Metal–organic cycle-based multistage assemblies

Significance

Different from polymers or peptides (lacking metals), metal–organic cycles (MOCs) have properties which arise from the combination of metals and common nonmetal elements and topologies. The development of MOC supramolecular materials is in its infancy, and how the coordination bonds work to make the corresponding suprastructures is unknown. This has limited the potential application of these MOC-based materials. Considering the applications of individual MOCs, the study and discovery of the unique factors in MOC-involved multilevel self-assembly are critical to further our knowledge of the underlying molecular mechanisms of metal-containing compounds. Here, a systematic study of MOC assembly in various solvent systems has confirmed the critical role of coordination linkers in tuning the shape and size of the MOC-derived suprastructures.

Abstract

It is well known that chemical compositions and structural arrangements of materials have a great influence on their resultant properties. Diverse functional materials have been constructed by using either biomolecules (peptides, DNA, and RNA) in nature or artificially synthesized molecules (polymers and pillararenes). The relationships between traditional building blocks (such as peptides) have been widely investigated, for example how hydrogen bonds work in the peptide multistage assembly process. However, in contrast to traditional covalent bond-based building blocks-based assembly, suprastructures formed by noncovalent bonds are more influenced by specific bond features, but to date only a few results have been reported based on noncovalent bond-based building block multistage assembly. Here, three metal–organic cycles (MOCs) were used to show how coordination bonds influence the bimetallacycle conformation then lead to the topology differences of MOC multilevel ordered materials. It was found that the coordination linker (isophthalate-Pt-pyridine) is an important factor to tune the shape and size of the MOC-derived suprastructures.

In nature, as a typical bottom-up method to create functional materials, self-assembly has been employed by all living organisms to provide important compounds to perform various biological functions. Inspired by nature, self-assembly has been used as a powerful tool to create artificial multifunctional materials with specific functions (1). In these dynamic systems, the as-obtained suprastructures emerge as products after countless self-corrections associated with the balance of weak interactions (2–5). In general, hydrogen bonds, aromatic stacking, metal coordination, hydrophobic/hydrophilic, and electrostatic interactions are involved in the assembly process. In the past, proteins (6), macromolecules (7, 8), and polymers (9) were used as building blocks to prepare nanofibers (10), nanovesicles (11), and networks (12). With the advance of the self-assembly method, multicomponent and multistage assembly have been developed for obtaining more complicated applications (13). As a result, these materials can be used to prepare semiconductors (14), smart devices (15), etc. As a consequence, the as-prepared material can be further used as electronic skin (16) to monitor dynamically pulsing cardiomyocytes (17).

Chemical components have a considerable influence on the resultant properties, and although a large number of suprastructures have been prepared from peptides and polymers, their properties have been restricted via the limited elements involved. Thus, the building blocks with and without metal show a dramatic difference in their material properties. Therefore, studying the relationship between metal–organic suprastructures (MOS) and the resultant multiscale well-organized materials is important and necessary for generating new materials with specific properties. However, different from the achievements in traditional building blocks-based functional materials (the first polymer was reported in 1907) (18), the knowledge of building blocks such as MOS-based (the first being reported in 1983) (19) on multilevel and multistage assembly is still in its initial stage.

As compared to a typical MOS, metal–organic cycles/cages (MOCs) (20–24) are discrete, well-defined molecular entities with various metal centers that can be used as noncovalent bond-based building blocks (NBBs) for fabricating functional suprastructures. Compared with the building blocks obtained by the traditional covalent bonds, noncovalent bond-based units, such as MOCs (25–27), exhibit attractive advantages in catalysis, fluorescent sensors, gas separation, and biomedical diagnosis and can serve as emissive materials and artificial biomolecular channels (28–38). For instance, MOCs exhibit a specific geometry depending on their precursor angle. To date, a series of MOCs with controllable geometries, including triangles (39), tetrahedra (40), rhomboids (41), hexagons (42), bimetallacycles (43), cuboctahedra (44), trigonal (45), and tetragonal prisms (46) have been prepared. Compared to their covalent bond-based building units, novel properties (47) appear due to their dynamic transformation features (48). For example, anthracene-based highly twisted metallacycles form instead of the expected square formed by coordination-driven self-assembly (49). Tetra-(4-pyridylphenyl)-ethylene (TPPE)-based MOCs maintain the tetragonal prism geometry in CHCl₃ but exist as a bimetallacycle in dichloromethane (DCM). This solvent-templated synthesis of MOCs with various geometries is based on the incorporation of metallacrown recognition units into a flexible nanostructure, which allows the generation of solvent-specific binding pockets (50). Everything from their chemical/physical properties to their biomedical activities and catalysis was found to be influenced by their suprastructures (51, 52).

Compared with covalent bond building blocks, CBBs-related assembly, suprastructure construction involving NBBs is more complicated and requires more delicate control for balancing weak interactions (53). Specifically, how to tune the balance between NBBs and solvents and facilitate the self-assembly/correction of NBBs during the dynamic process is a challenge (54–56). Different from the mature methods for preparation and use of polymer-based materials (57), the first MOC-based multiscale ordering materials were only recently reported (15). On the other hand, the exact mechanism of suprastructure formation in these self-assembly processes is not well understood. Thus, it is necessary and helpful to develop a systematic understanding of the relationships among MOCs, the resultant suprastructures, and the corresponding properties.

Results and Discussion

Preparation and Characterization of MOCs 1–3.

In this work, MOCs 1–3 were prepared using TPPE (4), m-dicarboxylate (5)/p-dicarboxylate (6)/(oligo(ethylene glycol))OEG-modified dicarboxylate (7), and cis-(PEt₃)₂Pt(OTf)₂ (8) via the coordination-driven self-assembly at 50 °C for 1 h (Scheme 1A). ¹H-NMR (SI Appendix, Figs. S1–S3), ³¹P-NMR (SI Appendix, Figs. S4–S6), and electrospray ionization time-of-flight mass spectrometry data (SI Appendix, Figs. S7–S9) confirm the formation of MOCs 1–3. Light yellow single crystals of these MOCs were obtained by slow diffusion in dichloromethane and ethyl acetate at 25 °C (SI Appendix, Table S1). It is shown that MOCs have differing space groups of triclinic P-1 (MOC 1), monoclinic c2/c (MOC 2), and triclinic P-1 (MOC 3) when the ligands change from m-dicarboxylate to p-dicarboxylate and then OEG-modified dicarboxylate. The atomic distances and dihedral angles change with the substituent employed (Scheme 1 B and C). Subsequently, MOCs 1–3 were designed as a model system to study how the MOCs’ structural difference influences the differences in morphology with changes in the ligand from m-dicarboxylate to p-dicarboxylate.

Scheme 1.

(A) Synthesis of MOCs 1–3 through coordination-driven assembly. (B) Cartoon showing the C–C, N–N, and Pt–Pt distances. (C) Dihedral angles between benzene rings and pyridine rings and other pyridine rings/dicarboxylate-related angles. SEM, TEM images (false color were used to label assemblies with different morphologies), and corresponding histograms of (D) MOC 1-based linear assemblies and (E) MOC 2-based nonlinear assemblies formed in DCM/TOL at DCM contents of 50% (10 μM).

Characterization of the Resultant Suprastructures.

The structural features of these three MOCs allow us to conclude with considerable confidence the characteristics that influence higher-level assembly. A DCM/toluene (TOL) solvent system of three different ratios was used to investigate further the assembly behaviors of MOCs 1–3. Optical microscopy, scanning electron microscopy (SEM), and transmission electron microscopy (TEM) were employed to characterize the resultant suprastructures.

At TOL 50%, assemblies obtained from MOC 1 have linear suprastructures (Scheme 1D and SI Appendix, Fig. S10), and the MOC 2-based aggregates show nonlinear (cubic) suprastructures (Scheme 1E and SI Appendix, Fig. S11). Specifically, SEM and TEM images (Scheme 1D, Left and Middle) show the linear assemblies with lengths around 10 µm, and the enlarged image in Scheme 1D also shows these aggregates were stacked in layers. For MOC 2, instead of m-dicarboxylate, if p-dicarboxylate is used as a linker in bimetallacycle the resultant suprastructures change from linear to cubic morphology (Scheme 1D, Left). In particular, the rhomboid suprastructure (Scheme 1E, Right) indicates not only cubic suprastructures but also other polyhedra were formed when p-dicarboxylate was used as a ligand. Histograms show the linear aggregates with a length of 11 ± 4 µm and width of 1.0 ± 0.3 µm (Scheme 1D, Right). For polyhedra, the length is 4.3 ± 0.9 µm (Scheme 1E, Right). However, no aggregates were observed from MOC 3 (using m-dicarboxylate, OEG-modified dicarboxylate as a linker) in the same solvent environment due to the better solubility of OEG.

It is found that the wire size increased from 15 ± 6 µm (length) and 1.3 ± 0.3 µm (width) to 33 ± 13 µm (length) and 2.0 ± 0.5 µm (width) as the TOL ratio increased from 60 to 80%. In particular, besides wires, small sheets (labeled purple) were also observed (Fig. 1 A and B). These results indicated that the morphological compositions and assembly purity of MOC 1 were influenced by the TOL ratio (SI Appendix, Figs. S12 and S13). However, for MOC 2, increasing TOL ratio resulted in the morphology changes from polyhedron (50% TOL, Scheme 1E) to polygon (80% TOL). As shown in Fig. 1C, MOC 2-based rhomboid prisms (red) were obtained at a TOL ratio of 60%. A continuous increase in the TOL content promoted the formation of hexagonal sheets (Fig. 1D and SI Appendix, Figs. S14 and S15), indicating the TOL content can be used to tune the morphology while it cannot be used as a parameter to tune the size of the nonlinear aggregates.

Fig. 1.

SEM and TEM images (false color were used to label assemblies with different morphologies) and corresponding histograms of MOC 1-based linear assemblies (A and B) and MOC 2-based nonlinear assemblies (C and D) formed in DCM/TOL at DCM contents of (A) 40% and (B) 20% (10 μM). X-ray single-crystal structure of dihedral angles between benzene rings and pyridine rings and 3D structure along the a-axis of MOC 1 (E and F) and MOC 2 (G and H).

The influence of the dicarboxylate linker on the resultant multilevel suprastructures was further confirmed by the second solvent system DCM/cyclohexane (CY). Fig. 2 A–C shows micrometer rods are the dominant morphology when using MOC 1 for multistage assembly (SI Appendix, Fig. S17–S19), confirming the relationship between MOC 1 and linear aggregates. Under the same experimental conditions, polygon and polyhedra were observed when using MOC 2 for multistage assembly (Fig. 2 D–F), confirming the relationship between MOC 2 and nonlinear aggregates. For MOC 2, just as with the DCM/TOL solvent system, a DCM/CY solvent system can be used to tune the size by changing the CY content from 50 to 80% (SI Appendix, Figs. S20–S22).

Fig. 2.

SEM and TEM images (false color were used to label assemblies with different morphologies) and corresponding histograms of MOC 1-based linear assemblies (A–C) and MOC 2-based nonlinear assemblies (D–F) formed in DCM/CY at DCM contents of 50%, 40%, and 20% (10 μM).

MOC conformation influence on multistage assembly was further confirmed by the third solvent system, DCM/dioxane (DX). As shown in SI Appendix, Figs. S23–S25, only linear suprastructures were formed by MOC 1, and no MOC 2 hierarchical aggregates were observed. Specifically, in the DCM/DX system, the length of MOC 1 wires decreased from 67 ± 17 µm to 28 ± 9 µm then to 20 ± 6 µm and the corresponding widths increased, when the DCM content increased from 20 to 40% then to 50%. Likewise, the fourth solvent system, DCM/ethyl acetate, brought about the formation of similar linear suprastructures for MOC 1 (SI Appendix, Fig. S26), whereas only nonlinear aggregates were observed for MOC 2 in the same solvent system.

Based on above results, the formation of MOC 1-based linear suprastructures is most likely to be structure-dependent rather than solvent-dependent, since similar linear morphologies have been observed in various solvent systems with a wide range of polarities and structural properties. To gain insight into the mechanism of structure-dependent morphogenesis, more information about the organization of metallacycles in three-dimensional (3D) suprastructures was provided by the stacking mode of MOCs crystals. After the formation of coordination bonds, MOCs 1–3 have different symmetries in atomic distances and dihedral angles (SI Appendix, Fig. S16). Specifically, for MOC 1, the same atomic distances (C to C, N to N, Pt to Pt, and P to P) and dihedral angles (Fig. 1E) demonstrate the symmetry of the bimetallacycle at the structural level, resulting in the parallel stacking mode of MOC 1 (Fig. 1F). As a result, linear aggregates can be observed in MOC 1-based multistage assembly (Fig. 1 A and B). For MOC 2, because of the difference in the dihedral angle twist after coordination bond formation, asymmetry can be easily observed. The green part of the model diagram represents the narrow portion of the molecule, and the purple part represents the wide portion of the molecule (Fig. 1G). The asymmetry of the two benzene rings on each side plays an important role in determining the stacking mode. To reduce steric hindrance, an alternating stacking mode is adopted (Fig. 1H). As a result, nonlinear aggregates were formed in this case (Fig. 1 C and D).

Variable-temperature NMR spectroscopy.

Variable-temperature NMR (VT-NMR) was applied to gain further insight into the multistage assembly process, providing evidence that carboxylate ligands serve as important moieties to adjust the organization of bimetallacyles through their flexible features. As shown in Fig. 3, in both DCM/CY and DCM/TOL solvent systems MOC 1 exhibited noticeable chemical shift changes (Fig. 3 A and B), whereas almost no changes were observed for MOC 2 except that a slight upfield shift in the protons of p-dicarboxylate was detected after increasing temperature from 15 °C to 30 °C in DCM/TOL (Fig. 3 C and D). This phenomenon indicates that the electron cloud density increases and the shielding increases, since the benzene rings of p-dicarboxylate are turned outward. Specifically, the proton NMR spectra of MOC 1 show smaller chemical shift changes in DCM/TOL than in DCM/CY (Fig. 3A and SI Appendix, Figs. S29 and S30), which indicates that MOC 1 is more flexible in cyclohexane than in TOL. Such solvent-dependent flexibility of MOC 1 could be attributed to the hydrophobic interactions between solvent molecules and MOCs, which influence the stacking mode of MOCs.

Fig. 3.

Histograms show the changes in chemical shifts of (A) MOC 1 and (C) MOC 2 in DCM/CY (CD₂Cl₂/cyclohexane-d₁₂ = 8:2, 100 μM) and DCM/TOL (CD₂Cl₂/toluene-d₈ = 8:2, 100 μM) at 15 to 30 °C; chemical shifts-based color span shows isophthalate-Pt-pyridine (green region, coordination bonds as linkage) are more flexible than TPE part in (B) MOC 1 and (D) MOC 2. DOSY NMR spectra of MOC 1 in DCM (E) and MOC 2 in (F) DCM, (G) DCM/CY, and (H) DCM/TOL.

Crystal data (SI Appendix, Fig. S31) provided further information on the flexibility of MOCs 1 and 2. By comparing the weak interaction distributions in MOCs 1 and 2, it was shown that COO⁻-PEt₃, OTf-PEt₃, and OTf-Py existed in both MOC 1 (SI Appendix, Fig. S31A) and MOC 2 (SI Appendix, Fig. S31 B–D). In MOC 2, these three interactions still acted as the main influence, whereas more interactions, such as pyridine rings and COO⁻, PEt₃ and benzene rings, and OTf and pyridine rings, were found. These results further indicate that the bimetallacycle MOC 2 is more hindered in the assembly process than MOC 1. In addition, more H-bonds in MOC 2 than MOC 1 were observed at the intramolecular level as well (SI Appendix, Tables S2–S4), and more hindrances for MOC 2 were also found at the intermolecular level. The association of these interlayer bonds strengthens the suprastructures, leading to the formation of a more stable staggered stacking system (SI Appendix, Figs. S31 E and F, S32, and S33).

DOSY and ROESY NMR studies.

Diffusion ordered spectroscopy (DOSY) experiments were used to get further insight into the MOC involved suprastructure systems construction. In DCM, the proton signals of MOC 1 are correlated with a diffusion coefficient (D) of ∼4.64 × 10⁻¹⁰ m²/s (Fig. 3E and SI Appendix, Fig. S34); for MOC 2, D is 6.90 × 10⁻¹⁰ m²/s (Fig. 3F and SI Appendix, Fig. S35). For MOC 1, due to the formation of microscale assemblies, resulting in the linear aggregates of these assemblies, no reasonable DOSY data can be obtained. For MOC 2, the results of the DOSY experiments show that the proton signals are correlated with a D value of ∼6.90 × 10⁻¹⁰ m²/s in DCM and decreased D of 3.80 × 10⁻¹⁰ m²/s in DCM/TOL and 3.81 × 10⁻¹⁰ m²/s in DCM/CY (Fig. 3 F–H and SI Appendix, Figs. S36 and S37), demonstrating the formation of larger aggregates in these solvent systems.

Rotating frame Overhauser enhancement spectroscopy (ROESY) NMR (SI Appendix, Figs. S38 and S39) was employed to study the influence of solvent systems and bimetallacycle structural features on the resultant molecular interactions. For both metallacycles (MOC 1 and MOC 2) and the two solvent systems (DCM/CY and DCM/TOL), the long-range H–H coupling was observed between PEt₃ and H_a in pyridine (labeled scarlet in the insets of SI Appendix, Figs. S38 C, D, F and S39, named as PEt₃-Pt-Py region), which confirmed the formation of Pt–N coordination bonds. In addition, for the MOC 1 DCM/TOL system, the long-range ¹H–¹H coupling is different from the rest of the systems and implies that the interactions among bimetallacycles change with the substituents/solvent systems through the hydrophobic interactions between MOCs and solvent molecules.

In addition to the similar interactions among TPPE moieties, ¹H–¹H coupling differences in DCM/TOL were observed, and thus solvent effects influence MOC 1 multilevel assembly. Specifically, the scarlet region in MOC 1 (SI Appendix, Fig. S38) coupling between H_CH2-/H_f instead of the previous couplings as found in DCM/CY, indicating there are interactions between methylene groups in the PEt₃ and the isophthalate. These observations provide more detail about how the solvent plays a special role in tuning the arrangement of MOCs through coordination bonds. For MOC 2, no such differences were observed in the two solvent systems (DCM/TOL and DCM/CY; SI Appendix, Figs. S38 E and F and S39). These outcomes are consistent with the observations during VT-NMR experiments. The different flexibility of MOC 1 in the two solvent systems is probably due to the different stacking modes, which is in agreement with its different PEt₃– isophthalate interactions observed in ROESY spectra, whereas no significant flexibility differences were observed for MOC 2 under the same experimental conditions, which is in line with its similar intramolecular interactions in the two solvent systems.

The cooperation of molecular structures and solvents in controlling the assembly was further confirmed by the chemical shift changes after the addition of CY or TOL (SI Appendix, Fig. S40). Generally, TOL was found to induce more significant chemical shifts than CY (SI Appendix, Figs. S41–S48).

Fig. 4 shows that the influence of MOCs’ structural features and solvent system on the assembly extends from ligand-influenced coordination level forming the MOCs to a higher intra-MOC level that leads to variations in topology as well as size. As shown on the left side in Fig. 4A, one-dimensional linear assemblies can be formed by MOC 1. In particular, the length decreases as the solvent system changes from DCM/DX to DCM/TOL to DCM/CY. The right side shows nonlinear aggregates in yellow such as two-dimensional polygons and 3D polyhedrons that can be formed by MOC 2. Scanning TEM (STEM) and energy-dispersive X-ray spectroscopy (EDS) images of the typical elements (Pt, F, O, N, P, and S) show the uniform distribution of MOCs across the assemblies (Fig. 4B), whereas no aggregates are formed by MOC 3 under the same conditions. Fig. 4C represents that isophthalate-Pt-pyridine (coordination bonds as linkage) plays an essential regulatory role in multistage self-assembly, which served as the most critical parameter for constructing supramolecular materials with various topological structures.

Fig. 4.

(A) Scheme of MOCs 1- and 2-based assemblies in DCM/DX, DCM/TOL, and DCM/CY. (B) STEM and EDS images of MOCs 1- and 2-based assemblies. (C) Isophthalate-Pt-pyridine plays an essential regulatory role in multistage self-assembly.

Conclusion

To date, various assemblies with complicated suprastructures have been obtained by different building blocks through self-assembly. With the advances in this field, more ligands with desired functions have been introduced to prepare building units, and a multistep assembly process was developed for more complex systems. However, in a system dominated by a series of weak interactions, how to best direct the coordination interactions, hydrogen bonds, and aromatic stacking is quite important. This work demonstrates how ligands influence the coordination bond formation and determine the steric conformation of noncovalent building blocks, as well as how the hydrogen bond distributions in the building blocks influence the higher-level assembly by controlling the stacking mode. It was found that isophthalate-Pt-pyridine (coordination bonds via linkage) plays an essential regulatory role in multistage self-assembly, which served as the critical parameter for controlling the construction of supramolecular materials with various topological structures. As a result, MOC-derived suprastructures with controllable topologies, from linear wires/rods to nonlinear polygons/polyhedra, can be predicted and synthesized based on the cooperation of weak interactions. These findings pave the way to understand and control the suprastructure in multilevel assemblies by rationally designing coordination bond-based building blocks.

Materials and Methods

All reagents were commercially available and used as supplied without further purification. Deuterated solvents were purchased from Cambridge Isotope Laboratory. NMR experiments were recorded on a spectrometer at room temperature. ¹H NMR spectra were recorded in the designated solvents on a Varian Inova 400-MHz and a Bruker Neo 500-MHz spectrometers. ³¹P{¹H} NMR spectra were recorded on a Varian Unity 300-MHz and a Bruker Neo 500-MHz spectrometers, and ³¹P{¹H} NMR chemical shifts were referenced to an external unlocked sample of 85% H₃PO₄ (δ 0.0 ppm). Mass spectra were recorded on Waters TQD Acquity UPLC w/Acquity H Class UPLC mass spectrometer.

Data Availability

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