A High Anisotropy Barrier in a Sulfur‐Bridged Organodysprosium Single‐Molecule Magnet

Abstract

The sulfur-bridged dimers [{Cp′2Ln(μ-SSiPh3)}2] (Ln=Gd (1), Dy (2); Cp′=η5-C5H4Me) were synthesized by the transmetalation reactions between [Cp′3Ln] and Ph3SiSLi. Compound 2 is a single-molecule magnet with slow relaxation of magnetization up to 40 K and an anisotropy barrier of Ueff=133 cm−1. Insight into the SMM properties of 2 and closely related SMMs has been obtained using ab initio calculations. Following the discovery in the early 1990s that individual molecules can exhibit slow relaxation of magnetization and magnetic hysteresis,1 studies of single-molecule magnets (SMMs) have been dominated by large-spin, exchange-coupled cages of 3d transition metals.2 In addition to the intrinsic interest in the physics of SMMs, such compounds have been proposed for applications in molecular spintronics and in quantum computing.3 However, a major obstacle to realizing the applications of SMMs is that energy barriers to magnetization reversal, Ueff, and blocking temperatures, TB, lead to slow relaxation of magnetization only at liquid helium temperatures. In recent years, there has been a growing realization that focusing on enhancements in magnetic anisotropy rather than on ever-larger spins may be a more effective strategy for increasing Ueff and TB. Therefore, there has been a rise to prominence of the f elements in SMM studies,4 and lanthanide-containing SMMs (Ln-SMMs) in particular have enabled many exciting developments. For example, in 2003 Ishikawa et al. reported that slow relaxation of magnetization is possible in molecules that contain a single lanthanide ion, namely the terbium(III) phthalocyanine sandwich complex [TbPc2]−, with Ueff=230 cm−1.5 The family of lanthanide single-ion magnets (SIMs) continues to grow,6 and indeed the terbium(III) SIM [TbiPc2] (iPc=isopropylidenedioxyphthalocyanine), has the highest anisotropy barrier of Ueff=556 cm−1.6e Many polymetallic Ln-SMMs, the majority based on dysprosium(III), have also been reported, some of which show impressive SMM properties, such as high values of Ueff and/or TB.7 The dilanthanide SMMs [Ln2{N(SiMe3)2}4(thf)2(μ:η2:η2-N2)]− with Ln=Dy or Tb, reported by Evans, Long et al., are arguably the most notable recent examples because the LnIII ions are strongly exchange-coupled by the radical [N2]3− ligand.8 In the case of the terbium derivative, hysteresis loops in the magnetization versus field plots were observed up to 14 K. In the majority of Ln-SMMs, the LnIII ions reside in environments that consist of hard oxygen or nitrogen donor ligands. Whereas the spin configurations of LnIII ions are insensitive to their environment, their overall electronic structure may be very sensitive. We reasoned that using less-conventional ligands (in the context of SMMs) could be a route to modifying their properties. For example, using soft donors could have two effects: first, magnetic exhange between LnIII ions could be enhanced; second, very different crystals fields could be generated. Thus, as crystal field effects are the main interactions that lead to slowly relaxing magnetization in Ln-SMMs, there is merit in exploring soft ligand environments. We now report the {Dy2} SMM [{Cp′2Dy(μ-SSiPh3)}2] (2, Cp′=η5-C5H4Me), in which the Dy ions are bridged by thiolate ligands. Compound 2 is the first sulfur-bridged SMM, and the magnetization reversal barrier is the highest yet reported for a {Dy2} SMM. Ab initio calculations on 2 and related compounds provide insight into the factors that influence the magnetization reversal barrier. The sulfur-bridged dimers [{Cp′2Ln(μ-SSiPh3)}2] (Ln=Gd (1), Dy (2)) were synthesized by the ligand substitution reactions of Cp′3Ln with lithium triphenylsilylthiolate (Scheme 1). Synthesis of 1 and 2. The centrosymmetric dimers crystallize as 1⋅(toluene) and 2⋅(toluene), and are isostructural (Figure 1; Supporting Information, Figures S1,S2 and Tables S1,S2). ORTEP representation (ellipsoids set at 50 % probability) of the molecular structure of 2. For clarity, only the ipso carbons of the phenyl rings are shown, and the hydrogen atoms are omitted. Each LnIII ion in 1 and 2 is complexed by two η5-Cp′ ligands and by the sulfur atoms of two μ-SSiPh3 ligands, resulting in formal coordination numbers of eight. Alternatively, with respect to Cp ligand centroids and the sulfur atoms, the DyIII centers reside in pseudo tetrahedral geometries, with local C2 symmetry at Ln (see Table 1 for bond lengths and angles). The asymmetric η5-Cp coordination mode is reflected in the range of LnC distances, which are 2.622(7)–2.707(14) for 1 and 2.599(15)–2.693(10) Å for 2. The Ln1S1 and Ln1S1A distances of 2.7755(14) and 2.7961(15) Å in 1 and 2.7500(15) and 2.7761(16) Å in 2, combined with the Ln1-S1-Ln1A angles of 101.53(5) and 101.43(6) in 1 and 2, respectively, show that the Ln2S2 rings are rhomboidal. The LnC and LnS bond lengths and angles in 1 and 2 are typical of compounds of the type [{Cp2Ln(μ-SR)}2] with Ln=Gd or Dy.9 1 (Ln=Gd) 2 (Ln=Dy) Ln–C 2.622(7)–2.707(14) 2.599(15)–2.693(10) Ln1–S1 2.7755(14) 2.7500(15) Ln1–S1A 2.7961(15) 2.7761(16) S1-Ln1-S1A 78.47(4) 78.57(5) Ln1-S1-Ln1A 101.53(5) 101.43(6) Lanthanide chalcogenolate complexes have been extensively studied because of their applications in organic synthesis and in the catalytic ring-opening polymerization of ε-caprolactone.10 However, the dynamic magnetic properties of such compounds have not, to the best of our knowledge, been studied previously. Direct current (dc) magnetic susceptibility measurements on polycrystalline samples of 1 and 2 were carried out in the temperature range 2–300 K in an applied field of 1 kG. For 1, the value of χM T (χM is the molar magnetic susceptibility) per GdIII at 300 K is 15.5 cm3 K mol−1 (Supporting Information, Figure S3), which is in good agreement with the predicted value of χM T=15.76 cm3 K mol−1 for two non-interacting GdIII free ions (8S7/2 ground term, g=2). On cooling, the value of χM T decreases very slowly until about 30 K, and then below 30 K χM T decreases rapidly to reach a value of 6.7 cm3 K mol−1 at 2 K. The absence of first-order spin–orbit coupling in GdIII enables the exchange coupling constant, J, for 1 to be readily determined. Fitting the susceptibility data for 1 using the isotropic spin Hamiltonian H=−2 J[SGd1⋅SGd1A], where SGd1 and SGd1A denote the spins on Gd1 and Gd1A, respectively, produced an excellent fit for J=−0.105 cm−1. The same parameters also reproduced the field dependence of the isothermal magnetization of 1 measured at several temperatures between 2–9 K (Supporting Information, Figure S4). These data indicate that the GdIII ions in 1 are weakly antiferromagnetically coupled by the μ-thiolate ligands, which is consistent with other exchange-coupled gadolinium compounds, and can be rationalized in terms of the core-like nature of their valence 4f orbitals.11 In 2, the value of χM T at 300 K is 27.9 cm3 K mol−1, which agrees well with the expected value of 28.34 cm3 K mol−1 for two magnetically noninteracting DyIII ions with a 6H15/2 ground term (g=4/3; Supporting Information, Figure S5). On lowering the temperature, χM T decreases gradually until about 70 K, and then more rapidly to reach 11.2 cm3 K mol−1 at 2 K; this decrease is likely to be due to depopulation of excited mJ sublevels and/or weak antiferromagnetic exchange. At low temperatures, the isothermal M(H) data of 2 are far from reaching saturation under magnetic fields of 0–7 T, and the corresponding M(H/T) curves do not superimpose (Supporting Information, Figure S6), indicating the presence of significant magnetic anisotropy and/or low-lying excited states. The value of the magnetization at 7 T and 1.8 K is 11.4 μB (5.7 per DyIII ion), and is similar to that observed in other Dy complexes.7 Both the temperature dependence (Figure 2; Supporting Information, Figure S7) and the frequency dependence (Supporting Information, Figure S8) of the alternating current (ac) susceptibilities of 2 reveal features characteristic of an SMM. In zero dc field, the in-phase (χ′) (Supporting Information, Figure S7) and the out-of-phase (χ′′) components (Supporting Information, Figure 2) of the ac susceptibility show a frequency dependence up to about 40 K, and a single χ′′(T,ν) peak is observed below 32 K. Semicircular Cole–Cole plots of χ′′ vs. χ′ at several temperatures below 32 K were obtained (Supporting Information, Figure S9), and were fitted by a generalized Debye model using α parameters in the range α=0.01–0.15. This indicates a narrow range of magnetization relaxation times in 2. Above 20 K, the relaxation in 2 is thermally activated and the Arrhenius plot of ln τ versus 1/T, where τ is the relaxation time constant, is linear (Figure 3). Fitting the higher-temperature data to the Arrhenius law τ=τ0 exp(Ueff/kB T) produced an effective energy barrier to magnetization reversal of Ueff=(133±3.5) cm−1 (or (192±5) K) and a pre-exponential factor of τ0=2.38×10−7 s. The SMM behavior of 2 was confirmed by the observation of a narrow hysteresis in magnetization versus field measurements on polycrystalline samples at 1.8 K (Supporting Information, Figure S10). Temperature dependence of χ′′ in 2 at various frequencies in zero applied dc field and an ac field of Hac=1.55 Oe. Arrhenius plot of ln τ versus (1/T) for 2. The solid line is the best fit of the data in the thermally activated regime (T>20 K). Compound 2 is the first sulfur-bridged SMM. Organometallic compounds are an interesting alternative chemical repertoire for developing SMMs with unusual physics and high magnetization reversal barriers. Notable recent SMMs based on the dianionic cyclooctatetraene ligand (C8H8, COT) include: [Dy{η8-1,4-(Me3Si)2(COT)}2]−, which exhibits multiple relaxation modes;6a [(η5-Cp*)Er(η8-COT)];6b and neptunocene, [Np(η8-COT)2].6c The SMMs most closely related to 2 are [{Cp2Dy(μ-Cl)}n] (n=2, 3 a; n=∞, 3 b), [Cp2Dy(thf)(μ-Cl)]2 (4) and [{Cp2Dy(μ-Bta)}2] (5; Bta=benzotriazolate), which have lower magnetization reversal barriers of Ueff=26, 68, 34, and 33 cm−1, respectively.12 Further insight into the magnetization reversal in 2–5 was obtained with ab initio calculations,7b,7d,7g using MOLCAS 7.6.13a Local magnetic properties on individual Dy ions were computed by the Single_Aniso module,13b while the magnetic properties of the dimers were simulated by the Poly_Aniso routine (Supporting Information).13c Table 2 shows the energies of the lowest Kramers doublets of the individual Dy ions in 2–5. No. 2 3 a 3 b 4 5 1 0 0 0 0 0 2 113 144 134 96 103 3 283 313 296 193 171 4 368 383 356 211 210 5 386 401 392 236 226 6 426 444 434 286 248 7 457 488 457 308 295 8 624 672 588 339 389 The orientation of the main anisotropy axes of the Dy ions in 2–4 is close to perpendicular to the {Dy2X2} planes (Figure 4; Supporting Information, Figures S17, S20, S21). The very anisotropic g matrices in the ground doublets (Table 3) mean that the magnetic interaction between Dy ions will be mostly of the Ising type, and so can be described by H=−J ss [Eq. (1)], where s and s are projections of the effective spin s̃=1/2 of lowest Kramers doublets of the two Dy ions on the local anisotropy axes. The magnetic coupling parameter J is a sum of the exchange interaction (Jexch) and the magnetic dipole–dipole interaction (Jdipolar) between the Dy ions. Knowledge of the ground state g matrix on the Dy sites allows for exact calculations of Jdipolar (Table 4), so that the only unknown parameters are Jexch, which are derived by fitting the magnetic data (Supporting Information, Figures S5, S18, S22). The intramolecular dipolar interaction favors antiferromagnetic coupling between magnetic moments, which is explained by the local anisotropy axes being almost perpendicular to the Dy1⋅⋅⋅Dy2 axis (Figure 4; Supporting Information, Figures S21 for 2 and 4, respectively). While the dipolar couplings in 2 and 4 are similar, the exchange interaction in 2 is twice as large as that in 4 (Table 4). This is not unexpected, because the valence orbitals of sulfur are more diffuse than those of chloride. Orientation of the main anisotropy axes (dashed lines) of the ground Kramers doublets of the DyIII ions (large spheres) in 2. 2 3 a 3 b 4 5 gx 0.0012 0.0004 0.0009 0.0224 0.0073 gy 0.0019 0.0009 0.0015 0.0479 0.0884 gz 19.3611 19.4090 19.3590 18.9208 19.0530 2 4 Jexch −2.19475 0.08550 Jdipolar −2.22550 −1.99075 J=Jexch+Jdipolar −4.42025 −1.90525 The Ueff value of 133 cm−1 for 2 agrees well with the calculated energy of the first excited state on Dy sites (113 cm−1). The Ueff value of 33 cm−1 for thf-solvated 4, which was extracted in a similar way, does not agree well with the calculated energy of the first excited Kramers doublet (96 cm−1). Analysis of the out-of-phase ac susceptibility of 4 in an applied dc field of 1200 Oe reveals an additional maximum, which is not reached at ac frequencies of less than 1000 Hz and, furthermore, this maximum is not detected in zero applied dc field.12b An explanation may be found by considering the two relaxation mechanisms possible in Ln-SMMs: first, a thermally activated (Orbach) mechanism via the first excited state, and; second, a direct mechanism via quantum tunneling of the magnetization (QTM) and/or a Raman process. The rate of the direct process generally correlates with the strength of QTM, and measurements on 4 show that this SMM is in a QTM regime already at T≈5 K.12b A dc field applied to 4 suppresses the QTM and other processes, and on this basis the activated relaxation becomes apparent. In 2, the QTM and direct relaxation are much less prominent; QTM is not approached even at 2 K and, accordingly, the Orbach process involving the first excited state is measured in the ac susceptibility studies, hence the good match of the energy barriers determined by experiment and by computation. This interpretation is supported by the calculated values of transverse g values on the Dy ions (gx and gy in Table 3), which are one order of magnitude larger in 4 relative to 2. This implies more prominent QTM (and a larger rate of direct relaxation) in 4. The same reasoning, based on the “axiality” of the ground Kramers doublet on the Dy sites, explains why the relaxation properties of 5 are similar to those of 4, whereas the polymeric chains of 3 b behave in a similar manner to 2. Spurious blocking barriers, which we find here for 4 and 5, have been encountered in other Ln-SMMs. For instance, the Arrhenius fit of ac susceptibility for {Dy3} triangles gave a blocking barrier of 25 cm−1 in zero dc field, which increased to about 83 cm−1 in a dc field of 3 kOe.7k The only deviation from this model are the dimeric units of 3 a, which show relatively small transverse g factors on the Dy sites, yet at the same time their magnetization relaxes relatively quickly as in 4 and 5. The reasons for this inconsistency are currently unclear. These results illustrate the potential and the complexity of organolanthanide SMMs with soft ligands. Our calculations and most of our observations suggest that the symmetry at the individual Ln sites is crucial in determining the observed Ueff value. The transverse g values decide the quantum tunneling rate, and hence whether the complex will fall into an activated or a direct relaxation regime. In low-symmetry cases, with high g factors, it is probable that the direct process is observed. Dedicated to Professor Dieter Fenske on the occasion of his 70th birthday Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.