Coordination-driven self-assembly is a powerful tool to construct cage-shaped metal complexes with a nano-sized interior space as a platform for space-specific molecular recognition and reactions. Although a large number of excellent examples of metallo-cage complexes have been reported so far,^1–7 it is still a challenge to precisely arrange molecular binding sites on the inner surface. A chemically functionalized confined space with internal molecular binding sites, as can be seen in the internal binding pockets of a variety of metallo-enzymes, is a necessary requirement to acquire functional spaces with high efficiency and selectivity.

Recently, porphyrin-based self-assembled architectures, in particular, cage complexes have attracted a lot of attention because of the porphyrin-specific photochemical and redox functions.^8–19 Typical examples are O-symmetric cubic cage M₈L₆ complexes with eight chemically equivalent metal centers as formed from 4-fold symmetric porphyrin-based ligands and metal ions.^10–12 In contrast, as we previously reported,¹⁷ a self-assembled cage complex Zn₁₁(L0·H₂O)₆(H₂O)₁₈(OTf)₂₂ (L0-Zn₁₁) with similar 4-fold symmetric porphyrin-based ligands (L0 = Zn^II-porphyrin-centered tetrakis-meso-(2,2′-bipyridyl) ligand, Figure 1a) has a lower symmetry with three types of metal centers.²⁰ This D₃ symmetric M₁₁L₆ type complex L0-Zn₁₁, formed from the C_4v symmetric porphyrin ligand L0 and Zn(OTf)₂, exhibits its ability to unsymmetrically encapsulate guest molecules (Tf = trifluoromethanesulfonyl).¹⁷ However, in the crystal structure, the axial ligands bind only from outside the cage to the Zn^II-porphyrin centers, and there is no convincing evidence for inward binding of coordinating molecules to the Zn^II-porphyrin center. Herein we report the effects of chemical modification of the porphyrin ligand from L0 to tetramethylated L1 to improve the solubility and coordination ability of the ligand. As a result, two isostructural M₁₁(L1·CH₃OH)₆(H₂O)₁₈(OTf)₂₂ (M = Cd^II or Zn^II) cage complexes, L1-Zn₁₁ and L1-Cd₁₁, having six sites to which an included molecule is bonded were obtained (Figure 1b).

Figure 1. a) Chemical structures of ligands, L0·H₂O and L1·H₂O. b) Synthetic schemes of L1-Zn₁₁ and L1-Cd₁₁ by complexation between L1·H₂O and Zn^II or Cd^II ions.

The Zn^II-porphyrin ligand L1·H₂O, a tetramethyl derivative of L0·H₂O, was synthesized from 5′-methyl-2,2′-bipyridyl-5-carboxyaldehyde and pyrrole via Adler synthesis and successive complexation with Zn(OAc)₂·(H₂O)₂ in 6.6% overall yield (Supporting Information S3), and characterized by NMR, ESI-TOF MS, and elemental analyses. The C_4v symmetric structure of L1·H₂O was confirmed by ¹H NMR spectroscopy (Figure 2a).

Figure 2. a)–c) ¹H NMR spectra of L1·H₂O, L1-Zn₁₁, and L1-Cd₁₁ (500 MHz, 300 K, CDCl₃/CD₃OD/D₂O = 10:10:1 (v/v/v)). d) X-ray crystal structure of [L1-Cd₁₁]²²⁺ in which one Zn^II-porphyrin is highlighted and color-coded in red, yellow, green, and blue to illustrate a C₁ symmetry in [Cd₁₁L1₆]²²⁺. e) Chemical structure and assignment of each hydrogen atom of L1. Colors are coded according to the structure in (d).

Firstly, L1-Zn₁₁ was synthesized by complexation of L1·H₂O with 11/6 eq. Zn(OTf)₂ in an aqueous mixed solvent, CDCl₃/CD₃OD/D₂O = 10:10:1 (v/v/v). The reaction mixture was heated at 50 °C for 24 h. L1-Zn₁₁ was isolated by precipitation with Et₂O as a dark-green solid in 78% yield. The ¹H NMR spectrum of L1-Zn₁₁ in CDCl₃/CD₃OD/D₂O = 10:10:1 (v/v/v) showed 32 signals in the aromatic region and 4 singlet signals in the alkyl region, suggesting the presence of a C₁ symmetric L1 ligand in the self-assembled structure (Figure 2b). This spectrum was in stark contrast with a simple spectrum of ligand L1 with a C_4v symmetry. Each ¹H signal was fully assigned by ¹H-¹H COSY (Figure S9) and ¹H-¹H NOESY spectra (Figures S10 and S11). The interligand NOE patterns of L1-Zn₁₁ (g₁-g′₂, g′₁-g₂, a₂-a₄, f₂-f₄, g₂-f₄, a₃-f₃, d₃-g₃, d₃-g′₃) were well matched with those of the previously reported L0-Zn₁₁. Moreover, the presence of [Zn₁₁L1₆(OTf)_22−n]ⁿ⁺ (n = 5–7) in solution was verified by ESI-MS spectrometry (Figure S14). L1-Zn₁₁ was further characterized by ¹H DOSY (Figure S12), ¹⁹F NMR (Figure S13), and elemental analyses (Supporting Information S13).

Next, to examine the effects of metal exchange on the self-assembly, we selected Cd^II as a zinc group element. Cd^II has a larger ionic radius (ca. 0.95 Å) in an octahedral hexacoordinate geometry than that of Zn^II (ca. 0.74 Å).²¹ In spite of the larger ionic radius of Cd^II and the smaller binding constants of Cd^II(bpy)_n compared with those of Zn^II(bpy)_n (log β for Cd^II(bpy)_n = 10.3 (n = 3) and 7.7 (n = 2); log β for Zn^II(bpy)_n = 13.2 (n = 3) and 9.5 (n = 2)),²² an isostructural L1-Cd₁₁ was successfully obtained. Ligand L1·H₂O was reacted with 11/6 eq. Cd(OTf)₂·H₂O in a mixed solvent, CDCl₃/CD₃OD/D₂O = 10:10:1 (v/v/v). The reaction mixture was heated at 50 °C for 17 h. L1-Cd₁₁ was isolated by precipitation with Et₂O as a purple solid in 70% yield. In contrast, a mixture of ligand L0·H₂O without methyl substituents of bipyridyl moieties and Cd^II did not afford any isostructural cage complexes (Figure S24). The ¹H NMR spectrum of isolated L1-Cd₁₁ in CDCl₃/CD₃OD/D₂O = 10:10:1 (v/v/v) showed 32 signals in the aromatic region and 4 singlet signals in the alkyl region, suggesting the presence of a C₁ symmetric L1·H₂O ligand in the self-assembled structure (Figure 2c). Each ¹H signal was fully assigned by ¹H-¹H COSY (Figure S16) and ¹H-¹H NOESY spectra (Figures S17 and S18). The interligand NOE patterns of L1-Cd₁₁, (g₁-g′₂, g′₁-g₂, a₂-a₄, f₂-f₄, g₂-f₄, a₃-f₃, d₃-g₃, d₃-g′₃) were identical with those of L1-Zn₁₁. However, in comparison with these two complexes, some proton signals were significantly shifted. For L1-Zn₁₁, the protons of the pyridyl groups (e, f) attached to the Zn^II-porphyrin ring at the meso positions showed downfield shift for f₃ and f₄ (0.28 and 0.19 ppm, respectively) and upfield shift for e₂ and e₄ (0.19 and 0.12 ppm, respectively). Some other signals also shifted, though to lesser extent. However, it is rather difficult to give a general explanation regarding the direction of the signal shifts. This is because there are a few factors, affecting the chemical environment of each proton, such as the different electronic perturbation from Zn^II and Cd^II, and the structural distortion due to the larger-sized Cd^II. The presence of [Cd₁₁L1₆(OTf)_22−n]ⁿ⁺ (n = 5–7) in solution was verified by ESI-MS spectrometry (Figure S21). L1-Cd₁₁ was further characterized by ¹H DOSY (Figure S19), ¹⁹F NMR (Figure S20), and elemental analyses (Supporting Information S14).

The structures of L1-Zn₁₁ and L1-Cd₁₁ were determined by single-crystal XRD analyses (Figure 3).²³ Violet crystals of L1-Zn₁₁ and L1-Cd₁₁ suitable for XRD analysis were obtained by vapor diffusion of Et₂O into a solution of L1-Zn₁₁ or L1-Cd₁₁ in CDCl₃/CD₃OD/D₂O = 10:10:1 (v/v/v), respectively. As shown in Figure 3, both [Zn₁₁(L1·CH₃OH)₆(H₂O)₁₈]²²⁺ and [Cd₁₁(L1·CH₃OH)₆(H₂O)₁₈]²²⁺ moieties contain six ligands L1 and eleven Zn^II or Cd^II ions. For their counter anions, only six TfO⁻ ions were solved and the remaining sixteen TfO⁻ ions were not due to their high disorder. L1-Zn₁₁ has three distinct Zn^II centers, that is, one Zn^II(bpy)₃ (Zn1) and two Zn^II(bpy)₂(H₂O)₂ (Zn2, Zn3). L1-Cd₁₁ has a similar structural framework with three distinct Cd^II centers, one Cd^II(bpy)₃ (Cd1) and two Cd^II(bpy)₂ (H₂O)₂ (Cd2, Cd3). Each racemic crystal with a space group R-3c (No. 167) contains two enantiomers with ΔΔΛΔΔ and ΛΛΔΛΛ configurations at the metal centers bound by bpy as shown in Figure 3. Notably, in both L1-Zn₁₁ and L1-Cd₁₁, one CH₃OH molecule coordinates to each Zn^II-porphyrin center at the axial position from inside the cavity, whereas in the previously reported L0-Zn₁₁, one H₂O molecule coordinates from outside the cavity in the crystal state (Figure 4).

Figure 3. X-ray crystal structures of a) L1-Zn₁₁ and b) L1-Cd₁₁. TfO⁻ ions are omitted for clarity. Zn^II and Cd^II ions are illustrated with the space filling model, Zn^II-porphyrin core skeletons with the bold stick model, and the other atoms with the stick model. Colors are coded according to the CPK coloring. Only a ΔΔΛΔΔ isomer is shown here.

Figure 4. Comparison between L0-Zn₁₁, L1-Zn₁₁, and L1-Cd₁₁ in the crystal structures: X-ray crystal structures (left: side view, right: top view) and the positions of the axial ligands on the Zn^II-porphyrins. a) X-ray crystal structures of L0-Zn₁₁ with six H₂O molecules as the axial ligands bound to the Zn^II-porphyrin from outside. b) X-ray crystal structures of L1-Zn₁₁ and L1-Cd₁₁, each with six CH₃OH molecules as the axial ligands bound to the Zn^II-porphyrin from inside. The core structures with Zn^II porphyrins and the axial ligands are shown in bold stick, and the other moieties in line.

The most attractive common feature of these two cage complexes, L1-Zn₁₁ and L1-Cd₁₁, is the presence of six inner molecular coordination sites in their crystal structure, which could enable the direct binding of included coordinating guest molecules to Zn^II-porphyrin. Indeed, the inner space of each cage complex has two sets of three adjacent CH₃OH molecules bound to each Zn^II-porphyrin center, which are distantly positioned with a distance of 11–12 Å. By the introduction of methyl groups onto the 5′-positions of bpy moieties of methyl-free ligand L0, the space group R32 (#155) for the crystal of L0-Zn₁₁ was changed into the space group R-3c (#167) for those of L1-Zn₁₁ and L1-Cd₁₁. More specifically, one bpy group of an L1-Zn₁₁ (or L1-Cd₁₁) complex gets close to the porphyrin ring of adjacent L1-Zn₁₁ (or L1-Cd₁₁) in the crystal packing, and therefore the coordination of axial CH₃OH ligands from inside should be more favorable with L1-Zn₁₁ and L1-Cd₁₁.

The structural frameworks of L1-Zn₁₁ and L1-Cd₁₁ appear to be maintained in solution as confirmed by NMR and ESI-MS analyses. However, the possibility of flip-out of the axia positions on the square-pyramidal Zn^II-porphyrins of L1-Zn₁₁ and L1-Cd₁₁ in solution remains unclear due to the fast axial ligand exchange in solution.

Figure 5 compares L1-Zn₁₁ and L1-Cd₁₁ (Figure 5a and 5b, respectively) for the distances between two metal centers bound by bpy: for L1-Zn₁₁, Zn1-Zn1 = 27.455 Å, Zn1-Zn2 = 12.331 Å, and Zn2-Zn3 = 11.840 Å and 12.581 Å, and for L1-Cd₁₁, Cd1-Cd1 = 27.745 Å, Cd1-Cd2 = 12.678 Å, and Cd2-Cd3 = 11.867 Å and 12.473 Å.

Figure 5. Structural comparison between L1-Zn₁₁ (a) and L1-Cd₁₁ (b) for the distances (Å) between two metal centers bound by bpy when seen from the C₂ and C₃ axes.

The angles between three metal centers were also calculated; for L1-Zn₁₁, ∠Zn2-Zn3-Zn2 = 64.33°, ∠Zn4-Zn3-Zn4 = 98.31°, and for L1-Cd₁₁, ∠Cd2-Cd3-Cd2 = 61.60°, ∠Cd3-Cd2-Cd3 = 100.07° (Figures S34 and S35). These results indicate that the whole structure of L1-Cd₁₁ is elongated along C₃-axis compared with L1-Zn₁₁ mainly due to the larger ionic radius of Cd^II. Regardless of this structural change, the inner volume of L1-Cd₁₁, 509 ų, was almost the same as that of L1-Zn₁₁, 511 ų (4.0 Å Connolly radius was used as a proof).

Guest encapsulation of L1-Zn₁₁ and L1-Cd₁₁ in CDCl₃/CD₃OD/D₂O = 10:10:1 (v/v/v) was then preliminarily examined with 2,7-dinitro-9-fluorenone as a guest molecule (Figures S25–S30). However, the quantitative analysis was practically difficult due to the low solubility of 2,7-dinitro-9-fluorenone in the solvent.

In conclusion, we have newly synthesized two isostructural M₁₁L₆ complexes, L1-Zn₁₁ and L1-Cd₁₁, from 4-fold-symmetric Zn^II-porphyrin-centered tetrakis-meso-(5′-methyl-2,2′-bipyridyl) ligands (L1) and metal ions. The self-assembled cage structures were characterized by NMR, ESI-MS, and XRD analyses. A common feature of these isostructural L1-Zn₁₁ and L1-Cd₁₁ complexes is the presence of their inner molecular coordination sites on the Zn^II-porphyrin centers in the nano-sized cage structures. Such an approach would develop metallo-porphyrins that work as catalysts in a confined space.