A Dinuclear Dysprosium(III) Single Molecule Magnet of Benzo[h]quinolin-10-ol

Abstract

To develop single molecule magnets, a dinuclear complex [Dy₂(HOBQ)₄Cl₆] (1) was prepared from the reaction of DyCl₃ with benzo[h]quinolin-10-ol (HOBQ). Each Dy(III) ion shows a compressed octahedral geometry and the two Dy(III) ions in 1 are bridged by two Cl⁻ ligands to construct a dinuclear structure with the four HOBQ ligands on the axial positions and six Cl⁻ ligands in the equatorial plane. Magnetic measurements showed that complex 1 is a field-induced single molecule magnet having an obvious magnetic hysteresis loop with an energy barrier of 71(2) K. These experimental results are corroborated by the ab initio complete active space self-consistent field (CASSCF) calculations which also interpret the magneto-structural correlation. It is a typical example to achieve Dy(III) SMM through regulating coordination geometry, i.e., lengthening equatorial coordination bonds and shortening axial ones to form a compressed octahedral geometry.

1. Introduction

In recent years, single-molecule magnet (SMM) has attracted a significant interest in many fields such as chemistry, physics, and material science [1,2,3,4,5,6,7]. It has undergone a considerable development, acquiring some surprising achievements such as large barrier energy, high blocking temperature and long relaxation time especially in mononuclear Dy(III) SMMs [8,9,10,11]. However, these achievements are still far from the requirement for their practical applications.

These studies revealed some rules on the synthetic strategy and SMM performance tuning. There are many factors influencing SMM performances, which include ligand design, symmetry control, counter ions and even coordinated or lattice solvent molecules [12,13,14,15,16]. In a word, any factors which can influence the anisotropy of the whole molecule will ultimately tune the SMM performance. For an example, the group of Sanping Chen found that lattice water molecules can tune SMM performance in Dy₂-SMMs through adjusting magnetic interactions [17]. In the latest several years, high SMM performances were also achieved in low-coordinate Dy(III) single-molecule magnets [18]. The group Eduard B. Yagubskii discussed the competition between the apical and equatorial crystal fields in tuning SMM performances in seven-coordinate pentagonal bipyramidal Dy(III) complexes [19]. They pointed out that their SMM performances are presumably improved by enhancing the apical crystal field through the use of more negatively charged ligands to obtain more short axial bonds, as well as by weakening equatorial crystal field through longer equatorial bonds and less equatorial negative charge.

Compared with transition metal ions, lanthanide ions, especially Dy(III) ions, have large magnetic moments and magnetic anisotropy due to their large unquenched orbital angular momentum. Thus, lanthanide ions are usually the optimum selection for developing SMMs with high performance. In addition to the anisotropy of individual lanthanide ion, intermetallic interactions can also influence the SMM performance in some extent. However, the 4f-electrons of lanthanide ions are shielded by their outer 5s and 5p electron shells, which leads to a weak inter-lanthanide magnetic interaction. So the high SMM performance of many lanthanide clusters and polymers are mainly from individual lanthanide ions [20,21,22,23]. In this sense, the design of ligands and the symmetry control of lanthanide ions in some special coordination geometries are of significant importance for optimizing SMM performances. In addition, a rational arrangement of lanthanide ions is also crucial for maximizing the SMM performances of complexes bearing two or more lanthanide ions. A significant success has been achieved in dilanthanide complexes, such as {Dy[15-MC_Cu-5]}₂ [24], [Er^III₂(COT″)₃] [25], phosphino-supported Er₂ SMMs [26] and (Cp^iPr5)₂Dy₂I₃ [27], highlighting the importance of intramolecular anisotropy axis alignment for enhancing SMM performance.

Based on the above-mentioned consideration, we aimed to use chloride ions to coordinate to Dy(III) ion in the equatorial plane and to use HOBQ to coordinate to Dy(III) at axial sites, attempting to achieve short axial coordination bonds and long equatorial coordination bonds. As expected, [Dy₂(HOBQ)₄Cl₆] (1) was finally prepared from the reaction of DyCl₃ with HOBQ. Its magnetic properties were investigated in detail, which demonstrated a field-induced single molecule magnet for complex 1.

2. Materials and Methods

2.1. Materials

All materials for this work were used as commercially received (benzo[h]quinolin-10-ol, Bide Pharmatech Ltd., Shanghai, China; DyCl₃·6H₂O, Macklin Biochemical Co., Ltd., Shanghai, China; ethanol and acetonitrile, Asia Chemical Co., Ltd., Chengdu, China). Their IR, powder and single crystal X-ray diffraction (PXRD and SCXRD) analyses were carried out on Spectrum Two FTIR equipment (PerKin-EImer, Waltham, MA, USA), Rigaku D/max 2500v/pc diffractometer (Rigaku, Tokyo, Japan) and XtaLAB Synergy-DS single crystal X-ray diffractometer (Rigaku, Tokyo, Japan), respectively. Elemental analyses (C, H and N) were made on an Elementar Micro cube CHN elemental analyzer (Elementar, Langenselbold, Germany). Magnetic measurements were performed on a MPMS 3 SQUID magnetometer (Quantum Design, San Diego, CA, USA) furnished with a 7 T magnet. Diamagnetic corrections were applied for dealing with their magnetic susceptibilities.

2.2. Synthesis of [Dy₂(HOBQ)₄Cl₆] (1)

After one end of a 20 cm long glass tube was sealed, DyCl₃·6H₂O (0.1 mmol, 0.0368 g), benzo[h]quinolin-10-ol (0.1 mmol, 0.0195 g), ethanol (0.5 mL) and acetonitrile (1.5 mL) were added into this tube. It was then evacuated and completely sealed before being heated in an oven (80 °C) for 10 h. Finally the sealed tube was cooled overnight, yielding pale yellow crystals with a yield of 42% (calculated from HOBQ). Elemental analysis calculated for C₅₂H₃₆Cl₆Dy₂N₄O₄: C, 47.37; H, 2.75; N, 4.25%. Found: C, 47.51; H, 2.83; N, 4.32. IR (Figure S1, KBr pellet, cm⁻¹): 3377(s), 3067(w), 2885(w), 1630(m), 1576(m), 1543(m), 1462(m), 1435(s), 1355(m), 1327(m), 1263(S), 1227(m), 1189(w), 1145(m), 1085(w), 1060(w), 1028(m), 993(w), 983(w), 920(w), 881(w), 861(w), 836(S), 808(m), 784(w), 761(w), 714(m), 681(w), 635(w), 585(m), 556(m), 522(w), 509(w).

2.3. X-Ray Data Collection and Structure Refinement

The structure of 1 was determined from single crystal X-ray diffraction analysis on a XtaLAB Synergy diffractometer with graphite-monochromate Cu K_α radiation (λ = 1.54184 Å). After being solved with SHELXS, its structure was then refined using the SHELXL program. Hydrogen atoms binding with C were geometrically added using a riding model and hydrogen atoms on N atoms were located from difference Fourier map. The detailed crystallographic data are listed in Table S1 with the selected bond distances and angles given in Table S2.

3. Results

3.1. Synthesis and Structural Characterization of 1

The crystalline sample of 1 was obtained from the solvothermal reaction of DyCl₃·6H₂O with HOBQ in a mixed solvent of ethanol and acetonitrile in a glass tube. Its purity was confirmed through the high agreement of experimental powder X-ray diffraction patterns with its simulated one derived from its single crystal structural data (Figure S2).

Complex 1 crystallizes in the Pī space group of triclinic system and shows a dinuclear structure (Figure 1a) with six Cl⁻ ligands in the equatorial plane and four HOBQ ligands on the axial positions. Two of the six Cl⁻ ligands bridge the two Dy(III) ions with the other four Cl⁻ ligands terminally coordinating to the two Dy(III) ions. There exist intramolecular H-bonds of N–H∙∙∙O (Table S3) in the HOBQ ligands which terminally coordinate to Dy(III) ions (Figure 1b). Between the adjacent two HOBQ ligands, π∙∙∙π stacking interactions were found as indicated by the dihedral angle of 3.806(2)° between the two benzo[h]quinoline planes and a shortest centroid-to-centroid distance of 3.5866(2) Å. Thus, each Dy(III) ion is six coordinated by four equatorial Cl⁻ ligands and two axial HOBQ ligands. The Dy–O and Dy–Cl bond lengths are in the range of 2.186(4)–2.209(4) and 2.5844(16)–2.7880(14) Å, respectively. This means that the equatorial coordination bonds are much longer than the axial ones. The axial O–Dy–O bond angle is 173.74(15)°, near to 180°. The Cl–Dy–Cl bond angles in the equatorial plane are in the range of 80.20(4)–101.41(6) and 167.64(5)–170.98(6)°. The O–Dy–Cl bond angles between axial and equatorial coordination atoms are in the range of 82.54(12)–93.56(12)°. All of these bond lengths and bond angles support that the Dy(III) ions have distorted compressed octahedral geometries (Figure 1c and Table S4).

The neighboring coordination molecules of 1 interact with each other through hydrogen bonds (C10–H10∙∙∙Cl1B, Table S3), leading to the construction of a hydrogen-bonded chain along axis a (Figure S3). Each coordination molecule further interacts with another four ones in the (1ī0) plane through π∙∙∙π stacking interactions as demonstrated by their parallel benzo[h]quinoline planes with the nearest centroid-to-centroid distance of 3.4924(2) and 3.6525(2) Å, building a π∙∙∙π stacking 2D framework in the (1ī0) plane (Figure S4). These hydrogen bonds and π∙∙∙π stacking interactions finally connect the coordination molecules of 1 into a 3D supramolecular framework (Figure 2).

3.2. Magnetic Properties

The magnetic susceptibilities of complex 1 from 2 to 300 K were measured under 1000 Oe, which are plotted into Figure 3a in the form of χ_MT versus T. The 300 K χ_MT value of 1 is 27.12 cm³ K mol⁻¹, a little lower than the theoretical spin-only values (28.34 cm³ K mol⁻¹) for two independent Dy(III) ions (g = 4/3, J = 15/2, L = 5, S = 5/2, ⁶H_15/2) [28,29]. The χ_MT values of complex 1 decrease with lowering temperature to 15.04 cm³ K mol⁻¹ at 2 K. The χ_MT-T profiles might result from the thermal depopulation of the Stark sublevels of Dy(III) ions and/or the crystal field splitting, and also possible weak antiferromagnetic interactions [30,31,32,33,34,35]. The linear fitting of χ_M⁻¹ versus T plot (Figure S5) for 1 in the temperature range of 2.0–300 K based on Curie-Weiss law gave C = 27.38 cm³ K mol⁻¹ and θ = −3.84 K. The negative Weiss constant supports the presence of dominant weak antiferromagnetic interaction.

Field-related magnetizations of 1 at 2, 3 and 5 K were measured and drawn as M–H and M–HT⁻¹ plots (Figure 3b,c). Its magnetization increases slowly in the high field region after a rapid increase in the low field region, reaching a maximum of 10.99 Nμ_β at an applied field up to 70 kOe measured at 2 K, which is consistent with those reported in the literature [36,37,38], but far from saturation values. The difference in M values at different temperatures become larger with the increasing HT⁻¹ values as revealed in Figure 3c, leading to the non-superposition of M–HT⁻¹ plots at different temperatures. All of the above-mentioned features suggest a large magnetic anisotropy of Dy(III) ions in 1. Interestingly, an obvious small hysteresis loop (Figure 3d) was observed for 1 at 2 K [39].

In order to study the slow magnetic relaxation behavior of complex 1 in detail, its dynamic magnetic properties were investigated. The optimal field for measuring its in-phase (χ′) and out-of-phase (χ″) alternating current (ac) magnetic susceptibilities (Figure 4) was first explored at a temperature of 2 K under a series of external fields (0–1800 Oe). At zero dc field, the χ″ values of 1 are very small with no maximum in its χ″–ν curve. When a dc field was applied, the χ″ values of 1 become larger and the χ″–ν curves show maximums which become more significant before 1200 Oe and then less obvious and move to a little higher frequency upon increasing the applied dc fields. After an overall consideration of the above-mentioned information, 1200 Oe was chosen as the optimal dc field for the further measurement of ac magnetic susceptibility.

Thus the frequency-dependent χ′ and χ″ of 1 were collected under 1200 Oe with its data and the derived data plotted into Figure 5 [40,41]. As shown in Figure 5b, 1 present clearly non-zero χ″ values under the optimal dc field and the χ″-ν curves show maximums which became less important upon increasing the measuring temperature and disappeared after 3.5 K, with the maximum at 2 K appearing at 284.71 Hz. These experimental facts told us that 1 exhibits slow magnetic relaxation behavior and is a field-induced SMM [42,43,44,45]. Then, the Cole-Cole diagram (Figure 5c) of 1 was drawn based on these frequency-dependent χ′ and χ″ data collected under the optimal dc field. The best fitting of these χ″ versus χ′ curves on Debye model gave the distribution coefficients α = 0.12 − 0.17 and relaxation time τ = 3.7 × 10⁻⁵ − 4.4 × 10⁻⁴ s. The derived relaxation times at different temperatures were plotted into a curve of ln(τ/s) versus T⁻¹ (Figure 5d) for analyzing its spin-lattice relaxation. Considering that the ac magnetic susceptibilities were measured under the optimal field, only direct relaxation process (AH^mT), Raman process (CTⁿ) and Orbach process (τ₀⁻¹exp(−U_eff/k_BT)) were taken into the fitting of ln(τ/s) versus T⁻¹ curve with the quantum tunneling process (QTM) omitted [46,47,48]. The best simulation based the equation of τ⁻¹ = τ₀⁻¹exp(U_eff/k_BT) + CTⁿ + AH^mT gave τ₀ = 1.3(7) × 10⁻¹³ s, U_eff/k_B = 71(2) K, C = 112(3) s⁻¹ K⁻ⁿ, n = 3.0(1), AH^m = 1356.1(4). The low Raman exponent (n = 3.0(1)) might be resulted from an acoustic-optic two-phonon Raman process as suggested in documents [49,50], which is presumably originated from the dinuclear skeleton of 1 bearing six equatorial Cl⁻ ligands and four axial ligands. A nearly linear slope can be found in the ln(τ/s) versus T⁻¹ curve above 3.2 K, suggesting a dominant Orbach relaxation mechanism in this temperature region. Raman process accounts for the curvature of the ln(τ/s) versus T⁻¹ curve at intermediate temperature region.

All of these investigations confirmed that we succeeded in obtaining a field-induced single molecule magnet. It might be associated with the structure which has six chloride ions coordinating to Dy(III) ions in the equatorial plane and four HOBQ ligands coordinating to Dy(III) ions in the axial positions, with the equatorial coordination bonds (Dy–Cl) much longer than the axial coordination bonds (Dy–O). This kind of structure is facile for improving the anisotropy of Dy(III) ion, achieving to obtain SMMs. However, an optimal dc field is necessary for suppressing the QTM and the energy barrier of 1 is not high as expected. It is presumably caused by two deficiencies in the structure. One is that the HOBQ ligands coordinate to Dy(III) ions in the neutral form with the H atoms still between O and N atoms, which leads to the absence of necessary negative charge on the axial coordinated O atoms for enhancing the axial anisotropy of Dy(III) ion. The second one is that the equatorial negative charge is superfluous although the equatorial bond lengths are long enough, which is not facile for improving the axial anisotropy of Dy(III) ion. This also gave us some hints in the performance improvement of this kind of SMMs. We can remove the protons on the axial ligands to increase axial negative charges and replace the four terminal Cl⁻ ligands using neutral chelating ligands to reduce the equatorial negative charge, achieving high axial anisotropy of Dy(III) through their cooperation. The further work is presumably to construct another perfect story.

3.3. Theoretical Calculation

To further understand the magnetic properties and magnetic exchange interactions of compound 1, we performed electronic structure calculations based on single-crystal diffraction data without optimizing the compound’s structure. Here, we used version 5.0.3 of the ORCA software [51,52] to perform ab initio complete active space self-consistent field (CASSCF) calculations [53]. To further calculate the magnetic interactions between the two dysprosium ions, our calculations were conducted in two steps: First, replacing one of the dysprosium ions with an antiferromagnetic Lu³⁺ ion, we constructed two fragments, 1a and 1b (Figure S6). Using the CASSCF method and the SINGLE_ANISO module [54,55], we calculated the anisotropic g-factor, the eight lowest-energy Kramers doublet energy levels, and the magnetic axis for 1a and 1b. Second, we calculated the magnetic interaction between the two dysprosium ions based on the results from the first step using the Lines model and the POLY_ANISO module [56,57,58,59]. The equations used for calculation are shown below:

$$\widehat{H}=-J_{total}{\widehat{\tilde{S}}}_{1z}{\widehat{\tilde{S}}}_{2z}=-(J_{exch}{\widehat{\tilde{S}}}_{1z}{\widehat{\tilde{S}}}_{2z}+J_{dip}{\widehat{\tilde{S}}}_{1z}{\widehat{\tilde{S}}}_{2z})$$

(1)

$$J_{dip}={\displaystyle \frac{{\mu }_{Bohr}^2}{|r{|}^3}}[{\overrightarrow{g}}_{1z}\cdot {\overrightarrow{g}}_{2z}-3({\overrightarrow{g}}_{1z}\cdot \overrightarrow{r})(\overrightarrow{r}\cdot {\overrightarrow{g}}_{2z})]$$

(2)

$$J_{total}=J_{exch}+J_{dip}$$

(3)

In the calculations, the SARC2-DKH-QZVP basis set for Dy and Lu atoms [60], DKH-def2-TZVP for O atoms, and DKH-def2-SVP [61] for other atoms. The “Autoaux” feature was used to automatically generate auxiliary basis sets for RIJCOSX approximations. In CASSCF calculations, the active space included 9 electrons in 7 orbitals for the Dy³⁺ ion (CASSCF(9,7)). In RASSI calculations, all 21 sextets were considered, and selecting all sextets to describe Dy-SMMs is reliable [62,63,64]. The relativistic effects were considered using the second-order Douglas-Kroll-Hess (DKH2) method [65]. Additionally, the effective energy barrier was calculated using the method proposed by Yin et al. [20,66], and the magnetic relaxation mechanism of 1a and 1b was analyzed, which involves Equations (4)–(8).

$$U_{\mathrm{eff}}(T)={\displaystyle \sum _{i=1}^n\frac{K_i(T)}{{\displaystyle \sum _i{K}_i(T)}}}{E}_i$$

(4)

$$K_i(T)={\displaystyle \frac{\exp (-E_i/k_{B}T)}{{\displaystyle \sum _i\exp (-E_i/k_{B}T)}}}{k}_{\mathrm{QT},i}$$

(5)

$$k_{\mathrm{QT},i}={\displaystyle \frac{g_{\mathrm{XY},i}^2}{2\sqrt{(g_{\mathrm{XY},i}^2+g_{\mathrm{Z},i}^2)}}}, g_{XY}=\sqrt{(g_{\mathrm{X}}^2+g_{\mathrm{Y}}^2)}$$

(6)

$$N={\displaystyle \sum _i{K}_i(T)}$$

(7)

$${\tau }_{\mathrm{QTM},i}^{-1}={\displaystyle \frac{\beta B_{ave}}{h}}{k}_{\mathrm{QT},i}$$

(8)

In 1, the two Dy³⁺ ions have similar coordination environments, so the calculated saturated effective barriers for fragments 1a and 1b are close, with values of 1a (U_eff(300 K) = 601.5 cm⁻¹) and 1b (U_eff(300 K) = 601.4 cm⁻¹) (Figure S7). From the energy gaps of the two fragments, 1a (856.6 cm⁻¹) and 1b (856.5 cm⁻¹) are also similar (Tables S5 and S6), further indicating that the two Dy³⁺ ions have similar coordination environments. From the perspective of magnetic relaxation mechanisms and wave functions, the lowest-energy Kramers doublets (KDs) of 1a and 1b are both pure | ± 15/2>, with wave function compositions of 99.8%. In the ground state, the probabilities of magnetic quantum tunneling between the ground state KDs of the two fragments are 2.63 × 10⁻⁴ μ_B for 1a and 2.64 × 10⁻⁴ μ_B for 1b (Figure S8), indicating that the QTM process is almost quenched in the ground state. The magnetic relaxation process occurs via the thermally activated third excited state (KD₂). At the same time, the calculated g_xy values are 1.151 × 10⁻³ and 1.156 × 10⁻³ for 1a and 1b, respectively, both of which are less than 0.015 [67]. This also indicates that 1 can be considered a candidate for zero-field single molecule magnets.

To further investigate the magnetic interaction between two Dy³⁺ ions, the magnetic coupling constant was calculated using the POLY_ANISO module [54,55] based on the Lines model. The J_total of −0.752 cm⁻¹ (J_exch = −0.187 cm⁻¹ and J_dip = −0.565 cm⁻¹) indicates that there is a weak antiferromagnetic coupling between the two Dy³⁺ ions, which is consistent with experimental measurements. From the perspective of the main magnetic axis direction of the ground state of compound 1, its main magnetic axis in the ground state is along the O–Dy–O axis and does not lie in the Dy₂Cl₆ plane (Figure S9).

4. Conclusions

In summary, we have successfully synthesized a field-induced dinuclear Dy(III) SMM. Each Dy(III) ion has a compressed octahedral geometry with four Cl⁻ ligands on the four equatorial positions and two oxygen atoms from two benzo[h]quinolin-10-ol ligands on the two axial positions. The two Dy(III) ions are bridged by two Cl⁻ ligands in the equatorial plane to construct the dinuclear structure with the four HOBQ ligands on the axial positions and six Cl⁻ ligands in the equatorial plane, which leads to much shorter axial coordination bonds than the equatorial ones. Its static and dynamic magnetic properties were investigated in detail. It revealed a field-induced single molecule magnet for 1 with an obvious magnetic hysteresis loop and an energy barrier of 71(2) K. The ab initio complete active space self-consistent field (CASSCF) calculation results support the experimental results. It is a typical example for the construction of Dy(III) SMM through regulating coordination geometry which was achieved by controlling axial coordination bonds shorter than the equatorial ones. Although the performance of this SMM is not as high as expected, it can give us some guiding for the construction of SMMs, as well as some hints for further modification to achieve high-performance SMMs.