A Dy^III Complex of a Pentadentate Schiff Base with Field-Induced Single-Ion Magnet Behaviour

Abstract

The influence of the solvent in the reaction of dysprosium(III) chloride hydrate with the N₃O₂ ligand H₂L (2,6-bis(2-hydroxyphenyliminomethyl)pyridine) was studied To this end, the new mononuclear chloride complex [Dy(L)Cl(H₂O)₂] (1) was isolated in absolute ethanol as solvent, without any evidence of the hydrolysis of the ligand. This clearly contrasts with previous results, where a similar reaction in methanol proceeds with the partial hydrolysis of the Schiff base, and the formation of a new hemiacetal donor to yield [Dy(HL’)₂)][Dy(L)(Cl₂)] (H₂L’ = (6-(2-hydroxyphenyliminomethyl)-2-methoxyhydroxymethyl)pyridine). The single crystal X-ray structure of the chloride complex 1 shows that the Dy^III ion is octacoordinated in a highly distorted N₃O₄Cl environment between triangular dodecahedral and biaugmented trigonal prisms. The full magnetic characterisation of 1 shows that it presents field-induced single ion magnet behaviour, with a thermal energy barrier U_eff of 113.5 K, which is the highest among dysprosium complexes derived from H₂L.

1. Introduction

Single-molecule magnets (SMMs), and within them, single-ion magnets (SIMs), which are quantum in nature, are at the vanguard of new technological developments in spintronics and quantum information processing [1,2]. This is because the properties of these magnets are inherent to the molecule, enabling an extraordinary degree of device miniaturization. Moreover, the molecular nature of these systems offers the possibility of introducing different functionalities into the devices, which is a great additional advantage. However, there are still many challenges to overcome before molecule magnets can be used at a technological level. The first commitment is to obtain SMMs that remain stable in air with higher blocking temperatures than has been achieved so far. The up-to-date record blocking temperature (T_B) is 80 K, and it is held by the SIM [(Cp^iPr5)Dy(Cp*)][B(C₆F₅) [3], a similar T_B to that recently achieved by a dinuclear mixed-valent dysprosium metallocene [4]. Therefore, the T_B of these compounds exceeds the temperature of liquid nitrogen, which allows a glimpse of the use of these magnets in nanotechnology. However, a major drawback of these metallocenes is that they are unstable in air. Consequently, the field of molecular magnets is still wide open, and obtaining air-stable coordination compounds with geometries that can improve the anisotropy of the systems, and, accordingly, their functioning, is an area of growing interest. In this way, some air stable complexes of oblate ions with highly axial D_5h symmetry, mainly derived from phosphine oxides and other auxiliary monodentate ligands [5,6,7], have been reported to show high anisotropy, and, consequently, good magnetic performance, with blocking temperatures of up to 20 K [5]. However, isolating compounds with a predetermined geometry from monodentate ligands is not an easy task, and obtaining a specific geometry often rather seems to be the result of serendipity. Therefore, a more rational strategy to predetermine highly axial geometries around oblate ions should be the use of rigid ligands that occupy the equatorial plane, and to fill the apical positions with monodentate ancillary donors. This is an emerging approach and, in this way, dysprosium compounds with fairly rigid macrocyclic ligands are beginning to show promising results [8]. However, the field of rigid acyclic ligands seems poorly explored [9,10,11,12], where results so far have been mixed. In this specific field, we began to study the coordination capability of a previously well-known air-stable pentadentate Schiff base, H₂L (Scheme 1) [13], with Dy^III ions [14], and accordingly, we have published the first lanthanoid coordination compounds with this type of ligand, many of which show slow relaxation of the magnetization. However, in this reported study [14], we were unable to obtain a dysprosium complex with the non-hydrolysed Schiff base and chloride as auxiliary donor. As a continuation of this work, the effect of the synthesis solvent on the complexes to be obtained was studied, and thus the new compound [Dy(L)Cl(H₂O)₂] (1), whose behaviour as a magnet is also analysed, was able to be isolated and it is presented herein.

2. Results and Discussion

2.1. Synthesis

The reaction of DyCl₃·6H₂O at room temperature with an absolute ethanol solution of H₂L, basified with triethylamine, leads to the isolation of the molecular mononuclear complex [Dy(L)Cl(H₂O)₂] (1), whereas a similar related reaction in methanol at reflux yields the dinuclear ionic compound [Dy(HL’)₂)][Dy(L)(Cl₂)], as shown in Scheme 1 [14].

In view of these results, it might seem that there are two factors that can affect the obtaining of the different compounds by favouring the hydrolysis of the H₂L ligand: the solvent of the reaction and the temperature. Thus, in order to discern which of the two factors promotes Schiff base cleavage, the same reaction leading to 1 was repeated at reflux temperature, and it yields the same complex. In addition, further evidence that temperature does not seem to influence the cleavage of the Schiff base is that similar complexes with nitrate [14] instead of chloride as an ancillary donor have also been obtained in 96% ethanol at reflux, without any sign of the hydrolysis of the ligand. Therefore, it appears that the temperature plays no role in the hydrolysis process, but that the solvent is crucial, this process being highly favoured for methanol. In principle, this could be attributed to the higher amount of water present in methanol compared to the absolute ethanol (98%) used in the synthesis of 1. However, this does not seem to be a determining factor, considering that dysprosium complexes containing H₂L and nitrate as ligands were able to be obtained from 96% ethanol solutions [14], and that the starting salts for the syntheses are always hydrated. Therefore, perhaps the slightly more acidic character of methanol over ethanol could contribute to the easier hydrolysis in methanolic solution.

Complex 1 is a deep red solid, which is air stable and does not lose water ligands, at least not at temperatures slightly above 100 °C (ca. 125 °C). The complex was completely characterized by means of elemental analysis, infrared spectroscopy, and single and powder X-ray diffraction studies. In addition, the full magnetic characterization of the sample and its behaviour as a magnet were also analysed.

The IR spectrum of compound 1 (Figure S1) shows two sharp bands located at 1533 and 1583 cm⁻¹, which could be assignable to the ν(C=N) vibrations of the pyridine and imine bonds [14], respectively. These bands experience negative displacements with relation to the ones of the free H₂L, and this shift agrees with the coordination of both the pyridine and imine nitrogen atoms to the dysprosium metal ion. In addition, a broad band centred at 3315 cm⁻¹ is in agreement with the existence of water coordinated to the metal atom.

2.2. X-ray Diffraction Studies

Figure 1 shows an ellipsoid diagram for complex 1, and the main distances and angles regarding the dysprosium ion for this compound are summarised in Table S1.

The asymmetric unit of 1 just contains neutral [Dy(L)Cl(H₂O)₂] molecules, without any solvate. In the [Dy(L)Cl(H₂O)₂] complex, both phenol oxygen atoms of the Schiff base are deprotonated. This dianionic L²⁻ base acts as a pentadentate N₃O₂ ligand, using all its potential donor atoms. Despite the conjugation of the deprotonated ligand, which should make it practically flat, it adopts a helical arrangement in this complex, as was observed in certain previously described complexes of this donor [14].

Consequently, O1 and O2 are the most protruding atoms from the mean N₃O₂ computed plane, which are ca. 0.35 Å above or below this plane. The dysprosium ion is also out ofthe calculated plane, from where it protrudes ca. 0.39 Å.

The coordination sphere of the Dy^III ion is fulfilled by three further monodentate ligands: two water molecules and a chloride anion. Hence, the Dy^III ion has coordination number 8, with an N₃O₄Cl environment. The calculations of the grade of distortion of the DyN₃O₄Cl core relative to an ideal polyhedron of eight vertexes with the SHAPE program [15] reveals a largely distorted geometry between a biaugmented trigonal prism and a triangular dodecahedron (Table S2). The main angles and distances of the dysprosium ion (Table S1) are comparable to those previously reported for this kind of compound [10,14,16], and do not deserve further consideration.

Finally, the mononuclear units [Dy(L)Cl(H₂O)₂] establish hydrogen bonds among them, involving both water and chloride ligands. These interactions expand the initial mononuclear block in a zigzag chain, as shown in Figure 2, and this leads to the shortest intermolecular distance between Dy^III ions being 5.4407(7) Å.

Moreover, powder X-ray diffraction measures for the raw sample of 1 (Figure S2) were performed. These studies show that the bulk sample of the isolated product and the solved single crystals are the same complex, and that it has been obtained with high purity, given that no other peaks can be observed in the experimental diffractogram.

2.3. Magnetic Properties

The temperature change in magnetic susceptibility data for [Dy(L)Cl(H₂O)₂] (1) was registered. Figure 3 represents the χ_MT product vs. T.

The value of this χ_MT product at 300 K is 14.42 cm³Kmol⁻¹, which is very similar to the calculated one for an isolated Dy³⁺ ion (14.13 cm³Kmol⁻¹). This χ_MT value remains virtually constant when the temperature drops to 35 K, and then it diminishes to 2 K, touching a minimum value of 9.71 cm³kmol⁻¹. This fall in the curve can be attributed primarily to the depopulation of the M_J sublevels of the Dy^III ion at low temperatures, although it cannot be ruled out that antiferromagnetic intermolecular interactions also contribute to this drop.

The value of reduced magnetisation at 2 K points to 5.58 Nµ_B at the highest applied field of 50,000 Oe (see Figure 3 inset). This value is much less than that calculated for a Dy^III ion with g = 4/3 and J = 15/2 (10 Nµ_B), but it is quite close to the value of a largely anisotropic Dy^III ion with an Ising-like ground doublet (5 Nµ_B) [17]. Accordingly, both the susceptibility and reduced magnetisation measurements seem to agree with a quite anisotropic system.

In the light of these results, the low-temperature magnetic relaxation behaviour of this complex was also studied. Accordingly, alternating current (ac) magnetic susceptibility data were registered at a frequency of 8000 Hz in a zero dc external field. Under these conditions, no in-phase (χ’_M) or out-of-phase ac susceptibility (χ″_M) peaks were observed (Figure S3). This could be because the relaxation of the magnetisation is extremely fast, or because there is a substantial quantum tunnelling of the magnetisation (QTM) [18,19]. A common way to see whether quantum tunnelling exists is to apply an external dc magnetic field [20], as this would destroy the bistability of the ground state, and thus the possibility of the quantum tunnel effect. Therefore, different static magnetic fields were applied in order to see whether they were able to remove the QTM with the aim of establishing the optimum field. Consequently, new ac susceptibilities as a function of the temperature were recorded under the presence of various dc fields, which allowed the establishment of an optimal field of 1500 Oe (Figure S4). Hence, new ac susceptibility data were again recorded under this dc field (χ′_M in Figure S5 and χ″_M in Figure 4). In these conditions, χ″_M (Figure 4) peaks between 4.5 and 8 K at different fixed frequencies (ranging between 10 and 10,000 Hz) can be observed.

Moreover, the Cole–Cole graph for 1 (Figure S6) shows semicircles with α parameters extracted using a generalised Debye model (Table S3) that range from 0.22 to 0.12. These α values seem to indicate that many relaxation processes are operative at low temperatures.

In addition, studies on the relaxation of the magnetization were additionally performed by examining the dependence of relaxation time (τ) on the temperature. Thus, the representation of the variation of τ with the temperature (Arrhenius plot, Figure 5) shows a deviation from linearity at the low-temperature region, and this, in combination with the α parameters obtained from the Cole–Cole graph, indicates that the line cannot be replicated with an Orbach relaxation model alone.

Therefore, many efforts were made to reproduce this graph, taking into consideration all the possible relaxation processes (Orbach, Direct, Raman, and QTM). Thus, the fit that best replicated this plot with rational parameters considering all these processes, individually or grouped, was obtained using Orbach and Raman relaxation (Equation (1)), and the different attempts at fitting the curve including the QTM relaxation process were all unsuccessful. This mentioned fit renders an energy barrier for the reversal of the magnetization of 113.5 K, with a τ₀ factor of 2.6 × 10⁻¹¹ s, and Raman values of C = 0.1 K⁻ⁿs⁻¹ and n = 5.8. This n value could seem much lower than anticipated for a Dy^III ion (n = 9), but the implication of acoustic and optical phonons in magnetic dynamics can justify the divergence from the foreseen value for a Dy^III Kramers ion [21,22]. In fact, this value lies between the range of 5 and 9 usually reported for several other dysprosium complexes [23].

$${\tau }^{-1}=C{T}^{-n}+{\tau }_0{}^{-1}{\mathrm{e}}^{-U_{eff}/k_B\mathrm{T}}$$

(1)

In an attempt to deepen the understanding of the magnetic behaviour, the anisotropy axis of the dysprosium ion was calculated using the MAGELLAN program [24] (Figure 6, left). Thus, it can be seen that the anisotropy axis is practically aligned with the Dy-O1 bond. Accordingly, the anisotropy axis of 1 coincides with the geometric axial axis of a biaugmented trigonal prism (Figure 6, right). Therefore, this marks an axiality that justifies that the compoundbehaves like a molecule magnet. However, the hydrogen bond interactions present in this complex (see crystallographic discussion) could enhance the quantum tunnelling of the magnetization and, as a result, avoid the observation of molecule magnet behaviour in the absence of an external magnetic field, as has already been demonstrated for other dysprosium compounds with the same ligand [14]. In spite of this, the U_eff barrier for 1 is the highest among the reported dysprosium complexes derived from H₂L (Table S4). It must be noted that all of the previously published non-ionic Dy^III compounds with this Schiff base ligand show the coordination number 9, while in complex 1, the dysprosium ion is in an octacoordinated environment. Consequently, it seems that the use of monodentate ligands instead of potentially bidentate ones (chloride instead of nitrate) decreases the coordination number, which seems to be accompanied by an increase in the energy barrier. However, given the different coordination numbers, which lead to very different geometries, it is not possible to establish further magneto-structural correlations.

In addition, the energy barrier for the spin reversal in 1 was compared with the one of other dysprosium complexes with other rigid N₃O₂ ligands. Not many complexes of this kind were found, and, to our knowledge, the only other planar rigid ligands that behave as N₃O₂ donors are hydrazones, all of which are derived from 2,6-diacetylpyridine and various hydrazides. The comparison of U_eff in 1 and in these Dy^III complexes derived from hydrazones under different magnetic fields shows that 1 still remains the compound with the highest energy barrier for the reversal of magnetization among those with a predetermined quasi-planar equatorial N₃O₂ environment and coordination numbers 8 [8,25] or 9 [10,26].

3. Materials and Methods

3.1. Materials and General Methods

All the chemical reagents used in the present work were purchased from trading houses, and they were used as supplied by the provider, without additional purification. The elemental analyses of C, H, and N for the Schiff base H₂L, and for the dysprosium complex 1 were performed on a Themoscientific Flash Smart analyser. The infrared spectrum of 1 was registered between 4000 and 500 cm⁻¹ on a Varian 670 FT/IR spectrophotometer operating in the ATR mode. A Varian INNOVA-400 spectrometer was used to record the ¹H NMR spectrum of H₂L in the DMSO-d₆ as solvent.

3.2. Syntheses

The well-known H₂L ligand was obtained via the modification of a method previously described [13], as published by us [14], and adequately characterized in solid state by elemental analysis, and in solution by ¹H NMR spectroscopy.

[Dy(L)(Cl)(H₂O)₂] (1): a solution of DyCl₃∙6H₂O (0.130 g, 0.347 mmol) in absolute ethanol (10 mL) was added to a solution of a mixture of H₂L (0.110 g, 0.347 mmol) and Et₃N (0.095 g, 0.941 mmol) in absolute ethanol (35 mL). This mix was stirred at room temperature for 3 h, and the small quantity of solid formed was separated by centrifugation and discarded. The obtained solution was filtered to remove any possible impurities, and then it was allowed to steadily evaporate until single crystals of [Dy(L)(Cl)(H₂O)₂] (1), suitable for X-ray diffraction studies, were obtained. The single crystals were filtered, and then dried in an oven. Yield: 0.072 (58%). Elemental analysis calcd. for C₁₉H₁₇DyN₃O₄Cl (549.31): C 41.54, N 7.65, H 3.12%. Found: C 42.11, N 7.68, H 3.41%. IR (ATR, $\tilde{\mathbf{\nu }}$/cm⁻¹): 1533 (vCN_py), 1583 (vCN_imine), 3315 (vOH).

The same compound was isolated when this reaction was repeated exactly with the same quantity of reactants and solvents but heating at reflux. Elemental analysis calcd. for C₁₉H₁₇DyN₃O₄Cl (549.31): C 41.54, N 7.65, H 3.12%. Found: C 41.64, N 7.82, H 3.50%.

3.3. Single X-ray Crystallographic Refinement and Structure Solution

Table S5 summarises the crystal data and the details of the refinement for the crystal structure of the metal complex 1. Single crystals of 1, suitable for single X-ray diffraction analysis, were isolated as detailed previously. The data were recorded on a Bruker D8 VENTURE PHOTON III-14 diffractometer at a temperature of 100 K, using graphite monochromatised Mo-Kα (λ = 0.71073 Å) radiation. Multi-scan absorption corrections were carried out through the use of the SADABS routine [27]. The structure was resolved by means of standard direct methods, with the software SHELXT, and next refined by full matrix least squares techniques on F² by utilising SHELXL [28] from the program package SHELX-2018 [28]. The atoms that were different to hydrogen were anisotropically refined, whereas hydrogen atoms were normally included in the structure factor calculations in geometrically idealised positions. Hydrogen atoms bonded to oxygen atoms (water) were found in the Fourier map, with the aim of obtaining the hydrogen bond scheme. These atoms were freely refined in this case.

3.4. Powder X-ray Diffraction Studies

X-ray powder measurements for 1 were also performed. The powder X-ray diffraction data of 1 were registered on a Philips diffractometer, equipped with a “PW1710” type unit control, a vertical “PW1820/00” type goniometer, and an “Enraf Nonius FR590” type generator, functioning at 40 kV and 30 mA, using monochromatised radiation (Cu-Kα, λ = 1.5418 Å) radiation. A scan in the range 2 < 2θ < 30° was performed, with t = 3 s and Δ2θ = 0.02°. LeBail refinement was achieved with the help of HighScore Plus Version 3.0d.

3.5. Magnetic Measurements

Experimental magnetic susceptibility dc and ac measurements were recorded for a microcrystalline sample of 1 on a PPMS Quantum Design susceptometer. The dc magnetic susceptibility data were registered between 2 and 300 K, under a magnetic field of 1000 Oe. Magnetisation measurements at 2.0 K were also performed, in this case under magnetic fields in the range 0 to 50,000 Oe. Diamagnetic corrections for the data were assumed on the basis of Pascal’s tables. Alternating current (ac) susceptibility measurements in the temperature range 0–10 K in the absence of an external dc field (H_dc = 0), and at H_dc = 1500 Oe, were recorded in an oscillating alternating current field of 3.5 Oe, with ac frequencies varying from 50 to 10,000 Hz. ac susceptibility measurements were also collected at the fixed temperature of 5 K at different magnetic fields (500 Oe, 800 Oe, 1000 Oe, 1500 Oe, 2000 Oe, 2500 Oe, 3000 Oe, and 3500 Oe) and at frequencies ranging between 50 and 2000 Hz in order to determine the optimum field for eliminating the quantum tunnel of the magnetization.

4. Conclusions

The isolation of the neutral mononuclear complex [Dy(L)Cl(H₂O)₂] (1) was favoured over that of the related ionic compound [Dy(HL’)₂)][Dy(L)(Cl₂)], in which the Schiff base was partially hydrolysed, simply by changing the methanol reaction solvent by absolute ethanol. Compound 1 behaves as a single ion magnet in the presence of an optimal external magnetic field of 1500 Oe, with an U_eff barrier of 113.5 K. This behaviour is promoted by the easy axis of the anisotropy, which points to the vertex of a biaugmented trigonal prism, rendering the highest barrier for the inversion of the magnetization among the scarcely reported mononuclear dysprosium complexes of H₂L, and the related Dy^III compounds with planar N₃O₂ hydrazones as ligands and coordination numbers 8 or 9. Therefore, this study presents a new example of a field-induced SIM with a rigid acyclic ligand, thus contributing to an increase in the knowledge of magnetic properties in this field where examples are still fairly scarce. Thus, 1 seems to show that the use of monodentate auxiliary ligands decreases the rate of coordination compared to ancillary bidentate ligands, so improving the energy barrier for magnetisation reversal.