Mixed valence mono- and hetero-metallic grid catenanes

Multicomponent self-assembly was employed to obtain, in the solid state, a series of mixed valence mono- and hetero-metallic grid catenanes, which were characterized by single crystal X-ray diffraction.


Introduction
In the last half century, the scientic community has heavily focused on the design and synthesis of structurally demanding molecular architectures following the desire for complexity, aesthetic beauty, topological novelty and, most importantly, novel emerging properties. 1 To this day, differently interlocked and/or entangled species, such as rotaxanes, 1a catenanes, 1j,k knots 1h and other complex structures 1c,d,f,i have been vigorously pursued. In particular, catenanes 3 consist of two or more mechanically interlocked macro-or metallocycles, which cannot be separated without breaking a bond. The rst discovery of a metal complex [2]-catenane selfassembled in one step was reported by Fujita and co-workers in 1994, 4 yet exploration into this new eld remained mostly limited to [M 2 L 2 ] 2 [2]-catenane species (M ¼ metal, L ¼ ligand). Only very recently have some examples of more complex catenanes been reported. 5 For instance, Clever et al. reported on a dimeric [M 2 L 4 ] 2 interpenetrated coordination cage, which forms by the assembly of Pd(II) metal ions with a simple ditopic pyridyl ligand. 5a Notably, the same group, by using a very similar but slightly shorter ligand, was able to achieve double ([M 2 L 4 ] 2 ) and triple ([M 2 L 2 ] 3 ) catenation in a stepwise process triggered by halide anions. 5b A fully organic [3]-catenane was, instead, obtained by Nitschke et al., by employing an elegant but intricate self-assembly procedure based on a metal ion template and utilizing ve different precursor molecules. 5c An increase in the number of catenated species can be achieved, as demonstrated in the dynamic systems comprised of poly-catenanes (up to [7]-catenane) based on a tetrahedrally-shaped cage framework. 5d Leigh et al. have recently demonstrated the preparation of a Solomon link, 5e a pentafoil knot 5f and a Star of David catenane 5g in one 5e,f or two steps, 5g through careful ligand design while also taking advantage of iron(II) and anion (chloride 5f or sulfate 5g ) templating effects. Finally, the assembly of a two-dimensional metal-organic network possessing Borromean links has recently been reported by Hardie et al. 5h In the pursuit of the construction of chemical objects of considerable structural complexity, multicomponent selfassembly has been proven to be a very successful technique, 6 as demonstrated by many of the examples described above. It relies on the designed organization of carefully selected subcomponents, which are linked together through the simultaneous formation of covalent and/or coordinative bonds, ultimately yielding the desired superstructure.
An assembly constituted by arrays of linear molecules held together and shaped into a discrete crisscrossed two-dimensionally ordered framework is usually termed a molecular grid or, when metal ions interconnect a set of organic ligands, a metallogrid. 1e Although grid-like systems 7 have become relatively common over the last decade, 1e and their potential for applications has been demonstrated, 1i,t,2 a higher-level organization of grids as components remains quite rare. 1e,g In the context of interlocked systems, only two notable examples involving metallogrids can be found, namely that of the so far only homometallic grid catenane by Thompson et al. 14 and the recent elegant use of a metallogrid as a scaffold in the synthesis of a Solomon link, reported by Leigh et al. 1b To the best of our knowledge, no examples of catenanes formed by heterometallic grids 7a,1m,n have been reported to this date.
Here, we report on the isolation in the solid state of three [2]catenane species where the macrocyclic ring units are made of [2 Â 2] metal grids. These metal-organic architectures are formally created by the self-assembly of 32 components (8 Â 1, 16 Â 2 and 8 Â metal, with 2 ¼ 2-formylpyridine and metal ¼ Zn(II) or Cu(II)/(I)) and they spontaneously precipitate as crystalline solids from a H 2 O/DMF mixture. The rst species, 4 (Scheme 1), is a mono-metallic mixed valence Cu(II)/Cu(I) grid catenane, while the other two, 5 and 6 (Scheme 1), are bimetallic Cu/Zn complexes. Among the eight metal ions that constitute the nodes of the two concatenated grid units, six of them can be considered as external while the remaining two can be viewed as internal with respect to the whole assembly (Scheme 1). Interestingly, in all cases, at least one of the internal copper atoms displays different coordination geometry with respect to the other external ones, along with a conspicuous conformational change of the organic ligands that ll its coordination sphere. Importantly, this conformational change renders the ligand's NH moieties available to establish additional hydrogen bonding (HB) interactions, which, as it will be shown, are important to allow for the catenation of the two [2 Â 2] grids in 4 and 5. In addition to this, while in the case of 4 and 5 we observed a tetrahedral coordination geometry for one of the internal copper ions, which led us to assume a Cu(II)-to-Cu(I) transformation, 9 in 6, both internal copper ions are instead found to be penta-coordinated. Indeed, the versatility of the coordination of the Cu(II)/Cu(I) couple appears to be fundamental for the formation of the [2]-catenane species, since the use of Zn(II) salts only led to more trivial [2 Â 2] grid-like structures, such as 3. To the best of our knowledge, structures 4-6 represent a unique set of [2]-catenane systems including the heterometallic grid catenanes 5 and 6, which are the rst of their kind.

Results and discussion
Compound 1 (Scheme 1) can be easily synthesized by the addition of hydrazine to the sodium salt of dimethyl 5-sulfoisophthalate in MeOH. 10 The slow evaporation of a 1 : 2 : 1 mixture of 1, 2-formylpyridine, 2, and Zn(II) acetate in a 2 : 1 H 2 O : DMF mixture produced good quality yellow single crystals of 3 aer approximately one week. Their X-ray diffraction analysis revealed the formation of the [2 Â 2] grid shown in Fig. 1.
Four Zn(II) cations are held in place in a square grid-like arrangement by four ditopic ligands, each generated by Schiff base condensation of 1 with two equivalents of 2, and each interacting in a tridentate fashion with two metals. The latter ions are found to be in an octahedral environment, coordinated by four nitrogen and two oxygen atoms. As far as the ligand is concerned, it is important to note that tridentate coordination can be attained by different congurations, depending on the degree of ligand deprotonation. In Scheme 2, the possible congurations for the metal binding site are shown. While both congurations A and B lead to a tridentate NNO coordination mode, deprotonation occurs in A. On the other hand, conguration C results in a NN bidentate coordination. On the whole, each ligand can be found in a combination of A-A, A-B, B-B and A/B-C global congurations, either symmetric or not, all of which are chemically feasible and have been observed in metal complexes with similar ligands. 11 In particular, coordinating C-O bond lengths are expected to show a change upon (de) protonation viz., C-O À (A) vs. C]O (B), with the latter being shorter than the former. However, in grid 3, due to high symmetry and possible positional disorder, the relevant C-O distance, but also C-N and N-N distances, are found to be averaged out and thus the exact conguration status of the ligand in the grid-like structure cannot be directly determined by the X-ray diffraction data. Component 1 is expected to be a sodium salt and we were unable to nd and model any metallic counter-ions in the lattice. This may be simply due to the high symmetry and disorder present in the structure, which prevented us from clearly identifying and differentiating between counterions and solvent. However, we notice that if all ligands Scheme 1 The chemical formulae of components 1 and 2 and a schematic representation of grid 3 and the [2]-catenanes 4-6. adopted conguration A-B, with each ligand being monodeprotonated, 3 would be neutral on the whole, with no additional counter-cations needed to balance the charge. Indeed, this would represent a quite simple and appealing hypothesis that we are poised to believe as real, yet unconrmed.
Packing of the [2 Â 2]-grids in 3 can be interpreted as directed by steric factors and by the polarity of the hydrophilic sulfonate groups (Fig. S5, ESI †). [2 Â 2] grids of this kind are quite common in the literature and have also been obtained with ligands of similar chemical structures. 7 It is well known that changing the identity of the metal ions can results in a complete change of coordination geometry and, hence, of the nal complex architecture. 12 As to our system, interestingly, the replacement of Zn(II) ions with Cu(II) ions led to a quite surprising outcome. Indeed, the slow evaporation of a 1 : 2 : 1 mixture of 1, 2 and Cu(OAc) 2 in a 2 : 1 mixture of H 2 O/DMF solution produced good quality single crystals of [2]-catenane 4, whose X-ray determined structure is shown in Fig. 2.
As clearly seen, 13 [2 Â 2] grids are formed also in this case. However, two grids are interlocked with each other to form a [2]catenane superstructure. The two concatenated grids are not symmetrical and one is slightly larger than the other, with the distances between adjacent copper ions in the 8.359-8.665Å and 8.549-9.054Å range for the smaller and larger grid, respectively. The concatenation leads to a rotation of the two grids with respect to each other of approximately 40 . The detailed inspection of its structural elements reveals that not all of the copper ions adopt the same coordination geometry. Indeed, one of the two internal copper ions is found to be tetracoordinated, while the other seven are in octahedral environments. The two ligands that bind this "anomalous" copper ion display a non-symmetrical structure with one of their binding sites having a different conguration (C in Scheme 2), and acting as a NN bidentate ligand. This coordination motif, as well as the hydrogen bonding around the internal Cu(II) atoms, implies that deprotonation is not occurring on those sites of the ligands. Since tetrahedral coordination of Cu(II) with NNbidentate ligands is not common and the crystallization conditions are known to induce reduction of Cu(II), 9 we propose that a Cu(II)-to-Cu(I) transformation is occurring in this case.
The coordination geometry of most of the other Cu(II) atoms shows the characteristic distortion from an ideal octahedron due to the Jahn-Teller effect (Table S7, ESI †). This feature, while conrming the oxidation state of Cu ions as +2, will also become important later.
As far as charge state is concerned, also in this case, as in the previously described Zn(II)-grid 3, we were not able to nd and model any additional counter-cations in the lattice (neither Na + , nor Cu 2+ /Cu + ions), possibly implying the neutrality of the whole structure. If that were the case, the presence of a lower valence state for the Cu(I) ion would be therefore counterbalanced, for example, by one of the eight ligands forming the [2]-catenanes adopting a fully protonated state.
Inter-macrocycle interactions, together with the necessity of efficient packing, are usually responsible for the formation of concatenated architectures. In this case, the two ligand sites that act as NN-bidentate do not make use of their carbonyl groups in the coordination with the metal and two additional inter-grid NH  of [2]-catenane, 14 where inter-macrocyclic interactions between amidic NH and carbonyl groups were rst observed. Notably, the ligand C-type partial conguration is stabilized by the presence of weak intramolecular hydrogen bonds (C 26A -H 26A /O 22A (2.751(10)Å, ang. 126 ), C 7B -H 7B /O 11B (2.796(13)Å, ang. 119 )). Additional structural stabilization is surely provided by the abundant water forming a network of hydrogen bonds, which, however, remain undened due to the less-than-ideal data quality and a large degree of disorder.
The [2]-catenane units pack with direct interactions by means of N-H/O hydrogen bonds, supported by weaker C-H/O hydrogen bonds, with sulfonate groups, forming centrosymmetric pairs. The same can be observed for the other, symmetrically independent catenane species.
To the best of our knowledge, there exists only one example of [2]-catenane systems made by interlocked grids. 8 In that case, a 1 : 1 Co(II)/Co(III) metal distribution within the grid was observed due to the oxidation of Co(II) to Co(III) by air. The ligand employed displayed two different conformations in the grid structure, however, no conguration change was observed, making our system unique.
Intrigued by the unexpected grid concatenation observed in 4 and by the different behaviour of Zn(II) and Cu(II) under these experimental conditions, we wondered if the process could be selective and we tried to crystallize a mixture of 1 and 2 with Cu(II) and Zn(II) salts in a 1 : 2 : 0.5 : 0.5 ratio. Aer several crystallization attempts, we were able to obtain good quality single crystals of two species, 5 and 6 (Scheme 1). They are both heterometallic species and while displaying strong similarities with 4, are still substantially different. Such mixed valence heterometallic [2]-catenane species made of grid units are unprecedented.
With the assignment of metal identity being of exceeding importance in this work, we devoted particular attention to establishing a set of criteria 15 that could be used to properly distinguish Zn(II) from Cu(II) in 5 (Fig. 3). First of all, the internal metal ion that displays a tetrahedral coordination environment was unambiguously assigned as Cu(I). As far as discriminating Zn(II) from Cu(II) ions, the observation of Jahn-Teller effectpresent in d 9 electronic conguration as in Cu(II) ions but not in d 10 Zn(II) ionswas very helpful. Indeed, in 5, 13 four metal ions display such an effect, which leads to the elongation of the metal-ligand bond distances in the apical coordination positions. For example, the mean apical Cu(II)-N(py) and Cu(II)-O(hydr.) distances in the catenane unit of 5 discussed here of 2.26 and 2.32Å, respectively, can be compared to the respective equatorial distances of 2.06 and 2.05Å.
The availability of monometallic Zn(II) grid 3 and the Cu(II)/ Cu(I) [2]-catenane 4 provided a solid ground for the comparison with 5, aimed at the discrimination between copper and zinc centres. Also, an analysis of the CSD for Zn(II) and Cu(II) complexes with similar ligands provided additional conrmation on the validity of the criteria chosen and allowed us to evaluate the effect of the strain due to the [2]-catenane complex formation on the coordination geometries around the metal ions. 16 Indeed, in simple Cu(II) and Zn(II) metal complexes, the M-N(hydr) distance (Scheme 2) seems to be the clearest indicator of the metal atom type, with the average distances for Zn(II) being largerin cases of both A and B type congurationthan those for Cu(II). Furthermore, we considered the N(py)-O(hydr) distance (Scheme 2) as revealing for the presence of Jahn-Teller effect. Indeed, its mean values are larger for Zn complexes than for Cu ones and distributed over a narrower interval (4.087-4.357Å, to be compared with the 3.850-4.539Å interval found in the case of Cu-complexes, see ESI † -CSD search).
In the [2]-catenane systems, strain due to concatenation is surely present and the divergence between the average Cu(II) and Zn(II) coordination environments is less profound. 17 However, it has been always possible to determine the Zn(II)/ Cu(II) quite condently (Table S8, ESI †).
Packing in 5 is essentially very similar to that found in 4, showing interdigitation of the neighbouring catenane units (Fig. S9 in ESI †). However, an additional solvated Zn(II) ion is found in the lattice. The latter ion, present as a DMF/water solvate, balances the overall charge. This feature again conrms the considerable adaptability of the ligand, which can adjust to the environment in terms of global charge distribution.
In any case, given the central importance of determining the metal composition in 5, we sought additional conrmation of the Zn/Cu ratio in 5. Energy-dispersive X-ray spectroscopy (EDS) and atomic absorption spectrophotometry (AAS) analyses made on the same single crystals analysed by X-ray diffraction brought in comforting results. EDS data show a Zn : Cu ratio ¼ 0.63 on average of multiple measurements (see ESI †). This ratio is in perfect agreement with the Zn : Cu ratio provided by the metric analysis on the X-ray diffraction data, also taking into account the presence of an additional Zn(II) ion (with 0.4 occupancy factor, total 6.4 : 10 Zn : Cu equals to 0.64) in the lattice (see ESI †). 18 Moreover, the AAS of Cu-Zn catenane 5 gives 5.87 (AE0.01%) for Cu and 3.63 (AE0.07%) for Zn, giving a ratio of 0.62 AE 0.01, in accordance with the EDS and the X-ray analysis made on 5.
An important aspect that is common to both 4 and 5 pertains to the observed necessity of a change in the coordination geometry of, at least, one the internal metal ions in order to produce the concatenated structures. Steric factors can be involved, however, we must stress that the tetra-coordination exerted on the Cu(I) ion by the C-type conguration ligands frees two carbonyl groups per [2]-catenane unit, and they can then be involved in stabilizing inter-grid interactions. In the case of structures 4 and 5, each grid unit is connected to the concatenated one by two such NH/O]C HB interactions.
From the data above, it might seem that the concatenation process has a slight preference for the incorporation of copper over zinc, however, the attainmentby a slightly modied crystallization procedureand characterization of 6 disproves this hypothesis. Indeed, the crystal structure of 6, shown in Fig. 4, features a [2]-catenane structure having six Zn(II) metals and only two Cu(II) ions. 13 Again, similarly to the situation found in 4 and 5, the catenation leads to a change in the conguration of the ligands involved in the coordination of the internal Cu(II) ions. This strongly suggests that the catenation happens in the presence of at least one internal copper atom and a change in the conguration of the ligands bound to it. In the case of 6, the symmetrical grids require both of the internal Cu(II) ions to be penta-coordinated (Table S9, ESI †), whereas in the structures of 4 and 5, only one tetra-coordinated internal Cu(I) suffices for catenation. Condent assignment of the metal identity was achieved by a careful metric analysis of the data in relation to 4 and 5, 19 and by the comparison with published structures of NNO-NN pentacoordinated Cu(II) complexes (ESI †). 20 A visual summary of the three different environments that the Cu(II/I) metal ions adopt in the structurally similar [2]-catenanes 4-6 is shown in Fig. 5. Hexa-and penta-coordinated Cu(II), and tetra-coordinated Cu(I) are indeed observed.
Twelve Na + counter-cations can be located and quite appropriately modelled in the structure (six for each [2]-catenane adduct, there are two of them in the asymmetric unit), 21 therefore, on average, each grid must be triply negative and the four ligands wholly bear 11 negative charges.
Interestingly, in this case, neither of the non-coordinating carbonyl groups (C 10H ]O 11H , C 10B ]O 11B ), freed by the change of conguration of the ligand, are involved in inter-grid interactions as in 4 and 5, but instead participate in binding to Na + ions mediated by water molecules.
It is important to add here that the crystals of 6 were obtained by a slightly modied crystallization procedure, that is, by ethanol vapour diffusion into a 1 : 1 H 2 O : DMF solution. Since the organic components of the [2]-catenanes are quite adaptable both in terms of their conformation and conguration, the combination of ligand and copper ion gives access to a large congurational and conformational space, and the crystallization of a different structure does not come as a surprise. Clearly, the difference in the experimental procedure (evaporation vs. anti-solvent diffusion) presented a signicant enough perturbation to lead to the formation of a completely different structure in 6, with all the distinguishing features as already highlighted above. As a nal comment, we note that complexes 4-6 were obtained in yields close to 30%, a gure which is far from optimal, and in striking variance with other reported systems, which can be obtained in signicantly higher yields, and sometimes even quantitatively. 5a,b

Conclusion
Structural complexity is sought by chemists to rene their synthetic capabilities and to discover new properties. This work describes three examples of [2]-catenane systems made of interlocked metal-organic grids obtained by multicomponent self-assembly, and characterized in the solid state by single crystal X-ray diffraction. These unique metal complex systems are either mono-(4) or hetero-metallic (5 and 6) and display multivalence copper (4 and 5). To the best of our knowledge, 5 and 6 represent the rst catenated heterometallic grids reported to date. Their apparent heterogeneity should not be misleading. Indeed, all the [2]-catenane systems possess common features. For example, at least one of the internal metal ions is copper, which displays a different coordination geometry with respect to the other external ones, along with a conspicuous conformational change of the organic ligand involved. The role of the Cu(II)/Cu(I) metal pair, especially in terms of its versatility, is here highlighted as a key factor responsible for the formation of such species. We would also like to emphasize that the presented work provides a simple route to obtain octanuclear complexes, some of which are non-symmetric, and with dened metal composition, which would be otherwise unaccessible. This feature stems from the particular environment surrounding the internal metal ions within the framework and from the catenation event. The possibility to lay a given metal ion into a precise position in such a complex structure could have indeed a large impact in the development of functional materials.
Further research is aimed at verifying the degree of control that can be attained over the systems in terms of overall charge distribution in the [2]-catenanes and nal Zn/Cu ratio within the hetero-metallic species. Also, structural modications of components 1 and 2 have been envisaged in order to improve control over metal identities and complex stoichiometry. Extensive crystallization trial experiments with a wider range of d-block metal salts and concentrations are currently ongoing in our laboratories.