Reduction of Rare‐Earth Stannole Sandwich Complexes to Tin‐Based Radical Ligands and Tin–Tin Bonds

Abstract

f‐Element organometallic chemistry is dominated by cyclopentadienyl ligands. In contrast, isoelectronic metallole ligands with the general formula [EC₄R₄]²⁻, where E is a heavier group 14 element, are rare in the f‐block, particularly stannole ligands. Here, we describe the synthesis of the dimetallic stannole complexes [(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)]₂ (1 _M ; M = Y, Gd, Dy; Cp^Sn = [SnC₄‐2,5‐(SiMe₃)₂–3,4‐Me₂]²⁻, Cp^ttt = [1,2,4‐C₅ ^t Bu₃H₂]⁻), which form by virtue of Sn→M dative bonds. One‐electron reduction of 1_M with KC₈/2.2.2‐cryptand produces the mono‐anionic complexes [{(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)}₂]⁻ (2_M ), and two‐electron reduction gives di‐anionic [{(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)}₂]²⁻ (3_M ) as [K(2.2.2‐crypt)]⁺ salts. Studies of the stannole complexes using crystallography, UV/vis and EPR spectroscopy, magnetometry and computational methods reveal that the reduction steps generate tin–tin bonds through population of a delocalized molecular orbital that spans the {M₂Sn₂} rings, with attendant dearomatization of the stannole rings. Complexes 2_M are the first tin‐radical ligands bound to rare earth elements. Spin density calculations of 2_Y and 2_Gd reveal significant build‐up of unpaired spin on the tin atoms, with magnetic measurements on 2_Gd yielding an unprecedentedly large tin–gadolinium exchange coupling constant of −112 cm⁻¹ (−2J formalism).

One‐electron reduction of dimetallic rare‐earth stannole complexes with KC₈ produces a di‐tin radical ligand that shows strong coupling between tin and yttrium, gadolinium or dysprosium. Two‐electron reduction of the stannole complexes further populates a delocalized {M₂Sn₂} molecular orbital, enhancing the tin–tin bonding and simultaneously dearomatizing the stannole rings.

Introduction

The synthesis and isolation of persistent main group radicals has emerged as a topic of considerable interest in recent years, particularly in the case of heavier p‐block metals and metalloids.^{[ 1} , ^{2 ]} Impressive progress has been made by stabilizing neutral, anionic, and cationic radicals of heavier main group elements using kinetic (steric) and/or electronic (orbital‐based) approaches, notably with carbene ligands such as cyclic alkyl–amino carbenes bound to the formal radical centre.^{[ 3 ]} Whilst motivation for further research into heavier main group radicals stems from their reactivity, another application of these species lies in molecular magnetism, where they can be deployed as ligands to interact with the unpaired electrons of, for example, lanthanides. Indeed, the use of radical ligands based on light main group elements (carbon, nitrogen, and oxygen) in lanthanide single‐molecule magnets (SMMs) has proven to be an effective strategy for targeting “hard” magnet behaviour.^{[ 4} , ^{5 ]} In some radical‐bridged multimetallic SMMs, strong lanthanide‐radical exchange coupling mitigates the low‐temperature rapid relaxation of magnetization that would otherwise occur,^{[ 6} , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ^{15 ]} forming the basis of proposals to use such molecular magnets in quantum technologies.^{[ 16 ]}

In contrast to radicals composed of lighter main group elements, the use of heavier p‐block radicals in lanthanide molecular magnets is rare. To the best of our knowledge, this chemistry is limited to a few lanthanide metallocenes bound to a phosphorus‐centred heterocyclic radical,^{[ 17 ]} dimetallic lanthanide metallocenes containing the exotic [Bi₂]³⁻ radical,^{[ 18 ]} and samarocene and ytterbocene complexes of an acyclic sulfur–nitrogen radical.^{[ 19 ]} Beyond magnetism, the diffuse orbitals of heavier p‐block elements also introduce the possibility of greater metal‐ligand covalency, a topic of fundamental importance in lanthanide chemistry.^{[ 20 ]}

Our interests in lanthanide complexes of heavy p‐block radicals was recently piqued by the discovery that one‐electron reduction of the dimetallic germole‐ligated sandwich complexes [(η⁵‐Cp^Ge)M(η⁵‐Cp^ttt)]₂, where Cp^Ge is the germole dianion [GeC₄‐2,5‐(SiMe₃)₂–3,4‐Me₂]²⁻, Cp^ttt is [1,2,4‐C₅ ^t Bu₃H₂]⁻, and M is yttrium, gadolinium, or dysprosium, produces the corresponding mono‐anionic complexes [{(η⁵‐Cp^Ge)M(η⁵‐Cp^ttt)}₂]⁻.^{[ 21 ]} Analysis of these germanium radical‐bridged species revealed that the unpaired electron occupies a highly delocalized molecular orbital (MO) spanning the {M₂Ge₂} core, including an unexpected germanium–germanium bond. Magnetic susceptibility measurements of [{(η⁵‐Cp^Ge)Gd(η⁵‐Cp^ttt)}₂]⁻ yielded a gadolinium‐radical exchange coupling constant of −95 cm⁻¹ (−2J formalism), i.e., roughly two orders of magnitude larger than typically found in gadolinium complexes of light main group radical ligands.^{[ 4} , ^{5 ]}

Having observed that further reduction of the radical‐bridged mono‐anion [{(η⁵‐Cp^Ge)M(η⁵‐Cp^ttt)}₂]⁻ to the di‐anion [{(η⁵‐Cp^Ge)M(η⁵‐Cp^ttt)}₂]²⁻ enhances the M─Ge and Ge─Ge bonding, we were interested to explore the analogous stannole chemistry. Whilst the use of group 14 metallole ligands in f‐element organometallic chemistry is an emerging area in general, the use of stannole ligands is currently limited to the erbium sandwich SMM [(η⁵‐Cp^Sn)Er(η⁸‐COT)]⁻, where Cp^Sn is [SnC₄‐2,5‐(SiMe₃)₂–3,4‐Me₂]²⁻ and COT is cyclo‐octatetraenyl.^{[ 22 ]} The greater size of tin relative to germanium and the attendant weaker bonding involving the heavier group 14 metal could mean that complexes of the type [{(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)}₂] ^{n −} (n = 0, 1, 2) show divergent structural and bonding properties, as well as providing a way of tuning the magnetism of paramagnetic versions via the group 14 element. The synthesis of mono‐anionic [{(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)}₂]⁻ would also provide the first example of a tin‐based radical ligand in lanthanide chemistry.

Results

Synthesis and Structural Studies

The synthesis of the dimetallic stannole complexes [(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)]₂ (1_M ) with M = Y, Gd, Dy was accomplished according to Scheme 1. The choice of yttrium was determined by its diamagnetic nature, facilitating characterization by NMR spectroscopy. Gadolinium was chosen to explore the magnetic exchange coupling using an isotropic 4f⁷ species, and dysprosium was selected for the potential SMM properties. The isolated yields of crystalline 1_M were typically in the region of 50%–55%. Subsequently, reduction of 1_M using one stoichiometric equivalent of KC₈/2.2.2‐cryptand gave the 1:1 salts [K(2.2.2‐crypt)][{(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)}₂] ([K(2.2.2‐crypt)][2_M ]) in isolated yields of 60%–65%. Adding two equivalents of KC₈/2.2.2‐cryptand to 1_M produced the 2:1 salts [K(2.2.2‐crypt)]₂[{(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)}₂] ([K(2.2.2‐crypt)]₂[3_M ]) in 50%–60% yields.

Scheme 1.

Synthesis of 1 _M , [K(2.2.2‐crypt)][2 _M ] and [K(2.2.2‐crypt)]₂[3 _M ] (M = Y, Gd, Dy).

The molecular structures of all nine compounds were determined by X‐ray crystallography (Tables S1–S6, Figure 1, S1–S3).^{[ 23 ]} The three 1 _M compounds are isostructural, consistent with their similar FTIR spectra, as are the singly reduced compounds [K(2.2.2‐crypt)][2 _M ] and doubly reduced [K(2.2.2‐crypt)]₂[3 _M ] (Figures S4–S6). The structures of the gadolinium versions are discussed in detail, with details of the yttrium and dysprosium analogues provided in the Supporting Information.

Figure 1.

Molecular structures of 1_Gd (top), the mono‐anion 2_Gd (centre), and the di‐anion 3_Gd (bottom). Thermal ellipsoids are at the 50% probability level. For clarity, hydrogens atoms are not shown.

In the series 1_Gd , 2_Gd , and 3_Gd , stepwise reduction results in a pronounced decrease of the Gd–(Cp^Sn)_cent distances from 2.3821(11)/2.3911(11) Å in 1_Gd to 2.320(3)/2.300(3) Å in 2_Gd and then to 2.253(4) Å in 3_Gd (cent is the centroid of the ligand). Simultaneously, the Gd–(Cp^ttt)_cent distances increase in the order 2.4273(13)/2.4247(15) Å in 1_Gd to 2.485(3)/2.465(3) Å in 2_Gd , and 2.514(4) Å in 3_Gd . Substantial shortening of the Gd–Sn distances to the η⁵‐Cp^Sn ligands occurs in the order 3.1826(6)/3.1882(5) Å, 3.1480(7)/3.1232(8) Å, and 3.1028 (7) Å for 1_Gd , 2_Gd , and 3_Gd , respectively. In contrast, only a slight variation occurs in the dative Gd–Sn interactions to the η¹‐Cp^Sn ligands, with distances of 3.1849(6)/3.1676(5) Å in 1_Gd , 3.1649(6)/3.1760(8) Å in 2_Gd , and 3.1799(7) Å in 3_Gd . The stepwise reductions are also accompanied by a marked decrease in the tin–tin distance, i.e., from 3.2154(4) Å in 1_Gd to 3.1780(6) Å in 2_Gd and 3.0989(13) Å in 3_Gd , whereas a slight lengthening in the Gd⋅⋅⋅Gd separation occurs from 5.3506(7) Å in 1_Gd to 5.3672(7) Å in 2_Gd , followed by a much larger increase to 5.4658(9) Å in 3_Gd . The increase in separation between the gadolinium ions with each reduction, which is also found in the yttrium and dysprosium analogues (Tables S4, S6), presumably occurs to accommodate a progressively stronger transannular tin–tin interaction.

Another distinct variation in the structures of the gadolinium complexes relates to the distribution of Sn–C and C–C distances within the stannole rings, as shown in Scheme 2. A significant increase in the Sn─C bond lengths from 1_Gd to 3_Gd occurs along with short–long–short alternating C─C bond lengths. In addition, puckering of the SnC₄ ring is observed, reflected in the C1‐Sn‐C3‐C4 torsional angles of 12.86(17)/11.97(17)°, 14.9(4)/15.7(5)°, and 22.4(7)° in 1_Gd , 2_Gd , and 3_Gd , respectively. These structural data could indicate a loss in the aromatic character of the stannole ring upon reduction to give 3_Gd , with similar patterns observed for the yttrium and dysprosium versions (Tables S4, S6).

Scheme 2.

Bond lengths (Å) within the stannole ligands of 1_Gd , 2_Gd , and 3_Gd (shown for one of two unique stannole ligands in 1_Gd and 2_Gd ).

Turning to solution‐phase studies of the diamagnetic yttrium complex 1_Y by NMR spectroscopy, a 1:1 doublet occurs at δ = 529.6 ppm in the ¹¹⁹Sn{¹H} NMR spectrum in toluene‐D₈, presumably due to coupling between ⁸⁹Y (I = 1/2, 100% abundance) and ¹¹⁹Sn via the η¹–Cp^Sn interaction, with ¹  J = 444.6 Hz (Figure S7). This observation is consistent with retention of the dimeric structure in solution. The ¹H NMR spectrum of 1_Y consists of resonances at δ = 0.06 ppm for the SiMe₃ groups, 1.28 and 1.49 ppm for the tert‐butyl groups, 2.70 ppm for the stannole methyl substituents, and at 6.42 ppm for the cyclopentadienyl protons (Figure S8). The ¹H/²⁹Si HMBC spectrum of 1_Y allows the ²⁹Si chemical shift to be identified at δ = −8.42 ppm (Figure S9). In contrast to 1_Y , the ¹H NMR spectrum of paramagnetic [K(2.2.2‐crypt)][2_Y ] in THF‐D₈ only shows resonances due to the cryptand ligand, which are partially obscured by residual proton solvent resonances (Figure S11). The ¹H NMR spectrum of [K(2.2.2‐crypt)]₂[3_Y ] in THF‐D₈ features resonances at −0.25 ppm for the SiMe₃ groups, 1.02 and 1.54 ppm for the tert‐butyl groups, 2.58, 3.54, 3.61 ppm for the cryptand ligand, and 5.70 ppm for the cyclopentadienyl protons (Figure S12). A single resonance was observed for the trimethylsilyl groups in the ²⁹Si/¹H HMBC NMR spectrum at δ = −11.41 ppm (Figure S13), but no signal was observed in the ¹¹⁹Sn{¹H} NMR spectrum (Figure S14).

Following characterization of the solution‐phase structures of the three yttrium–stannole complexes, ¹H NMR spectroscopy was used to determine if [K(2.2.2‐crypt)][2_Y ] could be converted into 1_Y and [K(2.2.2‐crypt)]₂[3_Y ] by means of simple oxidation and reduction reactions, respectively. Thus, adding one equivalent of AgPF₆ to [K(2.2.2‐crypt)][2_Y ] produced 1_Y (Figure S15), and adding one equivalent of KC₈/2.2.2‐crypt to [K(2.2.2‐crypt)][2_Y ] generates [K(2.2.2‐crypt)]₂[3_Y ] (Figure S16), with both reactions proceeding cleanly and essentially quantitatively in THF‐D₈.

Bonding and Electronic Structure

To gain insight into the bonding in the yttrium and gadolinium stannole complexes, density functional theory (DFT) calculations were performed using the coordinates obtained from the crystallographic studies. The ORCA 6.0.0 software package was used for these calculations,^{[ 24} , ^{25 ]} with full details provided in the Supporting Information.

The overall appearance and composition of the frontier MOs in compounds 1 _M , the singly reduced complexes 2 _M , and doubly reduced 3 _M (M = Y, Gd) are similar (Figures 2, S17). In the case of 1 _M , the HOMOs (highest‐occupied MOs) and LUMOs (lowest‐unoccupied MOs) consist of appreciable contributions from the yttrium 4d (30%) and gadolinium 5d (30%) orbitals in addition to the tin 5p orbitals (11% and 10%, respectively). Notably, the LUMOs in 1 _M feature overlap between the tin‐based orbitals, which effectively become the SOMOs (singly occupied MOs) following one‐electron reduction to give 2 _M . The SOMO for 2_Y features 27% yttrium 4d character and a tin 5p contribution of 21%, and for 2_Gd the gadolinium 5d and tin 5p orbitals contribute 33% and 18%, respectively. Adding a second electron to the SOMO in 2_Y and 2_Gd to give 3_Y and 3_Gd results in HOMOs with 25% 4d and 23% 5p character for the yttrium complex, and 31% 5d and 19% 5p character for the gadolinium complex. The Wiberg bond index (WBI) for the tin–tin interaction in 2_Y is 0.81, increasing to 1.06 in 3_Y (Figure 3). The WBIs for the yttrium–tin interactions were calculated for all three compounds and found to increase slightly with each reduction step. Overall, the pattern of WBIs for the yttrium–stannole complexes are consistent with the variation in the bond lengths within the {Y₂Sn₂} cores.

Figure 2.

Frontier molecular orbitals for 1_Y (left), the mono‐anion 2_Y (centre), and the di‐anion 3_Y (right). Isosurface value = 0.04 a.u.

Figure 3.

Wiberg bond index values for 1_Y (left), 2_Y (centre), and 3_Y (right).

Noting the structural changes that occur within the stannole rings with each reduction step, nucleus‐independent chemical shift (NICS) calculations were carried out on diamagnetic 1_Y and 3_Y to explore the impact on the aromaticity of these formally 6π‐electron systems.^{[ 26 ]} In 1_Y , the aromatic character of both the stannole and Cp^ttt ligands is reflected in the negative NICS values of −14.97 and −8.64 ppm, respectively (Figure S18). In 3_Y , whereas the Cp^ttt ligand retains a negative NICS value of −6.21 ppm, that of the stannole ligand is now + 25.29 ppm. The loss in the aromatic character of the stannole is, presumably, in favour of the delocalized bonding in the {Y₂Sn₂} ring system.

The electronic structure of the yttrium–stannole complexes was investigated further using UV/visible spectroscopy and time‐dependent DFT calculations. The UV/vis spectrum of 1_Y in THF (Figure 4) consists of a single major absorption centred on λ _max = 795 nm, which can be assigned to a HOMO‐to‐LUMO excitation calculated at 760 nm according to the TD‐DFT analysis (Figure S19, Table S7). The major absorption in the UV/vis spectrum of [K(2.2.2‐crypt)][2_Y ] in THF at λ _max = 918 nm (Figure 4) corresponds to a SOMO‐to‐LUMO transition calculated to occur at 886 nm (Table S8). The UV/vis spectrum of [K(2.2.2‐crypt)]₂[3_Y ] features three significant absorptions centered on λ _max = 965 nm, λ _max = 795 nm, and λ _max = 460 nm (Figure 4). The lower‐ and intermediate‐energy absorptions correspond to the HOMO–LUMO transition calculated at 1066 nm, and transitions from the HOMO and HOMO–1 to the LUMO, LUMO + 1 and LUMO + 2 calculated at 871 and 788 nm (Table S9). The higher energy transitions are from the from the HOMO to higher‐lying LUMOs calculated at 405–545 nm.

Figure 4.

UV/vis spectra of 1_Y ·toluene (blue), [K(2.2.2‐crypt)][2_Y ]·1.5(hexane) (red), and [K(2.2.2‐crypt)]₂[3_Y ]·3(THF) (green) in THF.

The UV/vis spectra of 1_Gd and 1_Dy are qualitatively similar to those of yttrium analogues, with major absorptions at λ _max = 800 and 805 nm, respectively (Figures S20, S21). Likewise, the UV/vis spectra of [K(2.2.2‐crypt)][2_Gd ] and [K(2.2.2‐crypt)][2_Dy ] are also comparable to the yttrium versions, with absorptions at λ _max = 922 and 889 nm, respectively. The UV/vis spectra of [K(2.2.2‐crypt)]₂[3_Gd ] and [K(2.2.2‐crypt)]₂[3_Dy ] display much weaker absorptions located around λ _max = 455, 670, and 980 nm for gadolinium, with barely discernible λ _max values for dysprosium. Overall, the similarities in the UV/vis spectra of the yttrium–, gadolinium– and dysprosium–stannole complexes indicate similarities in the orbital structure and relative energies, based on the TD‐DFT analysis of the yttrium versions.

EPR Spectroscopy and DC Magnetic Properties

The X‐band EPR spectrum of [K(2.2.2‐crypt)][2_Y ] was recorded as a frozen solution in 2‐Me‐THF at 100 K and is anisotropic in nature due to delocalization of the unpaired electron (Figure 5). A simulation was achieved using g_x  = 2.057, g_y  = 2.058, and g_z  = 1.932, the anisotropic hyperfine coupling constants A_x  = 161 MHz, A _y = 30 MHz and A _z = 140 MHz for coupling to ⁸⁹Y, and A_x  = 112 MHz, A_y  = 13 MHz and A_z  = 132 MHz for coupling to ¹¹⁹Sn, with anisotropic linewidths of 16, 30, and 22 MHz.

Figure 5.

X‐Band EPR spectrum of [K(2.2.2‐crypt)][2_Y ] in 2‐Me‐THF at 100 K (red) and simulated spectrum (blue) using the parameters stated in the text.

Ddelocalization of the radical electron in [K(2.2.2‐crypt)][2_Y ] is reflected in a DFT calculation of the spin density in the anion 2_Y , which shows appreciable unpaired spin on the yttrium and tin atoms, including along the tin–tin axis (Figure 6). Similar calculations on the three gadolinium complexes revealed the spin density to reside solely on the Gd³⁺ ions in 1_Gd and 3_Gd , as expected, with additional spin density located within the {Gd₂Sn₂} ring of 2_Gd (Figure 6). The spin density properties of 2_Gd therefore prompted measurement of the magnetic susceptibility properties to determine the nature of the exchange coupling between gadolinium and the tin radical ligand.

Figure 6.

Spin density plots for a) 2_Y , b) 1_Gd , c) 2_Gd , and d) 3_Gd . The isosurface value is 0.0015 for 2_Y and 0.0026 for the gadolinium complexes.

The molar magnetic susceptibility (χ _M) of the gadolinium– and dysprosium–stannole compounds was measured as a function of temperature in a DC field of 1000 Oe. For compounds 1_Gd and [K(2.2.2‐crypt)]₂[3_Gd ], the plots of χ _M T(T) in the temperature range 2–300 K are characteristic of weak antiferromagnetic exchange coupling between the gadolinium centres (Figure 7). At 300 K, χ _M T is 15.63 cm³ K mol⁻¹ for 1_Gd and 15.61 cm³ K mol⁻¹ for [K(2.2.2‐crypt)]₂[3_Gd ], similar to the theoretical value of 15.76 cm³ K mol⁻¹ for two uncoupled Gd³⁺ ions (⁸S_7/2 ground term). The susceptibility for both compounds then decreases gradually down to 20 K before reaching 3.07 and 3.77 cm³ K mol⁻¹, respectively, at 2 K. In contrast, the radical‐bridged compound [K(2.2.2‐crypt)][2_Gd ] has a markedly different χ _M T(T) profile, with the susceptibility increasing from 16.02 cm³ K mol⁻¹ at 300 K, close to the predicted value of 16.14 cm³ K mol⁻¹ for two Gd³⁺ ions and a single unpaired electron,^{[ 27 ]} to a maximum of 22.28 cm³ K mol⁻¹ at 18 K. The susceptibility then decreases to 18.17 cm³ K mol⁻¹ at 2 K. These data indicate much stronger antiferromagnetic exchange coupling in [K(2.2.2‐crypt)][2_Gd ].

Figure 7.

Plots of χ _M T(T) for the gadolinium–stannole (top) and dysprosium–stannole (bottom) complexes. Data were collected in an applied field of 1000 Oe. Solid lines for the gadolinium compounds represents fits to the data using the spin Hamiltonian parameters stated in the text. Solid lines for 1_Dy and 3_Dy represent ab initio simulations of the data using the parameters in the text.

For 1_Gd and [K(2.2.2‐crypt)]₂[3_Gd ], a −2J spin Hamiltonian formalism (Equation 1) was used to fit the data with terms to account for the gadolinium–gadolinium exchange (J _GdGd), the Zeeman interaction and an intermolecular exchange term (zJ'). For [K(2.2.2‐crypt)][2_Gd ], an additional term (J _GdSn) was used to account for the gadolinium‐radical exchange (Equation 2). The parameters used in the fits are presented in Table 1. The most striking result from this analysis is the very large J _GdSn value of –112 cm⁻¹ for [K(2.2.2‐crypt)][2_Gd ], which is accompanied by an unusually large J _GdGd value of −5.13 cm⁻¹. In stark contrast, the values of J _GdGd for 1_Gd and [K(2.2.2‐crypt)]₂[3_Gd ] are more characteristic of the weak exchange coupling normally found in polynuclear gadolinium compounds,^{[ 13 ]} being −0.23 and −0.24 cm⁻¹, respectively.

Table 1.

Parameters used to fit the χ _M T(T) data for the gadolinium–stannole compounds using a −2J spin Hamiltonian formalism.

	1Gd	2Gd	3Gd
g	2.003	1.94	1.98
J Gd‐Gd/cm−1	−0.23	−5.13	−0.24
J Gd‐Sn/cm−1		−112
zJ'/cm−1	−0.005	−0.006	−0.04

The χ _M T(T) data for the three dysprosium–stannole complexes follow a qualitatively similar profile to the gadolinium versions (Figure 7). For 1_Dy and [K(2.2.2‐crypt)]₂[3_Dy ], the values of χ _M T at 300 K are 28.76 and 27.88 cm³ K mol⁻¹, respectively, in good agreement with the expected value of 28.34 cm³ K mol⁻¹ for two uncoupled Dy³⁺ ions (⁶H_15/2 ground term). A steady decrease in χ _M T then occurs as the temperature is lowered to around 20 K, and then a sharper decrease in the susceptibility to 18.87 cm³ K mol⁻¹ for 1_Dy and 12.47 cm³ K mol⁻¹ for [K(2.2.2‐crypt)]₂[3_Dy ] occurs at 2 K. For [K(2.2.2‐crypt)][2_Dy ], the value of χ _M T at 300 K is 28.60 cm³ K mol⁻¹, close to the expected value of 28.7 cm³ K mol⁻¹ for two uncoupled Dy³⁺ ions and an unpaired electron.^{[ 27 ]}

(1)

(2)

The value of χ _M T then steadily decreases as the temperature is lowered to 100 K, before increasing sharply to reach a maximum of 34.26 at cm³ K mol⁻¹ 14 K, before decreasing to 28.48 cm³ K mol⁻¹ at 2 K.

Simulations of the susceptibility data were achieved for 1_Dy and 3_Dy using the POLY_ANISO routine and the Ising‐type spin Hamiltonian stated in Equation 3. The resulting dipolar exchange coupling constants are J _dip = −0.006 cm⁻¹ for 1_Dy and + 0.020 cm⁻¹ for 3_Dy , with exchange coupling constants of J _exch = −0.028 and −0.32 cm⁻¹, respectively. The total coupling in 1_Dy and 3_Dy is therefore J _tot = −0.034 and −0.30 cm⁻¹, respectively. Although a simulation of the susceptibility for 2_Dy proved to be computationally intractable, the similarities with the data for 2_Gd suggest that the dysprosium‐radical exchange in 2_Dy should be considerably stronger than in 1_Dy and 3_Dy . The susceptibility properties of 1_Dy and 3_Dy are also remarkably like their germole‐ligated cousins,^{[ 28 ]} whereas those of 2_Dy show a more pronounced maximum in the χ _M T(T) data than the radical‐bridged germanium complex. This observation may reflect greater enhancement of the bonding across the {Dy₂Sn₂} unit upon one‐electron reduction, presumably due to the more diffused character of the tin 5p orbitals relative to the 4p orbitals of germanium.

$$\widehat{H}=-\left(J_{dip}+J_{ex}\right)\widehat{{\stackrel{\sim }{S}}_{1,z}}·\widehat{{\stackrel{\sim }{S}}_{2,z}}$$ (3)

Dynamic (AC) Magnetic Properties

Di‐ and tri‐metallic dysprosium complexes have proven to be invaluable for studying the impact of exchange coupling on SMM properties, particularly the effective energy barrier to reversal of the magnetization (U _eff) and magnetic hysteresis.^{[ 29} , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ^{37 ]} Whilst previous studies have focused on N‐ and O‐bridged multimetallic systems, several examples with μ‐bridging heavier p‐block elements (P, As, Sb, Bi, S, Se) have also been described.^{[ 18} , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ^{42 ]} Furthermore, the only tin‐ligated SMM is the monometallic erbium–stannole sandwich complex [(η⁵‐Cp^Sn)Er(η⁸‐COT)]⁻ reported by us,^{[ 22 ]} with the properties of two lanthanide bis(stannole) complexes having been described in a recent preprint.^{[ 43 ]} Compounds 1_Dy , [K(2.2.2‐crypt)][2_Dy ] and [K(2.2.2‐crypt)]₂[3_Dy ] therefore provide an opportunity to investigate how the unique {Dy₂Sn₂} delocalized bonding and the tin‐mediated exchange impact on the dynamic magnetic properties.

In zero DC field, all three dysprosium–stannole dimetallic complexes give rise to maxima in the out‐of‐phase component of the AC susceptibility (χ″) as a function of frequency (ν) (Figures 8, S24–34). For 1_Dy , temperature‐dependent maxima were observed in the relatively wide temperature range 4–27 K, whereas for [K(2.2.2‐crypt)][2_Dy ] and [K(2.2.2‐crypt)]₂[3_Dy ] the maxima were observed in much narrower ranges of 2–4 and 2–3.5 K, respectively, and the maxima show only a weak temperature dependence. Magnetic relaxation times (τ) were then extracted from Cole–Cole plots of χ″ versus the in‐phase component of the AC susceptibility χ′. The plot of ln τ versus T ⁻¹ for 1_Dy is roughly linear at 22–27 K before showing curvature at lower temperatures (Figure S28, Table S10). These data suggest that at least Orbach and Raman relaxation processes occur, with some contribution from quantum tunnelling of the magnetization (QTM) at low temperatures. A fit of the relaxation time data for 1_Dy was obtained using Equation 4, in which τ ₀ is the attempt time, C is the Raman coefficient, n is the Raman exponent and ${\tau }_{\mathrm{QTM}}^{-1}$ is the rate of QTM. This analysis gave U _eff = 138 ± 7 cm⁻¹, τ ₀ = 1.44 × 10⁻⁷ s, C = 6.2 × 10⁻⁴ ± 10⁻⁴ s⁻¹ K^{− n} , n = 4.8 ± 0.1, and τ _QTM = 1.97 ± 0.34 s.

Figure 8.

Plots of χ″ versus AC frequency (ν) in zero DC field for: 1_Dy (top), [K(2.2.2‐crypt)][2_Dy ] (centre) and [K(2.2.2‐crypt)]₂[3_Dy ] (bottom) at the indicated temperatures. Solid lines indicate fits to a generalised Debye model.

$${\tau }^{-1}={\tau }_0^{-1}{e}^{-U_{\mathrm{eff}}/k_{\mathrm{B}}T}+C{T}^n+{\tau }_{\mathrm{QTM}}^{-1}$$ (4)

The ln τ versus T ⁻¹ plot for [K(2.2.2‐crypt)][2_Dy ] has more pronounced curvature (Figure S32, Table S11), with the fit using Equation 4 producing U _eff = 36 ± 1.4 cm⁻¹, τ ₀ = 1.14 × 10⁻⁹ s, C = 5.4 ± 1.6 s⁻¹ K^{− n} , n = 3.5 ± 0.25, and τ _QTM = 1.33 × 10⁻³ ± 1.41 × 10⁻⁵ s. In contrast, a fit of the relaxation time data for [K(2.2.2‐crypt)]₂[3_Dy ] was possible using only the Raman and QTM terms (Figure S36, Table S12), resulting in C = 19.9 ± 7.0 s⁻¹ K^{− n} , n = 4.3 ± 0.3, and τ _QTM = 1.035 × 10⁻³ ± 9.5 x 10⁻⁴ s.

The prominent QTM in all three dysprosium–stannole complexes indicates fast relaxation of the magnetization even at low temperatures and implies that none of these compounds should show magnetic hysteresis (memory) effects. This was confirmed by measuring the dynamic field‐dependence of the magnetization at 2 K in each case (Figures S37–S40). For 1_Dy , using field sweep rates in the range 0.23–11.6 mT s⁻¹, the magnetization (M) versus field (H) loops shows a very slight opening at 2 K around zero field, closing at higher temperatures. The M(H) loops for [K(2.2.2‐crypt)][2_Dy ] are butterfly‐shaped and closed around zero field, whereas for [K(2.2.2‐crypt)]₂[3_Dy ] no discernible opening of the hysteresis loops occurs.

The decrease in U _eff value from 1_Dy to [K(2.2.2‐crypt)][2_Dy ], and the absence of a measurable barrier in [K(2.2.2‐crypt)]₂[3_Dy ], can be interpreted qualitatively in terms of the changes in molecular structure that occur across the dysprosium–stannole units with each reduction. With the contraction of the {Dy₂Sn₂} rings across the series and the build‐up of electron density on and between the tin atoms, the equatorial component of the crystal field is enhanced. Simultaneously, the Dy–(Cp^ttt)_cent interactions lengthen, which weakens the axial crystal field from 1_Dy to 2_Dy , and again from 2_Dy to 3_Dy . For the oblate spheroidal ion Dy³⁺, the structural changes should be detrimental to the SMM properties, as found in other dysprosium metallocene SMMs.^{[ 44} , ⁴⁵ , ^{46 ]}

Multireference Calculations

To provide a more detailed basis for the variation in the SMM properties of the dysprosium–stannole complexes, multireference calculations were performed on 1_Dy and 3_Dy using the atomic coordinates obtained from the crystallographic studies. All calculations were performed using the ORCA 6.0.0 software package.^{[ 24} , ^{25 ]} Calculations on the radical‐bridged system 2_Dy were not carried out owing to the prohibitive computational cost.

For 1_Dy , the easy axes of magnetization in the ground Kramers doublet (KD) for each Dy³⁺ ion are oriented towards the centre of the [η⁵‐Cp^Sn]²⁻ ligand rather than the [η⁵‐Cp^ttt]⁻ ligand, presumably because of the greater formal charge on the former (Figure 9). The g‐tensors associated with the ground KD of Dy1 in 1_Dy are g_x  = 0.0018, g_y  = 0.0026, and g_z  = 19.70, and for Dy2 they are g_x  = 0.0020, g_y  = 0.0029, and g_z  = 19.69 (Tables S16, S17). The strong axial character of the ground KDs in 1_Dy is emphasized by the wavefunction compositions of greater than 96% |M_J | = 15/2 for both dysprosium centres. The first‐excited KDs in 1_Dy both lie at 200 cm⁻¹ above the ground KD and feature larger contributions from the transverse components of the g‐tensors, along with significant mixing of wavefunctions. These properties indicate that Orbach relaxation should occur via the first‐excited KD, as reflected in the calculated relaxation barriers for this system (Figures S42, S43). Although the experimental barrier is somewhat lower than predicted by the calculations, it is possible that the discrepancy is related to electron correlation effects outside of the dysprosium 4f manifold.

Figure 9.

Easy axes of magnetization (blue lines) in the ground Kramers doublets of the Dy³⁺ ions in 1_Dy (left) and 3_Dy (right).

In 3_Dy , the orientation of the easy axes of magnetization shifts towards the centre of the Cp^ttt ligands and, consequently, towards one of the Sn─C bonds of the stannole ligand (Figure 9), which could be due to delocalization of electron density from the stannole ring into the {Dy₂Sn₂} ring. Significantly reduced axial character of the ground KDs in 3_Dy is reflected in the associated g‐tensors of g_x  = 0.0365/0.0366, g_y  = 4.693/4.665 and g_z  = 14.51/14.54, with the associated wavefunctions being strong admixture of states (Tables S18, S19). As such, fast QTM within the ground KD is expected to be the dominant relaxation mechanism in 3_Dy , as shown in the calculated relaxation barriers (Figures S44, S45), consistent with experimental observations.

Discussion

Compounds containing bonds between tin and rare‐earth elements form a relatively small family that can be divided into two types. The first consists of coordination and organometallic compounds of interest for covalency in the metal–metal bonds and their associated reactivity.^{[ 47} , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ^{54 ]} The second type is composed of structurally elaborate intermetallic clusters with unusual electronic properties,^{[ 55} , ⁵⁶ , ^{57 ]} which can enable applications in lithography and as magnetic materials.^{[ 58} , ^{59 ]} By comparison, f‐element metallole chemistry, especially with stannole ligands, is conspicuously under‐developed despite the availability of reliable synthetic methods for a variety of these heavy cyclopentadienyl analogues.^{[ 28} , ⁶⁰ , ⁶¹ , ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ , ^{66 ]} Furthermore, surprisingly few η⁵‐stannole complexes of transition metals have been reported,^{[ 67} , ⁶⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ^{74 ]} emphasizing that the coordination chemistry of this ligand is still to be explored.

The one‐ and two‐electron reduction of 1 _M to give the tin–tin bonded complexes 2 _M and 3 _M , respectively, are unusual examples of stannole coupling to give di‐stannole compounds. The reactivity of 1 _M is reminiscent of the tin–tin coupling reactions of dilithio‐stannoles to give di‐stannole compounds, although these reactions typically occur under oxidative, rather than reductive, conditions.^{[ 75} , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ , ⁸¹ , ⁸² , ^{83 ]} Whilst 1 _M , 2 _M , and 3 _M are reminiscent of their germanium analogues^{[ 21 ]} and of a related titanium(III) germole complex with delocalized {Ti₂Ge₂} bonding,^{[ 84 ]} their similar chemistry is perhaps surprising given the expected weaker nature of tin‐tin bonds. The valence d‐orbitals of the rare‐earth metals evidently play a critical structural role in supporting the highly unusual delocalized {M₂Sn₂} bonding in 2 _M and 3 _M .

To the best of our knowledge, the tin‐based radicals in 2 _M are not only unprecedented as ligands in f‐element chemistry, but they are also unknown in transition metal chemistry. The tin–tin bond in 2 _M shares some characteristics with the unusual single‐electron bond formed upon one‐electron reduction of the di‐amido stannylene [Sn(NDipp)₂C₆H₄], forming the di‐tin radical anion [C₆H₄(DippN)₂Sn⋅⋅⋅Sn(NDipp)₂C₆H₄]⁻ (Dipp = 2,6‐diisopropylphenyl).^{[ 85 ]} Whereas structurally authenticated monometallic tin(I) and tin(III) radicals are relatively numerous,^{[ 86} , ⁸⁷ , ⁸⁸ , ⁸⁹ , ^{90 ]} complex 2 _M is a rare example of a di‐tin radical.^{[ 91} , ⁹² , ^{93 ]}

The gadolinium–tin exchange coupling constant of J = −112 cm⁻¹ in 2_Gd is one of the largest reported for a lanthanide complex. Although larger coupling constants have been determined, including a record of 387(4) cm⁻¹, these exceptionally strong interactions are based on direct lanthanide–lanthanide bonds rather than exchange between a lanthanide and a radical ligand.^{[ 94} , ⁹⁵ , ⁹⁶ , ⁹⁷ , ^{98 ]} Indeed, strong lanthanide‐radical exchange couplings involving S = 1/2 ligands such as [N₂]³⁻ or N‐heterocycles are typically in the range of 20–30 cm⁻¹, making the exchange in 2_Gd unusually strong for this type of complex.^{[ 99 ]} It is also noteworthy that the only significant difference in the magnetic properties of 2_Gd and its isostructural germole analogue is the smaller exchange coupling of −95 cm⁻¹,^{[ 21 ]} highlighting that lanthanide‐radical coupling can be tuned by varying the group 14 element. Furthermore, the ability of heavy p‐block radical ligands to enhance exchange interactions between lanthanide centres is also reflected in the large J _GdGd value of −5.13 cm⁻¹ for 2_Gd , which is comparable in magnitude to the analogous parameter of approximately −1.92 cm⁻¹ found in a bismuth‐based radical‐bridged di‐lanthanide complex.^{[ 18 ]}

Conclusion

The dimeric rare‐earth stannole complexes [(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)]₂ (1 _M , M = Y, Gd, Dy) undergo one‐ and two‐electron reduction to give the corresponding complex anions [{(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)}₂]⁻ (2 _M ) and [{(η⁵‐Cp^Sn)M(η⁵‐Cp^ttt)}₂]²⁻ (3 _M ), respectively, as salts of [K(2.2.2‐crypt)]⁺ in good yields. Crystallographic studies reveal a contraction of the central {M₂Sn₂} rings with each reduction, consistent with the population of a bonding molecular orbital. A loss in the aromaticity of the stannole rings in 3 _M is also suggested by the structural studies and is supported by the determination of a positive NICS(0) value for 3_Y . Bonding analysis of the yttrium‐ and gadolinium‐stannoles reveals the frontier MOs to have appreciable contributions from the valence d‐orbitals of the rare‐earth metals and the tin 5p orbitals. Reduction of 1 _M to give the tin radical complexes 2_M  results in population of a SOMO that spans the {M₂Sn₂} rings and the stannole ligand, with appreciable end‐on overlap between tin 5p orbitals to give a tin‐tin bond. Further reduction to give 3 _M enhances the M─Sn and Sn─Sn bonding, reflected in increased WBI values. Magnetic measurements on 2_Gd reveal an extremely large coupling of −112 cm─¹ between gadolinium and the tin radical, and EPR measurements of 2_Y reveal extensive hyperfine interactions spanning the yttrium and tin centres, consistent with extensive delocalization of spin density.

With the stannole ligand now established in rare‐earth chemistry, the next phase in our investigations will focus on the reactivity of these compounds as bimetallic reducing agents.

Supporting Information

Synthesis, FTIR, NMR and UV/vis spectra, X‐ray crystallography details, magnetic measurements, EPR spectroscopy, DFT and multireference calculation details. Additional research data supporting this publication are available as supplementary information at DOI: 10.25377/sussex.29634422. The authors have cited additional references within the Supporting Information.^{[ 100} , ¹⁰¹ , ¹⁰² , ¹⁰³ , ¹⁰⁴ , ¹⁰⁵ , ¹⁰⁶ , ¹⁰⁷ , ¹⁰⁸ , ¹⁰⁹ , ¹¹⁰ , ¹¹¹ , ¹¹² , ¹¹³ , ¹¹⁴ , ¹¹⁵ , ¹¹⁶ , ¹¹⁷ , ¹¹⁸ , ¹¹⁹ , ^{120 ]}

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Supporting Information

De S., Mondal A., Tang J., Layfield R. A., Angew. Chem. Int. Ed.. 2025, 64, e202516323. 10.1002/anie.202516323

Data Availability Statement

The data that support the findings of this study are openly available in Figshare at https://doi.org/10.25377/sussex.29634422, reference number 1.