Single-Crystal to Single-Crystal Reversible Transformations Induced by Thermal Dehydration in Keggin-Type Polyoxometalates Decorated with Copper ( II )-Picolinate Complexes : The Structure Directing Role of Guanidinium

Three new hybrid inorganic-metalorganic compounds containing Keggin-type polyoxometalates, neutral copper(II)-picolinate complexes and guanidinium cations have been synthesized in bench conditions and characterized by elemental analysis, infrared spectroscopy and single-crystal X-ray diffraction: the isostructural [C(NH2)3]4[{XW12O40}{Cu2(pic)4}]·[Cu2(pic)4(H2O)]2·6H2O [X = Si (1), Ge (3)] and [C(NH2)3]8[{SiW12O40}2{Cu(pic)2}3{Cu2(pic)4(H2O)}2]·8H2O (2). The three compounds show a pronounced two-dimensional character owing to the structure-directing role of guanidinium. In 1 and 3, layers of [{XW12O40}{Cu2(pic)4}]n4n− hybrid POM chains and layers of [Cu2(pic)4(H2O)] complexes and [C(NH2)3]+ cations pack alternately along the z axis. The hydrogen-bonding network established by guanidinium leads to a trihexagonal tiling arrangement of all copper(II)-picolinate species. In contrast, layers of [C(NH2)3]+-linked [{SiW12O40}2{Cu(pic)2}3]n8n− double chains where each Keggin cluster displays a {Cu2(pic)4(H2O)} moiety pointing at the intralamellar space are observed in 2. OPEN ACCESS Inorganics 2015, 3 195 The thermal stability of 1–3 has been studied by thermogravimetric analyses and variable temperature powder X-ray diffraction. Compounds 1 and 3 undergo single-crystal to single-crystal transformations promoted by reversible dehydration processes and the structures of the corresponding anhydrous phases 1a and 3a have been established. Despite the fact that the [Cu2(pic)4(H2O)] dimeric complexes split into [Cu(pic)2] monomers upon dehydration, the packing remains almost unaltered thanks to the preservation of the hydrogen-bonding network established by guanidinium and its associated Kagome-type lattice. Splitting of the dimeric complexes has been correlated with the electron paramagnetic resonance spectra.


Introduction
Over the past several years, the large family of anionic metal-oxygen clusters known as polyoxometalates (POMs) has been thoroughly employed as building blocks to construct a variety of inorganic-organic hybrid compounds [1][2][3][4][5][6][7].The assembly of POMs with transition metal complexes bearing organic ligands (TMCs) is an effective strategy for designing such type of compounds.The POMs may adopt a variety of roles in these types of hybrid systems: (1) charge compensating anions; (2) ligands directly bonded to TMCs; (3) templates inducing the self-assembly of MOFs [8][9][10][11][12][13].The clusters can act as peculiar inorganic ligands able to bind several TMCs through terminal or bridging oxygen atoms [14][15][16], and this often results in assemblies with extended structures.Thus, many high-dimensional POM-based hybrids have been successfully synthesized to date [17][18][19].
A critical factor for the construction of such architectures rests on the choice of appropriate organic ligands.For example, carboxylate derivatives of heterocyclic amines with mixed N,O-donor atoms are likely to afford polymeric structures with high dimensionalities among the vast library of polydentate ligands [20][21][22].One way of better controlling the structure of the hybrid compound is the use directing agents able to form extensive networks of weak intermolecular interactions.A great deal of attention has been paid to the structure-directing role of several organic species and a surprising variety of organically-templated inorganic frameworks are found in the literature [23][24][25].Guanidinium has shown up as an excellent template because it can establish massive hydrogen-bonding networks due to its high molecular symmetry and extremely weak acid character [26].This cation has been successfully applied in POM chemistry not only as a template of high-dimensional frameworks but also as a selective crystallizing agent for minor POM species in mixed solutions [27][28][29].
We have recently reported a series of hybrid compounds based on [XW12O40] 4− Keggin-type anions (X = Si, Ge) and copper(II) complexes of tetradentate bis(aminopyridil) ligands that can reversibly undergo thermal desorption of water via single-crystal to single-crystal (SCSC) transformations with significant modifications in the bonding and coordination geometry around the Cu II centers [30,31].To date, full studies on SCSC transformations are still scarce for POM-based compounds [32][33][34][35][36][37][38] and those involving the temperature as the external stimulus inducing the solid-state phase transition are limited to the low-temperature polymorphs of [Tm2(H2O)14(H6CrMo6O24)][H6CrMo6O24]•16H2O and [C(NH2)3]6[Mo7O24]•H2O, to the monitoring of the dehydration in the H5PV2Mo10O40•36H2O acid and in the porous [Co2(ppca)2(H2O)(V4O12)0.5] (ppca = 4-(pyridin-4-yl)pyridine-2-carboxylic acid) hybrid material [39][40][41][42].This scarcity is certainly remarkable because the study of solid-state phase transitions induced by external stimuli such as the temperature, redox processes, or the interaction with guest molecules is at the forefront of the crystal engineering [43].For example, several reports on SCSC transformations triggered by the removal, incorporation and/or exchange of solvent guest molecules can be found in the literature for related systems like metalorganic framework (MOF) materials [44][45][46][47].These processes are often referred to as dynamic structural changes associated to compounds classified as third generation materials with potential applications in gas storage and separation, chemical sensing or magnetic switching [48,49].
We now intend to explore the thermostructural behavior of other hybrid systems related to our previous Keggin/bis(aminopyridyl) compounds to evaluate the role of the organic component in facilitating such SCSC transitions.We have first focused our studies on N,O-polydentate heterocyclic ligands, which represent a great first choice for the preparation of extended structures [50][51][52], and their concerted action with templating cations like [C(NH2)3] + .Keggin-type anions have been kept as the inorganic building blocks in our systems because (i) these clusters and their numerous derivatives represent the most archetypal class of heteropolyoxometalates [53,54] and (ii) they are widely known to give rise to highly intricate hybrid structures by coordinating a large number of TMCs simultaneously [6,55].
In this work, we report the synthesis, crystal structure, thermal behavior and electron paramagnetic resonance (EPR) spectra of a series of guanidinium-templated compounds based on Keggin-typeanions and copper(II)-picolinate complexes:

Synthesis and Infrared Spectroscopy
Compounds 1-3 were prepared under mild bench conditions from the [C(NH2)3] + -directed self-assembly of [XW12O40] 4− (X = Si, Ge) and [Cu(pic)2] building blocks in acidic aqueous medium (pH 3-3.5) at room temperature.Both types of building blocks were generated in situ from [XW11O39] 8− POM precursors, a copper(II) source and the pic ligand in its acidic form.For X = Si, different POM:Cu:pic ratios were tested and the solid products obtained upon evaporation were characterized preliminarily by IR spectroscopy.For a 1:3:2 ratio, a mixture of blue crystals of a copper(II)-picolinate complex and a white powder corresponding to a guanidinium salt of the plenary [SiW12O40] 4− anion was obtained.Lowering the amount of Cu II ions to a 1:2:2 ratio led to co-crystallization of the complex with compound 1 as the minor fraction.Crystallization of the former was avoided by using a 1:1:2 ratio.Formation of 1 was maximized in these conditions, but crystallization of a small amount of a second crystalline phase (compound 2) was in turn observed.Crystals of 1 are formed before those of 2 and this fact could be explained on the basis of the different POM:Cu ratio in both compounds: 1:6 and 2:7, respectively.The initial POM:Cu ratio in the reaction mixture is 1:1, and hence the compound with the highest Cu II content (1) tends to crystallize first.The amount of the copper(II)-picolinate complex in solution decreases with respect to that of the POM when 1 crystallizes and this in turn favors the formation of a small amount of 2 with the highest POM content to re-equilibrate the POM:Cu ratio.All attempts of improving the synthetic procedure to avoid formation of mixtures were unsuccessful.While 1 is obtained as the major phase in a mixture of crystals with the side-product 2, the isostructural 3 is isolated as a single crystalline phase when [GeW11O39] 8− is used under the same synthetic conditions and no traces of a hypothetical Ge-containing analogue of 2 are observed by powder X-ray diffraction analysis (Figure S1).It is also worth noting that we never obtained any spectroscopic indication of a compound containing copper(II)-monosubstituted [XW11O39Cu(H2O)] 6− species in spite of using monolacunary Keggin-type anions as precursors.These species are known to be metastable in weakly acidic conditions (typically in the pH range 4-6 for heteropolyoxotungstates), and hence slow conversion into the plenary clusters seems reasonable after considering the pH values of our reaction mixtures.Since 1-3 contain plenary Keggin-type anions as the inorganic building block, we also performed a set of reactions using [XW12O40] 4− POMs as the precursors.In all cases, powders containing these clusters and copper(II)-picolinate complexes were obtained according to IR spectroscopy.These powders could not be recrystallized or unequivocally identified as compounds 1-3 on the basis of powder X-ray diffraction.Thus, the kinetically slow [XW11O39] 8− to [XW12O40] 84− conversion appears to be a key factor in isolating our compounds as single crystals suitable for further structural characterization.
The infrared spectra of 1-3 (Figure S2) show the characteristic features of the [α-XW12O40] 4− Keggin-type anion in the region below 1000 cm −1 with bands of strong intensity corresponding to the antisymmetric stretching of the W-Ot and W-Ob-W bonds that appear at 970 and ca.800 cm −1 for X = Si and 966 and 787 cm −1 for X = Ge, respectively.The grafting of the copper(II)-picolinate complexes onto the POM surfaces shift the above signals by 10 cm −1 compared to those of the clusters in the potassium salts and leads also to the appearance of additional peaks in the 760-660 cm −1 range related to the Cu-O and Cu-N stretching among other vibrations.The metalorganic region above 1000 cm −1 is dominated by signals of medium to strong intensity that are observed in the 1160-1684 cm −1 range and associate to C=C and C=N stretching vibrations in the pyridine rings.
The inorganic [XW12O40] 4− building block in all compounds shows the characteristic structure of the α-Keggin isomer consisting of a central XO4 tetrahedron surrounded by four edge-shared W3O13 trimers, all of them linked via corner-sharing in ideal Td symmetry.In the case of 1 and 3, the Keggin anion lies on a center of inversion with the tetrahedral XO4 group disordered over two crystallographic positions, which leads to its observation as a XO8 cube with half-occupied O sites (Figure 1, left).Table S1 displays ranges of W-O and X-O bond lengths compared to those of the DFT-optimized Keggin anion [56].
Figure 1.Connectivity between building blocks in the asymmetric units of 1-3.Color code: W, gray polyhedra; Si/Ge, yellow polyhedra; Cu, blue spheres; N, green spheres; O, red spheres for Opic or OPOM atoms and cyan spheres for terminal aqua ligands; C, black sticks.Symmetry code: (i) -x, -y, -z.

Copper(II)-Picolinate Complexes
The title compounds contain different types of neutral copper(II) complexes with the ligand 2-picolinate.In all of these complexes, the Cu II atom shows axial-type coordination geometry with two trans-related organic ligands forming the basal or equatorial plane (Figure 2).Selected bond lengths compared to those found in the anhydrous phases 1a and 3a are listed in Table 1.

Figure 2.
Copper(II)-picolinate complexes with atom labeling in 1-3 (for the symmetry codes i and v see Table 1).Two different types of dinuclear complexes coexist in the structures of 1 and 3.The metalorganic {Cu2(pic)4} subunit is composed of two centrosymmetrically related {CuA(pic)2} fragments where the CuA atom is involved in a CuN2O2O'2 chromophore with tetragonally elongated octahedral geometry.One of the axial positions in each CuA center is occupied by one of the Opic atoms forming the equatorial CuN2O2 plane of the neighboring fragment, in such a way that a dimeric complex with equatorial-axial Cu2(μ 2 -Opic)2 rhomboid core is formed.A terminal OPOM atom (disordered over two sites, e.g., O1/O1Z in Table 1) occupies the second axial position, and hence the {Cu2(pic)4} subunits link the Keggin clusters in hybrid [{XW12O40}{Cu2(pic)4}] 4− chains with alternate inorganic and metalorganic building blocks.In contrast, the [Cu2(pic)4(H2O)] moiety is formed by one {CuC(pic)2} and one {CuB(pic)2(H2O)} fragments where both Cu II atoms show CuN2O2O' chromophores with distorted square-pyramidal geometry.A water molecule is located at the apical position of CuB, whereas that of CuC is occupied by one of the Opic atoms that are not involved in the basal CuN2O2 plane of CuB.Therefore, the two Cu II centers in this moiety are linked by a single pic bridging ligand acting in μ 2 -κ 2 N,O 1 :κ 1 O 2 coordination mode.
In the case of 2, the structure contains one dinuclear and two crystallographically independent mononuclear complexes where the CuN2O2O'2 coordination environment around all Cu II centers is tetragonally elongated octahedral.In both monomeric subunits, the axial positions of the CuB and CuC atoms are occupied by terminal OPOM atoms.Thus, the {CuB(pic)2} subunits link the Keggin clusters in a one-dimensional assembly of alternate inorganic and metalorganic building blocks, whereas the {CuC(pic)2} subunits act as connectors between pairs of such hybrid chains to lead to the backbone of the [{SiW12O40}2{Cu(pic)2}3{Cu2(pic)4(H2O)}2]n 8n− polymer.The dinuclear {Cu2(pic)4(H2O)} subunit is made of one {CuA(pic)2} and one {CuD(pic)2(H2O)} fragments linked in equatorial-axial fashion by two pic ligands in μ 2 -κ 2 N,O 1 :κ 1 O 2 bridging mode.Thus, each Cu II center shows at axial positions one of the Opic atoms that are not involved in the equatorial plane of the neighboring fragment.The coordination geometry of CuD is completed with one aqua ligand, whereas CuA axially anchors to a terminal OPOM atom, in such a way that the double-chained backbone of the hybrid polymer results decorated with antenna {Cu2(pic)4(H2O)} subunits.

Crystal Packing of Compounds 1 and 3
The crystal packing of 1 and 3 has a pronounced two-dimensional character with alternating hybrid and metalorganic layers stacked along the [001] direction (Figure 3).The hybrid layers consist of the [{XW12O40}{Cu2(pic)4}] 4− chains running along the [110] direction and arranged in parallel fashion in the xy plane.The metalorganic sublattice contains the [Cu2(pic)4(H2O)] dimers, all water molecules of hydration and all guanidinium cations.The interstitial water molecules do not appear to play a significant structural role as they only establish few hydrogen bonds that either connect Keggin clusters from different layers-the [Cu2(pic)4(H2O)] dimers to the clusters or adjacent hybrid chains through the metalorganic subunits (Table S2).In contrast, the two guanidinium cations create an extended and massive network of N-H•••Opic hydrogen bonds with the carboxylate functionalities of the organic ligands.Each cation strongly interacts with the three crystallographically independent {Cu(pic)2} fragments and arrange them in the (11-1) plane to lead to a corrugated double trihexagonal tiling of Cu II atoms (Figure 4).The Keggin clusters are nested in the hexagonal motifs of this distorted Kagome-type double lattice, whereas the structure-directing guanidinium cations reside in the triangular cavities.

Crystal Packing of Compound 2
The crystal packing of 2 shows also a two-dimensional character with hybrid layers parallel to xy plane (Figure 5).These layers are formed by a double sheet of Keggin clusters arranged in two levels of z and held together by the monomeric {CuB(pic)2} and {CuC(pic)2} subunits.The former link clusters lying in the same z level to lead to a one-dimensional assembly of alternate inorganic and metalorganic building blocks parallel to the [010] direction.The connectivity between building blocks is such that the pyridinic ring of one of the picolinate ligands is sandwiched between the tetrameric {W4O18} faces of adjacent Keggin clusters with distances between the ring centroid and the average plane of the tetramers of 2.755 and 2.856 Å.These distances are comparable to those observed in related compounds with similar POM-aromatic interactions [30,31].The {CuC(pic)2} subunits connect in turn centrosymmetrically related clusters located at different z levels through long Cu-OPOM bonds typical of semi-coordination (Table 1).In this case, the two aromatic rings interact with Keggin anions as they place almost parallel to tetrameric faces of the contiguous clusters with a centroid-tetramer plane distance of 2.781 Å.The linkage through the {CuC(pic)2} subunits of pairs of one-dimensional hybrid assemblies running along the crystallographic b axis at different z levels results in double-chained [{SiW12O40}2{Cu(pic)2}3]n 8n− anions with rectangular cavities in the polymeric backbone and where each Keggin cluster is additionally decorated with a {Cu2(pic)4(H2O)} dimer grafted as antenna subunit.Two of the guanidinium cations (C1G and C2G) are hosted in the rectangular cavities and they establish multiple N-H•••O interactions with both the Keggin surfaces and the carboxylate groups of the picolinate ligands (Figure S3, N•••O distances in the range 2.33(3)-3.28(3)Å).The layers pack with the antenna subunits directed to the interlamellar space to give rise to an alternate sequence of hybrid and metalorganic regions along the [001] direction.The antenna complexes in the metalorganic region are arranged in such a way that all picolinate ligands are almost parallel to the (10-2) plane and dimers grafted at contiguous POM sheets are hydrogen bonded by the water molecules of hydration and the guanidinium cations C3G and C4G.

Thermostructural Behavior
The thermostructural behavior of the title compounds was investigated by a combination of thermal analyses and variable temperature X-ray diffraction.Thermal analyses show that all compounds decompose via three mass loss stages (Figures 6 and S4).The first stage starts at room temperature and it is associated with two endothermic processes that originate from the release of the water molecules.For 1 and 3, the dehydration stage extends up to ca. 130 °C and comprises the loss of a 2.24% of the total mass, which accounts for only 6 out of the 8 water molecules determined by single-crystal X-ray diffraction (calcd.for 6H2O: 1, 2.13%; 3, 2.11%).Analogously, dehydration of 2 is completed at ca. 95 °C with the release of eight out of the ten water molecules determined crystallographically [calcd.(found) for 8H2O: 1.69% (1.71)].It is likely that these compounds lose some weakly bound interstitial water molecules when crystals are removed from their mother liquors and filtered at room temperature prior to be analyzed thermogravimetrically.The resulting anhydrous phases all show a wide range of thermal stability, up to ca. 300 °C for 1 and 3 and to 280 °C in the case of 2. Above these temperatures, the anhydrous derivatives undergo further decomposition via two highly overlapping mass loss stages.The former originates from the combination of two endothermic and one exothermic consecutive processes that can be related to the release of [C(NH2)3] + cations as guanidine molecules and to the combustion of part of the picolinate ligands, respectively.For 1 and 3, this stage extends up to ca. 355 °C and involves the loss of a 18.22 and 18.74% of the respective total mass, which roughly corresponds to 4 cations and 6 picolinate ligands (calcd.for 4(CH6N3) + 6(C6H4NO2): 1, 4.74 + 14.45 = 19.19%;3, 4.70 + 14.33 = 19.03%).In the case of 2, the upper temperature limit and the mass loss are ca.380 °C and 12.10%, which roughly accounts for 4 ligands besides 8 cations (calcd.for 8(CH6N3) + 4(C6H4NO2): 5.64 + 5.73 = 11.37%).The final mass loss stage originates from a complex combination of exothermic processes that must associate with the combustion of the remaining organic matter and the crumbling of the Keggin framework.The final residues are obtained at temperatures in the 530-570 °C range and have been identified as mixtures of monoclinic WO3 (PDF 88-269) [59] and triclinic CuWO4 (PDF 43-1035) with Scheelite-type structure [60] according to powder X-ray diffraction (calcd.(found) for aCuWO4 + bWO3 + cXO2: 1, 65.5% (65.6), a = b = 6, c = 1; 2, 73.2% (76.1), a = 7, b = 17, c = 2; 3, 65.8% (65.9), a = b = 6, c = 1).
Variable temperature powder X-ray diffraction reveals that the title compounds retain crystallinity within the range of thermal stability upon dehydration (Figures 6 and S5).For 1 and 3, well-defined diffraction patterns are obtained up to 310 °C, which is in full agreement with the upper temperature limit of the stability range in the TGA curves.The diffraction pattern is preserved with negligible variations in the positions and intensities of the diffraction maxima for the resulting anhydrous phases (1a and 3a), and this fact indicates that dehydration does not result in drastic structural changes.Compound 2 also maintains crystallinity upon dehydration, but in contrast to 1 and 3, a phase transformation is unequivocally observed between 50 °C and 70 °C.All compounds become amorphous solids in the temperature range corresponding to the release of the guanidinium cations and the combustion of the picolinate ligands.New crystalline phases corresponding to the final residue originated upon breakdown of the Keggin framework start appearing at 510 °C and they reach complete formation at temperatures slightly beyond the end of the third mass loss stage in the TGA curves (ca.590 °C).Analogous single-crystal X-ray diffraction studies were also carried out.Crystals of the title compounds were mounted at room temperature on a diffractometer and the temperature was raised at a rate of 1 °C min −1 to 140 °C for 1 and 3 and to 100 °C for 2. The crystal of 1 preserved its integrity and crystallinity in the whole temperature range and darkening of its blue color was observed upon heating (Figure 7).This crystal stability allowed us to perform unit cell determinations at room temperature, 50, 80 and 140 °C (Table 2).In contrast, crystals of 3 cracked almost immediately after the temperature was ramped, but we could manually separate one of the resulting pieces to perform the experiment.Diffraction was of lower quality and much weaker than that observed for the isostructural 1.Thus, the unit cell parameters determined for 3 are significantly less accurate than those of 1, but nevertheless, they reproduce analogous trends acceptably.In the case of 2, the laminar crystal also cracked when the temperature was ramped, but unfortunately, we could not apply the strategy followed for 3 because the extreme fragility of the resulting pieces prevented us from their manipulation.As shown in Table 2, a significant shortening of the parameter c and consequent contraction of the unit cell volume is observed for both 1 and 3 when going from 80 to 140 °C, which indicates formation of the corresponding anhydrous derivatives.At this point, we lowered the temperature to 100(2) K to carry out the full data collections for both compounds and the structures of 1a and 3a were determined.Simple TGA/DTA experiments were performed to determine the reversibility of the dehydration processes (Figure S6).Crystalline samples of 1 and 3 were heated at a rate of 2 °C min −1 up to 200 °C and the so-generated anhydrous samples were exposed to the room atmosphere for one day, and then, heated again at the same rate.The recorded TGA profiles are almost identical for both heating cycles, and this fact shows that the anhydrous phases 1a and 3a are fully rehydrated to the original compounds simply after being in contact with moisture for a few hours.These observations were confirmed by single-crystal X-ray diffraction.The crystals used for determining the structures of 1a and 3a were kept on the goniometer head in contact with the room environment and the intensity data were collected back at 100(2) K after a few days.The crystals still diffracted acceptably enough and the unit cells of the initial hydrated phases 1 and 3 were again obtained.The structural solutions were of poorer quality than those determined originally most likely due to disorder affecting the water molecules upon resorption (note the differences in the DTA profiles in Figure S6), but nevertheless, we could locate the inorganic and metalorganic building blocks in their original positions.These observations demonstrate that dehydration of both 1 and 3 proceeds via SCSC transformations, but furthermore, that this process is reversible and the anhydrous 1a and 3a phases also undergo SCSC transformations promoted by consequent rehydration.

SCSC Transformations of Compounds 1 and 3 into the Anhydrous Phases 1a and 3a
Dehydration of compounds 1 and 3 into the phases 1a and 3a does not equally affect the hybrid layers and the metalorganic regions.The hybrid layers remain virtually unaltered: for example, variations in the bond lengths within the {CuA2(pic)4} subunit are negligible (Table 1) and the relative arrangement of Keggin anions in the hybrid [{XW12O40}{Cu(pic)2}2] 4− chains is only affected by a subtle lengthening of 0.1 Å in the X•••X distance between adjacent clusters.In contrast, significant changes take place in the metalorganic sublattice because all water molecules of coordination and hydration reside in this area (Figure 8).Their removal promotes a reorganization of the dimeric [Cu2(pic)4(H2O)] moieties, each of which split into two independent [Cu(pic)2] square-planar complexes as evidenced by the remarkable lengthening of the CuB-O4C distance from ca. 2.49 to 3.31 Å.The contraction of the parameter c by 1 Å is also consequence of this splitting.While the original metalorganic regions consist in a corrugated lattice of {Cu(pic)2} fragments in the crystallographic xy plane, the release of the water molecules force the newly generated [Cu(pic)2] monomers to spread on the plane in such a way that the corrugation degree decreases and the Cu II atoms become nearly coplanar.In spite of this rearrangement of complexes, the structure-directing network of N-H•••O hydrogen bonds remains almost intact upon dehydration because of a slight reorientation of the guanidinium cations that preserves almost all contacts (Table S2).Thus, the double trihexagonal tiling described above is maintained without noticeable alterations (Figure 9).

Electron Paramagnetic Resonance Spectroscopy for Compounds 1 and 1a
The EPR spectra of 1 (Figure 10) and 3 (Figure S7) are virtually identical in good agreement with the isostructurality of the compounds.The only difference worth to be mentioned is the largest line width observed for 3, which is likely due to the lower crystallinity of this compound when compared to 1.The spectra are relatively complex as a result of the overlapping of the contributions from the different copper(II)-picolinate dimeric entities coexisting in the crystal packing.The X-band spectra show multiple resonances in the range 2300-3800 Gauss and a half-field signal corresponding to the ΔMS = ±2 forbidden transition centered at ca. 1600 Gauss, which indicates the presence of a magnetically isolated triplet state (S = 1).Moreover, a partially resolved hyperfine structure originating from the interaction of an electron spin with a limited number of non-zero nuclear spins is also observed in both spectra.The number of detectable lines in this hyperfine structure is above the 4 lines that would correspond to a spin doublet interacting with a single I = 3/2 nucleus and this confirms the presence of an isolated S = 1 state.Both the X-and Q-band spectra display at least one signal for which the apparent g value is substantially lower than that of the free electron (3600 and 12200 Gauss, respectively).As all Cu II atoms in 1 and 3 are in octahedral or square-pyramidal coordination environments, the presence of such signals can only be attributed to a noticeable zero-field splitting (ZFS) within a multiplet state.The observation of such multiplet states can be well correlated with the coexistence of the dinuclear entities {Cu2(pic)4} and [Cu2(pic)4(H2O)].In both complexes, intradimeric magnetic exchange takes place through axial-equatorial pathways: µ 2 -O bridging atoms for {Cu2(pic)4} and O-C-O linkages in the case of [Cu2(pic)4(H2O)].The spectroscopic features suggest that the former are magnetically isolated by the bulky diamagnetic Keggin clusters and give rise to a common signal with the typical features of a triplet state with significant zero-field splitting, whereas the latter are coupled in an extended system of long-range, weak magnetic interactions that average their individual signals.Thus, the spectra were initially simulated as the sum of the following individual contributions: one signal of axial symmetry corresponding to a cooperative exchange g tensor (signal 1) and another signal originating from an S = 1 spin state with collinear D and g tensors of axial symmetry (Figure S8).The fit of signal 1 to the experimental profile was improved by breaking the axial symmetry with some equatorial anisotropy, whereas that of signal 2 required the introduction of certain ZFS also in the equatorial plane.The value of the parameter E used during the fitting should be considered only as a simple approximation or as a maximum value for the equatorial ZFS effect.
The calculated spin Hamiltonian parameters are shown in Table 3, together with those of the isostructural 3 and the anhydrous derivative 1a.The g values are consistent with those expected for the topology shown by the copper(II) chromophores in the title compounds and confirms that the ground state is mainly of d(x 2 -y 2 ) type.Therefore, the intradimeric coupling interactions must be very small in both cases considering the axial-equatorial pathways.The calculated hyperfine coupling constant (87×10 −4 cm −1 ) is actually half of that expected for a Cu II chromophore with g = 2.25 and N2O2-type equatorial plane, which implies that each unpaired electron must interact with the nuclear spins of two different Cu II ions as corresponds to the presence of magnetically isolated dinuclear entities.The D parameter obtained is relatively small and this fact is in good agreement with the strong deviation of the axial component of the g tensors with respect to the intradimeric Cu-Cu axis, which reduces the anisotropic exchange contribution.3) 470( 5) -Dehydration of 1 into 1a does not significantly affect the EPR spectra in spite of the fact that the release of the water molecules results in the splitting of one of the dimeric entities into independent monomers with consequent modification of the coordination geometry around the Cu II centers from square-pyramidal to square-planar.The two main contributions corresponding to the isolated dimer and the extended system can still be well appreciated in the spectra of 1a.For the former, the lines of the fine structure approach each other due to a decrease of the D parameter that may associate with a reduction of the anisotropy around the CuA centers.All of the fine and hyperfine lines become wider, resulting in a loss of resolution that might a priori be related to an increase of dipolar interactions or to a reduction of the exchange coupling.However, we believe that this phenomenon is simply due to a loss of crystallinity in the solid sample upon heating the starting material in an oven.It is worth highlighting that EPR spectroscopy finely demonstrates that rehydration of 1a into the initial 1 is fully achieved in a very short time under standard atmospheric conditions.Using a freshly prepared sample of 1a, the original experimental profiles of 1 were recovered within a few minutes when an open sample holder was used, but fortunately, sealing the latter slowed the process down enough for allowing us to collect the spectra of the anhydrous derivative.
Figure 11 displays the X-band (9.40 GHz) and Q-band (34.10 GHz) EPR spectra of 2 recorded at room temperature on a grained polycrystalline sample.The spectra are closely related to those of 1 and 3 as contributions of two different magnetic systems are also observed: a magnetically isolated S = 1 state with significant zero-field splitting and a poorly resolved hyperfine structure in the parallel region and a more intense rhombic signal that must originate from a cooperative exchange g tensor after considering its lack of hyperfine lines.The calculated g, A and D values are similar to those determined for 1 and 3 (Table 3).Taking into account the structural features of 2, it should be assumed that the signal of the isolated triplet state corresponds to the dinuclear {Cu2(pic)4(H2O)} antenna subunits, and hence the rhombic signal must then be ascribed to the presence of long-range, weak magnetic interactions involving the monomeric {Cu(pic)2} bridging subunits.Therefore, magnetic

Materials and Methods
The precursors K8[α-SiW11O39]•13H2O and K8[α-GeW11O39]•13H2O were prepared according to the literature [61,62] and identified by IR spectroscopy.All other chemicals were obtained from commercial sources and used without further purification.Carbon, hydrogen, and nitrogen were determined on a Perkin-Elmer 2400 CHN analyzer (Perkin Elmer, Waltham, MA, USA).Infrared spectra for solid samples were obtained as KBr pellets on a SHIMADZU FTIR-8400S spectrometer (Shimadzu, Kyoto, Japan).Thermogravimetric and Differential Thermal Analyses were carried out from room temperature to 750 °C at a rate of 5 °C min −1 on a TA Instruments 2960 SDT thermobalance (TA Instruments, New Castle, DE, USA) under a 100 cm 3 •min −1 flow of synthetic air.Electron Paramagnetic Resonance (EPR) spectra were recorded on Bruker ELEXSYS 500 (superhigh-Q resonator ER-4123-SHQ, (Bruker, Karlsruhe, Germany) and Bruker EMX (ER-510-QT resonator, Bruker, Karlsruhe, Germany) continuous wave spectrometers for Q-and X-bands, respectively.

X-ray Crystallography
Crystallographic data for 1-3 and the anhydrous phases 1a and 3a are given in Table 4. Intensity data were collected at 100(2) K on an Agilent Technologies SuperNova diffractometer (Santa Clara, CA, USA) equipped with an Oxford Cryostream 700 PLUS temperature device (Oxford, UK).Mirror-monochromated Mo Kα radiation (λ = 0.71073 Å) and an Eos CCD detector (Santa Clara, CA, USA) was used in all cases with the exception of 1 and 3a, for which data collection involved mirror-monochromated Cu Kα radiation (λ = 1.5418Å) and an Atlas CCD detector (Santa Clara, CA, USA).In the case of the anhydrous phases 1a and 3a, a single crystal of the corresponding hydrated compound was mounted on the goniometer and a preliminary data collection was performed at room temperature to check that its diffraction was of sufficient quality.The temperature was then ramped at a rate of 2 K min −1 and unit cell measurements were carried out at 323, 353 and 413(2) K to ensure whether the sample maintained its integrity as a single crystal during the structural transformation associated to dehydration.Once the temperature reached 413(2) K, it was lowered to 100(2) K at a rate of 6 K min −1 for performing a full data collection of the so generated anhydrous phases.The crystals were kept on the goniometer head and exposed to room atmosphere for several days, after which routine full data collections corresponding to the initial hydrated forms were carried out at 100(2) K.  Data frames were processed (unit cell determination, intensity data integration, correction for Lorentz and polarization effects, and analytical absorption correction with face indexing) using the CrysAlis Pro software package (Agilent Technologies UK Ltd., Oxford, UK) [63].The structures were solved using OLEX (OlexSys Ltd in Durham University, Durham, UK) [64] and refined by full-matrix least-squares with SHELXL-97 (University of Goettingen, Goettingen, Germany) [65].Final geometrical calculations were carried out with PLATON (Utrecht University, Utrecht, The Netherlands) [66] as integrated in WinGX (University of Glasgow, Glasgow, UK) [67].Thermal vibrations were treated anisotropically for heavy atoms (W, Cu, Si).Hydrogen atoms of the organic ligands were placed in calculated positions and refined using a riding model with standard SHELXL parameters.In all cases, the Keggin clusters displayed disorder originated from slight tilting in the crystal packing.This tilting was modeled by disordering the O atoms of the Keggin clusters over two positions labeled as O/OZ.The population factor within the O/OZ pairs was initially refined as a single free variable, resulting in the following occupancies: 50/50 for 1, 48/52 for 1a, 65/35 for 2, 51/49 for 3 and 52/48 for 3a.CCDC-1058791 (1), -1058793 (2), -1058794 (3), -1058792 (1a), and -1058795 (3a) contain the supplementary crystallographic data for this paper.These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
Powder X-ray diffraction patterns were collected on a Bruker D8 Advance diffractometer (Karlsruhe, Germany) operating at 30 kV and 20 mA and equipped with a Pt sample holder, Cu tube (λ = 1.5418Å), Vantec-1 PSD detector (Karlsruhe, Germany), and Anton Parr HTK2000 high-temperature furnace (Graz, Austria).The patterns were recorded in 2θ steps of 0.033° in the 5 ≤ 2θ ≤ 39° range using an exposure time of 0.3 s per step.Full data sets were recorded from 30 to 770 °C every 20 °C and a heating rate of 0.16 °C s −1 was applied between the temperatures.

Conclusions
The study presented herein represents a good indication of the fact that single-crystal to single-crystal transformations might be a common structural response to thermal dehydration in a wide scope of hybrid compounds composed of polyoxometalate anions and transition metal complexes bearing organic ligands.To date, such types of solid-state phase transition studies have only been developed for polyoxometalate-based hybrid compounds containing bis(aminopyridyl)-type ligands.In this work, we demonstrate that analogous behavior can also be found in related systems with completely different metalorganic subunits such as transition metal bis(picolinate) complexes.The aiding role of guanidinium cations as structure-directing agents appears to be a key factor in facilitating the crystal transformations because they are able to establish a massive network of intermolecular interactions that remains nearly unaltered upon dehydration.

Figure 3 .
Figure 3.View of the crystal packing of 1 and 3 along the crystallographic a axis with details of the arrangement of the [{XW12O40}{Cu(pic)2}2] 4− chains in the hybrid layers and the [Cu2(pic)4(H2O)] units and [C(NH2)3] + cations in the metalorganic regions.

Figure 7 .
Figure 7. Photographs of single crystals of 1 taken at room temperature (left) and upon dehydration at 140 °C (right).Insets: images of the crystals used for performing the full single-crystal X-ray diffraction data collections of compouunds 1 and 1a.

Figure 9 .
Figure 9. Guanidinium-templated Kagome-type double lattice of Cu II atoms in 1 and 3 compared to that found in the anhydrous phases 1a and 3a.
exchange pathways between the CuB and CuC ions must exist through the Keggin clusters, or most likely, through the strong N-H•••O hydrogen bonds established with the guanidiniums cations.

Table 2 .
Unit cell parameters of 1

Table 4 .
Crystallographic data for 1-3 and for the anhydrous phases 1a and 3a.