Targeted Synthesis of End-On Dinitrogen-Bridged Lanthanide Metallocenes and Their Reactivity as Divalent Synthons

High-yield syntheses of the lanthanide dinitrogen complexes [(Cp2tttM)2(μ-1,2-N2)] (1M, M = Gd, Tb, Dy; Cpttt = 1,2,4-C5tBu3H2), in which the [N2]2– ligands solely adopt the rare end-on or 1,2-bridging mode, are reported. The bulk of the tert-butyl substituents and the smaller radii of gadolinium, terbium, and dysprosium preclude formation of the side-on dinitrogen bonding mode on steric grounds. Elongation of the nitrogen-nitrogen bond relative to N2 is observed in 1M, and their Raman spectra show a major absorption consistent with N=N double bonds. Computational analysis of 1Gd identifies that the local symmetry of the metallocene units lifts the degeneracy of two 5dπ orbitals, leading to differing overlap with the π* orbitals of [N2]2–, a consequence of which is that the dinitrogen ligand occupies a singlet ground state. Magnetic measurements reveal antiferromagnetic exchange in 1M and single-molecule magnet (SMM) behavior in 1Dy. Ab initio calculations show that the magnetic easy axis in the ground doublets of 1Tb and 1Dy align with the {M–N=N–M} connectivity, in contrast to the usual scenario in dysprosium metallocene SMMs, where the axis passes through the cyclopentadienyl ligands. The [N2]2– ligands in 1M allow these compounds to be regarded as two-electron reducing agents, serving as synthons for divalent gadolinium, terbium, and dysprosium. Proof of principle for this concept is obtained in the reactions of 1M with 2,2′-bipyridyl (bipy) to give [Cp2tttM(κ2-bipy)] (2M, M = Gd, Tb, Dy), in which the lanthanide is ligated by a bipy radical anion, with strong metal–ligand direct exchange coupling.


X-ray Crystallography
Single-crystal X-ray diffraction measurements were carried out on an Agilent Gemini Ultra or a Rigaku HyPix 6000HE diffractometer using Cu-K radiation ( = 1.54184Å) at 100 K. Crystals were mounted on MiTiGen loops from pump oil kept over activated 4 Å molecular sieves in a glove box.Data collection and processing was handled by CrysAlis Pro.Structure solution and model refinement were performed using the Olex2 package and all software within. 4,5Anisotropic thermal parameters were used for non-hydrogen atoms and isotropic parameters for hydrogen atoms.Hydrogen atoms on carbons were added geometrically and refined using a riding model.Solvent masking was used in the refinement of 1Gd, 1Tb, and 1Dy due to the presence of highly disordered pentane molecules in the lattice.

DFT calculations
The DFT calculations on 1Gd were performed on coordinates obtained from the X-ray structure using Gaussian09 programme without optimizations, except for the hydrogen atoms.The hybrid B3LYP functional 6 was used along with the effective core potential 'Stuttgart RSC 1997' basis set 7 for gadolinium, and the Ahlrichs triple- TZV basis set 8 for other atoms.The NBO analysis was performed using NBO 3.0 as implemented in Gaussian09.The TD-DFT calculations were performed with the PBE0 hybrid functional 9 using ORCA 5.0.2. 10 The relativistic effects were included with the Douglas-Kroll-Hess (DKH) Hamiltonian, together with the scalar relativistic contracted version of the basis functions def2-QZVP for Gd, and def2-TZVP for N and def2-SVP for other atoms.The CPCM solvation model was used to consider solvent effects.
To gain insight into the magnetic exchange coupling, we also performed DFT calculations in combination with the broken symmetry (BS) approach in ORCA.Relativistic effects were included with the DKH Hamiltonian, together with the scalar relativistic contracted version of the basis functions def2-QZVP for gadolinium and def2-TZVP for other atoms.In these calculations, the well-known B3LYP functional was employed to extract the isotropic exchange coupling constant () using equation S1, where  ,  and  represent the energy of the triplet state, broken-symmetry state and total spin, respectively.
For 1Gd, the energy of the high-spin (triplet) state was determined as  = -25300.704939hartree and the broken symmetry state as  = -25300.707329hartree, leading to  = -9.37 cm -1 .
For 2Gd, the energy of the high-spin (triplet) state was determined as  = -13349.623384hartree and the broken symmetry state as  = -13349.624128hartree, leading to  = -8.16cm -1 .

Magnetic measurements
Magnetic measurements were recorded on a Quantum Design MPMS-XL7 SQUID magnetometer equipped with a 7 T (70 kOe) magnet.The samples were restrained in eicosane and sealed in 7 mm NMR tubes.Direct current (DC) magnetic susceptibility measurements were performed on polycrystalline samples in the temperature range 1.9-300 K and using an applied field of 1000 Oe.Alternating current (AC) susceptibility measurements were performed using an AC field of 3 Oe in zero DC field.Diamagnetic corrections were made using Pascal's constants for all the constituent atoms. 11

Simulation of Magnetic Susceptibility and Magnetization data
Simulations for [{(Cp ttt )2M}2(-1,2-N2)] (1M, M = Gd, Tb, Dy) To quantify the exchange interactions, we have simulated the molar magnetic susceptibility and magnetization data for all complexes using the PHI software. 9For 1Gd, the isotropic Hamiltonian stated as equation 1 in the main text was used, whereas the Hamiltonian stated as equation 2 in the main text was used for 1Tb and 1Dy.
For 1Tb and 1Dy, we considered the crystal field parameters obtained from the ab initio calculations (Table S11) and an intermolecular interaction term, , calculated to be +0.01 cm -1 and -0.05 cm -1 , respectively.The best simulation of both susceptibility and magnetization results in weak antiferromagnetic interactions for all complexes.
Simulations for [(Cp ttt )2M(bipy)] (2M, M = Gd, Tb, Dy) To quantify the exchange interaction between gadolinium and the bipy radical anion ligand in 2Gd, we performed a simulation using an isotropic Hamiltonian with an intermolecular term  = −0.15cm -1 , as expressed in equation 4 in the main text.The Hamiltonian stated as equation 5 in the main text, which includes the ab initio crystal field parameters, was used for 2Tb and 2Dy.
For 2Tb and 2Dy, the crystal field parameters were calculated for the hypothetical oxidized complexes [(Cp ttt )2Tb(bipy)] + and [(Cp ttt )2Dy(bipy)] + in order to simplify the active space (Table S15).Intermolecular terms of  = −0.45cm -1 and  = −0.60 cm -1 , respectively, were used in the simulations.For all 2M complexes, reasonable simulations of the  () data were obtained, but the () data could not accurately be reproduced.

Ab initio calculations
All calculations were carried out on the coordinates obtained from the relevant crystal structure using the ORCA 5.0.2 software package.The positions of hydrogen atoms were optimized at the DFT level using a pure GGA PBE exchange correlation functional, keeping the position of the other atoms constant.The DKH Hamiltonian was used throughout to consider relativistic effects.The lanthanide centre was modelled with the SARC2-DKH-QZVP basis set, and all other atoms were treated with the DKH-def2-TZVP basis set in combination with the 'AutoAux' auxiliary basis set. 12The active space CAS(8,7) was constructed from eight electrons in seven f-orbitals for terbium and CAS(9,7) was constructed from nine electrons in seven f-orbitals for dysprosium.In the configuration interaction procedure, seven septets, 76 quintets, and 52 triplets were computed for terbium whereas 21 sextets, 128 quartets, and 130 doublets were considered for dysprosium.To include the spin-orbit coupling, we also used the quasi-degenerate perturbation theory (QDPT) approach using SA-CASSCF wave functions. 13The SINGLE_ANISO module as implemented in ORCA was used to compute the g-tensor and crystal field parameters of the low-lying excited states using previously calculated spin-orbit states.

Figure S8 .
Figure S8.Real (left) and imaginary (right) components of the AC susceptibility as a function of temperature at 1000 Hz frequency in an AC field of 3 Oe and zero DC field for 1Tb.

Figure S11 .
Figure S11.Cole-Cole plot (left) and temperature-dependence of the relaxation time (right) for 1Dy, solid lines are fits to the data according to equations S2 and S3 and the parameters in TableS9.

Figure S13 .
Figure S13.Left:  () for 2Gd.  is 7.48 cm 3 K mol -1 at 300 K and 1.68 cm 3 K mol -1 at 2 K.The red points represent a of the data according to equation in the main text.Right: () data at 2 K.

Figure S14 .
Figure S14.Left:  () for 2Tb.  is 11.43 cm 3 K mol -1 at 300 K and 1.29 cm 3 K mol -1 at 2 K.The red points represent a fit of the data according to equation 5 in the main text.Right: () data at 1.9, 3.0 and 5.0 K.

Figure S18 .
Figure S18.Left:  () for 2Dy.  is 13.96 cm 3 K mol -1 at 300 K and 1.51 cm 3 K mol -1 at 2 K.The red points represent a fit of the data according to equation 5 in the main text.Right: () data at 1.9, 3.0 and 5.0 K.

Figure S20 .
Figure S20.Cole-Cole plot (left) and temperature-dependence of the relaxation time (right) for 2Dy, where solid lines are fits to the data according to equations S2 and S3 and the parameters in TableS10.

Table S1 .
Crystal data and structure refinement parameters for 1Gd, 1Tb, and 1Dy.

Table S2 .
Crystal data and structure refinement parameters for 2Gd, 2Tb, and 2Dy.

Table S5 .
DFT calculated spin densities on selected atoms for 1Gd.

Table S6 .
Selected bonding orbitals from the natural bond orbital (NBO) analysis of 1Gd.

Table S7 .
Computed excitation wavelengths () and oscillator strengths (f) in length representation for 2Gd.

Table S12 .
Ab initio calculated low-lying spin-orbit energy states for 1Dy and 1Tb.

Table S13 .
Computed energies, tunnel splittings, -tensors and wavefunction composition for the first six low-lying states in 1Tb.

Table S14 .
Kramers doublet (KD) energies, -tensors, angle between the anisotropic axis of the excited states and ground state, and wavefunction compositions for 1Dy.