review article

Reticular synthesis and the design

of new materials
Omar M. Yaghi*†, Michael O’Keeffe*‡, Nathan W. Ockwig*†, Hee K. Chae*†§, Mohamed Eddaoudi*† & Jaheon Kim*†

* Materials Design and Discovery Group, † Department of Chemistry, 930 N. University Avenue, University of Michigan, Ann Arbor, Michigan 48109-1055, USA
‡ Department of Chemistry, Arizona State University, Tempe, Arizona 85287-1604, USA
§ Permanent address: Hankuk University of Foreign Studies, Korea
...............................................................................................................................................................................................................................................................

The long-standing challenge of designing and constructing new crystalline solid-state materials from molecular building blocks is
just beginning to be addressed with success. A conceptual approach that requires the use of secondary building units to direct the
assembly of ordered frameworks epitomizes this process: we call this approach reticular synthesis. This chemistry has yielded
materials designed to have predetermined structures, compositions and properties. In particular, highly porous frameworks held
together by strong metal–oxygen–carbon bonds and with exceptionally large surface area and capacity for gas storage have been
prepared and their pore metrics systematically varied and functionalized.

A
lthough the synthesis of new materials has long been throughout the construction process—an important aspect that
recognized as the most essential element in advancing could help to realize fully the benefits of design in crystalline
technology, it generally remains more of an art than a solid-state frameworks. Similarly, reticular synthesis should be
science—in that the discovery of new compounds has distinguished from supramolecular assembly11, because in the
mostly been serendipitous, using methods referred to by former, building blocks are linked by strong bonds throughout
critics as ‘shake and bake’, ‘mix and wait’ and ‘heat and beat’. For the crystal.
much of the twentieth century, this worked well for the synthesis of In this context, it is worth noting that a plethora of extended
important solid-state materials, and we expect that it will continue structures successfully prepared by copolymerization of metal ions
to yield interesting compounds. However, it is becoming increas- with organic links have been reported and reviewed under the
ingly urgent to produce materials designed to perform highly rubric of ‘crystal engineering’, a term which has mutated somewhat
specific and cooperative functions1,2. over the years and which is now rather broadly defined12–14. We use
Recent extensive research into the design and synthesis of metal- the more precise term reticular synthesis (or chemistry) to describe
organic frameworks (MOFs) has led to numerous practical and the logical approach to the synthesis of robust materials with pre-
conceptual developments in that direction3–7. Specifically, the designed building blocks, extended structures, and properties.
chemistry of MOFs has provided an extensive class of crystalline Thus, it might be considered a subclass of crystal engineering.
materials with high stability, tunable metrics, organic functionality, Research in the area of MOFs and, to a lesser extent, inorganic
and porosity. Here we present some of the important developments frameworks, has recently progressed far enough to provide
that have shaped this rapidly growing field and propose a general examples which show the feasibility of the building-block approach
conceptual framework, which serves as a useful tool in designing and the scope of reticular synthesis. Here we take selected examples
materials constructed from molecular building blocks. to highlight the ideas involved, with emphasis on our own work on
MOFs in which structural and thermal stability characterization
Reticular synthesis have been combined with in-depth studies of surface area and gas
sorption capacities. The subject of metal-organic frameworks has a
General lack of control over the character of solids produced from
long history, however, so we begin by giving some perspectives on
the traditional synthetic methods is directly related to the fact
the field.
that the starting entities do not maintain their structure during
the reaction, leading to poor correlation between reactants and
products. We have shown that the design of an extended network Perspectives and challenges
can be realized by starting with well-defined and rigid molecular The Cambridge Structure Database (CSD) documents the crystal
building blocks that will maintain their structural integrity structures of more than 11,000 extended metal-organic compounds
throughout the construction process3,8. Alternatively, the use of in which a metal ion or cluster has been linked by an organic moiety
well-defined conditions that lead to the formation of such building in which the linking functionality is a cyanide, pyridyl, phosphate
blocks in situ is an equally viable approach to the design of extended or carboxylate. Of these, nearly 3,000 compounds have three-
structures and one that has been especially fruitful in MOF dimensional (3D) structures and about double that number have
chemistry8. We have shown8,9 that the practice of logical synthesis 2D structures. MOFs composed of the first-row transition metals
must begin with knowledge of the target network ‘blueprint’ and and organic links such as cyanide15,16, glutamate17, formate18,
identification of the required building blocks for its assembly. triazole19, oxalate20, 1,2,4,5-tetracarboxylates21 and squarates22
This process is central to our ability to achieve true design of have long been known. Also, expanded diamond structures based
solid-state materials: we refer to its implementation as reticular on linking of Cu(I) with ANT23, DCNQ24 and TCTPM25, and
synthesis. decorated diamond structures based on linking Mn(II) with
In essence, reticular synthesis can be described as the process of Ge4S42 9
10 (T2) clusters (see Box 1 for abbreviations of chemical
assembling judiciously designed rigid molecular building blocks names and clarification of terms) represent the numerous fore-
into predetermined ordered structures (networks), which are held runners of the extended structures constructed from molecular
together by strong bonding. It is different from retrosynthesis of building blocks. Subsequently, other compounds based on linking
organic compounds10, because the structural integrity and rigidity metal ions with ADC26, ETC27, INT28, BPY29,30 and BPCN31 have
of the building blocks in reticular synthesis remain unaltered been reported. Tetrahedral building blocks such as HMT32 and

NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature © 2003 Nature Publishing Group 705
review article
MCP33 have been linked by metal ions to form decorated diamond Consideration of this range of crystal structures, in particular
structures, while TPY34 and ATC35 are examples of links pre- those of MOFs, has led us to identify two critical questions for
designed so that their assembly into diamond networks is done by reticular synthesis. First, of the almost unlimited possible networks,
linking through hydrogen bonding. which can be expected to form and how can they be synthesized?

Box 1
Definitions and structures

Here we define some terms and illustrate organic species referred to in but not chemically linked. See refs 4 and 8 for reviews on this
the text by abbreviations in the Box 1 figure. property.
Reticular: (adjective) having the form of a (usually periodic) net. Expansion: Increasing the spacing between vertices in a network.
Isoreticular: Based on the same net (having the same topology). Decoration: Replacing a vertex in a net by a group of vertices.
MOF-n: Metal-organic framework (with n an integer assigned in SBU: The term ‘secondary building unit’ has been used for some
roughly chronological order). time to describe conceptual fragments of zeolites; in the context of
IRMOF-n: Isoreticular MOF (with n an integer referring to a member reticular chemistry it refers to the geometry of the units defined by the
of the series). points of extension (such as the carboxylate C atoms in most
Interpenetration: A term used to describe the mutual intergrowth of carboxylate MOFs).
two or more networks in a structure where the networks are physically

706 © 2003 Nature Publishing Group NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature
 review article
Second, with few exceptions, MOFs based on M-N linkages in which points of extension (in this example, carboxylate C atoms) defining
the vertex of the network is just a single atom have a tendency to geometrical shapes referred to as secondary building units (SBUs)26.
form structures which collapse upon removal of ‘guest’ atoms from The successful design of rigid frameworks based on such SBUs was
the pores, rendering the structure nonporous. How then can we demonstrated for the first time in MOF-2 (see ref. 43 and below)
ensure that our structures will be robust? and MOF-544. The design principles employed for MOF-5 will be
discussed here to show that SBUs have intrinsic geometric proper-
Secondary building units ties that facilitate network design and help us to address the issues of
To address these questions it is important to first appreciate that it is network synthesis and robustness.
difficult (although not impossible) to attempt a priori synthesis of In MOF-5 (Fig. 1a), Zn4O(CO2)6 units containing four ZnO4
structures, such as those presented above, from simple metal ions tetrahedra with a common vertex and six carboxylate C atoms that
and organic links because ions hold little directional information. define an octahedral SBU are joined together by benzene links. This
This relative lack of directionality often results in flexibility around leads to a cubic network in which the vertices are the octahedral
the metal ion, a multiplicity of possible structures and a general lack SBUs and the edges are the benzene struts. In practice, this
of control—as exemplified by frameworks based on Cu ions and compound was prepared from Zn(II) and BDC acid under con-
bipyridine and related links13,36–42. We have focused on using the ditions pre-determined to yield the octahedral SBU in situ. Because
carboxylate functionality to chelate metal ions and lock them into the SBU and benzene links are relatively large and rigid entities, the
rigid and thus directional metal–oxygen–carbon clusters, with the structure produced has exceptional porosity (indicated by its sorp-
tion), and stability (indicated by thermal analysis and single crystal
X-ray diffraction studies on the completely evacuated frame-
work8,44).
The exceptional stability of MOF-5 can be understood by
comparing its basic network, composed of single atom vertices
(Fig. 1b), with the actual structure of MOF-5, which has cationic
clusters at those vertices (Fig. 1a). The basic network has no
resistance to shear if the links are considered to be universal joints.
However, in the actual MOF-5 structure, the cationic clusters have
a truncated tetrahedral envelope (Fig. 1c), and the rigidly planar
O2C-C6H4-CO2 linkers have a planar slat envelope. The linkage of
these two groups produces an inherently rigid structure held
together by mutually perpendicular hinges.
This approach, based on the concept of SBUs, has been useful in
rationalizing the topologies of MOF structures26, and more impor-
tantly, it has allowed the synthesis and use of a large number of
inorganic and organic SBUs with varying geometries (Fig. 2)26,43–52.
In many of these cases, identifying the reaction conditions that
produce an SBU with a specific geometry in situ means that the
addition of a rigid organic SBU will result in the formation of a pre-
determined network. In other words, with this strategy it is now
possible to control the overall coordination number of the inorganic
and organic SBUs, and therefore the need to identify the networks
that are expected to form from different geometric shapes becomes
particularly acute.

Possible framework topologies

Which framework topology can be expected to form is of critical
importance because there are a large number of possible structures
for each geometrical shape. For example, more than 100 different
topologies (in addition to that of diamond)53 are possible for linking
tetrahedral building blocks together into structures with just one
kind of vertex (that is, all vertices related by symmetry), and the
would-be designer must find some criteria for deciding which will
be selected.
We have recently argued that in general only a small number of
simple, high-symmetry structures will be of overriding general
importance, and they would be expected to form most commonly54.
We call these default structures, in that they would be expected to
form if the assembly reaction involved simple building blocks,
especially single atoms (ions) joined by flexible linkers. More
Figure 1 The MOF-5 structure and its topology. a, The MOF-5 structure shown as complicated structures can be targeted by judicious use of appro-
ZnO4 tetrahedra (blue polyhedra) joined by benzene dicarboxylate linkers (O, red and C, priately shaped SBUs and linkers (we give examples below). A list
black) to give an extended 3D cubic framework with interconnected pores of 8 Å of some default structures54 and some criteria for their classifi-
aperture width and 12 Å pore (yellow sphere) diameter. (Yellow sphere represents the cation55–60 are given in Box 2.
largest sphere that can occupy the pores without coming within the van der Waals size
We have undertaken a systematic survey59 of structures with the
of the framework). b, The topology of the structure (primitive cubic net) shown as a ball-
goal of design in mind. There are just six three-periodic structures
and-stick model. c, The structure shown as the envelopes of the (OZn4)O12 cluster (red
truncated tetrahedron) and benzene dicarboxylate (BDC) ion (blue slat). Note that
with one kind of vertex, edge and ring—we call these regular and
opposing slats are all at 908.
quasi-regular. The next most regular are those with just one kind of
link. Their enumeration is mostly straightforward and—together

NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature © 2003 Nature Publishing Group 707
review article
with the regular structures—they are the most common structures using the considerable sophistication of organic synthetic
used in reticular chemistry, because one almost always attempts a methods10—see Box 1 and Fig. 2 for examples. For a given
synthesis with just one kind of linker. As an illustration, we48 have cationic SBU the geometry of the organic unit plays an
enumerated the nine principal ways in which square SBUs can be important role in directing the structure. This is illustrated
joined by equal linkers to form polyhedra, rods, layers or three- (Fig. 3) for four structures41,61–63 of different dimensionality
dimensional nets. In the next section we show how this enumeration (periodicity—0-D refers to an object within only point sym-
allows particular linked-square structures of each different dimen- metries): a truncated cuboctahedron (0D), a linear rod (1D) the
sionality to be successfully targeted for synthesis using appropriately square grid (2D), and the NbO network (3D) based on linkage
shaped linkers. of paddle-wheel clusters by ditopic linkers. The dihedral angle
between the paddle wheels (square SBUs) plays a crucial role in
Design and synthesis of pre-determined frameworks determining which one of these structures will be produced.
As we noted above, metal-containing (cationic) SBUs are not The reaction conditions that would yield the paddle-wheel
isolatable entities, and so it has been essential to establish the clusters were used in the presence of the appropriate link, to
exact chemical conditions that will yield a specific SBU in situ. give the desired structure—which is in each case the only
On the other hand, organic (anionic) SBUs are pre-assembled possible structure given identical linkers of the shape indicated.

Figure 2 Examples of SBUs from carboxylate MOFs. O, red; N, green; C, black. In shown in green. The geometry for i is an example of a tertiary building unit52 consisting
inorganic units8 metal-oxygen polyhedra are blue, and the polygon or polyhedron of four SBUs (green triangles). The carboxylate carbons of this unit are at the vertices of
defined by carboxylate carbon atoms (SBUs) are red. In organic SBUs the polygons or a trigonal prism.
polyhedrons to which linkers (all –C6H4– units in these examples) are attached are

708 © 2003 Nature Publishing Group NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature
 review article
This approach can be used in linking together other SBU shapes. that give a propeller-shaped binuclear M2(CO2)3 SBU (Fig. 2a)45.
For example, the PtS net is the default structure for linking squares Here, three carboxylate C atoms of the SBU are twisted with respect
(Pt) and tetrahedra (S). Assembly of the paddle wheel with ATC and to the C atoms on the benzene ring to give chiral crystals although
ATB (tetrahedral SBUs) yields MOF-1151 and MOF-3626, which are the bulk material is racemic. Recently, it has been demonstrated that
indeed based on the PtS net. There are only two ways of linking an enantiopure MOF can be prepared from a derivative D-tartaric
square and triangular SBUs—these are the nets named for Pt3O4 and acid (D-PTT) and Zn(II) to give a triangular SBU, which formed a
‘twisted boracite’—and they might both be considered default hexagonal layered network66 that was functionalized with pyridine
structures54 (Box 2). Again, linking a copper-based paddle wheel units to allow the pores to be used for chiral catalysis.
with triangular SBUs, either BTB or BTC, gives MOF-1449 with the It is worth noting that exquisite control over the outcome of
Pt3O4 topology and HKUST-164 with the twisted boracite topology, reticular synthesis has also been developed using site-blocked
respectively. Examination of the parent nets shows that the simpler metal complexes that express angular information. For example,
Pt3O4 net requires the more flexible BTB links. cis-PDA provides 908 binding angles such that with linear
The regular (and hence default) structure for linking triangles is ditopic organic links such as BPY67, the nitrate ligands are
the chiral 3D Si network in SrSi2, which has been a target of reticular displaced by BPY, leading to molecular squares. On the basis
synthesis45,65. Interestingly, it is composed of achiral triangular of this approach other metal complexes of various binding
vertices, which twist with respect to adjacent like vertices by a angles and a variety of branching and bent organic links have
dihedral angle of 70.58 to make an overall chiral arrangement. This been used in the design of other polygons, platonic and archime-
topology was assembled from BTC and Zn(II) under conditions dean polyhedra61,68–71.

Box 2
Framework topologies

The topology underlying periodic structures is that of a periodic net. of them (the exact number depends on some subtle details of what
Crystal chemists, particularly Wells15, have laboured to enumerate, we consider to be allowable structures). There are another twenty or
describe and classify such nets. We58 find the approach based on tiling so structures with two vertices and one edge. Structures with just one
to be particularly fruitful55,56. The tiles are generalized polyhedra (cages) kind of edge (edge transitive) are the most reasonable targets of
which generate the entire structure when packed together. For each designed reticular synthesis, and thus the most important from that
periodic net there is a unique ‘natural’ tiling58; examples are shown in point of view. Even these approximately 50 structures can quickly be
Fig. 5. The adamantane unit with ten vertices is the natural tile of the reduced to a smaller number. For example, there are seven edge-
diamond net. Its four faces are six-membered rings, which we transitive nets with tetrahedral vertices (examples are given in Table
symbolize as [64]. The natural tiling of the lonsdaleite net has equal 1) but the diamond net is the most regular, and thus the default net for
numbers of two tiles that are [63] and [65] (Fig. 5). tetrahedral coordination.
To classify nets we make use of the transitivity of the structure57. If Of course less regular nets can be forced by low-symmetry SBUs.
there are p kinds of vertex, q kinds of edge (link), r kinds of ring (faces For example, 5-coordinated SBUs cannot be linked by just one kind
of tiles) and s kinds of tile, the transitivity is pqrs. It turns out58 that the (that is, symmetry equivalent) of link. Two simple examples are given in
five structures whose natural tilings have transitivity 1111 have Table 1. There is only one 3-coordinated 3D net with symmetry that
coordination figures that are regular polygons or polyhedra and we requires the coordination figure to be an equilateral triangle (that of
call them ‘regular’ (see Table 1). Their importance in reticular SrSi2—see Table 1). The use of 3-coordinated SBUs with a T or Y shape
synthesis is comparable to that of the five platonic solids in other can accordingly lead to a lower-symmetry net such as that of Si in ThSi2
areas of chemistry. Structures 1112 are called ‘quasiregular’ and (also listed). The formation of the rutile net instead of the more
structures 11rs are ‘semiregular’. These are the structures that are symmetrical pyrite net in (6,3)-coordinated structures is often
most likely to form with one kind of SBU joined by one kind of link. ascribable to similar causes.
They can be enumerated fairly easily and there are only about twenty

Table 1 Some important nets and their natural tilings

CN LC Name Vertex figure Transitivity Tiles
...........................................................................................................................................................................................................................................................................................................................................................
þ
3 Y* SrSi2 Triangle 1111 [103]
4 J* NbO Square 1111 [68]
4 D Diamond Tetrahedron 1111 [64]
6 cP Primitive cubic Octahedron 1111 [46]
8 cI Body-centred cubic Cube 1111 [44]
12 cF Face-centred cubic Cuboctahedron 1112 2[34] þ [38]
3 ThSi2 T or Y 1211 [104]
4 CdSO4 Square 1221 [62.82]
4 W* Sodalite Tetrahedron 1121 [46.68]
4 Q Quartz Tetrahedron 1121 [62.82]
4 Lonsdaleite Tetrahedron 1222 [63] þ [65]
6 E Trigonal prism 1122 2[43] þ [43.62]
5 CaB6 Square pyramid 1222 [38] þ [38.46]
5 BN Trigonal bipyramid 1221 [46.62]
6 Polybenzene Hexagon 1121 [46.64]
4,8 Fluorite (CaF2) Tetrahedron, cube 2111 [412]
3,6 Pyrite (FeS2) Triangle, octahedron 2112 2[63] þ [66]
3,6 Rutile (TiO2) T or Y, octahedron 2232 2[4.62] þ [62.82]
3,4 Pt3O4 Triangle, square 2122 3[84] þ 2[86]
3,4 Boracite Triangle, tetrahedron 2122 [64] þ [64.86]
4,4 PtS Square, tetrahedron 2122 [42.82] þ [84]
6,6 NiAs Prism, octahedron 2122 [43] þ [49]
...........................................................................................................................................................................................................................................................................................................................................................
CN is the coordination number of the vertices. LC refers to the symbol for invariant lattice complexes59. The transitivity is defined in the text, and the entries under ‘tiles’ are the face symbols of the tiles.
The first six rows are the five regular and one quasi-regular tilings. For illustrations of these nets see refs 54 and 60.

NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature © 2003 Nature Publishing Group 709
review article
A new class of porous materials are, in the Oxford English Dictionary definition, a “minute…
We have just argued that the rigidity and directionality of SBUs opening… through which fluids or gases can pass”. In addition,
provide a means for the logical synthesis of a wide variety of demonstration of porosity through gas sorption isotherms is
frameworks. An important aspect of this approach is that the necessary to prove permanent porosity. Indeed, we find that
large sizes of SBUs inevitably lead to there being spaces within MOFs having structures based on inorganic SBUs are exceptionally
these structures where solvent and/or counter-ion guests reside. In porous with pore diameters and volumes which exceed those of the
many cases, such solids are interchangeably referred to as “porous” most porous and useful zeolites as shown in Fig. 473–84.
and “open-framework”, without definitive proof of porosity72. We Two important attributes of porous MOFs account for their
prefer to categorize them according to the mobility of the species structural stability and high porosity. First, they are essentially
occupying their cavities. It is inaccurate to use the term ‘open scaffoldings with all or most atoms on internal surfaces, which
framework’ to refer to any low-density or large-scale structure, lead to the highest known surface areas (2,900 m2 g21) and pore
because ‘open’ implies allowing passage. We note that in many cases volumes (1–2 cm3 g21). In other porous solids such as zeolites and
this property is absent (or at least undetermined in the structure in molecular sieves the interior of the pores is largely composed of
question) because passage is hindered by solvent molecules that walls leading to relatively lower specific capacity for sorption.
form strong interactions with the framework; as a result, such Second, in contrast to frameworks of the metal bipyridine type,
solvent molecules are necessary to maintain the crystal architecture. which are known to collapse upon removal of guests, MOFs
Thus we consider that ‘porous’ implies having pores, and pores constructed with oxide SBUs are far more robust, owing to the

Figure 3 The control of dimensionality of linked paddle-wheel units by use of precise MOF-242 (c) and MOF-10163 (d). C, black; O, red; Br, green; metal, blue. Above each
linker geometry. Shown are fragments of the assembled structures of MOP-161 (a), structure, one linker is shown joined to two paddle-wheel units. The yellow spheres are
MOF-222 (Chae, H.K., Rowsell, J.R., O’Keeffe, M., Yaghi, O.M., to be published) (b), at the centres of large cavities in the structures of MOP-1 and MOF-101.

710 © 2003 Nature Publishing Group NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature
 review article
presence of strong metal–oxygen bonds. For example, in MOF-5 diffraction studies82. The key element in synthesizing this series was
each link to an octahedral SBU involves two Zn–O bonds with to use the acid form of the link under reaction conditions known to
energy of 360 kJ mol21 per pair (based on the heat of formation of form the octahedral SBU. In this series, the per cent free volume in
ZnO)85 comparable to the C–C bond energy of 358 kJ mol21 in crystals varies in small increments (1 to 5%) starting from 55.8%
diamond. The corresponding energy for two Cu–O bonds (as in (R5-BDC) to an exceptional 91.1% (TPDC).
linked paddle wheels) is 372 kJ mol21. In contrast, we estimate from It is worth noting that the lowest per cent free volume obtained in
the heats of formation of ammoniates such as CuCl·3NH3 and this series exceeds that found in some of the most open zeolites (Fig.
CuCl2·2NH3 that the energy of the Cu(I)–N coordination bond is 4) in which the free space is 45–50% of the crystal volume. In fact,
about 55 kJ mol21 and the Cu(II)–N bond is about 90 kJ mol21: the fraction of free space in crystals of the most expanded member
very much weaker, and indeed the weak link in metal-bipyridine (TPDC) of this series has only been achievable in non-crystalline
type frameworks. porous systems such as SiO2 xerogels and aerogels87. Furthermore,
the calculated crystal densities (in the absence of guests) of these
Frameworks with adjustable metrics and properties materials also vary in small increments (around 0.1) in the range
An important step forward has been the use of expanded links such 1.00 g cm23 (R5-BDC) to 0.21 g cm23 (TPDC) having densities,
as ECN which can be linked into MOF structures in which the pores which are the lowest reported for any crystalline material. In
can be functionalized by modifying R86, and expanded by lengthen- comparison, the density of Li metal is 0.56 g cm23. The ability to
ing L (see Box 1). When assembled with Ag(I) this link gives control and incrementally vary the metrics in this series has
networks based on the Si network in ThSi2 (Box 2). It is interesting important technological implications. It has already been shown
to note that the unfunctionalized (R ¼ H) link also gives the same that the specific uptake of fuel gases such as methane in one member
network. This finding indicates that the presence of functional (R6-BDC) of the series is the highest known at present and comes
groups (R) does not alter the course of reticulation. In other close to meeting Department of Energy guidelines for use of natural
words, all frameworks produced by this method are isoreticular gas in mobile applications.
(having the same network topology). These compounds have not In cognate work an isoreticular series based on lamellar G and S
been shown to have permanent porosity or rigid frameworks, which (guanidine-sulphonate) hydrogen-bonded materials has been
is not surprising as they are not based on vertices of clusters (SBUs), achieved by varying the sulphonate functionality from DS-PGBDS
but rather on single Ag ion vertices, which do not provide for the (Box 1). Although these frameworks do not have permanent
prerequisite rigidity. porosity as in MOFs, they provide advantages for selective crystal-
On the other hand, frameworks isoreticular with MOF-5 have lization-based separations88.
been synthesized with functionalized structures and shown to have In addition to providing rigidity and structure information, SBUs
permanent porosity. Employing each of the links: o-Br-BDC, R3- can also impart physical properties to the framework. In the
BDC, R4-BDC, R5-BDC, R6-BDC, R7-BDC, 2,6-NDC, BPDC, examples based on the paddle wheel and other binuclear clusters,
HPDC, PDC and TPDC instead of BDC yielded IRMOF-2 to the axial ligands are weakly bound to the porous framework,
IRMOF-16, including the non-interpenetrating structures of allowing their removal by heating or evacuation. The resulting
BPDC, HPDC, PDC and TPDC. Each member of the IRMOF open metal sites act as molecular imprints and site-isolated
series has been isolated and characterized by single-crystal X-ray units51, exhibiting highly selective binding of organic molecules
and, in the case of Tb(III), luminescent behaviour that has formed a
basis for sensing small molecules74.
Although porosity is almost an inevitable outcome of the SBU
approach, other MOFs that have not been examined for their
porosity have shown other interesting properties. Acentric SBUs
can be assembled into MOFs having nonlinear optical properties, as
illustrated for PCA-189. Here, a triangular SBU is linked by PCA-1 to
give a decorated honeycomb-type framework. Similar frameworks
have been designed using paramagnetic metal ions to produce
ferromagnetic MOFs: links such as OX and OXM serve as magnetic
coupling bridges in 2D ferromagnetic frameworks90,91. Recently, 3D
magnetic frameworks have also been targeted using Mn(II), Fe(II),
Co(II), Ni(II), Cu(II) and Cr(III) metal ions and CN92, C(CN)393
and Re6S8(CN)694 links to produce rutile and Prussian-blue type
solids. A mixed-valence conductive framework based on BDC has
been constructed from V(III/IV)80, in which infinite vanadium
oxide SBUs lead to the same SrAl2 topology found for MOF-6995.

Logic of synthesis by design

On the conceptual level, the underlying process for logical synthesis
of MOFs starts with the designer choosing a specific target network,
which is then deconstructed into its component geometric units.
Molecular building blocks with the same geometry as those of the
units are then assembled into a MOF structure that has the target
network topology. The value of this approach is not limited to its
utility in accessing a vast number of materials with predetermined
structures: it also allows the designer to develop a deconstruction
strategy for a target network. Such a deconstruction strategy
becomes indispensable when designing non-default networks.
Figure 4 The porosity of MOFs compared to zeolites. Blue circles, sorption and crystal
MOF structures; red squares, crystal MOF structures; yellow diamonds, some typical
This aspect is also addressed by reticular synthesis as shown in
zeolites. Reference numbers (superscripted) are shown in green.
Fig. 5 for two basic networks; diamond and lonsdaleite (‘hexagonal
diamond’)54,60.

NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature © 2003 Nature Publishing Group 711
review article
The designer of a material having these topologies would recog- Complexity and non-default structures
nize that the simplest deconstruction scheme yields tetrahedral The discussion so far has been limited to the design of frameworks
building blocks in both cases. Thus the assembly of molecules in which the geometry of the links and/or vertices largely determines
with tetrahedral shape would be expected to yield at least one of the outcome. In the formation of neutral frameworks, such as those
these two possible arrangements. However, in practice most struc- of metal carboxylates, the role of the solvent in default structures is
tures (discussed above) obtained from simple tetrahedral building largely a space-filling one, although their function as structure-
blocks are based on the cubic diamond network (Fig. 5a), which directing agents (templates) becomes more important in the design
corresponds to the simplest, highest-symmetry structure (it is the of networks where the building blocks have more than one possible
only regular tetrahedral net) and unless the molecular building blocks default structure96.
contain information to the contrary, it is the default reticulation. In particular, assembly of charged frameworks with counter ions
Reticulations based on the lonsdaleite topology are rare. requires careful matching of the charge, size and volume of counter
We therefore investigated how reticular synthesis could be used to ion to the charge per unit area of the target framework. This was
access non-default structures such as lonsdaleite. It is necessary to illustrated in the assembly of T3 and T4 indium sulphide super-
deconstruct its structure into more elaborate units that express tetrahedra, whose tetrahedral symmetry leads not only to the
structural features unique to its structure. Although the two default diamond-like frameworks97, but also to the slightly less
structures are composed of fused six-membered rings, all such symmetric structures such as that of sodalite and the boron network
rings in diamond have the ‘chair’ conformation, whereas those in in CrB498. Here it is important to consider both the size and volume
the hexagonal form assume both chair and ‘boat’ conformations as of the organic ammonium cations and the charge they provide per
illustrated in Fig. 5. Thus it seems reasonable that to make frame- unit area of the prospective charged framework. We expect that
works based on lonsdaleite, the network should be deconstructed similar factors must be considered to explain the formation of
into units of fused chair and boat rings rather than of individual sodalite, feldspar and quartz MOF networks from tetrahedral Cu(I)
tetrahedra. In this way, the assembly of the corresponding molecular and HPMO99, PRM100 and 2,2 0 -bpy39, respectively. This analysis
indicates that for some target non-default networks, elaborate units
building blocks would yield exclusively the target network. We note
similar to those suggested for the reticular synthesis of lonsdaleite
that the above discussion does not preclude the possibility of
may not be the only option for their design.
achieving lonsdaleite from smaller units, if the structure formation
were driven by the templating effect of included guest molecules96 as
described in the next section. Future prospects
This treatment can be universally applied to the design of Even with only the structures that are based on default networks, the
crystalline molecular arrays of any dimensionality. We emphasize immense wealth and scope of reticular chemistry has become
that the achievement of complex low-symmetry topologies by apparent. In particular, design of the structural skeleton coupled
design requires the construction of complex molecular building with the ability to control its chemical functionalization and
blocks coded specifically for those topologies. adjustment of its metric dimensions present exciting prospects.
We have described here a conceptual approach to this subject, which
we believe has proved useful, and which will continue to guide the
designer.
Specific future challenges include the ability to design sophisti-
cated molecular building blocks that would act as carriers of
structural and functional information to be expressed in a specific
target material. This will be of great benefit to many emerging
technologies involving molecular electronics, miniature fuel cells,
chiral catalysis and materials with hybrid properties. A
doi:10.1038/nature01650.
1. Stein, A., Keller, S. W. & Mallouk, T. E. Turning down the heat: Design and mechanism in solid-state
synthesis. Science 259, 1558–1563 (1993).
2. Yaghi, O. M., O’Keeffe, M. & Kanatzidis, M. G. Design of solids from molecular building blocks:
golden opportunities for solid state chemistry. J. Solid State Chem. 152, 1–2 (2000).
3. Yaghi, O. M., Li, H., Davis, C., Richardson, D. & Groy, T. L. Synthetic strategies, structure patterns,
and emerging properties in the chemistry of modular porous solids. Acc. Chem. Res. 31, 474–484
(1998).
4. Batten, S. T. & Robson, R. Interpenetrating nets: Ordered, periodic entanglement. Angew. Chem. Int.
Edn Engl. 37, 1460–1494 (1998).
5. Férey, G. Building units design and scale chemistry. J. Solid State Chem. 152, 37–48 (2000).
6. Kitagawa, S. & Kondo, M. Functional micropore chemistry of crystalline metal complex-assembled
compounds. Bull. Chem. Soc. Jpn 71, 1739–1753 (1998).
7. Yaghi, O. M., O’Keeffe, M. & Kanatzidis, M. G. Special issue on the design of solids from molecular
building blocks: golden opportunities for solid state chemistry. J. Solid State Chem. 152, 1–321
(2000).
8. Eddaoudi, M. et al. Modular chemistry: Secondary building units as a basis for the design of
highly porous and robust metal-organic carboxylate frameworks. Acc. Chem. Res. 34, 319–330
(2001).
9. Yaghi, O. M., Sun, Z., Richardson, D. A. & Groy, T. L. Directed transformation of molecules to solids:
Synthesis of microporous sulfide from molecular germanium sulfide cages. J. Am. Chem. Soc. 116,
807–808 (1994).
10. Corey, E. J. Retrosynthetic thinking-essentials and examples. Chem. Soc. Rev. 17, 111–133 (1988).
11. Lehn, J. M. Supramolecular chemistry-scope and perspectives. Chem. Scr. 28, 237–262 (1988).
12. Discussion 1: Innovation in Crystal Engineering CrystEngComm [online]; available at khttp://
www.rsc.org/is/journals/current/crystengcomm/ceced.html (August 2002).
13. Moulton, B. & Zaworotko, M. J. From molecules to crystal engineering: Supramolecular isomerism
Figure 5 The deconstruction of diamond and lonsdaleite. a, diamond; b, lonsdaleite and polymorphism in network solids. Chem. Rev. 101, 1629–1658 (2001).
(‘hexagonal diamond’). The units outlined in red are the natural tiles for these 14. Braga, D., Grepioni, F. & Desiraju, G. R. Crystal engineering and organometallic architecture. Chem.
structures. Rev. 98, 1375–1406 (1998).
15. Wells, A. F. Structural Inorganic Chemistry, (Oxford University Press, New York, 1984).

712 © 2003 Nature Publishing Group NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature
 review article
16. Dunbar, K. R. & Heintz, R. A. Chemistry of transition metal cyanide compounds: Modern 51. Chen, B. et al. Cu2(ATC)·6H2O: Design of open metal sites in porous metal-organic crystals (ATC:
perspectives. Prog. Inorg. Chem. 45, 283–391 (1997). 1,3,5,7-adamantane tetracarboxylate). J. Am. Chem. Soc. 122, 11559–11560 (2000).
17. Gramaccioli, C. M. Crystal structure of zinc glutamate dihydrate. Acta Crystallogr. 21, 600–605 52. Chae, H. K. et al. Tertiary building units: Synthesis, structure, and porosity of a metal-organic
(1966). dendrimer framework (MODF-1). J. Am. Chem. Soc. 123, 11482–11483 (2001).
18. Okada, K., Kay, M. I., Cromer, D. T. & Almodovar, I. Crystal structure by neutron diffraction 53. O’Keeffe, M. & Brese, N. E. Uninodal 4-connected nets I: Nets without 3- or 4-rings. Acta
and antiferroelectric phase transition in copper formate tetrahydrate. J. Chem. Phys. 44, 1648–1653 Crystallogr. A 48, 663–669 (1992).
(1966). 54. O’Keeffe, M., Eddaoudi, M., Li, H., Reineke, T. & Yaghi, O. M. Frameworks for extended solids:
19. Jarvis, J. A. J. Crystal structure of a complex of cupric chloride and 1, 2, 4-triazole. Acta. Crystallogr. geometrical design principles. J. Solid State Chem. 152, 3–20 (2000).
15, 964–966 (1962). 55. Delgado Friedrichs, O., Dress, A. W. M., Huson, D. H., Klinowski, J. & Mackay, A. L. Systematic
20. Sterling, C. Crystal structure of wedellite. Science 146, 518–519 (1964). enumeration of crystalline networks. Nature 400, 644–647 (1999).
21. Robl, C. Water clustering in the zeolite-like channel structure of Na2Zn(C6H2(COO)4·9H2O. Mater. 56. O’Keeffe, M. Tiling by numbers. Nature 400, 617–618 (1999).
Res. Bull. 27, 99–107 (1992). 57. Delgado Friedrichs, O. & Huson, D. H. 4-regular vertex-transitive tilings of E3. Discr. Comput. Geom.
22. Weiss, A., Riegler, E., Alt, I., Bohme, H. & Robl, C. Transition metal squarates 1. Chain structures 24, 279–292 (2000).
M(C4O4)·4H2O. Z. Naturforsch. B. 41, 18–24 (1986). 58. Fischer, W. & Koch, E. in International Tables For Crystallography A (ed. Hahn, Th.) Ch. 14 (Kluwer,
23. Kinoshita, Y., Matsubara, I., Higuchi, T. & Saito, Y. The crystal structure of bis Dordrecht, 1983).
(adiponitrilo)copper(I) nitrate. Bull. Chem. Soc. Jpn 32, 1221–1226 (1959). 59. Delgado Friedrichs, O., O’Keeffe, M. & Yaghi, O. M. Three-periodic nets and tilings: regular and
24. Aumüller, A. et al. A radical anion salt of 2, 5-dimethyl-N, N 0 -dicyanoquinonediimine with quasiregular nets. Acta Crystallogr. A 59, 22–27 (2003).
extremely high electrical conductivity. Angew. Chem. Int. Edn Engl. 25, 740–741 (1986). 60. O’Keeffe, M. & Hyde, B. G. Crystal structures, I. Patterns and symmetry (Mineralogical Society of
25. Hoskins, B. F. & Robson, R. Infinite polymeric frameworks consisting of three dimensionally linked America, Washington, DC, 1996).
rod-like segments. J. Am. Chem. Soc. 111, 5962–5964 (1989). 61. Eddaoudi, M. et al. Porous metal-organic polyhedra: 25 angstrom cuboctahedron constructed from
26. Kim, J. et al. Assembly of metal-organic frameworks from large organic and inorganic secondary 12 Cu2(CO2)4 paddle wheel building blocks. J. Am. Chem. Soc. 123, 4368–4369 (2001).
building units: new examples and simplifying principles for complex structures. J. Am. Chem. Soc. 62. Moulton, B., Lu, J., Mondal, A. & Zaworotko, M. Z. Nanoballs: nanoscale faceted polyhedra with
123, 8239–8247 (2001). large windows and cavities. Chem. Commun., 863–864 (2001).
27. Brimah, A. K. et al. Towards organometallic zeolites: spontaneous self-assembly of Et3SnCN, CuCN 63. Eddaoudi, M., Kim, J., O’Keeffe, M. & Yaghi, O. M. Cu2[o-Br-C6H3(CO2)2]2(H2O)2·(DMF)8(H2O)2:
and (nBu4N)CN to supramolecular [(nBu4N)(Et3Sn)2Cu(CN)4]. J. Organometall. Chem. 475, 85–94 A framework deliberately designed to have the NbO structure type. J. Am. Chem. Soc. 124, 376–377
(1994). (2002).
28. Evans, O. R., Wang, Z. Y., Xiong, R. G., Foxman, B. M. & Lin, W. B. Nanoporous interpenetrated 64. Chu, S. S. Y., Lo, S. M. F., Charmant, J. P. H., Orpen, A. G. & Williams, I. D. A chemically
metal-organic diamondoid networks. Inorg. Chem. 38, 2969–2973 (1999). functionalizable nanoporous material Cu3(TMA)2(H2O)3. Science 283, 1148–1150 (1999).
29. Yaghi, O. M., Richardson, D. A., Li, G., Davis, C. E. & Groy, T. L. Open-framework solids with 65. Kepert, C. J., Prior, T. J. & Rosseinsky, M. J. A versatile family of interconvertible microporous chiral
diamond-like structures prepared from clusters and metal-organic building blocks. Mater. Res. Soc. molecular frameworks: the first example of ligand control of network chirality. J. Am. Chem. Soc. 122,
Symp. Proc. 371, 15–19 (1995). 5158–5168 (2000).
30. MacGillivray, L. R., Subramanian, S. & Zaworotko, M. J. Interwoven two- and three-dimensional 66. Seo, J. S. et al. A homochiral metal-organic porous material for enantioselective separation and
coordination polymers through self-assembly of Cu (I) cations with linear bidentate ligands. Chem. catalysis. Nature 404, 982–986 (2000).
Commun., 1325–1326 (1994). 67. Fujita, M. et al. On the structure of transition-metal-linked molecular squares. Chem. Commun.,
31. Hirsch, K. A., Venkataraman, D., Wilson, S. R., Moore, J. S. & Lee, S. Crystallization of 4, 4 0 - 1535–1536 (1996).
biphenyldicarbonitrile with silver (I) salts- a change in topology concomitant with change in 68. Stang, P. T. & Olenyuk, B. Self-assembly, symmetry, and molecular architecture: coordination as the
counterion leading to a nine diamond network. Chem. Commun., 2199–2200 (1995). motif in the rational design of supramolecular metallacyclic polygons and polyhedra. Acc. Chem. Res.
32. Carlucci, L. et al. Self-assembly of a three-dimensional network from two dimensional layers via 30, 502–518 (1997).
metallic spacers: the (3,4)-connected frame of [Ag3(hmt)2][ClO4]3·2H2O 69. MacGillivray, L. R. & Atwood, J. L. A chiral spherical molecular assembly held together by 60
(hmt ¼ hexamethylenetetramine). Chem. Commun., 631–632 (1997). hydrogen bonds. Nature 389, 469–472 (1997).
33. Klein, C., Graf, E., Hosseini, M. W. & De Cian, A. Design and structural analysis of interpenetrated 70. Caulder, D. L. & Raymond, K. N. Supermolecules by design. Acc. Chem. Res. 32, 975–982 (1999).
3-D coordination networks formed from self-assembly using tetrapyridinocyclophane and silver 71. Cotton, F. A., Lin, C. & Murillo, C. A. Supramolecular arrays based on dimetal building units. Acc.
cations. New J. Chem. 25, 207–209 (2001). Chem. Res. 34, 759–771 (2001).
34. Simard, M., Su, D. & Wuest, J. D. Use of hydrogen bonds to control molecular aggregation. 72. Gregg, S. J. & Sing, K. S. W. Adsorption, Surface Area and Porosity (Academic Press, New York,
Self-assembly of three-dimensional networks with large chambers. J. Am. Chem. Soc. 113, 4696–4698 1982).
(1991). 73. Reineke, T. M., Eddaoudi, M., O’Keeffe, M. & Yaghi, O. M. A microporous lanthanide-organic
35. Ermer, O. Fivefold-diamond structure of adamantane-1,3,5,7-tetracarboxylic acid. J. Am. Chem. Soc. framework. Angew. Chem. Int. Edn Engl. 38, 2590–2594 (1999).
110, 3747–3754 (1988). 74. Reineke, T., Eddaoudi, M., Fehr, M., Kelley, D. & Yaghi, O. M. From condensed lanthanide
36. Blake, A. J., Champness, N. R., Hubbertstey, P., Li, W., Withersby, M. A. & Schroder, M. Inorganic coordination solids to microporous frameworks having accessible metal sites. J. Am. Chem. Soc. 121,
crystal engineering using self-assembly of tailored building-blocks. Coord. Chem. Rev. 183, 117–138 1651–1657 (1999).
(1999). 75. Eddaoudi, M., Li, H. & Yaghi, O. M. Highly porous and stable metal-organic frameworks: Structure
37. Carlucci, L., Ciani, G., Macchi, P. & Proserpio, D. M. An unprecedented triply interpenetrated design and sorption properties. J. Am. Chem. Soc. 122, 1391–1397 (2000).
chiral network of ‘square planar’ metal centres from the self assembly of copper(II) nitrate and 76. Kitaura, R., Fujimoto, K., Noro, S., Kondo, M. & Kitagawa, S. A pillared-layer coordination polymer
1,2-bis(4-pyridyl)ethyne. Chem. Commun., 1837–1838 (1998). network displaying hysteretic sorption: [Cu2(pzdc)2(dpyg)]n (pzdc ¼ pyrazine-2,3-dicarboxylate;
38. Carlucci, L., Ciani, G., Proserpio, D. M. & Rizzato, S. New examples of self-catenation in three dpyg ¼ 1,2-Di(4-pyridyl)-glycol). Angew. Chem. Int. Edn Engl. 41, 133–135 (2002).
dimensional co-ordination networks. J. Chem. Soc. Dalton Trans., 3821–3827 (2000). 77. Szostak, R. Molecular Sieves: Principles of Synthesis and Identification (Van Nostrand Reinhold, New
39. Carlucci, L., Ciani, G., Proserpio, D. M. & Rizzato, S. Interlinked molecular squares with [Cu(2,2 0 - York, 1989).
bipy)]2þ corners generating a three-dimensional network of unprecedented topological type. Chem. 78. Bhatia, S. Zeolite Catalysis: Principles and Applications (CRC Press, Boca Raton, Florida, 1990).
Commun., 1198–1199 (2001). 79. Seki, K., Takamizawa, S. & Mori, W. Design and gas adsorption property of a three-dimensional
40. Evans, O. R. & Lin, W. Crystal engineering in NLO materials based on metal-organic coordination coordination polymer with a stable and highly porous framework. Chem. Lett., 332–333 (2001).
networks. Acc. Chem. Res. 35, 511–522 (2002). 80. Millange, F., Serre, C. & Férey, G. Synthesis, structure determination and properties of MIL-53as and
41. Biradha, K., Hongo, Y. & Fujita, M. Open square-grid coordination polymers of the dimension MIL-53ht: the first CrIII hybrid inorganic-organic microporous solids: CrIII(OH)·{O2C-C6H4-
20 £ 20 Å: remarkably stable and crystalline solids even after guest removal. Angew. Chem. Int. Edn CO2}·{HO2-C6H4-CO2H}x. Chem. Commun., 822–823 (2002).
Engl. 39, 3843–3845 (2000). 81. Noro, S., Kitagawa, S., Kondo, M. & Seki, M. A new methane adsorbent, porous coordination
42. Yaghi, O. M., Li, H. & Groy, T. L. A molecular railroad with large pores: synthesis and structure of polymer [{CuSiF6(4,4 0 -bipyridine)2}n]. Angew. Chem. Int. Edn Engl. 39, 2082–2084 (2000).
Ni(4,4 0 -bpy)2.5(H2O)2(ClO4)2·1.5(4,4 0 -bpy)·2H2O. Inorg. Chem. 36, 4292–4293 (1997). 82. Eddaoudi, M. et al. Systematic design of pore size and functionality in isoreticular MOFs and their
43. Li, H., Eddaoudi, M., Groy, T. L. & Yaghi, O. M. Establishing microporosity in open metal-organic application in methane storage. Science 295, 469–472 (2002).
frameworks: gas sorption isotherms for Zn(BDC) (BDC ¼ 1,4-benzenedicarboxylate). J. Am. Chem. 83. Davis, M. E., Montes, C., Hathaway, P. E., Arhancet, J. P., Hasha, D. L. & Garces, J. M.
Soc. 120, 8571–8572 (1998). Physicochemical properties of VPI-5. J. Am. Chem. Soc. 111, 3919–3924 (1989).
44. Li, H., Eddaoudi, M., O’Keeffe, M. & Yaghi, O. M. Design and synthesis of an exceptionally stable and 84. Davis, M. E. Ordered porous materials for emerging applications. Nature 417, 813–821 (2002).
highly porous metal-organic framework. Nature 402, 276–279 (1999). 85. O’Keeffe, M. & Stuart, J. A. Bond energies in solid oxides. Inorg. Chem. 22, 177–179 (1983).
45. Yaghi, O. M., Davis, C. E., Li, G. & Li, H. Selective guest binding by tailored channels in a 3-D porous 86. Kiang, Y. H., Gardner, G. B., Lee, S., Xu, Z. & Lobkovsky, E. B. Variable pore size, variable chemical
zinc(II) benzenetricarboxylate network. J. Am. Chem. Soc. 119, 2861–2868 (1997). functionality, and an example of reactivity within porous phenylacetylene silver salts. J. Am. Chem.
46. Barthelet, K., Riou, D. & Ferey, G. [VIII(H2O)]3O(O2CC6H4CO2)3·(Cl, 9H2O) (MIL-59): A rare Soc. 121, 8204–8215 (1999).
example of vanadocarboxylate with a magnetically frustrated three-dimensional hybrid framework. 87. Husing, N. & Schubert, U. Aerogels airy materials: Chemistry, structure, and properties. Angew.
Chem. Commun., 1492–1493 (2002). Chem. Int. Edn Engl. 37, 23–45 (1998).
47. Yang, S. Y., Long, L. S., Jiang, Y. B., Huang, R. B. & Zheng, L. S. An exceptionally stable metal-organic 88. Holman, K. T., Pivovar, A. M., Swift, J. A. & Ward, M. D. Metric engineering of soft molecular host
framework constructed from the Zn8(SiO4) core. Chem. Mater. 14, 3229–3231 (2002). frameworks. Acc. Chem. Res. 34, 107–118 (2001).
48. Eddaoudi, M. et al. Geometric requirements and examples of important structures in the assembly of 89. Lin, W. B., Wang, Z. & Ma, L. A novel octupolar metal-organic NLO material based on chiral 2D-
square building blocks. Proc. Natl. Acad. Sci. USA 99, 4900–4904 (2002). coordination network. J. Am. Chem. Soc. 121, 11249–11250 (1999).
49. Chen, B., Eddaoudi, M., Hyde, S. T., O’Keeffe, M. & Yaghi, O. M. Interwoven metal-organic 90. Farrell, R. P., Hambley, T. W. & Lay, P. A. A new class of layered microporous materials: crystal
framework on a periodic minimal surface with extra-large pores. Science 291, 1021–1023 structure of disodium pentakis(trimethylphenylammonium) bis[tris(oxalato(2))-chromate(III)]
(2001). chloride pentahydrate. Inorg. Chem. 34, 757–758 (1995).
50. Diskin-Posner, Y., Dahal, S. & Goldberg, I. Crystal engineering of metalloporphyrin zeolite 91. Kahn, O. Chemistry and physics of supramolecular magnetic materials. Acc. Chem. Res. 33, 647–657
analogues. Angew. Chem. Int. Edn Engl. 39, 1288–1292 (2000). (2000).

NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature © 2003 Nature Publishing Group 713
review article
92. Sato, O., Iyoda, T., Fujishima, A. & Hashimoto, K. Electochemically tunable magnetic phase 98. Li, H., Laine, A., O’Keeffe, M. & Yaghi, O. M. Supertetrahedral sulfide crystals with giant cavities and
transition in a high-T c chromium cyanide thin film. Science 271, 49–51 (1996). channels. Science 283, 1145–1147 (1999).
93. Manson, J. L., Campana, C. & Miller, J. S. Interpenetrating three-dimensional rutile-like frameworks. 99. Tabares, L. C., Navarro, J. A. R. & Salas, J. M. Cooperative guest inclusion by a zeolite analogue
Crystal structure and magnetic properties MnII[C(CN)3]2. Chem. Commun., 251–252 (1998). coordination polymer: sorption behaviour with gases and amine and group 1 metal salts. J. Am.
94. Bennett, M. V., Beauvais, L. G., Shores, M. P. & Long, J. R. Expanded prussian blue analogues
Chem. Soc. 123, 383–387 (2001).
incorporating [Re6Se8(CN)6]3-/4- clusters: adjusting porosity via charge balance. J. Am. Chem. Soc.
100. Keller, S. W. An acentric, three dimensional coordination polymer: synthesis and structure of
123, 8022–8032 (2001).
95. Rosi, N. L. et al. Infinite secondary building units and forbidden catenation in metal-organic [Cu(pyrimidine)2]BF4. Angew. Chem. Int. Edn Engl. 36, 247–248 (1997).
frameworks, N. L. Angew. Chem. Int. Edn Engl. 41, 284–287 (2001).
96. Kitazawa, T., Kikuyama, T., Takeda, M. & Iwamoto, T. Silica-mimetic polymorphism of the Cd(CN)2 Acknowledgements O.M.Y. and M.O’K. thank the NSF and the DOE (O.M.Y.) for their support
host lattice depending on the guest G in Cd(CN)2.xG clathrates. J. Chem. Soc. Dalton Trans. 22, of research in their respective laboratories on the subject of this contribution.
3715–3720 (1995).
97. Li, H., Eddaoudi, M., Laine, A., O’Keeffe, M. & Yaghi, O. M. Noninterpenetrating indium sulfide Correspondence and requests for materials should be addressed to O.M.Y.
supertetrahedral cristobalite framework. J. Am. Chem. Soc. 121, 6096–6097 (2001). (oyaghi@umich.edu).

714 © 2003 Nature Publishing Group NATURE | VOL 423 | 12 JUNE 2003 | www.nature.com/nature