There is nothing wrong with being soft: using sulfur ligands to increase axiality in a Dy( III ) single-ion magnet †

A new air-stable sulfur-ligated Dy( III ) single-ion magnet has been successfully isolated with U eﬀ = 638 K and hysteresis loops open up to 7 K. In silico studies show that the S co-ligands significantly boost the axiality and that Te- and Se-donors have the potential to further enhance the magnetic properties.

point group symmetries are very promising. Our group recently reported the blueprint for generating strong uniaxial magnetic anisotropy for Dy(III) in BD 6h symmetry (hexagonal bipyramidal), by combining a strong linear axial ligand field with a weak equatorial ligand field, by using a polydentate ligand L N6 (L N6 = N6-hexagonal plane from the neutral Schiff base ligand formed from 2,6-diacetylpyridine and ethylenediamine). 11 Most of the reported 4f-SIMs incorporate traditional ligands with O-, N-, C-and halogen-donor atoms. 3 Rarer are the examples of SMMs exploring ligands with more ''exotic'' donor atoms from the main group, as was reviewed recently by Guo et al. 12 Recently we have explored how ligand electronics can tune SIM properties 13 and herein, we sought to explore how different donor groups originating from the p block affect the magnetisation dynamics. We report for the first time the synthesis, structure, magnetic characterisation and ab initio studies of [Dy III L ON3 (C 5 H 10 NS 2 ) 2 ]Á0.5THF (1) (Fig. 1) which is a S-ligated single-ion magnet with hysteresis loops open up to 7 K and a magnetisation reversal barrier of U eff = 638 K, which is unprecedented in the very small family of S-ligated Dy(III) singleion magnets (see Table 1). 14-20 Compound 1 was isolated by using the cage-like ligand N-(3,5-di-tert-butyl-2-hydroxybenzyl)-N,N-bis(2-pyridylmethyl)amine (HL ON3 ) (see ESI †). 21 Our synthetic strategy generates one short Dy-O bond, to direct the magnetic anisotropy, 22 and three longer Dy-N bonds. 23 The rest of the coordination sphere is completed with soft S-donor groups, by using diethyldithiocarbamate co-ligands, (Fig. 1) giving longer Dy-S bonds ( Fig. S1 and Table S2, ESI †). Importantly, through a detailed in silico study we also examine how O-, Se-and Te-based co-ligands affect the calculated magnetisation reversal barrier, U cal , of 1.
Complex 1 (Fig. 1) was isolated from a dry THF solution (see ESI †) with the phase purity of the bulk sample confirmed by powder X-ray diffraction (Fig. S2, ESI †). Continuous shape measures 24 analysis gives a value of 2.7 for a biaugmented trigonal prism geometry (C 2v symmetry) ( Measurements of the variable temperature alternating current (ac) susceptibility between 1-940 Hz were performed in order to investigate the dynamics of the magnetisation for 1  25 The a parameters found are in the range of 0-0.3 (2-40 K) for 1. The t À1 vs. T data were fitted using the equation t À1 = t QTM À1 + CT n + t 0 À1 exp(ÀU eff /T), in which C and n are parameters of the Raman process and t QTM is the rate of the quantum tunnelling of magnetisation (QTM). 26 The best fit gives a magnetisation reversal barrier, U eff of 638 K, t 0 = 2.99 Â 10 À12 s, n = 3.24, C = 0.02 K Àn s À1 , t Q t M = 0.017 s, under zero dc field (Fig. S17, ESI †) and U eff = 656 K, t 0 = 1.94 Â 10 À12 s, n = 3.96 and C = 3.95 Â 10 À5 K Àn s À1 under an optimum field of 1200 Oe (Fig. S18, ESI †). The observed values of the pre-factor t 0 , 10 C and n are within the commonly observed range for Dy(III) SMMs. 3 The exponent n of the Raman process has a smaller value than expected for a Kramers ion (n = 9) suggesting the presence of Raman processes involving optical acoustic phonons. 26 To the best of our knowledge, this is the largest magnetisation reversal barrier observed for a Dy(III) single-ion magnet that has S-donor ligands (see Table 1). In order to gain insight into the mechanism that governs the magnetic relaxation of 1, we have performed ab initio calculations using the CASSCF/RASSI-SO/SINGLE_ANISO approach implemented in MOLCAS 8.2 27 (see ESI †). The eight Kramers Doublets (KDs) in 1 span an energy range of 964 K (Table S4, ESI †). The ground state (m J = AE15/2) of the Dy(III) ion in 1 is highly anisotropic with nearperfect axiality (g zz = 19.859, g xx = g yy = 0.001, Table S4, ESI †). The main anisotropy axis is nearly collinear with the relatively short Dy-O bond (Fig. S19, ESI †) resulting from our synthetic strategy.  Using the CASSCF wavefunction, the computed Loprop 28 charge on the oxygen atom is found to be nearly three times larger than the nitrogen atoms of the L ON3 ligand and twice as large as the sulfur atoms of the diethyldithiocarbamate ligands (Fig. S20, ESI †). The axial nature is also observed for the first and second excited states (m J = AE13/2, g xx = 0.023, g yy = 0.028, g zz = 17.359 and m J = AE 11/2, g xx = 0.281 g yy = 0.380, g zz = 14.372, Table S4, ESI †), with the higher KDs showing relatively stronger admixtures (Fig. 3). The maximum calculated relaxation barrier, U cal , is estimated at 651 K (Fig. 3), which is in excellent agreement with the experimentally determined magnetisation reversal barrier (U eff ) of 638 K found in zero applied dc field. A relatively small transverse magnetic moment is calculated for the first three KDs (0.35 Â 10 À3 , 0.88 Â 10 À2 , 1.2 Â 10 À1 m B , respectively), which indicates relaxation via the third excited state (Fig. 3). In addition, the Orbach processes for the m J and m J + 1 excited states of opposite magnetisation between the first four KDs are found to be very small (r0.43 m B , Fig. 3).
To investigate the importance of the coordination environment and the ligand electronics on the magnetisation dynamics of 1, we have changed the co-ligand coordination environment in silico. We have created a family of three different model systems and used ab initio calculations to examine how O-, Te-and Se-based co-ligands affect the calculated magnetisation reversal barrier of 1 ( Fig. 4 and Fig. S21, S22, ESI †). Importantly, replacing the S-atoms of the diethyldithiocarbamate co-ligands with more commonly used oxygen donors (i.e. common carboxylate ligands, model 1-O, Fig. S22 upper, ESI †) gives stronger transverse components, with larger g xx /g yy values obtained for the ground and excited states (see Table S5 (ESI †) for model 1-O). Specifically, the QTM probabilities are calculated to be larger for the first three KDs of model 1-O (0.63 Â 10 À3 , 0.31 Â 10 À1 and 0.85 m B see Fig. S21 upper, ESI †) compared to 1 (Fig. 3), leading to a smaller calculated barrier of U cal = 528 K (see Fig. 4 and Fig. S21, Table S5, ESI †). These results are similar to earlier observations for Dy-O vs. Dy-S substitution 14 suggesting a likely generality of such behaviour in Dy(III) complexes.
In contrast, the g xx /g yy values obtained for model systems 1-Te and 1-Se, where the S-atoms in 1 are replaced with Te-and Se-atoms (Fig. S22 lower, ESI †), suggest that the magnetisation relaxes via the fourth KD, giving higher calculated barriers of U cal = 718 K for 1-Te and 752 K for 1-Se (see Fig. 4 and Fig. S21, Table S5, ESI †). Importantly, our results suggest that substitution of the S-atoms in 1 with O-atoms favours a stronger transverse anisotropy, while substitution with Te-and/or Se-atoms stabilises stronger axiality, with smaller transverse magnetic moments calculated for the first four KDs and smaller g xx /g yy values (see Fig. 4 and Fig. S21, Table S5, ESI †). This is in excellent agreement with a study performed on pnictogenligated compounds. 30 The ratio between B 0 2 and the corresponding non-axial crystal field parameters increases in the following order 1-O o 1 o 1-Te o 1-Se (Table S7, ESI †) in line with the increasing U cal barrier (Fig. 4).
In conclusion, [Dy III L ON3 (C 5 H 10 NS 2 ) 2 ]Á0.5THF (1) is the first S-ligated single-ion magnet with hysteresis loops open up to 7 K and a magnetisation reversal barrier of 638 K, which is significantly higher than any reported Dy(III) SIM that has S-donor ligands (Table 1). This novel complex was isolated by a carefully designed synthetic strategy that generates one short Dy-O bond, 22 which directs the magnetic anisotropy, combined with three longer Dy-N bonds, with the remainder of the coordination sphere completed with soft S-donor groups, giving longer Dy-S bonds. Furthermore, through detailed in silico studies we examine how O-, Se-and Te-based co-ligands affect the calculated magnetisation reversal barrier and magnetisation dynamics in 1, finding higher U cal values for Te-and Se-based co-ligands. We hope that this study will generate further interest in the investigation of S-ligated SIMs and prompt the study of new Te-and Se-ligated SIMs.

Conflicts of interest
There are no conflicts to declare.