A Cu₁₂ Metallacycle Assembled from Four C₃-Symmetric Spin Frustrated Triangular Units

Abstract

Assembling metallacycles with interesting topological arrangements is a critical task for chemists. We report here a novel dodecanuclear Cu^II compound, [{Cu₃L(µ-N₃)}₄(Py)₁₄]·2Py (Cu₁₂) (where Py = pyridine and [H₆L]Cl = tris(2-hydroxybenzylidine)triaminoguanidinium chloride, respectively), with the topology of a cycle accomplished by four two-connecting approximately flat C₃-symmetric guanidine-based ligands. Each ligand affords three tridentate metal-binding cavities and the four node-to-node connections through single azido bridges are provided by pairs of metal centers. A theoretical investigation using CASSCF in addition to DFT calculations showed strong antiferromagnetic coupling within the Cu₃-triangles, resulting in spin-frustrated systems. However, these calculations were not able to properly reproduce the very weak antiferromagnetic couplings between the triangle units, highlighting the challenge of describing the magnetic behavior of this compound.

1. Introduction

Coordination chemistry provides an opportunity for construction by self-assembling an apparently limitless variety of structures [1,2]. Specifically, the construction of the supramolecular coordination chemistry of transition metal complexes has received much attention over the last couple of decades, due to their vital roles in biology [3,4,5], catalysis [6,7], and more importantly, in magneto-chemistry [8,9,10,11,12,13,14]. In this context, polytopic threefold symmetric ligands based on triaminoguanidine are of particular interest in crystal engineering and supramolecular chemistry [15,16,17,18] due to their structural diversity and the chemical versatility of the resulting systems [19,20,21,22,23,24]. Ligands possessing triaminoguanidine core have been successfully employed to develop high-nuclear supramolecular cages, but with diamagnetic metal ions [2,25,26,27,28,29]. Only a couple of open-shell metal ion systems [30,31], copper(II) coordination polymers [32,33], and a series of trinuclear cobalt(II) [34], nickel(II) [35,36], Iron (III) [37], and copper(II) [38,39] triangles have been studied magnetically, with the latter two showing the phenomenon of spin frustration, having significant applications in molecular spintronics [40].

Triaminoguanidine-based ligands with three equivalent coordinating sites could coordinate with three metal ions, leading to an almost planar structure style and rather small metal∙∙∙metal distances bridged through the N−N diazine channels of the central triaminoguanidine moiety [30]. Consequently, the established complexes display strong antiferromagnetic exchange interactions through the σ bond of the N−N diazine arms of the ligand among the metal centers [34,35,36]. In this regard, Plass and co-workers [38,39] developed two trinuclear copper(II) triangles, [Cu₃L(py)₆]ClO₄ (Cu₃py) and [Cu₃L(bpy)₃]ClO₄·3DMF (Cu₃bpy) (py and bpy are pyridine and bipyridine, respectively). Both complexes exhibit strong antiferromagnetic coupling between the copper(II) ions, leading to a spin-frustrated system. Very recently, by fluctuating the synthetic approach while keeping the same metal–ligand combination, we synthesized two new antiferromagnetic hexa-nuclear Cu^II complexes, [Cu₆L₂Cl(µ-OAc)(DMF)₃]·DMF (Cu₆) and [Cu₆L₂(µ-Cl)₂(DMF)₄] (Cu₆Cl), in which two planar triangles with strict C₃-symmetry were deliberately bridged through acetate and chloride anions in cis and trans fashions, respectively [41]. It was assumed that both Cu₆ and Cu₆Cl would also present the phenomenon of spin frustration based on their similar structural and experimental magnetic properties to the reference compounds, Cu₃py and Cu₃bpy.

Coordination supramolecular cages with tetrahedral topologies have been constructed using triaminoguanidine-based ligands, tris (2-hydroxybenzylidine) triaminoguanidinium [H₆L]⁺, and tris (5-bromo-2-hydroxybenzylidine) triaminoguanidinium [H₆L₁]⁺, in which the triangular units were connected through (−CdO−)₂ four-membered rings (O from the phenolate of the ligand itself), resulting in [{(CdCl)₃L}]⁸⁻ and [{(CdCl)₃L¹}]⁸⁻, respectively [28,42]. The authors concluded that the construction of a comparable discrete single-walled supramolecular tetrahedron with Zn²⁺ ions was impossible due to the much smaller size of the Zn²⁺ ion relative to the Cd²⁺ ion (Zn²⁺, 0.74 Å; Cd²⁺, and 0.97 Å) [29]. In the case of a Zn²⁺ single-walled tetrahedron, the triangular faces would result in closer contact in view of its smaller size, which is sterically unfavorable. Hence, based on this assumption, they developed a double-walled tetrahedron using two different triaminoguanidine-based mixed ligands [29]. Keeping this task in mind, this work planned to use an azide anion as a bridging co-ligand, owing to its linear structure and least steric hindrance. Surprisingly, we synthesized a dodecanuclear metallacycle based on a Cu²⁺ ion (0.73 Å), [{Cu₃L(µ-N₃)}₄(Py)₁₄]·2Py (Cu₁₂), assembled from four C₃-symmetric isosceles triangular faces in close contact. Furthermore, the experimental and theoretical magnetic calculations revealed very strong intra-triangular antiferromagnetic exchange interactions between the copper ions through the σ bond pathway of the N−N diazine channels of the ligand, very similar to that of the reference compounds, Cu₃py and Cu₃bpy. To our knowledge, this is the first example of a supramolecular Cu₁₂ metallacycle based on triaminoguanidine ligands with four isosceles triangular sub-units that has been synthesized and studied magnetically.

2. Materials and Methods

2.1. Materials and General Information

Tris(2-hydroxybenzylidine)triaminoguanidinium chloride [H₆L]Cl (Scheme S1) was synthesized according to the methods in the literature [28,29]. Sodium azide was purchased and used without further purification. The elemental analyses of C, H, and N were completed on a PerkinElmer 2400 analyzer. The powder X-ray diffraction analysis was achieved on a Burker-D8 advance diffractometer using Cu-Kα (λ = 1.5418 Å) radiation at room temperature.

Synthesis of [{Cu₃L(µ-N₃)}₄(Py)₁₄]·2Py·4[CH₃OH]·0.8[H₂O] (Cu₁₂)

Two freshly prepared solutions, copper(II) salt Cu(OAc)₂·6H₂O (43.8 mg, 0.15 mmol) in 10 mL of methanol and ligand [H₆L]Cl (22.6 mg, 0.05 mmol) in 5 mL of pyridine, were mixed slowly with continued stirring. Then, the sodium azide (0.013 mg, 0.2 mmol) dissolved in 1 mL of water was added dropwise. After 3 h of further stirring, the final solution was filtered out and put aside for slow evaporation. After three days, well-shaped dark green cube-like crystals of Cu₁₂ in a quantitative yield suitable for single crystal measurements were obtained. The elemental analysis (%) calcd for C₁₆₈H₁₄₀Cu₁₂N₅₂O₁₂: C: 52.52, H: 3.67, and N: 18.95; and found C: 52.54, H: 3.65, and N: 18.93.

2.2. Crystallography

The details of the crystal data and structure refinement parameters of the Cu₁₂ are summarized in Table S1, while the obtained bond lengths and angles are given in Table S2. To measure the single crystals of the titled complex, they were mounted on glass fiber under a microscope and the diffractional data were collected at 120 (2) K using a Bruker AXS D8 Venture single-crystal diffractometer equipped with graphite-monochromatized Mo Kα (λ = 0.71073 Å) (Table S1). The drawing of the molecular structure and mean plane analysis were obtained using the DIAMOND (version 3.1) software. The structure was solved using direct methods and SHELXT and refined using full-matrix least-squares methods based on F2 (SHELXL) in the Olex2 package [43,44].

2.3. Magnetic Measurements

The magnetic susceptibility measurements were performed on a polycrystalline sample using a Quantum Design MPMS-XL7 SQUID magnetometer in the 2−300 K temperature range, equipped with a 7 T magnet. The diamagnetic corrections of the constituent atoms were determined from Pascal’s constants [45].

2.4. Theoretical Calculations

Based on the wavefunction theory (WFT), using the OpenMolcas software package with the complete active space (CAS) self-consistent field (SCF) approach [46], the electronic structures and magnetic properties of each Cu center were first calculated [47]. This was achieved by replacing the eleven other Cu^II ions with diamagnetic Zn^II centers using the X-ray structures. In combination with the all-electron atomic natural orbital relativistically contracted basis set (ANO-RCC) [48,49], the calculations were first carried out at the scalar (SR) level using the second-order Douglas–Kroll–Hess scalar relativistic Hamiltonian [50,51,52]. The basis sets were contracted to the triple-ζ plus polarization (TZP) quality for the Cu (21s15p10d6f4g2h/6s5p3d2f1g) and Cl (17s12p5d4f2g/5s4p2d1f) atoms, as well as for the N and O atoms coordinated to the paramagnetic center (N, O = 14s9p5d3f2g/4s3p2d1f). The Zn atoms were treated with a double-ζ plus polarization (DZP) basis set (21s15p10d6f4g2h/5s4p2d1f), whereas the rest of the C, N, O, and H atoms were treated with a double-ζ basis set (C, N, O = 14s9p5d3f2g/3s2p; H = 8s4p3d1f/2s). By considering the five lowest spin-doublet states and an active space formed by 11 electrons spanning 11 orbitals, the state-average calculations were performed accordingly. These 11 orbitals contained five 3d orbitals of the Cu^II ion, five 3d’ orbitals by considering the double-shell effect, and one ligand-centered orbital that could form a bonding σ interaction with the metal-centered orbitals. By using the restricted active space state interaction (RASSI) approach [53], spin–orbit coupling (SOC) was introduced afterwards via a state interaction within the basis of the spin–orbit free states. The EPR g-factors were calculated based on the Reference [54], as implemented in the RASSI module of OpenMolcas, whereas the magnetic susceptibility and magnetization were calculated using the Single-Aniso and Poly-Aniso modules of OpenMolcas, as detailed in Reference [55].

Using the Amsterdam Density Functional (ADF) software package [56,57,58], Kohn–Sham density functional theory (DFT) calculations were carried out to evaluate the magnetic coupling between the Cu^II centers The calculations of the magnetic coupling constants were performed based on the X-ray structures, in which ten of the paramagnetic Cu^II atoms were replaced by diamagnetic Zn^II centers. Such a replacement produced a model system of two Cu^II centers with a spin of 1/2, leading to a triplet or a singlet spin state.

These calculations utilized the scalar all-electron zeroth-order regular approximation (ZORA) [59] in combination with the spin-unrestricted formalism. The hybrid functionals, PBE0 [60,61] (Perdew–Burke–Ernzerhof) with 25% of exact Hartree–Fock exchange (HFX), B3LYP [62,63,64,65] (Becke, three-parameter Lee–Yang–Parr) with 20% of HFX, and B3LYP* [66] with 15% of HFX, were employed along with the triple-ζ polarized Slater-type orbital (STO) all-electron basis set, with one set of polarization function for all the atoms (TZP) [67].

The magnetic exchange coupling J was determined from broken symmetry (BS) calculations first proposed by Noodleman [68], in addition to the spin decontamination scheme from Yamaguchi [69,70],

$$J=2\frac{E_{\mathrm{BS}}-E_{\mathrm{T}}}{{\langle {\widehat{S}}^2\rangle }_{\mathrm{T}}-{\langle {\widehat{S}}^2\rangle }_{\mathrm{B}\mathrm{S}}}$$

where $E_{\mathrm{BS}}$ and $E_{\mathrm{T}}$ are the energy of the BS configuration and the triplet state, respectively, and ${\langle {\widehat{S}}^2\rangle }_{\mathrm{BS}}$ and ${\langle {\widehat{S}}^2\rangle }_{\mathrm{T}}$ are the expectation values of the ${\widehat{S}}^2$ operator.

The magnetic susceptibility of Cu₁₂ was then calculated using the PolyAniso module, as implemented in the OpenMolcas package, with the following spin Hamiltonian,

$$\widehat{H}={\mu }_B\sum _i{g}_i·{\widehat{S}}_i·B-\sum _{i,j}{J}_{ij}·{\widehat{S}}_i·{\widehat{S}}_j$$

where the magnetic moments and magnetic coupling constants of each Cu^II center were obtained from the WFT and DFT calculations, respectively. It is worth noting that the validity of such a computational strategy has been successfully confirmed on a related trinuclear copper(II) complex [39] and closely related Cu₆ and CuCl₆ complexes [41].

3. Results and Discussions

3.1. Crystal Structure of Cu₁₂

The structural data obtained for [{Cu₃L(µ-N₃)}₄(Py)₁₄]·2Py·4[CH₃OH]·0.8[H₂O] (Cu₁₂) revealed that the complex crystalizes in the triclinic space group P$\stackrel{-}{1}$. The unit cell contains one crystallographically independent dodeca-nuclear neutral complex [{Cu₃L(µ-N₃)}₄(Py)₁₄] and two co-crystallized pyridine molecules (Figure 1, left and Figure S1). The supporting information provides the selected crystallographic data, bond lengths, and angles for Cu₁₂ in Tables S1 and S2.

The asymmetric unit of Cu₁₂ contains half of the molecule [{Cu₃L(µ-N₃)}₂(Py)₇]·Py (Figure S2). Each asymmetric unit contains two almost-planar C₃-symmetric triangular fragments, which are bridged via Cu3 and Cu6 (6.11 Å) through azide anions (Figure 1, right and Figure S2). Finally, both asymmetric units are again node-to-node connected via Cu2 and Cu4 (5.95 Å) through azide anions to construct a dodeca-nuclear Cu₁₂ metallacycle. In total, four two-connecting node-to-node connections via single azide bridges are provided by pairs of metal centers (Figure 1, right). Within each triangle, the tri-topic ligand leads to an approximately planar symmetrical triangular arrangement through the N−N diazine bridges among the copper centers. In Cu1 to Cu5, all five copper centers are five-coordinated, in which three equatorial coordination sites (NNO) are occupied by the tridentate pockets of the ligand, along with two trapped pyridine molecules for both the Cu1 and Cu5 centers to provide trigonal bipyramid (D_3h) arrangements, while in the case of the Cu2 to Cu4 centers, the remaining two coordination sites are filled by one pyridine and one bridging azide anion to accomplish spherical square pyramid (C_4v) geometries, respectively (Figure S3 and Table S3). Cu6 is the only metal ion that is four-coordinated and whose fourth position is occupied by the bridging azide anion in cooperating with the planar tridentate pocket (NNO) of the ligand, displaying a square planar (D_4h) geometry (Figure S3). The distances within intra-triangular copper ions range between 4.70 Å and 4.90 Å (Figure 1 right). The Cu−N−N−Cu dihedral angles are almost linear, ranging from 147.32° to 177.97°. The calculated bond lengths and dihedral angles of Cu₁₂ are very consistent with the reported spin-frustrated triangles, Cu₃py and Cu3bpy [38,39].

In addition, at first glance, an offset face-to-face π···π interaction may exist between the two perfectly parallel benzene rings of the ligands related by an inversion center (Figure 2, left), proved by the short distance of 4.06 (1) Å between the centroids of the benzene rings. Although the vertical distance of 3.36 (4) Å between them corresponds to the formation of a weak π···π interaction between the benzene rings, the planes of the two adjacent benzene rings hardly overlap with each other at all, seeing that the centroid shift distance of 2.26 (9) Å between the two benzene rings is almost equal to the centroid distance of 2.42 (1) Å between the two adjacent side-by-side benzene rings (Figure 2, right). Thus, there was nearly no π···π interaction in the Cu₁₂ and a further negligible exchange interaction attached to the centers of Cu6 and Cu6a can be transmitted along this path.

3.2. Magnetic Properties

Temperature-dependent direct current (dc) molar susceptibility (χ_MT) measurements for Cu₁₂ were performed in the temperature range of 2 to 300 K (Figure 3). The room temperature χ_MT value of 3.30 cm³ K mol⁻¹ is significantly lower than the spin-only value expected for the twelve independent Cu^II ions (χ_MT = 0.375 cm³ K mol⁻¹ for S = 1/2 and g = 2.0). The χ_MT products decreased gradually with lowering the temperature, reaching a plateau value of about 1.8 cm³ K mol⁻¹ at around 40 K. This behavior indicates strong antiferromagnetic interactions within the intra-triangular Cu^II ions, with a ground spin state of S = 1/2. Below 15 K, the χ_MT value dropped down sharply to reach a value of 1.51 cm³ K mol⁻¹ at 2 K, indicating the existence of weak inter-triangular interactions between the four Cu₃ triangles in the Cu₁₂ [34,35,36,38,39]. The field dependence of the magnetization curve was also measured for Cu₁₂ under 2.0 K between 0 and 70 kOe (Figure S4). The obtained magnetization plateau value was ca. 3.96 μ_B at 70 kOe, which is consistent with the S = 2 spin ground state for the four triangular units (each with S = 1/2) and fully in line with the recently reported spin-frustrated triangles, Cu₃py and Cu₃bpy, based on the same ligand.

3.3. Theoretical Calculations

The calculated magnetic susceptibility obtained with the different density functionals and experimental data are reported in Figure 3. Between them, a relatively good qualitative agreement was reached for the high-temperature range. This resulted from a relevant determination of the strong antiferromagnetic intra-couplings (J_intra) between the Cu^II ions of the same triangular unit (Table S4). However, the couplings determined from the different functionals used in this work were not able to properly reproduce the trend of (χ_MT) in the low-temperature range. The magnetic behavior at low temperatures mainly resulted from the interactions between the triangular units through the cis or trans azido bridges. These (J_azido) couplings were determined as very weakly ferromagnetic, with values in the order of magnitude of the cm⁻¹ (Table S4). This order of magnitude is basically the expected accuracy of the DFT BS evaluation. Hence, one may have some doubts about the ability of this level of theory to properly evaluate these couplings. In addition, the magnetic couplings between copper ions mediated by such end-to-end azido bridges have been deeply investigated in the literature and are expected to be weakly antiferromagnetic [71,72,73,74]. However, using higher levels of theory such as WFT-based methods is untractable regarding the size of the system considered here.

In order to confirm the possible antiferromagnetic behavior of J_azido, we simulated the χ_MT curve using an average value of J_intra = −250 cm⁻¹ and different values of J_azido. Presented in Figure 4, these results show that using a weak antiferromagnetic J_azido qualitatively allows the reproduction of the magnetic behavior of Cu₁₂, even at the low-temperature range. Using these different antiferromagnetic J_azido results in a non-magnetic ground state. The χ_MT product increased rapidly with respect to the temperature as the low-lying excited states were populated. For J_azido = −2, −5, and −15 cm⁻¹, the χ_MT curves reached a plateau at about 5 K, 15 K, and 40 K, respectively, corresponding to populating the triply degenerated excited states at 0.96, 2.37, and 6.88 cm⁻¹, respectively.

The determination of J_azido highlights the difficulty in computing such subtle interactions within complex systems. To explain this difficulty, one may wonder about the relevance of reducing the interactions between the triangular units with the interactions between the closest Cu^II ions of each unit. However, a more sophisticated treatment appears difficult in the BS-DFT framework, while WFT-based approaches would be untractable.

4. Conclusions

Through synthetic engineering, for the first time, based on a triaminoguanidine ligand, we deliberately constructed a supramolecular [{Cu₃L(µ-N₃)}₄(Py)₁₄]·2Py (Cu₁₂) metallacycle with four triangular sub-units. Each ligand traps three Cu^II ions in its tridentate-binding cavities (NNO), forming rigid planar triangular structures. This is also the first example of a higher nuclear compound in this metal–ligand combination that has been studied magnetically. While comparing Cu₁₂ with reported spin-frustrated Cu₃py and Cu₃bpy systems, similar strong antiferromagnetic interactions within the intra-triangular fragments can be observed and we can also expect spin frustration in Cu₁₂. This type of phenomenon in molecular systems is expected to give rise to spin–electric coupling, which allows for a manipulation of the molecular spin states. Hence, Cu₁₂ can be seen as a potential candidate in the future for molecular spintronics.