Formation of Defect-Dicubane-Type NiII 2LnIII 2 (Ln = Tb, Er) Clusters: Crystal Structures and Modeling of the Magnetic Properties.

In the field of molecular nanoclusters, cubane and defect-dicubane, or butterfly structures, are typical examples of tetranuclear metal core architectures. In this work, a halogenated and anionic Schiff-base ligand (L2-) is utilized as it is predisposed to chelate within a cluster core to both 3d and 4f metal ions, in different binding configurations (H2L = 4-chloro-2-(2-hydroxy-3-methoxybenzyliden amino)phenol). The phenolate oxygen atoms of the deprotonated ligand can act in μ-O and μ3-O bridging binding modes for the intramolecular assembly of metal ions. Based on that, two tetranuclear and isostructural compounds [Ni2Tb2(L)4(NO3)2(DMF)2]·2CH3CN (1) and [Ni2Er2(L)4(NO3)2(DMF)2]·0.5CH3CN (2) were synthesized and structurally characterized. Magnetic susceptibility and magnetization data indicate the occurrence of dominant intramolecular ferromagnetic interactions between the spin centers. Particular emphasis is given to the theoretical description of the magnetic behavior, taking into account the Ln-Ni and Ni-Ni coupling paths and the magnetic anisotropy of the LnIII and NiII ions. The study is distinguished for its discussion of two distinct models, whereby model A relies on the uniaxial B 20 Stevens term describing the lanthanide anisotropy and model B is based on point-charge model calculations. Importantly, the physical meaning of the obtained parameters for both models was critically scrutinized.

Introduction

In the field of molecular magnetism,[1−7] heteronuclear nanoclusters comprising 3d and 4f metal ions are attractive candidates for studying the factors that govern the strength and type of intracluster magnetic coupling between the metal ions.[8−30] The inherent anisotropy of the lanthanide ions and their large magnetic moments give reason to combine them with 3d metal ions, which are expected to exhibit stronger magnetic couplings. In addition, in recent decades, synthetic methods for polynuclear systems have reached a level of efficiency attained with mononuclear compounds, and consequently, the factors governing a specific core structure can often be elucidated.[31] Accordingly, a large diversity of heteronuclear structures of cluster compounds has been reported in the literature.[8−30] As a case in point, a class of π-conjugated Schiff-base ligands has been designed to be particularly suitable for assembling metal ions into a “butterfly” or defect-dicubane core structure.[24] An important property of such ligands is the availability of different coordination pockets, each of which can act as a chelating unit. Therefore, the ligands are predisposed to bond to both transition metal ions and lanthanide ions. Regarding the analysis of the magnetic properties of polynuclear coordination compounds, magneto-structural correlations are sought in order to gain insights into the pathways of the magnetic interactions. In the context of this work and to cite one example, such a correlation, which involves NiII ions, is given by relating the Ni–O–Ni angles formed via doubly bridged or triply bridged oxygen atoms within a cubane-type structure. For a value of this angle above 98–99°, the interaction between the NiII ions is mostly found to be of antiferromagnetic character but of ferromagnetic in the case of smaller angles.[32−34] However, one must also take into account that structural distortions within such a cluster can substantially affect these kinds of guidelines.[35]

In this paper, we report the synthesis, structure, and magnetic characterization of two tetranuclear and isostructural compounds with the stoichiometries [Ni2Tb2(L)4(NO3)2(DMF)2]·2CH3CN (1) and [Ni2Er2(L)4(NO3)2(DMF)2]·0.5CH3CN (2). Four metal ions are assembled into a defect-dicubane cluster core by four halogenated and anionic Schiff-base ligands L2– (Chart ), while nitrate anions and solvent DMF molecules complete their coordination spheres. For both complexes, the magnetic susceptibility and magnetization data were experimentally determined and modeled with a critical view on the physical significance of the obtained magnetic parameters.

Chart 1

Chemical Structure of the Ligand H2L: Its Deprotonated Form L2– Ligates in the NiII and LnIII Ions of the Cluster Core

Results and Discussion

Synthesis

The Schiff-base ligand was prepared by a condensation reaction of the corresponding aldehyde and amine in methanol. The reaction of the Schiff-base ligand with nickel acetate or nickel nitrate and rare-earth nitrate salts in a mixture of MeOH, MeCN, and DMF, in the presence of triethylamine, leads to [Ni2Ln2(L)4(NO3)2(DMF)2] stoichiometry for the cluster compounds. Solvent DMF molecules and nitrate anions were found as terminal ligands in the coordination environment (vide infra). The IR data of the compounds are in agreement with their X-ray structures. The free Schiff-base features an intense band at 1638 cm–1, which is attributed to the stretching vibration of the azomethine [ν(C=N)] group.[36] As a coordinated ligand, this stretching frequency appears at 1608 cm–1. This downshift of the frequency indicates the coordination of the imino nitrogen to the metal atoms.

Structural Descriptions of the Complexes

The heterometallic complexes 1 and 2 crystallize in the monoclinic space groups P21/n and P21/c, respectively. The two isostructural compounds contain a centrosymmetric Ni2Ln2O6 cluster core with a defect-dicubane-type structure. The crystal structure of complex 1 contains two acetonitrile molecules of crystallization, while complex 2 has half of an acetonitrile molecule. The metal ions are connected by six phenolate oxygen atoms from four anionic Schiff-base ligands L2–, exhibiting four μ-O and two μ3-O binding modes (Figure ). The bridging Ni–O–Ni angles for 1 and 2 are 97.0 and 97.1° for 1 and 2, respectively. Figure illustrates the different binding modes of the deprotonated ligands with the metal ions. Two symmetry-independent ligands L2– chelate in an almost coplanar fashion to a NiII and a TbIII ion from opposite sides. The O^N^O coordination pocket of one ligand binds to the NiII ion, and the same pocket of the opposite ligand binds to the TbIII ion. The TbIII coordination sphere is completed by the O^O coordination pocket of one ligand, which remains empty on the other ligand. Two of these fragments, related by a center of symmetry, form the cluster core (Figure ). In the cluster, the NiII ion resides in a slightly distorted octahedral NO5 coordination geometry. The coordination sphere comprises four phenol oxygen atoms, one nitrogen atom from the chelating ligands, and one terminal monodentate DMF ligand. The LnIII ion shows a distorted square antiprismatic NO7 coordination geometry, formed by phenol oxygens, aminophenol oxygens, a methoxy oxygen, and a peripheral chelating NO3– ligand. Figure shows the cluster of 1 (for 2, see Figure S1), and selected bond lengths and angles for 1 and 2 are given in Table S1. Within the cluster, the aminophenol oxygens (O5) of two ligands L2– form the two μ3-O bridges, each over a Ni2Ln triangle, and the phenol oxygens (O4) of the same Schiff-base ligate, each in a monodentate manner, to the two LnIII ions. The methoxy oxygens (O6) of these two ligands remain nonbonding. The aminophenol oxygens (O2) together with the phenol oxygens (O1) of the other two ligands form the four μ-O bridges, each linking a NiII to a LnIII ion, and their methoxy oxygens (O3) bind monodentately to a LnIII ion. The crystal packing of compounds 1 and 2 shows no special feature, and due to the bulky ligand shell around the Ni2Ln2O6 core, the spin centers of neighboring molecules are quite distant (>9.5 Å), which minimizes any intermolecular magnetic coupling.

Figure 1

Ni2Ln2O6 defect-dicubane core of 1 (Ln = Tb) and 2 (Ln = Er). Alternatively, the core structure can be described as a butterfly topology, where the two NiII ions form the body and the two LnIII ions are in the wing positions (O, red; Ni, dark green; Tb, light blue).

Figure 2

Fragment of complex 1 emphasizing the two symmetry-independent ligands L2– with different binding modes for the NiII and TbIII ions (O, red; N, blue; Cl, light green; Ni, dark green; Tb, light blue).

Figure 3

ORTEP structure of complex 1 drawn with 30% ellipsoid probability. Atoms labeled with the suffix A are related to the symmetry position −x, 1 – y, −z. Hydrogen atoms and the acetonitrile molecules are omitted for clarity.

Magnetic Properties

The temperature dependence of the magnetic susceptibility of complexes 1 and 2 is shown in Figures a and 5a. At room temperature, the χMT products of 1 and 2 of 27.0 and 24.7 cm3 K mol–1 are in good agreement with the calculated values for two independent NiII ions (S = 1 with gNi = 2) and two TbIII ions (7F6) or two ErIII ions (4I15/2) of 25.6 and 25.0 cm3 K mol–1, respectively.[3] For both compounds, the χMT product initially does not change markedly with decreasing temperatures. Below ca. 10–20 K, a sharp increase in the χMT products is observed, which increase to 66.3 and 38.3 cm3 K mol–1, respectively, for 1 and 2, at 1.9 K. This observation is indicative of intramolecular ferromagnetic interactions between the paramagnetic centers in each molecule. The field dependence of the magnetization at 1.9 K is shown for both compounds in Figures b and 5b. Up to a field of ca. 10 kOe, the magnetization shows a rapid increase, continues at higher fields with a slight almost linear increase, and at 50 kOe reaches values of 14.2 and 13.0 μB for 1 and 2, respectively, without fully saturating. Assuming that the dominant magnetic coupling is given by the dimeric NiII2 subunit and further considering that the increase in the χMT product occurs at very low temperatures, the strength of the corresponding coupling constant JNi can be of the order of only a few cm–1 at most.[37] The bridging Ni–O–Ni angles of 97.0 and 97.1° for 1 and 2, respectively, are in the range where the interaction is expected to have a ferromagnetic character, in agreement with the experimental finding.[35]

Figure 4

(a) Temperature dependence of the χMT product and (b) field dependence of the magnetization at 1.9 K for compound 1. The experimental data are shown as black solid circles. The results of simultaneous fits to the data using model A are shown as red solid lines. The model parameters are JNi = 207(14) K, JLn = 0.67(2) K, DNi = 0.4(5) K, and χ2 = 85.1.

Figure 5

(a) Temperature dependence of the χMT product and (b) field dependence of the magnetization at 1.9 K for compound 2. The experimental data are shown as black solid circles. The results of simultaneous fits to the data using model A are shown as red solid lines. The model parameters are JNi = −1.0(5) K, JLn = 0.15(2) K, DNi = −2.4(4) K, and χ2 = 86.1.

The following models were considered in the analysis of the experimental magnetic data: The lanthanide magnetic moments were described by J-multiplets. The interactions involving them can thus be described by the standard Heisenberg model. The models further assume two different interaction strengths for the Ln–Ni and Ni–Ni coupling paths. The Ln–Ln interaction, due to the large separation between the ions, is expected to be small and was therefore ignored. This approximation may not be fully justified since the dipole–dipole interactions are long-range and found to be of significance in some cases for LnIII ions. However, this would introduce an additional parameter, of which there are already too many in the models. The coupling scheme is depicted in Figure . Various models were considered, which differed in the description of the magnetic anisotropy of the LnIII and NiII ions. However, we eventually settled on two models, which were exploited in detail. In model A, it is assumed that the uniaxial B20 Stevens term is the only nonzero parameter describing the lanthanide anisotropy. The model then readsHere, the subscripts 1 and 2 denote the NiII spins (S = 1) and 3 and 4 the lanthanide magnetic moments (J = 6 or 15/2 for the TbIII and ErIII magnetic moments, respectively). In the actual fitting, the model was simplified further by assuming that the B20 parameter is large. This effectively means that the anisotropy of the lanthanide magnetic moment is either highly Ising-like or highly XY-like.

Figure 6

Assumed exchange coupling scheme for the magnetic models discussed in the text.

In the second model B, the choice of Stevens terms for describing the lanthanide anisotropy was inspired by parameters arising from point-charge model calculations (vide infra). The anisotropy of the NiII ions was set to zero for simplicity. Model B can then be written asIn the actual fitting, it was further assumed that B20 = B22, as also suggested by the point-charge model calculations.

In order to advance our understanding of the anisotropy of the lanthanide magnetic moment in these clusters, point-charge model (PCM) calculations of the ligand-field parameters were performed using in-house software. It is important to stress that these calculations did not aim at yielding quantitative values for the parameters of the single-ion lanthanide Hamiltonian; the PCM is usually not sufficiently accurate for such an attempt. Instead, the sole purpose of these calculations was to obtain generic information about the single-ion lanthanide spectrum and, most importantly, to hopefully identify the most relevant Stevens terms and to exclude those that do not contribute significantly. The goal of the effort was of course to overcome issues with overparametrization.

The results of the PCM calculations are expressed in terms of what we call the bare ligand-field parameters, Ωkl, which are proportional to the usual Stevens parameters, Bkl, but are dependent on only the ligand field and not on the type of the lanthanide ion. The relationship between the Bkl and Ωkl parameters is given as followsHere, θk are the ion-dependent Stevens factors, frequently labeled in tables as α, β, and γ for k = 2, 4, and 6, respectively.[38] ⟨rk⟩ are the ion-dependent radial averages (constants) of the f-electron wavefunction,[38] and a0 is the Bohr atomic radius.

It is important to note that given the same ligand environment of the TbIII and ErIII ions (as it can be assumed with some approximation for isostructural compounds) and due to the opposite signs of their ionic α parameters, the resulting B2 parameters for these ions will be of opposite signs for equal values of the bare ligand-field parameters Ω2 (αTb = −0.0101 and αEr = 0.0025, see Table 1.4 in ref (41)). Thus, if the anisotropy is of Ising-type for TbIII, then it is expected to be of XY-type for ErIII and vice versa. The PCM assumed the local structure around the LnIII ions based on the X-ray crystal structure data. The oxygens were modeled by charges of −2e and the nitrogens by charges of −3e. This is certainly not a fully realistic model, but it is a reasonable attempt at getting insights into the generic trends. By means of varying the charge of the nitrogens systematically, it was confirmed that the conclusions below are representative in the sense that they are not affected by the assumed charge value.

The result of these calculations is a complex anisotropy scheme with all the possible 27 Stevens parameters present. However, for both the ErIII and TbIII cases, the largest contributions (compared to other parameters of the same order, for example, Ω65 compared to Ω60, and so on) are the terms Ω20 ≈ Ω22 = 2300 K, Ω43 ≈ −1000 K, and Ω65 ≈ 16 K for ErIII, with similar values for TbIII (all calculated values are listed in Table S2). This finding suggested model B.

For both models A and B, least-squares fits were performed using in-house software, which simultaneously included the experimental magnetization data at T = 1.9 K and the magnetic susceptibility data. For model A, the parameter B20 was fixed to a large value of 1500 and −1500 K for TbIII and ErIII, respectively. The reason for this and the expected opposite sign of B20 has been discussed above. All other possible combinations for the sign (++, –+, and −−) were also tested but yielded worse results. The fits were reasonably fast due to the uniaxial nature of the model, which significantly simplifies the numerical averaging required for simulating powder samples (one fit by model A takes about four days on a modern personal computer). The best fits to the magnetic susceptibility and magnetization data using this model are shown in Figures and 5. A modest agreement with experimental data is observed. While the fits for the two compounds are of the same quality, χ2 ≈ 85 in both cases, the best-fit parameters for the NiII ions differ markedly and assume an unrealistically large ferromagnetic exchange value JNi = (207 ± 14) K in the TbIII case and a too small value JNi = (−1 ± 0.5) K in the ErIII case. The large difference in the obtained JNi is noteworthy since it is not expected to differ much for the two compounds. For both compounds, reasonable and roughly consistent JLn values were obtained, JLn = (0.67 ± 0.02) K for the TbIII case and JLn = (0.15 ± 0.02) K for the ErIII case. The anisotropy (D) of the NiII ion is zero within the uncertainty given by the fit for the TbIII case, which can be considered acceptable given that the magnetic response is dominated by the larger lanthanide magnetic moments. For the ErIII case, D = (−2.4 ± 0.4) K was obtained. Given such differences in the parameters between the two isostructural compounds, one should not accept these results as reflecting the physical truth. It would appear more likely that two basically unrelated minima are found in a high-dimensional parameter space for the two compounds. As John von Neumann put it: “With four parameters I can fit an elephant, and with five I can make him even wiggle his trunk.”

With regard to model B, the biaxial nature of the model and the required detailed powder averaging resulted in exceptionally long fit times of about 8 weeks per fit. The obtained best fits are shown in Figures and 8. It is obvious that the best fits by model B are significantly worse than those by model A. The large χ2 ≈ 130 for both the TbIII and ErIII compounds confirms this. Model B appears to be able to reasonably reproduce the magnetic susceptibility; however, the fits for the magnetization data are systematically off. In the TbIII case, a small ferromagnetic exchange JNi = (0 ± 2) K and a reasonable JLn = (0.9 ± 0.1) K were obtained. The best-fit Stevens parameters for the TbIII compound are B20 = B22 = (6 ± 1)·10–2 K and an enormously large B65 = (−780 ± 70)·10–6 K. For the ErIII case, the best-fit exchange couplings are JNi = (−34 ± 1) K and JLn = (3.0 ± 0.3) K, and the best-fit Stevens parameters are B20 = B22 = (25 ± 3)·10–2 K and B65 = (3 ± 1)·10–6 K. One again faces the situation that the fit parameters do not resemble one another for the two compounds, and the impression of “fitting an elephant” holds for fits by model B as well.

Figure 7

Results of the simultaneous fits by model B to (a) magnetic susceptibility and (b) magnetization data taken at 1.9 K for TbIII compound 1 (black solid circles = experimental data; red solid lines = fits). The parameters are given in the text.

Figure 8

Results of the simultaneous fits by model B to (a) magnetic susceptibility and (b) magnetization data taken at 1.9 K for ErIII compound 2 (black solid circles = experimental data; red solid lines = fits). The parameters are given in the text.

In order to understand the results better, the powder averaged magnetic susceptibility and low-temperature magnetization curves were calculated assuming a single-ion lanthanide cluster with only nonzero Ω20 and Ω22 parameters (which are proportional to the Stevens parameters B20 and B22, respectively). The results for the magnetization curves at T = 2 K are shown for TbIII and ErIII in Figures and 10, respectively, for various ratios of Ω22/Ω20. It can clearly be seen that for the cases TbIII with Ω20 > 0 and ErIII with Ω20 < 0, which corresponds to B20 < 0 and thus Ising-type anisotropy for both ions, the details of the magnetization curves are not sensitive even to the quite drastic changes of the ratio Ω22/Ω20 and appear to show a similar, generic behavior. This is somewhat expected given that in both cases, the Ising-type anisotropy results in ground states that are largely insensitive to orthorhombic anisotropies. Interestingly, also for the ErIII case with Ω20 > 0, only a weak sensitivity of the Ω22 parameter is obtained. Only for the TbIII case with Ω20 < 0 is a pronounced effect on Ω22 found. These unfortunate findings are obviously an effect of the powder averaging, which smoothens out any features that would normally be expected from single-crystal data. It is, however, also an effect of the insensitivity of magnetic data to details of the ground and excited states. Given that the above analysis indicates that Ω20 > 0 is evident in complexes 1 and 2, it is thus not surprising that the anisotropy parameters cannot be very well determined based on the powder magnetic data alone. On the other hand, one would then also expect that these magnetic data can be accurately described by simple models and a small parameter set. Surprisingly, this does not appear to be the case. Further investigations into this seeming contradiction should be of much interest.

Figure 9

Single-ion magnetization calculations at T = 2 K for a TbIII ion assuming (a) positive and (b) negative signs of Ω20 and different ratios of Ω22/Ω20 (|Ω20| = 100,000 K).

Figure 10

Single-ion magnetization calculations at T = 2 K for an ErIII ion assuming (a) positive and (b) negative signs of Ω20 and different ratios of Ω22/Ω20 (|Ω20| = 100,000 K).

Conclusions

A halogenated Schiff-base ligand L2– was utilized to assemble NiII and LnIII (Ln = Tb, Er) ions into a heteronuclear defect-dicubane-type structure. Magnetic properties of both isostructural cluster compounds were determined, and steps have been taken to carefully elucidate them by modeling the experimental data with different approaches. Point-charge model calculations for the lanthanide ligand-field parameters and fits for the magnetization and magnetic susceptibility data were performed. The point-charge model suggests a complex anisotropy of the lanthanide ions. Selecting the few most dominant Stevens parameters inspired model B, which in this sense could be viewed as a sort of a “realistic” model. In addition, model A was studied, which cannot be justified from the actual ligand-field environment present in the studied clusters, but was introduced merely for its simplicity and the underlying idea that the TbIII ions might show a strong Ising anisotropy. Interestingly enough, fits by model B resulted in worse agreement with the experimental data than the simple model A. Normally, one would expect that the more parameters one uses in a fit, the better agreement with the data is obtained. Curiously, this does not seem to hold. It was emphasized that the large scatter in the best-fit parameters puts the physical significance of the obtained parameter values into question. Obviously, as also demonstrated by additional calculations, even coarse aspects of the anisotropy in lanthanide-containing magnetic molecules can often be washed out nearly completely in powder samples. While the effect itself is of course not surprising, the severity of the effect is somewhat surprising. Magnetization data taken on small single crystals, when available, would go a long way to discern details of the anisotropy.

Experimental Section

Materials

Starting materials, reagents, and solvents were purchased from commercial suppliers with AR grade and used as received.

General Methods

Elemental analyses were performed on a PerkinElmer 240C elemental analyzer. IR spectra were recorded on a Jasco FT/IR-4000 spectrometer as KBr pellets in the 4000–400 cm–1 region. UV–vis spectra were recorded on a PerkinElmer Lambda 900 spectrometer. 1H NMR and 13C NMR were performed with a Bruker 500 MHz spectrometer. Single-crystal structures were determined on a Bruker D8 Venture single-crystal diffractometer.

Synthesis of 4-Chloro-2-(2-hydroxy-3-methoxy benzyliden amino)phenol (H2L)

3-Methoxysalicylaldehyde (0.15 g, 1.0 mmol) and 2-amino-4-chlorophenol (0.14 g, 1.0 mmol) were mixed in methanol (50 mL). The mixture was stirred for 30 min at reflux, and the solvent was evaporated by distillation. The solid was recrystallized from methanol to give an orange crystalline product. Yield: 92%. Elemental analysis (%) calcd. for C14H12ClNO3: C, 60.55; H, 4.36; N, 5.04. Found: C, 60.41; H, 4.27; N, 5.13. IR data (cm–1): 1638 (vs), 1505 (s), 1435 (w), 1347 (w), 1247 (m), 1210 (s), 1105 (w), 1068 (w), 1014 (w), 972 (w), 906 (w), 739 (m), 568 (m), 514 (m). UV–vis data in methanol (λ (nm), ε (L·mol–1·cm–1)): 263, 3.12 × 10;[4] 298, 1.50 × 10;[4] 345, 1.20 × 10;[4] 450, 2.45 × 10.[3]1H NMR (500 MHz, DMSO-d6): 13.65 (s, 1H, OH), 10.01 (s, 1H, OH), 8.98 (s, 1H, CH=N), 7.48 (s, 1H, ArH), 7.20 (d, J = 7.8 Hz, 1H, ArH), 7.16 (d, J = 8.6 Hz, 1H, ArH), 7.10 (d, J = 7.9 Hz, 1H, ArH), 6.96 (d, J = 8.7 Hz, 1H, ArH), 6.88 (t, J = 7.9 Hz, 1H, ArH), 3.81 (s, 3H, OCH3). 13C NMR (126 MHz, DMSO-d6): 162.94, 151.38, 150.05, 148.07, 135.91, 127.29, 123.95, 123.09, 119.21, 119.14, 118.18, 117.75, 115.56, 55.87.

Syntheses of the Complexes 1 and 2

The Schiff-base ligand H2L (0.14 g, 0.5 mmol) was dissolved in MeOH (7.5 mL), MeCN (2.5 mL), and DMF (10 mL). Then, triethylamine (0.10 g, 1.0 mmol) was added to the solution, which was stirred at room temperature for 10 min. Then, a methanolic solution (5 mL) of Ni(NO3)2·6H2O (0.15 g, 0.5 mmol) and a methanolic solution (5 mL) of Ln(NO3)3·6H2O (0.5 mmol; Ln = Tb for 1, Er for 2) were added. The reaction mixture was further stirred for 30 min at room temperature to afford a clear brownish-green solution. The solution was allowed to slowly evaporate at room temperature for a period of a week, while well-shaped single crystals of the respective complexes were formed and collected by filtration.

[Ni2Tb2(L)4(NO3)2(DMF)2]·2CH3CN (1)

Yield: 37%. Elemental analysis (%) calcd. for C66H60Cl4N10Ni2O20Tb2: C, 41.94; H, 3.20; N, 7.41. Found: C, 41.72; H, 3.28; N, 7.37. IR data (KBr, cm–1): 3440 w, 3060 w, 2932 w, 2839 w, 1666 s, 1608 s, 1546 m, 1475 sh, 1447 m, 1452 sh, 1382 s, 1328 w, 1275 m, 1224 s, 1175 m, 1113 s, 1084 m, 1023 w, 973 m, 912 s, 855 m, 822 sh, 783 w, 740 s, 679 m, 596 m, 521 m, 492 m, 446 w. UV–vis data in methanol (λ, ε): 235 nm, 2.71 × 104 L·mol–1·cm–1; 309 nm, 1.16 × 104 L·mol–1·cm–1; 360 nm, 8.91 × 103 L·mol–1·cm–1; 425 nm, 9.73 × 103 L·mol–1·cm–1.

[Ni2Er2(L)4(NO3)2(DMF)2]·0.5CH3CN (2)

Yield: 33%. Elemental analysis (%) calcd. for C63H55.5Cl4Er2N8.5Ni2O20: C, 41.00; H, 3.03; N, 6.45. Found: C, 41.06; H, 3.14; N, 6.34. IR data (KBr, cm–1): 3450 w, 3063 w, 2934 w, 2836 w, 1663 s, 1608 s, 1550 m, 1477 sh, 1448 m, 1384 s, 1329 w, 1274 m, 1224 s, 1178 m, 1108 s, 1085 m, 1027 w, 971 m, 911 s, 861 w, 824 sh, 782 w, 737 s, 679 m, 595 m, 521 m, 445 w. UV–vis data in methanol (λ, ε): 235 nm, 2.56 × 104 L·mol–1·cm–1; 305 nm, 1.23 × 104 L·mol–1·cm–1; 375 nm, 9.37 × 103 L·mol–1·cm–1; 415 nm, 9.03 × 103 L·mol–1·cm–1.

General X-ray Crystallography

Diffraction intensities for complexes 1 and 2 were collected at 298(2) K using a Bruker D8 Venture diffractometer with Mo Kα radiation (λ = 0.71073 Å). The collected data were reduced with SAINT,[39] and multiscan absorption correction was performed using SADABS.[40] Structures of the complexes were solved by direct methods and refined against F2 by a full-matrix least-squares method using SHELXL.[41] All of the nonhydrogen atoms were refined anisotropically. The hydrogen atoms were placed in calculated positions and constrained to ride on their parent atoms. Crystallographic data for complexes 1 and 2 are summarized in Table . Selected bond lengths and angles for 1 and 2 are given in Table S1.

Table 1

Details of the Data Collection and Refinement Parameters for Complexes 1 and 2

complex	1	2
empirical formula	C66H60Cl4N10Ni2O20Tb2	C63H55.5Cl4N8.5Ni2O20Er2
formula weight/g mol–1	1890.30	1845.40
temperature/K	298(2)	298(2)
wavelength/Å	0.71073	0.71073
crystal system	monoclinic	monoclinic
space group	P21/n	P21/c
a/Å	13.7343(11)	12.1038(8)
b/Å	13.7633(11)	19.1062(19)
c/Å	18.8592(12)	16.2693(15)
α/°	90	90
β/°	96.8840(10)	109.133(2)
γ/°	90	90
volume/Å3	3539.2(5)	3554.6(5)
Z	2	2
Dc/g·cm–3	1.774	1.724
μ/mm–1	2.729	3.085
F(000)	1880	1826
θ limit/°	1.74–25.50	2.13–25.50
unique reflections	6593	6626
observed reflections [I ≥ 2σ(I)]	4589	4833
parameters	474	469
restraints	0	18
Rint	0.0590	0.0564
goodness of fit on F2	1.027	1.067
R1, wR2 [I ≥ 2σ(I)]	0.0393, 0.0717	0.0436, 0.1078
R1, wR2 (all data)	0.0729, 0.0835	0.0700, 0.1239

Magnetic Measurements

Magnetic susceptibility and magnetization data were recorded for powdered samples of complexes 1 and 2 on a Quantum Design MPMS-5XL SQUID magnetometer. Magnetic susceptibility data were taken in the temperature range from 1.9 to 300 K in a magnetic field of 1 kOe. Magnetization measurements were performed at 1.9 K in magnetic fields of 0 to 50 kOe. The magnetic data were corrected for the empty sample holder and diamagnetic contributions from the sample (−0.45 × 10–6 cm3/g·molar weight). Transmission powder X-ray analysis was utilized to ensure that the single-crystal data were representative of the bulk material.