2.1 Crystal structures of [M^IV{(μ-CN)₄Mn^II(H₂O)₂}₂·4H₂O]_n (M^IV = Nb^IV (1); Mo^IV (2) and W^IV (3))

The three bimetallic, three-dimensional networks 1–3 are isostructural and crystallize in the tetragonal space group I4/m. Crystallographic data and selected bond distances and angles for 1–3 are given in Tables 1 and 2, respectively. The M^IV ions and Mn^II ions are alternately linked through μ-cyanido ligands in a 1:2 ratio. The Nb^IV and Mn^II coordination spheres of 1 are shown as ORTEP drawing in Fig. 1. The [001] projection of the crystal structure of 1 is shown in Fig. 2. The M^IV ion is surrounded by eight carbon atoms from μ-cyanido ligands in approximately square antiprismatic coordination. A fourfold rotation axis is passing through the M^IV center. The coordination around the Mn^II ion is a slightly distorted octahedron: four nitrogen atoms from μ-cyanido ligands bind in equatorial positions whereas the axial positions are occupied by oxygen atoms from two water molecules. A mirror plane is passing through the Mn^II ion and the two oxygen atoms. The average bond distances (in Å, (1)/(2)/(3)) Mn–N (2.222/2.203/2.22) and Mn–O (2.186/2.16/2.20) are similar in the three compounds. Similar conclusions can be drawn for C–N distances (in Å, (1)/(2)/(3), 1.137/1.13/1.14) in [M^IV(CN)₈]. The only significant difference between 1 and 3 is related to the M–C bond distances (in Å), Nb–C (2.248), Mo–C (2.157), W–C (2.17). The Nb–C distances are slightly longer than those of Mo–C or W–C. Such trend was previously observed [7d–e]. The bond angles M–C1–N1 (in °, (1)/(2)/(3), 176.1/176.5/175.5) and M–C2–N2 (177.1/176.8/179.1) deviate only slightly from linearity, whereas the bond angles Mn–N1–C1 (166.6/166.2/164.6) and Mn–N2–C2 (154.1/155.0/153.1) are much less than 180°, as found in other μ-cyanido-bridged compounds [3–6]. The shortest distance between the M^IV and Mn^II ions across the μ-cyanido bridge is 5.57 Å in 1, 5.49 Å in 2 and 5.51 Å in 3. These distances and angles are similar to those found in other [M^IV(CN)₈]/M^II compounds [5,6]. We would like to point out that our crystallographic results are a very clean and simple demonstration of the structural trends usually observed in the periodic table: (i) contraction of the atomic radii when increasing Z in the same line [from Nb to Mo, in Å: Nb–C (2.248), Mo–C (2.157)] thanks to the increase of the effective atomic number Z_eff (due to a reduced screening of the valence electrons of the same valence shell); (ii) moderate increase of the atomic radii in the same column when going to 4d to 5d elements [from Mo to W, in Å: Mo–C (2.157), W–C (2.17)] because of the “lanthanide contraction” due to the filling of the f orbitals before the 5d ones. These data are peculiarly reliable since for the three elements (Nb, Mo, W), the oxidation state (IV), the coordination number (8) and the symmetry of the coordination sphere (square antiprism) are exactly the same and make easy the comparison.

2.2 Magnetic properties

Variable-temperature magnetization data, collected for 1 with field-cooled (FCM), zero-field-cooled (ZFCM) and remnant (RM) measuring protocols, are shown in Fig. 3. The FCM shows a rapid increase when the temperature is lowered below 50 K, corresponding to the onset of a magnetically ordered phase. When the field is switched off at 5 K, a remnant magnetization (RM) is observed which vanished as the temperature is increased up to 50 K. The ZFCM curve merges with the FCM one at T_C ≈ 47 K. The field dependence of the magnetization at T = 2 K for 1 and 2 is presented in Fig. 4. The magnetization curve shows a very high zero-field susceptibility and reaches a saturation value at 8.9 Bohr magnetrons per NbMn₂ formula unit, close to the value of 9 (g = 2.0), expected for antiferromagnetically coupled Nb^IV and Mn^II ions. The magnetization data demonstrate a three-dimensional ferrimagnetic ordering in 1.

Instead, compounds 2 and 3 present no magnetic order. The magnetization of 2, in Fig. 4, in contrast to the high zero-field susceptibility shown by compound 1, shows the expected behaviour for practically uncoupled Mn^II ions, approaching 9.7 Bohr magnetrons per formula unit, at 50 kOe, close to the expected value of 10 (g = 2). Furthermore, the variable-temperature magnetic susceptibility data for 2–3 are very similar and consistent with two uncoupled Mn^II ions as shown in Fig. 5. At T > 50 K, χ_MT is constant with a 8.8 cm³ K mol⁻¹ value, close to the χ_MT value of 8.75 cm³ K mol⁻¹ calculated for two uncoupled Mn^II (S = 5/2, g = 2). Below 50 K, the χ_MT decrease points out the existence of a weak antiferromagnetic exchange interactions between next nearest Mn^II neighbours. The Mn^II–NC–M–CN–Mn^II pathway is rather long since the Mn–M–Mn distances are around 10.9 Å but the overlap between Mn^II magnetic orbitals is not strictly zero.

The antiferromagnetic interaction Nb^IV–Mn^II is simply explained by a “trough-bond” interaction, i.e. the overlap between the magnetic orbital centered on the 4dz² orbital of niobium and some of the 3d orbitals of manganese [1b,7f]. The magnetic orbital (singly occupied molecular orbital) centered on the dz² orbital of Nb(IV) is partially delocalized on the carbon and nitrogen (π symmetry) of the cyanide (Scheme 1a). Among the five magnetic orbitals centered on the d orbitals of manganese, at least one of them (xz for example, t_2g type) is also delocalized on the nitrogen and the carbon of the cyanide (with the same π symmetry) (Scheme 1a). These two basis orbitals overlap to give two molecular orbitals in the complex Nb–Mn. For two electrons, the Kahn's model [J = 2 k + 4βS, where J is the coupling constant, energy difference between the singlet and the triplet, k is the exchange integral (>0), β is the resonance integral (<0), and S is the overlap integral] [9] allows to foresee an antiferromagnetic interaction (2k << |4βS|). Furthermore, for two metals located at ≈5.5 Å the ferromagnetic terms arising from the orthogonality of the dz² niobium orbital with some of the manganese orbitals are weak and can be neglected.

The antiferromagnetic coupling between nearest neighbour spins with different values (Nb^IV, S = 1/2; Mn^II, S = 5/2) leads then to ferrimagnetism (↓↑) and to the observed long range ferrimagnetic order.

3.1 Materials

All solid reactants were reagent grade, purchased from Acros or Aldrich and used as-received. Absolute methanol was obtained over magnesium before use.

3.2 Syntheses

The precursors K₄[M(CN)₈]·2H₂O (M = Mo, W) were made according to previously published procedures [10]. The precursor K₄[Nb(CN)₈]·2H₂O was prepared using a procedure slightly modified compared to the literature [11]. In 25 mL of methanol was dissolved 20 g of NbCl₅ under inert argon atmosphere using Schlenk technique. After the vigorous HCl evolution stopped, the light-yellow solution was diluted with an additional 50 mL of methanol. The resulting solution was electrochemically reduced on carbon electrodes for 24 h at a constant voltage of 35 V. After reduction, the green-brown solution was slowly added to a solution of 100 g KCN dissolved in 150 mL H₂O, cooled in an ice/NaCl bath. The mixture was stirred for 1 h under cooling and for additional 24 h at ambient temperature. Brick red K₅[Nb^III(CN)₈] precipitated by this time. After centrifugation, K₅[Nb^III(CN)₈] was added under stirring to 100 mL of H₂O and then 15 mL of a 10% aqueous solution of hydrogen peroxide were slowly added. Yellow K₄[Nb^IV(CN)₈]·2H₂O precipitated by addition of methanol. The product was recrystallized from water/methanol. Yield ∼25%.

[Nb^IV{(μ-CN)₄Mn^II(H₂O)₂}₂·4H₂O]_n (1) was prepared by the addition, in the dark, of 0.4 mmol (197.2 mg) of K₄[Nb(CN)₈]·2H₂O in 15 mL of H₂O to 1.0 mmol (0.198 g) of MnCl₂·4H₂O in 15 mL of H₂O (without stirring). The resultant yellow suspension was kept at 4 °C. After a few days, dark-red polyhedral crystals formed. The crystals were washed with a few small portions of distilled water and ethanol and dried in air. Calcd for 1: C, 17.3; H, 2.9; N, 20.2. Found: C, 17.2; H, 3.1; N, 20.1.

Mo^IV{(μ-CN)₄Mn^II(H₂O)₂}₂·4H₂O]_n (2) and [W^IV{(μ-CN)₄Mn^II(H₂O)₂}₂·4H₂O]_n (3) were prepared by reacting aqueous solutions, in the dark, of MnCl₂·4H₂O and K₄[Mo(CN)₈]·2H₂O for 2 or K₄[W(CN)₈]·2H₂O for 3 by slow diffusion in an H-tube. After a few weeks, yellow-orange polyhedral crystals formed. The crystals were washed with a few small portions of distilled water and ethanol and dried in air. Calcd for 2: C, 17.2; H, 2.9; N, 20.1. Found: C, 17.1; H, 3.0; N, 20.0. Calcd for 3: C, 14.9; H, 2.5; N, 17.3. Found: C, 14.8; H, 2.9; N, 17.3.

3.3 Crystallographic data collection and structure determination

The determination of the unit cell parameters and the collection of intensity data for 1, 2 and 3 were made at 295 K on a Enraf Nonius CAD4 diffractometer using crystals of dimensions 0.2 × 0.2 × 0.4 mm³ (1), 0.4 × 0.4 × 0.35 mm³ (2) and 0.2 × 0.2 × 0.4 mm³ (3). Of the 902 (1), 1183 (2) and 902 (3) unique reflections, 838 (1), 856 (2) and 720 (3) with I > 3σ(I) were used to solve the structure with SHELXS-86 program [12]. Computations were performed by using PC versions of CRYSTALS [13]. Least-squares refinements of 68 (for 1), 67 (for 2) and 67 (for 3) parameters with approximation to the normal matrix were carried out by minimizing the function Σw(|F_o| − |F_c|)², where F_o and F_c are the observed and calculated structure factors. Corrections were made for Lorentz and polarization effects. Empirical absorption corrections based on Ψ scan curves were applied. The models reached convergence with R = 0.0516 and R_w = 0.050 for 1, R = 0.081 and R_w = 0.062 for 2 and R = 0.0564 and R_w = 0.0487 for 3. The hydrogen atoms were not included in the refinement. All non-hydrogen atoms were refined anisotropically. The crystallographic data for 1, 2 and 3 are summarized in Table 1. Main bond lengths and angles are listed in Table 2.

3.4 Magnetic measurements

Magnetic susceptibility data of polycrystalline samples of 1, 2 and 3 were collected with an MPMS Quantum Design SQUID magnetometer (XL5S) in the temperature range 300–302 K. Pascal's constants were used for the diamagnetic corrections [14].