Synthesis and Characterization

The synthesis of nanocone 3 starts with tetrahydroxycalixarene 4,^[38] which is fixed in the cone conformation by having four n-propoxy chains at the lower rim (Scheme 1). The first step was the methylation of the phenolic oxygens with methyliodide to give calixarene 5 in 60% yield. Subsequent bromination with NBS by adopting a previously reported protocol on aminocalixarenes,^[39] gave a mixture of fourfold brominated calixarenes 6–8, having one bromo substituent per aromatic ring. After column chromatography, these could be isolated in 53% (6), 24% (7), and 22% (8) yield, and their constitutional structures were unambiguously proven by NMR spectroscopy and single crystal X-ray diffraction (Figure 2).^[40]

C_2h-symmetric tetrabromocalixarene 9 was not isolated but probably has formed in traces together with 8 (see NMR discussion below). In solution as well as in the solid state, the C_2v-symmetric pinched cone conformation is preferred in comparison to a C_4v-symmetric cone conformation, and the latter is the transition state between two pinched cone conformations.^[41] At room temperature for most calix[4]arenes, the energy of the transition is low, and on the NMR time scale, an averaged signal set is observed.^[42] By temperature-dependent NMR measurements the coalescence of the transition and thus the correlated energy can be determined. At room temperature, the signals of C₄-symmetric tetrabromo calixarene 6 are broad, revealing that interconversion between two pinched cone conformations is relatively slow on the NMR timescale (at T_c = 263 K coalescence is found, which corresponds to ΔG = 11.9 kcal mol⁻¹). C_2h-symmetric tetrabromo calixarene 7 already shows at room temperature very sharp signals of a pinched-cone conformation. Even at 140 °C coalescence was not reached, which shows that the transition barrier is higher than ΔG > 20.3 kcal mol⁻¹. At first glance, the ¹H NMR spectrum of tetrabromo compound 8 appears impure, with several small signals aside from the main peaks. It must be noted that the two C₁-symmetric pinched cone conformations are diastereoisomers, and by integration of the NMR peaks, a ratio of ∼ 10:1 toward the isomer 8a is found, revealing that this one is thermodynamically more favored by about 1.4 kcal mol⁻¹ in comparison to 8b. This corroborates with PM6 based calculations (ΔG_f = 1.0 kcal mol⁻¹). By temperature-dependent NMR spectroscopy in CD₂Cl₂, a non-linear behavior of pinched-cone to pinched-cone transformation (8a to 8b) was found with ΔG^‡_298K = 4.0 kcal mol⁻¹. It is worth mentioning that the small peaks at 5.61 and 6.75 ppm (denoted x and y) probably stem from minor impurities of calixarene 9, which we were not able to remove. According to the ratio of integrals, this is formed overall in less than 0.6% in the reaction.

To synthesize calix[4]nanocone 3 (Scheme 2), the tetrabromo calixarene 6 was used in a Suzuki–Miyaura cross-coupling with ethoxy vinyl boronic pinacol ester 10 to give 11 in 70% yield. The tetravinyl ether is very unstable and needs directly to be converted to nanocone 3 with Bi(OTf)₃ under high dilution conditions (approx. 1 mg mL⁻¹) in dry DCM.^[43] After purification, nanocone 3 was isolated in 33% yield, besides traces (5%) of partially closed nanocone 12 with a missing propylchain. The olefinic double bonds of nanocone 3 can be hydrogenated with Pd/C (5 wt%) to give reduced nanocone 13 in 96% yield.

Both nanocone 3 and reduced nanocone 13 were clearly identified by MS, NMR spectroscopy, and single crystal X-ray diffraction. Nanocone 3 shows an m/z peak for [M]^∙+ at 808.3973 (calc. 808.3975). It is worth noting that the basis peak is found at m/z 831.3869, which corresponds to [M+Na]⁺, giving already a hint that the tendency of this compound toward binding of sodium ions is probably high. The reduced nanocone 13 has eight more atomic units and the basis peak at m/z 816.4567 fits to the calculated one at m/z 816.4601. Here, the corresponding [M+Na]⁺ is also found (m/z 839.4494), however less pronounced. Both ¹H NMR spectra are rather simple (Figure 3). Besides the typical pattern of the propoxy chains, the bridging methylene protons are still diagnostic of the cone conformation and appear as two doublets at δ = 4.70 and 2.79 ppm, each with geminal coupling of J = 11.4 Hz for 3 and at δ = 4.52 and 3.52 ppm with J = 12.8 Hz for 13. The vinylene bridge protons of 3 resonate as a singlet at δ = 7.07 ppm, whereas in the reduced nanocone 13, the corresponding ethylene bridges give two multiplets at δ = 3.06 and 2.91 ppm with an integral 8 each for the inner and outer protons.

Single crystals of nanocone 3 were grown by slow evaporation of solvent from a saturated solution of the compound in acetonitrile (Figure 4). The compound crystallizes in the monoclinic space group C2/c. Due to the flexible alkoxy chains, the molecule is C₁-symmetric. The vinylene bridges are 1.33 Å long, and the six-membered rings have bonds that are non-alternating in length, as it is typical for aromatic benzene derivatives. More interesting is the dimension of the nanocone, especially the distances of the oxygens at the upper and lower rims. Whereas the overall dimensions of the backbone are comparable to the nanocone reported by Wang and coworkers,^[29] the distances between distal-oriented oxygens are at the upper rim 8.24 and 8.32 Å and at the lower rim 4.62 and 4.94 Å (average: 4.78 Å). The latter two seem ideal for complexation of lithium or sodium ions, because they are in the range of the distances calculated from van der Waals and ion radii of oxygen and the ions (d_{(O∙∙∙Li+∙∙∙O)} = 4.56 Å and d_{(O∙∙∙Na+∙∙∙O}) = 5.08 Å.^[44]

By reduction of the double bonds, as done to achieve 13, the structure becomes more flexible, leading to larger alternation of distances between opposite atoms or groups. Single crystals of reduced nanocone 13 were grown from evaporation of dichloromethane from a saturated solution, and the compound crystallized in the triclinic space group P-1. The distances between distal-oriented oxygens are at the upper rim 7.59 and 8.40 Å and at the lower rim 4.61 and 5.01 Å (average: 4.81 Å). The reduction of the vinylene bridges to ethylene bridges has been proven by their elongation from 1.33 Å in 3 to 1.51 Å in 13. Comparison of electron density maps of calixarene 5 with nanocones 3 and 13 (Figure 5) clearly shows that in the latter two, the highest electron densities are located in the center of the plane between the oxygens at the lower rim, which should be beneficial for hosting the smaller alkaline ions.

Complexation Studies

We were interested, if the decrease of structural flexibility and the nearly ideal distances between the oxygens at the lower rim influence the complexation of alkaline metal ions Li⁺, Na⁺, and K⁺ also in comparison to the flexible calixarene 5, which can in principle adopt any preferred conformation, but 3 and 13 cannot. To a solution of nanocone 3 in CDCl₃, solid LiOTf was added and sonicated. LiOTf was fully complexed. Please note that without nanocone 3, the triflate is nearly insoluble in CDCl₃. The same occurs with NaOTf and KOTf. The ¹H NMR spectra of the complexes show that all peaks of the nanocone are significantly shifted (Figure 6). The protons most affected by cation complexation are those of the methylene units attached to the oxygens of the propoxy chains (H^d).

These shift from δ = 3.83 ppm in nanocone 3 to δ = 4.47 ppm in Li⁺-complex Li⁺⊂3, δ = 4.23 ppm in Na⁺-complex Na⁺⊂3, and δ = 4.01 ppm in K⁺-complex K⁺⊂3. Reduced nanocone 13 behaved very similarly and dissolved any alkaline triflates in CDCl₃ solutions thereof (for details, see Supporting Information), forming complexes Li⁺⊂13, Na⁺⊂13, and K⁺⊂13. From ¹H,¹⁹F HOESY NMR measurements it is assumed that in solution the triflate anion is dissociated since no correlation peak of the nanocone with the triflate was found. Furthermore, another salt (NaClO₄) gave exactly the same shifts of the methylene protons H^d, once more suggesting that no contact ion pairs are formed.

From (Li⁺⊂3)OTf⁻, (Na⁺⊂3)OTf⁻, and (K⁺⊂3)OTf⁻, single crystals of high quality were obtained from CDCl₃ by evaporation of solvent (Figure 7). (Li⁺⊂3)OTf⁻, crystallizes in the space group P4/n. The lithium cation is centered between the four oxygens of the propoxy chains of the lower rim at a distance of d = 2.22–2.23 Å. The lithium is placed only 0.39 Å below the plane that the four oxygen atoms build. The triflate is bound as a fifth ligand to the lithium at a distance of 1.90 Å. (Na⁺⊂3)OTf⁻ crystallizes in the space group P21/c. Here, the distance between the alkaline metal ion and the oxygens is d = 2.34 and 2.37 Å somewhat larger, and the distance to the triflate is with 2.17 Å also larger. Most important, the sodium cation is found at 0.41 Å below the plane that is formed by the oxygens. This is only somewhat lower than for the lithium, so it is assumed that both cations fit well to the preorganized host molecules. This situation is different for the larger potassium ion. (K⁺⊂3)OTf⁻ crystallized in the spacegroup Pnma, the distance between K⁺ and the oxygens is with d = 2.67 and 2.71 Å substantially larger than in the two other complexes. The cation is placed 1.06 Å below the plane that is formed by the oxygen atoms, which is substantially longer than for the other cations. The triflate is acting as a bifurcating ligand with a distance of 2.83 Å between the two oxygens. (Li⁺⊂13)OTf⁻ shows very similar distances (d_Li∙∙∙O = 2.23 Å) compared to (Li⁺⊂3)OTf⁻ (for details see Supporting Information).

We were interested in how well and selectively the nanocones bind alkaline cations in solution and how important the rigidity and pre-organization of the alkoxy oxygens at the lower rim is.^[45] Although the used alkaline triflates are nearly insoluble in CDCl₃ (all < 0.015 mmolL⁻¹), as soon as nanocone 3 is present, the salts dissolve immediately and complexes Li⁺⊂3, Na⁺⊂3, and K⁺⊂3 are formed quantitatively. By adding understoichiometric amounts of triflates, we tried to determine the association constants, but again, all triflate is bound within the nanocone 3, assuming that the binding constants K_a,_CDCl3 exceeded >10⁵ M⁻¹. Reduced nanocone 13 behaved very much the same and was therefore not further investigated in detail.

The electronic situation of more flexible calixarene 5 is comparable to nanocones 3 and 13 and gives insight about the impact of ligand preorganization.^[46] Here, although triflates were added in large excesses, the formation of Li⁺⊂5 never exceeded 95%, and Na⁺⊂5 never exceeded 84%. K⁺ is not taken up at all by calixarene 5.

In the experiments with understoichiometric amounts of salt, two well-separated sets of peaks were found in the NMR spectra, suggesting that the exchange is slow on the NMR time scale. This allowed us to clearly distinguish Li⁺⊂3, Na⁺⊂3, and K⁺⊂3 and nanocone 3 and quantify the equilibrium concentration of those compounds. Therefore, 1:1:0.8 mixtures of two different alkaline triflates and the nanocone 3 were mixed, and NMR spectra were measured until they reached equilibrium (no further change in relation of integrals). In the mixture of LiOTf, NaOTf, and nanocone 3, exclusively Na⁺ was taken up, which means that this one is at least 100 times better bound than Li⁺. A similarly observation toward the Na⁺ in favor of K⁺ was made with their triflates: Only traces of K⁺⊂3 were found in the ¹H NMR spectrum with integrals below the detection limit (< 5%). When LiOTf, KOTf, and nanocone 3 were mixed 1:1:0.8, the equilibrium showed a ratio of 45% of Li⁺⊂3 and 55% of K⁺⊂3.

The advantage of ligand preorganization to the binding of Na⁺ was confirmed by competition experiments of CDCl₃ solutions of pure (Na⁺⊂3)OTf⁻, treated with defined amounts of calix[4]arene 5. Here an equilibrium constant of K_e = 2.2∙10⁴ in favor of Na⁺⊂3 was determined. We also compared the nanocone 3 with some of the best ligands for Na⁺ ions,^[45] namely 18-crown-6^{[47, 48]} and Lehn's cryptand 2.2.1.^{[49, 50]} When adding stoichiometric amounts of cryptand 2.2.1, all Na⁺ is immediately removed, and only signals of free nanocone 3 are observed in the ¹H NMR spectrum and none of the complex Na⁺⊂3, clearly demonstrating that the cryptand has an association constant several orders of magnitude higher than the nanocone 3. With 18-crown-6, the nanocone binds Na⁺ better with K_e = 3.2. By the knowledge of the association constant of 18-crown-6 for Na⁺ in CDCl₃ (K_a = 1.3·10⁶ M⁻¹)^[51] this results in a K_a = 4.2·10⁶ M⁻¹ for binding sodium by nanocone 3.

THF is known to be a good solvent/ligand for alkaline cations. Therefore, complexation studies in THF-d8 were performed too. With the exception of KOTf, the other two salts were very good soluble. Nanocone 3 binds Na⁺ with an association constant of 2.5·10³ M⁻¹ and again, two signal sets are found in the NMR spectrum, revealing a slow exchange on the NMR time scale. Although KOTf is badly soluble, binding is confirmed by two signal sets and K⁺ is bound only with K_a ∼ 150 M⁻¹. With LiOTf in THF, neither a second signal set nor a significant shift of NMR signals was observed, suggesting that either Li⁺ in THF is weakly bound or its exchange is very fast on the NMR time scale at room temperature. Even if an NMR sample with 100 equiv. of LiOTf was cooled down to −100 °C, no splitting of signal sets was observed. Nevertheless, we did an NMR titration with LiOTf, treated as if it binds by fast exchange. The change in shift over the course of titrations is very small (Δδ_max(H^a) = +0.002 ppm) and in the range of the standard deviation (σ_δ(Ha) = ± 0.001 ppm), which is why the discussion of the results needs to be understood as a trend only. Within six NMR titrations of two series (each measured three times), a mean binding constant of K_a = 15.2 M⁻¹ ± 13.5 was estimated. The large error is representative of the large uncertainty of the binding constant, resulting in a conservatively estimated selectivity of S_Na/K > 165). Calixarene 5 in contrast, binds all cations (K_ass,Li < 10⁻² M⁻¹; K_ass,Na = 1.35 M⁻¹, K_ass,K < 10⁻² M⁻¹) very weakly in THF-d8, once more emphasizing the beneficial pre-orientation of the oxygens for the complexation process as it is typically found for spherands and cryptospherands.^[46]