A supramolecular lanthanide separation approach based on multivalent cooperative enhancement of metal ion selectivity

Multivalent cooperativity plays an important role in the supramolecular self-assembly process. Herein, we report a remarkable cooperative enhancement of both structural integrity and metal ion selectivity on metal-organic M4L4 tetrahedral cages self-assembled from a tris-tridentate ligand (L1) with a variety of metal ions spanning across the periodic table, including alkaline earth (CaII), transition (CdII), and all the lanthanide (LnIII) metal ions. All these M4L14 cages are stable to excess metal ions and ligands, which is in sharp contrast with the tridentate (L2) ligand and bis-tridentate (L3) ligand bearing the same coordination motif as L1. Moreover, high-precision metal ion self-sorting is observed during the mixed-metal self-assembly of tetrahedral M4L4 cages, but not on the M2L3 counterparts. Based on the strong cooperative metal ion self-recognition behavior of M4L4 cages, a supramolecular approach to lanthanide separation is demonstrated, offering a new design principle of next-generation extractants for highly efficient lanthanide separation.

The titration experiments of L 2 with metal ions (M = Eu III , Ca II , Cd II ) indicated the formation of more than one kind of metal-organic assembly species with the ratio R M/L1 ranging from 0 to 1.0, suggesting a rather low stability of the monometallic complexes, which hindered its application of in metal ion separation. Figure 114. 1   ] 12+ (B) and La III /Pr III mixed-metal self-assembled complexes with L 1 (ClO 4 − salt) (C) with spectra zoomed in over some of the peak ranges selected for the integrals. Non-absolute metal ion self-recognition assembly (at 40°C for 1h) of L 1 (0.01 M) with La III /Pr III mixtures (0.01 M of each), with 88.50 percent Pr III and 11.50 percent La III in the assembled complexes, was observed according to the 1 H NMR characterization and no spectra change in (C) was observed even after 2 weeks. Figure 156. 1  − salt) (C) with spectra zoomed in over some of the peak ranges selected for the integrals. Non-absolute metal ion self-recognition assembly (at 40°C for 1h) of L 1 (0.01 M) with Pr III /Eu III mixtures (0.01 M of each), with 95.24 percent Eu III and 4.76 percent Pr III in the assembled complexes, was observed according to the 1 H NMR characterization and no spectra change in (C) was observed even after 2 weeks. ] 12+ (B) and Nd III /Eu III mixed-metal self-assembled complexes with L 1 (CF 3 SO 3 − salt) (C) with spectra zoomed in over some of the peak ranges selected for the integrals. Non-absolute metal ion self-recognition assembly (at 40°C for 1h) of L 1 (0.01 M) with Nd III /Eu III mixtures (0.01 M of each), with 92.59 percent Eu III and 7.40 percent Nd III in the assembled complexes, was observed according to the 1 H NMR characterization and no spectra change in (C) was observed even after 2 weeks. ] 12+ (B) and Nd III /Y III mixed-metal self-assembled complexes with L 1 (CF 3 SO 3 − salt) (C) with spectra zoomed in over some of the peak ranges selected for the integrals. Non-absolute metal ion self-recognition assembly (at 40°C for 1h) of L 1 (0.01 M) with Nd III /Y III mixtures (0.01 M of each), with 83.33 percent Y III and 16.67 percent Nd III in the assembled complexes, was observed according to the 1 H NMR characterization and no spectra change in (C) was observed even after 2 weeks. ] 12+ (B) and Sm III /Y III mixed-metal self-assembled complexes with L 1 (CF 3 SO 3 − salt) (C) with spectra zoomed in over some of the peak ranges selected for the integrals. Non-absolute metal ion self-recognition assembly (at 40°C for 1h) of L 1 (0.01 M) with Sm III /Y III mixtures (0.01 M of each), with 68.49 percent Y III and 31.51 percent Sm III in the assembled complexes, was observed according to the 1 H NMR characterization and no spectra change in (C) was observed even after 2 weeks. Pr III /Sm III mixed-metal self-assembled complexes with L 3 (CF 3 SO 3 − salt) (C) with spectra zoomed in over some of the peak ranges selected for the integrals. Non-absolute metal ion self-recognition assembly (at 40°C for 1h) of L 3 (0.0075 M) with Pr III /Sm III mixtures (0.005 M of each), with 88.50 percent Sm III and 11.50 percent Pr III in the assembled complexes, was observed according to the 1 H NMR characterization and no spectra change in (C) was observed even after 2 weeks. (1.5 μmol) in CD 3 CN, with the total volume of CD 3 CN as 0.6mL, the substitution process was monitored by 1 H NMR spectroscopy at room temperature, and progressively substitution of La(III) by Pr(III) was clearly observed, as new 1 H NMR signals corresponding to Pr(III) coordination environment appeared. During the time of 6min to 14h, the highly complicated 1 H NMR spectra suggested a mixture of [(La n Pr 4-n )L 1 4 ] 12+ (n = 1 to 4) and at the equilibrium state, exactly the same 1 H NMR spectrum was observed as that seen in the one-pot self-assembly process. It is worth mentioning that the substitution rate for each metal combination depends on the difference in the ionic radii, and a larger difference results in faster substitution and vice versa. For example, it takes only 30 minutes for Y(III) to fully substitute all four La(III) vertices in the La 4 L 1 4 cage, whereas it takes more than 20 hours for substitution with Pr(III).  By carefully comparing the observed and simulatedisotope patterns of ESI-TOF-MS, it can be seen that a mixture of [Ln n Pr (4-n) L 1 4 ] 12+ (n=0-4) complexes were observed at 7min and La(III) was gradually substituted by Pr(III) over ca. 30min. The metathesis experiments reached equilibrium faster than that in the NMR experiment, which is possibly due to a low reaction concentration in this case.

Supplementary Tables
Supplementary Table 1. Comparison of the 1 H NMR chemical shifts (δ i ) for L 1 and tetrahedral complexes Ln 4 (L 1 ) 4 (CD 3 )] was calculated from the integrals in the 1 H NMR spectra of mixed-metal self-assembly complexes, with a systematic error of 5%.  represents complete metal selectivity,  represents highly efficient metal selectivity and  represents no obvious metal selectivity. b r Ln(III) /Å is ionic radii for coordination number as 9 S8 . c No Lna(III) complexes were detected by NMR and ESI-MS. d The ionic radii of Ln(III) ions shrinks gradually along the series as a result of lanthanide contraction. While our results on metal ion selectivity mainly follow the ionic radii sequence, other factors, such as electronegativity, could not be neglected. According to the Pauling's rule in electronegativity, lanthanide ions with smaller ionic radii tend to have bigger electronegativity. However, there are always exceptions. For example, Yb(III), which has an ionic radius just bigger than the smallest Lu(III), bears the minimum electronegativity among all the trivalent lanthanides. So there is a difference in metal ion selectivity between Yb(III) and Lu(III). Moreover, ESI-TOF-MS measurements indicate non-absolute metal ion self-recognition assembly of L 1 with Ln III /Yb III (Ln = Ce, Pr, Nd, Sm, Eu) mixtures. However, the severe line-broadening of the 1 H NMR spectrum caused by paramagnetism of Yb III hinders precise quantitation of different lanthanide ions in the assembled complexes through integration of the NMR spectra.

Supplementary Methods Synthesis and physical properties of ligands L 1-6
Synthesis of L 1-3 : Enantiomeric pure L 1-3 were synthesized according to reported procedures with spectroscopic data consistent with the literature. S1  13   Synthesis of L 4-6 : Compound 6 (1.86 g, 3.70 mmol, 1 equiv.) was dissolved in thionyl chloride (SOCl 2, 10 mL), followed by the addition of 2 drops of DMF at room temperature. After stirring the reaction mixture at 50°C for 3h, excess SOCl 2 was removed in vacuo and the resulting residue was used without further purification. The acid chloride was dissolved in anhydrous DCM (50 mL) and treated with 1,3,5-tris(4aminophenyl)benzene (325.1 mg, 0.92 mmol, 0.25 equiv.) and N-ethyldiisopropylamine (1 mL) at 0°C. The resulting solution was allowed to warm to room temperature and the reaction was monitored by TLC until all starting materials were consumed. After removing the solvent under reduced pressure, the product was purified with column chromatography (DCM:MeOH = 100:1) to give L 6 as pale yellow solid (1.30 g, 0.72 mmol) in 78% yield. 1 13   Single crystal X-ray diffraction studies.
X-ray data was collected at room temperature. The crystal was transferred and sealed inside a glass capillary with an atmosphere of the mother liquor without exposure to air. X-ray data collection at cryogenic conditions for these compounds resulted in the deterioration of crystallinity due to unknown reasons and gave worse quality of data. The reason for the deterioration of the crystals under cryogenic conditions possibly has something to do with the large cavities existing inside the big unit cells that are filled with amorphous organic solvents such as diethyl ether, THF and so on, which may still slowly diffuse/evaporate under liquid N 2 temperature. This is a quite different feature from biological samples which are always grown from water. So it is quite common to see X-ray data collected at room temperature for supramolecular systems. Our own experiences also suggest that to seal the crystals inside a glass capillary and collect the data quickly at room temperature may be a general protocol to follow for fragile crystals grown from volatile organic solvents. S1, S4,
Crystal data for La 4 L 1 4 (ClO 4 ) 12 : Space group P2 1 3, a = b = c =42.908(2) Å, V = 78995(13) Å 3 , Z = 8, T = 293K. Anisotropic least-squares refinement for the framework atoms and isotropic refinement for the other atoms on 27430 independent merged reflections (R int = 0.0838) converged at residual wR 2 = 0.2553 for all data; residual R 1 = 0.0940 for 21686 observed data [I > 2σ(I)], and goodness of fit (GOF) = 1.090. Additional Comments: The crystals of these kinds of giant supramolecular assemblies diffract very weakly in nature. The diffractions for compound Cd 4 L 1 4 (ClO 4 ) 8 and La 4 L 1 4 (ClO 4 ) 12 are very limited even though we optimized the measurement based on synchrotron radiations. The final R factor was converged to very high