Docking rings in a solid: reversible assembling of pseudorotaxanes inside a zirconium metal–organic framework

An unprecedented zirconium metal–organic framework featuring a T-shaped benzimidazole strut was constructed and employed as a sponge-like material for selective absorption of macrocyclic guests. The neutral benzimidazole domain of the as-synthesized framework can be readily protonated and fully converted to benzimidazolium. Mechanical threading of [24]crown-8 ether wheels onto recognition sites to form pseudorotaxanes was evidenced by solution nuclear magnetic resonance, solid-state fluorescence, and infrared spectroscopy. Selective absorption of [24]crown-8 ether rather than its dibenzo counterpart was also observed. Further study reveals that this binding process is reversible and acid–base switchable. The success of docking macrocyclic guests in crystals via host–guest interactions provides an alternative route to complex functional materials with interpenetrated structures.


Introduction
Rotaxanes and pseudorotaxanes are important interpenetrated molecular architectures consisting of macrocyclic wheels and axle molecules. 1 The dynamic and switchable nature of their structures facilitates wide applications in the design of articial molecular machines 2 and advanced catalysts. 3 Rotaxane-derived rotors, 4 shuttles, 5 pumps, 6 transporters, 7 and a peptide synthesizer 8 have been successfully constructed to mimic the motions and functions of biological machines in Nature. Moreover, the reversible threading and de-threading of a pseudorotaxane enables advanced systems for drug delivery, 9 uorescence sensing, 10 and self-healing polymers. 11 Despite these widespread applications, organizing pseudorotaxanes or rotaxanes into crystal lattices to develop solid-state molecular machinery 12 or stimuli-responsive materials 13 is relatively less explored.
Metal-organic frameworks (MOFs), also known as porous coordination polymers (PCPs), are porous materials with periodic and tailorable structures. 14 The high stability and large pore size have proven themselves ideal platforms for accommodating and operating bulky molecular assemblies. 15 The incorporation of rotaxanes into MOFs affords metal-organic rotaxane frameworks (MORFs) which are potentially dynamic materials furnishing motions of molecular machines (Fig. 1a). 12a,16 The most representative work has been achieved by Loeb and co-workers demonstrating that both rotational and translational motions can be unambiguously accessed in UWDM serial materials, paving the way to solid-state molecular machinery. 17 In 2009, an outstanding example of assembling pseudorotaxanes inside MOFs was reported by Stoddart and coworkers. 18 With macrocyclic recognition modules incorporated into a framework, zinc MOF-1001 is capable of docking and sieving dication paraquat guests (Fig. 1a). Although this process can be reversed by rinsing with solvents, efficiently docking and releasing a threaded component in a controlled fashion remains a great challenge.
The UiO series of zirconium-based MOFs are known for their robustness which facilitates post-synthetic modications and switchable applications. 19 A mechanized Zr MOF for photoresponsive cargo release was demonstrated by Wang et al. 13a More recently, Zr MOFs incorporated with benzimidazole rotaxane shuttles exhibit extraordinary acid-base stability which should enable the further design and installation of molecular switches in crystals. 20 These inspiring studies have led us to propose a post-synthetic approach to form pseudorotaxanes in a pre-constructed framework i.e. docking rings in a solid (Fig. 1b). Specically, a naphthalene-elongated T-shaped benzimidazole ligand was designed and synthesized to form a three-dimensional crystalline framework with zirconium (T-MOF, T represents thread). 21 The robustness and large cavities of the material facilitate further acid-base doping modication. Cite this: Chem. Sci., 2022, 13, 6291 All publication charges for this article have been paid for by the Royal Society of Chemistry benzimidazolium, crystal lattices with recognition sites were readily produced. Finally, reversible self-assembly of pseudorotaxanes with crown ether wheels in crystals was accomplished realizing the docking of rings in a solid.

Results and discussion
Key steps for the synthesis of the T-shaped dicarboxylic acid ligand, 4,7-bis(6 0 -carboxynaphthalen-2 0 -yl)-2-phenyl-1H-benzimidazole (BNPB), are outlined in Scheme 1. The diester intermediate 1 was readily obtained by Pd-catalyzed Suzuki coupling of 4,7-dibromo-2-phenyl-1H-benzimidazole with 6ethoxycarbonylnaphthalene-2-boronic acid pinacol ester. Hydrolysis of 1 afforded BNPB in excellent yield (99%). The structure of BNPB was unambiguously conrmed by NMR, mass spectrometry and single crystal X-ray diffraction (SCXRD) analyses (see ESI, Table S1 †). 22 The structure of BNPB shows that the two naphthalene wings are almost co-planar to the central benzene ring due to conjugation (Fig. 2a). The anti-coconformation of two naphthalene wings results in a distance of 19.9Å (d O/O ) between the two carboxyl groups which is fairly close to that of the ligand reported for UiO-69. 23 Accordingly, either an interpenetrated or non-interpenetrated framework could be obtained when combining BNPB with the 12-connection Zr 6 (m 3 -O) 4 (m 3 -OH) 4 secondary building unit (SBU).
Upon heating BNPB and ZrCl 4 in DMF with triuoroacetic acid as the modulator at 100 C for 5 days, octahedral crystals suitable for single-crystal X-ray diffraction were harvested and designated as T-MOF (Fig. 2b). The similar morphology of T-MOF to many other Zr-MOFs constructed with linear dicarboxylic linkers indicates that it is isoreticular to UiO-66. 19 Despite the good crystal quality, T-MOF showed weak diffraction due to the porous nature and its inherent disorder in the framework. SCXRD analysis reveals that T-MOF crystallizes in the cubic space group Fm 3 with a lattice parameter a ¼ 38.49Å (Table S1 †), which is slightly larger than that reported for UiO-69. 23 The Zr 6 (m 3 -O) 4 (m 3 -OH) 4 SBUs are linked by BNPB to afford the 12-connected 3D network of fcu topology. It should be noted that, with the central benzimidazole moiety perpendicular to the struct, a non-interpenetrated framework is obtained exclusively with a variety of synthetic conditions. As depicted in Fig. 2c and d, two types of cages are found in the structure. The cavity sizes are estimated to be ca. 18.2Å and 25.1Å (in diameter) for the tetrahedron and octahedron, respectively. The central benzimidazole moiety is disordered over four positions and located in either of the two different cavities. The total solvent accessible volume of T-MOF is estimated to be 57.2% by analysis using the Olex2 (ref. 24) solvent mask (Olex2 implementation of the Platon Squeeze routine 25 ). The large void implies potential application of T-MOF in accommodating large guest molecules.
To prove the phase purity of T-MOF as well as the integrity of the framework aer solvent exchange, the powder X-ray diffraction (PXRD) patterns of the as-synthesized and dichloromethane (DCM) soaked phases were compared with the pattern calculated from the single crystal data (Fig. S2 †). Slight shiing (0.9 to higher angles) was observed for assynthesized T-MOF measured at 298 K. The discrepancy can be attributed to the different temperatures used to determine the PXRD of bulk and single crystals. Accordingly, Rietveld renement of the cell parameter was further applied (Fig. S3 †). The rened lattice parameter (a 0 ¼ 38.35Å) afforded a simulated pattern in good consistence with those of as-synthesized and DCM soaked materials. Thermal gravimetric analysis (TGA) revealed that activated T-MOF starts to undergo slow degradation above 140 C indicating a lower thermal stability compared to those reported for Zr MOF analogues (Fig. S4 †). This is presumably due to the non-interpenetrated framework with a long strut. N 2 sorption isotherms at 77 K of the activated T-MOF show a Brunauer-Emmett-Teller (BET) surface area of 200 m 2 g À1 (Fig. S5 †), a much smaller number than its theoretical BET surface area (Fig. S22 †), suggesting partially impaired porosity upon removal of the solvents. Nevertheless, pseudorotaxanes are usually formed at ambient temperature in the presence of a solvent. Under such mild operating conditions, the stability of bulk T-MOF crystals should be sufficient to ensure the self-assembling process.
T-shaped benzimidazoliums have been proven to efficiently form pseudorotaxanes with 24-membered crown ethers. 21 To probe the ability of T-MOF to bind crown ethers, a model compound [1-H][BF 4 ] was rst tested for its host-guest chemistry with [24]crown-8 (24C8) and dibenzo [24] S6-9 †). The higher binding affinity for DB24C8 to [1-H] + could be attributed to p-stacking interactions resulting from the clamping of the crown around the axle as previously reported. 21 In addition, disassembling both pseudorotaxanes was readily realized by neutralization with a base.
Accordingly, the neutral as-synthesized T-MOF was subjected to protonation (Fig. 3a). To obtain the protonated material [T-MOF-H] + , CH 2 Cl 2 pre-exchanged T-MOF was soaked in a 0.01 M CH 2 Cl 2 solution of HBF 4 and continuously monitored by X-ray photoelectron spectroscopy (XPS) analysis ( Fig. 3b and Table S2 †). The N 1s spectrum of the neutral T-MOF exhibits peaks at $398.5 and $400.5 eV which can be attributed to the imine (] N-) and amine (-NH-) moieties of the imidazole rings, respectively. Upon protonation, a new band at $401.5 eV for (-NH + ]) appears (Fig. 3b) (Table S2 †). Full protonation was achieved aer soaking for 3 h at room temperature. The completely protonated T-MOF crystal retains its octahedral shape but some surface cracks are observed (Fig. S13 †). Although peak broadening was detected in the PXRD pattern (Fig. 3c) which implies some loss of crystallinity, this did not signicantly impede the study of pseudorotaxane formation in the protonated crystals.

indicated host-guest interactions between [T-MOF-H]
[BF 4 ] and 24C8. 26 This was corroborated by infrared spectroscopy (IR) of the soaked material as shiing of the N-H stretching band from 3401 to 3214 cm À1 infers the formation of N-H/O hydrogen bonding (Fig. S21 †). Further analysis of a digested sample (K 3 PO 4 /D 2 O/DMSO-d 6 ) by 1 H NMR spectroscopy gave a molar ratio of ca. 0.4 for wheels to benzimidazolium domains ( Fig. 4d and S14 †). Conversely, performing the same docking experiments with 18C6, which is known to be too small to thread a phenyl group (Fig. S17a †), 27 resulted in no change in either the uorescence or IR spectrum, and the absence of a crown ether signal in the 1 H NMR (Fig. S17b †). This result rules out the possibility that absorption of the 24C8 of rings is due to non-specic electrostatic interactions or adherence to the crystal surface, proving that docking of 24C8 in the protonated MOF is dictated by mechanical bonding associated with the formation of pseudorotaxanes. As outlined in Fig. 4d,  fast loading of 24C8 was achieved in the rst 6 h (18.7%) and the molar fraction increased to 36% aer immersion for 40 h. A maximum of 37.5% (ca. 0.4 wheel per benzimidazolium domain) was reached in 48 h. No signicant change was observed for prolonged processing indicating the saturation of binding. T-MOF consists of two types of cavities, a large octahedral cage and a smaller tetrahedral cage. Each octahedral cage is face sharing with 8 tetrahedral cages, that gives a ratio of 2 : 1 for the tetrahedron/octahedron in the crystal. Since the volume of a regular octahedron is four times that of a regular tetrahedron, the statistical distribution ratio of "side arms" in cavities should be 1 : 2 for the tetrahedron/octahedron. Considering that the smaller tetrahedral cavity is likely to exhibit a lower binding energy due to the steric restriction of the smaller pore, 20 the ring/linker ratio of 0.4 presumably means that most of the rings are trapped in the octahedral cages to form pseudrotaxanes. To our surprise, attempts to construct a similar material utilizing DB24C8 as the wheel component result in no absorption ( Fig. 4d and S16 †). This could be accounted for by the bulky size and rigidity of DB24C8 as compared to 24C8. Based on reported [2]pseudorotaxane structures, DB24C8 usually adopts the C-shaped clamped conformation when it is threading on a benzimidazolium axle. 21 An estimated dimension of ca. 13.8Å is about the smallest size for a complexed DB24C8 (Fig. S24 †), while it is bigger than the predominant pores with diameters ranging from 10.0 to 12.6Å according to the calculated pore size histogram (Fig. S23 †).
Finally, the reversibility of the docking process was evaluated. As previously noted, disassembly of benzimidazolium pseudorotaxanes can be readily realized by neutralization with a base. To verify this, a ring (24C8) threading experiment was rst conducted on neutral T-MOF with no detectable signal of crown ether observed by 1 H NMR (Fig. S15 †). A ring dethreading experiment was then executed by suspending [24C83T-MOF-H] [BF 4 ] in a triethylamine/CH 2 Cl 2 solution (0.01 mol mL À1 ). 1 H NMR analysis revealed that disassembling of pseudorotaxane inside the solid was rapid and completed within 30 min (Fig. S19 †). The recovery of T-MOF, as conrmed by the PXRD pattern (Fig. 4b), further proves that reversible assembly of pseudorotaxanes inside a metal-organic framework, i.e. docking rings in solids, can be achieved.

Conclusions
To conclude, a novel T-shaped benzimidazole ligand was designed and utilized as struts for the preparation of a zirconium-based metal-organic framework. The neutral domains of the as-synthesized framework could be readily converted to recognition sites for templating pseudorotaxanes with appropriately sized crown ethers. The resulting sponge-like material exhibits selective and acid-base switchable absorption of 24C8 while rebuffing its dibenzo counterpart DB24C8 or the smaller 18C6. The success of reversibly docking and releasing rings inside a crystalline solid provides further impetus for exploration into related phenomena such as mechanisorption. 15d

Data availability
All experimental and computational data associated with this article are included in the main text and ESI. †

Author contributions
K. Z. supervised the project. X. L. performed all the synthetic experiments. X. L. collected and analyzed the NMR, PXRD, TGA, FT-IR, and XPS data with assistance from K. Z., J. X., Z. D., and G. Li. L. J. and K. Z. collected and analyzed the SCXRD data. S. L. supervised theoretical computation study, analysis and interpretation. K. Z. wrote the manuscript with input from X. L. and S. L.

Conflicts of interest
There are no conicts to declare.