Effect of Heteroatoms on Field-Induced Slow Magnetic Relaxation of Mononuclear FeIII ( S = 5/2) Ions within Polyoxometalates.

In this paper, the synthesis and magnetic properties of mononuclear FeIII-containing polyoxometalates (POMs) with different types of heteroatoms, TBA7H10[(A-α-XW9O34)2Fe] (IIX, X = Ge, Si; TBA = tetra- n-butylammonium), are reported. In these POMs, mononuclear highly distorted six-coordinate octahedral [FeO6]9- units are sandwiched by two trivacant lacunary units [A-α-XW9O34]10- (X = Ge, Si). These POMs exhibit field-induced slow magnetic relaxation based on the single high-spin FeIII magnetic center ( S = 5/2). Combining experiment and ab initio calculations, we investigated the effect of heteroatoms of the lacunary units on the field-induced slow magnetic relaxation of these POMs. By changing the heteroatoms from Si (IISi) to Ge (IIGe), the coordination geometry around the FeIII ion is mildly changed. Concretely, the axial Fe-O bond length in IIGe is shortened compared with that in IISi, and consequently the distortion of the [FeO6]9- unit in IIGe from the ideal octahedral coordination geometry becomes larger than that in IISi. The effective demagnetization barrier of IIGe (11.4 K) is slightly larger than that of IISi (9.2 K). Multireference ab initio calculations predict zero-field splitting parameters in good agreement with experiment. Although the differences in the coordination geometries and magnetic properties of IIGe and IISi are quite small, ab initio calculations indicate subtle changes in the magnetic anisotropy which are in line with the observed magnetic relaxation properties.


INTRODUCTION
Polyoxometalates (POMs) are a class of anionic metal oxide clusters that exhibit large structural diversity, and their properties can be controlled by selecting structures, constituent elements, and charges.1 Various types of metal cations can be introduced into the structurally well-defined vacant sites of lacunary POMs, where precise structural design of metal oxo clusters is possible; i.e., numbers, arrangements, and coordination geometries of metal cations can be designed within the vacant sites. Therefore, lacunary POMs are useful ligands for the rational investigation of catalytic, redox, and magnetic properties of metal oxo clusters.2 In particular, since the coordination geometries of metal cations directly affect their electron configurations, rigid multidentate lacunary POM ligands are useful to design and control their magnetic properties. 3 We have recently reported that various mononuclear and multinuclear metal oxo clusters can be selectively synthesized using lacunary POMs in organic solvents.4,5 In particular, a mononuclear FeIII ion possessing an unusually distorted sixcoordinate octahedral geometry within a trivacant lacunary POM exhibited a unique field-induced single-molecule magnet (SMM) property for a high spin FeIII ion. 4 Ab initio calculations showed that the axial ligand elongations of the octahedrally coordinated [FeO6]9-unit destabilized the dx2-y2 orbital, which lowered the sextet-quartet gap, resulting in a large magnetic anisotropy of the FeIII ion required for SMMs. 6 It is noteworthy that although a number of reports of mononuclear SMMs consisted of 3d transition metals have been reported,7 there have been only two reports on mononuclear FeIII complexes exhibiting SMM properties8 other than our mononuclear FeIII-POM, to the best of our knowledge: i.e., Mossin, Mindiola, and  Instruments. IR spectra were measured on JASCO FT/IR-4100 using KBr disks. UV/vis spectra were measured on JASCO V-570. Cold-spray ionization (CSI) mass spectra were recorded on JEOL JMS-T100CS. Thermogravimetric and differential thermal analyses (TG-DTA) were performed on Rigaku Thermo plus TG 8120. ICP-AES analyses for Fe, Ge, and W were performed on Shimadzu ICPS-8100. Elemental analyses were performed on Elemental vario MICRO cube (for C, H, and N) at the Elemental Analysis Center of School of Science of the University of Tokyo.
X-ray Crystallography. Diffraction measurements were made on a Rigaku MicroMax-007 Saturn 724 CCD detector with graphic monochromated Mo K radiation ( = 0.71069 Å, 50 kV, 24 mA) at 123 K. The data were collected and processed using CrystalClear12 and HKL2000.13 Neutral scattering factors were obtained from the standard source. In the reduction of data, Lorentz and polarization corrections were made. The structural analyses were performed using CrystalStructure14 and WinGX.15 All structures were solved by SHELXS-97 (direct methods) and refined by SHELXL-2014.16 The metal atoms (Fe, Ge, and W) and oxygen atoms in the POM frameworks were refined anisotropically. CCDC-1824491 (IIGe) and CCDC-1035329 (IISi) contains the supplementary crystallographic data for this paper. The data can be obtained free of charge via www.ccdc.cam.ac.uk/conts/retrieving.html (or from the Cambridge Crystallographic Data Centre, 12, Union Road, Cambridge CB2 1EZ, UK; Fax: (+44) 1223-336-033; or deposit@ccdc.cam.ac.uk).

Bond Valence Sum (BVS) Calculations.
The BVS values were calculated by the expression for the variation of the length rij of a bond between two atoms i and j in observed crystal with valence Vi.
where B is constant equal to 0.37 Å, r'0 is bond valence parameter for a given atom pair.17 Magnetic Susceptibilities. Magnetic susceptibilities of polycrystalline samples were measured on Quantum Design MPMS-XL7. Direct current (dc) magnetic susceptibility measurements were carried out between 1.9 and 300 K under 0.1 T magnetic field. Diamagnetic corrections were applied by the diamagnetisms of the sample holder and IGe. Alternating current (ac) magnetic susceptibility measurements were carried out under 0.1 T dc field and 3.96 × 10-4 T ac oscillating field. Variablefield (1-7 T) magnetization measurements were carried out in the temperature range of 1.9-10 K.
Then, diethyl ether (50 mL) was added. Pale yellow precipitates formed were filtered off, followed by recrystallization from a mixture of 1,2-dichloroethane and diethyl ether. Pale yellow crystals of IIGe suitable for X-ray crystallographic analysis were obtained (33.8 mg, 36% yield based on IGe). The coordination geometry of the introduced FeIII ion was six-coordinate distorted octahedral:
The UV/vis absorption spectrum of IIGe in acetonitrile showed a broad band at 350 nm assignable to LMCT transition of the [FeO6]9-unit. The LMCT absorption band was observed at slightly shorter wavelength in the case of IISi (345 nm) ( Figure S4). All the above mentioned results, the elemental analyses, and TG-DTA data showed that the formula of IIGe is TBA7H10[(A--GeW9O34)2Fe]·3H2O.  FeIII ion. We attempted to fit the magnetization curves using an isotropic g-values and found a reasonable match with experiment, although not as accurate as the former fit due to the lower number of fitting variables.
The ac magnetic susceptibility of IIGe under the external dc magnetic field of 0.1 T showed temperature-and frequency-dependence of ' and '', indicating the field-induced slow relaxation of magnetization characteristic for SMMs (Figures 2a and S6). The Cole−Cole plots for IIGe in the form of '' versus ' were fitted by means of the generalized Debye model,28 which showed the small  values in the range of 0.06−0.09, thus indicating a single relaxation process ( Figure S7). According to the Arrhenius plot, the energy barrier for the magnetization reversal of IIGe was 11.4 K (0 = 1.3 × 10-6 s). This value was slightly higher than that of IISi (9.2 K, 0 = 3.3 × 10-6 s) presumably owing to the larger D value of IIGe (Figure 2b).4  and dz2 orbitals. Dynamical correlation was taken into account by the NEVPT2 method and spin-orbit effects were considered by a quasi-degenerate perturbation theory (QDPT) step for state interaction due to the electronic spin-orbit coupling operator (see Computational Details section for further information). Ab initio ligand field theory analysis pictures a qualitatively similar d-orbital splitting for IIGe and IISi (Figure 4b). From the CASSCF(5,5) energies and wave functions, AILFT yields dorbital energies which are more consistent with a square planar coordination than with an octahedral geometry. This is in line with the axially elongated Fe-O distance associated to the more exposed position of equatorial oxygen donor atoms in comparison to the axial oxygen atoms (Table 2).29 Given the more pronounced axial elongation of IISi, the total orbital energy splitting for this complex is larger than IIGe. In both cases, dxy and dyz orbitals are lower in energy and almost degenerate, with an energy separation of 397.7 cm-1 (IIGe) and 239.3 cm-1 (IISi). The third orbital is mainly dxy, with an energy of 4316.9 cm-1 (IIGe) and 4382.8 cm-1 (IISi). The dz2 orbitals are the most differing, as they appear at 9211.9 cm-1 (IIGe) and 7962.3 cm-1 (IISi). Finally, the antibonding dx2-y2 orbitals are at 17733.9 cm-1 (IIGe) and 18771.3 cm-1 (IISi) (Figure 4b). As noted in our previous theoretical investigations,6 zero-field splitting parameters of high spin d5 complexes tend to be underestimated by CASSCF calculations as they chiefly depend on the sextet-quartet energy gap, which is overestimated under this methodology. As expected, NEVPT2 calculations tend to improve results, although their correction is only partial. To account for these limitations, CASSCF(9,7) calculations were performed including the bonding complement of the antibonding dx2-y2 and dz2 orbitals. 11 sextet and 24 quartet roots where included in the state average orbital optimization. In this way, 10 of the sextet roots are LMCT states, which will give a higher weight of charge transfer configurations and lower the sextet-quartet gap. Of course, the inclusion of more LMCT roots will increasingly bias the result in this direction, potentially leading to calculation artifacts. To prevent from this possibility, LMCT transitions were calculated, which are in reasonable agreement with experimental UV/vis spectra ( Figure S4, Table S2). Considering NEVPT2 energies, the sextet-quartet gaps are 8784 cm-1 and 8478 cm-1, while the zero-field splitting parameters (D) are -1.07 cm-1 and -1.01 cm-1 for IIGe and IISi, respectively (Table S3). These values are in close agreement with the experimental fit for IIGe (-1.26 cm-1) and IISi (-1.22 cm-1). Then, axial magnetic anisotropy is slightly smaller for IISi in comparison to IIGe. In addition, calculations indicate that E/D is lower for IIGe (0.056) in comparison to IISi (0.083), in line with the observed enhancement of SMM properties for IIGe (Figure 2b). Although the differences in spin Hamiltonian parameters are small for both complexes, the trend of a higher axiality for IIGe shows up independently of the active space (5,5 or 9,7) for both CASSCF and NEVPT2 energies.
The orientation of the anisotropy axis matches the vector connecting the axial donor atoms ( Figure   5). The alignment with the z-axis can be rationalized if we consider the sextet-quartet excitations which are responsible for magnetic anisotropy in a high spin d5 ion. In the free ion limit, the ground 6S term mixes with the 4P excited term by the SOC operator. Under a square planar ligand field, the 4P term will split in 4A2 and 4E states and mix with levels stemming from the other quartet terms. Spinorbit coupling mixing to the 4A2 level will be related with lz and 4E state interaction will be related to lx,y (see Figure 6). Thus, the difference in the x and y directions due to the structural distortion will be related with the presence of a nonzero component for the rhombic zero-field splitting parameter (E).
This rhombic contribution favors the quantum tunneling contribution resulting in a faster magnetic relaxation worsening SMM properties. In the case of IISi and IIGe, several excitations contribute to D and E, given their departure from square planar geometry. As expected, sextet-quartet excitations contribute more than sextet-sextet transitions to zero-field splitting parameters, especially for E (Tables S3). Table S4 shows the contribution of all quartet and sextet states to D and E for IISi and IIGe. Although the final D and E parameters results for the sum (or cancellation) of many contributions, it is possible to appreciate how cancellation of contributions to E is less effective in IISi when comparing with IIGe. Finally, we stress that the relatively good agreement between the calculated magnetic anisotropy properties with the experimental data does not guarantee a quantitative prediction due to the small energy differences involves for these systems.  Changes in the coordination geometries engendered the slightly larger magnetic anisotropy for IIGe, likely resulting in the increase in the energy barrier of the magnetization reversal. Although the difference in coordination geometries and the magnetic properties of IIGe and IISi were small, the small changes of magnetic properties were consistent with that of the theoretical results.
Multireference calculations provide similar values of the zero-field splitting parameters for both compounds, with IIGe presenting moderately larger and more axial parameters. E/D ratio shows to be sensitive to the deviation of the coordination environment from the ideal elongated octahedral geometry.

Supporting Information
The Supporting Information is available free of charge on the ACS Publications website.
Crystallographic data, CSI mass spectrum, UV/vis spectra, magnetic data, and cartesian coordinates of molecular models for electronic structure calculations (PDF)

Notes
The authors declare no competing financial interest.

TABLE OF CONTENTS
The effect of heteroatoms on single-molecule magnet properties of mononuclear high-spin FeIII magnetic center within lacunary polyoxometalates was investigated by combining experiment and multireference ab initio calculations. By changing the heteroatoms from Si to Ge, the coordination geometry around the FeIII ion is mildly changed, which engendered the slightly larger magnetic anisotropy, likely resulting in a small enhancement of the single-molecule magnet properties.