The codriven assembly of molecular metalla-links ($6_1^3$, $6_2^3$) and metalla-knots ($4_1$, $3_1$) via coordination and noncovalent interactions

Significance

The appeal of topologically nontrivial mechanically interlocked molecules (MIMs) exists not only in their fascinating structures but also in their potential applications as functional molecular machines and smart materials. However, the synthesis of these materials faces limited assembly methodologies, thus significantly hindering further structural diversity and complexity through fundamental chemical principles. In this study, a facile assembly strategy based on the synergistic effect of coordination and noncovalent interactions was proposed and successfully applied to construct four types of MIMs ($6_1^3$, $6_2^3$ metalla-links, $4_1$, $3_1$ metalla-knots). Our work not only proposes an efficient strategy for the systematic preparation of complex interlocked topologies but also showcases the potential of noncovalent interactions, offering inspiration for future supramolecular self-assembly design strategies.

Abstract

Although mechanically interlocked molecules (MIMs) display unique properties and functions associated with their intricate connectivity, limited assembly strategies are available for their synthesis. Herein, we presented a synergistic assembly strategy based on coordination and noncovalent interactions (π–π stacking and CH⋯π interactions) to selectively synthesize molecular closed three-link chains ($6_1^3$ links), highly entangled figure-eight knots ($4_1$ knots), trefoil knot ($3_1$ knot), and Borromean ring ($6_2^3$ link). $6_1^3$ links can be created by the strategic assembly of nonlinear multicurved ligands incorporating a furan or phenyl group with the long binuclear half-sandwich organometallic Cp*Rh^III (Cp* = η⁵-pentamethylcyclopentadienyl) clip. However, utilizing much shorter binuclear Cp*Rh^III units for union with the 2,6-naphthyl-containing ligand led to a $4_1$ knot because of the increased π–π stacking interactions between four consecutive stacked layers and CH⋯π interactions. Weakening such π–π stacking interactions resulted in a $3_1$ knot. The universality of this synergistic assembly strategy for building $4_1$ knots was verified by utilizing a 1,5-naphthyl-containing ligand. Quantitative conversion between the $4_1$ knot and the simple macrocycle species was accomplished by adjusting the concentrations monitored by NMR spectroscopy and electrospray ionization mass spectrometry (ESI-MS). Furthermore, increasing the stiff π-conjugated area of the binuclear unit afforded molecular Borromean ring, and this topology is a topological isomer of the $6_1^3$ link. These artificial metalla-links and metalla-knots were confirmed by single-crystal X-ray diffraction, NMR and ESI-MS. The results offer a potent strategy for building higher-order MIMs and emphasize the critical role that noncovalent interactions play in creating sophisticated topologies.

Since the discovery of mechanically interlocked molecules (MIMs) in proteins (1, 2) and DNA (3, 4) that play crucial biological roles, the creation of wholly synthetic analogs of these molecules to comprehend and harness their intrinsic properties and functions has inspired scientists for several decades. Following the first controlled synthesis of a Hopf link (5) by Sauvage et al. in 1983 using a Cu(I)-template approach, and subsequently the first practical synthesis of a nontrivial molecular trefoil knot (6) (the simplest prime $3_1$ knot, a single ring with three crossings) in 1989, various MIMs featuring tanglesome topologies took off like a rocket (7–10). The construction of increasingly complicated, topologically nontrivial MIMs has become attractive targets for synthetic chemists. To date, considerable efforts have been made to create higher-order catenane molecules (with >2 rings) (11–14) and orderly entangled knotted molecules (with a minimum number of crossing points >3) (15–17), to more closely approach structurally complex biological macromolecules at the molecular level.

Closed three-link chains ($6_1^3$ links) (Fig. 1A) are a prime link of [3]catenane with six crossings among molecular links, and to date, this topology has only been realized by Komiyama et al. (18) Chi et al. (19) and our group (20). Despite the accessibility of such topologies, their synthesis from three rings is extremely difficult because alternative topological isomers, including Borromean rings ($6_2^3$ links) (21–24) (Fig. 1B), cyclic [3]catenanes ($6_3^3$ links) (25–27) (Fig. 1C), linear [3]catenanes ($4_1^3$ links) (28–31) (Fig. 1D), and other topologies, are equally likely to arise during the self-assembly process. Among the knotted molecules, unlike the fastest-growing trefoil knots (Fig. 1E), the realization of figure-eight knots ($4_1$ knots, a single strand with four crossings) (32–35) (Fig. 1F) is exceedingly tough. The first possible synthesis of $4_1$ knots was accomplished by Sanders et al. in 2014, the structure of which was determined primarily by symmetry and NMR data (32). Until 2019, only our group obtained $4_1$ knots with well-defined crystallographic structures despite their serendipitous synthesis (33). Therefore, the artwork of constructing such topologically nontrivial catenanes and knots is a formidable task for synthetic chemists because of the very limited assembly methodology.

Fig. 1.

Topological structures of molecular links and knots. (A) Closed three-link chain ($6_1^3$ link). (B) Borromean ring ($6_2^3$ link). (C) Cyclic [3]catenane ($6_3^3$ link). (D) Linear [3]catenane ($4_1^3$ link). (E) Trefoil knot ($3_1$ knot). (F) Figure-eight knot ($4_1$ knot).

The utilization of coordination-driven self-assembly is efficacious in designing and creating various topologies (36–39). However, due to the mechanical constraints on the relative positions of the constituent components and the difficulty of predicting the extent of weak interactions, higher-order MIMs cannot be built exclusively through coordination interactions (8). Noncovalent interactions, a weak force whose synergistic impact with the coordination interactions allows components to be incorporated and modified and accelerates intra- or inter-ring interactions, are essential for generating nontrivial supramolecular topologies that are troublesome to achieve by previous synthetic methods (19, 40). Therefore, the combination of noncovalent and coordination interactions appears to be a promising strategy for constructing supramolecular topologies with increasing complexity (41–43). So far, the cooperative noncovalent interactions emerging from self-assembling units, such as π-stackings, H-bonds, anion–π interactions, and Ag⋯O═C interactions, have been used to construct discrete coordination assemblies, but their scope for building intricate supramolecular topologies is still limited (41–45). Hence, constructing complicated MIMs through the cooperative interplay of noncovalent and coordination interactions remains an intriguing challenge.

Herein, utilizing a coordination and noncovalent interaction (π─π stacking and CH⋯π interactions) codriven assembly strategy, four types of MIMs, molecular closed three-link chain ($6_1^3$ metalla-link), figure-eight knot ($4_1$ metalla-knot), trefoil knot ($3_1$ knot), and Borromean ring ($6_2^3$ metalla-link), were selectively constructed. Our approach involved the strategic use of a series of nonlinear multicurved dipyridine ligands (44, 46–48) that not only make spontaneous conformational adjustments by rotating pyridyl arms but also potentially engender favorable noncovalent interactions through the π-conjugated plane, thus allowing for the subsequent mechanical interlocking of specific positions to form these metalla-links and metalla-knots. As shown in Fig. 2, the coordination-driven self-assembly of the ligands L^1/2 and the binuclear half-sandwich organometallic Cp*Rh^III (Cp* = η⁵-pentamethylcyclopentadienyl) clip E1 afforded two $6_1^3$ metalla-links in high yields. In contrast to E1, the combination of the much shorter binuclear unit E2 with the ligand L³, which introduced a 2,6-naphthyl group, considerably strengthened the π–π stacking interactions between the consecutive stacked layers and CH⋯π interactions, leading to the highly entangled $4_1$ metalla-knot. The self-assembly of L³ with E3 weakened the abovementioned noncovalent interactions, thus resulting in a $3_1$ knot instead of a $4_1$ knot despite the use of a similar synthetic route. Moreover, the universality of the formation of the $4_1$ metalla-knot was verified by the assembly of the 1,5-naphthyl-containing ligand L⁴ with E4. The combination of ligand L⁴ with E5, which bears a large, rigid conjugated plane, allows the formation of molecular Borromean ring. These results, in turn, highlight the importance of noncovalent interactions in the coordination-driven assembly of MIMs.

Fig. 2.

Synthetic route to molecular closed three-link chains ($6_1^3$ metalla-links) 1 and 2, figure-eight knots ($4_1$ metalla-knots) 3 and 5, trefoil knot ($3_1$ metalla-knot) 4, and Borromean ring ($6_2^3$ metalla-link) 6.

Results and Discussion

Self-Assembly of Molecular Closed Three-Link Chains (${\mathbf{6}}_{\mathbf{1}}^{\mathbf{3}}$ Metalla-Links) 1 and 2.

As depicted in Fig. 2, yellow crystalline complex 1 was obtained in a high yield of 92% by the coordination-driven assembly of ligand L¹ bearing a furan group and binuclear half-sandwich organometallic unit E1 (Rh–Rh distance of 12.9 Å) in methanol (MeOH) followed by slow diffusion of isopropyl ether into the above solution for crystallization.

Single-crystal X-ray diffraction (SCXRD) analysis revealed complex 1 crystallized in the trigonal space group P$\overline{3}$1c and featured a closed three-link chain structure (Fig. 3 A–C) interlaced with three equivalent but deformed tetranuclear monocycles. Each independent monocycle is represented in a distinct color. The presence of any two of the three rings constituting 1 that latched together from the beginning of one ring to the end of another ring resulted in the topology of the head-to-tail cyclic [3]catenane with six crossing points (Fig. 3D, where n represents the number of minimum crossing points), denoted by the $6_1^3$ link according to the Alexander−Briggs notation (49). Consequently, the other two rings remain interlocked if any one ring is broken. Furthermore, neither of the two organic linkers L¹ on each ring exhibited an M-shaped conformation. Rather, a distinct conformation akin to a water spoon (Fig. 3E) was produced because the C─C single bond attached to the central furan ring allowed the 4-pyridinophenyl arm on linker L¹ to rotate. In addition, linker L¹ established inter-ring face-to-face π–π stacking interactions with E1 in the range of 3.33 to 3.44 Å (Fig. 3F, red dotted lines). The inter-ring edge-to-face CH⋯π interactions between the protons on the benzene and furan rings of the L¹ linker and the pyridine rings of two lateral L¹ linkers were also observed, with primary distances ranging from 2.76 to 3.16 Å (Fig. 3F, green dotted lines). Therefore, the accumulation of multiple inter-ring π–π stacking and CH⋯π interactions may have motivated this unique form of ring interlocking in a closed three-link chain, stabilizing the structure of 1.

Fig. 3.

Single-crystal X-ray structure of $6_1^3$ metalla-link 1. (A) Structure of 1 viewed along the c axis and (B) viewed along the b axis. (C) View of the space-filling representation of 1. (D) Reduced representation of 1 and the $6_1^3$ link topology. (E) A single macrocycle component constituting 1. (F) Inter-ring noncovalent interactions in 1. All triflate anions and solvent molecules were omitted for clarity. Color code: C, gray; N, navy blue; O, red; H, magenta; Rh, turquoise.

The molecular structure of 1, as revealed by a crystallographic study, was also confirmed in solution by NMR spectroscopy and electrospray ionization mass spectrometry (ESI-MS) analyses. All ¹H NMR signal assignments in 1 were validated by ¹H–¹H correlated spectroscopy (SI Appendix, Figs. S9 and S10). The consistent diffusion coefficient D (3.08 × 10⁻⁶ cm² s⁻¹) (Fig. 4A) determined by diffusion-ordered spectroscopy (DOSY) for aromatic and aliphatic signals (excluding solvent signals) revealed that these signals originated from a single species of $6_1^3$ metalla-link 1. The ESI-MS spectra in acetonitrile (MeCN) show six prominent peaks at m/z 949.7172, 1,224.3843, 1,420.5751, 1,682.1621, 2,048.3842, and 2,597.7151, which can be assigned to the intact species with the formulas [1–10 OTf⁻]¹⁰⁺ (calc. 949.7136), [1−8 OTf⁻]⁸⁺ (calc. 1,224.3801), [1−7 OTf⁻]⁷⁺ (calc. 1,420.5704), [1−6 OTf⁻]⁶⁺ (calc. 1,682.1576), [1−5 OTf⁻]⁵⁺ (calc. 2,048.3796), [1−4 OTf⁻]⁴⁺ (calc. 2,597.7126), respectively, in perfect accordance with their theoretical isotope distribution patterns (Fig. 4B and SI Appendix, Figs. S12–S17). The above results indicated that 1 has excellent stability in MeCN, and the solution structure mirrored the solid-state structure.

Fig. 4.

NMR spectroscopy and ESI-MS analyses of $6_1^3$ metalla-link 1. (A) ¹H DOSY NMR spectrum (400 MHz, 298 K, MeCN-d₃, ppm) of 1. (B) ESI-MS spectra of 1 in positive mode. Inset: the simulated (red) and experimental (blue) isotope distribution patterns of 1.

To verify the universality of this synergistic assembly strategy for constructing $6_1^3$ metalla-links utilizing nonlinear multicurved dipyridyl ligands, ligand L² with three phenyl groups was synthesized. As anticipated, closed three-link chain 2 (SI Appendix, Fig. S18), which is isomorphic to 1, was formed via self-assembly with E1 utilizing ligand L². Likewise, 2 comprises three individual but mechanically interlocking tetranuclear macrocycles (SI Appendix, Fig. S18E) in a head-to-tail arrangement. The cationic skeleton of 2 is also held together by rich π–π interactions between stacked rings, and CH⋯π interactions between the protons of the phenyl rings and pyridine rings (SI Appendix, Fig. S18F), indicating the essence of inter-ring noncovalent interactions responsible for the generation of $6_1^3$ metalla-link topology and its structural stability.

Self-Assembly of Molecular Figure-Eight Knot ($4_1$ Metalla-Knot) 3 and Trefoil Knot (${\mathbf{3}}_{\mathbf{1}}$ Metalla-Knot) 4.

The accomplishment of $6_1^3$ metalla-links utilizing coordination-driven self-assembly and synergistic noncovalent interactions prompted us to further investigate the feasibility of creating knotted architectures. Reinforcing the noncovalent interactions of the constituents, particularly the π–π stacking interactions between consecutive stacked layers, may enable the constituent parts to pack closer together in spatial arrangement, therefore linking the constituent parts might generate highly intertwined, topologically complicated knotted assemblies.

To strengthen noncovalent interactions between interlayers, we introduced a large π-conjugated 2,6-naphthyl group, which can potentially form multiple aromatic stacking contacts and other interactions, into ligand L³. Moreover, the binuclear unit E2 with a nonbonding Rh–Rh length of 8.4 Å, which is much shorter than that of E1, was purposefully chosen. By the direct assembly of E2 with ligand L³ and the subsequent slow diffusion of isopropyl ether into the solution, dark green crystals of 3 (Fig. 5A) were obtained with 88% yield.

Fig. 5.

Single-crystal X-ray structure of $4_1$ knot 3. (A) Chemical structural representation of 3. (B) Cationic structure of 3 viewed along the b axis, exhibiting its high symmetry with a rotary inversion axis of S₄. (C) Reduced representation of 3 and the $4_1$ knot topology. (D) Cationic structure of 3 viewed along the c axis. Partial structures highlighting (E) the π–π stacking interactions between four stacked layers and (F) CH⋯π interactions in 3. Color code: H, white. All triflate anions and solvent molecules were omitted for clarity.

SCXRD analysis revealed that the structure of the resulting 3 crystallizing in the monoclinic space group C2/c is a molecular figure-eight knot (Fig. 5 B–D) comprising a single molecular strand, denoted as the $4_1$ knot topology in the Alexander−Briggs notation. In the solid-state molecular structure of 3, four L³ linkers and four binuclear E2 units are alternately connected to form an octanuclear complex in the form of a closed loop. Viewing the structure of 3 along the b axis (Fig. 5B), upon first inspection, $4_1$ knot 3 seems to have eight crossing points in total, four of which are nugatory, indicating that Reidemeister moves (8, 50) could be executed on the loop. In other words, the loop could be moved past other loops to eliminate those nugatory crossing points without rupturing and rejoining the strand, assisting itself in converting into a totally different conformation without altering its topology. Through such moves, each knot can be deformed into a configuration that has the minimal number of crossing points. Herein, the operations of Reidemeister moves involve $4_1$ knot 3, which transforms from its initial eight-crossing conformation to the simplest representation of four crossing points (Fig. 5C). Intriguingly, the addition of four nugatory crossing points enables 3 to adopt an achiral conformation with an S₄ rotary inversion axis (Fig. 5B). The structure of 3 shows strong quadruple π–π stacking interactions (3.46 to 3.57 Å) between the pyridine and the benzene rings of four L³ linkers (Fig. 5E). In addition, three distinct sets of edge-to-face CH⋯π interactions between the protons on the benzene rings of L³ and the 1,4-naphthoquinone moieties of E2 (2.77 to 3.27 Å), the pyridine and the 2,6-naphthyl group (2.90 Å), and two 2,6-naphthalene moieties (2.65 and 2.75 Å) (Fig. 5F) were clearly observed. Close-contact structural analysis demonstrated that the highly interwoven topology of the $4_1$ knot was caused by the elongated binuclear building blocks and increased π-conjugated areas on nonlinear multicurved ligands, which enhanced interlayered aromatic stacking and CH⋯π contacts. Most likely, the crucial noncovalent interactions propel such complex molecular entanglements.

The composition of $4_1$ metalla-knot 3 was also confirmed using ESI-MS and NMR analyses. An evident signal at m/z 1,713.5695 was attributed to the entire species, with the formula [3−3OTf⁻]³⁺ (calc. 1,713.5689) (SI Appendix, Fig. S27). Further results from the ¹H DOSY NMR study in MeOH-d₄ showed the presence of a single species in solution, with a single D value of 2.20 × 10⁻⁶ cm² s⁻¹ (SI Appendix, Fig. S30). The above experimental findings revealed that the solution structure of assembly 3 perfectly matched its solid-state structure.

A trefoil knot is the simplest nontrivial knot that can conceptually be formed by cutting one of the crossing points of a figure-eight knot and then connecting the cleaved moieties (Fig. 6A). Pursuant to this concept, we speculate that lengthening binuclear half-sandwich organometallic clips could effectively diminish π–π stacking interactions between successively stacked interlayers, which would lower the number of crossings and ultimately reduce structural entanglement. As envisioned, coordination-driven self-assembly of the linker L³ and the binuclear building blocks E3 afforded the target molecular trefoil knot 4, which comprises three binuclear E3 units and three L³ linkers linked alternately in a single loop (Fig. 6 B and C). All of the E3 units in 4 displayed trans configurations (Fig. 6D), with a Rh–Rh nonbonding distance of 11.8 Å, which is much longer than that of E2. The results indicated that the topologies of the $4_1$ and $3_1$ knots could be fine-tuned by simply adjusting the length of the binuclear building blocks.

Fig. 6.

Formation process of the $3_1$ knot and single-crystal X-ray structure of $3_1$ knot 4. (A) Schematic diagram of the formation process of a $3_1$ knot from a $4_1$ knot by cutting a crossing point and reconnecting the cleaved knot. (B) Cationic structure of 4 and its reduced representation of the $3_1$ knot topology. (C) View of the space-filling model representation of 4. (D) Structure of the trans-E3 unit in 4. Color code: C, gray; N, navy blue; O, red; Rh, turquoise. All triflate anions, solvent molecules, and H atoms were omitted for clarity.

Self-Assembly of Molecular Figure-Eight Knot (${\mathbf{4}}_{\mathbf{1}}$ Metalla-Knot) 5 and Borromean Ring (${\mathbf{6}}_{\mathbf{2}}^{\mathbf{3}}$ Metalla-Link) 6.

The assembly process from ligand L³ to intricate $4_1$ knot 3 prompted us to further investigate the universality of multicurved ligands for the construction of molecular figure-eight knots. Hence, ligand L⁴, which contains a 1,5-naphthyl group and is an isomer of ligand L³, was synthesized. A synthetic procedure similar to that used for the $4_1$ knot 3 generated the navy-blue crystalline complex 5 in 85% yield using the ligand L⁴ and the binuclear unit E4 of the same length as E2.

SCXRD analysis confirmed the molecular figure-eight knot ($4_1$ knot) structure of complex 5 (Fig. 7 A and B) in the solid state, which crystallized in the tetragonal space group I4₁/a. Similar noncovalent interactions involving multiple aromatic stacking (SI Appendix, Fig. S39) and CH⋯π contacts (SI Appendix, Fig. S40) were also observed for 5. Furthermore, the solution structure, as observed by NMR spectroscopy and ESI-MS analyses (SI Appendix, Figs. S41–S44), reflected the solid-state structure. The above results validated the real possibility of artificially synthesizing entangled $4_1$ metalla-knots using coordination-driven self-assembly and cooperative intralock noncovalent interactions from multicurved dipyridine donors and binuclear acceptors of the appropriate length.

Fig. 7.

Single-crystal X-ray structure of $4_1$ knot 5 and the structural transformation between 5 and the corresponding metalla-macrocycle 5a, as monitored by NMR and ESI-MS. (A) Cationic structure of 5. (B) Reduced representation of 5 and the $4_1$ knot topology. (C) ¹H NMR (400 MHz, CD₃OD, ppm) spectrum of 5 (blue triangles), showing the transformation of the corresponding metalla-macrocycle 5a (green circles) with decreasing concentration from 6.7 mM to 0.25 mM. (D) Mass spectra peaks of 5 and (E) 5a.

Remarkably, the quantitative conversion of $4_1$ metalla-knot 5 to simple metalla-macrocycle 5a was well observed using concentration dilution experiment of 5 in MeOH-d₄ (Fig. 7C and SI Appendix, Fig. S45). As illustrated in Fig. 7C, when the concentration decreased from 6.7 mM to 0.25 mM, the intensity of the initial proton signals of 5 at 9.46, 9.40, 9.20, 8.96, 8.76, 8.54, 8.32, 8.13, 1.87, and 1.60 ppm gradually decreased, accompanied by the appearance of a set of new, simple proton signals at 9.25, 9.08, 8.62, 8.30,7.90, 7.78, 7.67, and 1.80 ppm attributable to the corresponding metalla-macrocycle species 5a (SI Appendix, Fig. S46). Moreover, the coexistence of 5 and 5a in MeOH was also confirmed by the ESI-MS spectrum, in which the peaks at m/z 1,846.9692 and 848.8444 could be assigned to [5−3 OTf⁻]³⁺ (calc. 11,846.9739) (Fig. 7D) and [5a−2 OTf⁻]²⁺ (calc. 8,848.8437) (Fig. 7E), respectively, with the correct theoretical isotope distribution patterns. When the solution concentration decreased to 0.25 mM, the original proton resonances of 5 were completely eliminated, which indicated that the transformation into 5a was almost quantitative. The above results followed Le Chatelier’s principle that concentration changes accelerate the structural transformation between a complex figure-eight knot and the corresponding macrocycle.

Inspired by the topologically structural transition from $4_1$ knot 3 to $3_1$ knot 4 by prolonging binuclear building blocks, choosing more rigid and large conjugated anthraquinone-containing binuclear units might conceivably foster the formation of distinct topologies when using the same ligand, L⁴. Therefore, ligand L⁴ and binuclear unit E5 were assembled, forming the molecular Borromean ring 6 ($6_2^3$ metalla-link) (Fig. 8A). Three mechanically interpenetrating but chemically identical tetranuclear macrocycles (Fig. 8F), with an average Rh–Rh length and width of 22.2 and 8.4 Å, respectively, that constitute the Borromean structure of 6, as presented in Fig. 8 B–E, have been validated by SCXRD analysis. This type of $6_2^3$ link is a topology in which the rings are interspersed such that disrupting any one of them causes all of the other rings to unlink. In the cationic structure of 6, a pair of linkers in each ring is arrayed in parallel, although both linkers displayed severely deformed conformations. The successful synthesis of Borromean ring 6 exemplifies that the topological transition from a complex $4_1$ metalla-knot to a $6_2^3$ metalla-link can be controlled by selecting the appropriate binuclear building blocks bearing large, rigid π-conjugated planes.

Fig. 8.

Single-crystal X-ray structure of $6_2^3$ metalla-link 6. (A) Chemical structural representation of 6. (B) Cationic structure of 6 viewed along the a axis and (C) viewed along the c axis. (D) View of the space-filling representation of 6. (E) Reduced representation of 6 and the $6_2^3$ link topology. (F) A single macrocycle component constituting 6. All triflate anions, solvent molecules, and H atoms were omitted for clarity. Color code: C, gray; N, navy blue; O, red; Rh, turquoise.

Conclusion

In summary, a coordination and noncovalent interaction (π–π stacking and CH⋯π interactions) codriven assembly strategy was proposed and implemented to create complex closed three-link chains ($6_1^3$ metalla-links), figure-eight knots ($4_1$ metalla-knots), trefoil knot ($3_1$ metalla-knot) and Borromean ring ($6_2^3$ metalla-link), by self-assembly of a set of nonlinear multicurved organic linkers with the appropriate binuclear Cp*Rh^III clips. By combining the ligands bearing a furan or phenyl group with the long binuclear Cp*Rh^III unit, two $6_1^3$ metalla-links were obtained. Conversely, when a much shorter binuclear unit was utilized for assembly with a ligand containing a 2,6-naphthyl group, the $4_1$ metalla-knot was generated, which could be attributed to the enhanced successive interlayered π–π stacking and CH⋯π interactions between the constituents, allowing for spatially closer and more entwined arrangements. Increasing the length of the binuclear clip could generate a $3_1$ metalla-knot. When utilizing the 1,5-naphthyl-containing ligand, delicately adjusting the conjugated plane of the binuclear units allowed the selective formation of the $4_1$ metalla-knot and the $6_2^3$ metalla-link. This work not only provides a methodology for systematically preparing MIMs but also highlights the power of noncovalent interactions in the formation of interlocked and entangled species with increasing complexity.

Materials and Methods

General Procedure for the Preparation of Molecular Closed Three-Link Chains (${\mathbf{6}}_{\mathbf{1}}^{\mathbf{3}}$ Metalla-Links) 1 and 2.

A MeOH (5 mL) solution of ligand L¹ (7.5 mg, 0.02 mmol) and E1 (29.2 mg, 0.02 mmol) was stirred at room temperature for 12 h. The resulting yellow solution was filtered through a membrane filter, and the obtained filtrate was crystallized via isopropyl ether diffusion to obtain orange crystals of complex 1, which were subsequently washed with diethyl ether and dried under vacuum. The synthesis method of 2 was consistent with that of 1, in which 2 was obtained by substituting ligand L¹ for ligand L² (7.7 mg, 0.02 mmol).

General Procedure for the Preparation of Molecular Figure-Eight Knot (${\mathbf{4}}_{\mathbf{1}}$ Metalla-Knot) 3 and Trefoil Knot (${\mathbf{3}}_{\mathbf{1}}$ Metalla-Knot) 4.

A MeOH (5 mL) solution of ligand L³ (8.7 mg, 0.02 mmol) and E2 (19.3 mg, 0.02 mmol) was stirred at room temperature for 12 h. The reacted solution was filtered through a membrane filter, and the obtained filtrate was crystallized via isopropyl ether diffusion to obtain dark green crystals of complex 3, which were subsequently washed with diethyl ether and dried under vacuum. 4 was synthesized following the same procedure as 3, except that E2 was replaced with E3 (21.4 mg, 0.02 mmol).

General Procedure for the Preparation of Molecular Figure-Eight Knot (${\mathbf{4}}_{\mathbf{1}}$ Metalla-Knot) 5 and Borromean Ring (${\mathbf{6}}_{\mathbf{2}}^{\mathbf{3}}$ Metalla-Link) 6.

A MeOH (5 mL) solution of ligand L⁴ (8.7 mg, 0.02 mmol) and E4 (21.2 mg, 0.02 mmol) was stirred at room temperature for 12 h. The reacted solution was filtered through a membrane filter, and the obtained filtrate was crystallized via isopropyl ether diffusion to obtain dark blue crystals of complex 5, which were washed with diethyl ether and dried under vacuum. The synthesis method of 6 was consistent with that of 5, in which 6 was obtained by substituting E4 for E5 (20.2 mg, 0.02 mmol).

X-ray Crystal Structure Determination and Crystallographic Data.

All the crystal structure data were obtained using a Bruker APEX-II CCD diffractometer with Ga Kα radiation (λ = 1.34138 Å). The crystal structures of 1–6 were determined using direct methods and refined using full-matrix least-squares techniques on Fo² via the SHELXL program through the OLEX2 interface. In these data, the highly disordered solvent molecules and anions that could not be restrained properly were removed using the SQUEEZE route in the Platon program. The X-ray crystallographic data for 1–6 are shown in SI Appendix, Tables S1–S3. The CCDC numbers of 1–6 were 2340339 (1), 2340340 (2), 2340341 (3), 2340342 (4), 2340343 (5), and 2343598 (6).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.