Metallocyclic CuII-LnIII Single-Molecule Magnets from the Self-Assembly of 1,4-Diformylnaphthalene-2,3-diol.

We report the synthesis and characterization of seven new tetranuclear 3d-4f complexes derived from the 3:3:1 reaction of 1,4-diformylnaphthalene-2,3-diol (H2 L) with copper(II) nitrate and a lanthanide salt, Ln = Tb [L 3Cu3TbCl2(NO3)2(H2O)2] (C1), Ho [L 3Cu3HoCl3(H2O)3(MeOH)](H2O) (C2), Er [L 3Cu3ErCl3(H2O)3.5(MeOH)0.5](H2O) (C3), Gd [L 3Cu3Gd(NO3)2(H2O)2(MeOH)](NO3) (C4), Dy [L 3Cu3Dy(NO3)2(H2O)2(MeOH)](NO3) (C5), Yb [L 3Cu3Yb(NO3)2(H2O)2(MeOH)](NO3) (C6), and La [L 3Cu3La(NO3)2(H2O)2(MeOH)](NO3) (C7). Structural elucidation showed that the self-assembly using the acyclic ligand system was successful for all seven complexes, which exhibit the same near-planar Cu3LnO12 core. Five complexes (C1, C2, and C4-C6) were magnetically characterized at 300 K and 1.8 K. Complexes C1, C4, and C5 were observed to have ferromagnetic ground states and showed appreciable frequency dependence in their AC magnetic measurements, which yielded effective barriers between 7.82(4) and 13.2(3) K, confirming the presence of single-molecule magnet properties.

Introduction

Since the slow relaxation of magnetization in the archetypal Mn12 single-molecule magnet (SMM) was first reported nearly three decades ago,[1] the study and application of molecular magnetism have flourished and attracted interest across the fields of chemistry, physics, nanoscience, and materials science.[2−10] At low temperatures, the thermal energy kBT becomes comparable to the energy barrier for the reversal or relaxation of magnetization, Ueff, consistent with a magnetically bistable ground state.[11] A decade after the report of Mn12, the first lanthanide-based SMM was reported, [LnPc2].[12] The heavy rare earth metals have become popular nuclei used to slow or prevent the reversal of magnetization in SMMs due to their large ground-state magnetic anisotropies and spin multiplicities.[13−16] Because of the shielding of 4f electrons in lanthanide ions, the exchange interaction between 4f centers is inherently weak, and thus there has been a focus on combining paramagnetic first row transition metal ions with the lanthanide ions.[17−19] The inclusion of 3d metals in proximity to 4f ions establishes magnetic exchange pathways which can act to suppress the quantum tunnelling of magnetization, a common pitfall of purely 4f SMMs, and increase Ueff.[19−21]

A combination of three transition metals and one lanthanide ion have, in the past, been used to template the macrocyclization of 2,3-dihydroxybenzene-1,4-dicarbaldehyde with various diamine linkers.[22−26] The Plieger group have previous experience in preparing multinuclear metal complexes with potentially interesting magnetic properties,[27−30] and the Ishida group have expertise in the magnetic characterization of 3d–4f complexes, particularly those involving copper(II).[31−35] Inspired by the Zn3La system first reported by Nabeshima et al.[22] and later extended to other metal combinations by Brooker et al.,[24,25,35−40] we investigated the self-assembly and successful crystallization of L3Cu3Ln complexes using the acyclic ligand system 1,4-diformylnaphthalene-2,3-diol (H2L). While there are numerous examples of magnetically coupled Cu–Ln systems,[41,42] throughout the literature, there are very few Cu3Ln complexes featuring four in-plane metal ions;[43−45] and a CCDC search indicated that, outside of the previously mentioned complexes of Nabeshima and Brooker, there are no other structurally characterized examples which exclusively contain Cu–O–Ln bridges. The Brooker group have reported macrocyclic Cu3Ln complexes for the entire available lanthanide family, excluding promethium. Of the numerous compounds, only variations of the macrocyclic Cu3Tb complex showed promise as SMMs. Although isolation and structural elucidation of analogous systems prior to Schiff base/macrocycle formation have been reported in the literature, magnetic characterization of such “intermediate” species is notably absent; however, we recently reported the structures and magnetic properties of four acyclic Ni3Ln complexes containing planar Ni3LnO12 cores.[30] Herein, we extend this to include a series of acyclic Cu3Ln complexes with near planar Cu3LnO12 cores and report the first magneto–structural correlations for this class of complexes (Figure ).

Figure 1

Schematic showing the general metal—ligand connectivity of the complexes reported in this work. Carbon = gray, oxygen = red, CuII = purple, and LnIII = green.

Results and Discussion

The addition of a methanolic Cu(NO3)2 solution (3 equiv.) to the ligand suspension (3 equiv.) was accompanied by rapid dissolution and a subsequent color change from a murky orange-brown to a deep red-brown solution. There were no notable changes upon addition of the lanthanide containing solution (1 equiv) nor were there any changes when the lanthanide metal was added prior to the transition metal solution. IR spectroscopy revealed a new band in the 1610–1616 cm–1 region corresponding to the coordinated aldehyde groups, and no bands corresponding to unreacted aldehyde (1673 and 1641 cm–1) were present. X-ray quality single crystals of C1–C7 were grown by vapor deposition of diethyl ether into a concentrated methanolic solution of the complex. Two complexes (C2 and C3) crystallized in the orthorhombic space group Pbcn, and the other five complexes crystallized in the monoclinic space groups (C1 in P21, and the remaining five complexes in C2/c), with π–π stacking being the dominant supramolecular interaction in all seven complexes. Selected bond lengths and angles are given in Table as ranges over all copper ions in each complex. Crystal packing diagrams and details on interactions can be found in Supporting Information, Figures S1–S8. Bulk purity was confirmed by microanalytical analysis. Continuous shape measurement (CShM) calculations[46,47] indicate that all copper ions have approximately C4v square pyramidal geometries (average CShM values for SPY-5 geometry: C1 1.01, C2 1.08, C3 1.07, C4 0.646, C5 0.649, C6 0.771, and C7 0.689). CShM values below unity correspond to complexes which crystallize in the monoclinic space group C2/c. The geometry of Cu4 in each complex also tends toward C4v vacant octahedron geometry (C1 1.13, C2 1.34, C3 1.33, C4 1.07, C5 1.22, C6 1.48, and C7 1.26).

Table 1

Selected Structural Parameters for C1–C7

Distance/Å	C1	C2	C3	C4	C5	C6	C7
LnIII–Ophenol	2.415(12)–2.531(12)	2.420(3)–2.463(3)	2.410(3)–2.462(3)	2.488(8)–2.524(6)	2.458(7)–2.517(8)	2.435(6)–2.513(6)	2.477(8)–2.524(8)
LnIII–Onitrate	2.434(20)–2.532(18)			2.415(14)–2.488(15)	2.395(13)–2.456(12)	2.305(13)–2.395(14)	2.367(10)–2.444(12)
LnIII–Owater	2.389(15)–2.513(18)	2.298(3)–2.374(3)	2.276(4)–2.362(4)
CuII–Ophenol	1.900(15)–1.935(14)	1.909(3)–1.941(3)	1.906(4)–1.948(3)	1.883(7)–1.914(8)	1.880(8)–1.907(9)	1.843(7)–1.900(8)	1.865(7)–1.921(8)
CuII–Oformyl	1.898(19)–2.014(15)	1.916(3)–1.978(3)	1.916(3)–1.978(4)	1.884(9)–1.950(9)	1.902(10)–1.930(9)	1.880(9)–1.946(9)	1.904(9)–1.946(9)
CuII–Owater			2.368(9)	2.259(16)–2.320(20)	2.246(14)–2.273(17)	2.207(17)–2.230(20)	2.283(13)–2.302(15)
CuII–Omethanol	2.212(19)	2.371(4)	2.368(9)a	2.220(20)	2.270(20)	2.378(18)	2.287(18)
CuII–Cl	2.502(7)–2.578(8)	2.478(1)–2.532(1)	2.473(2)–2.530(2)
Av. CuII···CuII	6.162(4)	6.113(1)	6.111(1)	6.225(3)	6.215(3)	6.182(3)	6.211(3)
Av. CuII···LnIII	3.562(4)	3.538(1)	3.537(1)	3.594(2)	3.589(2)	3.570(2)	3.586(2)
Min. LnIII···LnIII	9.988(2)	9.757(1)	9.752(1)	9.668(1)	9.710(1)	9.741(1)	9.675(1)

Crystallographically disordered H2O/MeOH cap with the oxygen atom occupying the same coordinate.

The crystal structure of C1 (Figure ) revealed a relatively nonplanar system with a mixture of nitrate and chloride caps. Although C1 was not the only complex prepared using a LnCl3·xH2O/Cu(NO3)2·3H2O combination, it is the only complex to feature both chloride and nitrate capping groups–as confirmed by single-crystal X-ray diffraction and elemental analyses. As expected, the large TbIII sits in the middle of an O6 equatorial binding environment formed from the three sets of catechol-like groups with its axial sites occupied by two water molecules on one side of the Cu3 plane and a η2-NO3 group on the other side of the plane. Each CuII sits in a square pyramidal coordination environment bound between a phenol and an aldehyde each from two ligand units. In the case of C1, the copper ions are axially capped by either a chloride (Cu2 and Cu3) or a methanol molecule (Cu4), with all three copper ions being capped on the same side of the Cu3 plane. While the entire ligand unit that bridges Cu2 and Cu4 sits below the Cu3 plane, the remaining two ligand units form a separate plane, and the terbium ion is displaced above the Cu3 plane by 0.162(3) Å.

Figure 2

X-ray crystal structure of C1. Thermal ellipsoids of metal atoms shown at 50% probability. Carbon = gray, oxygen = red, nitrogen = blue, and chlorine = green. Hydrogen bonds shown as a segmented yellow bond.

The two complexes which crystallized in the orthorhombic space group Pbcn, C2 (Figure ) and C3, are isomorphous and feature only chloride capping groups. The LnIII center in each of these complexes is displaced above the Cu3 plane by 0.245(1) Å, with the ligand unit bridging Cu2 and Cu3 in each sitting well below the plane. Much like C1, Cu2 and Cu3 are each axially capped by a chloride on one side of the plane, while Cu4 is axially capped below the Cu3 plane. The Cu4 cap is the only structural difference between C2 and C3, with the Cu4 cap of C2 being a methanol molecule, whereas for C3, it was crystallographically determined as a disordered one-to-one mixture of water and methanol. Holmium and erbium, being slightly smaller than terbium, occupy nine-coordinate sites in C2 and C3 each with two water caps above the plane and one below. In C1, the Cu–Cl bonds can be thought of as diverging away from the catechol moiety which bridges Cu2 and Cu3 (forming a 129.6° angle from the centroid of the two oxygen atoms, O2 and O3) allowing for a parallel arrangement of water caps on Tb1 with a Cl–Cl separation of 8.327 Å. This allows the chloride caps to each accept a hydrogen bond from a different water molecule (O14–H14A···Cl1 = 2.494 Å and O15–H15A···Cl2 = 2.047 Å). In contrast, the Cu–Cl bonds in C2 and C3 converge relative to their respective catechol O2–O3 centroid (forming 77.1 and 76.9° angles, respectively). The much smaller Cl–Cl separations (4.828 Å for C2 and 4.914 Å for C3) means the two water caps on Ho1 and Er1 run perpendicular with a single water molecule forming hydrogen bonds to both Cl1 and Cl2 (average of 2.029 Å for C2 and 2.122 Å for C3). Considering the charge balance of the elements present in the crystal structure, an additional negative charge is required per unit complex. A third, non-coordinated chloride anion was crystallographically observed in both C2 and C3 but was heavily disordered over multiple sites which were not stable to refinement, and has thus been included within identical solvent masks for each complex of 38 electrons in 98 Å3 which is consistent with the presence of one chloride and two water molecules per unit complex. This final charge being present as a chloride rather than a nitrate (as in C1) is also consistent with the nitrogen content determined by elemental analyses.

Figure 3

X-ray crystal structure of C2. Thermal ellipsoids of metal atoms shown at 50% probability. Carbon = gray, oxygen = red, nitrogen = blue, and chlorine = green. Hydrogen bonds shown as a segmented yellow bond.

Complexes C4, C5, C6, and C7 (Figure ) are isomorphous and crystallize in the monoclinic space group C2/c. They form relatively planar complexes with only minor deviations of the LnIII center and ligands from the Cu3 plane. Each LnIII center is 10-coordinate with an axially bound η2-NO3 group on both sides of the Cu3 plane, with the CuII centers of each complex being square pyramidal and all bound on the same side of the Cu3 plane. Gd1 is displaced below the Cu3 plane by 0.050(2) Å, while Dy1 and La1 are each displaced below their respective Cu3 plane by 0.053(2) Å and Yb1 is displaced below its Cu3 plane by 0.060(2) Å. Each complex contains two water caps (on Cu2 and Cu3) and a methanol cap (on Cu4), with the axial sites of each copper in C6 being poorly ordered and modeled with only 50% chemical occupancy. All four of these complexes exhibited further crystallographic disorder with the location of a non-coordinated nitrate which is required to balance all charges. This nitrate group was identified to occupy a special position on a 2-fold rotation axis in C4, C5, and C6, with 50% occupancy per unit complex. For C4, a disordered fragment of a nitrate anion was located but was only stable to refinement at 25% occupancy, with the remaining 25% occupancy being contained within a solvent mask. In C5, the remaining half occupancy nitrate was located and stable to refinement only when very hard isotropic restraints were applied. The remaining half occupancy nitrate for C6 and the one nitrate per unit complex for C7 could not be located via the difference map and thus were included in a solvent mask. The formulation of these complexes to include the non-coordinated nitrates within a solvent mask is consistent with elemental analyses.

Figure 4

X-ray crystal structure of C7. Thermal ellipsoids of metal atoms shown at 50% probability. Carbon = gray, oxygen = red, and nitrogen = blue.

Magnetic Analysis

Five of the complexes reported here (C1, C2, C4, C5, and C6) have been magnetically characterized (Table ). The other two complexes, C3 and C7, could not be prepared in sufficient quantities to facilitate magnetic characterization.

Table 2

Magnetic Properties of C1, C2, C4, C5, and C6

Complex	gJ of LnIII	χmT/cm3 K mol–1 (exp. at 300 K)a	χmT/cm3 K mol–1 (calculated)	M/NAμB (exp. at 7 T)b	Msat/NAμB (calculated)
C1	3/2	11.70	12.9	7.52	12c
C2	5/4	13.79	15.2	7.81	13c
C4	2	9.06	9.00	9.84	10
C5	4/3	14.06	15.3	7.98	13c
C6	8/7	3.63	4.39	4.39	7c

Measured at 500 Oe, fixed with eicosane.

Measured at 1.8 K, fixed with eicosane.

Calculated for strongly anisotropic LnIII ions where |J| ≈ J.

DC Magnetic Measurements

It was found that upon decreasing the temperature, C2 and C6 both exhibited a monotonic decrease in the χmT product (Figure a) which may be associated with the progressive depopulation of the Stark sublevels of each lanthanide ion. This suggests an apparent weak antiferromagnetic coupling between the CuII and LnIII ions, which is further evidenced by the M – H data collected (vide infra).

Figure 5

Plots of χmT vs T for (a) C2 and C6 and (b) C1, C4, and C5 measured at 500 Oe below 100 K and at 5000 Oe above 100 K.

Complexes C1, C4, and C5 all exhibited a sharp increase in the χmT product when cooled below ca. 50 K, with C1 reaching a peak of 14.6 cm3 K mol–1 at 2.2 K and C4 reaching a peak of 12.5 cm3 K mol–1 at 7.5 K (Figure b). These peaks strongly suggest ferromagnetic coupling between the LnIII and CuII centers. The M vs H measurement for C4 (Figure ) unequivocally evidences a high-spin ground state of Stotal = 10/2. The χmT vs T plot for C4 shows a monotonic increase down to 7.5 K originating from the three CuII–GdIII interactions. Quantitative separation of J-coupling terms is difficult due to the approximate C3 symmetry about the GdIII center of C4, and thus to avoid overparameterization, C4 has been approximated to feature only a single Cu–GdIII coupling term, that is, J12 = J13 = J14 = J, where J can be estimated with the following van Vleck equation (eq ) based on the spin Hamiltonian = −2J(1·2 + 1· + 1·4)where

Figure 6

Magnetization plots for (a) C2 and C6 and (b) C1, C4, and C5 (b) measured at 1.8 K.

The Weiss mean field parameter, θ, has also been included to account for intermolecular interactions. Optimization of eq gave 2J/kB = +5.78(6) K, gave = 1.976(2), and θ = −0.898(7) K. The calculated curve reproduced the trend of the experimental data well (Figures b and 6b). The final χmT drop below 7 K could not be reproduced when modeling the Cu–Cu interactions (J23, J24, and J34; see Figure S9 in the Supporting Information) with MAGPACK.[48,49]

Figure shows the magnetization curves for fixed polycrystalline samples of C1, C2, C4, C5, and C6, as measured at 1.8 K. The magnetization of C4 exhibited the theoretical saturation magnetization (Msat) with 10 NAμB, and those of C1 and C5 did not reach the theoretical Msat values at 7 T (Table ) owing to the presence of strong magnetic anisotropy. However, these values exceeded the antiferromagnetic limit values, thus supporting the presence of a ferromagnetic ground state. The magnetizations of C2 and C6 similarly exceeded the antiferromagnetic limit, but practically noncoupled systems would also show similar behavior due to the additive property. Combining the susceptibility results, we can cautiously conclude that the Ho–Cu and Tb–Cu exchange couplings in C2 and C6, respectively, are weak.

AC Magnetic Measurements

As expected from the DC measurements, only C1, C4, and C5 exhibited meaningful frequency dependence for the AC susceptibility measurements (Figure for C1 and Figure S10, Supporting Information for C1, C4, and C5). Complex C1 exhibited slow relaxation of magnetization as indicated by the appreciable frequency dependence of the AC susceptibility below ca. 5 K (Figure ). The effective barrier to the reversal of magnetization, Ueff, for C1 was estimated using a modified Arrhenius plot (Figure a) according to eq (50)

Figure 7

Temperature and frequency dependence of the AC susceptibilities in the zero applied DC field.

Figure 8

(a) Modified Arrhenius plots for the 1000 Hz data of C1 in the zero applied DC field, C4 in a 2000 Oe static field, and C5 in a 1000 Oe static field. (b) Cole–Cole plot for C1 with traces for data sets recorded at 2.0, 2.1, and 2.3 K.

The Ueff values are 10.49(8) K (zero applied bias) and 13.3(2) K (1000 Oe applied bias; Figure S11a, Supporting Information), with τ0 values of 5.02 × 10–8 and 3.1 × 10–8 s, respectively. Similarly, C4 and C5 showed appreciable frequency-dependent AC susceptibilities below 5 K with applied DC bias fields of 2000 and 1000 Oe, respectively. As Figure a shows, optimization of eq for C4 and C5 leads to Ueff values of 10.8(2) and 7.82(4) K and τ0 values of 8.8 × 10–8 and 6.05 × 10–8 s, respectively. Details on various DC bias fields are shown in Figure S11, Supporting Information.

Figure b shows the Cole–Cole plot for C1 in the presence of 1000 Oe static magnetic field at the indicated temperatures. The solid line represents the best fit obtained at the given temperature by considering the generalized Debye model[51]where χS is the adiabatic susceptibility, χT is the isothermal susceptibility, ω is the frequency, τ is the relaxation time, and α is the Cole–Cole parameter.[52] The best fit lines at 2.0, 2.1, and 2.3 K almost trace quarter circles where α = 0.24(2), 0.21(2), and 0.19(5), respectively. In this low-temperature regime, the thermal energy is of the order of the exchange-coupled excited states, thus allowing thermal population of these states and serving as a magnetization relaxation pathway, where the Orbach process may be operative. Unfortunately, a normal Arrhenius plot cannot be drawn from the data, so the operative relaxation mechanisms cannot be determined with accuracy; however, the relatively small Cole–Cole parameter values indicate that only a single relaxation process is likely to be occurring.

Table summarizes the results of the AC magnetic susceptibility measurements. In general, superexchange between the 3dy2 and 4f spins of heavy lanthanide metals depends on the geometry of the Cu–O–Ln–O four-membered chelate ring, where more planar MO2 and LnO2 dihedral angles will lead to stronger ferromagnetic coupling.[53−56] Based on the macrocyclic Cu3Ln analogues previously prepared by Brooker et al., C1, containing the Cu3TbO6 core, was expected to show the greatest potential as an SMM with the Cu3TbL complex[57] (L = 3 + 3 Schiff base of 1,4-diaminobutane and 1,4-diformyl-2,3-dihydroxybenzene) containing a 10-coordinate Tb ion reported to have a barrier of 19.5(5) K. A smaller Cu3TbL analogue[24] (L = 3 + 3 Schiff base of 1,3-diaminopropane and 1,4-diformyl-2,3-dihydroxybenzene), which contained a nine-coordinate Tb ion, was also determined to be an SMM although a barrier height could not be extracted. C1, being a Kramers-type molecule, indeed shows the most promising SMM properties as indicated by the magnitude of χ′ and its frequency dependence. C4 and C5 also showed appreciable frequency dependence in an applied field but yielded lower effective barriers, which may be related to the fact they are not Kramers-type molecules.[32,58] Since GdIII itself is an isotropic ion, the origin of the slow relaxation of magnetization for C4 is unclear; however, the observed anisotropy could arise from an asymmetric crystal field environment where J is not equal to J and J.[59]

Table 3

Summary of the AC Susceptibility Measurements

Complex	H/Oe	τ0/10–8 s	Ueff/kB K
C1	0	5.02(14)	10.49(8)
	1000	3.1(2)	13.3(2)
	2000	3.6(3)	13.2(3)
C4	2000	8.8(6)	10.8(2)
C5	1000	6.05(11)	7.82(4)

Conclusions

The reaction between the ligand, H2L, and Cu(NO3)·3H2O with various lanthanide salts has yielded seven new complexes which have been characterized by single-crystal X-ray diffraction. These self-assembled metallo–cyclic complexes had previously been overlooked in favor of macrocyclic analogues where peripheral diimine chelators are typically used to stabilize complexes involving a transition metal ion with five- or six-membered chelation rings. Macrocyclic ligand formation was found to be unnecessary in the construction of the self-assembled tetranuclear clusters in the present system. The compartmental ligand in this work being acyclic in nature is in sharp contrast to the work by Nabeshima et al.[8] Complexes C2 and C3 were both prepared using LnCl3·6H2O salts [Ln = Ho (C2) and Er (C3)] and were found to be isomorphous containing only the chloride ions present from the lanthanide salt. Similarly, complexes C4–C7 are all isomorphous in the monoclinic space group C2/c, containing only nitrate ions despite the presence of the acetate from Yb(OAc)3·6H2O in the preparation of C6. Five of the complexes (C1, C2, and C4–C6) were prepared in sufficient quantity and purity to elicit magnetic characterization. DC susceptibility measurements implied a ferromagnetic ground state in three of the complexes (C1, C4, and C5). Furthermore, the three compounds exhibited appreciable frequency dependence in their AC magnetic measurements. Macrocyclic analogues indicated that the TbIII-centered complex was likely to exhibit SMM properties; however, observation of the slow relaxation of magnetization in monomeric Cu3Ln (Ln = Gd or Dy) complexes of this type had yet to be reported. Fitting of the AC data yielded small values of the Cole–Cole parameter, α, implying that only a single relaxation pathway is present for the three SMM candidates. Complex C1, containing TbIII, is a Kramers-type molecule and had an observable barrier to relaxation even with zero applied bias, with the maximum barrier determined to be 13.3(2) K in a 1000 Oe applied field. Complexes C4 and C5 are not Kramers-type molecules but still yielded meaningful frequency dependence in their AC susceptibility measurements. The Gd-containing complex, C4, was found to have a barrier to relaxation of 10.8(2) K in a 2000 Oe applied bias, while C5, the Dy containing complex, had a barrier energy of only 7.82(4) K in a 1000 Oe applied bias. Observation of the slow relaxation of magnetization in this type of Cu3Ln complex which utilizes an acyclic ligand system has not yet been reported in the literature. Investigation of analogous systems using 3d metals with a greater number of unpaired electrons such as MnII/III, FeIII, or CoIII could yield even more promising results.

Experimental Section

General Experimental Section

All experiments were carried out in air. Solvents and reagents were used as received from commercial sources (Merck/Sigma-Aldrich, ThermoFisher) without further purification. The synthesis of ligand H2L has recently been reported[30] and was adapted from the method originally reported by MacLachlan et al.[60] IR spectra were collected on a Nicolet 5700 FT-IR spectrometer with an ATR sampling accessory. Elemental analyses were determined by the Campbell Microanalytical Laboratory at the University of Otago.

Single-crystal X-ray diffraction experiments were carried out on a Bruker D8 Venture diffractometer equipped with an IμS Diamond microfocus Cu Kα source (λ = 1.54187 Å) and a Photon III detector. Single crystals were mounted on MiTeGen mylar loops using Fomblin Y perfluoroether oil and cooled to 100 K with an Oxford Cryostream 800. Data were collected and processed using the APEX3 software package.[61] The structures were solved by intrinsic phasing with SHELXT[62] and refined against least-squares using SHELXL[63] as implemented in Olex2.[64] Non-hydrogen atoms were refined anisotropically, and hydrogen atoms were calculated to their ideal positions unless otherwise stated and refined using a riding model with fixed Uiso values. Crystal structure and refinement details for C1–C7 can be found in Table S1, Supporting Information. CCDC deposition numbers 2103942–2103948.

Magnetization and magnetic susceptibility measurements were carried out with a Quantum Design MPMS-XL7 SQUID magnetometer using a static field of 0.05 T. A field of 0.5 T was used to measure the susceptibility of samples with small magnetic moments. Measurements were corrected with diamagnetic blank data from the empty sample holder. The diamagnetic contribution of the sample itself was estimated using Pascal’s constants. The AC magnetic susceptibilities were recorded on a Quantum Design PPMS apparatus equipped with an AC/DC magnetic susceptibility option. To avoid possible field-alignment effect, the polycrystalline samples were fixed with a small amount of eicosane.

General Procedure for the Preparation of Complexes C1–C7

Cu(NO3)2·3H2O (121 mg, 0.5 mmol) was reacted with H2L (108 mg, 0.5 mmol) and the corresponding lanthanide salt (0.167 mmol) in methanol (10 mL) in a 3:3:1 equiv. Each mixture was stirred for 24 h and then filtered, and diethyl ether was allowed to passively diffuse into the methanolic solutions.

C1 [L3Cu3TbCl2(NO3)(H2O)2]

Orange blocky crystals, 77% yield based on TbCl3·6H2O. ATR-IR ν̅: 1612, 1569, 1485, 1450, 1432, 1382, 1351, 1314, 1259, 1141, 1030, 975, 756, 718 cm–1. Microanalytical data found (calcd) for C1 (calculated for [L3Cu3TbCl2(NO3)(H2O)2], Mr = 1161.03 g mol–1) C, 37.19(37.24); H, 1.71(1.91); N, 1.22(1.21).

C2 [L3Cu3HoCl3(H2O)4(MeOH)](H2O)

Orange blocky crystals, 56% yield based on HoCl3·6H2O. ATR-IR ν̅: 1611, 1568, 1486, 1433, 1382, 1352, 1313, 1259, 1141, 1029, 975, 755, 717 cm–1. Microanalytical data found (calcd) for C2 (calculated for [L3Cu3HoCl3(H2O)4MeOH], Mr = 1208.55 g mol–1) C, 36.47(36.77); H, 2.38(2.50).

C3 [L3Cu3ErCl3(H2O)3.5(MeOH)0.5](H2O)

Orange platelike crystals, 84% yield based on ErCl3·6H2O. ATR-IR ν̅: 1612, 1567, 1484, 1428, 1381, 3152, 1310, 1264, 1143, 1027, 981, 756, 710 cm–1. Microanalytical data found (calcd) for C3 (calculated for [L3Cu3ErCl3(H2O)4.5MeOH0.5], Mr = 1203.87 g mol–1) C, 36.70(36.42); H, 2.48(2.43).

C4 [L3Cu3Gd(NO3)2(H2O)2(MeOH)](NO3)

Orange platelike crystals, 88% yield based on Gd(NO3)3·6H2O. ATR-IR ν̅: 1616, 1566, 1485, 1429, 1382, 1351, 1314, 1258, 1140, 1029, 975, 756, 718 cm–1. Microanalytical data found (calcd) for C4 (calculated for [L3Cu3Gd(NO3)3(H2O)2MeOH], Mr = 1244.50 g mol–1) C, 36.07(35.71); H, 1.97(2.11); N, 3.59(3.38).

C5 [L3Cu3Dy(NO3)2(H2O)2(MeOH)](NO3)

Orange platelike crystals, 47% yield based on Dy(NO3)3·6H2O. ATR-IR ν̅: 1611, 1568, 1486, 1432, 1382, 1351, 1313, 1259, 1141, 1030, 974, 755, 711 cm–1. Microanalytical data found (calcd) for C5 (calculated for [L3Cu3Dy(NO3)3(H2O)2MeOH], Mr = 1249.75 g mol–1) C, 35.72(35.56); H, 1.97(2.10); N, 3.31(3.36).

C6 [L3Cu3Yb(NO3)2(H2O)2(MeOH)](NO3)

Orange blocky crystals, 44% yield based on Yb(OAc)3·6H2O. ATR-IR ν̅: 1610, 1567, 1488, 1434, 1383, 1355, 1316, 1260, 1143, 1032, 975, 752, 699 cm–1. Microanalytical data found (calcd) for C6 (calculated for [L3Cu3Yb(NO3)3(H2O)2(MeOH)], Mr = 1260.30 g mol–1) C, 35.08(35.26); H, 1.90(2.08); N, 3.36(3.33).

C7 [L3Cu3La(NO3)2(H2O)2(MeOH)](NO3)

Orange platelike crystals, 28% yield based on La(NO3)3·6H2O. ATR-IR ν̅: 1614, 1566, 1488, 1434, 1387, 1351, 1317, 1261, 1144, 1030, 975, 751, 710 cm–1. Microanalytical data found (calcd) for C7 (calculated for [L3Cu3La(NO3)3(H2O)2MeOH], Mr = 1226.15 g mol–1) C, 36.40(36.24); H, 1.80(2.14); N, 3.41(3.43).