Tunable Porosities and Shapes of Fullerene-Like Spheres

The formation of reversible switchable nanostructures monitored by solution and solid-state methods is still a challenge in supramolecular chemistry. By a comprehensive solid state and solution study we demonstrate the potential of the fivefold symmetrical building block of pentaphosphaferrocene in combination with CuI halides to switch between spheres of different porosity and shape. With increasing amount of CuX, the structures of the formed supramolecules change from incomplete to complete spherically shaped fullerene-like assemblies possessing an Ih-C80 topology at one side and to a tetrahedral-structured aggregate at the other. In the solid state, the formed nano-sized aggregates reach an outer diameter of 3.14 and 3.56 nm, respectively. This feature is used to reversibly encapsulate and release guest molecules in solution.

Different molar ratios of CuCl also influence the completeness of the supramolecules: A deficit of CuCl (e.g. 3 mg, 0.03 mmol) leads to more incomplete spheres (2d-Cl).
Analytical data of [{Cp Bn Fe(μ6-η 5 :η 1 :η 1 :η 1 :η 1 :η 1 -P5)}12{CuBr}20-n] (2b-d): In a schlenk tube complex 1 (100 mg, 0.14 mmol) was dissolved in CH2Cl2 (25 mL) to give an intensive green solution and added to a colorless solution of CuBr (110 mg, 0.78 mmol) in 4 mL CH3CN. The reaction mixture immediately turns red-brown and was stirred for two further hours. Afterwards first 2 mL of a solvent mixture of CH2Cl2/CH3CN (2:1) and subsequently 20 mL of toluene were layered on top of it. The reaction mixture was allowed to stand in an undisturbed area at r.t.. Within two days the crystallization process starts with the formation of small red rods of 3 at the phase boundary. After complete diffusion the almost completely discoloured mother liquor decanted, the crystals were washed with hexane (4 x 5 mL) and dried under vacuum at r.t. to afford 180 mg (0.011 mmol, 96%) of 3.
A higher molar ratio of CuBr (e.g. 6 eq.; 120 mg) referred to 1 leads to the crystallization of a small amount of CuBr besides 3, a lower molar ratio of CuBr (e.g. 5 eq.; 100 mg) at the other hand leads to the crystallization of 2 besides 3. Therefore, the molar ratio of 5.5 equivalents CuBr (110 mg) proved to be the ideal stoichiometry for the selective crystallization of 3. CuK radiation ( = 1.54178Å) using ω scans of 1 (0.5 in the case of 3) frames. Absorption corrections were applied analytically from crystal faces using CrysAlisPro software. The structure 2b was solved by direct methods and refined by full-matrix least-squares method against F  2 in anisotropic approximation using SHELX97 programs set (see Table S). The structures 2a-d were found to be isostructural and therefore were refined using the earlier obtained structural model. All non-hydrogen atoms were refined anisotropically. Hydrogen atoms were refined as riding on pivot atoms.

NMR Experiments in Solution
NMR spectra were recorded on a Bruker Avance 400 MHz spectrometer equipped with a BBO probe with z-gradient and BVT 2000 temperature control unit at 300 K, a Bruker Avance III 600 MHz spectrometer equipped with a TBI 1 H/ 31 P-BB z-gradient probe and BVT 3000 unit at 300 K and an Avance III 600 MHz spectrometer equipped with a 5 mm TCI z-gradient cryo 17 probe and BVT 3000 unit at 298 K. The spectra were processed with the Bruker program Topspin® and the diffusion coefficients were calculated with the Bruker software T1/T2 package. All experimental diffusion coefficients were within a standard deviation of ±3 % and are stated as temperature-and viscosity-corrected diffusion coefficients in the manuscript.
For the calibration of the 31 P chemical shifts, the Ξ value corresponding to TMS was applied. For the calibration of the 1 H and 13 C chemical shifts and for the temperature-and viscosity-correction of the diffusion coefficients, TMS (tetramethylsilane) was added to each sample.

CuX (X = Cl, Br) titration in CD2Cl2 with 9 vol% CD3CN
By changing the amount of CuX the formation trends of the different complexes in solution can be clearly followed. The experiment with CuBr is shown in Fig. S 5 and explained in more detail here, the one with CuCl is illustrated in Fig. S 6. As shown in Fig. S 5a, without CuBr only 1 was present in solution. Upon the addition of 0.5 CuBr (see Fig. S 5b), several other broad and overlapping signals in the 1 H spectrum (6.80 ppm for Hpara and 6.40-6.60 ppm for Hortho and Hmeta and at least three signals at 5.10, 4.84 and 4.67 ppm for the methylene protons (data not shown)) were detected, which all showed the same 1 H diffusion coefficient (see Table S 7). In the 31 P spectrum, at least three very broad signals in the range of 65-125 ppm were of CuBr: a) 0.0 eq., b) 0.5 eq., c) 1 eq., d) 1.7 eq., e) 4 eq., f) 8 eq., each in CD2Cl2 with 9 vol% CD3CN. 1 H spectra at 298 K and 600 MHz, 31 P spectra at 300 K and 400 MHz. of CuCl: a) 0.0 eq., b) 0.5 eq., c) 1 eq., d) 1.7 eq., e) 4 eq., f) 8 eq., each in CD2Cl2 with 9 vol% CD3CN at 300 K and 400 MHz.

Pure CD2Cl2: CuBr titration and dissolving crystals 2a-Br and 2b-d-Br
The spectra in pure CD2Cl2 confirm these assignments and especially the assignment of 2aCD2Cl2. The solubility of CuBr in pure CD2Cl2 is quite low, therefore the samples were prepared by stirring a solution of 1 over solid CuBr for two hours or one, two, five and ten days and subsequent filtration (whereby removing CuBr stopped the reaction). The resulting 1 H and 31 P spectra are shown in Fig. S 7a-e. In case of the CuBr titration mainly the 31 P spectra are discussed due to the severe chemical shift overlap of the supramolecules in the 1 H spectra.
The low solubility of CuBr in pure CD2Cl2 is directly reflected in the slow formation of the supramolecules. After two hours (see Fig. S 7a), beside 1 only several species of 2b-d were detected showing several subspecies in the range of 65-125 ppm in the 31 P spectrum. After one day (see Fig. S 7b), 1 was completely consumed. The distribution of the 2b-d species is clearly shifted to those with higher copper contents (species at higher field increase at the expense of 22 those at lower field). In addition, the sharp signal of 2aCD2Cl2 appeared. After two days (see Fig. S 7c) the amount of 2aCD2Cl2 further increased at the expense of 2b-d and a small signal C was detected, which could not be assigned so far. At longer reaction times (see Fig. S 7d and e), just a further small shift between 2b-d and 2aCD2Cl2 was observed. C remained constant and 3 requiring small amounts of CD3CN was not detected.

Acetonitrile titrations to 2a and 3 crystals dissolved in CD2Cl2
The     In the corresponding NOESY spectrum, the signals of 1 were in the extreme narrowing limit (negative sign of the cross peaks) indicating a fast reorientation typical for small molecules. In accordance with the DOSY results, the signals of 2b-d were in the slow tumbling limit (positive cross peaks) indicating a large hydrodynamic radius. Furthermore, exchange peaks between 1 and 2b-d were detected in the NOESY/ROESY spectra of this sample. 28 The DOSY data, the sum of the NMR spectra and the crystallographic data assemble now to a picture of the formation and the relative stabilities of these supramolecules. In the 31 P spectra always a distinct set of signals were detected for 2b-d with some intensity variations depending on the actual CuBr concentration. The DOSY data of the averaged 1 H signal of several 2b-d species show a large hydrodynamic volume 15 times larger than that of 1 (even in the presence of chemical exchange with 1 leading usually to apparently faster diffusion) and in the crystals all the time the same scaffold with different occupation of the CuBr sites is observed. These data suggest a preferred formation and a high stability of the 2b-d scaffold in solution and speak clearly against a continuous stepwise formation of this supramolecular scaffold.

Encapsulation of guest molecules
First the encapsulation of ferrocene was tested in a solvent mixtures providing 2aCD3CN and 3 (1 and 2 eq. CuBr in CD2Cl2 with about 3 vol% CD3CN) and indeed, a small signal at 0.66 ppm was detected indicating the encapsulation of ferrocene in 2aCD3CN (see Fig. S 12b). No NOESY cross peaks from 0.66 ppm to the signals 2aCD2Cl2 or 3 were observed. Please be aware that due to the dilution procedure the relative intensities of the guests have to be corrected by the absolute intensities of the guest molecules. Therefore, a direct "vertical" comparison is misleading (e.g. the percentage of encapsulated guests is similar in j) and f)). 31 Together with the encapsulation also a significant shift from 2aCD2Cl2 to 2aCD3CN was observed (see Fig. S 13). However, from these experiments, it was not clear whether this might be attributed to small variations of the amount of CD3CN in the sample with ferrocene or whether 2aCD2Cl2 is not able to encapsulate ferrocene. Therefore, a titration row with increasing amounts of CD3CN and ferrocene was measured (see Fig. S 13; 1, 2.0 eq. CuBr, 0.25 eq. ferrocene in CD2Cl2 with ca. 1 vol% CD3CN stirred for several hours). In this setup, a potential crystallization of one of the supramolecules and, thus, potential shift of the thermodynamic equilibrium was avoided by the use of relatively low concentrations and could be finally excluded by the use of an external NMR standard. Therefore, all changes in the relative amounts of the supramolecules are clearly attributable to a transformation of the clusters into one another.
With 1 vol% CD3CN as expected a very high amount of 2aCD2Cl2 was detected beside 2aCD3CN and 3. In addition, three small singlets at 0.66, 0.31 and 0.13 ppm were detected (see ppm to Cp2Fe@2aCH2Cl2 and of 0.13 ppm to Cp2Fe@2b-dCH2Cl2. To test this, CuBr was added (see Fig. S 12b and c). Indeed, as expected the signal of Cp2Fe@2b-dCH2Cl2 at 0.13 ppm decreased strongly, while that of Cp2Fe@2aCH2Cl2 at 0.31 ppm increased together with that of Cp2Fe@2aCD3CN at 0.66 ppm most probably due to a slightly higher amount of CD3CN.
Next the reversibility of the encapsulation was tested by reducing again the amount of CD3CN as well as adding again 1 and CuBr to compensate the dilution effect (see Fig. S 13). The signals of the encapsulated ferrocene Cp2Fe@2aCD3CN at 0.66 ppm, Cp2Fe@2aCH2Cl2 at 0.31 and Cp2Fe@2b-dCH2Cl2 at 0.13 ppm follow exactly the formation trends of the respective supramolecules (for Cp2Fe@2b-dCH2Cl2 a species close to Cp2Fe@2aCH2Cl2 seems to be 33 required because spectra S 12 and S 14 did not show any encapsulation signals of these species) and the spectra show a complete reversibility in this cyclus. Note, that a release of ferrocene is also enabled by the switch of Cp2Fe@2-Br to 3. To do so, crystals of Cp2Fe@2-Br are dissolved in CH2Cl2 and 9 vol% CH3CN and an excess of CuBr is added. Layering with toluene leads to a quantitative crystallization of the supramolecule 3, which has no accessible void for the guest molecule, while ferrocene remains in solution.