Confined Spaces in [n]Cyclo-2,7-pyrenylenes

Strained aromatic macrocycles have distinct properties compared to their linear counterparts such as shape persistence, enhanced solubility, and the ability to complex guests in their cavities.^1-3 Among them, cycloparaphenylenes (CPPs) have gained much interest in the last decade due to their size-dependent photophysical and electronic properties.^4-7 Several bottom-up strategies have been developed to synthesize CPPs or derivatives containing hetero- or polycyclic aromatic moieties.^8-12 Incorporation of such moieties allows to study new functionality within this class of macrocycles and alter their photophysical and electronic properties.^13-17 Moreover, the developed synthetic methodology can be used to extend the molecular shape from a macrocycle to a cage.^18-21 While molecular cages intrinsically favor guest inclusion, CPPs and related strained carbon macrocycles have been of limited use as supramolecular hosts, most prominently for fullerenes.^22-27 However, a higher degree of functionalization in combination with spatial extension of the cyclic structure could enable turning CPPs into a relevant class of supramolecular hosts.

Regarding their rigidity and their size-dependent UV/Vis absorption and emission, carbon nanorings are unique compared to other supramolecular host structures, such as calix[n]arenes, resorcin[n]arenes, and cucurbit[n]urils.^28-31 Improved guest binding should therefore enhance the applicability of these intrinsically porous and shape persistent molecules in a variety of research areas they are currently investigated for.^{32, 33}

Herein, we report on a series of highly functionalized [n]cyclo-2,7-(4,5,9,10-tetrahydro)pyrenylenes (CPYs) as supramolecular hosts with confined binding cavities. The rich chemistry developed on the pyrene scaffold and the reported [4]cyclo-2,7-pyrenylene by Yamago and co-workers (Figure 1, top) make this polycyclic aromatic hydrocarbon a feasible starting point for highly functionalized aromatic macrocycles.^{34, 35} The molecular shape of the pyrene subunit leads to a π-conjugated concave inner host surface, creating an accessible volume suitable for host–guest interactions.

The K-region of pyrene was symmetrically functionalized via Ru-catalyzed oxidation to obtain pyrene-4,5,9,10-tetrone (5).³⁶ Acid-catalyzed condensation with ethylene glycol readily yielded the wall fragment 4 of the desired macrocyclic structure (Figure 2 a).^{37, 38} We turned to regioselective Ir-catalyzed C−H borylation of 4 to efficiently produce 3.³⁹ Subsequent transmetalation with dichloro(1,5-cyclooctadiene)platinum(II) gave the macrocyclic precursor 2.^{11, 40, 41} This organometallic precursor is accessible in gram-scale within four steps from pyrene without the need of column chromatography. Reductive elimination using triphenylphosphine resulted in the targeted CPYs 1_[n] as a crude reaction mixture of five different macrocycle sizes 1_[4–8] as observed by MALDI-TOF MS (Figure 2 b). The different congeners were separated by recycling gel permeation chromatography (GPC).

Formation of various ring sizes is a common feature, observed in several examples following the platinum-mediated macrocyclization.^42-46 While the origin of this effect is still not fully understood, the result is beneficial in the presented example, as a systematic series of hosts with various sizes are obtained in a one-pot synthesis. During the analysis of the Pt-containing macrocyclic precursor, we have detected exclusively the mass of tetrameric 2 via MALDI-TOF MS (see Figure S21 in the Supporting Information), nevertheless formation of other Pt-ring sizes appears plausible in regard of the literature.^42-46 The present triphenylphosphine leads to the formation of Pt(PPh₃)_4−m, which is known to insert into strain-activated C−C bonds. Dynamic insertion of active Pt(PPh₃)_4−m into the C−C bond between pyrenylene units and subsequent ligand exchange of cyclic fragments can therefore be an explanation for the observed variety of ring sizes.^{12, 47-49}

The ¹H NMR spectra show a single signal in the aromatic region for each macrocycle 1_[n] owing to their average D_nh (n=4–8) symmetry (Figure S8–S19).⁵⁰ Due to the limited solubility of 1_[8] full characterization by NMR was not possible, pointing towards the feasible limitations in terms of ring size for the presented macrocycles. Unambiguous structural proof was obtained by single-crystal X-ray diffraction for 1_[4–6].⁷⁷ Carbon-rich CPP-type macrocycles often exhibit a dense herringbone crystal packing, reducing the intrinsic porosity (Figure S34 and S39 for a detailed analysis).^{33, 51, 52} In contrast, the solid-state structures of 1_[4] and 1_[6] exhibit a slipped–stacked tubular packing presumably caused by the sterically demanding ethylene glycol groups preventing a self-filling assembly in the solid state. Crystals suitable for X-ray analysis of 1_[5] could only be obtained in the presence of a guest. The defined tubular stacking of this structure can therefore not be compared to the other series members. The crystal structure of 1_[6] shows well resolved chloroform molecules, structurally positioned by weak Cl₃C−H⋅⋅⋅O_host hydrogen bonds (3.1–3.3 Å heavy atom distance) and several remarkable C−Cl⋅⋅⋅O_host halogen bonds (2.7–3.3 Å; Figure 2 c, for a detailed analysis on the complex interaction network between 1_[6] and the solvent CHCl₃ see Supporting Information Figure S36–S38).

The UV/Vis absorptions of the series 1_[4–8] show similar absorption maxima λ_max for all macrocycle sizes (Figure 3 and Table 1), which is a typical property for this class of macrocycles originating from a formally Laporte-forbidden optical HOMO–LUMO transition.^{6, 53, 54} Exceptions to this forbidden transition result in a red-shifted shoulder of all absorption maxima (400–450 nm). While the entire series of 1_[4–8] shows rather large apparent Stokes shifts, taking into account the forbidden HOMO–LUMO transition the genuine Stokes shifts are 28–45 nm (0.20–0.31 eV, Table 1). Using time-dependent density functional theory (TD-DFT) the calculated oscillator strengths for the HOMO–LUMO transitions of 1_[4–8] were rather low (f<0.2) verifying the forbidden nature of the HOMO–LUMO transition (Figure S79–S83).

The unique photophysical properties of these CPP-type macrocycles are assigned to their cyclic symmetric structure in combination with the strain necessary to bend their otherwise planar aromatic repeating units. We calculated the strain energy of 1_[4–8] using an homodesmotic approach and also assessed the ring-strain energy with StrainViz (Figure 3 b,c).^{55, 56} The results of these calculations generally follow the same and expected trend for the series 1_[5–8] except for the shortest, most strained tetramer 1_[4], for which the homodesmotic equation calculates a higher strain energy than the results obtained from the StrainViz analysis (Figure 3 b). Strain energies cannot be determined experimentally, yet the calculated results allow for comparison within this set of macrocycles.

A great advantage of StrainViz is the visualization of the strain energy at affected bonds, while classifying the strain energy into bond-, angle-, and dihedral strain energy. For CPPs in general, dihedral strain has the highest impact because an average biphenyl angle of circa 44° cannot be fulfilled between all Clar-sextet units (Figure S84–S88).⁵⁷ Comparison of 1_[4] to the equivalently sized parent [8]CPP revealed an evenly distributed dihedral strain energy on all eight biphenyl units for [8]CPP but a localized dihedral strain energy over only four pyrene units in 1_[4] (visualized by red bonds, Figure 3 c).

Confined spaces within the volumetric centers of 1_[4–8] are established by the radially converging ethylene groups into the cavity gates of the macrocycles. A crucial parameter in this concept is the ratio of the gate diameter formed by the ethylene glycol groups to the cavity diameter in the center of the macrocycle. If the gate diameter becomes very small the energetic barrier to pass the gate of the cavity might be too high, whereas for a large opening the effect of tight binding to the confined space vanishes (see Supporting Information, Section S7).^58-60 These considerations are experimentally supported by our finding that 1_[5] shows no binding with C₆₀ using ¹H NMR spectroscopy (Figure S75), while the structurally similar [10]CPP (with the same amount of phenylene units as 1_[5]) shows very strong binding towards fullerene (K_a=10¹⁰ m⁻¹).⁶¹ Besides that ratio, the cavity volume is the second crucial parameter, which we visualized for each ring size using the MS Roll suite implemented in X-Seed⁶² (Figure 4 and Section S7). Shape and size of this volume preclude several potential guests according to earlier reports on space filling.^{63, 64} While the ratio between the gate diameter and the central diameter for [n]CPPs is 1, this ratio is with 0.64 and 0.67 considerably smaller for 1_[4] and 1_[5] (Table S8). To investigate the influence on confinement we decided to first study the guest properties of 1_[5] since the average gate diameter at the van der Waals surface of 1_[4] was calculated to be 3.65 Å, exhibiting a too small opening into the macrocycle's cavity.

Host-in-host complexes are an intriguing starting point in the construction of more complex superstructures.^65-67 A well-studied example is the encapsulation of crown ethers or their nitrogen-containing equivalents into cages or capsular hosts.^68-72 We selected crown ethers as guests because they are conformationally flexible allowing them to pass the gate of the macrocycle, and being readily available in various sizes. We first studied the binding of several crown ether metal cation complexes to 1_[5] using UV/Vis absorption spectroscopy in CHCl₃ at 25 °C. Addition of 12-crown-4⋅LiCl did not lead to a change in absorption (Figure S61). In contrast, upon addition of 15-crown-5⋅Na(BF₄) a bathochromic shift combined with a narrowing of the absorption band and multiple distinct isosbestic points were observed. The new absorption maximum λ_max of the complex 1_[5]⋅⋅⋅[15-crown-5⋅Na(BF₄)] is red-shifted by 9 nm as compared to 1_[5]. Isothermal UV/Vis absorption titration with addition of up to 54 equivalents of 15-crown-5⋅Na(BF₄) and subsequent non-linear least-square curve fitting afforded an association constant K_a=8.0×10⁴ m⁻¹ (Figure 5 a and Figure 5 b, top). A perfect fit in size is further suggested by DFT calculations (Figure S54 and Table S9). To the best of our knowledge, this is the strongest binding observed for a non-fullerene guest to a CPP or a CPP-related macrocycle.

To get more structural information on the binding we turned to isothermal ¹H NMR binding titration, but the necessity for higher concentrations than in the UV/Vis experiments led to peak broadening and precipitation upon addition of 15-crown-5⋅Na(BF₄) to a solution of 1_[5] in CDCl₃ (Figure S74). To investigate the size-dependency of the binding, we studied the next larger crown ether 18-crown-6 and its K⁺ complexes.

Isothermal UV/Vis absorption titration upon addition of 18-crown-6⋅K(BF₄) (Figure 5 c) gave a bathochromic shift of 5 nm for λ_max, accompanied by a narrowing of the absorption, and several distinct isosbestic points. Non-linear least-square curve fitting of the absorption titration data afforded an association constant of K_a=1.2×10⁴ m⁻¹ (Figure 5 b, bottom). This only slightly lower K_a value compared to 15-crown-5⋅Na⁺ with 1_[5] results from the conformational flexibility of crown ethers. 18-Crown-6 can adapt to the rigid host cavity by bending out one ethylene glycol unit, as confirmed by X-ray crystallography (see below). In contrast, the equally sized but unconfined parent [10]CPP shows significantly smaller binding constants of K_a=103 m⁻¹ and K_a=24 m⁻¹ for 15-crown-5⋅Na(BF₄) and 18-crown-6⋅K(BF₄), respectively (25 °C, CHCl₃, Figure S70–S73). To complete the series, we investigated the binding of the larger dibenzo-18-crown-6—a guest even too large to be included inside 1_[7]—and we observed no binding event (neither for 1_[5] nor for 1_[7], see Figure S59 and Figure S69, respectiveley).

Investigating the host abilities of 1_[6] towards crown ethers, no pronounced binding was observed for the addition of neither 15-crown-5⋅Na(BF₄) (K_a=28 m⁻¹) nor 18-crown-6⋅K(BF₄) (K_a too weak to be detected, Figure S66–S68). We assign this result to the drastic increase of the cavity volume by 23 % from 1_[5] to 1_[6] with the latter therefore requiring much larger guests for reasonable strong binding. Further, no binding event could be determined for 1_[4] with the smallest crown ether 12-crown-4⋅LiCl, which we assign to the already mentioned very small gate opening, presumably requiring much higher energies to overcome this barrier (Figure S56).

Crystals suitable for synchrotron-based X-ray crystallography were obtained for the complex 1_[5]⋅⋅⋅[18-crown-6⋅K(BF₄)] (Figure 5 d). The cationic 18-crown-6⋅K⁺ is located in the center of the cavity of 1_[5] and the ethylene glycol groups of 1_[5] embrace the guest complex as a secondary coordination sphere.⁷² 18-crown-6⋅K⁺ is bound in an unexpected non-planar conformation with one ethylene glycol unit twisted out-of-plane. The energetic penalty of this bent conformation compared to its optimized planar conformation was calculated to be 2.0 kcal mol⁻¹ (DFT:B3LYP-D3/QZVP in vacuum, Figure S45), well compensated by a non-covalent interaction free enthalpy. Measuring the distance between centroids of the phenyl subunits of 1_[5] and the C-atoms of the crown ether shows ten close contacts (3.5–4.0 Å) assigned to C−H⋅⋅⋅π interactions (Figure S43). Several further C−H⋅⋅⋅O hydrogen bonds were determined between the strongly polarized C−H groups of the crown ether as hydrogen bond donors and the oxygens of the ethylene glycol groups of 1_[5] acting as hydrogen bond acceptors (Figure S45). The crystal structure shows a tubular packing of 1_[5], where the void space is filled with 18-crown-6⋅K⁺ and the BF₄⁻ anion. These columnar stacks form a hexagonal circle-packing motif (Figure S42).^{73, 74}

In summary, we successfully synthesized a series of five [n]cyclo-2,7-(4,5,9,10-tetrahydro)pyrenylenes 1_[4–8] carrying ethylene glycol groups at the outer rim to create a confined space in the center of the macrocycles. The effect of confinement was demonstrated by a significantly weaker binding of 15-crown-5⋅Na(BF₄) in [10]CPP or 18-crown-6⋅K(BF₄) in [10]CPP. We assign the improved binding to the confined space of 1_[5] and the additional hydrogen bond acceptors created by the introduction of the dipolar ethylene glycol groups to form a cooperative interaction network.⁷⁵ The observed association constant of 15-crown-5⋅Na(BF4) to 1_[5] is the largest of a non-fullerene guest to a CPP-type macrocycle, thereby illustrating the potential of our reported extension strategy to develop these highly fluorescent and shape persistent macrocycles into a new class of supramolecular hosts beyond fullerene guests.^{61, 76}