Antimony-ligated dysprosium single-molecule magnets as catalysts for stibine dehydrocoupling

The synthesis of antimony-ligated dysprosium SMMs is described in addition to the unexpected reactivity of the SMMs in stibine dehydrocoupling catalysis.


Introduction
The synthesis of complex magnetic materials from simple chemical building blocks encapsulates the intrinsic fascination of molecular magnetism. Molecular magnets are typically designed using bottom-up approaches that provide access to experimental testbeds for theoretical models of magnetism, and that enable modular approaches to applications based on the properties of well-dened magnetic units. For example, carefully constructed transition metal complexes display characteristics that could lead to their implementation as molecular qubits for quantum computing. 1 The magnetocaloric effect, in which the entropy of a magnetic system is modulated by a magnetic eld, introduces the possibility of using molecular magnets as refrigerants that function more efficiently than conventional cryogens. 2 Spin-crossover materials, which have been intensively studied for many years, 3 have been proposed for applications in displays, sensors and information storage devices. 4 Lanthanide complexes continue to play important roles in enhancing our understanding of ligand eld theory, 5 and many such species nd applications in NMR spectroscopy as shi reagents 6 and in magnetic resonance imaging. 7 Single-molecule magnets (SMMs) are coordination compounds that can be dened by an effective energy barrier (U eff ) to reversal of their magnetization. 8 The pioneering work on SMMs focused on exchange-coupled transition metal cage compounds, 9,10 and monometallic 3d complexes have recently emerged as another important class of SMM. [11][12][13] Many of the most exciting developments in single-molecule magnetism have been accounted for by the lanthanides terbium, dysprosium and erbium, [14][15][16][17][18][19] and lanthanide SMMs have been described with very high U eff values and magnetic blocking temperatures. [20][21][22] Studies of the interactions between electrical currents and SMMs on surfaces has also led to the development of prototype molecular spintronic devices. 23,24 Despite the remarkable progress with SMMs, challenges remain, including overcoming the need for liquid-helium temperatures to observe slow relaxation of the magnetization, and the need to organise and stabilise molecules on surfaces for devices to become viable. To address these challenges, novel synthetic coordination chemistry strategies are of prime importance. Ligand environments in SMMs, especially those containing lanthanides, are dominated by hard, oxygen-and nitrogen-donor atoms. [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22] Targeting lanthanide SMMs with ligands in which the donor atoms have metallic character would introduce new ways of inuencing the metal-ligand bonding and hence the electronic structure of the metal ion, potentially providing a way of enhancing the magnetic relaxation properties. Furthermore, using metalloid donor ligands as building blocks in SMMs could unearth new reactivity, which could itself be manipulated further for the synthesis of new molecular magnets.

Results and discussion
We now describe two dysprosium-containing SMMs based on the metallocene building block {Cp 2 Dy(E) 2 4 Mes 3 ] 3À , respectively (Cp 0 ¼ methylcyclopentadienyl; Mes ¼ mesityl). Compound 1-Dy and the yttrium analogue 1-Y were synthesized by adding three equivalents of MesSbH 2 to a 3 : 3 mixture of Cp 0 3 M and n BuLi over 30 minutes at À50 C (Scheme 1). Compounds 2-M were synthesized in a similar fashion with four equivalents of MesSbH 2 , with the reaction being warmed from À78 C to room temperature overnight. Without careful control of reaction time and temperature, stibine dehydrocoupling occurs to give the 1,2distibane Sb 2 H 2 Mes 2 ( Fig. S1 †), the tetrastibetane Sb 4 Mes 4 , 25,26 and H 2 . This unanticipated reactivity introduced the possibility of converting 1-M into 2-M via cross-dehydrocoupling of the former with MesSbH 2 : in the case of 2-Y, the reaction is quantitative by 1 H NMR spectroscopy (Fig. S5 †), and for 2-Dy the isolated yield was 45%. The dehydrocoupling reactivity is considered further aer discussion of the structural and magnetic properties of 1-Dy and 2-Dy.

Magnetic properties
The magnetic susceptibilities of 1-Dy and 2-Dy were measured in a d.c. eld of 1 kOe. The plots of c M T(T) in the range 2-300 K for both compounds are similar (Fig. S7 †) and consistent with the presence of three Dy 3+ ions with 6 H 15/2 ground terms and g J ¼ 4/3 (theoretical c M T ¼ 42.5 cm 3 K mol À1 at 300 K). For 1-Dy, c M T is 40.57 cm 3 K mol À1 at 300 K before gradually decreasing upon cooling to 50 K; at lower temperatures the decrease in c M T is more rapid, reaching 10.01 cm 3 K mol À1 at 2 K. The values of c M T for 2-Dy at 300 K and 2 K are 42.69 cm 3 K mol À1 and 10.10 cm 3 K mol À1 , respectively. The eld (H) dependence of the magnetization (M) at 1.8 K is also similar for 1-Dy and 2-Dy, with  Comparing the experimental and calculated magnetic properties for 1-Dy and 2-Dy in the absence of intramolecular exchange interactions, it is clear that the experimental decrease in c M T at low temperatures cannot be due to ligand eld effects alone. Similarly, the increase in magnetization at low elds is slower than calculated. These observations imply non-negligible antiferromagnetic exchange interactions between the dysprosium centres, which were simulated by implementing the Lines model 29,30 and the Hamiltonian shown in eqn (1) using PHI. 31 Here, theÔ q k i operator equivalents act on the |J, m J i i basis of the 6 H 15/2 term of each Dy 3+ ion where the B q k i crystal eld terms are xed from CASSCF calculations (see below) taking into account the relative orientations of the local reference frames of each Dy 3+ ion. The only variable is the single isotropic Lines exchange constant J iso , which acts on the true S ¼ 5/2 spins of the Dy 3+ ions via a Clebsch-Gordan decoupling; we use this term to account for both the exchange and dipolar coupling. Modelling the interactions in this way, the best simulations are obtained for 1-Dy and 2-Dy using J iso ¼ À0.121 cm À1 and À0.150 cm À1 , respectively ( Fig. S7 and S8 †).
The SMM properties of 1-Dy and 2-Dy were investigated using a.c. magnetic susceptibility measurements, employing a weak a.c. eld of 1.55 Oe and zero d.c. eld. In order to explore the impact of exchange interactions on the SMM properties, we also studied the magnetically dilute analogues [(Cp 3 Sb] (Dy@2-Y). Dilution levels of 5% were achieved by combining Cp 0 3 Y and Cp 0 3 Dy in 19 : 1 ratio and performing the syntheses according to Scheme 1, which produced Dy@1-Y and Dy@2-Y in matrices of 1-Y and 2-Y, respectively. The frequency (n) dependence of the in-phase (c 0 ) ( Fig. S9 and S10 †) and the out-of-phase (c 00 ) (Fig. 2) magnetic susceptibilities reveal prominent SMM behaviour for 1-Dy and 2-Dy. The c 00 (n) plots for both systems show well-dened maxima in the temperature range 5-36 K and 4-33 K, respectively, using a.c. frequencies up to 1400 Hz. The plots of c 00 vs. c 0 for the undiluted SMMs are semi-circular in nature, and were tted using a modied Debye model with a parameters of 0.20-0.52 and 0.19-0.40 for 1-Dy and 2-Dy, respectively, indicating broad distributions of relaxation times (Fig. S11 †). The diluted systems Dy@1-Y and Dy@2-Y also show pronounced SMM behaviour, with maxima in c 00 (n) being observed up to slightly higher temperatures relative to the undiluted SMMs (Fig. 2, S12 and S13 †). The a parameters for the dilute SMMs are 0.25-0.44 and 0.03-0.43 for Dy@1-Y and Dy@2-Y, respectively (Fig. S14 †).
Insight into the relaxation dynamics of the SMMs was obtained by plotting ln s versus T À1 (Fig. 2), where s is the relaxation time. The four SMMs display similar properties, where the high-temperature regimes show a linear dependence of ln s on T À1 , indicating relaxation via Orbach and/or thermally assisted quantum tunnelling of the magnetization (TA-QTM) mechanisms. At lower temperatures, the relaxation shows a weaker temperature dependence, suggesting relaxation via a Raman process; as the experiment was conducted in zero eld, relaxation via the direct process is expected to be negligible. Notably, the relaxation dynamics do not enter a temperatureindependent regime (usually assigned to ground-state QTM) at the lowest temperatures attainable by our SQUID magnetometer. The data was modelled for each SMM using the equation s À1 ¼ s 0 À1 e ÀUeff/kBT + CT n , where s 0 and U eff are the Orbach parameters, and C and n are the Raman parameters ( Table 1).
The U eff value of 345 cm À1 for 1-Dy is one of the largest yet determined for a polymetallic SMM in zero applied eld. The highest anisotropy barriers in SMMs based on lanthanide ions with oblate electron density in the most magnetic m J statessuch as Dy 3+typically occur when strong crystal elds are applied on high-order symmetry axes. 8 Thus, the current record anisotropy barrier is 1261 cm À1 , which was determined for a D 5h -symmetric dysprosium complex with a pentagonal bipyramidal arrangement of donor atoms. 21 In light of this, a remarkable observation on 1-Dy is that a very large barrier can still be obtained when the Dy 3+ occupies a much lower symmetry environment of approximately C 2v (assuming ring whizzing of the Cp 0 ligands). The Raman exponents n are similar to those in other metallocene-based SMMs. 32 Variable-eld magnetization measurements on the SMMs revealed marked differences between the non-dilute and dilute systems. For 1-Dy, a sweep rate of 2 mT s À1 produced a narrow S-shaped hysteresis loop at 1.8 K (Fig. S15 †), whereas butteryshaped loops were observed for Dy@1-Y at 1.8-5.4 K (Fig. 2). The hysteresis properties of 2-Dy and Dy@2-Y ( Fig. 2 and S15 †) mirror those of the stibine-ligated compounds, albeit with the M(H) loops for the diluted system remaining open up to 4.0 K. The likeliest explanation for the closed hysteresis loops in the nondilute SMMs is that exchange interactions between the Dy 3+ ions provide tunnelling pathways that close upon replacement with diamagnetic Y 3+ . The precipitous drop in magnetization for the diluted SMMs around zero eld is characteristic of the vast majority of SMMs and can be attributed to single-ion effects such as hyperne coupling to spin-active isotopes of dysprosium. 33

Theoretical characterization
Deeper insight into the magnetic properties of 1-Dy and 2-Dy was obtained by performing complete active space self-consistent eld (CASSCF) calculations. 34 For both complexes, the electronic structure of the individual Dy 3+ ions is dominated by the [Cp 0 ] À ligands, which creates a strong axial potential and leads to the ground Kramers doublet at each Dy 3+ ion being well described as m J ¼ AE15/2. The main magnetic anisotropy axis of each Dy 3+ ion is therefore oriented along the local [Cp 0 ]/[Cp 0 ] direction, where all three form a teepee-like arrangement (Fig. 3). The dominant axial potential generated by the [Cp 0 ] À ligands also results in the rstand second-excited states being highly axial in nature and collinear with the ground-state axis for all sites in 1-Dy and 2-Dy (Tables 2 and S3-S8 †).
The C 2v symmetry of the dysprosium environments renders a rhombic third excited state in both complexes; this is likely to be the origin of the most efficient thermal relaxation pathway in 1-Dy since the rhombic state is calculated to lie at 416(3) cm À1 , which is comparable to the experimental U eff value of 345 cm À1 . For 2-Dy, the rhombic third excited state lies at 413(17) cm À1 , which is much larger than the experimental barrier of 270 cm À1 . Although relaxation via higher-lying Kramers' doublets is known, 35-37 it remains a relatively uncommon phenomenon, with thermally activated relaxation thought to proceed via the rst-excited doublet in most SMMs. 38 In both cases, magnetic dilution does not signicantly alter the a.c. susceptibility properties, hence the discrepancy between theory and experiment for 2-Dy cannot arise from intramolecular interactions. Despite the differing ligand environments in 1-Dy and 2-Dy, the properties of the low-lying Kramers doublets in both complexes  are remarkably similar, as are the orientations of the groundstate anisotropy axes. The LoProp charges on the antimony atoms bonded to the Dy 3+ centres range from À0.17 to À0.23 for 1-Dy and from À0.28 to À0.29 for 2-Dy, respectively. 39 Although the accumulation of charge on the donor atoms is not large in either case, the negligible difference in the average Dy-Sb bond lengths of 0.036Å between the two systems combined with the slightly greater charge density on the antimony atoms in the equatorial plane in 2-Dy relative to 1-Dy can account for the lower U eff value in the former, which is consistent with observations on related SMMs containing [MesE(H)] À and [MesE] 2À ligands (E ¼ P, As). 40,41 The U eff value determined for 1-Dy of 345 cm À1 is markedly larger than those determined for the isostructural phosphide-and arsenide-bridged analogues [(h 5 -Cp 0 2 Dy){m-E(H)Mes}] 3 (E ¼ P, As), of 210 cm À1 and 256 cm À1 , respectively. The only signicant differences in the molecular structures of 1-Dy and the two lighter congeners are the dysprosium-pnictogen bond lengths, which increase signicantly with the radius of pnictogen (those in 1-Dy are, on average, 0.168Å longer than those in the As-bridged analogue). Since the main magnetic axes in the phosphide-, arsenide-and stibinide-bridged SMMs all adopt similar orientations along the [Cp 0 ]/[Cp 0 ] directions, the pnictogens occupy equatorial sites; as the Dy-E bond lengths increase, the inuence of the pnictogen on the splitting of the Dy 3+ crystal eld levels diminishes, leading to a more dominant axial crystal eld and hence larger U eff values.
Being intrigued by the unusual [Sb 4 Mes 4 ] 3À ligand, we endeavoured to determine the electronic structure of this species. The Dy 3+ ions in 2-Dy were replaced with Lu 3+ to ensure a well-dened active space for the antimony-containing ligand, and the restricted active space (RAS) probing approach was employed with 2-Lu to identify an appropriate orbital manifold near the Fermi level to describe the Sb 4 unit. The resulting CAS, which consisted of 12 electrons in 9 orbitals for the lowest lying ten S ¼ 0 and ten S ¼ 1 states delocalized over the Sb 4 unit, is dominated by the antimony 5p orbitals (Fig. S16 †). The ground state of [Sb 4 Mes 4 ] 3À is a well-isolated S ¼ 0, as expected, however aer the rst excitation to the S ¼ 1 state at ca. 26 000 cm À1 , there is a continuum of states up to at least 45 000 cm À1 (Table S9 and Fig. S17 †). This delocalized set of continuum states is reminiscent of a semi-conductor, and it is possible that this feature also contributes to diminishing U eff in 2-Dy. Unfortunately, however, all efforts to calculate the properties of the individual Dy 3+ ions while allowing excitation into the Sb 4 continuum failed owing to the extremely large active space required for the calculation.

Stibine dehydrocoupling reactivity
In optimizing the synthesis of 1-M and 2-M, it was apparent from 1 H NMR spectroscopic studies of the yttrium derivatives that 1-Y, 2-Y, Sb 2 H 2 Mes 2 , Sb 4 Mes 4 and H 2 all form during the same reaction. Furthermore, the relative amounts of each component depend on reaction time and temperature, with longer times and higher temperatures producing greater amounts of Sb 4 Mes 4 . These observations implied that yttrium mediatesor even catalysesthe dehydrocoupling of MesSbH 2 . To investigate this possibility, 1 H NMR spectroscopy was used to study the reactions of MesSbH 2 with 10 mol% of Cp 0 3 Y at 40, 50, 60 and 70 C (Fig. S18-S21 †). The initial 1 H NMR spectrum of the 40 C reaction shows the resonances of the two starting materials (Fig. S18 †), whereas aer 20 hours MesSbH 2 , Sb 2 H 2 -Mes 2 (both diastereomers) and Sb 4 Mes 4 , account for 40%, 43% and 17% of the antimony-containing species (Fig. 4). The 1 H NMR spectrum also shows H 2 at 4.47 ppm. The amount of Sb 2 H 2 Mes 2 then gradually decreases, accounting for 20% aer 170 h, whereas the amount of Sb 4 Mes 4 increases to reach 71%, with 7% MesSbH 2 remaining. Also noteworthy in the 1 H NMR spectrum is the rapid emergence of a series of broad resonances in the region d z 6.1-6.6 ppm, which correspond to the methine CH resonances of 1-Y. Aer 170 h, all the signals due to 1-Y have been replaced by those of 2-Y. At 50 C, the conversion of MesSbH 2 to Mes 4 Sb 4 increases to 80% at a faster rate, but at higher temperatures the conversion level decreases and, at 70 C, an appreciable amount of mesitylene was observed due to decomposition of MesSbH 2 (Fig. S22 †). Thus, Cp 0 3 Y does catalyse the dehydrocoupling of MesSbH 2 to give Sb 2 H 2 Mes 2 and then Sb 4 Mes 4 .  The initial yttrium-containing product of the dehydrocoupling is 1-Y, which is subsequently converted into 2-Y. Since our stoichiometric (Scheme 1) and catalytic reaction studies have established that 1-Y reacts quantitatively with MesSbH 2 to give 2-Y ( Fig. S5 †), the fate of 2-Y once formed is of interest. This was probed by adding 3.33 mol% of 2-Y (i.e. 10 mol% yttrium) to MesSbH 2 and following the reaction by 1 H NMR spectroscopy at 40 C (Fig. S23 †). The resulting spectra acquired over 345 h reveal that, although the reaction is slower than with Cp 0 3 Y as the catalyst, 2-Y does dehydrocouple MesSbH 2 to give Sb 2 H 2 Mes 2 and H 2 , and then Sb 4 Mes 4 .
A mechanism for the catalytic dehydrocoupling of MesSbH 2 by Cp 0 3 Y is proposed in Scheme 2. The variation in the relative amounts of MesSbH 2 , Sb 2 H 2 Mes 2 and Sb 4 Mes 4 as a function of time, in addition to the formation of H 2 , suggests: (i) that the distibine is formed from dehydrocoupling of MesSbH 2 , and; (ii) that the tetrastibetane is formed subsequently from further reactivity of the distibine. The formation of Sb 2 H 2 Mes 2 also implies that the dehydrocoupling does not occur via stibinidene (i.e. RSb) elimination, which would only produce cyclic oligomers of the type [MesSb] n . Thus, we envisage deprotonation of Dehydrocoupling catalysis has emerged as one of the most important methods for the synthesis of homo-or hetero-nuclear bonds between p-block elements. 44 Considerable attention has focused on the synthesis of inorganic polymers, especially poly(ammonia-borane) and poly(amine-boranes), owing to their proposed applications as hydrogen storage and delivery materials. 45 Notably, only one example of metal-catalysed stibine dehydrocoupling has previously been reported, which employed the group 4 metallocenes [(Cp*)(Cp)M(H)Cl] (M ¼ Zr, Hf) as catalysts at 5 mol% loading for the formation of Sb 4 Mes 4 from MesSbH 2 . 46 This reaction is thought to proceed via a mechanism that involves a-elimination of highly reactive stibinidene (SbR) fragments, which subsequently cyclo-oligomerize to Sb n R n . Many catalysts based on main group metals and transition metals are well established for the dehydrocoupling of a range of element-element bonds, 42,44 however surprisingly few examples employ rare earth elements. A recent study has shown that divalent rare earth alkyl complexes are effective catalysts for the cross-dehydrocoupling of silanes and amines to give silazanes. 47 The cross-dehydrocoupling of 1-Dy with mesitylstibine to give 2-Dy is the rst example of such reactivity being used to synthesize an SMM. Our observations therefore represent a new catalytic transformation in rare-earth chemistry and a new synthetic strategy in molecular magnetism. The observation of SMM behaviour for 1-Dy, 2-Dy and their magnetically dilute analogues in light of the role of 1-Y and 2-Y in stibine dehydrocoupling is also signicant. Although the paramagnetism of the dysprosium systems precludes detailed study by NMR spectroscopy, crystalline Sb 2 H 2 Mes 2 , Sb 4 Mes 4 can be isolated from the dehydrocoupling of MesSbH 2 catalysed by 10 mol% Cp 0 3 Dy. In light of the similar chemistry of Y 3+ and Dy 3+ , 1-Dy and 2-Dy should therefore also be intermediates in the catalytic stibine dehydrocoupling. Thus, the dysprosium-antimony compounds display two functions that can be accessed by varying the temperature, since cooling 1-Dy and 2-Dy below 40 K leads to SMM behaviour, and heating them in solution above 313 K results in catalytic stibine dehydrocoupling.

Conclusions
In summary we have synthesized the rst antimony-ligated SMMs. The anisotropy barriers of 1-Dy, 2-Dy in zero applied eld, and of their diluted analogues, are U eff ¼ 345 cm À1 and 270 cm À1 , respectively, placing them amongst the highest yet reported. The conversion of 1-Dy into 2-Dy via cross dehydrocoupling with mesitylstibine represents a novel synthetic strategy in molecular magnetism. Indeed, our initial aim of targeting SMMs with lanthanide-metalloid bonds has resulted in the identication new catalytic reactivity for the rare earth elements. Given the broad scope of dehydrocoupling chemistry, the synthetic strategy has considerable potential to be extended to incorporate many new and unconventional chemical environments into molecular magnets. The next challenge is to extend the reactivity to synthesize SMMs that can be regarded as molecular alloys, i.e. systems in which the magnetic centres are bonded to the heaviest stable metallic elements. Based on the periodic trend in the anisotropy barrier unearthed during this study, i.e. that U eff tends to increase with increasing radius of the pnictogen, substantial increases in U eff can be expected for SMMs ligated by the 6p elements thallium, lead and bismuth, provided the chemical environments can be stabilized. Ongoing work in our laboratory will pursue these targets.