A Walk across the Lanthanide Series – Trend in Affinity for Phosphate and Stability of Lanthanide Receptors from La(III) to Lu(III)

Abstract

The trend in affinity of two 1,2-hydroxypyridinonate lanthanide(III) receptors—Ln^III-2,2-Li-HOPO and Ln^III-3,3-Gly-HOPO (Ln^III = La^III, Pr^III, Nd^III, Sm^III, Eu^III, Gd^III, Tb^III, Dy^III, Ho^III, Er^III, Tm^III, Yb^III, and Lu^III)—for phosphate across the series was investigated by luminescence spectroscopy via competition against the central europium(III) analog. Regardless of the ligand, the rare earth receptors display a steep and continuous increase in affinity for their phosphate guest across the series, with the later lanthanides displaying the highest affinity for the oxyanion. This trend mirrors that of the stability of the lanthanide receptors, which also increases significantly and continuously from La^III to Lu^III. For these two ligands, the ionic radius of a rare earth, a parameter directly linked to its Lewis acidity, correlates strongly with its affinity for anions, regardless of whether that anion is the one coordinating it (in this case the 1,2-hydroxypyridinonate ligand) or the guest targeted by the lanthanide receptor (in this case phosphate). These observations are indicative of a lack of steric hindrance for coordination of phosphate. Advantageously, increased efficacy of the lanthanide receptor comes with increased stability. The remarkably high stability of Lu^III-2,2-Li-HOPO, combined with its high affinity for phosphate, makes it a particularly promising candidate for translational application to medical or environmental sequestration of phosphate since the higher stability will further reduce the risk of the rare earth leaching during anion separation. The unusually large difference in stability between lanthanide complexes—the Lu^III complex of 2,2-Li-HOPO is at least seven orders of magnitude more stable than the La^III one—bodes well for potential applications in rare earth separation.

SYNOPSIS

The affinity for phosphate and the stability of tripodal lanthanide(III) complexes increase steadily and steeply from lanthanum(III) to lutetium(III) across the lanthanide series as the ionic radii decreases and Lewis acidity of the metal increases. These trends indicate that steric hindrance is not a significant contributing factor to either ligand or phosphate coordination.

INTRODUCTION

The characteristic magnetic and spectroscopic properties of lanthanides that arise from their unfilled 4f electrons orbitals shielded by the filled 5s and 5p ones are key to their numerous applications, such as in MRI contrast agents,¹ bioresponsive probes,^2-4 and single-molecule magnets (SMMs).⁵ In particular, the rich supramolecular chemistry of lanthanide complexes has inspired their use as receptors for the detection and sequestration of hard anions such as phosphate,^6-9 an ion of both medical and environmental significance. Advantageously, the similar chemistry that rare earth ions and complexes display throughout the series enables the facile interchange in a receptor of one lanthanide for another as needed to exploit the unique spectroscopic and/or magnetic properties of each metal² and achieve the desired luminescence and/or magnetic output. Nonetheless, their chemistries are not identical and the extent of the response will be affected by the metal. The lanthanide contraction, characterized by a decrease in ionic radius from 1.160 Å for La^III to 0.977 Å for Lu^III affects the metals’ Lewis acidities.¹⁰ Since the affinity for anions of lanthanide complexes with sterically unhindered open coordination sites are governed by the acidity and basicity of the metal host and anionic guest, respectively,⁷ differences in the affinity for anionic ligands and guests should be anticipated between rare earths.

The effects of the lanthanide contraction on the structure of rare earth complexes,^11,12 as well as on their thermodynamic stability,¹³ kinetic inertness,¹⁴ and water exchange rates¹⁵ have been studied extensively. Investigation into the trend in affinity for anionic guests by lanthanide receptors across the series, on the other end, has barely been explored. The few studies reported so far evaluated only some of the rare earths, often not enough to make general trends. The most extensive one by Grenthe and coworkers determined, based on eight lanthanides complexes (Pr^III, Nd^III, Sm^III, Gd^III, Dy^III, Er^III, Yb^III, and Lu^III), that the affinity of picolinate for the Ln^III-EDTA was higher for the later lanthanides.¹⁶ Despite the fact that anion coordination was not affected by any steric congestion at the open site, the trend observed was not linear: Sm^III also had high affinity for the anion. The affinity for dipicolinate of lanthanide complexes of another hexadentate polyaminocarboxylate ligand, Ln^III-DO2A, also does not follow lanthanide ionic radii. Instead, the affinity of the tridentate anion for Sm^III-, Eu^III-, Tb^III-, and Dy^III-DO2A correlate strongly to the Ln^III → Ln^IV ionization energy of these four rare earths.¹⁷ Gray and coworkers postulated that the DO2A ligand polarized to a greater extent the lanthanide ions with lower ionization energies, thereby leading to higher affinities for the anionic guest. This observation is not universal. Pecoraro, for instance, observed that the affinity of lanthanide metallacrowns (Ln^III = La^III, Nd^III, Gd^III, and Dy^III) for benzoate has a linear relationship with the ionic radii, and hence the Lewis acidity, of the rare earth ion.¹⁸ The basic guest having higher affinity for the later, more acidic, lanthanides. In view of these apparent contradictions, a more complete studies across the lanthanide series in the trend in affinity of lanthanide complexes for anionic guests is needed to fully understand, predict and control the different parameters governing this class of receptors.

Our group has recently developed receptors with exceptional selectivity and affinity for inorganic phosphate in water.⁶ Prior studies demonstrated the effect of ligand basicity,⁷ overall charge,⁸ and geometry⁹ of tripodal tris-bidentate lanthanide complexes on their affinity and selectivity for phosphate. The translation of these receptors to applications in wastewater remediation¹⁹ and medical treatment of hyperphosphatemia²⁰ requires us to consider, for reasons of cost, earlier and less expensive lanthanide ions. In the absence of studies investigating the trend across the lanthanide series in binding affinity of small oxyanions for rare earth receptors, we investigated the relationship between the Lewis acidity of the lanthanide ion and the affinity of its complex for inorganic phosphate (Pi). This trend was compared to that of the stability of the complexes across the series. This study was conducted with two different lanthanide receptors that we previously demonstrated had high affinity and selectivity for inorganic phosphate in water: Ln^III-2,2-Li-HOPO and Ln^III-3,3-Gly-HOPO (Figure 1).⁹ Although both employ tripodal tris-bidentate ligands of 1,2-hydroxypyridinonate, the europium(III) complexes of each ligand have different coordination number and form ternary complexes with phosphate of different stoichiometry. Indeed, whereas Eu^III-3,3-Gly-HOPO is nine-coordinate and binds phosphate in 1:3 ratio, Eu^III-2,2-Li-HOPO is eight coordinate and binds the oxyanion in a 1:2 ratio. While the number of open coordination sites of lanthanide complexes, q, was demonstrated not to correlate with their affinity for anions,⁹ comparing the solution thermodynamics of lanthanide complexes of these two ligands enables us to evaluate the possible effect of steric hindrance and coordination number on the trend in affinity for phosphate across the series.

Figure 1.

Chemical structures of lanthanide-based phosphate receptors.

RESULTS AND DISCUSSION

Synthesis

The two ligands 2,2-Li-HOPO (1) and 3,3-Gly-HOPO (2) were synthesized as previously reported.^8,9 Their La^III, Pr^III, Nd^III, Sm^III, Eu^III, Gd^III, Tb^III, Dy^III, Ho^III, Er^III, Tm^III, Yb^III and Lu^III complexes were synthesized following standard protocols using the corresponding lanthanide chloride salts in the presence of base, either pyridine or NaOH, as needed to deprotonate the ligand. Every complex was characterized by high resolution mass spectroscopy and their purity assessed by high performance liquid chromatography (see Supporting Information for details). The Pm^III complexes were not synthesized due to their radioactive nature. The Ce^III complexes were also not incorporated in this study due to their partial decomposition under the experimental conditions of aqueous analytical studies.

Periodic trend in affinity for phosphate

One of the challenges in studying anion binding across the lanthanide series is that a same technique cannot readily be applied to all rare earths. Indeed, whereas both Eu^III complexes are bright and display a substantial luminescence response upon phosphate coordination, 1,2-HOPO can sensitize only three other lanthanides: Sm^III, Tb^III, and Dy^III.²¹ Anion coordination to Gd^III in water are best studied by relaxivity,^6,7,22 a method that, given its limited sensitivity, requires substantially different experimental conditions. NMR, a technique commonly used to study host:guest binding, is unfortunately not amenable to this study. Lanthanide complexes of these two linear ligands are highly fluxional such that interconversion between numerous isomers occurs faster than the NMR time scale. Combined with a rapid equilibrium between bound and unbound phosphate and the paramagnetic nature of most rare earths, this technique did not enable us to accurately monitor phosphate coordination across the series. The trend in affinity for anions across the series was thus determined by a new protocol: direct competition between a lanthanide complex of a ligand and its europium(III) analog. This enables rapid evaluation of the trend across all stable and non-radioactive lanthanide ions by luminescence spectroscopy under identical conditions.

In this experiment each Eu^IIIL·Pi_n complex is competed with increasing concentration of an analogous lanthanide receptor; the resulting equilibrium is monitored via the Eu^III-centered luminescence emission (Figure 2). The europium(III) complexes have either 2 (for Eu^III-2,2-Li-HOPO) or 3 (for Eu^III-3,3-Gly-HOPO) open coordination sites that, in the absence of phosphate, are filled by water molecules. Those coordinated water molecules efficiently quench the Eu^III-centered luminescence via non-radiative energy transfer (ET) of the excited ⁵D₀ triplet state of the lanthanide to the fourth overtone of the O-H vibration.² As a result, in the off state, the Eu^III complexes are poorly emissive. Coordination of phosphate to the Eu^III center occurs via displacement of the quenching inner-sphere water molecule, according to the equilibrium in equation (1), resulting in a significant 10-fold (for Eu^III-2,2-Li-HOPO) or 20-fold (for Eu^III-3,3-Gly-HOPO) increase in time-gated luminescence intensity.⁹

$${\text{Ln}}^{\text{III}}\mathrm{L}\cdot n{\mathrm{H}}_2\mathrm{O}+n\text{Pi}\rightleftharpoons {\text{Ln}}^{\text{III}}\mathrm{L}\cdot {\text{Pi}}_n+n{\mathrm{H}}_2\mathrm{O}$$ (1)

Figure 2.

Monitoring the trend in affinity for phosphate of lanthanide receptors by competition with the Eu^III analog. Efficient sensitization of Eu^III via its hydroxypyridinonate antenna results in a brightly luminescent Eu^IIIL·Pi_n ternary complex (left). Exchange of the phosphate anions from the Eu^III complex to a competing receptor, composed of the same ligand but a different lanthanide ion, results in the coordination of water molecules to the Eu^III center, which quench the latter’s luminescence (right).

The overall equilibrium constant, β_1n for this equilibrium is defined as follows where n = 2 for Ln^III-1 and n = 3 for Ln^III-2:

$$\log {\beta }_{1n}=\log \left(\frac{[{\text{Ln}}^{\text{III}}\mathrm{L}\cdot {\text{Pi}}_n]}{[{\text{Ln}}^{\text{III}}\mathrm{L}][\text{Pi}{]}^n}\right)$$ (2)

Note that the greater the number of water molecules displaced, the greater the increase in Eu^III-centered luminescence. Of note, previous studies have demonstrated that each phosphate anion binds the lanthanide in a monodentate manner. We previously demonstrated that through a combination of luminescence lifetime studies, Job plots, and titration curve fitting, that the complex Eu^III-2,2-Li-HOPO forms a receptor:phosphate ternary complex with a 1:2 stoichiometry whereas Eu^III-3,3-Gly-HOPO forms one with a 1:3 ratio.⁹

A second receptor composed of the same ligand but of a different lanthanide ion can compete with the Eu^III complex for the phosphate anions (Figure 2). As the phosphate exchange from the Eu^III receptor to the competing one, they are replaced by water molecules that quench the Eu^III-centered luminescence, resulting in an overall decrease in time-gated luminescence intensity. The stronger the competing receptor, the more the equilibrium is shifted to the right, the steeper the Eu^III-centered luminescence decreases upon addition of increasing concentration of Ln^IIIL. Likewise, the weaker the competing receptor, the more the equilibrium is shifted to the left, the less the Eu^III-centered luminescence decreases upon addition of increasing concentration of Ln^IIIL. The luminescence competition titration enables the determination of the constant corresponding to the equilibrium depicted in Figure 2. Since the affinity of both Eu^III receptors have independently been determined by direct titration with phosphate,^8,9 the affinity for phosphate of each lanthanide receptor can thus be calculated.

Of note, since Eu^III-2,2-Li-HOPO binds two phosphate anions, the competition experiments with this complex were performed in the presence of two equivalents of phosphate. Competitions with Eu^III-3,3-Gly-HOPO, a receptor that binds three phosphate anions, were performed in the presence of three equivalents of the anions. Importantly, this competition experiment is simplified greatly by the high affinity of the Eu^III receptors for phosphate (logβ₁₂ = 10.4 for Eu^III-2,2-Li-HOPO; logβ₁₃ = 14.5 for Eu^III-3,3-Gly-HOPO).⁹ Indeed, those high affinity ensure that in a 1:2 (for Eu^III-2,2-Li-HOPO) or 1:3 (for Eu^III-3,3-Gly-HOPO) ratio of Eu^III receptor:phosphate, almost all of the phosphate is coordinated to either the europium center or the competing lanthanide receptors. The concentration of unbound phosphate in solution in this experiment is negligible. Kinetic studies determined that competition for phosphate between two different lanthanides occurs within minutes (Figure S6). Nonetheless, in order to ensure that thermodynamic equilibrium was reached, each solution was allowed to equilibrate for 10 minutes before measuring the luminescence.

Importantly, for each data point in these titrations, the time-gated europium(III) luminescence intensity of the Eu^IIIL/Ln^IIIL/Pi_n mixture (I) is standardized to the intensity of the Eu^III complex in the absence of phosphate but in the presence of the competing lanthanide receptor (I₀). The spectra of a representative titration is shown in Figure 3 and S4. Indeed, even in the absence of phosphate, the Ln^III-HOPO receptor (Ln^III≠Eu^III) increases the luminescence intensity of the Eu^III analog present in the same solution (Figures S1 and S2). This increase in Eu^III luminescence is not due to the presence of dimers. Rather it is due to collisional energy transfer, which, in this case, enables sensitization of Eu^III by HOPO antennas coordinating other lanthanide ions. Such an energy transfer is strongly favored by coordination of the HOPO ligand to a lanthanide (Ln^III≠Eu^III), since the heavy atom facilitates intersystem crossing of the HOPO chromophore from its singlet to its triplet excited state, which in turn increases the efficiency of collisional energy transfer to Eu^III.²³ This assertion is supported by temperature-dependent studies. Indeed, enhancement of the Eu^III-centered luminescence caused by the analogous La^III- or Lu^III-HOPO complexes increases as temperature increases from 10 to 70°C (Figure S3). Such an observation is characteristic of a collisional energy transfer. Since dimerization is entropically disfavored, if dimerization was the cause of the increase in Eu^III emission, this increase would instead become smaller as the temperature increases and dimers break apart. Interestingly, the Ln^III-HOPO complexes do not increase the emission of their analogous Eu^III complex present in the same solution by the same amount. Any energy transfer from the excited ³T state of the HOPO chromophore to an excited state (emitting or not) of the lanthanide ion it coordinates decreases the efficiency of collisional energy transfer from that same HOPO moiety to a Eu^III it does not coordinate. As such, the complexes of lanthanides whose excited states energy level match poorly with the ³T state of HOPO (La^III, Gd^III, Lu^III) display higher collisional energy transfer to Eu^III.

Figure 3.

A) Time-delayed luminescence spectra of Eu^III-1 and 2 equivalents of phosphate upon addition of increasing concentrations of Sm^III-1 (I). B) Time-delayed luminescence spectra of Eu^III-1 alone upon addition of increasing concentrations of Sm^III-1 (I₀). Experimental conditions: 10 mM HEPES(aq) pH 7.4, 25°C, λ_excitation=325 nm, excitation and emission slit widths=10 nm, time delay=0.1 ms.

The results of the phosphate competition titrations are shown in Figures 4 and 5 for Ln^III-2,2-Li-HOPO and Ln^III-3,3-Gly-HOPO, respectively. A snapshot of these titrations when one equivalent of competing Ln^III complex is added to either receptor is shown in Figure 6. Except for Dy^III, the two series of complexes behave remarkably similarly. As discussed above, the higher the affinity of the competing Ln^III receptor for phosphate, the more phosphate are displaced from Eu^III and exchanged by quenching coordinated molecule, the lower the Eu^III-centered emission. In both cases, the affinity of the receptor increases down the series as the atomic number of the lanthanide increases. This is a different trend than what is observed for the solubility product (K_sp) of Ln^IIIPO₄ for the lanthanide series²⁴. Solubility constants are defined as follows:

$${\text{LnPO}}_4(s)\rightleftharpoons {\text{Ln}}^{3+}(aq)+{\text{PO}}_4^{3-}(aq)$$ (3)

$$K_{\text{sp}}=[{\text{Ln}}^{3+}][{\text{PO}}_4^{3-}]$$ (4)

Figure 4.

Decrease in Eu^III-centered emission of Eu^III-2,2-Li-HOPO·Pi₂ (1·Pi₂) in the presence of increasing concentrations of competing Ln^III receptor (Ln^III = Pr^III, Nd^III, Sm^III, Gd^III, Tb^III, Dy^III, Ho^III, Er^III, Tm^III, Yb^III, or Lu^III). I=integrated time-gated luminescence intensity from 550 to 750 nm of the Eu^III complex in the presence of the competing Ln^III receptor and 2 equivalents of phosphate; I₀=integrated time-gated luminescence intensity of the Eu^III complex in the presence of the competing Ln^III receptor. Experimental conditions: 10 mM HEPES(aq) pH 7.4, 25°C, λ_excitation=325 nm, excitation and emission slit widths=10 nm, time delay=0.1 ms.

Figure 5.

Decrease in Eu^III-centered emission of Eu^III-3,3-Gly-HOPO·Pi₃ (2·Pi₃) in the presence of increasing concentrations of competing Ln^III receptor (Ln^III = Sm^III, Gd^III, Dy^III, Er^III, or Lu^III). I=integrated time-gated luminescence intensity from 550 to 750 nm of the Eu^III complex in the presence of the competing Ln^III receptor and 3 equivalents of phosphate; I₀=integrated time-gated luminescence intensity of the Eu^III complex in the presence of the competing Ln^III receptor. Experimental conditions: 10 mM HEPES(aq) pH 7.4, 25°C, λ_excitation=335 nm, excitation and emission slit widths=10 nm, time delay=0.1 ms.

Figure 6.

Decrease in Eu^III-centered emission of Eu^III-2,2-Li-HOPO·Pi₂ (1·Pi₂) and Eu^III-3,3-Gly-HOPO·Pi₃ (2·Pi₃) in the presence of 1 equivalent of competing Ln^III receptor. I=integrated time-gated luminescence intensity from 550 to 750 nm of the Eu^III complex in the presence of the competing Ln^III receptor and 2 equivalents of phosphate for 2,2-Li-HOPO (1) or 3 equivalents of phosphate for 3,3-Gly-HOPO (2); I₀=integrated time-gated luminescence intensity of the Eu^III complex in the presence of the competing Ln^III receptor. Experimental conditions: 10 mM HEPES(aq) pH 7.4, 25°C, λ_excitation=325 nm for 2,2-Li-HOPO (1) or 335 nm for 3,3-Gly-HOPO (2), excitation and emission slit widths=10 nm, time delay=0.1 ms.

Of note, the solubility products of LnPO₄ do not display a continuous trend across the lanthanide series. It stays constant from La^III to Sm^III , then spikes for the mid-lanthanides with a maximum for Tb^III, decreases again until Yb^III, then spikes again for the late lanthanide Lu^III. This highlights our observations that trend in the solubility of metal complexes do not necessarily match trend in their stability.²⁴ La^III receptors were not able to compete for phosphate and did not cause a decrease in Eu^III-centered emission. The earlier lanthanides, including Pr^III and Nd^III competed poorly with Eu^III and were only able to displace some of the phosphate when present in significant excess. The middle lanthanides adjacent to Eu^III—Sm^III and Gd^III—had slightly higher affinity for phosphate. The later lanthanides, from Tb^III to Lu^III, required only one equivalent of competing Ln^III-HOPO receptor to remove most or nearly all phosphate from their Eu^III analog. In general, the later the lanthanide, the smaller its ionic radius, the higher its Lewis acidity and the higher its affinity for phosphate. The same trend is apparent regardless of whether the Eu^III complex is eight (Eu^III-2,2-LiHOPO) or nine (Eu^III-3,3-Gly-HOPO) coordinate, with one exception: Dy^III-3,3-Gly-HOPO does not fit the trend. Comparison of the two phosphate competition titration curves of the two Eu^III complexes with their analogous Gd^III receptors (Figure 7) demonstrate the remarkably similar trend in binding affinity of the receptors of the two ligands across neighboring lanthanides. This comparison indicates that the substantial increase in affinity of the receptor with increasing atomic number is a general characteristic of the HOPO-class of phosphate receptor and that steric hindrance at the phosphate binding site is likely negligible.

Figure 7.

Comparison of phosphate competition curves between the competing receptor Gd^III-2,2-Li-HOPO (Gd^III-1) and Gd^III-3,3-Gly-HOPO (Gd^III-2). I=integrated time-gated luminescence intensity from 550 to 750 nm of the Eu^III complex in the presence of the competing Ln^III receptor and 2 equivalents of phosphate for 2,2-Li-HOPO (1) or 3 equivalents of phosphate for 3,3-Gly-HOPO (2); I₀=integrated time-gated luminescence intensity of the Eu^III complex in the presence of the competing Ln^III receptor. Experimental conditions for Gd^III-1 curve: 10 mM HEPES(aq) pH 7.4, 25°C, λ_excitation=325 nm, excitation and emission slit widths=10 nm, time delay=0.1 ms. Experimental conditions for Gd^III-2 curve: 10 mM HEPES(aq) pH 7.4, 25°C, λ_excitation=335 nm, excitation and emission slit widths=10 nm, time delay=0.1 ms.

Using solutions of Eu^IIIL and Eu^IIIL·Pi_n at the same concentration and conditions as references, the concentration of competing receptor needed to generate an equal partition of Pi between the Eu^III and Ln^III receptors gives the difference in phosphate association constant (see ESI). The overall association constants, defined in equation (2), for each competing complexes were then calculated from the cumulative association constants reported for Eu^III-1 and Eu^III-2 with phosphate (Table 1).⁹ As anticipated based on the titration curves, a small decrease in ionic radii (−0.013 Å)¹⁰ results in a substantial increase in overall binding affinity logβ (2.2) from Eu^III-1 to Gd^III-1. The change in phosphate affinity is so great that the logβ of complexes before Nd^III-1 and after Tb^III-1 can only be estimated by this method. The magnitude in change of association constant for both Ln^III-1 and Ln^III-2 is, to the best of our knowledge, the largest observed for lanthanide complexes for any anion in water. Other work on anion binding to lanthanide complexes find variation of the association constant less than an order of magnitude between adjacent lanthanides or even across the entire series.^16-18

Table 1.

pK_a for the hydrated Ln^III ion, solubility product, K_sp of LnPO₄ (s), and overall phosphate association constants, logβ_1n = [Ln^IIIL·Pi_n]/([Ln^IIIL][Pi]ⁿ) , of Ln^III-2,2-Li-HOPO and Ln^III-3,3-Gly-HOPO.

	pKaa	logKsp (LnPO4)b	LnIII-2,2- Li-HOPO logβ12	LnIII-3,3- Gly-HOPO logβ13
PrIII			< 9.8	< 14.5
NdIII	8.37	−26.0	9.8	-
SmIII	8.05	−25.8	10.4	14.7
EuIII	8.02	−25.4	10.4c	14.5c
GdIII	8.08	−25.1	12.6	17.0
TbIII	7.91	−25.2	13.6	-
DyIII	7.91	−25.6	> 14	15.1
HoIII – LuIII			> 14	> 17

^aRef. 25, ionic strength of 0.1 M.

^bRef. 24.

^cRef. 9.

Stability.

As demonstrated above, the affinity of lanthanide receptors for phosphate increases significantly and progressively across the series. This trend follows the trend in Lewis acidity of the rare earths and indicates that steric hindrance likely does not affect anion coordination. Since the coordinating hydroxypyridinonate ligand is also anionic, a similar trend should thus be anticipated regarding the stability of the lanthanide receptor. The stability of each lanthanide complex of 2,2-Li-HOPO at pH 7.4 was determined by direct competition with diethylenetriaminepentaacetic acid (DTPA), a ligand for which the stability constant of all lanthanide complexes is known.²⁶ As previously reported, the equilibrium, depicted in equation (5), enables direct determination of the stability of the complex at that pH in terms of pM (pM = pLn for M=Ln^III).^22,27

$${\text{Ln}}^{\text{III}}{\mathrm{L}}_{(aq)}+{\text{DTPA}}_{(aq)}\rightleftharpoons {\text{Ln}}^{\text{III}}-{\text{DTPA}}_{(aq)}+{\mathrm{L}}_{(aq)}$$ (5)

where the pM is defined by the concentration of free (uncomplexed) metal in solution as follows:

$$\text{pM}=\text{-log}[\mathrm{M}{]}_{\text{free}}\text{when}[\mathrm{L}{]}_{\text{total}}={10}^{\text{-}5}\mathrm{M}\text{and}[\mathrm{M}{]}_{\text{total}}={10}^{\text{-}6}\mathrm{M}$$ (6)

The more stable the complex, the higher its pM value, and vice versa. The pM of all the lanthanide complexes of DTPA can be calculated from the known formation constant, K_ML, and the ligand protonation constants,²² most easily using the software Hyss.²⁸ In this protocol previously established by Raymond,²⁷ the concentration of competing DTPA ligand needed to partition equally the metal ion between the hydroxypyridinone ligand and DTPA (log([DTPA]/[L]) when log([M-DTPA]/[M-L]) = 0) directly gives the pM of the complex Ln^IIIL at that pH. In this case, concentrations of free and complexed of 2,2-Li-HOPO (1) were determined by UV-Visible spectroscopy using spectra of 1 and Ln^III-1 as reference. Advantageously, this technique can be employed for all lanthanide complexes of the series. Of note, we could not determine the stability of lanthanide complexes of 3,3-Gly-HOPO (2) in this manner, since addition of DTPA results in the formation of ternary Ln^III-2·(DTPA) complexes as opposed to a strict metal exchange.

As can be seen in Figure 8 and Table 2, the stability of lanthanide complexes of 2,2-Li-HOPO increases continuously and significantly across the series; the pM values for the late lanthanide complexes do not level off. The trend in stability of lanthanide coordination complexes across the series has been investigated with numerous ligands^13,29 and can be categorized into three general categories. Coordination complexes belonging to category 1 display a consistent increase in stability from La^IIIL to Lu^IIIL. To belong in this series, a ligand must be able to wrap itself around the lanthanide ion in such a way that no steric hindrance affects coordination, particularly for the smaller, later lanthanides. Complexes of this class generally employ ligands of lower denticity, as exemplified by the Ln^III complexes of the hexadentate EDTA³⁰ (Figure 9), although some octadentate ligands with binding pockets large and flexible enough to accommodate both large and small rare earths also belong in this category.³¹ Complexes belonging to category 2 present the opposite trend: a consistent decrease in stability from La^IIIL to Lu^IIIL. The ligands in this series, exemplified by MACROPA,³² are typically rigid with larger binding pockets better suited for the earlier Ln^III ions but not flexible enough to accommodate the later ones. The third category combines the previous two. Complexes in this category initially display an increase in stability with increasing atomic number. As the rare earth become smaller, though, steric hindrance impeding ligand coordination becomes significant and the trend in the stability of the complexes levels out and eventually reverses. This is most famously observed with DTPA²⁶ and some if its amide derivatives³³ but also more rigid ligands whose binding pockets is better suited for the middle Ln^III ion, such as DOTA³⁴ and DO2A.³⁵ Importantly, though, ligands and complexes are categorized based on the trends in stability, not on the absolute value of the stability constants. Each category contains both highly and poorly stable complexes. The observed trend of increasing stability with decreasing ionic radii thus places the Ln^III-2,2-Li-HOPO complexes among the first category. This strongly indicates that steric hindrance does not negatively affect coordination of the rare earth. In comparison, another HOPO-based ligand, 3,4,3-LI(1,2-HOPO), showed a trend of increasing stability with decreasing ionic radii, except for Lu^III complex, which was less stable than the Yb^III one.³⁶ The significantly smaller difference in selectivity between lanthanide ion for the octadentate 3,4,3-LI(1,2-HOPO) (pLu – pLa = 4.8) compared to the hexadentate 2,2-Li-HOPO (pLu – pLa > 7.0) indicate that ligands with smaller denticity can achieve greater selectivity between lanthanide ions. This is likely due to the greater steric hindrance that negatively affect the stability of the smaller late lanthanides.

Figure 8.

Trend in the stability across the lanthanide series of pLn values of the Ln^III complexes of 2,2-Li-HOPO (red square) compared to EDTA (open square),³⁰ MACROPA (closed triangle),³² DTPA (star).²⁶

Table 2.

Ionic radii and pK_a for the hydrated Ln^III and their stability constants (logβ and pLn) for DPTA and 2,2-Li-HOPO (1).

	Ionic Radii (Å)a	pKab	logβ LnDTPAc	pLn (DTPA) d	ΔpLn (1 – DTPA)	pLn (1)
LaIII	1.160	9.24	19.48	16.1	− 1.1	15.0
PrIII	1.126	8.58	21.07	17.7	− 0.5	17.2
NdIII	1.109	8.37	21.60	18.2	0.1	18.3
SmIII	1.079	8.05	22.34	19.0	0.1	19.1
EuIII	1.066	8.02	22.39	19.0	0.6	19.6
GdIII	1.053	8.08	22.46	19.1	0.7	19.8
TbIII	1.040	7.91	22.71	19.4	0.6	20.0
DyIII	1.027	7.91	22.82	19.5	0.9	20.4
HoIII	1.015	7.76	22.78	19.4	1.1	20.5
ErIII	1.004	7.78	22.74	19.4	1.5	20.9
TmIII	0.994	7.91	22.72	19.4	1.8	21.2
YbIII	0.985	7.44	22.62	19.3	2.4	21.7
LuIII	0.977	7.52	22.44	19.1	> 3	>22

^aRef. 10, coordination number of 8.

^bRef. ²⁵, ionic strength of 0.1 M.

^cRef. ²⁶

^dCalculated using Hyss from the stability constant values and the protonation constants of DTPA.²⁸

Figure 9.

Chemical structures of ligands exemplifying the three categories of stability down the lanthanide series.

There are two notable aspects to this trend. The first is that despite their lack of rigidity, these complexes display exceptionally high stability, especially so for the late lanthanides. Although the earlier lanthanides of 2,2-Li-HOPO have pM values comparable to those of DTPA, the later ones are noticeably more stable. In fact, Lu^III-2,2-Li-HOPO is so much more stable than Lu^III-DTPA that only a lower estimate of its pLu^III could be accurately determined. The remarkably high stability of the later lanthanide complexes of 2,2-Li-HOPO, combined with their high affinity for phosphate, make them particularly promising candidates for translational applications to medical or environmental sequestration of phosphate. Increased efficacy of the lanthanide receptor comes with increased stability. This high stability will further reduce the risk of the rare earth leaching during anion separation, thereby preventing heavy metal contamination.

The second notable aspect is the significant difference in stability between complexes of the late versus the early lanthanide. The Lu^III complex of 2,2-Li-HOPO is at least seven orders of magnitude more stable than the La^III one (pLu^III – pLa^III = 7). In comparison, pLu^III-pLa^III for EDTA, another category 1 ligand, is barely 4.4,³⁰ and for 3,4,3-LI(1,2-HOPO), another 1,2-HOPO ligand, is 4.8.³⁶ Such a large difference in stability values between lanthanide complexes is unusual²⁹ and bodes well for potential applications in rare earth separation.

CONCLUSION

Previous work on tripodal tris-bidentate complexes of Gd^III demonstrated the significant role that the basicity of a ligand has on both the stability of the complex and on its affinity for anions.⁷ The more basic the ligand, the more stable the Gd^III complex and the lower its affinity for anions. Any Gd^III complexes with open coordination site that did bind anions did so with a selectivity that matched the basicity of the ligand: in the absence of steric hindrance, the affinity and/or selectivity of lanthanide complexes with wide open coordination sites occurred in the order phosphate>bicarbonate>fluoride. It follows from these prior studies that the ionic radius of a rare earth, a parameter directly linked to its Lewis acidity, would thus also correlate with affinity for anions, regardless of whether that anion is the one coordinating it (in this case the 1,2-hydroxypyridinonate ligand) or the guest targeted by the lanthanide receptor (in this case phosphate). This is indeed the trend observed. Both the affinity of the receptor for phosphate and its stability (pM) increases going from left to right across the lanthanide series, as the ionic radii of the rare earth ion decreases and their acidic character increases. The trend in phosphate affinity holds for both HOPO ligands investigated, even though the Eu^III complex of 2,2-Li-HOPO is 8-coordinate whereas that 3,3-Gly-HOPO is 9. Previous crystal structures of lanthanide complexes of this class demonstrated that, for both 8 and 9 coordinate lanthanides, the inner-sphere water molecules are far removed from the HOPO ligand, and the faces of the complex that are open for phosphate coordination lack any steric hindrance.³⁷ This lack of steric hindrance accounts for the continuous increase in both affinity for phosphate and stability. Advantageously, both the efficacy of the receptor and its stability can be increased simultaneously. Altogether, the trends highlight the translational potential of the late rare earth for anion sequestration. Lutetium(III), in particular, stands out for its very high affinity for phosphate in water, higher than that of any other receptor, and its very high stability—seven orders of magnitude higher than that of lanthanum(III). These benefits, unfortunately, come at the cost of the increased price of the late lanthanides, thereby highlighting the need to develop novel methodology to establish recyclable receptors that can catch-and-release at will.

EXPERIMENTAL

General Considerations.

Unless stated otherwise, all chemicals were purchased from commercial suppliers and used without further purification. Deuterated solvents were obtained from Cambridge Isotope Laboratories (Tewskbury, MA). Distilled water was further purified by a Millipore Simplicity UV system (resistivity of 18×10⁶ Ω). All organic extracts were dried over anhydrous MgSO₄(s). Flash chromatography was performed on Merck Silica Gel. ¹H NMR and ¹³C NMR spectra were recorded on a Bruker Advance III 400 instrument at 400 and 100 MHz, respectively, or a Bruker Advance III AV 500 instrument at 500 and 125 MHz, respectively, at the LeClaire-Dow instrumentation facility of the Department of Chemistry of the University of Minnesota. The residual solvent peaks were used as internal references. Data for ¹H NMR are recorded as follows: chemical shift (δ, parts per million), multiplicity (s, singlet; d, doublet; t, triplet; q, quartet; br, broad; m, multiplet), coupling constant (Hz), integration. Data for ¹³C NMR are recorded as follows: chemical shift (δ, parts per million). Low-resolution (LR) and high-resolution (HR) electrospray ionization time-of-flight mass spectrometry (ESI/TOF- MS) data were recorded on a Bruker BioTOF I instrument at the LeClaire-Dow instrumentation facility of the Department of Chemistry of the University of Minnesota. Ultraviolet–visible spectra were recorded on a Varian Cary 100 Bio Spectrophotometer. Data were collected over the range of 250-400 nm. Luminescence data were acquired on a Varian Cary Eclipse fluorescence spectrophotometer using a quartz cell with a path length of 1 cm and a chamber volume of 400 μL. Time-gated luminescent spectra were recorded with a time delay of 0.1 ms and a gate time of 5 ms. Unless otherwise stated, sample solutions were allowed to equilibrate for 10 min before measurement of their luminescence spectra, as we have demonstrated that this time was sufficient to achieve thermodynamic equilibrium (Figure S6). Luminescence data were processed with Scilab 6.0.2 and QtiPlot 0.9.8.9 software. All pH measurements were performed using a Thermo Scientific Ag/AgCl refillable probe and a Thermo Orion 3 Benchtop pH meter. High-performance liquid chromatography (HPLC) data were collected on a Varian Prostar model 210 instrument, coupled with an Agilent ZORBAX Eclipse XDB-C18 column, and a Varian ProStar 335 diode array detector. Unless specified otherwise, HPLC measurements were performed at a flow rate of 1.0 mL min⁻¹ with the following elution condition: 15% CH₃CN/85% water from 0 to 2 min, followed with a linear gradient to 85% CH₃CN/15% water from 2 to 23 min, 85% CH₃CN/15% water from 23 to 26 min, with a linear gradient to 15% CH₃CN/85% water from 26 to 30 min, and 15% CH₃CN/85% water from 26 to 32 min.

Synthesis.

The free ligands 1 (2,2-Li-HOPO) and 2 (3,3-Gly-HOPO) were synthesized according to published procedures with successful synthesis established by ¹H NMR, ¹³C NMR and ESI-MS.⁹

Representative procedures for the synthesis of Ln^III-1 (Ln^III-2,2-Li-HOPO):

Nd^III-2,2-Li-HOPO (Nd^III-1). A solution of NdCl₃·6H₂O (7 mg, 0.02 mmol, 1 eq.) in methanol (5 mL) was slowly added to a solution of the free ligand 2,2-Li-HOPO (1, 10. mg, 0.019 mmol) in methanol (5 mL). Pyridine (16 μL, 0.19 mmol) was subsequently added to the reaction mixture that was stirred 60°C for 6 h. The reaction mixture was cooled down to room temperature, and the resulting white precipitate was filtered, rinsed with ethanol, and dried under reduced pressure to yield the Nd^III complex, Nd^III-1, as a white solid (11 mg, 87%). ESI-HRMS: m/z = 678.0185 ([M+Na]⁺), (Calcd. 678.0211), m/z = 732.0487 ([M+DMSO+H]⁺), (Calcd. 732.0528).

La^III-1. (9 mg, 80%). ESI-HRMS: m/z = 651.0457 ([M+H]⁺), (Calcd. 651.0350), m/z = 729.0556 ([M+DMSO+H]⁺), (Calcd. 729.0489).

Pr^III-1. (10 mg, 79%). ESI-HRMS: m/z = 653.0401 ([M+H]⁺), (Calcd. 653.0363), m/z = 731.0536 ([M+DMSO+H]⁺), (Calcd. 731.0502).

Nd^III-1. (11 mg, 87%). ESI-HRMS: m/z = 678.0185 ([M+Na]⁺), (Calcd. 678.0211), m/z = 732.0487 ([M+DMSO+H]⁺), (Calcd. 732.0528).

Sm^III-1. (10 mg, 78%). ESI-HRMS: m/z = 742.0622 ([M+DMSO+H]⁺), (Calcd. 742.0621).

Eu^III-1. (10 mg, 78%). ESI-HRMS: m/z = 695.0479 ([M+CH3OH-H]⁻), (Calcd. 695.0606), m/z = 709.0388 ([M+COOH₂-H]^-), (Calcd. 709.0399).

Gd^III-1. (11 mg, 85%). ESI-HRMS: m/z = 670.0459 ([M+H]⁺), (Calcd. 670.0532).

Tb^III-1. (11 mg, 85%). ESI-HRMS: m/z = 671.0527 ([M+H]⁺), (Calcd. 671.0540).

Dy^III-1. (10 mg, 77%). ESI-HRMS: m/z = 754.0783 ([M+DMSO+H]⁺), (Calcd. 754.0721), m/z = 776.0128 ([M+DMSO+Na]⁺), (Calcd. 776.0540).

Ho-^III1. (10 mg, 76%). ESI-HRMS: m/z = 699.0246 ([M+Na]⁺), (Calcd. 699.0409), m/z = 755.0509 ([M+DMSO+H]⁺), (Calcd. 755.0729).

Er-^III1. (9 mg, 69%). ESI-HRMS: m/z = 780.0484 ([M+DMSO+Na]⁺), (Calcd. 780.0575).

Tm^III-1. (12 mg, 92%). ESI-HRMS: m/z = 759.0566 ([M+DMSO+H]⁺), (Calcd. 759.0768), m/z = 781.0346 ([M+DMSO+Na]⁺), (Calcd. 781.0587).

Yb^III-1. (12 mg, 90%). ESI-HRMS: m/z = 708.0461 ([M+Na]⁺), (Calcd. 708.0498).

Lu^III-1. (10 mg, 76%). ESI-HRMS: m/z = 687.0670 ([M+H]⁺), (Calcd. 687.0694).

Representative procedures for the synthesis of Ln^III-2 (Ln^III-3,3-Gly-HOPO):

La^III-3,3-Gly-HOPO (La^III-2). A solution of 1 M NaOH (6 μL, 6 μmol) was added to a solution of the free 3,3-Gly-HOPO ligand (2, 2 mg, 3 μmol) in water (5 mL). A solution of LaCl₃·7H₂O (1 mg, 3 μmol) in water (5 mL) was subsequently added slowly to the reaction mixture, which was heated to 60°C for 6 h. The reaction mixture was then cooled to room temperature and the resulting complexes were characterized and used without further purification.

La^III-2. ESI-HRMS: m/z = 736.0910 ([M+H]⁺), (Calcd. 736.0877).

Sm^III-2. ESI-HRMS: m/z = 749.0993 ([M+H]⁺), (Calcd. 749.1012).

Eu^III-2. ESI-HRMS: m/z = 771.0860 ([2M+2Na]²⁺), (Calcd. 771.0841).

Gd^III-2. ESI-HRMS: m/z = 755.1041 ([M+H]⁺), (Calcd. 755.1060).

Dy^III-2. ESI-HRMS: m/z = 782.0851 ([2M+2Na]²⁺), (Calcd. 782.0922).

Er^III-2. ESI-HRMS: m/z = 786.0919 ([2M+2Na]⁺), (Calcd. 786.0952).

Lu^III-2. ESI-HRMS: m/z = 872.1099 ([M+Na]⁺), (Calcd. 872.1180).

Phosphate Competition Titrations.

The overall phosphate affinity constants for the competing complexes (logβ₁₂ for Ln^III-2,2-Li-HOPO and logβ₁₃ for Ln^III-3,3-Gly-HOPO) were determined by competition with the Eu^III analog. Samples were prepared with known concentrations of Eu^III-1 and Eu^III-2, 2 or 3 equivalents of phosphate for 2,2-Li-HOPO or 3,3-Gly-HOPO, respectively, and the corresponding Ln^III-1 or Ln^III-2 (0 to 4 eq.) in HEPES buffer (10 mM, pH 7.4). To avoid the inner filter effect, the total concentration of HOPO ligand was kept sufficiently low to keep the overall absorbance at the excitation wavelength below 0.1. For samples with < 1 equivalent of Ln^IIIL, the concentration of Eu^IIIL was 3.7 μM. For samples with ≥ 1 equivalent of Ln^IIIL, the concentration of Eu^IIIL was 1.8 μM. The time-gated luminescence (delay time of 0.1 ms, excitation slit width = 10 nm, emission slit width = 10 nm, T = 298 K) was measured for each sample after 10 min of mixing to ensure that thermodynamic equilibrium was reached (Figure S6). The Eu^III luminescence was integrated from 550 to 750 nm. The concentrations of phosphate bound and phosphate free Eu^III complexes in each solution were determined by using solutions of Eu^IIIL and Eu^IIIL·Pi_n at the same concentration and conditions as references. The concentration of competing Ln^IIIL necessary generate an equal partition of Pi between Eu^IIIL and Ln^IIIL occurs at the x-intercept of the log-log plot (Figures S8 and S9), which is equal to the difference in overall phosphate affinity constants (logβ_Ln – logβ_Eu) . Using the known logβ₁₂ or logβ₁₃ for Eu^IIIL, the constants for Ln^IIIL can be determined.

Stability of lanthanide complexes.

The stability constants of the lanthanide complexes of 2,2-Li-HOPO (1, Ln^III = La^III, Pr^III, Nd^III, Sm^III, Eu^III, Gd^III, Tb^III, Dy^III, Ho^III, Er^III, Tm^III, Yb^III and Lu^III) were determined by competition with diethylenetriaminepentaacetic acid (DTPA) following a previously reported procedure.²⁷ Briefly, samples were prepared containing known concentrations of Ln^III complex (Ln^III-1) and DTPA (1 to 50 eq.) in HEPES buffer (10 mM, pH 7.4) with KCl (0.1 M). All samples were equilibrated at 60°C for 24 h, after which UV-visible spectra were recorded. The concentrations of free and complexed ligand in each solution were averaged over 20 wavelengths by using solutions of 1 and Ln^III-1 at the same concentration and conditions as references. Using the known stability constants for DTPA with the Ln^III ions tested,²⁶ their hydrolysis constants³⁸ and the protonation constants of DTPA,³⁹ the pLn^III values for each lanthanide were calculated using Hyss.²⁸ The concentration of competing ligand necessary to generate an equal partition of Ln^III between the ligand of interest and competitor occurs at the x-intercept of the log-log plot (Figures S10-S12).