Slow magnetisation relaxation in tetraoxolene-bridged rare earth complexes.

Three families of tetraoxolene-bridged dinuclear rare earth (RE) complexes have been synthesised and characterised, with general formula [((HB(pz)3)2RE)2(μ-tetraoxolene)] (HB(pz)3- = hydrotris(pyrazolyl)borate; tetraoxolene = chloranilate (1-RE), the dianionic form of 2,5-dihydroxy-1,4-benzoquinone (2-RE), or its 3,6-dimethyl analogue (3-RE)). In each case, the bridging tetraoxolene ligand is in the diamagnetic dianionic form and species with selected lanthanoid(iii) ions from Eu(iii) to Yb(iii) have been obtained, as well as the diamagnetic Y(iii) analogues. Use of the 3,6-dimethyl substituted tetraoxolene ligand (Me2-dhbq2-) has also afforded the two byproducts [((HB(pz)3)(MeOH)(B(OMe)4)Y)2(μ-Me2dhbq)] (4-Y) and [{((HB(pz)3)(MeOH)Y)2(μ-B(OMe)4)}2(μ-Me2dhbq)2]Cl2 (5-Y), with the B(OMe)4- ligands arising from partial decomposition of HB(pz)3-. Electrochemical studies on the soluble 1-RE and 3-RE families indicate multiple tetraoxolene-based redox processes. Magnetochemical and EPR studies of 3-Gd indicate the negligible magnetic coupling between the two Gd(iii) centres through the diamagnetic tetraoxolene bridge. Alternating current magnetic susceptibility studies of 1-Dy and 3-Dy reveal slow magnetic relaxation, with quantum tunnelling of the magnetisation (QTM) dominant in the absence of an applied dc field. The application of a dc field suppresses the QTM and relaxation data are consistent with an Orbach relaxation mechanism playing a major role in both cases, with effective energy barriers to magnetisation reversal determined as 47 and 24 K for 1-Dy and 3-Dy, respectively. The different dynamic magnetic behaviour evident for 1-Dy and 3-Dy arises from small differences in the local Dy(iii) coordination environments, highlighting the subtle structural effects responsible for the electronic structure and resulting magnetic behaviour.


Introduction
The last few years have seen impressive advances in the field of single-molecule magnets (SMMs), with slow magnetic relaxation that arises from intrinsic molecular properties reported at unprecedentedly high temperatures. [1][2][3][4][5] Two recent spectacular examples are mononuclear Dy complexes with high axial symmetry, each with experimentally measured energy barriers to magnetisation reversal (U eff ) over 1200 cm −1 (1700 K). 1,2 A salt of the complex [(Cp ttt )Dy] + (Cp ttt = 1,2,4-tri(tertbutyl)cyclopentadienide) exhibits magnetisation hysteresis up to 60 K. 1 Thus the practical target of molecules that can act as magnets at liquid nitrogen temperatures for applications, for example in molecular spin valves and spin transistors, seems closer to reality than ever. Before such applications can be realised, it must become possible to deposit the molecules onto a surface or otherwise incorporate them into a matrix in such a way that the SMM properties are maintained and the molecules can be addressed individually, for instance via a magnetic STM tip. Advances in these areas of SMM deposition and addressing are also being vigorously pursued. [6][7][8][9][10][11] The energy barrier to magnetisation reversal for lanthanoid (Ln) SMMs arises from crystal field (CF) splitting of the ground spin-orbit coupled J state of the Ln(III) ion into microstates. The relative order, energies and m J composition of these states depend on the local symmetry and CF of the Ln(III) ion. 12 In situations where the CF splitting affords a bistable ground state, with dominant contributions from large m J values and large energy separations between the microstates, an energy barrier to magnetisation reversal gives rise to the slow magnetic relaxation that is a signature of SMMs. For lanthanoid SMMs, magnetisation relaxation typically occurs through the first few higher energy microstates. Quantum tunnelling of the magnetisation (QTM) between degenerate microstates is an efficient relaxation pathway for many Ln-SMMs, 13,14 although the application of an external field can suppress the QTM, as it removes the degeneracy of the microstates involved. 15 Dysprosium(III) features most prominently in Ln-SMMs due to its potentially high magnetic anisotropy and Kramers nature that affords a doubly degenerate ground state. [16][17][18][19][20] Efforts to improve SMM properties by incorporating more than one lanthanoid(III) ion in polynuclear complexes have also been fruitful. 16,[21][22][23] The intrinsically small exchange coupling between Ln(III) ions can often mean that the different ions act as discrete non-interacting magnetic units within polynuclear complexes. Nevertheless, several examples demonstrate how the ground state magnetic moment and/or magnetic anisotropy, and therefore U eff , can be enhanced by linking lanthanoid(III) ions through bridging ligands in dinuclear or polynuclear complexes. [24][25][26][27] As a further advantage, the weak exchange interactions in these systems provides an effective small bias field, which can result in suppression of the QTM in zero external applied field that can prohibit bistability. These examples generally rely on the specific coordination mode of the ligand bridge enforcing ferromagnetic coupling or alignment of the individual local anisotropy axes.

Synthesis
The rare earth (RE) complexes [((HB( pz) 3 ) 2 RE) 2 (μ-ca)] (1-RE) with RE = Y, Eu, Gd, Tb, Dy, Ho, Er and Yb were synthesised following modification of the literature procedure to enhance the purity of the product. 36 Four equivalents of KHB( pz) 3 were added to a stirred ethanol/dichloromethane solution containing two equivalents of rare earth salt. Slightly less than one equivalent of the caH 2 ligand was doubly deprotonated with Et 3 N and added dropwise to an unstirred suspension of the rare earth mixture and swirled gently, causing dissolution of the suspension. A fine white solid precipitated, which was removed by filtration and the purple filtrate left to stand, affording a crystalline sample of 1-RE. The elemental analysis reported previously for these compounds indicated that the bulk samples are often impure, 36 which we have been able to overcome by multiple recrystallisations from dichloromethane/ n-hexane to access analytically pure samples of 1-RE·2CH 2 Cl 2 in 20-50% yield. A single crystal of another solvate, 1-Y·2Me 2 CO, was also obtained by recrystallisation from acetone/ n-hexane. The complexes [((HB( pz) 3 ) 2 RE) 2 (μ-dhbq)] (2-RE) with RE = Y, Eu, Gd, Tb, Dy, Ho, Er and Yb were synthesised as per 1-RE, however the pink microcrystalline products are highly insoluble and cannot be purified by recrystallisation. The crude yields of these complexes were in the range 58-76%. The purple complexes [((HB( pz) 3 ) 2 RE) 2 (μ-Me 2 dhbq)] (3-RE) with RE = Y, Eu, Gd, Tb, Dy, Ho, Er and Yb were synthesised in a manner analogous to that employed for 1-RE and obtained in generally slightly higher yields of 43-49%. Of the three families of compounds, 3-RE is the easiest to purify, with a single recrystallisation from dichloromethane/n-hexane affording analytically pure products of 3-RE·xCH 2 Cl 2 (x = 0-2). Both the 1-RE and 3-RE dinuclear complexes undergo some decomposition in alcoholic solutions and as such the crude products were rapidly isolated from the ethanol-containing reaction solution prior to recrystallisation. Powder X-ray diffraction data (  4-Y in the refrigerator, and 5-Y in the freezer. Large wellformed crystals of compounds containing these complexes were manually separated from 3-Y for X-ray structural analysis, but it has not been possible to obtain pure bulk samples for further study. Both 4-Y and 5-Y incorporate B(OMe) 4 − ligands that form in situ from reaction of the HB( pz) 3 − with methanol.
Compound 4-Y·3.5MeOH crystallises in the triclinic space group P1 with half a dinuclear complex per asymmetric unit. The dinuclear structure of neutral 4-Y (Fig. 2) is related to that of the 3-RE family, with one HB( pz) 3 − ligand per metal centre replaced by a bidentate B(OMe) 4 − ligand and the eighth yttrium coordination site occupied by a monodentate MeOH ligand. Compound 5-Y·5MeOH crystallises in the monoclinic P2 1 /c space group with half a tetranuclear complex per asymmetric unit. Complex 5-Y is related to 4-Y and can be con- 2.504 ( (2) a In brackets are atom-labelling for Y2 and tetraoxolene in 3-Y. b Shortest distance.

Dalton Transactions Paper
This  (Fig. S5 †). The tetranuclear complex in 5-Y is overall dicationic, with charge balance provided by chloride counterions. The yttrium coordination geometry is best described as triangular dodecahedral in both 4-Y and 5-Y (Table S1 †) and the bond lengths are in the ranges evident for the other complexes (Table S2 †). The B(OMe) 4 − ligand has been observed in complexes and coordination polymers previously, binding in the bidentate or bisbidentate modes observed in 4-Y and 5-Y. 39,45 The structure of complex 5-Y can be described as a supramolecular square or grid, which are much less common for rare earth metals than for transition metals. 46

Infrared spectroscopy
The infrared spectra of 1-RE·2CH 2 Cl 2 , 2-RE and 3-RE·xCH 2 Cl 2 were measured as pressed KBr disks ( Fig. S6 and S7 †). The spectra of the different rare earth analogues of the three structural families vary little, consistent with the isostructural nature of the complexes. All spectra exhibit strong vibrational modes attributed mostly to stretches associated with the HB(pz) 3 − ligand. A single B-H stretch of the HB(pz) 3 − is seen at ∼2450 cm −1 , confirming the presence of the coordinated tripodal ligand. Bands attributed to modes involving the tetra-oxolene ligands include the band at 1540 cm −1 in 2-Y which shifts to lower energy with increasing electron donation from the substituents to 1527 cm −1 in 3-Y, which is attributed to a ν CO stretch, as well as the bands at 1404 and 1437 cm −1 .

Electronic spectroscopy
Diffuse reflectance UV-visible spectra were measured for solid samples of all compounds 1-RE·2CH 2 Cl 2 , 2-RE and 3-RE·xCH 2 Cl 2 dispersed in KBr, while solution absorbance spectra were also acquired for the soluble 1-RE and 3-RE in acetonitrile, as well as for the deprotonated tetraoxolene ligands (Fig. S8-S10 †). The solution spectra for 1-RE and 3-RE remain unchanged over at least 24 hours, consistent with stability of the complexes in acetonitrile and the solution and solid state spectra are in good agreement. The three major bands in the diffuse reflectance spectra of 1-RE, 2-RE and 3-RE are attributed to ligand-centred π-π* transitions for the bands at ∼220 nm and ∼350 nm and to a ligand-centred n-π* transition for the broad band with a low extinction coefficient over the visible range (Fig. S9 †). The bands are assigned to transitions based on the tetraoxolene ligands by comparison with the solution spectra of the deprotonated tetraoxolene ligands, as well as similar literature complexes. 36,47 The spectra for the different members of the 1-RE, 2-RE, and 3-RE families with different rare earth metals are very similar (Fig. S9 †), with little variation with the metal. There are no obvious charge-transfer bands involving the rare earth metal and redox-active tetraoxolene ligand. 48 For the three Ho 3+ analogues, additional sharp bands corresponding to f-f transitions in the Ho 3+ ion are evident in the diffuse reflectance spectra (Fig. S10 †). The three features that are observed are assigned as 5 I 8 → 5 G 5 (418 nm), 5 I 8 → 5 G 6 , 5 F 1 (452 nm) and 5 I 8 → 5 F 4 (538 nm) transitions. 49

Electrochemistry
Following confirmation of solution stability of 1-RE and 3-RE by electronic absorption and 1 H NMR spectroscopy (for the Y analogues), the voltammetric behaviour of acetonitrile solutions of 1-RE (RE = Y, Eu, Tb, Dy, Yb) and 3-Y with 0.1 M Bu 4 NPF 6 as the supporting electrolyte were examined using a glassy carbon working electrode (Fig. 3 and S11 †). The measured oxidation (E pa ) and reduction (E pc ) peak potentials (E p ) from the cyclic voltammograms (100 mV s −1 scan rate) are tabulated in Table 3, together with the mid-point potentials (E m ) and peakto-peak separations (ΔE p ) for the quasi-reversible processes. All potentials are quoted versus the ferrocene/ferrocenium (Fc/Fc + ) couple.
The 1-RE complexes measured all exhibit two reductions ( processes I and II) and an oxidation ( process III) at similar potentials (Table 3, Fig. 3 and S11 †), which are all assigned as tetraoxolene ligand-based processes (Scheme 1), consistent with similar literature complexes. 47 For both reductions I and II in 1-Y, the E m values are independent of scan rate at −0.90 V and −1.73 V, respectively. The current ratios i pc /i pa for both processes are close to unity, where i pa and i pc are the peak anodic and peak cathodic currents respectively. A plot of the cathodic peak current versus the square root of the scan rate is linear for process I only, following the Randles-Sevcik model for reversible systems. 50 For an ideal reversible one-electron process, ΔE p should be 59 mV and for process I the ΔE p of 75 mV at a scan rate of 100 mV s −1 is the same as that measured for the Fc/Fc + couple under these conditions, consistent with the assignment of I as a diffusion controlled and reversible one-electron process. Process II has a ΔE p of 95 mV and the plot of the cathodic peak current versus the square root of the scan rate is non-linear, consistent with assignment of process II as quasi-reversible. The oxidation process III shows no current response on the reverse sweep and is clearly irreversible. It is tentatively assigned to a one-electron oxidation to the monoanionic form of the tetraoxolene ligand, although steady-state voltammetry or coulometry would be required to confirm this assignment. The voltammograms measured for the other analogues of 1-RE are very similar to that for 1-Y, with no additional metalcentred redox processes observed for the commonly redoxactive Eu(III), Tb(III) and Yb(III) ions; however, the potentials of the lanthanoid complexes are shifted around 100 mV more positive than for 1-Y. The complexes 1-Eu and 1-Yb seem to be less stable in solution than the other 1-RE complexes, and the reductive processes exhibit decreased reversibility than the other analogues (Fig. S11 †).
The voltammograms measured for 3-Y are qualitatively similar to those for 1-Y (Fig. 3, Table 3; Scheme 1), with all processes shifted more negative by around 200-300 mV. This is consistent with the electron donating/withdrawing characteristics of the substituents on the bridging tetraoxolene ligands and the more difficult reduction of the Me 2 dhbq 2− ligand versus the ca 2− analogue. The cyclic voltammogram of 3-Y shows one diffusion-controlled reversible reduction at E m = −1.22 V (I), which fulfils the same requirements of reversibility as those of the first reduction process in 1-Y. The second reduction process (II) for 3-Y at E pc = −2.11 V exhibits a smaller anodic current on the reverse sweep than the equivalent process in 1-Y, consistent with less chemical reversibility or slower electron transfer kinetics.

Efforts to access a radical-bridged analogue
The ultimate target of this work at the outset was an analogue of compounds 1-RE, 2-RE or 3-RE that incorporates the bridging tetraxolene in a radical oxidation state, to explore the SMM properties. From the voltammetric studies the most promising approach would appear to be through one-electron reduction of the 1-RE family to access the tetraoxolene in the radical trianionic form, as the first reduction potential ( process I) is more positive for the 1-RE family than for the 3-RE family and the 2-RE family is insoluble. Oxidation to a radical monoanionic form might also be possible, although the irreversibility of the first oxidation process (III) in the voltammetry of 1-RE and 3-RE suggests that the oxidised form might be unstable. To our knowledge the radical monoanionic redox state of the ligand has never been reported in a metal complex. We attempted reduction of 1-RE (RE = Y, Gd, Tb and Dy) using one equivalent of cobaltocene in rigorously anaerobic and anhydrous conditions. An immediate colour change from purple to green occurred, suggesting that the desired reaction had proceeded. However, the reduced compounds are unstable and all our efforts to isolate and purify the target radical-bridged dinuclear complex have been unsuccessful so far, with efforts to recrystallise the obtained solid instead affording crystals of a mononuclear decomposition product.

Static magnetic properties and EPR spectroscopy
Magnetic susceptibility data were acquired for analytically pure bulk samples of the easier to purify Me 2 dbhq 2− compounds 3-Gd·2CH 2 Cl 2 , 3-Tb·0.7CH 2 Cl 2 and 3-Dy·1.1CH 2 Cl 2 as well as the ca 2− analogue 1-Dy·2CH 2 Cl 2 . The Tb(III) and Dy(III) complexes were of interest due to their potential SMM behaviour and the Gd(III) analogue provides a useful spin-only comparison for evaluation of any exchange interaction. The magnetic properties of the other 1-RE and 3-RE analogues were not measured and bulk samples of 2-RE of sufficient purity for magnetic analysis could not be obtained. Variable temperature magnetic susceptibility measurements are shown in Fig. 4 and magnetisation versus field data are available in Fig. S12  The thermal behaviour of the Gd(III) derivativefor which the CF effects do not mask the exchange ones due to the absence of orbital degeneracyclearly indicates that the two Gd(III) centres are non-interacting: the χ M T value is constant throughout the investigated temperature range, following Curie-Weiss behaviour (Fig. S14 †) with respective Curie and Weiss constants of C = 15.2 cm −3 mol K −1 and θ = 0.15 K. The powder X-band (ν ∼ 9.4 GHz) EPR spectrum measured at 10 K (Fig. S15 †) confirms this interpretation. Preliminary analysis of the EPR spectrum on the basis of the spin Hamiltonian: where D and E are the axial and rhombic zero-field splitting parameters, respectively, indicates 0.1 cm −1 < |D| < 0.15 cm −1 and E/D ∼ 0.1, 51 which is in the same order of magnitude previously reported for related [Gd(HB( pz) 3 ) 2 (L)] (L = tropolonate and 3,5-di-tert-butyl-seminquinonate) complexes. 52 This picture is further confirmed by the measured isothermal fielddependent magnetisation curves (Fig. S13 †), which are consistent with two non-interacting S = 7/2 spins with a small zero field splitting. These data were fit using the computer program PHI, 53 constraining E = 0 to reduce overparameterisation, providing the following best fit parameters: |D| = 0.09 ± 0.02 cm −1 , which is very close to the value estimated from EPR, and g = 1.94 ± 0.01. The somewhat low g value compared to the expected value of 1.99 probably reflects a 2-3% error in sample weighing and can be regarded as physically irrelevant. 54 Having established the absence of a measurable exchange interaction for 3-Gd·2CH 2 Cl 2 , the temperature dependence of χ M T observed for the derivatives containing anisotropic lanthanoid ions can be unequivocally attributed to the progressive depopulation, upon decreasing temperature, of the excited levels arising from the CF splitting of the ground J multiplets of Dy(III) and Tb(III). In the case of 1-Dy and 3-Dy this results in relatively well isolated ground doublets, as witnessed by the saturation of M vs. H curves measured at low temperature (Fig. S12 †). The non-Kramers nature of the Tb(III) ion, on the other hand, affords some non-negligible field induced interaction among the lowest lying levels, which is evident from the linear increase of the magnetisation versus field curves at higher field.
No EPR spectrum was observed for the two Dy(III) derivatives, possibly due to fast relaxation or a low intra-doublet transition probability because of the ground-state composition. A broad, uninformative band at zero field, quite common for Tb(III) molecular systems, 55 is evident for 3-Tb (Fig. S15 †).

Dynamic magnetic properties
Alternating current (ac) magnetic susceptibility data were measured for 3-Tb·0.7CH 2 Cl 2 , 3-Dy·1.1CH 2 Cl 2 and 1-Dy·2CH 2 Cl 2 to probe for SMM behaviour in the temperature range 2-14 K and frequency (ν) range 1-10 000 Hz. No frequency-dependence of the magnetic susceptibility was observed for 3-Tb·0.7CH 2 Cl 2 either without an applied dc field or with a field up to 3 kOe, which is likely due to fast QTM or the lack of a doubly degenerate ground state for this non-Kramers ion, given the relatively low symmetry crystal field. In contrast, the two Dy(III) derivatives exhibit rich dynamic behaviour, which is both field-and temperature-dependent. At low temperature and zero applied dc field, a relatively fast process is evident, occurring at ν > 10 000 Hz for 1-Dy·2CH 2 Cl 2 at 4 K and at ν ∼ 1000 Hz for 3-Dy·1.1CH 2 Cl 2 at 2 K (Fig. S16 †). For 1-Dy·2CH 2 Cl 2 , the zero-field relaxation was simply too fast to characterise the behaviour in the range of frequencies available on our instrument. However, variable temperature measurements for slower-relaxing 3-Dy·1.1CH 2 Cl 2 (Fig. 5 left) clearly show a temperature-independent peak in χ″ M in the temperature range 1.9-4.5 K due to QTM, which is also apparent from the Arrhenius plot of the temperature dependence of the relaxation times (Fig. S17 †).
For both derivatives, the fast relaxation channel is suppressed by the application of a dc field (Fig. 5 centre and  right), which triggers the onset of a much slower process, reaching the slowest rate for 1-Dy·2CH 2 Cl 2 upon application of a 1.6 kOe dc field (Fig. S16 †). This behaviour is quite common in lanthanoid-based SMMs and can be attributed to the competing effects of two different relaxation processes: the faster one, dominant in zero field, is directly related to relaxation of the magnetisation via QTM, induced by dipolar and hyperfine interactions; the slower process is linked to thermally-activated relaxation (direct, Raman or Orbach). 20,[56][57][58] Since determination of the optimum dc field was not possible for 3-Dy·1.1CH 2 Cl 2 , as the in-field relaxation is too slow at low temperature, we performed the study of the temperature dependence of the relaxation rate of the magnetisation for both complexes with an applied field of 1.6 kOe. Quantitative determination of the relaxation time τ was obtained by fitting the χ″ M versus ν curves with the generalised Debye equation, 59 and the corresponding temperature dependences are shown as Arrhenius plots for the two compounds (Fig. 6). It is evident that while the relaxation of 3-Dy·1.1CH 2 Cl 2 is linear, this is not the case for 1-Dy·2CH 2 Cl 2 . Accordingly, data for 3-Dy·1.1CH 2 Cl 2 could be fit by assuming a simple Orbach process with τ 0 = 1.0 × 10 −5 s and U eff = 47 K, while a fit of the temperature dependence of 1-Dy·2CH 2 Cl 2 required testing with different combinations of relaxation processes according to: 14 where the first term represents the rate of QTM, the second is the rate of the Orbach process, the third is the Raman process ( parameterised by C and n) and the fourth term is the rate of the direct process ( parameterised by A). The best fit was obtained by assuming τ 0 = 3 × 10 −5 s, U eff = 24 K, C = 2.7 × 10 −3 s K −n and fixing n = 7. No reasonable fit could be obtained by including either QTM and/or direct relaxation, possibly because these are minimised at the "optimum" dc field employed. We note that while the value of the Raman exponent n is as expected for the Kramers Dy(III) ions, for both derivatives τ 0 is at the higher end of expected values. This suggests that the obtained parameters should be considered as purely phenomenological, in the absence of further theoretical and/or experimental investigations to confirm the presence of a real state at the 47/24 K necessary to promote Orbach relaxation. [60][61][62] Ideally a quantitative study of the mechanisms involved in the magnetic relaxation would require investigation of the samples diluted with a diamagnetic rare earth ion, which is beyond the scope of the present study. 63

Conclusions
Variation of the substituents on the 3-and 6-positions of the bridging tetraoxolene ligand has afforded three families of neutral dinuclear rare earth complexes with hydrotris( pyrazolyl)borate blocking ligands. Two structurally interesting byproducts that also feature bridging tetraoxolene ligands were also obtained, following partial decomposition of the hydrotris( pyrazolyl)borate ligands in methanol to afford tetramethoxyborate   6 Arrhenius plots of relaxation rates of 1-Dy·2CH 2 Cl 2 (squares) and 3-Dy·1.1CH 2 Cl 2 (circles) in a 1.6 kOe dc magnetic field with fits (lines) as described in the text. ligands. For all complexes, the tetraoxolene ligands are in the diamagnetic dianionic form. Solution electrochemical studies of dinuclear complexes confirm the redox-activity of the tetraoxolene ligands, with the electrochemical reversibility of a tetraoxolene-based one-electron reduction suggesting the possibility of accessing an analogue with a trianionic radical bridge following chemical reduction. However, isolation of the one-electron reduced analogue has proved elusive to date. Magnetochemical and EPR characterisation of one of the dinuclear Gd complexes indicate negligible magnetic coupling between the two Gd(III) ions, either as exchange coupling through the diamagnetic tetraoxolene bridging ligand, or from through-space dipolar coupling. Two Dy(III) analogues both exhibit slow magnetic relaxation, with QTM dominant during ac magnetic susceptibility measurements performed in the absence of an applied dc field. The application of a dc field effectively suppresses the QTM and relaxation data are consistent with an Orbach mechanism playing a major role in relaxation for both compounds. The effective energy barriers to magnetisation reversal are determined as 47 and 24 K for the analogues with the dichloro-and dimethyl-substituted tetraoxolene ligands, respectively. The molecular structures for the two Dy(III) complexes are similar, but the local Dy(III) coordination environments are distinctly different, dodecahedral for 1-Dy·2CH 2 Cl 2 and square antiprismatic for 3-Dy·1.1CH 2 Cl 2 . These structural differences are reflected in the different dynamic magnetic behaviour measured for the two compounds, highlighting once again the number of subtle parameters which might influence the electronic structure and resulting low temperature magnetisation dynamics in lanthanoid complexes.

Dalton Transactions Paper
That slow magnetic relaxation is evident in the present Dy(III) complexes, accompanied by significant zero-field QTM, is promising for future radical-bridged analogues, in which the zero-field QTM may be suppressed by metal-ligand exchange. Regardless of whether the exchange coupling between the Dy(III) ion and the radical ligand is ferro-or antiferromagnetic, radical-bridged dinuclear complexes could well exhibit enhanced SMM properties. Ongoing work in our lab is focused on fine-tuning the chemical properties of the bridging and blocking ligands to facilitate access to related radicalbridged dinuclear lanthanoid(III) complexes.

Synthesis
All manipulations were performed under aerobic conditions and all chemicals purchased were of reagent grade or higher and used as received. The Me 2 dhbqH 2 and KHB( pz) 3 proligands were prepared according to literature procedures. 64,65 [((HB( pz) 3 ) 2 Y) 2 (μ-ca)] (1-Y). A modified literature procedure was followed. 66 A suspension of KHB( pz) 3 (97.1 mg, 0.385 mmol) in CH 2 Cl 2 /EtOH (6 mL) was added to a solution of Y(NO 3 ) 3 ·6H 2 O (73.5 mg, 0.192 mmol) in EtOH (1 mL), CH 2 Cl 2 (10 mL) and the resultant colourless suspension stirred for 10 minutes. A red-purple solution of ca 2− was prepared from addition of Et 3 N (26.8 μL, 0.192 mmol) to caH 2 (19.8 mg, 0.095 mmol) in CH 2 Cl 2 (10 mL). The deprotonated ligand was added dropwise to the solution of Y(HB( pz) 3 ) 2 (NO 3 ) resulting in a deep violet solution with some fine precipitate, which was removed by filtration. The solution volume was decreased, and a microcrystalline powder collected. The product was recrystallized from a CH 2 Cl 2 /n-hexane mixture and then vacuum-filtered, washed with n-hexane and air-dried, yielding purple microcrystals of 1-Y·2CH 2 Cl 2 (53 mg, 39%). Purple plates of 1-Y suitable for X-ray diffraction were also obtained from slow evaporation of a solution of 1-Y in acetone/ n-hexane, maintained in contact with mother liquor to prevent desolvation and identified crystallographically as 1-Y·2Me 2 CO. 1 H NMR (400 MHz, CDCl 3 ) δ, ppm: 6.00 (t, 12H), 7.01 (d, 12H), measured as KBr disc on a Bruker Tensor 27 FTIR spectrometer and normalised as absorbance spectra. X-band (ν = 9.41 GHz) spectroscopic studies on the microcrystalline powder samples were carried out at low temperatures using a Bruker E500 spectrometer equipped with an ESR900 (Oxford Instruments) continuous-flow 4 He cryostat. All 1 H NMR spectra were acquired on a Varian MR400 400 MHz spectrometer and referenced to residual protic solvent.

Magnetic measurements
All magnetic measurements were performed on powder samples pressed into pellets with PTFE tape to avoid preferential orientation of crystallites induced by magnetic torque. The dc susceptibility and magnetisation measurements were performed on a Quantum Design SQUID magnetometer, with dc susceptibility measurements measured between 1.8 and 300 K. Measurements were corrected for the diamagnetic contribution of the PTFE tape. The ac susceptibility measurements were measured on a Quantum Design PPMS magnetometer equipped with the ac measurement system (ACMS) option.

Electrochemical measurements
Cyclic voltammetry was performed using a standard three-electrode cell configuration under a N 2 atmosphere, using a 3 mm diameter glassy-carbon working electrode, a Pt-wire auxiliary electrode and a leak-free Ag/AgCl reference electrode calibrated against the ferrocene/ferrocenium (Fc/Fc + ) couple. All electrochemical measurements were undertaken with 1 or 0.5 mM analyte in MeCN with 0.1 M Bu 4 NPF 6 as the supporting electrolyte.

Other measurements
Elemental analyses (CHN) were performed at the Campbell Microanalytical Laboratory, University of Otago. Thermogravimetric analyses were performed on a Mettler Toledo thermal analyser using a ramp rate of 5°C per minute up to a maximum temperature of 400°C and a N 2 atmosphere.

Conflicts of interest
There are no conflicts of interest to declare.