Size-Induced Highly Selective Synthesis of Organometallic Rectangular Macrocycles and Heterometallic Cage Based on Half-Sandwich Rhodium Building Block

The controlled synthesis of organometallic supramolecular macrocycles cages remains interesting and challenging work in the field of supramolecular chemistry. Here, two tetranuclear rectangular macrocycles and an octuclear cage were designed and synthesized utilizing a rigid and functionalized pillar linker, 2,6-bis(pyridin-4-yl)-1,7-dihydrobenzo [1,2-d:4,5-d′]diimidazole (BBI4PY) based on three half-sandwich rhodium building blocks bearing different sizes. X-ray crystallography in combination with 1H NMR spectroscopy elucidated that the two building blocks with shorter spacers only result in rectangular macrocycles. However, the building block of bulkier size to avoid the π-π stacking interactions between two ligands BBI4PY led to the formation of an octuclear cage complex. The latter cage contains two types of metal ions, namely Rh3+ and Cu2+, showing significant characteristics of heterogeneous metal-assembling compounds. In addition, the cage accommodates two free isopropyl ether solvent molecules, thus displaying host–guest behavior.


Introduction
In recent years, the design and synthesis of coordination supramolecular compounds have made tremendous advances due to their various potential applications in catalysis, host-guest chemistry, nanomaterials, separation, etc. [1][2][3][4][5][6][7][8]. These studies have attracted the attention of many scientists and have gradually become a hot topic in the field of supramolecular chemistry, as demonstrated by the work of Sauvage, Stoddart, Stang, Jin and others [9][10][11][12][13]. Accompanied by the development of this field, a series of molecular knots [14][15][16][17], catenanes [17][18][19], Borromean rings [20][21][22], Solomon links [23][24][25] and molecular cages [26][27][28] have been designed reasonably and reported continuously. In the process of research, the main synthetic methods were divided into two classes: metal-iontemplated and template free based on coordination-driven methods [29][30][31][32][33]. The former has been utilized to construct various coordination supramolecular compounds. The latter can not only achieve the construction of supramolecular structures but can also study the structural dynamic transformation characteristics among molecular catenanes, Borromean rings, Solomon links or knots and organometallic macrocycles [22]. These studies have promoted the rapid development of supramolecular chemistry to a certain extent. However, the research on the selective formation of macrocycle and cage induced merely by the size effect of building blocks has never been reported. Additionally, in order to realize the selective construction of organometallic macrocycle and cage, the judicious selection of metallic building blocks and organic ligands with functional sites is vitally important.

Synthesis of Metallarectangles 1, 2 Based on E1 and E2
The ligand L1 and two binuclear complexes [Cp*RhCl(µ-Cl)] 2 (E1) [38] [Cp* 2 Rh 2 (BiBzIm)]Cl 2 (E2) [39] were prepared according to literature methods. The metallarectangles 1, 2 were acquired by the combination of L1 ligand and the binuclear complexes following a two-step construction method. As shown in Scheme 2, firstly, the binuclear complex was treated with two equivalents of silver trifluoromethanesulfonate (AgOTf) in the dark so that it could be connected to the bridging donor ligand. Followed by addition of the pyridine linker L1 ligand, the crystalline products 1, 2 were obtained after extraction with methanol/isopropyl ether in yields of 86% and 83%, respectively. Single crystals of sufficient quality for X-ray diffraction were obtained by slow diffusion of isopropyl ether into solutions of both complexes in methanol at ambient temperature over three days.

Analysis of Metallarectangle 1
Single-crystal X-ray structure determination and NMR spectroscopy confirmed the nature of complex 1. As shown in Figure 1, the two binuclear (RhIII) molecular clips E1 are connected by two rigid dipyridyl ligands L1, adopting a rectangular macrocyclic compound 1. The structural analysis of this complex shows short and long Rh-Rh nonbonding distances of 3.66 and 19.17 Å, respectively. The separation between central conjugated planes from two ligand L1 is about 3.62 Å, displaying strong π-π stacking interaction possibly due to the electron delocalization effect of benzimidazole units. In addition to the intramolecular π-π stacking, similar interactions could also be found between these discrete metallarectangle molecules, in which the separation between Cp* group of one metallarectangle molecule and phenyl unit from another metallarectangle molecule is about 3.51 of 3.55 Å, demonstrating that these discrete molecules coexist stably through a series of π-π stacking interactions. The presence of compound 1 in solution was also determined by the 1 H NMR spectra ( Figure 2), along with the combination of 1 H− 1 H COSY NMR and 1 H− 1 H DOSY NMR spectra (Figure 3a). In complex 1, two obvious double peaks could be observed in 8.20 and 8.11 ppm, which could be attributed to two pyridine protons of ligand L1. Meanwhile, a strong singlet could be found in 7.29 ppm, belonging to phenyl protons of ligand L1. Moreover, the peak of the Cp* group proton was located at 1.65 ppm. The 1 H diffusionordered spectroscopy (DOSY) NMR spectrum of 1 clearly indicated that these signals for the aromatic and Cp* units showed a single diffusion coefficient (2.38 × 10 −10 m 2 s −1 ), further confirming stable presence of structure 1 (Figure 3b).  Moreover, the results of IR spectrum of complex 1 exhibits a strong band at 3441 cm −1 owing to the stretching vibration of the N-H bond of imidazole group [40][41][42], the peaks at 1030 and 1279 are likely the asymmetric and symmetric sulfonate stretches and the peaks at 1160 are due to CF 3 asymmetric and symmetric stretches [43][44][45]. The peaks at 1614 and 1444 are likely attributed to the vibration of aromatic ring structures of ligand L1 (Figure 4).

Analysis of Metallarectangle 2
The solid-state structure of complex 2 was also demonstrated by single-crystal X-ray diffraction analysis, which shows a centrosymmetric rectangular macrocyclic compound with four rhodium atoms located at the corners. Here, the short Rh-Rh nonbonding distance is 5.39 Å, which is longer than that in 1 (3.66 Å) due to the larger spacer. The long Rh-Rh nonbonding distance of 18.92 Å is closely comparable to that measured in compound 1. Interestingly, the distance of central conjugated units from two ligands L1 in complex 2 is 3.59 Å, which is even shorter than that in complex 1, reflecting that the central conjugated planes tend to be close to each other for forming π-π stacking interaction ( Figure 5). Therefore, the imidazole units in L1 are not coordinated with the rhodium atom, possibly due to the building blocks E1, E2 providing a suitable distance for the central conjugated planes from two L1 ligands, which is more conducive to forming strong π-π stacking interactions from electron delocalization effects of benzimidazole units. In addition, compared to that in compound 1, the intermolecular force is different. Two sets of obvious π-π stacking interactions could be found between Cp* groups and benzimidazole units in complex 2, accompanied by a distance of 3.70 Å. From Figure 5, we could find that each molecular rectangle 2 is able to form four π-π stacking interactions from another two rectangles, resulting in a stable solid structure.
Moreover, the solution behavior of solid complex was also important for the various performance studies. Therefore, the solution characterization of complex 2 was also explored via the 1 H NMR spectra accompanied by the two-dimensional spectra of 1 H-1 H COSY NMR and 1 H-1 H DOSY NMR (Figures 6 and 7). Similarly, compared with that of complex 1, two obvious multiple peaks 8.18-8.17, 7.64-7.62 ppm bearing clear coupling interaction can be observed in complex 2, which belongs to the phenyl protons of E2 building units. In addition, the signals of two pyridine and one phenyl proton of ligand L1 could be found in 7.87, 7.39, 7.08 ppm, which is pretty close to that in complex 1, albeit with a slight offset. The chemical shift of Cp* proton is at 1.92 ppm. Finally, a single diffusion coefficient from the 1 H-1 H DOSY NMR spectrum (2.45 × 10 −10 m 2 s −1 ) clearly demonstrated the presence of complex 2 (Figure 7b).  Moreover, the IR spectrum of complex 2 was also carefully explored. Similar to complex 1, the strong band at 3474 cm −1 could be observed owing to the stretching vibration of the N-H bond of imidazole group; the stretching vibrations of the OTf − anions appeared at 1255, 1029, and 1163 cm −1 , respectively. Additionally, the peaks at 1612 and 1445 are due to the vibration of aromatic ring of L1. [46][47][48][49][50]. In addition, a strong signal at 1474 cm −1 could be attributed to imidazole groups of building block E2 (Figure 8).

Synthesis and Analysis of Heterometallic Cage 3@2iPr 2 O Based on E3
Then, a longer heterometallic building block [Cp* 2 Rh 2 L Cu ]Cl 2 (L Cu = [Cu(opba)] 2− [opba = o-phenylenebis(oxamato)] (E3) was chosen deliberately to build a larger assembly, where the Rh-Rh nonbonding distance is 10.10 Å, providing enough space to form additional coordination of two E3 ligands through imidazole donors to Rh atoms. Through the self-assembly of ligand L1 and building block E3, a new green centrosymmetric complex 3@2iPr 2 O was obtained in a yield of 82% (Scheme 3). The structure was finally characterized by single-crystal X-ray diffraction, elemental analysis and UV-Vis spectra. Single crystals suitable for X-ray structure determination were obtained by slow vapor diffusion of isopropyl ether into a methanol solution of 3@2iPr 2 O at room temperature, showing that a heterometallic cage was obtained in which two kinds of metal-positive ions, rhodium and copper ions, were included. The single-crystal structure displayed that the short Rh . . . Rh distance are ca. 8.67 Å from the two rhodium atoms coordinated to imidazole sites. In the building block E3, the Rh . . . Rh separation are ca. 10.67 Å. The longest Rh . . . Rh separation of 19.12 Å is similar to the distances found in complexes 1 and 2. Interestingly, two isopropyl ether molecules are included in the host cage as two guest molecules by a series of C-H . . . π interactions, which was demonstrated by the thermogravimetric analysis (TGA). A careful structural analysis of complex 3@2iPr 2 O displayed that each isopropyl ether molecule could form three C-H . . . π interactions with two building blocks E3. The distance between the H proton and the conjugated plane is about 2.79, 2.99 and 2.88 Å, and three interactions stabilize just one isopropyl ether molecules. Additionally, another isopropyl ether was stabilized by the other two building blocks E3 by three C-H . . . π interactions. In addition, the shortest separation between the two isopropyl ether molecules is 3.03 Å. The distance avoids the interference of van der Waals forces between the two isopropyl ether molecules. Therefore, cage 3 can be used as a good host to selectively wrap the isopropyl ether molecules (Figure 9).
Then, a series of tests, such as UV-vis (Figure 10), TGA and IR spectra were performed carefully to explore the structural characteristic of cage 3@2iPr 2 O. The TGA test also showed that a significant loss was found in 30-100 • C, reflecting the loss of isopropyl ether and coordinated H 2 O molecules. Then, when the temperature rose from 100-300 • C, no mass loss was found, showing the structural stability. After that, an obvious weak loss (17.4%) can be observed, which is consistent with the mass of the two ligands L1 (17.1%), suggesting the skeletal collapse of complex 3@2iPr 2 O (Figure 11). The results also reflect a stronger interaction between the host framework and guest molecules contained in a cage than these free isopropyl ether molecules. The UV-vis in solution and solid state and IR spectrum tests both determined the structure of this complex 3@2iPr 2 O (Figure 12). The structural analysis of IR spectrum for complex 3@2iPr 2 O showed a strong band at 3442 cm −1 , attributed to the stretching vibration of the N-H bond of imidazole group. Additionally, the stretching vibrations of the OTf − anions appeared at 1278, 1026, and 1159 cm −1 , respectively [51][52][53][54][55]. Additionally, the vibration of aromatic ring of ligand L1 is likely at 1615 and 1421 cm −1 [56][57][58][59][60]. In addition, a strong signal at 575 cm −1 could be attributed to the Cu-O bonding signal of building block E3 (Figure 12).

The Host−Guest Chemistry Exploration of Heterometallic Cage 3@2iPr 2 O
In addition, when the crystals 3@2iPr 2 O was allowed to vacuum under low pressure for 3h, after that, the isopropyl ether molecules were removed and the cage structure could be maintained integrally, which was demonstrated by the TGA tests ( Figure 13). After that, the empty cage 3 was dissolved in a methanol solution. Then, the isopropyl ether solvents were allowed to diffuse into the above solution. After three days, good green crystals were recovered; the crystals could be determined by the single X-ray diffraction analysis, demonstrating the reformation of host-guest cage 3@2iPr 2 O containing two isopropyl ether molecules. The results reflect that the cage 3@2iPr 2 O packaging isopropyl ether molecules has cyclicity.

Experimental Section Materials
All reagents and solvents were obtained commercially and used without further purification. The starting materials [Cp*RhCl(µ-Cl)] 2 (E1), [Cp* 2 Rh 2 (BiBzlm)](Cl) 2 (E2) and [Cp* 2 Rh 2 (L Cu )]Cl 2 (E3) were prepared according to literature methods. NMR spectra were recorded on Bruker AVANCE I 400 spectrometers at room temperature and referenced to the residual protonated solvent. Elemental analyses were performed on an Elementar Vario EL III analyzer. IR spectra of the solid samples (KBr tablets) in the range 400-4000 cm −1 were recorded on a Nicolet AVATAR-360IR spectrometer.

Conclusions
In summary, two metallarectangles were acquired successfully by a combination of ligand L1 and two Cp*Rh-based building blocks E1 and E2 through a coordination-driven self-assembly strategy. Interestingly, through deliberately choosing a reasonable copper based building unit E3, a novel heterometallic cage was obtained smoothly, reflecting that size effect could affect the formation of different topological complexes. In addition, the cavity of the host cage contained two isopropyl ether molecules by multiple C-H . . . π interactions, showing an obvious host-chemistry effect. Additionally, these complexes were characterized by single-crystal X-ray diffraction analysis, NMR spectra and UV-Vis spectra. We believe that our results may be useful for controlling the synthesis of complicated structures and potential applications of organometallic macrocycles and a heterometallic cage, in addition to encouraging the design of further novel building blocks and functional architectures.