Thermally Stable Terbium(II) and Dysprosium(II) Bis-amidinate Complexes

Abstract

The thermostable four-coordinate divalent lanthanide (Ln) bis-amidinate complexes [Ln(Piso)₂] (Ln = Tb, Dy; Piso = {(NDipp)₂C^tBu}, Dipp = C₆H₃ⁱPr₂-2,6) were prepared by the reduction of parent five-coordinate Ln(III) precursors [Ln(Piso)₂I] (Ln = Tb, Dy) with KC₈; halide abstraction of [Ln(Piso)₂I] with [H(SiEt₃)₂][B(C₆F₅)] gave the respective Ln(III) complexes [Ln(Piso)₂][B(C₆F₅)]. All complexes were characterized by X-ray diffraction, ICP-MS, elemental analysis, SQUID magnetometry, UV-vis-NIR, ATR-IR, NMR, and EPR spectroscopy and ab initio CASSCF-SO calculations. These data consistently show that [Ln(Piso)₂] formally exhibit Ln(II) centers with 4fⁿ5d_z2¹ (Ln = Tb, n = 8; Dy, n = 9) valence electron configurations. We show that simple assignments of the f-d coupling to either L-S or J-s schemes are an oversimplification, especially in the presence of significant crystal field splitting. The coordination geometry of [Ln(Piso)₂] is intermediate between square planar and tetrahedral. Projecting from the quaternary carbon atoms of the CN₂ ligand backbones shows near-linear C···Ln···C arrangements. This results in strong axial ligand fields to give effective energy barriers to magnetic reversal of 1920(91) K for the Tb(II) analogue and 1964(48) K for Dy(II), the highest values observed for mononuclear Ln(II) single-molecule magnets, eclipsing 1738 K for [Tb(C₅ⁱPr₅)₂]. We tentatively attribute the fast zero-field magnetic relaxation for these complexes at low temperatures to transverse fields, resulting in considerable mixing of m_J states.

Scheme 1. Synthesis of 2-Ln and 3-Ln from 1-Ln (Ln = Tb, Dy)

Figure 1

Molecular structures of 1-Dy, 2-Dy, and 3-Dy, with selected atom labeling (counteranion not shown for 2-Dy). The displacement ellipsoids are set at the 50% probability level, and hydrogen atoms and lattice solvent are omitted for clarity. Key: Dy(III) = green, Dy(II) = purple, I = brown, N = blue, and C = gray.

Figure 2

Experimental UV–vis-NIR spectra of 0.2 mM solutions of 3-Ln (benzene), 2-Ln (DCM), 1-Ln (benzene), and KPiso (benzene) at room temperature; Ln = Dy (a) and Tb (b).

Figure 3

Plots of the natural log of the inverse relaxation time vs temperature for (a) powder and solution sample of 3-Tb under 0 Oe and (b) 3-Dy under 0 (circle points) and 1000 (square points) dc field (violet, purple, cyan, and green points are from ac data; orange and wine points are from dc data) at a temperature range from 2 to 150 K. (c) The enlarged scales of panel a from 60 to 120 K for 3-Tb. (d) The enlarged scales of panel b from 70 to 120 K for 3-Dy. The dashed orange, blue, and purple lines show isolated Orbach, Raman, and QTM fitting, respectively; black lines show their sum.

Figure 4

Magnetization M (μ_B) vs applied dc field H (T) plots for (a) 3-Tb and (b) 3-Dy, at temperature (K) intervals between 2 and 30 K at an average sweep rate of 22 Oe/s, to show the shapes of hysteresis loops. (Inset) Expanded view of the variable-field magnetization near the zero field at 20 to 50 K.

Figure 5

Singly degenerate 5d orbital for 3-Tb (left) and 3-Dy (right), calculated with CASSCF-SO, shown at an isosurface value of 0.04 au. The phases of the wave functions (colored lobes) are not observable and could be arbitrarily reversed.

Figure 6

Schematic for L–S (left) and J–s (right) coupling schemes, ignoring CF splitting; energies not to scale.

Figure 7

χ_MT values at 300 K for 4f⁸5d¹ Tb(II) (red) and 4f⁹5d¹ Dy(II) (blue) as a function of the isotropic f–d coupling parameter Ω, assuming a nondegenerate 5d orbital. These curves are simulated using eq 2 with fixed λ = −297 and −396 cm^–1 for Tb(II) and Dy(II), respectively. The dotted lines indicate where Ω = λ for each ion.