Coordination-driven self-assembly of a molecular figure-eight knot and other topologically complex architectures

Over the past decades, molecular knots and links have captivated the chemical community due to their promising mimicry properties in molecular machines and biomolecules and are being realized with increasing frequency with small molecules. Herein, we describe how to utilize stacking interactions and hydrogen-bonding patterns to form trefoil knots, figure-eight knots and [2]catenanes. A transformation can occur between the unique trefoil knot and its isomeric boat-shaped tetranuclear macrocycle by the complementary concentration effect. Remarkably, the realization and authentication of the molecular figure-eight knot with four crossings fills the blank about 41 knot in knot tables. The [2]catenane topology is obtained because the selective naphthalenediimide (NDI)-based ligand, which can engender favorable aromatic donor-acceptor π interactions due to its planar, electron-deficient aromatic surface. The stacking interactions and hydrogen-bond interactions play important roles in these self-assembly processes. The advantages provide an avenue for the generation of structurally and topologically complex supramolecular architectures.

In addition, the 1 H NMR signals at 2.11 ppm and 1.87 ppm also have a coupling interaction, which belong to benzyl protons. Large upfield shifts of phenyl and benzyl protons indicate tight π-π stacking between phenyl and pyridyl groups, which contributes to the formation of Figure

Geometries:
All geometries presented here were optimized with long-range dispersion correction, i.e. using the PBE-D method.

Supplementary Tables: X-ray crystallography details
Single crystals of 1, 2a, 2b, 3, 3′, 4 and 5, suitable for X-ray diffraction study were obtained at room temperature. X-ray intensity data of 1 and 2b were collected at 203 K, data of 2a was collected at 193 K and data of 3, 3′ and 4 were collected at 173 K on a CCD-Bruker SMART APEX system. In these data, the disordered solvent molecules which could not be restrained properly were removed using the SQUEEZE route.
In asymmetric unit of 1, C44 and O12 were refined isotropically and other non-hydrogen atoms were refined anisotropically. 16 ISOR, 1 DELU and 1 DFIX instructions were used to restrain anions and solvents so that there were 98 restraints in the data. Hydrogen of one methanol molecule could not be found and others were put in calculated positions.
In asymmetric unit of 2a, there were disordered anions and solvents (three triflate anion and four methanol molecules) which could not be restrained properly. Therefore, SQUEEZE algorithm was used to omit them.
One pentamethylcyclopentadienyl ligand (Cp* for short) was disordered and it was divided into two parts (50:50). O30 and O31 were refined isotropically and other non-hydrogen atoms were refined anisotropically. 42 ISOR and 13 DFIX instructions were used to restrain anions, solvents and Cp* fragments so that there were 265 restraints in the data. Hydrogen of methanol molecules could not be found and others were put in calculated positions.
In asymmetric unit of 2b, there were disordered anion and solvents (one triflate anion, four methanol and five water molecules) which could not be restrained properly. Therefore, SQUEEZE algorithm was used to omit them. One metalla-edge (Rh3 and corresponding Cp* fragment) was disordered and it was divided into two parts (44:56). F4, O11 and F5 were refined isotropically and other non-hydrogen atoms were refined anisotropically. 51 ISOR, 3 SIMU, 10 DELU and 26 DFIX instructions were used to restrain anions, ligands and Cp* fragments so that there were 488 restraints in the data.
In asymmetric unit of 3, there were disordered solvent molecules (six methanol and five water molecules) which could not be restrained properly. Therefore, SQUEEZE algorithm was used to omit them. 6 ISOR and 2 DFIX instructions were used to restrain anions and solvents so that there were 38 restraints in the data.
Hydrogen of methanol molecules could not be found and others were put in calculated positions.
In asymmetric unit of 3′, there were disordered anions and solvents (two triflate anions, half of a diisopropyl ether and half of a methanol molecules) which could not be restrained properly. Therefore, SQUEEZE algorithm was used to omit them. 21 ISOR, 3 DELU and 4 DFIX instructions were used to restrain ligand and Cp* fragments so that there were 136 restraints in the data.
In asymmetric unit of 4, there were disordered solvents (two and a half methanol molecules) which could not be restrained properly. Therefore, SQUEEZE algorithm was used to omit them. One triflate anion and one diethyl ether molecule were disordered and they were divided into two parts (65:35 for anion and 56:44 for Et2O). 20 ISOR and 9 DFIX instructions were used to restrain anions, solvent molecule and Cp* fragments so that there were 129 restraints in the data.
In asymmetric unit of 5, there were one disordered triflate anion which could not be restrained properly.
Therefore, SQUEEZE algorithm was used to omit them. Hydrogen of methanol and water molecules could not be found and others were put in calculated positions.