Magnetic Anisotropy of Homo- and Heteronuclear Terbium(III) and Dysprosium(III) Trisphthalocyaninates Derived from Paramagnetic ¹H-NMR Investigation

Abstract

¹H-NMR spectroscopy of lanthanide complexes is a powerful tool for deriving spectral–structural correlations, which provide a clear link between the symmetry of the coordination environment of paramagnetic metal centers and their magnetic properties. In this work, we have first synthesized a series of homo- (M = M* = Dy) and heteronuclear (M ≠ M* = Dy/Y and Dy/Tb) triple-decker complexes [(BuO)₈Pc]M[(BuO)₈Pc]M*[(15C5)₄Pc], where BuO- and 15C5- are, respectively, butoxy and 15-crown-5 substituents on phthalocyanine (Pc) ligands. We provide an algorithmic approach to assigning the ¹H-NMR spectra of these complexes and extracting the axial component of the magnetic susceptibility tensor, ${\chi }_{ax}$. We show how this term is related to the nature of the lanthanide ion and the shape of its coordination polyhedron, providing an experimental basis for further theoretical interpretation of the revealed correlations.

1. Introduction

Single-molecule magnetism (SMM) is one of the most intriguing manifestations of the anisotropic coordination environment of paramagnetic lanthanides in sandwich complexes with tetrapyrrolic ligands, which was actually first described for this family of elements using the example of terbium(III) bisphthalocyaninate, Tb(Pc)₂. Since the seminal report by Ishikawa et al. two decades ago [1], numerous examples of sandwiches containing mainly Tb³⁺ and Dy³⁺ metal centers have been synthesized and extensively studied in terms of magnetic relaxation dynamics, and the potential for fabrication of spintronic devices based on them has been clearly demonstrated [2,3,4].

Among such complexes, we can distinguish a prominent class of triple-decker phthalocyaninates, in which two metal ions bind three ligands, providing a wide range of combinations of different macrocycles and metal centers to achieve desired functional characteristics [5,6,7]. While the synthesis and properties of a large number of heteroleptic triple-deckers have been reported [8], the family of heteronuclear derivatives containing two different rare-earth elements (REEs) is still relatively underexplored. At the same time, such a combination provides unique possibilities for studying the subtle effects of intramolecular interactions between paramagnetic ions separated by a distance of only 3.4–3.5 Å.

The first vivid contribution to such studies was made by Ishikawa et al., who synthesized three series of heteroleptic triple-deckers [(BuO)₈Pc]M*(Pc)M(Pc), where [(BuO)₈Pc]²⁻ and (Pc)²⁻ were octa-butoxy-substituted and unsubstituted phthalocyaninato ligands, respectively [9,10]. One series consisted of homonuclear complexes with M* = M = Tb³⁺, Dy³⁺, Ho³⁺, Er³⁺, Tm³⁺ and Yb³⁺, and the other two families were isomeric heteronuclear complexes containing a diamagnetic Y³⁺ ion at either the M* or M site, while the other site was occupied by one of the listed lanthanides. A comparison of the temperature dependence of magnetic susceptibilities between homo- and heteronuclear complexes has provided the first detection and characterization of the dipolar f-f interactions between Ln³⁺ ions.

Later, the influence of these interactions on the behavior of SMMs based on triple-deckers was revealed. It was demonstrated that heteronuclear complexes containing one dia- and one paramagnetic ion, Tb³⁺ and Y³⁺, acted as field-induced SMMs, while the homonuclear di-terbium complexes showed slow magnetic relaxation even in a zero dc field. Such a behavior was attributed to the f-f interactions acting as an exchange bias that suppresses the undesired quantum tunneling mechanism of magnetic relaxation [11,12,13]. This effect was first demonstrated on the examples of the aforementioned [(BuO)₈Pc]M*(Pc)M(Pc) complexes [14], and a deeper insight into this phenomenon was gained with the 15-crown-5-substituted sandwiches [(15C5)₄Pc]M*[(15C5)₄Pc]M(Pc) reported by R. Holmberg et al. [15]. In both cases, M* = M = Tb or M* ≠ M = Tb/Y.

The synthesis of heteronuclear trisphthalocyaninates is relatively straightforward. It implies raise-by-one-story reactions of bisphthalocyaninates bearing one metal center with monophthalocyaninates bearing another metal center. The latter precursors can be either synthesized beforehand [16,17,18] (Figure 1a), generated in situ from metal-free ligands [9] (Figure 1b) or by transmetalation of other complexes [15,19] (Figure 1c).

In general, these approaches can be used to obtain thermodynamically and kinetically stable heteronuclear triple-deckers with any pair of middle and late lanthanide ions, including complexes with two paramagnetic ions. Such an interplay between different combinations of nonequivalent magnetically active nuclei placed in well-defined positions can be used to study f-f interactions and provides a tool for fine-tuning the SMM behavior of such complexes. A rare example of such complexes—[(n-C₈H₁₃)₈Pc]Tb(Pc)Dy(Pc)—has been reported by Lan et al. [20]. A weak but significant interaction between the two lanthanides was clearly detected, modifying the magnetic behavior of the single lanthanide as observed in the parent mononuclear complexes.

Trisphthalocyaninates also attract attention as conformationally flexible scaffolds where the twist angle between the ligands can be controlled by various inter- and intramolecular factors, including host–guest interactions [21,22], solvation [23,24] and redox transformations [25,26]. The variation in the twist angle inevitably changes the symmetry of the coordination polyhedron of the lanthanide ion sandwiched between these ligands, which has a profound effect on its magnetic properties due to the influence of this symmetry on magnetic anisotropy [27] and energetic characteristics and preferable mechanisms of magnetic relaxation [11,28,29].

Recently, we reported a series of works on conformationally flexible triple-decker complexes composed of butoxy- and crown-substituted phthalocyaninato ligands [(BuO)₈Pc]M[(BuO)₈Pc]M*[(15C5)₄Pc], including homonuclear complexes M = M* = Tb³⁺ and Y³⁺ [23] and their heteroleptic counterparts M ≠ M* = Tb³⁺ and Y³⁺ [24]. We showed that the pairwise conformations of the ligands are solvation-dependent: in halogenated alkanes, the pair of BuO- and 15C5-substituted ligands adopts the staggered conformation with a twist angle of ~45°, while the pair of BuO-substituted ligands forms the gauche conformation with a twist angle of ~22° (Figure 2). In turn, it provides metal centers M* and M with square-antiprismatic (SAP) and distorted prismatic (DP) coordination surroundings, respectively. In contrast, solvation with aromatic solvents leads to the stabilization of staggered conformations for both ligand pairs, so that both metal centers exist in SAP environments. The stabilization of certain conformers was explained by X-ray diffraction analysis of the solvates with dichloromethane and toluene, where weak contacts with the solvent molecules were found and analyzed in terms of the quantum theory of atoms in molecules [23]. Moreover, these conformations were stable not only in the crystalline state but also in solution, which was confirmed by ¹H-NMR characterization of Tb³⁺ complexes—their magnetic anisotropy ${\chi }_{ax}$ strongly increases upon lowering the coordination symmetry from SAP to DP, as evidenced from the structure-based analysis of lanthanide-induced shifts.

In the present work, we extended the developed synthetic and analytical approaches to Dy³⁺ complexes, including a homonuclear triple-decker with M = M* = Dy³⁺ and pairs of isomeric heteronuclear analogues with M/M* = Dy³⁺/Y³⁺ and Y³⁺/Dy³⁺, as well as Dy³⁺/Tb³⁺ and Tb³⁺/Dy³⁺. For the sake of brevity, we will henceforth refer to butoxy- and crown-substituted ligands as [B₄] and [C₄], respectively, following our previous notation [23,24] (Figure 2). Assuming that the axial anisotropy of the magnetic susceptibility tensor ${\chi }_{ax}$ deduced from NMR correlates with ligand field parameters [30,31,32,33,34], which, in turn, determine SMM behavior [26,35], these studies serve to verify the theoretical models and to design molecular magnets with improved characteristics [36].

2. Results

2.1. Synthesis and Characterization

The synthesis of the [B₄]M[B₄]M*[C₄] complexes with the aforementioned combinations of metal centers followed the procedure reported for Y³⁺ and Tb³⁺ complexes (Figure 3). It started with the preparation of bis(octa-n-butoxyphthalocyaninates) M[B₄]₂, M = Tb³⁺, Dy³⁺ and Y³⁺ by the interaction of H₂[B₄] with the corresponding metal acetates in a refluxing mixture of 1-chloronaphthalene (CN) and 1-octanol (OctOH) and DBU. Previously, we demonstrated that the use of this mixture is beneficial for the fast and efficient conversion of the starting ligand into metal complexes [23].

The resulting complexes underwent a reaction with tetra-15-crown-5-phthalocyanine, H₂[C₄], in the presence of acetylacetonates M*(acac)₃·nH₂O in a refluxing mixture of 1,2,4-trichlorobenzene (TCB) and 1-octanol. Rapid conversion of the starting reagents into the desired triple-deckers was observed using UV-vis spectroscopy, and the target complexes were isolated by column chromatography on alumina. Due to the difference in polarity, [B₄]M[B₄]M*[C₄] could be easily separated from the unreacted M[B₄]₂ and the side homoleptic products M*₂[C₄]₃.

UV-vis spectra measured in toluene and chloroform demonstrate that dysprosium-containing complexes exhibit pronounced solvatochromism, similarly to the previously reported Tb- and Y-containing [B₄]M[B₄]M*[C₄] complexes [23,24] (Figure 4). In particular, the spectra of the newly synthesized complexes in toluene contain intense Q-bands with well-resolved splitting (642–643 and 695–698 nm) and less intense Soret and N-bands at 363–364 and 292–294 nm, while in the spectra in chloroform, the Q-bands are severely broadened, although the maxima of the Q-bands in the two different solvents are very close. This has been previously explained by the difference in the conformational state of molecules, as in the spectrum of the conformer with DP/SAP coordination polyhedral, more electronic transitions are symmetrically allowed in comparison with the SAP/SAP conformer [25,37]. Since the positions and splitting of Q-bands in the UV-vis spectra of sandwich complexes are governed by the interligand distances, which in turn depend on the size of the metal centers [19,37,38], the faintly small difference in the spectral appearances of the synthesized complexes in each of the solvents is explained by the similarity of the radii of Y³⁺, Tb³⁺ and Dy³⁺ ions [39].

The very weak absorption of lanthanide ions in the synthesized complexes could not be detected due to the much stronger absorption of tetrapyrrolic ligands. Also, we could not detect f-luminescence of Tb³⁺ and Dy³⁺ ions due to reabsorption of weak lanthanide-centered emission by tetrapyrrolic ligands, which have absorption bands with extinction coefficients of 1.4−2.0 × 10⁴ L mol⁻¹ cm⁻¹ in the range of 400–600 nm, where the emission of these ions is typically observed in coordination compounds with colorless or weakly absorbing ligands [40]. Nevertheless, there are reports of f-luminescence observed in the near-IR for Nd(III), Ho(III), Er(III) and Yb(III) and complexes where Pcs and related macrocycles typically do not absorb light [41].

While optical methods are not suitable to confirm the presence of lanthanides in complexes with phthalocyanine ligands, the chemical composition of the synthesized triple-deckers is unambiguously confirmed by MALDI-TOF mass spectrometry due to the good agreement between the calculated and experimentally observed isotopic distributions (Figures S1–S5). The exact arrangement of the ligands and metal centers is determined by NMR spectroscopy, as discussed in the following sections.

2.2. Analysis of Lanthanide-Induced Shifts in ¹H-NMR Spectra

The ¹H-NMR spectra of lanthanide complexes typically have strikingly different appearances from the spectra of diamagnetic organic compounds since the presence of paramagnetic metal centers causes up- or downfield shifts of resonance signals by tens or even hundreds of ppm. The signs and magnitudes of these lanthanide-induced shifts (LISs) depend on both the nature of the lanthanide ion [42,43,44,45] and the overall geometry of the complex [46,47].

Thus, for the k-th proton, the LIS can be expressed as the difference between its chemical shifts in isostructural para- and diamagnetic complexes (1):

$$\mathsf{\Delta }{\delta }_k={\delta }_k^{para}-{\delta }_k^{dia}$$

(1)

On the other hand, the LIS can be presented as the sum of typically negligible contact (through-bond, ${\delta }_k^{con}$) and predominant dipolar (through-space, ${\delta }_k^{dip}$) contributions:

$$\mathsf{\Delta }{\delta }_k={\delta }_k^{con}+{\delta }_k^{dip}\approx \frac{{\chi }_{ax}}{12\pi }·G_{k,} G_k=\frac{3{\cos }^2{\theta }_k-1}{r_k^3}$$

(2)

Here, $G_k$ is a geometrical parameter depending on the distance $r_k$ between the k-th proton and the lanthanide ion, and ${\theta }_k$ is the angle between the vector ($\overline{{\mathrm{H}}_k;{\mathrm{Ln}}^{3+}}$) and the main symmetry axis, D₄ in our case (Figure 5a). The spectral–structural correlation (2) renders lanthanide ions as perfect probes for the elucidation of the solution structures, for example, in structural biology [48,49].

The proportionality factor ${\chi }_{ax}$ is the axial component of the magnetic susceptibility tensor. Thus, the observation of lanthanide-induced shifts in NMR spectra is a manifestation of lanthanide magnetic properties associated with crystal field parameters [42,43,44,46,47], and the value of ${\chi }_{ax}$ complements the easily affordable NMR with much more sophisticated time- and resource-consuming magnetochemical measurements [10,31,50,51,52,53].

The dipolar approximation of the LIS using Equation (2) suggests that the ratio of LISs for the pair of k-th and l-th protons can be approximated with the ratio of their geometrical parameters, $R_{kl}$:

$$\frac{\mathsf{\Delta }{\delta }_k}{\mathsf{\Delta }{\delta }_l}\approx R_{kl}, R_{kl}=\frac{G_k}{G_l}$$

(3)

A combination of Equations (1) and (3) written for a pair of protons gives Equation (4), which can be used to estimate the positions of the resonance signals in the ¹H-NMR spectra of paramagnetic complexes using the spectrum of a diamagnetic complex with a well-established structure and the shift of at least one firmly assigned resonance signal in the spectrum of a paramagnetic complex:

$${\delta }_k^{para}\approx {\delta }_k^{dia}+\mathsf{\Delta }{\delta }_l·R_{kl}$$

(4)

Yttrium(III) complexes are typically used as diamagnetic references, providing ${\delta }_k^{dia}$ values, and the ratio of geometrical parameters $R_{kl}$ can be taken from the structural models obtained from either X-ray characterization or DFT modeling [22].

The almost perfect agreement of the UV-vis spectra of the newly synthesized complexes with those reported previously (Figure 4) suggests an analogy in the solution structure of the entire series of [B₄]M[B₄]M*[C₄], which is generally expected given the close values of the ionic radii of Y³⁺ (1.019 pm), Tb³⁺ (1.040 pm) and Dy³⁺ (1.027 pm) [39], together with the similarity of their coordination chemistry [4]. This conclusion justifies the following application of the previously determined X-ray structures of [B₄]Y[B₄]Y[C₄] solvated either with dichloromethane (DP/SAP conformer) or with toluene (SAP/SAP conformer) [23] for the analysis of ¹H-NMR spectra of paramagnetic complexes containing Tb³⁺ and Dy³⁺ ions using Equation (4).

It is worth noting that the X-ray structures of the solvates with CHCl₃ are not yet available, although we attempted to grow single crystals by diffusion of acetonitrile, heptane or vapors of pentane into [B₄]Y[B₄]Y[C₄] solutions in CHCl₃ but failed to obtain material of sufficient quality to perform XRD experiments with the required precision. On the other hand, we previously demonstrated that chloroform also stabilizes staggered and gauche conformers of 15C5- and BuO-substituted triple-deckers, similarly to dichloromethane (see Figure S6 for more details). Moreover, we previously showed that the spectra of [B₄]Tb[B₄]Tb[C₄] in CDCl₃ and CD₂Cl₂ are almost indistinguishable [23]. Altogether, this suggests that the X-ray structure of the DP/SAP conformer solvated with CH₂Cl₂ is a reasonable structural model for the analysis of LISs measured in chloroform.

2.2.1. Analysis of Lanthanide-Induced Shifts in ¹H-NMR Spectra of Heteronuclear Complexes [B₄]Dy[B₄]Y[C₄] and [B₄]Y[B₄]Dy[C₄]

The ¹H-NMR spectra of the pair of isomeric trisphthalocyaninates [B₄]Dy[B₄]Y[C₄] and [B₄]Y[B₄]Dy[C₄] measured in CDCl₃ and toluene-d₈ evidence that the spectral range (SR) from the most upfield-shifted to the most downfield-shifted signals depends on both the position of the Dy³⁺ ion and the solvent applied (Figure 5b,c).

Only aromatic protons and methylene protons proximal to the Pc ligands—1^i,o-CH₂ groups of butoxy-substituents and α^o-CH₂ groups of crown-ether rings (black labels in Figure 5b)—were taken for further analysis as they refer to the most rigid part of the molecules, assuring the consistency between the X-ray and solution structures [23,24]. Although CH₂ groups can rotate around single σ-bonds, the interconversion between the exo- and endo-H protons of the considered methylenes is slow on the NMR timescale; their resonances appear as pairs of signals separated by ca. 10 ppm and coupled in ¹H-¹H COSY spectra. Such hindered rotation can be explained by numerous weak interactions of the aliphatic substituents with the substituents in the neighboring ligands [23] and solvent molecules [25]. Similar splitting patterns were reported previously by Enders and Yamashita et al. [26,54], Jiang et al. [55] and Ishikawa et al. [56], together with our publications [19,23,24].

The splitting is much smaller for more distant methylene groups (grey labels in Figure 5b), and it almost vanishes for the 3^i,o-CH₂ and CH₃ groups, suggesting that these groups rotate nearly freely around the corresponding sigma bonds. Thus, these distant protons were not taken into consideration so that the deduced ${\chi }_{ax}$ value is not disturbed by the conformational flexibility of the peripheral substituents. Good correlations between the experimentally observed chemical shifts and the shifts calculated from X-ray data (Figure 5c–f) validate the applied assumptions and approximations.

The multidirectional nature of LISs observed in Figure 5c–f in such complexes originates from the fact that different protons have either positive or negative $G_k$ depending on the location of these protons in the principle magnetic framework of the Dy³⁺ ion (Figure 6b,c), and the largest absolute value of $G_k$ deduced from X-ray data corresponds to the aromatic proton of the inner BuO-substituted ligand bH_Pcⁱ [24]. This observation allowed us to assign the resonance signals of the other protons using Equation (4), which was confirmed by further ¹H-¹H COSY spectra (Figures S7–S14).

Plotting LIS from NMR spectra vs. the geometrical parameters found from X-ray data, followed by linear regression analysis, afforded axial anisotropies of Dy³⁺ ions in different coordination surroundings (Figure 6a). This analysis demonstrates that the magnetic properties of this ion are sensitive to the symmetry of the coordination surrounding. Thus, switching from an SAP to a DP polyhedron in toluene and chloroform, respectively, causes a notable increase in axial anisotropy, similar to the results reported previously for Tb³⁺ ions in isostructural complexes.

2.2.2. Analysis of Lanthanide-Induced Shifts in ¹H-NMR Spectra of the Homonuclear Complex [B₄]Dy[B₄]Dy[C₄]

Due to the presence of two structurally nonequivalent paramagnetic centers in [B₄]Dy[B₄]Dy[C₄], Equation (4) cannot be applied for the assignment of its ¹H-NMR spectra, as these centers might give different contributions to the total LIS value. Therefore, in the present case, the assignment is made by approximating the chemical shift of the k-th proton in the homonuclear complex with the sum of the shifts of analogous protons in the heteronuclear complexes according to Equation (5) [56].

$${\delta }_k^{\left[B_4\left]Dy\right[B_4\left]Dy\right[C_4\right]}\approx {\delta }_k^{\left[B_4\left]Dy\right[B_4\left]Y\right[C_4\right]}+{\delta }_k^{\left[B_4\left]Y\right[B_4\left]Dy\right[C_4\right]}$$

(5)

The assignment is confirmed by ¹H-¹H COSY. In addition, due to the proximity of all the considered protons to the Dy³⁺ ions, their signals are the most broadened, providing additional verification of their correct assignment.

A comparison of the ¹H-NMR spectra of the homonuclear complex [B₄]Dy[B₄]Dy[C₄] measured in CDCl₃ and toluene-d₈ (Figure 7a,b) suggests that switching from the SAP/SAP conformer to the DP/SAP conformer also causes an increase in anisotropy, as evidenced by the broadening of the spectral range. To confirm this conclusion, we assumed that each of the Dy³⁺ ions has its own ${\chi }_{ax}$ value in accordance with Equation (6); thus, we used two-dimensional minimization of the Wilcott factor (AF, Equation (7)) across the various pairs of ${\chi }_{ax}$ to find the best agreement between the calculated and experimental LIS values (Figure 7c,d).

$$\Delta {\delta }_k^{\left[B_4\left]Dy\right[B_4\left]Dy\right[C_4\right]}\approx {\left[\frac{{\chi }_{ax}}{12\pi }·G_k\right]}^{\left[B_4\left]Dy\right[B_4\left]Y\right[C_4\right]}+{\left[\frac{{\chi }_{ax}}{12\pi }·G_k\right]}^{\left[B_4\left]Y\right[B_4\left]Dy\right[C_4\right]}$$

(6)

$$AF=\sqrt{\frac{{{\displaystyle \sum }}_k{\left({\delta }_k^{calc}-{\delta }_k^{exp}\right)}^2}{{{\displaystyle \sum }}_k{{\delta }_k^{exp}}^2}}$$

(7)

The results presented in Figure 7 suggest that the change in anisotropy in the homonuclear complex [B₄]Dy[B₄]Dy[C₄] follows the same trends observed in heteronuclear complexes. In particular, switching between SAP/SAP (in toluene-d₈) and DP/SAP (in CDCl₃) causes an increase in the anisotropy of the ion in the conformationally flexible [B₄]/[B₄] site from 3.95 × 10⁻³¹ to 4.28 × 10⁻³¹ m³, while the ${\chi }_{ax}$ value of the Dy³⁺ ion in the conformationally invariant [B₄]/[C₄] site remains almost the same—3.9 × 10⁻³¹ m³ (Figure 8a).

Plotting the contour maps of $\mathsf{\Delta }{\delta }_k^{para}$ for the molecules of the [B₄]Dy[B₄]Dy[C₄] complex in CDCl₃ and toluene-d₈ using the derived ${\chi }_{ax}$ values explains why all signals in the spectra have shifts of the same sign—all protons fall into the region with the same sign of the net $G\left(\theta ,r\right)$ function, in contrast to the heteronuclear derivatives, where protons fall into regions with different signs of the geometrical parameter (Figure 8b,c).

2.2.3. Analysis of Lanthanide-Induced Shifts in ¹H-NMR Spectra of Heteronuclear Complexes [B₄]Dy[B₄]Tb[C₄] and [B₄]Tb[B₄]Dy[C₄]

Following the procedure described above, we assigned and analyzed the spectra of two isomeric complexes, [B₄]Dy[B₄]Tb[C₄] and [B₄]Tb[B₄]Dy[C₄], in CDCl₃ and toluene-d₈ (Figure 9) to trace the change in the axial anisotropy of both lanthanide ions simultaneously using Equations (5) and (6). The assignment was confirmed by ¹H-¹H COSY (Figures S13–S16).

It can be clearly seen that the change from CDCl₃ to toluene-d₈ has the most pronounced influence on the ¹H-NMR spectra of [B₄]Tb[B₄]Dy[C₄], where the Tb³⁺ ion is placed into the switchable [B₄]/[B₄] site (Figure 9a,b), which is expectedly followed by a significant decrease in its axial anisotropy—from 9.40 × 10⁻³¹ to 7.75 × 10⁻³¹ m³—in agreement with the behavior of the previously studied Tb³⁺ complexes [23,24]. Switching the coordination polyhedron of the Dy³⁺ ion from DP to SAP also causes a decrease in axial anisotropy (Figure 9c,d), although this effect is not so strong compared to the Tb³⁺ metal centers (Figure 10a,b).

Finally, the availability of ${\chi }_{ax}$ for each metal center allows plotting the contour maps of the overall LIS values for different conformers of [B₄]Tb[B₄]Dy[C₄] and [B₄]Dy[B₄]Tb[C₄] (Figure 10c–f). These plots vividly explain the difference between the spectral appearances of these complexes and their homonuclear analogues. For example, it can be seen that a combination of two metal ions with essentially different anisotropies results in deformation of the zero $\mathsf{\Delta }\delta$ isosurface, which approaches the protons of the terminal ligands bound to the Dy³⁺ ion; this is clearly seen in the experimental spectra, in which the less downfield-shifted signals refer to aromatic protons (either bH_Pc^o or cH_Pc^o) and exo-protons of methylene groups (1^o′ or α^o′, respectively).

3. Discussion

The presented results give several guidelines for further work on lanthanide complexes with phthalocyanine ligands:

• The trisphthalocyanine scaffold affords the synthesis of heteronuclear complexes with a precise arrangement of rare-earth ions due to its thermodynamical and kinetic stability. Complexes with different combinations of paramagnetic lanthanides can be efficiently obtained, which provides the basis for further investigation of intramolecular f-f interactions and the elaboration of molecular magnetic materials. In the present work, the combinations of two strongly paramagnetic Tb³⁺ and Dy³⁺ ions were used to obtain isomeric heteronuclear complexes, but, obviously, other combinations of middle and late lanthanides can also be used in this type of chemistry to obtain complexes with the required number of unpaired f-electrons.
• The comprehensive ¹H-NMR spectroscopic characterization of strongly paramagnetic complexes is based on the appropriate structural model; therefore, this work provides algorithms for dealing with the spectra of complexes containing either one or two lanthanide ions, which are not necessarily equivalent. In this regard, the results presented demonstrate that the application of paramagnetic ¹H-NMR spectroscopy should not be limited to routine identification but can be used to extract the magnetic properties of lanthanide ions [53]. In this context, our report follows the strategies applied by Enders and Yamashita, where Tb(III) and Dy(III) sandwich phthalocyaninates were comprehensively studied using NMR spectroscopy [26,35,54,57,58], and the influence of electronic and structural effects on their magnetic properties, especially ${\chi }_{ax}$, was revealed. In summary, it is expected that further magnetochemical studies of the newly synthesized lanthanide phthalocyaninates will provide more correlations between the ${\chi }_{ax}$ term and the energetic properties of slow magnetic relaxation.
• The addition of controllable conformational flexibility gives one more degree of freedom to control the magnetic properties of sandwich lanthanide complexes. The previously discovered correlations between the symmetry of the coordination polyhedron of the Tb³⁺ ion and its magnetic properties are also valid for the Dy³⁺ ion, and the effect of the conformational switching can be studied for other lanthanides to find the capabilities and limitations of theoretical models.
• Importantly, our study evidences that the ${\chi }_{ax}$ values for Dy³⁺ are nearly twice smaller than those for Tb³⁺; however, this observation does not match the expectations from Bleaney’s theory, where the largest anisotropy in the lanthanide series is expected for dysprosium [43]. Moreover, it contradicts the results of theoretical modeling obtained by Mironov et al. [32] for various polyhedra of lanthanide complexes, where the most pronounced influence of the surrounding coordination was anticipated for dysprosium complexes. The reason for this discrepancy may be a violation of the theory’s basic assumption that the thermal energy is larger than the ligand field splitting; thus, further theoretical modeling using ab initio methods might be particularly helpful [59,60].

4. Materials and Methods

4.1. Materials

Starting phthalocyanines H₂[B₄], Y[B₄]₂, Tb[B₄]₂, Dy[B₄]₂ and H₂[C₄] were synthesized according to the previously reported procedures [61,62]. 1,2,4-Trichlorobenzene (TClB, for synthesis, 1-octanol (for synthesis), rare-earth acetylacetonates (Sigma-Aldrich, Burlington, MA, USA), and neutral alumina (50–200 μm, Macherey-Nagel, Düren, Germany) were used as received from the commercial suppliers. Chloroform (reagent grade, Ekos-1, Staraya Kupavna, Russia) was distilled over CaH₂.

4.2. Methods

Matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectra were measured on a Bruker Daltonics Ultraflex spectrometer. Mass spectra were registered in positive ion mode using 2,5-dihydroxybenzoic acid as a matrix. UV-vis spectra in the range of 250–900 nm were measured using a Thermo Evolution 210 spectrophotometer in quartz cells with 0.5–1.0 cm optical path lengths.

¹H NMR spectra were recorded at 303 K on a Bruker Avance III 600 MHz spectrometer equipped with a 5 mm Z-gradient BBO probe (zg30 pulse program from Topspin library). A total of 128 scans with a 30° excitation pulse (90° pulse width was 13.8 μs, DE 6.50 μs) and 1 s delay were accumulated for each system. Spectral width was taken to be 90–100 kHz, depending on the combination of lanthanides. Gaussian multiplication was used for processing. Automatic polynomial baseline correction was applied. For recording 2D COSY spectra (cosygpqf program from Topspin library), spectral width was selected to be 40–75 ppm, depending on the combination of lanthanides. The residual solvent resonances were used as internal references (δ toluene 7.09 ppm, chloroform 7.26 ppm). Typically, 5 mg of complexes was dissolved in 0.6 mL of a corresponding deuterated solvent to provide a concentration of ca. 2.3 mM. The applied deuterated chloroform (99.8 atom% D, ZEOchem, Uetikon am See, Switzerland) was filtered prior to use through Pasteur pipettes filled with alumina to remove possible acidic impurities. Deuterated toluene (99.8 atom% D, ABCR, Karlsruhe, Germany) was used without additional purification.

4.3. Synthesis and Characterization of the Triple-Decker Complexes

Trisphthalocyaninate [B₄]Dy[B₄]Y[C₄]: A solution of phthalocyanines Dy[B₄]₂ (30.0 mg, 12.8 μmol) and H₂[C₄] (20.4 mg, 16.0 μmol) in a mixture of 2.7 mL 1,2,4-trichlorobenzene and 0.3 mL n-octanol was heated to reflux at 230 °C under a stream of argon, and solid yttrium (III) acetylacetonate hydrate Y(acac)₃·nH₂O (19.4 mg, 48.1 μmol) was added. After 10 min, the reaction mixture was cooled to ambient temperature. The reaction mixture was transferred to a chromatographic column packed with alumina in a mixture of CHCl₃ and hexane (1:1 v/v). The target complex was isolated by elution with a CHCl₃-hexane mixture (4:1 v/v), followed by a mixture of CHCl₃ with 0→1% MeOH as a dark-blue solid (29 mg, yield 61%).

MALDI TOF: m/z calculated for C₁₉₂H₂₃₂DyN₂₄O₃₆Y 3703.5, found 3704.2 [M⁺].

UV-vis (CHCl₃) λ_max (nm) (log ε): 647 (5.06), 546 (4.50), 355 (5.21), 295 (5.16).

UV-vis (Toluene) λ_max (nm) (log ε): 698 (4.64), 642 (5.40), 363 (5.24), 294 (5.09).

¹H-NMR (600 MHz, CDCl₃) δ −31.71 (br s, 8H, bH_Pcⁱ), −27.34 (br s, 8H, bH_Pc^o), −17.16 (br s, 8H, 1^ib), −12.32 (br s, 8H, 1^o), −10.87 (br s, 8H, 1^ic), −8.88 (s, 8H, 2^ib), −8.62 (s, 8H, 2^ic), −7.92 (s, 8H, 3^ib), −7.65 (s, 8H, 3^ic), −7.14 (br s, 8H, 1^o′), −5.91 (s, 32H, CH₃ⁱ + 2^o), −5.69 (s, 8H, 2^o′), −4.89 (s, 16H, 3^o,o′), −3.50 (s, 24H, CH₃^o), 2.77 (s, 8H, β^o), 2.83 (m, 8H, γ^o), 2.96 (m, 8H, γ^o′), 3.13 (d, 25 Hz, 8H, α^o), 3.62 (m, 16H, δ^o,o′), 4.13 (s, 8H, β^o′), 6.95 (d, 25 Hz, 8H, α^o′), 16.34 (br s, 8H, cH_Pc^o).

¹H-NMR (600 MHz, Toluene-d₈) δ −25.33 (br s, 8H, bH_Pcⁱ), −22.01 (br s, 8H, bH_Pc^o), −14.89 (br s, 8H, 1^ib), −11.06 (br s, 8H, 1^o), −8.37 (br s, 8H, 1^ic), −6.40 (s, 8H, 2^ib), −6.07 (s, 8H, 2^ic), −5.95 (s, 8H, 3^ib), −5.70 (s, 8H, 3^ic), −5.12 (s, 8H, 1^o′), −4.88 (s, 8H, 2^o), −4.80 (s, 8H, 2^o′), −4.33 (br t, 24H, CH₃ⁱ), −3.99 (s, 16H, 3^o,o′), −2.79 (br t, 24H, CH₃^o), 2.68 (s, 8H, β^o), 3.08 (d, 30 Hz, 8H, α^o), 3.18 (br m, 8H, γ^o), 3.62 (br m, 8H, δ^o), 3.74 (br m, 8H, γ^o′), 3.81 (s, 8H, β^o′), 4.03 (s, 8H, δ^o′), 6.81 (d, 30 Hz, 8H, α^o′), 16.86 (br s, 8H, cH_Pc^o).

Trisphthalocyaninate [B₄]Y[B₄]Dy[C₄]: A solution of phthalocyanines Y[B₄]₂ (23.0 mg, 10.0 μmol) and H₂[C₄] (16.0 mg, 12.6 μmol) in a mixture of 2.7 mL 1,2,4-trichlorobenzene and 0.3 mL n-octanol was heated to reflux at 230 °C under a stream of argon, and solid dysprosium (III) acetylacetonate hydrate Dy(acac)₃·nH₂O (18.0 mg, 39.0 μmol) was added. After 8 min, the reaction mixture was cooled to ambient temperature. The reaction mixture was transferred to a chromatographic column packed with alumina in a mixture of CHCl₃ and hexane (1:1 v/v). The target complex was isolated by elution with a CHCl₃-hexane mixture (4:1 v/v), followed by a mixture of CHCl₃ with 0→1% MeOH as a dark-blue solid (21.2 mg, yield 57%).

MALDI TOF: m/z calculated for C₁₉₂H₂₃₂DyN₂₄O₃₆Y 3703.5, found 3704.3 [M⁺].

UV-vis (CHCl₃) λ_max (nm) (log ε): 646 (4.95), 546 (4.43), 354 (5.14), 294 (5.10).

UV-vis (Toluene) λ_max (nm) (log ε): 698 (4.65), 642 (5.40), 363 (5.25), 293 (5.11).

¹H-NMR (600 MHz, CDCl₃) δ −29.74 (br s, 16H, bH_Pcⁱ + cH_Pc^o), −13.94 (br s, 8H, 1^ib), −12.05 (br s, 8H, α^o), -8.73 (br s, 8H, 1^ic), −6.83 (br s, 8H, α^o′), −6.68 (m, 16H, 2^ib,ic), −5.45 (m, 16H, 3^ib,ic), −4.31 (s, 8H, β^o), −4.13 (s, 24H, CH₃ⁱ), −2.86 (s, 8H, β^o′), −1.04 (m, 8H, γ^o), −0.21 (m, 16H, γ^o′ + δ^o), 0.05 (s, 24H, CH₃^o), 0.53 (br m, 32H, δ^o′ + 2^o + 3^o,o′), 1.03 (m, 2^o′), 2.19 (d, 25 Hz, 8H, 1^o), 6.41 (d, 25 Hz, 8H, 1^o′), 16.55 (br s, 8H, bH_Pc^o).

¹H-NMR (600 MHz, Toluene-d₈) δ −28.26 (br s, 8H, bH_Pcⁱ), −22.88 (br, 8H, cH_Pc^o), −16.45 (br s, 8H, 1^ib), −10.96 (br s, 8H, α^o), −9.68 (br s, 8H, 1^ic), −7.17 and −6.89 (2s, 2×8H, 2^ib and 2^ic), −6.61 and −6.46 (2s, 2×8H, 3^ib and 3^ic), −4.86 (s, 32H, CH₃ⁱ + α^o′), −3.56 (s, 8H, β^o), −2.13 (s, 8H, β^o′), −0.52 (s, 8H, γ^o), 0.25 (d, 8H, γ^o′), 0.41 (s, 24H, CH₃^o), 0.56 (s, 8H, δ^o), 0.83 (s, 8H, δ^o′), 0.92 and 0.98 (2s, 2×8H, 3^o,o′), 1.08 and 1.14 (2s, 2×8H, 2^o,o′), 2.75 (d, 29 Hz, 8H, 1^o), 6.65 (d, 29 Hz, 8H, 1^o′), 17.15 (br s, 8H, bH_Pc^o).

Trisphthalocyaninate [B₄]Dy[B₄]Dy[C₄]: A solution of phthalocyanines Dy[B₄]₂ (25.5 mg, 10.9 μmol) and H₂[C₄] (17.4 mg, 13.6 μmol) in a mixture of 2.7 mL 1,2,4-trichlorobenzene and 0.3 mL n-octanol was heated to reflux at 230 °C under a stream of argon, and solid Dy(acac)₃·nH₂O (18.8 mg, 40.9 μmol) was added. After 8 min, the reaction mixture was cooled to ambient temperature. The reaction mixture was transferred to a chromatographic column packed with alumina in a mixture of CHCl₃ and hexane (1:1 v/v). The target complex was isolated by elution with a CHCl₃-hexane mixture (4:1 v/v), followed by a mixture of CHCl₃ with 0→1% MeOH as a dark-blue solid (27 mg, yield 66%).

MALDI TOF: m/z calculated for C₁₉₂H₂₃₂Dy₂N₂₄O₃₆ 3776.6, found m/e—3777.2 [M⁺].

UV-vis (CHCl₃) λ_max (nm) (log ε): 645 (5.04), 547 (4.53), 355 (5.28), 293 (5.22).

UV-vis (Toluene) λ_max (nm) (log ε): 696 (4.65), 642 (5.40), 364 (5.27), 292 (5.13).

¹H NMR (600 MHz, CDCl₃) δ −71.54 (br s, 8H, bH_Pcⁱ), −31.56 (br s, 8H, 1^ib), −30.69 (br s, 8H, 1^ic), −22.39 (br s, 8H, cH_Pc^o), −19.74 (br s, 8H, bH_Pc^o), −18.64 and −18.29 (2s, 2×8H, 2^ib and 2^ic), −15.95 and −15.75 (2s, 2×8H, 3^ib and 3^ic), −15.43 (br s, 8H, 1^o), −14.05 (br s, 8H, α^o), −11.86 (s, 24H, CH₃ⁱ), −7.87 and −7.25 (2s, 2×8H, 3^o and 3^o′), −6.46 (s, 16H, 3^o,o′), −6.11 (s, 8H, β^o), −5.42 (br s, 8H, 1^o′), −4.89 (s, 24H, CH₃^o), −5.49 (br s, 8H, α^o′), −3.06 (s, 8H, β^o′), −2.34 (m, 8H, γ^o), −1.29 (s, 8H, δ^o), −0.65 (m, 8H, γ^o′), −0.12 (s, 8H, δ^o′).

¹H NMR (600 MHz, Toluene-d₈) δ −64.60 (br s, 8H, bH_Pcⁱ), −30.13 (br s, 8H, 1^ic), −29.97 (br s, 8H, 1^ib), −15.96 (s, 16H, 2^ib,ic), −15.07 (br s, 8H, cH_Pcⁱ), −14.67 (s, 16H, 3^ib,ic), −14.02 (br s, 8H, bH_Pcⁱ), −12.87 (br s, 8H, 1^o), −12.39 (br s, 8H, α^o), −10.84 (s, 24H, CH₃ⁱ), −5.75 (m, 16H, 2^o,o′), −4.88 (s, 8H, β^o), −4.79 (s, 16H, 3^o,o′), −3.55 (s, 24H, CH₃^o), −2.32 (br s, 8H, 1^o′), −2.05 (s, 8H, β^o′), −1.99 (br s, 8H, α^o′), −1.37 (m, 8H, γ^o), 0.22 (m, 8H, γ^o′), 0.92 (s, 16H, δ^o,o′).

Trisphthalocyaninate [B₄]Tb[B₄]Dy[C₄]: A solution of phthalocyanines Tb[B₄]₂ (27.0 mg, 11.5 μmol) and H₂[C₄] (18.0 mg, 14.1 μmol) in a mixture of 2.7 mL 1,2,4-trichlorobenzene and 0.3 mL n-octanol was heated to reflux at 230 °C under a stream of argon, and solid Dy(acac)₃·nH₂O (21.0 mg, 45.7 μmol) was added. After 8 min, the reaction mixture was cooled to ambient temperature. The reaction mixture was transferred to a chromatographic column packed with alumina in a mixture of CHCl₃ and hexane (1:1 v/v). The target complex was isolated by elution with a CHCl₃-hexane mixture (4:1 v/v), followed by a mixture of CHCl₃ with 0→1% MeOH as a dark-blue solid (20 mg, yield 45%).

MALDI TOF: m/z calculated for C₁₉₂H₂₃₂DyN₂₄O₃₆Tb 3773.6, found 3774.3 [M⁺].

UV-vis (CHCl₃) λ_max (nm) (log ε): 646 (5.07), 545 (4.52), 354 (5.24), 293 (5.19).

UV-vis (Toluene) λ_max (nm) (log ε): 695 (4.68), 643 (5.47), 363 (5.26), 293 (5.13).

¹H NMR (600 MHz, CDCl₃) δ −118.51 (br s, 8H, bH_Pcⁱ), −60.39 (br s, 8H, bH_Pc^o), −57.29 (br s, 8H, 1^ib), −49.37 (br s, 8H, 1^ic), −35.40 (br s, 8H, 1^o), −32.17 and −31.58 (2s, 2×8H, 2^ib and 2^ic), −27.90 and −27.42 (2s, 2×8H, 3^ib and 3^ic), −20.68 (s, 24H, CH₃ⁱ), −19.23 (br s, 8H, 1^o′), −17.55 and −16.65 (2s, 2×8H, 2^o and 2^o′), −16.47 (s, 8H, α^o), −14.60 (s, 16H, 3^o,o′), −13.11 (br s, 8H, cH_Pc^o), −10.70 (s, 24H, CH₃^o), −8.11 (s, 8H, β^o), −4.02 (d, 8H, γ^o), −3.27 (s, 8H, δ^o), −2.77 (s, 8H, β^o′), −1.99 (br s, 8H, α^o′), −1.15 (d, 8H, γ^o′), −0.53 (s, 8H, δ^o′).

¹H NMR (600 MHz, Toluene-d₈) δ −97.98 (br s, 8H, bH_Pcⁱ), −48.46 (br s, 8H, 1^ib), −43.16 (br s, 8H, bH_Pc^o), −42.45 (br s, 8H, 1^ic), −27.43 (br s, 8H, 1^o), −24.44 and −24.08 (2s, 2×8H, 2^ib and 2^ic), −22.21 and −21.96 (2s, 2×8H, 3^ib and 3^ic), −16.24 (s, 24H, CH₃ⁱ), −13.35 (br s, 8H, α^o), −12.18 and −12.04 (2s, 2×8H, 2^o and 2^o′), −10.98 (br s, 8H, 1^o′), −10.19 (s, 16H, 3^o,o′), −7.28 (s, 24H, CH₃^o), −5.94 (s, 8H, β^o), −2.08 (m, 8H, γ^o), −1.86 (s, 8H, β^o′), 0.00 (s, 8H, δ^o), 0.20 (m, 8H, γ^o′), 0.93 (br s, 8H, α^o′), 1.06 (s, 8H, δ^o′).

Trisphthalocyaninate [B₄]Dy[B₄]Tb[C₄]: A solution of phthalocyanines Dy[B₄]₂ (29.0 mg, 12.4 μmol) and H₂[C₄] (19.7 mg, 15.5 μmol) in a mixture of 2.7 mL 1,2,4-trichlorobenzene and 0.3 mL n-octanol was heated to reflux at 230 °C under a stream of argon, and solid terbium (III) acetylacetonate hydrate Tb(acac)₃·nH₂O (22.0 mg, 46.5 μmol) was added. After 10 min, the reaction mixture was cooled to ambient temperature. The reaction mixture was transferred to a chromatographic column packed with alumina in a mixture of CHCl₃ and hexane (1:1 v/v). The target complex was isolated by elution with a CHCl₃-hexane mixture (4:1 v/v), followed by a mixture of CHCl₃ with 0→1% MeOH as a dark-blue solid (29 mg, yield 62%).

MALDI TOF: m/z calculated for C₁₉₂H₂₃₂DyN₂₄O₃₆Tb 3773.6, found 3774.3 [M⁺].

UV-vis (CHCl₃) λ_max (nm) (log ε): 645 (5.00), 550 (4.46), 354 (5.18), 293 (5.14).

UV-vis (Toluene) λ_max (nm) (log ε): 695 (4.61), 643 (5.39), 363 (5.20), 294 (5.07).

¹H NMR (600 MHz, CDCl₃) δ −107.81 (br s, 8H, bH_Pcⁱ), −57.71 (br s, 8H, cH_Pcⁱ), −48.90 (br s, 8H, 1^ic), −44.48 (br s, 8H, 1^ib), −30.28 (br s, 8H, α^o), −27.57 and −27.15 (2s, 2×8H, 2^ib and 2^ic), −23.25 (m, 16H, 3^ib,ic), −17.83 (br s, 8H, 1^o), −17.32 (s, 24H, CH₃ⁱ), −15.56 (br s, 8H, α^o′), −14.33 (s, 8H, β^o), −11.63 (br s, 8H, bH_Pc^o), −9.78 (s, 8H, β^o′), −9.38 and −8.51 (2s, 2×8H, 2^o and 2^o′), −7.73 (s, 16H, 3^o,o′), −7.75 (m, 8H, γ^o), −6.05 (s, 24H, CH₃^o), −5.33 (s, 8H, δ^o), −4.60 (m, 8H, γ^o′), −3.54 (br s, 8H, 1^o′), −3.15 (s, 8H, δ^o′).

¹H NMR (600 MHz, Toluene-d₈) δ −100.40 (br s, 8H, bH_Pcⁱ), −50.04 (br s, 8H, 1^ic), −44.57 (br s, 8H, cH_Pc^o), −43.19 (br s, 8H, 1^ib), −26.68 (br s, 8H, α^o), −24.04 and −24.76 (2s, 2×8H, 2^ib and 2^ic), −22.70 (s, 16H, 3^ib,ic), −16.69 (s, 24H, CH₃ⁱ), −14.12 (br s, 8H, 1^o), −11.80 (s, 8H, β^o), −10.39 (br s, 8H, α^o′), −7.45 (s, 8H, β^o′), −6.45 and −6.37 (2s, 2×8H, 2^o and 2^o′), −5.58 (m, 8H, γ^o), −5.40 (s, 16H, 3^o,o′), −4.17 (s, 32H, bH_Pc^o + CH₃^o), −3.07 (m, 8H, γ^o′), −2.93 (s, 8H, δ^o), −2.03 (s, 8H, δ^o′), 0.46 (br s, 8H, 1^o′).