Novel open-framework architectures in lanthanide phosphonates

Novel open-framework architectures in lanthanide phosphonates

Solid State Sciences 8 (2006) 397–403 www.elsevier.com/locate/ssscie Novel open-framework architectures in lanthanide phosphonates John A. Groves, Ni...

532KB Sizes 0 Downloads 40 Views

Solid State Sciences 8 (2006) 397–403 www.elsevier.com/locate/ssscie

Novel open-framework architectures in lanthanide phosphonates John A. Groves, Nicholas F. Stephens, Paul A. Wright, Philip Lightfoot ∗ EaStChem, School of Chemistry, University of St Andrews, St Andrews, Fife, KY16 9ST, UK Available online 2 March 2006

Abstract Two novel three-dimensional lanthanide coordination polymers have been prepared hydrothermally with the phosphonic acid N ,N  -piperazine bis(methylenephosphonic acid), H2 O3 PCH2 N(C2 H4 )2 NCH2 PO3 H2 (LH4 ). The structures of Gd2 (LH2 )3 ·3H2 O (I) and Nd2 (LH2 )3 ·9H2 O (II) have been characterised by single crystal X-ray techniques. One-dimensional ‘lanthanide-phosphate’ chains are a key feature in both structures, although there are major structural differences between the chains, with (I) displaying octahedral GdO6 coordination and (II) showing eightcoordinate NdO8 polyhedra. In each case, three-dimensional connectivity is completed by coordination of the phosphonate group resulting in open framework structures encapsulating loosely bound water molecules. Isostructural Y3+ and Yb3+ analogues of (I) have been prepared, suggesting that cation size is a key factor in controlling the differing reaction products. In the case of Y2 (LH2 )3 ·5H2 O, isostructural to (I), it is shown that the extra-framework water molecules may be removed reversibly without framework collapse. Structural relationships to other known lanthanide phosphonates are discussed. © 2006 Elsevier SAS. All rights reserved.

1. Introduction There is considerable current interest in the preparation of microporous organic-inorganic hybrids, because they are expected to offer alternative adsorptive properties and applications to those for more traditional, fully inorganic molecular sieves. Most attention currently focuses on di- and tricarboxylate [1,2] and diamine-based [3] metal organic frameworks, but there are also interesting examples of porous solids prepared by linking inorganic units (isolated polyhedra or clusters, chains or sheets of polyhedra) by diphosphonic acids or phosphonocarboxylic acids. The best known of these are the pillared zirconium phosphonates of Clearfield [4,5]. Other recent examples include framework diphosphonates of divalent transition metals such as cobalt [6], zinc [7] and pillared layered iron and aluminium diphosphonates [8–10]. We are currently investigating the chemistry of rare earth and related trivalent metal diphosphonates, because the large and variable coordination spheres of these metals, and their ability to form clusters, offer different ways by which diphosphonate ligands can be coordinated to give extended frameworks. * Corresponding author. Tel.: +44 1334 463841.

E-mail address: [email protected] (P. Lightfoot). 1293-2558/$ – see front matter © 2006 Elsevier SAS. All rights reserved. doi:10.1016/j.solidstatesciences.2006.02.018

A growing series of lanthanide dicarboxylates and carboxylatephosphonates which exhibit such features has already been reported [11,12]. In addition, our own studies on use of the ligand N ,N  -piperazine bis(methylenephosphonic acid) in the preparation of two novel lanthanum framework structures have recently been described [13]. The incorporation of specific lanthanides into porous silicates and carboxyphosphonates has also been demonstrated to introduce luminescent properties that are sensitive to the state of hydration, and are therefore being investigated as sensors [12–15]. Our aim is therefore to prepare novel microporous lanthanide phosphonates with applications in adsorption and catalysis. 2. Experimental 2.1. Synthesis N ,N  -piperazine (bis methylene phosphonic acid) was synthesised using a new variation of the modified Mannich reaction [16]. As before [13], HCl was used as a catalyst, and in addition a second reaction was carried out using a comparable aqueous solution of HBr, made up from 37.5 cm3 HBr (48% aq., Aldrich) in 62.5 cm3 water. In both cases the white precipitate was washed with cold water until the pH was raised from 1 to 5.5. The white residue was dried and analysed based on a

398

J.A. Groves et al. / Solid State Sciences 8 (2006) 397–403

Table 1 Representative synthetic conditions and products Reaction

Metal source

Metal (mol)

Ligand (mol)

Temp (◦ C)

Time (hours)

Initial pH

Products

1a

GdCl3 GdCl3 GdCl3 GdCl3 Gd(NO3 )3 Y(Ac)3 Y(Ac)3 YCl3 YCl3 YbCl3 YbCl3 YbBr3 NdCl3

0.002 0.002 0.001 0.001 0.001 0.002 0.002 0.001 0.002 0.001 0.001 0.001 0.001

0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015 0.0015

220 160 140 190 190 160 190 160 220 190 160 190 190

120 200 120 200 200 200 200 200 120 200 200 200 160

6 7.5–8 5.5 5.5 5.5 6 6 6 6 5.5 6 5.5 5.5

I + unknown Some I unknown unknown unknown I unknown I unknown I I + unknown I II + unknown

2 3 4 5 6c 7 8 9 10 11 12 13b

All syntheses were carried out in 1 mol H2 O. Reaction pH was adjusted by dropwise addition of 2M NaOH. a Single crystal structure of phase I from this batch. b Single crystal structure of phase II from this batch. c Dehydration study carried out on this batch.

composition LH4 ·H2 O (calc. for C6 H18 N2 P2 O6 ·H2 O: C 24.67, H 6.20, N 9.59%). For HCl catalysed: C 25.08, H 6.11, N 9.73%; for HBr catalysed: C 24.88, H 6.01, N 9.41%. Based on our previous work with lanthanum compounds, similar reaction conditions were used with GdCl3 as the metal salt. All reactions used reagents of 99.9% purity or higher. After several attempts using different reaction conditions (Table 1) a single crystal of (I) suitable for structure determination was isolated. Although the bulk material obtained from this reaction was not pure, the phase of interest could be concentrated by sonication and separation of the product, and this was confirmed by powder X-ray diffraction and elemental analysis to be of high purity: C 17.49%, H 3.42%, N 6.62% (calcd. for C18 H48 Gd2 N6 O21 P6 : C 18.24%, H 4.08%, N 7.09%). Repeated attempts to prepare the phase using the expected stoichiometric metal:ligand ratio were unsuccessful. In order to explore the stability field of this phase as a function of lanthanide cation size, similar reactions were carried out with salts of Nd3+ , Eu3+ , Y3+ and Yb3+ . For the smaller cations, phase (I) could be prepared more easily than with Gd3+ ; for example, for Y3+ both YCl3 and Y(Ac)3 produced a pure phase at 160 ◦ C, in metal:ligand molar ratios of either 4 : 3 or 2 : 3, with higher temperatures being somewhat less successful. For Yb3+ the range of successful conditions was even larger. Pure samples of both Y3+ and Yb3+ phases could be produced by direct reaction (for Y2 (LH2 )3 ·5H2 O, found C 18.9%, H 4.6%, N 7.2% (calcd. for C18 H52 Y2 N6 O23 P6 : C 19.94%, H 4.83%, N 7.75%); For Yb(I)·3H2 O, found C 16.8%, H 4.0%, N 6.9% (calcd. for C18 H48 Yb2 N6 O21 P6 : C 17.77%, H 3.98%, N 6.91%). Thermogravimetric analyses for all pure phases were carried out on a TA Instruments SDT2960 dual TGA/DTA, in the temperature range 25–500 ◦ C, in air, at a heating rate of 5 ◦ C min−1 . In the case of Eu3+ , all attempts at preparing phase (I) proved unsuccessful. There is therefore a clear indication that cation size is a major factor in allowing phase (I) to be produced, with lanthanides smaller than Gd3+ forming the phase

more easily, and those larger failing to form the phase. In the case of Nd3+ , a product different from both the current phase (I) and the previously characterised La3+ phases was isolated, although we have not yet managed to optimise conditions for phase purity. Crystals of Nd2 (LH2 )3 ·9H2 O (II) were prepared in a similar manner (see Table 1) to those described above. 2.2. Crystallography Single crystal X-ray diffraction studies were carried out on a Rigaku Mercury CCD diffractometer with graphite monochromated Mo Kα radiation. Intensity data were collected using 0.3◦ steps to give at least a full hemisphere of coverage. All data sets were corrected for absorption via multiscan methods. Data analyses used the SHELXS and SHELXL packages. Details of the crystal structure determinations are given in Table 2 and selected bond lengths in Table 3. Table 2 Crystallographic data for Gd2 (LH2 )3 ·3H2 O (I) and Nd2 (LH2 )3 ·9H2 O (II) Formula Crystal system Space group a (Å) b (Å) c (Å) α (deg) β (deg) γ (deg) V (Å3 ) Z Crystal morphology Data collection temperature λ (Å) Total / unique reflections Ind. reflections (I > 2σ (I )) Parameters Final R1, wR2

I

II

C18 H48 Gd2 N6 O21 P6 orthorhombic P221 21 (No. 18) 8.712(3) 9.859(3) 23.176(8) 90 90 90 1991(1) 2 Needle 93 K Mo Kα 11565 / 3465 3381 229 0.045, 0.131

C18 H60 Nd2 N6 O27 P6 triclinic P1¯ 9.547(3) 11.499(3) 11.592(4) 108.941(2) 111.851(4) 92.174(3) 1098.7(6) 1 Needle 93 K Mo Kα 7190 / 3515 3398 279 0.044, 0.129

J.A. Groves et al. / Solid State Sciences 8 (2006) 397–403

399

Table 3 Selected bond lengths (Å) for Gd2 (LH2 )3 ·3H2 O (I) and Nd2 (LH2 )3 ·9H2 O (II) (I)

(II)

Gd1 O1 Gd1 O3 Gd1 O4 Gd1 O5 Gd1 O7 Gd1 O8

2.269(7) 2.259(7) 2.253(7) 2.292(7) 2.263(7) 2.304(7)

Nd1 O1 Nd1 O1 Nd1 O2 Nd1 O4 Nd1 O5 Nd1 O7 Nd1 O8 Nd1 O9b Nd1 O10

2.394(5) 2.599(5) 2.535(5) 2.353(5) 2.402(5) 2.454(6) 2.379(6) 2.522(12) 2.568(13)

P1 O1 P1 O2 P1 O3 P1 C1 P2 O4 P2 O5 P2 O6 P2 C6 P3 O7 P3 O8 P3 O9 P3 C7

1.519(7) 1.516(8) 1.505(8) 1.816(11) 1.518(8) 1.515(7) 1.531(7) 1.844(10) 1.510(8) 1.509(7) 1.550(9) 1.824(11)

P1 O1 P1 O2 P1 O3 P1 C1 P2 O4 P2 O5 P2 O6 P2 C4 P3a O7 P3a O8 P3a O9a P3a C7 P3b O7 P3b O8 P3b O9b

1.521(5) 1.544(5) 1.502(5) 1.834(7) 1.505(6) 1.517(5) 1.520(6) 1.827(8) 1.439(7) 1.618(7) 1.538(14) 1.859(6) 1.579(7) 1.503(7) 1.555(16)

Hydrogen bond contacts O6· · ·O9 O9· · ·O12i O10· · ·O12ii O11· · ·N1iii O12· · ·N3iv

2.47(1) 2.81(1) 3.10(1) 2.91(1) 3.09(1)

i 1 − x, 1/2 + y, 1/2 − z; ii 1 − x, 1/2 − y, −1/2 + z; iii x, 1 − y, 1 − z; iv −x, −1/2 + y, 1/2 − z.

Powder X-ray diffraction data were collected on a Stoe STADI/P transmission diffractometer using CuKα1 radiation. In addition to confirming phase purity for the as-made materials, a study of the dehydration behaviour of Y2 (LH2 )3 ·5H2 O (isomorphous with (I)) was carried out. The material was dehydrated under vacuum at 100 ◦ C for 2 hours, sealed in a capillary, and examined by powder diffraction. Subsequently, the same sample was opened to moist air and allowed to rehydrate over 48 hours, before again being examined by powder diffraction. 3. Results and discussion 3.1. Crystal structure of Gd2 (LH2 )3 ·3H2 O (I) The crystal structure of (I) may be considered as a threedimensional open framework, including loosely-bound guest water molecules within a two-dimensional channel system. The asymmetric unit consists of one unique Gd site, one phosphonate ligand on a general position and a further phosphonate on an inversion centre. This second phosphonate group exhibits some disorder and the final model involves splitting of the nitrogen and one carbon atom into two equally occupied sites. Three extra-framework sites are identified as water molecules.

Fig. 1. Portion of the inorganic chain in (I), showing octahedral coordination around the Gd centre.

A key building unit of the structure is an essentially inorganic gadolinium ‘phosphate’ chain which runs along the b-axis of the unit cell. As shown in Fig. 1, this chain exhibits Gd3+ in almost regular octahedral coordination; a coordination number of six for a trivalent lanthanide is rather unusual in coordination complexes and polymers. There are no direct Gd–O–Gd links within the chain; rather, the GdO6 octahedra are linked by three bridging phosphonate groups on opposite octahedral faces. Chains of this type (see also Fig. 8 for a polyhedral view) are analogous to those in purely inorganic phosphate chains such as those in [In8 (HPO4 )14 (H2 O)6 ](H2 O)5 (H3 O)(C3 N2 H5 )3 [17] and [(H3 NC4 H8 NH3 )3 ][(Sc(OH2 ))6 Sc2 (HPO4 )12 (PO4 )2 ] [18]. All three crystallographically distinct phosphonate groups act as bidentate donors to Gd, with P(1)–O(2) clearly representing a terminal P=O bond, at 1.516(8) Å. The P(2)– O(6) and P(3)–O(9) bonds, in contrast, exhibiting a significant lengthening, to 1.531(7) and 1.550(9) Å, respectively. This, together with the very short O(6)–O(9) contact of 2.47(1) Å clearly suggests a strong hydrogen bond and, indeed, a proton was successfully located on O(9). For overall charge balance, this leads to formulation of (I) as Gd2(O3 PCH2 NH(C2 H4 )2 NHCH2 PO3 )2 (HO3 PCH2 N(C2 H4 )2 NCH2 PO3 H)·3H2 O, with N-protonation of the P(1)/P(2) phosphonate, but O-protonation of the P(3) phosphonate. The inorganic chains are further linked into relatively condensed sheets in the ab plane via the P(1)/P(2) phosphonate group, and thence into a three-dimensional framework via the P(3) phosphonate (Fig. 2). A relatively low density part of the framework in the inter-layer region around the P(3) phosphonate accommodates three water molecules, which are not coordinated to Gd, but partake in hydrogen-bonding to the framework O and N atoms. The most significant contacts are given in Table 3. 3.2. Dehydration of Y2 (LH2 )3 ·5H2 O Thermogravimetric analysis (in air) was carried out for each of the three phase (I) materials (Gd, Y and Yb). Each showed

400

J.A. Groves et al. / Solid State Sciences 8 (2006) 397–403

Fig. 4. Powder X-ray diffraction for Y(I)·5H2 O: as-made (bottom); dehydrated (middle); rehydrated (top).

Fig. 2. Open framework structure of (I) projected along the b-axis. Note the extra-framework water positions with H-bonds shown as dotted lines.

drated phase was successfully indexed using the Visser algorithm [19] which produced a monoclinic unit cell, a = 8.757, b = 9.901, c = 19.965 Å, β = 94.40◦ . This displays a remarkable similarity to the original orthorhombic cell of the as-made material, with very little change in the a- and b-axes but a substantial decrease in the c-parameter. A quantitative structural analysis is underway to establish the mechanism by which this dehydration is accommodated by the framework. 3.3. Crystal structure of Nd2 (LH2 )3 ·9H2 O (II)

Fig. 3. TGA trace showing dehydration of Y(I)·5H2 O, resulting in a thermally stable dehydrated phase.

a weight loss corresponding to dehydration at relatively low temperature, followed by a stability plateau. The pure yttrium material was studied in more detail. The TGA revealed a continuous weight loss of about 7.8% between 50 and 105 ◦ C, followed by a stability plateau up to 350 ◦ C and a further abrupt weight loss of 13% up to 380 ◦ C (Fig. 3). The first weight loss is consistent with a composition of Y2 (LH2 )3 ·5H2 O, although this has not been verified by single crystal X-ray diffraction. A dehydrated polycrystalline sample was studied by powder X-ray diffraction (see Section 2). Fig. 4 shows the data for as-made Y2 (LH2 )3 ·5H2 O, together with those from the dehydrated material and a sample rehydrated by leaving in air at room temperature for 48 hours. It is clear that a fully reversible dehydration occurs. The powder pattern of the dehy-

The structure of (II) may also be described as a threedimensional framework built up from phosphonate crosslinking of essentially inorganic chains. The framework consists of one unique Nd site and three independent phosphonate moieties, all situated on inversion centres. A portion of the neodymium ‘phosphate’ chain is shown in Fig. 5 (see also Fig. 8 for a polyhedral view). This structure exhibits disorder around one of the phosphonate groups (P(3)), and the views in Fig. 5 highlight both configurations, which occur as a 50 : 50 disorder. In both cases the Nd is eight-fold coordinated and the chain consists of pairs of NdO8 polyhedra sharing a common edge, separated by the bridging phosphonate group at P(3), leading to alternating short and long Nd· · ·Nd distances of 3.99 and 5.72 Å, respectively. The P(1) tetrahedron acts as both a bidentate and monodentate ligand to neighbouring Nd atoms, with O(1) triply-bridging. The P(1)–O(2) bond length of 1.544(5) Å may provide evidence of protonation, though, as in compound (I), charge balance simply requires protonation at all the N-atoms; there is no evidence for H-bonding from O(2). P(2) acts as a monodentate ligand to two adjacent Nd atoms. Both P(1) and P(2) provide one terminal P=O bond. In the case of P(3), the disorder provides two situations: the disorder can be regarded as a simple ‘inversion’ of the tetrahedron corresponding to positions P(3a)/O(9a) and P(3b)/O(9b), with atoms O(7), O(8) and C(7) common to both orientations. For the P(3b) option, this tetrahedron acts as a bidentate ligand to

J.A. Groves et al. / Solid State Sciences 8 (2006) 397–403

401

Fig. 7. Open framework structure of (II) projected along the c-axis. Note the extra-framework water positions with H-bonds shown as dotted lines.

Fig. 5. Portion of the inorganic chain in (II), showing the two different conformations around the P3 site: (a) O(9a) terminal, with additional H2 O site at O(10); (b) O(9b) acting as an additional donor to Nd.

Fig. 6. Linking of neighbouring inorganic chains in (II) via the P3 phosphonate group.

one Nd atom and a monodentate ligand to the next, whereas for the P(3a) option, this tetrahedron acts as monodentate ligand to two neighbouring Nd atoms, and an additional water molecule is present (O(10)) in order to complete the coordination around Nd. The chains extend parallel to the crystallographic a-axis. Direct links to neighbouring chains within the ac plane are made via the P(3) phosphonate group (Fig. 6). Small channels within the resulting layer are occupied by water molecules (O(102)), which are held in place by strong hydrogen bonds to the terminal P=O groups at O(3) and O(9a) (hydrogen bond contacts: O(102)· · ·O(9a) 2.48(1); O(102)· · ·O(3) 2.70(1); O(102)· · ·O(3) 2.79(1)). Perpendicular to this direction the layers are further linked in the b-direction by the remaining two phosphonate groups (Fig. 7). Here, larger channels are

produced, which are occupied by water molecules at several distinct sites, again held in place by extensive H-bonding. 3.4. Relationship to existing lanthanide phosphonates There are relatively few structurally characterised lanthanide diphosphonates, indeed only three previous papers report examples having extended lattice structures. All of these structures may be regarded as having ‘inorganic’ chain-like building blocks. In 1998 Nash et al. [20] reported the crystal structures of three different trivalent lanthanide complexes of 1hydroxyethane-1,1-diphosphonic acid. The larger lanthanides, Nd and Eu, adopt differing structures based on the same type of chain, shown in Fig. 8(a). Square-antiprismatic LnO8 polyhedra are linked into linear chains by bridging PO3 C-tetrahedral, with four tetrahedra on each face of the LnO8 polyhedron. For the smaller lanthanides (Tb–Lu) the coordination reduces to LnO7 , monocapped trigonal prismatic, but a similar linear chain arrangement occurs, albeit with alternately 3 and 4 linking tetrahedral (Fig. 8(b)). In all these structures the overall architecture may be regarded as one-dimensional, with no direct interchain covalent links. Soon after, Serpaggi and Férey [21] reported the series LnIII H[O3 P(CH2 )n PO3 ] (n = 1–3) for all lanthanides. All adopt the same basic pillared-layer structure, with the interlayer distance being dictated by the alkyl chain length. The inorganic chains in all these structures consist of linear chains of edge-sharing LnO8 dodecahedra with the phosphonate groups acting as bidentate ligands to two adjacent Ln atoms (Fig. 8(c)). Our own recent work [13] produced two closely related lanthanum phosphonates using N,N -piperazine bis(methylenephosphonic acid). Both these structures exhibit LaO7 polyhedra, with no direct La–O–La links, but linked via four PO3 C-tetrahedra per face into zig-zag chains (Fig. 8(d)). These may be compared and contrasted with the linear chains in the structure of Nash et al.; the zig-zag chains arise from the differing nature of the PO3 C-bridging groups, which in

402

J.A. Groves et al. / Solid State Sciences 8 (2006) 397–403

(a)

(b)

(c)

(d)

(e)

(f)

(g) Fig. 8. Polyhedral representations of the ‘inorganic’ chains present in all the known polymeric lanthanide phosphonates. (a) Nash structure [20] for the larger lanthanides, e.g., Nd, showing separated LnO8 polyhedra; (b) Nash structure [20] for the smaller lanthanides, e.g., Tb, showing separated LnO7 polyhedra; (c) Serpaggi structure [21] showing edge-sharing LnO8 polyhedra; (d) Groves La structure [13] showing separated LnO7 polyhedra; (e) Present phase (I) showing separated LnO6 polyhedra (f, g). Present phase (II) showing two different linkage modes of separated edge-sharing LnO8 dimers.

the case of Fig. 8(d) bridge three adjacent LnO7 polyhedra rather than two. In these cases there is inter-chain connectivity via the phosphonate groups, resulting in three-dimensional framework structures. Polyhedral representations of the inorganic chains present in the two title compounds are shown in Figs. 8(e)–(g). The linear chains in (I) clearly have similarities to those in Figs. 8(a) and (b), but with a reduction

of coordination number around the lanthanide. The two differing chain types in (II) have aspects in common with both the Serpaggi and Nash structure types, containing both direct edge-sharing links and phosphonate-bridges between the LnO8 polyhedra. It is difficult to draw any firm conclusions from the limited data currently available on the structural properties of lanthanide diphosphonates. Certainly there is no obvious cor-

J.A. Groves et al. / Solid State Sciences 8 (2006) 397–403

relation of lanthanide ion size on coordination number, or of the chemical nature or chain-length of the diphosphonic acid with the overall connectivity of either the inorganic chains or the resultant three-dimensional structure. Further chemical and structural explorations are underway to develop these systematics.

403

Acknowledgements We thank Prof. A.M.Z. Slawin and Dr. Yang Li for assistance in X-ray data collection, and the University of St Andrews for funding. References

4. Conclusion Two novel lanthanide phosphonate framework materials have been prepared and structurally characterised. Both crystal structures are characterised by continuous ‘inorganic’ chains, bridged into three-dimensional frameworks by the phosphonate ligands. The existence of two distinct structure types prepared under similar conditions is a direct consequence of the cation size effect, whereby smaller lanthanides (smaller than Eu3+ ) adopt the phase (I) structure, with octahedral Ln3+ , whereas the larger Nd3+ adopts the phase (II) structure, containing eight-fold coordinated Ln3+ . Gd2 (LH2 )3 ·3H2 O (I) and its Y and Yb analogues undergo a reversible dehydration, and indexing of the dehydrated Y form by powder diffraction shows a unit cell exhibiting a large uniaxial compression relative to the as-made material. Further studies to prepare a phasepure sample of Nd2 (LH2 )3 ·9H2 O (II) and to determine the crystal structure, and adsorption properties of phase (I) are in progress. CCDC (deposition numbers 299040 and 299041) contains the supplementary crystallographic data for this paper. These data can be obtained online free of charge (or from the Cambridge Crystallographic Data Centre, 12, Union Road, Cambridge CB2 1EZ, UK; fax: (+44) 1223 336 033; or [email protected]).

[1] M. Eddaoudi, J. Kim, N. Rosi, D. Vodak, J. Wachter, M. O’Keeffe, O.M. Yaghi, Science 295 (2002) 469. [2] G. Férey, C. Mellot-Draznieks, C. Serre, F. Millange, Acc. Chem. Res. 38 (2005) 217. [3] X. Zhao, B. Xiao, A.J. Fletcher, K.M. Thomas, D. Bradshaw, M.J. Rosseinsky, Science 306 (2004) 1012. [4] H. Byrd, A. Clearfield, D. Poojary, K.P. Reis, M.E. Thompson, Chem. Mater. 8 (1996) 2239. [5] A. Clearfield, Z. Wang, J. Chem. Soc. Dalton Trans. (2002) 2937. [6] D.L. Lohse, S.C. Sevov, Angew. Chem. Int. Ed. 36 (1997) 1619. [7] J.A. Groves, P.A. Wright, P. Lightfoot, Dalton Trans. (2005) 2007. [8] H.G. Harvey, S.J. Teat, M.P. Attfield, J. Mater. Chem. 10 (2000) 2632. [9] R.N. Devi, P. Wormald, P.A. Cox, P.A. Wright, Chem. Mater. 16 (2004) 2229. [10] C.A. Merrill, A.K. Cheetham, Inorg. Chem. 44 (2005) 5273. [11] C. Serre, J. Marrot, G. Férey, Inorg. Chem. 44 (2005) 654. [12] C. Serre, N. Stock, T. Bein, G. Férey, Inorg. Chem. 43 (2004) 3159. [13] J.A. Groves, P.A. Wright, P. Lightfoot, Inorg. Chem. 44 (2005) 1736. [14] D. Ananias, M. Kostova, F.A.A. Paz, A. Ferreira, L.D. Carlos, J. Klinowski, J. Rocha, J. Amer. Chem. Soc. 126 (2004) 10410. [15] D. Ananias, A. Ferreira, J. Rocha, P. Ferreira, J.P. Rainho, C. Morais, L.D. Carlos, J. Amer. Chem. Soc. 123 (2001) 5735. [16] K. Moedritzer, R.R. Irani, J. Org. Chem. 31 (1966) 1603. [17] A.M. Chippindale, S.J. Bench, A.R. Cowley, Chem. Mater. 9 (1996) 259. [18] S.R. Miller, A.M.Z. Slawin, P. Wormald, P.A. Wright, J. Solid State Chem. 178 (2005) 1738. [19] J.W. Visser, J. Appl. Crystallogr. 2 (1969) 89. [20] K.L. Nash, R.D. Rogers, J. Ferraro, J. Zhang, Inorg. Chim. Acta 269 (1998) 211. [21] F. Serpaggi, G. Férey, J. Mater. Chem. 8 (1998) 2749.