Macrocyclic Ligands with an Unprecedented Size-Selectivity Pattern for the Lanthanide Ions

Abstract

Lanthanides (Ln³⁺) are critical materials used for many important applications, often in the form of coordination compounds. Tuning the thermodynamic stability of these compounds is a general concern, which is not readily achieved due to the similar coordination chemistry of lanthanides. Herein, we report two 18-membered macrocyclic ligands called macrodipa and macrotripa that show for the first time a dual selectivity towards both the light, large Ln³⁺ ions and the heavy, small Ln³⁺ ions, as determined by potentiometric titrations. The lanthanide complexes of these ligands were investigated by NMR spectroscopy and X-ray crystallography, revealing the occurrence of a significant conformational toggle between a 10-coordinate Conformation A and an 8-coordinate Conformation B that accommodates Ln³⁺ ions of different sizes. The origin of this selectivity pattern was further supported by density functional theory (DFT) calculations, which show the complementary effects of ligand strain energy and metal-ligand binding energy that contribute to this conformational switch. This work demonstrates how novel ligand design strategies can be applied to tune the selectivity pattern for the Ln³⁺ ions.

Graphical Abstract

Introduction

The unique physical properties of the lanthanides are critical for use in magnets, superconductors, catalysts, luminescent phosphors, and medicinal agents.^1–6 Despite their diverse magnetic and electronic properties, the lanthanides possess similar chemical properties dictated by their strong preference for the +3 oxidation state and tendency to engage in ionic bonding.^7,8 The major distinguishing feature among them is their ionic radius, which decreases across the series, a phenomenon known as the lanthanide contraction.⁹

For many applications of the lanthanides (Ln³⁺), a ligand is required to chelate these ions to control and modify their chemical properties. The availability of a range of chelators with different affinities and selectivity patterns for the Ln³⁺ ions, measured by the stability constant K_LnL, is valuable for the implementation of these ions.¹⁰ Generally, most ligands possess a higher affinity for the heavier, smaller Ln³⁺ ions because the increased charge density on these smaller ions enhances metal-ligand electrostatic interactions.¹¹ Recent ligand design efforts, however, have led to systems with other Ln³⁺-selectivity patterns. To date, three types of selectivity patterns have been identified (Figure 1). As noted above, the most common trend shows a systematic increase in K_LnL across the lanthanide series (Type I). This Type I behavior is observed for many well-known ligands including EDTA,¹² (Figure 1, Chart 1), 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA),¹³ and diethylenetriaminepentaacetic acid (DTPA).^14,15

Figure 1.

Stability constants of Ln³⁺ complexes formed with EDTA, OBETA, and macropa plotted versus ionic radii.

Chart 1.

Structures of Ligands Discussed in this Work.

Another less frequently observed trend, Type II, has the stability reach a maximum prior to dropping again. The ligands OBETA^16,17 (Figure 1, Chart 1) and 1,4,7,10-tetrakis(carbamoylmethyl)-1,4,7,10-tetraazacyclododecane (TCMC)¹⁸ follow this pattern. The rarely observed behavior, Type III, occurs with ligands having reverse-size selectivity, an unusual thermodynamic preference for large over small Ln³⁺ ions. This pattern has been observed in ligands containing the large diaza-18-crown-6 macrocycles, such as macropa¹⁹ (Figure 1, Chart 1) and 1,10-diaza-4,7,13,16-tetraoxacyclooctadecane-N,N′-diacetic acid (dacda).²⁰ A common feature of these three selectivity patterns is that they all have the maximum affinity for only one Ln³⁺ ion.

Inspired by macropa and our ongoing research efforts on chelating agents for large metal ions of biomedical and industrial relevance,^21–24 we sought to explore other ligands containing 18-membered macrocycles. Specifically, we investigated an isomer of macropa, called macrodipa, and a triaza-18-crown-6 ligand^25,26 bearing three pendent picolinate donors, macrotripa (Chart 1). Notably, these macrocycles are cyclic analogues of the previously reported ligand OxyMepa²⁷ (Chart 1) that provides a similar albeit truncated set of donor atoms. Over the course of our studies on their Ln³⁺ coordination chemistry, we found that macrodipa and macrotripa undergo dramatic conformational changes upon binding large versus small ions. These conformational changes manifest in a previously unreported Type IV selectivity pattern with one minimum and two maxima of stability across the Ln³⁺ series. This work demonstrates how novel ligand design strategies can be employed to differentiate these ions in new ways.

Results and Discussion

The syntheses of macrodipa and macrotripa (Schemes S1–S2) involve the assembly of the tosyl-protected 18-crown-6 macrocycles, deprotection of the tosyl groups, and subsequent alkylation of the picolinate donor arms. They were characterized by NMR spectroscopy, mass spectrometry, and HPLC (Figures S1–10).

Potentiometric titrations were performed to determine their protonation constants (K_i, Table S1). To probe the thermodynamic affinities of these ligands for Ln³⁺ ions, we conducted potentiometric titrations to obtain their stability constants (K_LnL and K_LnHL, Table 1). These protonation and stability constants are defined in Eqs 1–3, with the concentrations of all species at chemical equilibrium.

Table 1.

Stability Constants of the Lanthanide Complexes Formed with macrodipa, macrotripa, OxyMepa, macropa, EDTA, and OBETA.

Ln3+	macrodipaa	macrotripaa		OxyMepab	macropac		EDTAd	OBETAe
	log KLnL	Log KLnL	log KLnHL	log KLnL	Log KLnL	log KLnHL	log KLnL	log KLnL
La3+	12.19(2)	12.57(1)	3.67(2)	9.93	14.99	2.28	15.46	16.89
Ce3+	12.50(4)	12.82(2)	3.66(4)	10.74	15.11	2.07	15.94	17.34
Pr3+	12.41(2)	12.65(1)	3.65(2)	11.25	14.70	2.96	16.36
Nd3+	12.25(3)	12.25(2)	3.88(1)	11.49	14.36	2.08	16.56	18.39
Sm3+	11.52(2)	11.41(1)	3.99(1)	12.13	13.80	2.70	17.10	19.02
Eu3+	10.93(1)	10.86(1)	4.24(2)	12.15	13.01	1.97	17.32	19.13
Gd3+	10.23(2)	10.19(2)	4.51(1)	12.02	13.02	2.48	17.35	19.37
Tb3+	9.68 (1)	10.19(2)	4.82(1)	12.17	11.79	2.91	17.92
Dy3+	9.36(2)	10.49(3)	4.86(2)	12.18	11.72	2.42	18.28	18.87
Ho3+	9.36(1)	10.70(3)	4.90(1)	12.10	10.59		18.60	18.93
Er3+	9.71(4)	11.05(3)	4.92(4)	12.00	10.10		18.83	18.46
Tm3+	10.13(1)	11.43(5)	4.89(2)	12.05	9.59		19.30
Yb3+	10.48(1)	11.80(3)	4.88(1)	12.14	8.89		19.48	18.31
Lu3+	10.64(4)	11.90(1)	4.93(2)	12.21	8.25		19.80	17.93

^a0.1 M KCl, this work. The values in the parentheses are one standard deviation of the last significant figure.

^b0.1 M KCl, ref 27.

^c0.1 M KCl, ref 19.

^d0.1 M, ref 12.

^e0.1 M KCl, ref 16,17.

$$K_i=\left[{\text{H}}_{\text{i}}\text{L}\right]/\left[{\text{H}}^{+}\right]\left[{\text{H}}_{\text{i}-1}\text{L}\right]$$ (1)

$$K_{\text{LnL}}=\left[\text{LnL}\right]/\left[{\text{Ln}}^{3+}\right]\left[\text{L}\right]$$ (2)

$$K_{\text{LnHL}}=\left[\text{LnHL}\right]/\left[{\text{H}}^{+}\right]\left[\text{LnL}\right]$$ (3)

Figure 2 shows a plot of log K_LnL versus the Ln³⁺ ionic radius²⁸ for both macrodipa and macrotripa, revealing them to be the first known ligands that exhibit Type IV selectivity. In comparing the log K_LnL values between Ln³⁺-macrodipa and Ln³⁺-macrotripa systems, they are similar for early lanthanides, La³⁺–Gd³⁺. Ln³⁺-macrodipa reaches a minimum for Dy³⁺ and Ho³⁺, whereas for macrotripa the minimum occurs earlier in the series, between Gd³⁺ and Tb³⁺. For late lanthanides, the macrotripa complexes are significantly more stable than those of macrodipa.

Figure 2.

Stability constants of Ln³⁺ complexes formed with macrodipa and macrotripa plotted versus ionic radii.

The log K_LnHL values, which represent the protonation of LnL complex, are also noteworthy. These values were not found for Ln³⁺-macrodipa complexes but were observed for Ln³⁺-macrotripa complexes. Comparing the chemical structures of macrodipa and macrotripa and their complexes (vide infra), it can be reasonably inferred that this protonation event occurs on the third picolinate arm in macrotripa. The K_LnHL values of the macrotripa complexes remain steady from La³⁺ to Pr³⁺, but then increase abruptly from Nd³⁺ to Tb³⁺, before leveling off from Dy³⁺ to Lu³⁺. The sudden change in log K_LnHL implies that a conformational change may be present for Ln³⁺-macrotripa complexs when crossing the Ln³⁺ series.

To gain insight on the Type IV selectivity of macrodipa and macrotripa, we analyzed their complexes of the largest and smallest lanthanides, La³⁺ and Lu³⁺, by NMR spectroscopy. These diamagnetic La³⁺ and Lu³⁺ complexes were characterized by ¹H, ¹³C{¹H}, and 2D (HSQC, HMBC, COSY, ROESY) NMR spectroscopy in D₂O at pD = 7 (Figures 3, S41–S80). The ¹H NMR spectra of all four complexes indicate that a single species is present in solution. However, significant differences are apparent in comparing the La³⁺ and Lu³⁺ complexes. For example, the La³⁺-macrodipa complex is 2-fold symmetric, indicated by half the number of ¹H and ¹³C resonances relative to the asymmetric Lu³⁺-macrodipa complex.

Figure 3.

¹H NMR spectra of macrodipa and macrotripa complexes formed with La³⁺ and Lu³⁺ (600 MHz, D₂O, pD = 7, 25 °C). Selected peaks are labeled using the numbering scheme in Chart 1.

Additionally, the hydrogen resonances from the methylene groups linking the crown and picolinate donors (H-13, H-20 for macrodipa, and H-13, H-20, H-27 for macrotripa; Chart 1), which are informative due to their proximities to the picolinate donors, are significantly different between the La³⁺ and Lu³⁺ complexes. For example, the peaks for the diastereotopic H-27′ and H-27′′ in La³⁺-macrotripa case are well-separated, whereas for Lu³⁺-macrotripa they have near-identical chemical shifts. Collectively, these NMR data suggest that there is a significant conformational difference between La³⁺ and Lu³⁺ complexes of these ligands.

To further explore these different conformations, we characterized these complexes by X-ray crystallography. Crystal structures of [La(macrodipa)]⁺, [Lu(macrodipa)(OH₂)]⁺, and [La(macrotripa)]⁺ are shown in Figure 4. A weakly diffracting and partially twinned crystal of [Lu(macrotripa)(OH₂)] was also obtained, but the connectivity information was reliably ascertained (Figure S89). Confirming the NMR spectroscopic data, the La³⁺ complexes attain a significantly different conformation than the Lu³⁺ complexes. In both La³⁺ structures, this ion is encapsulated into the 18-membered macrocyclic core, interacting with all its six donor atoms. Moreover, the pendent picolinate groups bind to La³⁺ from two opposite faces of the macrocycle, resulting in 10-coordinate complexes. The third picolinate donor of macrotripa does not participate in coordination. Consistent with its NMR spectra, [La(macrodipa)]⁺ attains a slightly distorted C₂ symmetry. By contrast, both Lu³⁺ structures show significantly different coordination environments. Specifically, only two tertiary nitrogens and one ethereal oxygen from the macrocycle act as donors. The coordination sphere is completed with the four donor atoms from two picolinate groups and an inner-sphere water molecule, yielding 8-coordinate Lu³⁺ centers. In [Lu(macrotripa)(OH₂)], the third unbound picolinate donor is positioned to interact with the bound water molecule through hydrogen bonding. Except for the macrocycle, these Lu³⁺ structures are highly comparable to those found for the acyclic ligand OxyMepa,²⁷ which displays Type I selectivity, indicating that the complete 18-crown-6 macrocycles of macrodipa and macrotripa are critical for their unique Ln³⁺-selectivity profiles.

Figure 4.

Crystal structures of (a) [La(macrodipa)]⁺, (b) [Lu(macrodipa)(OH₂)]⁺, and (c) [La(macrotripa)]⁺ complexes. Thermal ellipsoids are drawn at the 50% probability level. Solvent, counterions, and non-acidic hydrogen atoms are omitted for clarity. Only one of the two [La(macrodipa)]⁺ units in the asymmetric cell is shown.

As a further validation on the proposed intramolecular hydrogen bond in [Lu(macrotripa)(OH₂)], we optimized its structure (Figure S98) using validated DFT methods (vide infra), and carried out a topological analysis of the electron density using the quantum theory of atoms in molecules (QTAIM).²⁹ Specifically, we found all bond critical points (BCP) with the Multiwfn³⁰ program (Figure S99). A BCP was located between a hydrogen atom of the coordinated water molecule and the oxygen atom of the pendent picolinate group. At this BCP, the magnitude of its local electron density (ρ) is 0.079 a.u. and the Laplacian of the electron density (∇²ρ) is 0.16 a.u. The positive value for ∇²ρ reflects a closed-shell hydrogen bond, and the magnitude of ρ suggests that this interaction is strong, comparable to that found between OH^– and H₂O.^31–33 This analysis supports the proposed intramolecular hydrogen-bonding interaction present in [Lu(macrotripa)(OH₂)].

Based on the NMR and X-ray crystallographic data, it is clear that macrodipa and macrotripa attain distinct conformations depending on whether they bind large or small Ln³⁺ ions (Scheme 1). Large ions, like La³⁺, attain Conformation A, in which the ion is fully encapsulated by the macrocycle, whereas small ions, like Lu³⁺, sit in Conformation B, held by only part of the macrocycle. The ability of these ligands to drastically alter their conformations to match the sizes of metal ions accounts for the Type IV selectivity pattern. The structures may also explain the difference in thermodynamic stability of macrodipa and macrotripa for the late, but not early Ln³⁺ (Figure 2). Both ligands give rise to identical coordination spheres for the large early lanthanides, like La³⁺, and therefore exhibit only minor differences in their thermodynamic stabilities. However, for the small late lanthanides, like Lu³⁺, the inner coordination spheres are nearly identical between macrodipa and macrotripa, but the outer sphere differs due to the hydrogen-bonding interaction with the coordinated water molecule. Thus, the differences in thermodynamic stability between the macrodipa and macrotripa complexes of the late lanthanides is most likely a consequence of the hydrogen bonding of the pendent picolinate donor arm. This result highlights how modifying the outer coordination sphere of lanthanide complexes fine-tunes their thermodynamic properties.

Scheme 1.

Depiction of the Conformational Toggle Present in Ln³⁺-macrodipa and Ln³⁺-macrotripa complex systems.

As a further test of this conformational toggle, we investigated the complexes of Y³⁺, a diamagnetic Ln³⁺ analogue with a comparable ionic radius to that of Ho³⁺,^7,28 by NMR spectroscopy. The ¹H and ¹³C{¹H} NMR spectra of Y³⁺-macrodipa and Y³⁺-macrotripa were acquired in D₂O at pD = 7 (Figure S81–S88). Both Conformations A and B are detected for Y³⁺-macrodipa, in a molar ratio of 1:15. For the Y³⁺-macrotripa complex, only Conformation B is observed. The 90-pm ionic radius of Y³⁺ places its macrodipa complex near the local minimum of log K_LnL, but its macrotripa complex rather far from the minimum (Figure 2). Thus, these NMR data show that the conformational switch occurs for complexes of ions with their radii near the minimum of stability; larger and smaller ions show preferences for Conformations A and B, respectively.

DFT has been extensively used to investigate the properties of Ln³⁺ coordination compounds.^34–36 In this study, we took advantage of this powerful tool to help understand the origin of the Type IV selectivity pattern of these ligands. We focused exclusively on the Ln³⁺-macrodipa system. These complexes lack the third non-coordinated picolinate arm of the Ln³⁺-macroptripa, and therefore provide a straightforward system to model the inner coordination spheres of these complexes. DFT calculations were executed using Gaussian 09³⁷ with the ωB97XD functional.^38,39 This functional, which is long-range corrected and includes dispersion corrections, has been shown to give accurate geometries of Ln³⁺ complexes.⁴⁰ Because of the importance of relativistic effects in Ln³⁺ ions,⁴¹ we used the large-core relativistic effective core potential (LCRECP) by Dolg⁴² to account for these effects in a computationally efficient manner. For light atoms, the 6–31G(d,p) basis set^43,44 was applied. The SMD solvation model^45,46 was implemented to take the solvent effects into consideration. The ΔG° for the conformational equilibrium

$$\begin{gathered} {\left[\text{Ln}\left(\text{macrodipa}\right)\right]}^{+}\left(\text{Conformation}\hspace{0.17em}\text{A},aq\right)+{\text{H}}_2\text{O}\left(l\right) \\ \rightleftharpoons {\left[\text{Ln}\left(\text{macrodipa}\right)\left({\text{OH}}_2\right)\right]}^{+}\left(\text{Conformation}\hspace{0.17em}\text{B},aq\right) \end{gathered}$$ (4)

was calculated for Ln³⁺-macrodipa complexes. The ΔG° (Figure 5) is positive for light Ln³⁺ and negative for heavy Ln³⁺. This observation is consistent with the experimental results with La³⁺-macrodipa and Lu³⁺-macrodipa complexes attaining Conformations A and B, respectively. Additionally, ΔG° changes its sign between Gd³⁺ and Tb³⁺, indicating the switch of favored conformation. This crossover suggests that the Type IV behavior of macrodipa is a consequence of the significant conformational changes that occur when binding Ln³⁺ ions of different sizes.

Figure 5.

DFT-computed free energies for Ln³⁺-macrodipa complexes.

Furthermore, ΔG° for this conformation change can be broken up into three contributors. Specifically, it can be expressed as the sum of the relative ligand strain energies (ΔΔG_S°), relative metal-ligand binding energies (ΔΔG_B°), and relative solvation energies (ΔΔG_solv°) between Conformations A and B, as described in the Supporting Information. As shown in Figure 5, ΔΔG_solv° is positive for all Ln³⁺ complexes, revealing that Conformer A and the non-coordinated water ligand are better solvated in aqueous solution than Conformer B. Likewise, ΔΔG_B° is positive for all Ln³⁺, indicating that Conformation A is better suited to neutralize the electrostatic charges of these ions than Conformation B. This observation can be rationalized by the fact that Conformation A interacts with the Ln³⁺ with two more donor atoms. However, ΔΔG_B° decreases as the Ln³⁺ gets smaller, suggesting that Conformation A is less effective at binding the smaller ions. By contrast, ΔΔG_S° is negative across the entire series, showing that Conformation B requires less ligand strain than Conformation A. Among the three values, ΔΔG_S° shows the most significant changes as a function of the Ln³⁺ ionic radius, becoming more negative for smaller ions. Importantly, the strain energy is the only exothermic term for the switch from Conformation A to B, and thus it is the driving factor in the conformational switch of macrodipa. This result suggests that modifications of this ligand scaffold to alter the strain energy term could lead to a significant shift in the Ln³⁺-stability pattern for this ligand class.

Conclusion

In summary, this work describes the first ligands that display an unprecedented Type IV selectivity pattern with one minimum and two maxima of K_LnL across the Ln³⁺ series. This novel selectivity pattern may have key applications in Ln³⁺ separations and nuclear medicine, where it is often desirable to stably chelate a range of metal ions with disparate ionic radii. Our structural data by NMR spectroscopy and X-ray crystallography show that both macrodipa and macrotripa undergo a significant conformational shift in moving from large to small Ln³⁺ ion, which allows for high thermodynamic stability for both early and late Ln³⁺. In comparing macrodipa and macrotripa, we have shown that one can modify these ligands to tune the position of the minimum and the overall magnitude of thermodynamic stability. Our DFT calculations further support the presence of this conformational toggle, and show that the ligand strain and metal-ligand binding energies are complementary factors contributing to the conformational switch. Between them, we believe that the ligand strain energy is more easily modified, such as including groups in the macrocyclic backbone that prevent conformational flexibility, thus presenting a path for tuning these ligands for different applications. This work demonstrates how the incorporation of conformational flexibility in ligands can be used to satisfy the coordination environments of different metal ions, a principle of great value for chelator design efforts.