Integrative Assembly of Heteroleptic Tetrahedra Controlled by Backbone Steric Bulk

A bent fluorenone-based dipyridyl ligand LA reacts with PdII cations to a solvent-dependent dynamic library of [PdnL2n] assemblies, constituted by a [Pd3LA6] ring and a [Pd4LA8] tetrahedron as major components, and a [Pd6LA12] octahedron as minor component. Introduction of backbone steric hindrance in ligand LB allows exclusive formation of the [Pd6LB12] octahedron. Combining equimolar amounts of both ligands results in integrative self-sorting to give an unprecedented [Pd4LA4LB4] heteroleptic tetrahedron. Key to the non-statistical assembly outcome is exploiting the structural peculiarity of the [Pd4L8] tetrahedral topology, where the four lean ligands occupy two doubly bridged edges and the bulky ligands span the four remaining, singly bridged edges. Hence, the system finds a compromise between the entropic drive to form an assembly smaller than the octahedron and the enthalpic prohibition of pairing two bulky ligands on the same edge of the triangular ring. The emission of luminescent LA is maintained in both homoleptic [Pd3LA6] and heteroleptic [Pd4LA4LB4].

C oordination-driven self-assembly provides a powerful tool to design and synthesize discrete nanostructured objects with accessible cavities. 1,2 The resulting metallo-supramolecular assemblies are promising candidates for mimicking functional host systems found in nature, such as enzymes. The dynamic nature of many transition metal−ligand interactions, characterized by precise geometry and directionality, combined with a polytopic ligand structure, allow us to design and self-assemble a plethora of compounds with different shapes, sizes, and properties. Embedded functions, depending on either individual building blocks or their synergistic interaction, 3−7 may involve host−guest interactions, 8−10 photoswitching, 11,12 chirality, 13−15 chromophore effects, 16−19 or catalysis, 20−23 just to name a few.
Besides the formation of single components, dynamic systems consisting of several structures with different topologies may be the result of a self-assembly reaction. 24−29 So far, most reported metallo-supramolecular compounds carry only one type of ligand, limiting the possibility to exploit applications arising from the implementation of multiple functionalities. To overcome this restriction, we propose to increase structural complexity via the non-statistical integration of a set of different ligands. A first step in this direction is represented by homoleptic assemblies where the same ligand occupies two or more non-identical positions. For example, Luẗzen reported a [Pd 2 L 4 ]@[Pd 4 L 8 ] cage-in-ring assembly. 30 Shionoya differentiated metal positions, thus desymmetrizing a porphyrin ligand. 31 Our group investigated the controlled formation of [Pd 2 L 3 X 2 ] bowls (X = solvent, halides) featuring two different ligand environments. 32,33 Recently, structural complexity has been increased using non-symmetric ligands. 34−37 A further approach relies on the structural diversity of [Pd 4 L 8 ] assemblies with bis-pyridyl ligands, making it possible to form rings, 38 interpenetrated double cages, 39,40 or a tetrahedron-like arrangement, featuring four edges composed of a single ligand and two doubly bridged edges. 5,41−43 Complexity further increases when chemically different ligands are placed in defined positions, yielding heteroleptic species. To overcome the formation of a statistical mixture, 44 several strategies have been applied, e.g., exploiting hydrogenbonding, 45 templating guests, 46 shape complementarity, 47−53 or covalent bridges between ligands. 54 Herein, we report a system where a bis-monodentate, flat ligand L A self-assembles with Pd II to give a series of [Pd n L 2n ] (n = 3, 4, 6) architectures in a solvent-dependent process. Introduction of steric congestion into its backbone gives the bulky ligand L B , allowing us to exclusively form a large [Pd 6 L 12 ] octahedron. A similar approach was reported by Severin and Hiraoka based on clathrochelate metallo-ligands. 55,56 We now show that combining lean ligand L A and bulky derivative L B opens a new strategy to form unprecedented [Pd 4 L A 4 L B 4 ] heteroleptic structures. Key to clean, integrative self-sorting is the presence of two non-equivalent edge types in the [Pd 4 L 8 ] tetrahedron, combined with control over steric pressure in the ligand backbones.
Ligands L A and L B were synthesized by Suzuki crosscoupling starting from 2,7-dibromo-9-fluorenone and 2,7dibromo-9,9-dihexylfluorene, respectively, with 3-pyridineboronic acid pinacol ester (Supporting Information (SI)). Using 9-substituted fluorene-based backbones makes it possible to obtain non-linear bis-pyridyl ligands, bearing the CO or alkyl substituents pointing toward one side of the molecule. This generates two binding modes: convex (θ ≈ 90°) with nitrogen donors pointing away from the substituent(s), and concave (θ ≈ 40°) pointing in the same direction (Scheme S4). A similar backbone design was reported to lead to Febased helicates and tetrahedra, 57−60 as well as knots and Borromean rings. 61 At first, we studied the self-assembly of homoleptic species. Combination of ligand L A with Pd II in a 2:1 ratio led to a solvent-dependent dynamic library of compounds with different nuclearity (Figure 1a). In CD 3 CN, two major components are formed, a [Pd 3 L A 6 ] triangular ring and a [Pd 4 L A 8 ] tetrahedron. After coordination to Pd II , 1 H NMR signals are downfield shifted and split into three sets, as clearly observed for proton H IV , with a 1:1:2 ratio (Figure 1b). NOESY-NMR allows us to identify two independent sets of signals (SI), while DOSY-NMR shows the presence of two species in solution, with hydrodynamic radii of 11.04 and 12.19 Å, respectively (SI). This is consistent with the formation of a [Pd 3    3.07 Å, suggesting the incompatibility of a bulkier backbone substituent with this structure. Diffusion of ethyl acetate into a DMF solution yielded single crystals of [Pd 4 L A 8 ], yielding a tetrahedral structure with Pd II centers on the vertices and ligands bridging the edges (Figure 2b). In this case, L A occupies two non-equivalent positions: four edges ("edge 1") are composed of one ligand, while the two remaining edges ("edge 2") accommodate a pair of ligands. In edge 1, the carbonyl substituent points outside the tetrahedral cavity, with L A adopting a convex binding mode (θ ≈ 80°). In edge 2, the carbonyl group of one L A points toward the π-surface of its neighbor (CO−π-C 5 centroid = 3.53 Å), while the other carbonyl group points inside the cavity, featuring a concave binding mode (θ ≈ 40°). Finally, diffusion of 1,4-dioxane into the [Pd n L A 2n ] (n = 3, 4, 6) CH 3 CN solution resulted in single crystals of [Pd 6 L A 12 ], suitable for synchrotron diffraction analysis. The compound crystallizes in the R3̅ space group as a pair of enantiomeric [Pd 6 L A 12 ] octahedra (Figure 2c). Pd II cations occupy the vertices, while the edges feature one ligand L A each, with all carbonyl groups pointing outside the cavity. Comparing the three [Pd n L A 2n ] structures suggests that only the octahedron, the sole structure without CO−π interactions, should be able to accommodate sterically demanding ligands on all edges.
To explore this possibility, L B was synthesized by replacing the 9-fluorenone with a 9,9-dihexylfluorene backbone. Selfassembly of L B with Pd II cations in a 2:1 stoichiometry led to the formation of a single species, identified as a [Pd 6 L B 12 ] octahedron (Figure 3a). Upon complexation of Pd II , the 1 H NMR signals of L B are downfield shifted and slightly broadened (Figure 3b). DOSY analysis confirmed a single species in CD 3 CN (r H = 15.22 Å) and DMSO-d 6 (r H = 18.66 Å, SI), while pointing to differences in the solvent shell dynamics around the heavily alkyl-decorated [Pd 6 L B 12 ] species. The high nuclearity was confirmed by HR-ESI-MS, where a series of peaks for [Pd 6 L B 12 (BF 4 ) n ] m+ (n = 3−6; m = 9−6) were identified (Figure 3c). Moreover, single crystals were obtained from vapor diffusion of toluene into a DMSO solution. The compound crystallizes as an octahedron with a structure analogous to that of [Pd 6 (Figure 4a). To the best of our knowledge, this is the first example of such a [M 4 L A 4 L B 4 ] heteroleptic assembly topology. As postulated, four ligands L A are accommodated on edges 2, with the CO group either pointing inside the cavity or facing the π-surface of neighboring L A (CO−π-C 5 centroid = 3.38 Å, Figure 4a). In addition, four ligands L B are sitting on edges 1, adopting a convex binding mode with hexyl chains pointing outside the cavity (Figure 4a, purple backbone).
Next, we investigated guest binding of one aliphatic and two aromatic bis-sulfonates (Scheme S5) with [Pd 3  Finally, we investigated the photophysical properties of the systems ( Figure 5). In DMSO, L A shows an emission band centered at 555 nm that is blue-shifted to 532 nm upon Pd II complexation in either homoleptic [Pd 3 L A 6 ] or heteroleptic [Pd 4 L A 4 L B 4 ]. While this indicates that the emissive properties of ligand L A are retained in the heteroleptic tetrahedron, our platform allows to introduce additional functionality through modification of L B . It is worth noting that in Pd-mediated assemblies luminescence quenching is frequently observed, 15,39 and only few examples of emissive cages have been reported so far. 18,62−65 To conclude, we report a new strategy for the non-statistical, integrative self-assembly of a previously unreported [M 4 L A 4 L B 4 ] heteroleptic cage topology. Key factors are the use of bis-monodentate ligands, able to adopt a concave or convex binding mode, the precise introduction of backbone steric hindrance, and a balance between the entropic tendency to form small assemblies and the enthalpic disadvantage to pair bulky substituents on a single edge. The preservation of ligand emission properties in the Pd-mediated assemblies opens potential toward application as multifunctional devices and materials in fields such chiroptical sensing, donor−acceptor systems, and photoredox catalysis.