There are ever-growing requirements for molecular structural complexity to achieve functions and applications [1, 2]. Self-assembly is an essential process in nature, which plays a critical role in the construction of sophisticated supramolecular materials. Self-assembly is of great importance in chemistry, materials science and biology [3-6]. The advanced integrative self-assembly systems are capable of integrating different components into one complex system with precise control of the structure. This supramolecular strategy enhances the complexity and diversity of synthetic supramolecular architectures and realizes the implementation of complex functions [7-9]. Macrocyclic hosts, including crown ethers [10-12], cyclodextrins [13], calixarenes [14], cucurbiturils [15-17] and pillararenes [18-21], have emerged as important components for the construction of self-assembly systems and materials. These hosts have been widely used to fabricate supramolecular aggregates with tailor-made topology and functions.

Cucurbit[n]urils (CB[n]), a family of highly symmetrical pumpkin shaped macrocycle host molecules with rigid hydrophobic cavity, have tight and selective binding towards various guest molecules, have great applications [15-17]. In particular, CB[8] is capable of simultaneously binding two guests to form ternary complexes. This CB[8]-based ternary complexation has been used as a supramolecular "cross-linker" to construct supramolecular polymers, supramolecular organic frameworks and supramolecular materials [22-39]. A nor-seco-cucurbit[10]uril (ns-CB[10]) is a double cavity host molecule [40-47], which is able to form ternary complexes. We decide to use the ns-CB[10]-based ternary complexation to construct supramolecular structures of different dimensions. By applying ns-CB[10] and other CB[n] homologues, we are able to construct [5]rotaxane, linear supramolecular dynamic rotaxane polymers and cubic 3D supramolecular organic framework (SOF).

Self-assembled (pseudo)rotaxane and rotaxane-type polymers based on CB[n] host molecules have been widely used in various applications [48]. Disulfide bonds are common in biological systems, especially in proteins, and are important in dynamic covalent chemistry [49-52]. CB[6] could selectively encapsulate the sulfhydryl or disulfide part of the guest molecules and have certain inhibitory or protective effect [53]. We designed and synthesized linear compounds M1, M2 and M3 with sulfhydryl or disulfide moiety, to study its self-assembly with ns-CB[10], tetramethyl CB[6] (TMeCB[6]) and CB[7] towards the formation of [5]rotaxane and linear supramolecular dynamic rotaxane polymers (Fig. 1). Tetrahedral compound T1 was synthesized to study its assembly with ns-CB[10] towards the formation of SOF (Fig. 1). Compound G1 was used as a model guest to study the ternary complexation with ns-CB[10].

The binding between compound M1 and TMeCB[6] was investigated. The crystal structure of the complex formed between M1 and TMeCB[6] revealed a 1:1 binding stoichiometry and the S-S bond of M1 was encapsulated in the cavity of the TMeCB[6] cavity (Fig. 2a). Combined with the stacking of bipyridine, a quadrilateral two-dimensional structure was formed, which was similar to structures reported in literature (Fig. 2b) [54]. The S-S bond of M1 was encapsulated in TMeCB[6] was confirmed by ¹H NMR spectroscopy in D₂O. 1:1 binding stoichiometry was confirmed by absorption spectrum in solution (Figs. S1 and S2 in Supporting information). Due to the bulky tert-butyl benzene group, compound M3 was unable to thread through the cavity of TMeCB[6]. Instead, compound M3 binded to the carbonyl oxygen portal of TMeCB[6], which was confirmed by ¹H NMR spectroscopy and absorption spectrum (Figs. S3 and S4 in Supporting information). Disulfide bond was a dynamic covalent bond under weak alkalescent conditions. We discovered that disulfide bond was cleaved when growing single crystals with the solution of M3, TMeCB[6] and ns-CB[10] (M3: TMeCB[6]: ns-CB[10] = 1:1:1), which might be due to the residual alkaline from the synthesis of ns-CB[10] (Fig. 2c and Fig. S3). TMeCB[6] and ns-CB[10] encapsulated the sulfhydryl moiety and tert-butyl benzene moiety of the cleaved product M2, respectively. Due to the outer-surface interactions [17], a very regular arrangement structure was formed, ns-CB[10] and TMeCB[6] were showed hexagonal and rhomboid quadrilateral distribution, respectively (Fig. 2d). The complexation of compound M2, TMeCB[6] and ns-CB[10] (M2: TMeCB[6]: ns-CB[10] = 1:1:0.5) was studied by ¹H NMR spectroscopy (Fig. S5 in Supporting information). TMeCB[6] encapsulated sulfhydryl moiety and ns-CB[10] encapsulated tert-butyl benzene moiety to form a [5]rotaxane, which was in line with the crystal structure (Fig. 2c).

To extend the [5]rotaxane into supramolecular polymer, TMeCB[6] was replaced by CB[7], whose cavity was spacious enough to be threaded by tert-butyl benzene group. The self-assembly of compound M1, compound M3, ns-CB[10] and CB[7] was evaluated (Fig. 3). ¹H NMR and absorption spectra of M1 and CB[7] showed similar observations as M1@TMeCB[6], the S-S bond of M1 was encapsulated in the cavity of the CB[7] to form a complex with 1:1 binding stoichiometry (Figs. S6-S8 in Supporting information). While for M3 and CB[7], ¹H NMR spectroscopy revealed a formation of rotaxane or pseudorotaxane. When the binding stoichiometry of M3 and CB[7] was 1:1, the S-S bond of M3 was encapsulated by CB[7] (Fig. 3a) display upfield shifts (Δδ = −0.66 and −0.84 for H_i and H_j, respectively). By contrast, when the binding stoichiometry was 1:3, a [4]rotaxane was formed (Fig. 3a). The 1:3 binding stoichiometry was confirmed by absorption spectrum in solution (Figs. S9 and S10 in Supporting information). In this [4]rotaxane, CB[7] encapsulated S-S bond and then tert-butyl benzene group (Δδ = −0.30, −0.90 and −0.70 for H_c, H_b and H_a, respectively). We further increase the complexity by adding ns-CB[10] into the system [46]. Compound M3, CB[7] and ns-CB[10] were mixed in solution with a molar ratio of 1:1:1. ¹H NMR spectroscopy revealed that S-S bond was encapsulated in CB[7], and tert-butyl benzene was encapsulated in ns-CB[10], displayed upfield shifts (Δδ = −0.66, −0.84, −0.73, −0.87 and −0.55 for H_i, H_j, H_c, H_b and H_a, respectively) (Fig. 3b). Comparing to the crystal structure of M2@TMeCB[6] & ns-CB[10], we believe M3, CB[7] and ns-CB[10] formed a linear supramolecular dynamic rotaxane polymers. Dynamic light scattering (DLS) experiments for the solution of M3, CB[7] and ns-CB[10] (1:1:1) ([M3] = 1.0 mmol/L, CB[7] = 1.0 mmol/L, ns-CB[10]=1.0 mmol/L) gave rise to a hydrodynamic diameter (D_H) of 531 nm, significantly larger than that for [4]rotaxane (18 nm, [M3] = 1.0 mmol/L, CB[7] = 3.0 mmol/L) (Fig. S11 in Supporting information). 2D diffusion-ordered NMR spectroscopy (DOSY) further confirmed the supramolecular polymers. The diffusion coefficient of M3/CB[7]/ns-CB[10] (1:1:1, 0.5 mmol/L) in water was 8.9 × 10⁻¹¹ m²/s, and based on the DOSY results, the molecular weight of the supramolecular polymer was calculated to be 4.3 × 10⁴ Da, the result were similar to nor-seco-cucurbit[10]uril based linear supramolecular polymer by Xu et al. [46] (Fig. S12 in Supporting information). These observations indicated a ns-CB[10] based supramolecular dynamic rotaxane polymers formed.

Before ns-CB[10] was used to construct SOF, the binding between ns-CB[10] and compound G1 was studied. The crystal structure for the complex of G1 and ns-CB[10] revealed a 2:1 binding stoichiometry and the formation of [3]rotaxane. In the [3]rotaxane, the two butyl of G1 were antiparallelly encapsulated in the two cavities of the ns-CB[10] (Fig. 4a), which was also confirmed in D₂O by ¹H NMR spectroscopy (Fig. S13 in Supporting information). Next, compound T1 and ns-CB[10] was mixed to construct 3D SOF (ns-SOF-Bu) (Fig. 4b). ¹H NMR spectra showed that compound T1 formed complex with ns-CB[10] with a 1:2 binding stoichiometry, which was confirmed by a maximum proton resonance upfield shift when 2.0 equivalent ns-CB[10] was added (Fig. S14 in Supporting information). DLS experiments for the solution of T1 and ns-CB[10] (1:2) gave rise to a D_H ranging from 342 nm to 615 nm, depending on the concentration of the mixture ([T1] = 0.03‒1.0 mmol/L), indicating the formation of nanostructures (Fig. 4c). The structural ordering of ns-SOF-Bu was investigated with synchrotron small-angle X-ray scattering (SAXS) and selected area electron diffraction (SAED). This scattering signal of ns-SOF-Bu corresponded to the simulated {001} and {111} spacing 2.48 nm and 1.83 nm of a modeled network (Figs. 4d and e), simulated on the basis of the crystal structure of the 2:1 complex of G1 and ns-CB[10] and the interpenetrated crystal structure 1:2 complex of T1 and CB[8] (SOF-Bu) [34]. According to the larger structural characteristics of ns-CB[10], the unit-cell parameters of modeled interpenetrated structure ns-Bu-SOF are slightly larger than crystal structure SOF-Bu, and the aperture of the cyclohexane-like pore of interpenetrated ns-Bu-SOF in modeled interpenetrated structure was measured to be 2.5 nm, larger than 2.1 nm of crystal structure of SOF-Bu [34]. The SAED pattern of different microcrystals of ns-SOF-Bu showed the {040} and {004} lattice spacings as being viewed from the [100] lattice direction (Figs. 5a-d). Elemental mapping analysis of the microcrystals confirmed the compositions of the C, N, O, Br elements (Fig. S15 in Supporting information). All the above observations showed a high regularity of ns-SOF-Bu in the solid state. The structural ordering of ns-SOF-Bu in solution was also confirmed by SAXS, the scattering signal corresponded to the simulated {222} spacing 1.79 nm of a modeled network (Fig. 4f). And referenced to literatures methods [34], to test if the ns-SOF-Bu formed non-interpenetrating pores in solution, dialysis experiments were carried out for two dyes acid red 27 and 4, 4′, 4′′, 4′′′-porphyrin-5, 10, 15, 20-tetrabenzoate (PTB, as sodium salt) using dialysis bags (MWCO: 1000 Da). The result showed that ns-SOF-Bu was able to completely retained the dyes from escaping (≤1.0%) (Fig. S16 in Supporting information), indicating SOF was a porous structure in water. Therefore, a non-interpenetrated porous framework was formed in solution (Fig. 5f). This solid of ns-SOF-Bu by evaporation was also used to adsorb acid red 27, the adsorption percentage was 40%, which may be rationalized by considering partial formation of the interpenetrated structures (Fig. S17 in Supporting information).

Diluting the solution of [5]rotaxane (M2/ns-CB[10]/TMeCB[6]), [4]rotaxane (M3/CB[7]), linear supramolecular dynamic rotaxane polymers (M3/CB[7]/ns-CB[10]), and ns-SOF-Bu (T1/ns-CB[10]) from 1.0 mmol/L to 0.05 mmol/L did not cause observable shifting of the signals in ¹H NMR in D₂O (Figs. S18-S21 in Supporting information). The ¹H NMR spectrum of the sample in D₂O did not exhibit any change after being placed at room temperature for more than 96 h (Figs. S18-S21). Both results confirmed the high stability of the new supramolecular framework in solution.

In summary, ns-CB[10]-based ternary complexation was used to construct rotaxanes, linear supramolecular polymers and 3D supramolecular organic frameworks. It demonstrated that ns-CB[10] could act as a supramolecular cross-linker to building finite molecular devices and infinite supramolecular materials. This discovery has a major implication for CB[n]-based crystal engineering and material assembly.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We thank the supports from the National Natural Science Foundation of China (Nos. 21890732, 21890730 and 21921003). SAXS data were collected with beamline BL16B1 at Shanghai Synchrotron Radiation Facility.

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.06.055.