Insight into the Gd–Pt Bond: Slow Magnetic Relaxation of a Heterometallic Gd–Pt Complex

Lanthanide (Ln) compounds are common research targets in the field of magnetism and optics. Their properties arise from the electron localized in the f-orbital. Moreover, the effect of the covalency between lanthanide and ligands on magnetism attracted significant attention. We provided insight into the {[Pt(PhSAc) single-crystal polarized X-ray absorption near edge structure (XANES) reveal the electronic states around metal ion, where spectra of Gd- L III edges show the Gd – Pt direction has the highest covalency (less ionic) around Gd ion in 2 . In addition, calculating natural bonding (NBO) analysis, natural population analysis (NPA), LOL, and atoms in molecules (AIM), ab initio calculations reveal the role of metallic and organic ligands in the electronic and magnetic properties of Ln complexes. The slow magnetization relaxation of the Gd complex, which has not been reported previously in the Pt – Gd – Pt system, was observed up to 45K , the highest temperature reported to date among isolated Gd-complexes. The study about the effect of the covalency between lanthanide and ligands on magnetism attracted significant attention. We synthesized the heterometallic Ln−Pt complexes, in which the diamagnetic Pt(II) ions interact with the Gd(III) ion. We evaluated the Ln−Pt bond. Also, this complex showed a slow magnetic relaxation up to 45K which is the highest temperature reported to date for an isolated Gd-complex.


TOC GRAPHICS
The study about the effect of the covalency between lanthanide and ligands on magnetism attracted significant attention. We synthesized the heterometallic Ln−Pt complexes, in which the diamagnetic Pt(II) ions interact with the Gd(III) ion. We evaluated the Ln−Pt bond. Also, this complex showed a slow magnetic relaxation up to 45K which is the highest temperature reported to date for an isolated Gd-complex.

INTRODUCTION
The properties of lanthanide (Ln) compounds have made them prime research candidates in the field of magnetism, 1 optics, 2 and catalysis. 3 Control of magnetism and optical properties, which arise from the 4f electrons, is highly desired in materials science. The stereotypical understanding of the 4f electrons is often considered to be localized (core-like). They do not undergo large crystal fields (CF) splitting; thus, Ln(III) ions (except for Ce(III) or Ln(II) ions) in coordination complexes mainly behave like free ions.
Ln-based single-molecule magnets (SMMs) are a crucial subject for molecular spintronics, in which the molecular spin is utilized for information processing. Previous studies have shown that Dy complexes have strong magnetic anisotropies and large activation barriers (E) of spin flipping among spin up and spin down states. As a result, Dy complexes have been developed as high-performance molecular magnets. 1a-h Recently, organometallic Dy complexes that have a strong magnetic anisotropy or high symmetry around the Dy ion with a large E have been reported. In this system, a (weak) CF is a dominant source of the perturbation to resolve the degeneration of ground multiplet ( 6 H15/2) so that the axial magnetic states are stabilized. 1a-c, 4, 5 Therefore, an electrostatic CF has been used to control the physical properties of the f-electrons.
In addition, significant efforts have been made to enhance the features by changing the CF, magnetic interactions, and ligands.
In contrast, the reports in which the f orbital itself interacts with the orbitals of other atoms (s-f, p-f, d-f interaction) have been reported for intermetallics, metal oxides, and metal salts, and superconducting behavior was demonstrated in the 1980s. 6 Recently, Kuga et al. 9 Although the details of the interaction between f-orbitals and ligands for other complexes are still unclear, the effect of the heterometal bonds needs to be considered when discussing the magnetic properties of heterometallic bonded systems.
Thus, the perturbations caused by the nature of bonds cannot be ignored necessitating combined theoretical and experimental investigations.
Considering the difference in the size of the orbital and the electronegativity between the main group element and the transition metal element, it seems that the metallic ligand can enlarge the dative bond or covalency contribution. Bendix et al. estimated that minor changes in the orbital overlap among Ln and Pt/Pd affect the magnitude of zero magnetic fields splitting. 10a, 10h On the other hand, the theoretical calculation (localized orbital locator (LOL) analysis) and experimental data (X-ray absorption spectra: XAS) of Ln-Pd and Ln-Pt complexes indicated the electron donation from the Pd and Pt ions to Ln. 10b, 10c Electron spin resonance, luminescence spectra and inelastic neutron scattering are strong experimental methods to reveal the splitting of mJ states of Ln ion by fitting with the CF parameters. 10a,d,h However, theoretical approaches should also be considered with the ligand field theory and covalency in calculating the splitting energies of the mJ states of Ln ions to circumvent the risk of overparameterization or wrong fitting value for systems involving a small splitting of the mJ states (especially in Gd ion).
In this work, insight into the Gd-Pt bond (of the heterometallic Ln-Pt complexes: (1), Gd(2); PhSAc = thiobenzoate, NEt4 = tetraethylammonium)) was provided; single-crystal polarized X-ray absorption near edge structure (XANES) reveals the electronic states around the metal ion, where spectra of Gd-LIII edges show the Gd-Pt direction has the highest covalency (less ionic) around Gd ion in 2. In addition, natural bonding (NBO) analysis, natural population analysis (NPA), LOL, atoms in molecules (AIM), and ab initio calculations revealed the role of metallic and organic ligands in the electronic and magnetic properties of Ln complexes. PhSAc ligand is bulky, increases the intermolecular distance, and is effective in suppressing magnetic and intermolecular interactions between the Ln ions. Furthermore, it is effective for adding optical characteristics such as luminescence. The slow magnetization relaxation of the Gd complex was observed up to 45K, whic is the highest temperature reported to date among isolated Gd-complexes.

RESULTS & DISCUSSION
Crystal Structure. The crystal structure of 2 is shown in Figure 1 (crystal data are summarized in Table S1, CCDC No. 2054549, 20544545, 2054548). The three complexes are isostructural, and therefore only 2 is described here as a representative of the other examples. The complex crystallized in the Cc space group, with eight PhSAc − ligands, two Pt ions, one Gd ion, one NEt4 countercation, and two DMF crystalline solvents per asymmetry unit. The Gd ion forms a paddle-  (Table S2).
Each Pt ion is further coordinated with four sulfur atoms to form a square pyramidal structure with the Gd as the apex. The intramolecular Gd-Pt distances are 3.605(6) and 3.631(6) Å. These are shorter than the sum of the atomic radii (4.10 Å) but longer than the covalent (3.32 Å) and ionic  interactions involving the f-orbitals of metal complexes by using X-ray absorption. 13,14 The XANES spectra at the Pt L3-edge are shown in Figure S1a. A white line shift from 11562.5 eV (peak top of the edge of the standard sample of Pt foil) to 11563.5 eV (for 1), and 11563.4 eV (for 2) was observed. The spectra were simulated by FDMNES package (Figure S1b-d). 15  At the Pt LIII-edge, 2 showed that a weak white line (WL) absorption in the Pt-Ln-Pt direction (c direction: the direction perpendicular to Pt dx2-y2), but a strong WL absorption in the vertical direction (The results of density of states (DOS) calculation also supported the fact ( Figure   S1b,c)). On the other hand, at the Ln LIII-edge, contrast results were obtained. In 2, the WL absorption is the strongest in the Pt−Ln−Pt (parallel) direction, but weak at the perpendicular direction as well as at 45 degrees from c axis. The 2p→5d transition is originally non-oriented (The results of DOS calculation also supported the fact ( Figure S1d).), but the introduction of electrons into the 4f band shields the nuclear potential from the 5d band. 16 In other words, if the covalency between Ln and the ligand is high (or less ionic), the absorption intensity will increase.
It can be said that the axial direction has higher covalency than the in-plane direction. Since the signal of the powder sample is an average of all orientations, the Pt direction has the highest covalency in 2. Therefore, it was possible to experimentally clarify the anisotropy of the Ln complex's covalency in which it may be overall means ligand field splitting considering covalency and electron repulsion, using the single-crystal polarized X-ray absorption spectrum. In this work, only the two-dimensional orientation is considered, and the LnPt molecule orientation (Pt-Ln-Pt direction) is also slightly leaning from c axis. Therefore, the comparison with the isotropic powder sample is not simple. However, further research will enable us to visualize the three-dimensional anisotropy of covalency.      the LOL function is a simple sum of the LOL functions of each atom, the bond between the atoms is ionic. If a new peak exceeding 0.5 is observed between the two atoms, the bond between these atoms is covalent. A dative bond (footnote: Although it is labeled as polar covalent bond in the references, 18a it is labeled as dative bond according to the AIM classification below. 18b ) will appear as a shoulder peak or as a plateau below 0.5. The LOL results for 2 show the existence of a shoulder, indicating that the Pt center was strongly polarized by the Ln center ( Figure 4). These results may support the results of AIM. However, the LOL value ( = 0.2) is smaller than that obtained for the reported complexes (cf.  = 0.3); it means that the Gd−Pt bond in 2 is more ionic than that found for the reported complexes. 10b,10c Even perfect covalent bond, which has 50% covalency and 50% ionic.)" 19 The contour plots on the Pt−Ln−Pt plane and the calculated value of  (electron density), and Etotal (total energy density) were shown in Figure 5, S5, and Table 2  Ab initio calculations were performed using the ORCA 4.2.1 program. 20 The active space consists of 7 electrons in 7 orbitals (CASSCF (7,7)). For the state-averaged CASSCF procedure, 1 octet, 48 sextets, 392 quartets, and 784 doublets were considered 21 for the configuration state functions (CSFs) and used for the spin-orbit coupling calculations. The low-lying energy states from spin-orbit coupling are summarized in Table 3. Ab initio ligand field theory (AILFT) 11b was applied to check the ligand field splitting of 4f-orbitals of 2'-Me ( Figure 6). The results showed a ligand field splitting energy of ~450 cm − 1 . In terms of orbital stability, the axial (Pt-Gd-Pt direction) fz 3 , fxz 2 and fyz 2 orbitals are most stabilized, the equatorial fx(x 2 -3y 2 ) and fy(3x 2 -y 2 ) orbitals are intermediate while the diagonal fxyz, fz(x 2 -y 2 ) orbitals are the least stabilized. This order (axial, equatorial, diagonal) is similar to the results of the single-crystal XAFS -less ionic orientation order (parallel, perpendicular, 45 degrees). In addition, fz 3 orbital is less stable than fxz 2 , fyz 2 orbital, because fz 3 orbital also undergoes electrostatic repulsion from the lone pair of the Pt ion.  23 However, the spin-lattice relaxation is only observed below 10 K, which is a much lower temperature range than the 36 K observed for heterometal complexes. 10g For 20.1, a slow magnetic relaxation was observed up to 45 K (Figure 8b). This is the highest temperature reported to date for a discrete Gd complexes.
mT values for 2 and 20.1(10% of 2 diluted in 90 % of 0) remained constant at 8.168 cm 3 mol −1 K and 7.872 cm 3 mol −1 K, respectively, which are consistent with the expected value for an uncoupled Gd ion (7.875 cm 3 mol −1 K) (Figure 8a and S6). Moreover, 2 and 20.1 did not exhibit a magnetic ordering down to 2 K. The magnetization vs. field plots for 2 and 20.1 are shown in Figure. S7. There is butterfly-type hysteresis for 2 and 20.1 at 1.8 K, indicating a rapid quantum tunnelling effect. The hysteresis temperature is higher than those reported for Gd complexes (cf. 0.7 K). 10g No clear frequency dependence of the magnetization was observed for 2 without a dc field because of the fast QTM (the quantum tunnelling of the magnetization) process. However, 2 shows the explicit frequency dependency of the magnetization with applied dc field was applied ( Figure   S8). Furthermore, as the dc field was gradually increased, the maximum " shifted to a lower frequency, and an additional peak was observed above 4000 Oe dc fields. This second peak was not observed for 20.1 ( Figure S9) performed on this data under various conditions. In conclusion, the process involving both the Orbach and two-phonon Raman ( ∝ 1/T 3 ) processes afforded the best fit ( Figure S10). Lowenergy phonon modes (15-60 cm -1 ) that may resonate with the magnetic Raman process have been experimentally observed, and DFT calculations have revealed that they correspond to the PtS4−LnO8−PtS4 oscillation modes ( Figure S12, S13. The Raman spectra discuss in SI.). The smaller resonant phonon mode that resonates with the Orbach process is thought to be caused by the lattice vibration of the crystal packing ( Figure S11).   To investigate the zero-field splitting of Gd ion, we measured ESR spectrum of 20.05 (5% of 2 diluted in 95 % of 0) ( Figure. 7 and S17). The data were fitted using the Hamiltonian Ĥ = ̂+ 2 0̂2 0 + 2 2̂2 2 (Details are shown in SI.) with EasySpin package. 23

Experimental Section
General information. K2[PtCl4] was obtained from Kanto Chemical Co., Inc. and used as received.
LnCl3·nH2O was obtained from Strem Chemicals Inc. and used as received. Thiobenzoic acid ((PhSAc)H) was obtained from Tokyo Chemical Industry Co., Ltd. and used as received. All synthetic processes were performed under air.
Calculations. TDDFT, NBO, NPA, LOL, simulation of Raman spectra analyses were performed using the Gaussian16 program package. Quantum theory of atoms in molecules (QTAIM) analysis was performed in the AIMAll package. 26 The molecular structure of the model is made from the crystal structure of 1, 2 and simplified using GaussView5.0. The B3LYP functional 27 29 was used to generate the auxiliary basis sets for resolution of identity approximation. 30 The active space consists of the seven orbitals that have 4f character and seven electrons ((7,7) CASSCF). One octet state, 48 sextet states, 392 quartet states, and 784 doublet states were considered 21 and used for the following spin-orbit coupling calculations. The low-lying energy states from spin-orbit coupling are summarized in Table S8. Ab initio ligand field theory (AILFT) 11b was applied to check the ligand field splitting of 4f-orbitals ( Figure S11).
Orbitals were visualized using Multiwfn 3.7 and VESTA 3. 4.4. 31 The coordination geometry as determined using SHAPE2.1. 32 Charge decomposition analysis (CDA) is one of the methods to analyze donor-acceptor interactions of each fragment of complexes. 33 Here, We took complex 2 to three parts: Gd ion, Pt ions, and PhSAc ligands. CDA was performed on Multiwfn 3.7. 31a X-ray absorption near-edge structure. X-ray absorption spectroscopy was carried out at the BL-12C and BL9A, High Energy Accelerator Research Organization (KEK) (the edge position was calibrated using that of Pt foil). The analysis was performed using the Demeter software platform. 34 The simulated spectra are calculated with 6 Å radius by FDMNES package under Quadrupole, Relativiste, Spinorbite, Density_all options. 18 Magnetic susceptibility measurements. Both dc and ac magnetic susceptibility measurements were performed on solid polycrystalline samples with a Quantum Design MPMS3 SQUID magnetometer in applied dc fields. ESR spectra were conducted on solid polycrystalline samples using a JEOL JES-FA100.

ASSOCIATED CONTENT
Supporting Information.
The following files are available free of charge.
Some additional figure and plots.

Author Contributions
Authors may have contributed in multiple roles.

Notes
The authors declare no competing financial interests.