug adsorption and (ii) the
characterizations and the induced effects of the host–drug
interactions (including kinetics of delivery in a simulated
physiological environment). Two flexible materials were used:
(i) MIL-53(Cr) solid as a model material despite the well-
known toxicity of chromium compounds and which, when
activated, presents its open form; (ii) the much less-toxic iron
analogue MIL-53(Fe) (rat oral dose: DL50(Fe) = 30 g kg�1,
DL50(terephthalic acid 46.4 g kg�1) which is closed after
activation.
Both solids adsorb ca. 20 wt% of ibuprofen100 (Fig. 24). The
filled samples are non-porous. The final cell volume after
treatment is roughly the same for the two compounds
(V E 1400 A3), corresponding to a slight contraction (6%)
for the opened MIL-53(Cr) and an important expansion
(55%) for the Fe analogue. Using IR spectroscopy, the shift
of the n(CQO) band of the carboxylic group of Ibu (1695 -
1725 cm�1) correlated to that of the vibrational n(O–H) band
of the inorganic chains (3656 - 3639 cm�1) indicates the
formation of a strong hydrogen bond between these two
groups. The different 1H and 13C solid-state NMR character-
izations confirm this fact and also provide information about
the dynamics of ibuprofen within the tunnel. 13C NMR
indicates a slight conformational distribution of the drug in
the pores, which concerns essentially the aromatic part of Ibu.
However, it shows also that the mean conformation is close to
that of the pure form of ibuprofen. All these features were
confirmed by DFT calculations which also suggest (Fig. 25),
beside the main interaction, the possibility of van der Waals
and/or CH–p interactions between the hydroxyls of the skele-
ton and the methyl groups of Ibu. This provides an estimate of
the drug–matrix interaction energy of �57 kJ mol�1, larger
than that observed for CO2 in MIL-53(Cr) (�35 kJ mol�1).
The most striking feature of the phenomenon concerns the
kinetics of delivery. In the first place, the delivery is complete.
The kinetic study was performed using a simulated body fluid
(SBF) with similar inorganic composition as human plasma, at
37 1C under continuous stirring. Surprisingly, a very slow
delivery, complete only after 3 weeks, is observed with clearly
two steps in the process. Despite the slight changes of slope,
the results were treated as a first approximation, as having a
zero-order kinetics. A model was proposed to explain this fact.
These facts suggest a specific behaviour of flexible frame-
works on the adsorption of species (gases, vapours, drugs...),
compared to rigid ones. They are adaptative and take, within
the same topology, a configuration where the interactions
between guest molecules and the framework are optimised
while taking also into account the steric hindrance of the guest
molecules and their number. If the latter increases due to a
additional stimulus (increase of pressure for instance), the
structure will open more. This adaptability might explain the
long time of delivery. Indeed, due to its flexibility, MIL-53 can
be considered as an intrinsically tailor-made container which
fits with the geometrical and energetical characteristics of the
guest and therefore enhances the confinement effects. If the
Fig. 24 Ibuprofen delivery performances of flexible MIL-53(Cr) and -(Fe) (left) and of rigid MIL-100 and MIL-101 (right).
Fig. 25 (a) Localization of ibuprofen in the pores of MIL-53(Fe);
comparison of the conformations of ibuprofen confined in the tunnels
(b) and in its pure solid form (c).
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1397
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
latter are optimized, they will not favour rapid evacuation of
the guests and therefore explain the long times for release.
Finally, we could expect very long therapies using flexible
MOFs for drug delivery. Also, the possibility of having drug
carriers with zero-order kinetics represents an important
progress since a unique administration could be provided.
This would lead to a stable blood concentration, a mini-
misation of the toxicity effects as well as a decrease in patient
discomfort. Moreover, the slow release will protect the
drug from degradation processes by increasing its plasmatic
half-life, its bioavailability and therefore its efficiency.
5. Conclusion
At the outset, breathing effects were considered as laboratory
curiosities. Their careful academic study has allowed to
understand not only the structural requirements for such a
phenomenon, but also its different manifestations and their
reasons. They result from the juxtaposition of structural
characteristics of the skeleton (mainly the existence of weak
points in the structure which allow flexibility), of the nature of
the internal surface, on the complex and competitive nature of
the different interactions (guest–guest and host–guest). Beside
these academic features, some potential applications of this
effect have appeared, due to the large amounts of species that
the expansion of the cells allow to store compared to rigid
frameworks. These applications relate to energy and sustain-
able development (adsorption, separation, storage. . .), and to
health (drug delivery). Their development is ongoing.
Acknowledgements
The authors are grateful to their colleagues in Versailles and
also to the groups of M. Daturi (Caen), H. Jobic (Lyon),
P. L. Llewellyn (Marseille) and G. Maurin (Montpellier) for
their outstanding contributions to the knowledge of the
breathing effect.
References
1 J. P. Collin, C. Dietrich-Buchecker, P. Gavina, M. C. Jimenez-Molero and J. P. Sauvage, Acc. Chem. Res., 2001, 34, 477.
2 M. C. Jimenez-Molero, C. Dietrich-Buchecker and J. P Sauvage,Chem.–Eur. J., 2002, 8, 1456.
3 K. Barthelet, J. Marrot, D. Riou and G. Ferey, Angew. Chem.,Int. Ed., 2002, 41, 281.
4 C. Serre, F. Millange, C. Thouvenot, M. Nogues, G. Marsolier,D. Louer and G. Ferey, J. Am. Chem. Soc., 2002, 124, 13519.
5 T. Loiseau, C. Serre, C. Huguenard, G. Fink, F. Taulelle,M. Henry, T. Bataille and G. Ferey, Chem.–Eur. J., 2004, 10,1373.
6 K. Barthelet, J. Marrot, D. Riou and G. Ferey, Chem. Commun.,2004, 520.
7 K. Uemura, S. Kitagawa, M. Kondo, K. Fukui, R. Kitaura,H.-C. Chang and T. Mizutani, Chem.–Eur. J., 2002, 8, 3587.
8 K. Biradha andM. Fujita, Angew. Chem., Int. Ed., 2002, 41, 3392.9 D. Maspoch, D. Ruiz-Molina, K. Wurst, N. Domingo,M. Cavallini, F. Biscarini, J. Tejada, C. Rovira and J. Veciana,Nat. Mater., 2003, 2, 190.
10 G. Ferey, Nat. Mater., 2003, 2, 136.11 G. Alberti, M. Murcia-Mascaros and R. Vivani, J. Am. Chem.
Soc., 1998, 120, 9291.12 R. H. Mitchell, Perovskites Modern and Ancient, Almaz Press
Inc., Thunderbay, Canada, 2002.
13 G. Ferey and J. Pannetier, Eur. J. Solid State Inorg. Chem., 1994,31, 697.
14 N. Khosrovani and A. W. Sleight, J. Solid State Chem., 1996,121, 2.
15 J.-M. Lehn, Science, 2002, 295, 2400.16 L. Girondini, A. G. Stendardo, S. Geremia, M. Campagnolo,
P. Samori, J. P. Rabe, R. Fokkens and E. Dalcanale, Angew.Chem., Int. Ed., 2003, 42, 1384.
17 S. Kitagawa and K. Uemura, Chem. Soc. Rev., 2005, 34, 109.18 S. R. Batten and R. Robson, Angew. Chem., Int. Ed., 1998, 37,
1460.19 G. Ferey, Chem. Soc. Rev., 2008, 37, 191.20 T. Reneke, M. Eddaoudi, D. B. Moler, M. O’Keeffe and
O. M. Yaghi, J. Am. Chem. Soc., 2000, 122, 4844.21 M. Eddaoudi, D. B. Moler, H. Li, B. Chen, T. Reneke,
M. O’Keeffe and O. M. Yaghi, Acc. Chem. Res., 2001, 34, 319.22 M. E. Thomson, Chem. Mater., 1994, 6, 1168.23 G. Alberti, in Solid State Supramolecular Chemistry: two- and
three-dimensional inorganic networks, ed. G. Alberti and T. Bein(vol. 7 of Comprehensive Supramolecular Chemistry, J.-M. Lehnchairman ed.), Pergamon-Elsevier Science, Oxford, 1996, ch. 5.
24 A. Clearfield, Prog. Inorg. Chem., 1998, 47, 373.25 H. Byrd, A. Clearfield, D. Poojary, K. P. Reis and
M. E. Thomson, Chem. Mater., 1996, 8, 2239.26 G. Ferey, unpublished results.27 G. Ferey, Chem. Mater., 2001, 13, 3084.28 G. Ferey, C. Mellot-Draznieks, C. Serre and F. Millange, Acc.
Chem. Res., 2005, 38, 317.29 F. Millange, N. Guillou, R. I. Walton, J.-M. Greneche,
I. Margiolaki and G. Ferey, Chem. Commun., 2008, 4732.30 T. Loiseau, C. Mellot-Draznieks, H. Muguerra, M. Haouas,
F. Taulelle and G. Ferey, Comptes-Rendus Chimie, 2005, 8, 765.31 M. O’Keeffe and B. G. Hyde, Philos. Trans. R. Soc. London, Ser.
A, 1980, 295, 553.32 S. Kitagawa and K. Uemura, Chem. Soc. Rev., 2005, 34, 109.33 K. Uemura, R. Matsuda and S. Kitagawa, J. Solid State Chem.,
2005, 178, 2420.34 D. Bradshaw, T. J. Prior, E. J. Cussen, J. B. Claridge and
M. J. Rosseinsky, J. Am. Chem. Soc., 2004, 126, 6106.35 X. Zhao, B. Xiao, A. J. Fletcher, K. M. Thomas, D. Bradshaw
and M. J. Rosseinsky, Science, 2004, 306, 1012.36 A. J. Fletcher, K. M. Thomas andM. J. Rosseinsky, J. Solid State
Chem., 2005, 178, 2491.37 K. Seki and W. Mori, J. Phys. Chem. B, 2002, 106, 1380.38 D. N. Dybtsev, H. Chun and K. Kim, Angew. Chem., Int. Ed.,
2004, 43, 5033.39 H. Chun, D. N. Dybtsev, H. Kim and K. Kim, Chem.–Eur. J.,
2005, 11, 3521.40 S. Horike, R. Matsuda, D. Tanaka, S. Matsubara, M. Mizuno,
K. Endo and S. Kitagawa, Angew. Chem., Int. Ed., 2006, 45, 7226.41 K. Uemura, Y. Yamasaki, Y. Komagawa, K. Tanaka and
H. Kita, Angew. Chem., Int. Ed., 2007, 46, 6662.42 H. J. Choi, M. Dinca and J. R. Long, J. Am. Chem. Soc., 2008,
130, 7848.43 D. Tanaka, K. Nakagawa, M. Higuchi, S. Horike, Y. Kubota,
T. C. Kobayashi, M. Takata and S. Kitagawa, Angew. Chem., Int.Ed., 2007, 46, 6662.
44 K. Birhada andM. Fujita, Angew. Chem., Int. Ed., 2002, 41, 3392.45 R. Kitaura, K. Seki, G. Akiyama and S. Kitagawa, Angew.
Chem., Int. Ed., 2003, 42, 428.46 R. Kitaura, K. Fujimoto, S.-I. Noro, M. Kondo and S. Kitagawa,
Angew. Chem., Int. Ed., 2002, 41, 133.47 A. J. Fletcher, E. J. Cussen, T. J. Prior, M. J. Rosseinsky,
C. J. Kepert and K. M. Thomas, J. Am. Chem. Soc., 2001, 123,10001.
48 E. J. Cussen, J. B. Claridge, M. J. Rosseinsky and C. J. Kepert,J. Am. Chem. Soc., 2002, 124, 9574.
49 C. Mellot-Draznieks, J. Dutour and G. Ferey, Angew. Chem., Int.Ed., 2004, 43, 6290.
50 C. Mellot-Draznieks, C. Serre, S. Surble and G. Ferey, J. Am.Chem. Soc., 2005, 127, 16273.
51 S. Surble, C. Serre, C. Mellot-Draznieks, F. Millange andG. Ferey, Chem. Commun., 2006, 284.
52 C. Serre, C. Mellot-Draznieks, S. Surble, N. Audebrand,Y. Filinchuk and G. Ferey, Science, 2007, 315, 1828.
1398 | Chem. Soc. Rev., 2009, 38, 1380–1399 This journal is �c The Royal Society of Chemistry 2009
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online
53 C. Serre, F. Millange, S. Surble and G. Ferey, Angew. Chem., Int.Ed., 2004, 43, 6286.
54 C. Serre, S. Surble, C. Mellot-Draznieks, Y. Filinchuk andG. Ferey, Dalton Trans., 2008, 5462.
55 H. Li, M. Eddaoudi, M. O’Keeffe and O. M. Yaghi,Nature, 1999,402, 276.
56 M. Eddaoudi, J. Kim, N. Rosi, D. Vodak, J. Wachter,M. O’Keeffe and O. M. Yaghi, Science, 2002, 295, 469.
57 G. Ferey, Dalton Trans., 2009, DOI: 10.1039/b817360p.58 H. K. Chae, D. Y. Siberio-Perez, J. Kim, Y. Go, M. Eddaoudi,
A. J. Matzger, M. O’Keeffe and O. M. Yaghi, Nature, 2004, 427,523.
59 T. Devic, C. Serre, N. Audebrand, J. Marrot and G. Ferey, J. Am.Chem. Soc., 2005, 127, 12788.
60 S. Surble, C. Serre, F. Millange, T. Duren, M. Latroche andG. Ferey, J. Am. Chem. Soc., 2006, 128, 14889.
61 G. Ferey, C. Mellot-Draznieks, C. Serre, F. Millange, J. Dutour,S. Surble and I. Margiolaki, Science, 2005, 309, 2040.
62 F. Millange, C. Serre, N. Guillou, G. Ferey and R. I. Walton,Angew. Chem., Int. Ed., 2008, 47, 4100.
63 T. R. Whitfield, X. Wang, L. Liu and A. J. Jacobson, Solid StateSci., 2005, 7, 1096.
64 E. V. Anokhina, M. Vougo-Zanda, X. Wang and A. J. Jacobson,J. Am. Chem. Soc., 2005, 127, 15000.
65 X. Wang, L. Liu and A. J. Jacobson, Angew. Chem., Int. Ed.,2006, 45, 6499.
66 L. Alaerts, C. E. A. Krschhock, M. Maes, M. A. van der Veen,V. Finsy, A. Depla, J. A. Martens, G. V. Baron, P. A. Jacobs,J. E. M. Denayer and D. E. De Vos, Angew. Chem., Int. Ed., 2007,46, 4293.
67 M. Vougo-Zanda, J. Huang, E. V. Anokhina, X. Wang andA. J. Jacobson, Inorg. Chem., 2008, 47, 11535.
68 F. Millange, N. Guillou, J.-M. Tarascon and G. Ferey,unpublished results.
69 R.Matsuda, R. KItaura, S. Kitagawa, Y. Kubota, R. V. Belosludov,T. C. Kobayashi, H. Sakamoto, T. Chiba, M. Takata, Y. Kawazoeand Y. Mita, Nature, 2005, 436, 238.
70 R. Kitaura, S. Kitagawa, Y. Kubota, T. C. Kobayashi, K. Kindo,Y. Mita, A. Matsuo, M. Kobayashi, H. C. Chang, T. C. Ozawa,M. Suzuki, M. Sakata and M. Takata, Science, 2002, 298, 2358.
71 M. Albrecht, M. Lutz, A. I. Spek and G. van Koten, Nature,2000, 406, 970.
72 R. Kitaura, K. Seki, G. Akiyama and S. Kitagawa, Angew.Chem., Int. Ed., 2003, 42, 428.
73 M. Kondo, T. Okbo, A. Asami, S.-I. Noro, T. Yoshitomi,S. Kitagawa, T. Ishii, H. Matsuzaka and K. Seki, Angew. Chem.,Int. Ed., 1999, 38, 140.
74 H. J. Choi, M. Dinca and J. R. Long, J. Am. Chem. Soc., 2008,130, 7848.
75 S. Bourrelly, P. L. Llewellyn, C. Serre, Millange, T. Loiseau andG. Ferey, J. Am. Chem. Soc., 2005, 127, 13519.
76 K. Yamada, H. Tanaka, S. Yagishita, K. Adachi, T. Uemura,S. Kitagawa and S. Kawata, Inorg. Chem., 2006, 45, 4322.
77 K. Uemura, Y. Yamasaki, Y. Komagawa, K. Tanaka andH. Kita, Angew. Chem., Int. Ed., 2007, 46, 6662.
78 G. Ferey, M. Latroche, C. Serre, T. Loiseau and A. Percheron-Guegan, Chem. Commun., 2003, 2976.
79 F. Salles, H. Jobic, G. Maurin, M. M. Koza, P. L. Llewellyn,T. Devic, C. Serre and G. Ferey, Phys. Rev. Lett., 2008, 100,245901.
80 N. Rosenbach Jr, H. Jobic, A. Ghoufi, F. Salles, G. Maurin,S. Bourrelly, P. L. Llewellyn, T. Devic, C. Serre and G. Ferey,Angew. Chem., Int. Ed., 2008, 47, 6611.
81 C. Serre, S. Bourrelly, A. Vimont, N. A. Ramsahye, G. Maurin,P. L. Llewellyn, M. Daturi, Y. Filinchuk, O. Leynaud, P. Barnesand G. Ferey, Adv. Mater., 2007, 19, 2246.
82 A. Vimont, H. Leclerc, P. Bazin, J.-C. Lavalley, M. Daturi,C. Serre, G. Ferey, S. Bourrelly and P. L. Llewellyn, Chem.Commun., 2007, 3291.
83 A. Vimont, A. Travert, M. Daturi, J.-C. Lavalley, S. Surble,C. Serre and G. Ferey, J. Phys. Chem. C, 2007, 111, 383.
84 N. A. Ramsahye, G. Maurin, S. Bourrelly, P. L. Llewellyn,T. Loiseau and G. Ferey, Phys. Chem. Chem. Phys., 2007, 9, 1059.
85 N. A. Ramsahye, G. Maurin, S. Bourrelly, P. L. Llewellyn,T. Loiseau, C. Serre and G. Ferey, Chem. Commun., 2007, 3261.
86 F. Salles, A. Ghoufi, G. Maurin, R. G. Bell, C. Mellot-Draznieks,P. L. Llewellyn, C. Serre and G. Ferey, Angew. Chem., Int. Ed.,2008, 47, 8487.
87 P. L. Llewellyn, G. Maurin, T. Devic, S. Loera-Serna,N. Rosenbach Jr, C. Serre, S. Bourrelly, P. Horcajada,Y. Filinchuk and G. Ferey, J. Am. Chem. Soc., 2008, 130, 12808.
88 P. Llewellyn, S. Bourrelly, C. Serre, Y. Filinchuk and G. Ferey,Angew. Chem., Int. Ed., 2006, 45, 7751.
89 G. Ferey, F. Millange, M. Morcrette, C. Serre, M.-L. Doublet,J.-M. Greneche and J.-M. Tarascon, Angew. Chem., Int. Ed.,2007, 46, 3259.
90 S. S. Han and W. A. Goddard III, J. Am. Chem. Soc., 2007, 129,8422.
91 A. Blonquist, C. M. Araujo, P. Srepusharawoot and R. Ahuja,Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 20173.
92 K. L. Mulfort and J. T. Hupp, J. Am. Chem. Soc., 2007, 129, 9604.93 K. L. Mulfort and J. T. Hupp, Inorg. Chem., 2008, 47, 7936.94 S. Freiberg and X. X. Zhu, Int. J. Pharm., 2004, 282, 1.95 K. S. Soppimath, T. M. Aminabhavi, A. R. Kulkarni and
W. E. Rudzinski, J. Controlled Release, 2001, 70, 1.96 M. Vallet-Regı, Chem.–Eur. J., 2006, 12, 5934, and references
therein.97 B. Munoz, A. Ramila, J. Perez-Pariente, I. Diaz and M. Vallet-
Regi, Chem. Mater., 2003, 15, 500.98 P. Horcajada, A. Ramila, G. Ferey and M. Vallet-Regı, Solid
State Sci., 2006, 8, 1243.99 P. Horcajada, C. Serre, M. Vallet-regi, M. Sebban, F. Taulelle
and G. Ferey, Angew. Chem., Int. Ed., 2006, 45, 5974.100 P. Horcajada, C. Serre, G. Maurin, N. A. Ramsahye, F. Balas,
M. Vallet-Regi, M. Sebban, F. Taulelle and G. Ferey, J. Am.Chem. Soc., 2008, 130, 6774.
This journal is �c The Royal Society of Chemistry 2009 Chem. Soc. Rev., 2009, 38, 1380–1399 | 1399
Dow
nloa
ded
by N
AT
ION
AL
TE
CH
NIC
AL
UN
IVE
RSI
TY
OF
AT
HE
NS
on 2
4 Fe
brua
ry 2
013
Publ
ishe
d on
26
Febr
uary
200
9 on
http
://pu
bs.r
sc.o
rg |
doi:1
0.10
39/B
8043
02G
View Article Online