Fullerene-Functionalized Halogen-Bonding Heteroditopic Hosts for Ion-Pair Recognition

Abstract

Despite their hydrophobic surfaces with localized π-holes and rigid well-defined architectures providing a scaffold for preorganizing binding motifs, fullerenes remain unexplored as potential supramolecular host platforms for the recognition of anions. Herein, we present the first example of the rational design, synthesis, and unique recognition properties of novel fullerene-functionalized halogen-bonding (XB) heteroditopic ion-pair receptors containing cation and anion binding domains spatially separated by C₆₀. Fullerene spatial separation of the XB donors and the crown ether complexed potassium cation resulted in a rare example of an artificial receptor containing two anion binding sites with opposing preferences for hard and soft halides. Importantly, the incorporation of the C₆₀ motif into the heteroditopic receptor structure has a significant effect on the halide binding selectivity, which is further amplified upon K⁺ cation binding. The potassium cation complexed fullerene-based receptors exhibit enhanced selectivity for the soft polarizable iodide ion which is assisted by the C₆₀ scaffold preorganizing the potent XB-based binding domains, anion−π interactions, and the exceptional polarizability of the fullerene moiety, as evidenced from DFT calculations. These observations serve to highlight the unique properties of fullerene surfaces for proximal charged guest binding with potential applications in construction of selective molecular sensors and modulating the properties of solar cell devices.

Introduction

Fullerenes are unique molecules with a large spherical surface, strong electron acceptor properties, high electric polarizability, and curved π-electron systems capable of forming noncovalent interactions with electron-rich molecules.¹⁻¹³ As a result, they have been incorporated in numerous functional molecular assemblies and supramolecular arrays with a wide variety of applications in photochemistry, medicinal chemistry, and organic electronics.¹⁴⁻²⁶ Although molecular electrostatic potential (MEP) surfaces of simple fullerenes are positive, surprisingly, their interaction with anions has been largely overlooked.^6,27 Only recently, Matile and co-workers demonstrated remarkable examples of the stabilization of anionic transition states in anion−π catalysis on a fullerene surface,²⁸⁻³⁰ while Lei and co-workers reported facilitated charge transfer in solid-state aggregates of self-n-doped fullerene ammonium iodide, which is believed to be a result of iodide–C₆₀ interactions.³¹ However, thus far fullerene surfaces have not been exploited as potential supramolecular host platforms for the recognition of simple anions (e.g., halides) in molecular receptor structural design. Nevertheless, it is worth noting that an open-cage fullerene was demonstrated as a molecular container for F^–, Cl^–, Br^–, and I^–.³²

The MEP surface of C₆₀ reveals highly localized areas of positive potential, π-holes (Figure 1), which can be presumably used in anion recognition. Anion−π interactions are widely recognized and frequently exploited in the design of selective anion binding receptors.³³⁻³⁶ Strong attraction between an anion and a π-system can be achieved by electron-withdrawing substituents that further polarize the molecule and lead to a positive quadrupole moment along the axis perpendicular to the π-system, increasing the depth of a π-hole.³³ Interestingly, fullerenes are known for their remarkable polarizability³⁷ which, in principle, may enable a significant enhancement of anion−π interactions by exposure to an external electric field, produced for example by a proximate anion (so-called dynamic contribution) or cation. Moreover, the well-defined bulky architecture of fullerenes provides a potential scaffold for preorganization of binding motifs and a hydrophobic shield, which can create a microenvironment that excludes solvent molecules and enhances strength of noncovalent interactions.^38,39 Due to these features, C₆₀ constitutes an exceptional and unexplored platform for the design and construction of heteroditopic ion-pair receptors with increased affinity and selectivity. The positive cooperativity associated with the simultaneous proximal binding of oppositely charged species has been crucial in augmenting the ion-pair binding properties of heteroditopic receptors relative to their monotopic receptor counterparts. As such, heteroditopic receptors have been increasingly employed in a myriad of applications including salt extraction and solubilization,^40,41 membrane transport,^42,43 and biological zwitterion binding.^44,45

Figure 1.

Molecular electrostatic potential (MEP) surfaces of C₆₀ (left), C₆₀ functionalized with methylene (center left) or two cyano electron-withdrawing groups (center right), and C₆₀ in association with naphthalene diimide (right). The MEP surfaces are rendered at the 0.001 electrons Bohr^–3 contour and the π-holes on the surface of C₆₀ are identified as black dots.

The vast majority of reported heteroditopic receptors contain well-established recognition motifs such as crown ethers for cation recognition and hydrogen bond donors for anion complexation.⁴⁶⁻⁴⁸ In recent years, however, halogen bonding (XB), an interaction between a Lewis base and the σ-hole of an electron-deficient halogen atom, has emerged as a powerful addition to the anion supramolecular host–guest chemistry toolbox, due to its stringent linearity, comparable binding strength to hydrogen bonding (HB), and distinctive selectivity.⁴⁹⁻⁵² Herein, we describe for the first time the rational design, synthesis, and unique recognition properties of novel fullerene-functionalized halogen-bonding heteroditopic ion-pair receptors containing cation and anion binding domains spatially separated by C₆₀ (Figure 2).

Figure 2.

Cartoon representation of a tetradentate XB fullerene-functionalized heteroditopic host for ion-pair recognition. Purple spheres represent iodine halogen bond donors.

Results and Discussion

Synthesis of Fullerene Heteroditopic Ion-Pair Receptors

Interest in crown ether–fullerene adduct materials has been stimulated primarily by their photophysical, electrochemical, and superconducting properties.⁵³ In particular, Echegoyen, Pretsch, Diederich, and co-workers obtained the C₆₀-dibenzo-18-crown-6 (DB18C6) adduct in a highly regioselective double cyclopropanation (Bingel addition) taking place exclusively in the trans-1 positions on the opposite poles of C₆₀ (Figure 3a).^54,55 Potassium cation crown ether binding in the proximity of the fullerene surface was shown to elicit significant perturbations of the fullerene host’s reduction potentials, proving that alkali metal cation complexation can alter the physicochemical properties of C₆₀. We hypothesized that a complexed cation could further polarize the fullerene surface, resulting in anion binding enhancement on the opposite size of the molecule (Figure 2). Therefore, we adapted the regioselective DB18C6 double Bingel fullerene addition for the synthesis of heteroditopic ion-pair receptors 1 and 2 containing neutral acyclic XB donors based on 1,3-bis(iodotriazole)nitroaryl motifs in the anion binding domains,⁵⁶ spatially separated from the polarizable fullerene surface by linkers of different lengths (Figure 3b).

Figure 3.

(a) C₆₀-dibenzo-18-crown-6 adduct obtained by Echegoyen, Pretsch, Diederich, and co-workers. (b) Heteroditopic ion-pair receptors 1 and 2 with different linkers separating anion binding domains from the C₆₀ surface.

The separate appropriately functionalized crown ether–fullerene cation and halogen-bonding anion binding domain synthons were prepared according to Schemes 1 and 2. The synthesis of the bis-alkyne appended crown ether–fullerene synthon 9 was achieved via modification of the regioselective procedure reported by Diederich and co-workers (Scheme 1).⁵⁴ 3,4-Dihydroxybenzaldehyde 3 was alkylated with an excess of bis(2-chloroethyl) ether to obtain 4 (21%), which could be readily separated from the other regioisomer. Macrocyclization of 4 in the presence of the K⁺ template afforded the poorly soluble trans-dialdehyde of DB18C6 5 (31%), which upon reduction using NaBH₄ gave the diol 6 (62%). Monomalonate 7 was prepared either by treating 4-pentyn-1-ol with Meldrum’s acid (68%) or via an alternative approach involving selective monohydrolysis of a symmetric malonic ester (see the Supporting Information for details). Diol 6 was coupled with excess 7 using EDC to obtain bis-malonate ester crown ether 8 (76%). Bingel reaction of 8 with C₆₀ in the presence of K⁺ and I₂ led exclusively to doubly substituted fullerene adduct 9 (25%). The ¹H NMR and ¹³C NMR spectra of 9 (Figures S5 and S6)were in agreement with the trans-1 addition pattern (C₂ symmetry), which was later unambiguously confirmed by single crystal X-ray diffraction structural analysis (Figure 4).⁵⁷ Solid-state analysis also revealed that the DB18C6 ester groups of the cyclopropane rings are situated on the same side of the fullerene (out–out isomer). Interestingly, rotation of the crown ether moiety is significantly limited, causing planar chirality and splitting of the benzylic and ether CH₂ signals in the ¹H NMR spectrum (see Figure S5).

Scheme 1. Synthesis of C₆₀-DB18C6 Adduct 9.

Reagents and conditions: (a) (ClCH₂)₂O, K₂CO₃, DMF, 80 °C, 36 h, 21%; (b) K₂CO₃, DMF, 80 °C, 24 h, 31%; (c) NaBH₄, THF/MeOH 9:1, 0 °C, 2 h, 62%; (d) EDC·HCl, DMAP, KPF₆, CH₂Cl₂/CH₃CN 4:1, 0 °C, 48 h, 76%; (e) C₆₀, I₂, DBU, KPF₆, toluene, RT, 6 h, 25%.

Scheme 2. Synthesis of Anion Binding Domain and Receptor 1.

Reagents and conditions: (a) [Cu(CH₃CN)₄]PF₆, TBTA, CH₂Cl₂, RT, 16 h, 65%; (b) [Cu(CH₃CN)₄]PF₆, TBTA, CH₂Cl₂, RT, 16 h, 72%; (c) NaN₃, DMF, RT, 24 h, 60%; (d) [Cu(CH₃CN)₄]PF₆, TBTA, CH₂Cl₂, RT, 48 h, 36%.

Figure 4.

Solid-state structure of C₆₀-DB18C6 adduct 9 with a water molecule (green color) hydrogen bonded to a crown ether.

The synthesis of the azide-functionalized XB anion binding domain 15 required desymmetrization of bis-iodoalkyne building block 10 (Scheme 2). An excess of 10 was treated with azide 11 in the presence of [Cu(MeCN)₄]PF₆ and Cu(I) stabilizing ligand TBTA to obtain mono-iodotriazole 12 (65%). It was then reacted with azide 13 to obtain asymmetric bis-iodotriazole 14 (72%), which could be readily transformed into azide 15 (60%). In the final step, the anion binding moiety was doubly “clicked” with bis-alkyne crown ether fullerene adduct 9 to obtain ion-pair receptor 1 following purification by flash and size-exclusion chromatography (36%).

The synthesis of receptor 2 with a shorter linker between the anion binding domain and the fullerene surface was initially attempted in an alternative approach with a Bingel reaction conducted on precursor 16 (Scheme 3). Unfortunately, this resulted in a complex mixture of products, suggesting that the anion binding motif is not compatible with the conditions of the cyclopropanation reaction. An alternative route involving Cu(I)-catalyzed azide–alkyne cycloaddition (CuAAC) reaction between alkyne 12 and DB18C6-fullerene bis-azide 18 proved successful (Scheme 3). Fullerene azides are known to be highly unstable due to possible reactions at the fullerene surface.⁵⁸⁻⁶⁰ However, a Bingel reaction of short duration time between 17 and C₆₀, followed by rapid chromatographic purification afforded 18, which was used immediately in the next step to obtain final receptor 2 (14%).

Scheme 3. Receptor 16 and the Synthesis of Receptor 2.

Reagents and conditions: (a) C₆₀, I₂, DBU, KPF₆, toluene, RT, 1 h; (b) 12, [Cu(CH₃CN)₄]PF₆, TBTA, CH₂Cl₂, RT, 18 h, 14%.

Anion and Ion-Pair Binding Studies

The anion binding properties of fullerene containing heteroditopic receptors 1 and 2 were investigated by ¹H NMR titration experiments in 3:1 CDCl₃:CD₃CN. Addition of TBA halides (Cl^–, Br^–, I^–) to solutions of the free receptors 1 or 2 caused significant shifts of the XB anion binding domain proton signals, and no changes (Δδ < 0.01 ppm) of the crown ether cation binding protons were observed. Such a behavior strongly indicates that the ditopic binding domains of receptors 1 and 2 are electronically and spatially well-separated from each other. Notably, the respective receptor’s internal nitroaryl proton (a) shifted downfield, which is indicative of halide binding in a cavity formed by the iodotriazole XB donors (Figure 5). Bindfit analysis of the titration isotherm data revealed that 1:1 and 1:2 stoichiometric host–guest complexes are formed (Table 1).^74,75 In the 1:1 complex, the halide anion is most likely bound by both bidentate XB motifs, contributing up to four halogen bond donors. Upon addition of excess halide anion, each appended XB bidentate recognition site binds an individual anion with two halogen bond donors to form a 1:2 stoichiometric host–guest complex. Unsurprisingly then, K_1:1 association constant values are more than an order of magnitude larger than those of K_1:2. Receptor 1 binds halides more strongly than 2 with a selectivity trend of Br^– > I^– > Cl^–. The enhanced halide anion binding by 1 may be attributed to the longer, flexible linker providing the conformational freedom to facilitate the formation of stronger XB–halide anion interactions in the 1:1 complex. Control receptor 16, which does not contain fullerene, exhibits a halide selectivity trend mirroring that of 1, however with lower K_1:1 values. Interestingly, receptor 2 exhibits a unique, however modest, preference for I^– over Br^– and Cl^– which may be a result of a shorter distance between the XB binding units and the fullerene surface. The combination of the C₆₀ fullerene scaffold’s preorganization of the two XB anion binding arms proximal to the hydrophobic fullerene surface and possible additional anion−π interactions are most likely responsible for the enhanced binding of 1 and a unique iodide binding selectivity of 2.

Figure 5.

Truncated ¹H NMR spectra of receptor 2 in 3:1 CDCl₃:CD₃CN with an increasing amount of TBAI. Corresponding binding isotherms were used for determining values of binding constant.

Table 1. Halide Anion Association Constants (K_a, M^–1) for Receptors 1, 2, and 16.

anion		1a	2a	16a
Cl–	K1:1	3600 ± 300	2400 ± 100	2700 ± 100
	K1:2	150 ± 10	200 ± 30	140 ± 10
Br–	K1:1	5200 ± 800	2900 ± 300	4000 ± 400
	K1:2	190 ± 10	160 ± 10	170 ± 30
I–	K1:1	4600 ± 100	3400 ± 100	3200 ± 300
	K1:2	180 ± 20	140 ± 20	160 ± 10

^aSolvent: 3:1 CDCl₃:CD₃CN at 298 K. Values reported as the mean and the standard error of the mean from independently repeated experiments.

Dibenzo-18-crown-6 is known for its high affinity for potassium cations. Therefore, the K⁺ binding properties of the fullerene containing receptors 1 and 2 were also investigated by ¹H NMR titration experiments in 3:1 CDCl₃:CD₃CN. Addition of KBAr₄^F to solutions of free receptors 1 or 2 resulted in significant perturbations of the crown ether cation binding domain chemical shifts. With the first aliquots of KBAr₄^F, notable signal broadening was observed, and a new set of signals emerged due to slow exchange on the NMR time scale (Figure 6). In particular, the aromatic proton signals of the receptor’s DB18C6 motif experienced notable downfield shifts (Δδ ≈ 0.10 ppm), concomitant with −OCH₂– crown ether perturbations, which due to significant broadening were difficult to follow. After 1.4 equiv of K⁺, however, no further changes were observed, which is indicative of strong binding in a 1:1 stoichiometric host–guest complex. Importantly, no significant changes (Δδ < 0.01 ppm) were observed in the proton signals of the XB anion binding domains of both receptors. This observation further corroborates a good separation of the anion binding domain from the cation one.

Figure 6.

Truncated ¹H NMR spectra of receptor 2 in 3:1 CDCl₃:CD₃CN with an increasing amount of KBAr₄^F. Broadening of the aliphatic crown ether signals hinders their assignment.

To investigate the ion-pair binding properties of receptors 1 and 2, halide anion ¹H NMR titration experiments in the presence of 1 equiv of KBAr₄^F were undertaken. Addition of TBAI to a solution of K⁺ complexed 1 or 2 caused perturbations of the XB anion binding sites, confirming that XB donors are involved in anion binding. In fact, these shift patterns were qualitatively similar to those observed during titrations of the free receptors. However, small changes were also observed in the −OCH₂– proton signals of the crown ether cation binding domain. Notably, the downfield perturbations of the −OCH₂– crown ether signals around 3.90 ppm were larger during titrations with bromide and even greater with chloride (Figure 7). This perturbation pattern cannot be explained by simple potassium cation decomplexation of the crown ether and precipitation of the potassium halide salt (compare with the spectra shown in Figure 6).

Figure 7.

Comparison of crown ether regions during ¹H NMR titrations of receptors 9 (top) and 2 (bottom) in the presence of 1 equiv of K⁺ with TBA salts of Cl^–, Br^–, and I^–.

The aforementioned observations suggest that crown ether bound K⁺ is directly involved in complexation of the anions. Interestingly, a qualitatively similar perturbation pattern of the crown ether signals was also observed during analogous titrations of K⁺ complexed C₆₀-DB18C6 adduct 9, which does not contain a XB anion binding domain (Figure 7). In this case, however, the overall signal shifts were more pronounced. In the presence of 1 equiv of KBAr₄^F receptor 9 forms 1:1 stoichiometric halide complexes with the preference for hard small anions: Cl^– > Br^– > I^– in 3:1 CDCl₃:CD₃CN, as determined by the Bindfit analysis of binding isotherms (Table 2). This assembly is driven predominantly by electrostatic interactions with the crown ether complexed potassium cation resulting in the formation of a close contact ion-pair.

Table 2. Overall Anion Association Constants (K_1:1, K_1:2/M^–1) for Receptors 1, 2, and 9 in the Presence of 1 equiv of K⁺.

anion		1a	2b	9b
Cl–	K1:1	15100 ± 300	9300 ± 300	5000 ± 100
	K1:2	490 ± 30	400 ± 50	–b
Br–	K1:1	25600 ± 1500	8300 ± 300	2700 ± 100
	K1:2	330 ± 50	280 ± 30	–b
I–	K1:1	36000 ± 2800	19800 ± 700	1400 ± 100
	K1:2	600 ± 30	610 ± 60	–b

^aSolvent: 3:1 CDCl₃:CD₃CN at 298 K. Values reported as the mean and the standard error of the mean from independently repeated experiments.

^bNot formed.

We suspect that such a binding mode is also present during titrations of receptors 1 and 2 with halides, particularly harder ones such as Cl^– and Br^–. However, due to spatial separation of the cation and anion binding domains in receptors 1 and 2, two distinct types of 1:1 stoichiometric complex A and B can be simultaneously formed, contributing to the experimentally determined overall 1:1 association constants (Figure 8). In binding mode A, the anion is exclusively bound by the XB binding site, while in mode B the anion associates solely with the crown ether bound potassium cation in a contact ion-pair recognition fashion.^61,62 Similarly, upon excess addition of anion, two types of 1:2 stoichiometric complexes are possible: C, with two anions bound individually by the XB donor arms, and D, in which one anion is bound in a tetradentate XB fashion and the other anion is associating with the potassium cation (Figure 8).

Figure 8.

Proposed anion binding modes of fullerene-functionalized halogen-bonding heteroditopic hosts in the presence of crown ether bound potassium cation.

This hypothesis was further corroborated by the ¹H NMR titration of 2·K⁺ with NO₃^–, which exhibits strong preference for the close contact ion-pair formation (mode B). During the titration with TBANO₃ no changes were observed in the XB anion binding domain, while significant perturbation of the crown ether signals was observed until ca. 1.5 equiv of anion was added (Figure S38). Addition of up to 10 equiv of NO₃^– had no further impact on the proton signals of 2·K⁺, suggesting the oxoanion forms only the 1:1 stoichiometric anion complex of mode B. Interestingly, addition of 2 equiv of TBACl at the end of the titration with NO₃^– (10 equiv) induced perturbations of XB binding domain signals, and no changes of crown ether signals were observed. This confirms the independence of the anion binding sites and formation of type D complex with chloride occupying the XB binding domain and nitrate forming the contact ion-pair with the crown ether bound K⁺.

Quantitative analysis of the halide binding isotherms in the presence of 1 equiv of KBAr₄^F, using a 1:2 stoichiometric host–guest model, gave overall association constant values shown in Table 2, where K_1:1 represents the sum of 1:1 stoichiometric anion binding modes A and B and K_1:2 the sum of 1:2 stoichiometric anion binding modes C and D. Comparing Tables 1 and 2, the presence of the complexed K⁺ in the respective C₆₀-DB18C6 cation binding domain of the XB receptors 1 and 2 results in a significant enhancement of halide association constants, particularly of K_1:1. Importantly, it was possible to deconvolute and estimate the individual contributions of A and B halide binding modes to the overall K_1:1 association constant value through the analysis of the chemical shifts of the crown ether protons of receptors 1 and 2 and control receptor 9 during halide anion titrations in the presence of KBAr₄^F (see the Supporting Information for details). In the case of 1, the binding mode A accounts for approximately 63% of the overall 1:1 chloride association constant, 83% of 1:1 bromide association constant, and more than 95% of 1:1 iodide association constant, clearly showing the preference of the heavier softer halide anions toward fullerene-assisted XB binding. Further corroborating these estimates, the values of 1:1 association constants for mode B of receptors 1 and 2, obtained using this method, are in good agreement with the values obtained during titrations of control receptor 9, which is able to bind anions only via mode B.

Deconvolution of the A and B binding modes of 1:1 association enabled a direct comparison of K⁺ coordination effects on the fullerene-assisted halide binding in the XB domain. The K_1:1 association constant values for anion binding mode A (K_1:1^A) of both heteroditopic receptors 1 and 2 are significantly increased in the presence of cobound K⁺ (Table 3). Notably, the binding enhancement for iodide is particularly strong, resulting in a remarkably increased selectivity for this anion. In the case of receptor 2, K_1:1^A association constants increased by a factor α = 2.2 (α = K_1:1^A(K⁺)/K_1:1) and 2.1 for chloride and bromide, respectively, while for iodide α = 5.2. Even stronger I^– enhancement was observed for receptor 1 (α = 7.4); however, overall selectivity of 1 for I^– vs other halides is reduced in comparison with 2. In the presence of cobound K⁺, control heteroditopic receptor 16, without the fullerene scaffold, binds all the halides significantly more strongly (α = 12.7–13.5), however notably at the expense of much lower selectivity. This is most likely due to the formation of a close contact ion-pair, resulting in stronger electrostatic interactions. The remarkable properties and influence of C₆₀ for ion recognition are particularly evident in comparison of iodide binding affinity exhibited by 1 and 16. Impressively, the iodide K_1:1 association constant value of receptor 1, whose anion and cation binding domains are separated by the fullerene, almost matches the magnitude for receptor 16, which is capable of anion binding assisted by close contact with the crown ether cobound potassium cation. Importantly, this suggests that the polarizing C₆₀ surface can elicit particularly strong interactions with polarizable soft anion species such as iodide and can transfer electrostatic effects over significant distances within the fullerene heteroditopic host design.

Table 3. Anion Association Constants for Receptors 1 and 2 (K_1:1^A, M^–1, Anion Binding Mode A) and 16 (K_1:1, M^–1) in the Presence of 1 equiv of K⁺.

	1			2			16
anion	K1:1A a	αb	Ac (%)	K1:1A a	αb	Ac (%)	K1:1A a	αb
Cl–	9500 ± 200	2.6	63	5300 ± 200	2.2	57	34200 ± 600	12.7
Br–	21200 ± 1300	4.1	83	6100 ± 200	2.1	73	54000 ± 4600	13.5
I–	34200 ± 2700	7.4	>95	17800 ± 600	5.2	>90	43100 ± 3700	13.5

^aSolvent: 3:1 CDCl₃:CD₃CN at 298 K. Values reported as the mean and the standard error of the mean from independently repeated experiments.

^bBinding enhancement factors in the presence of K⁺.

^cContributions of anion binding mode A to overall 1:1 binding.

Computational Analysis

Having demonstrated the unique ion-pair recognition properties of fullerene-containing heteroditopic receptors 1 and 2, DFT calculations were undertaken to gain insight into the electronic and structural aspects of ion-pair complexation. In the computational analysis, we focused on receptor 2, which manifested the highest selectivity, presumably due to the closer proximity of the fullerene surface. DFT calculations in the gas-phase were performed with Gaussian16,⁶³ using the M06-2X functional, chosen for its ability to accurately describe halogen bonds and π–π stacking interactions.⁶⁴⁻⁶⁶ The Def2-SVP basis set was selected,⁶⁷ except for the anions and iodine binding units, which were described with the Def2-TZVPD basis set,^67,68 taken from the Basis Set Exchange website.⁶⁹⁻⁷¹ This combination was employed to balance the accurate description of the noncovalent interactions and the structural features of the large receptors 1 and 2.

The MEP surface of C₆₀ (Figure 1) exhibits distinct electrophilic regions, with π-holes situated above its 12 five-membered (C₅) rings and 20 six-membered (C₆) rings. These π-holes have molecular surface electrostatic potential (V_S) values ranging between 7.2 and 7.9 kcal mol^–1. For comparison, 9 and its K⁺ complex were also optimized by DFT (Figure 9). The DB18C6 motif in adduct 9 has a significant effect on the electrostatic potential map, resulting in a nearly negatively charged fullerene surface. The exposed surface of the fullerene displays several π-holes with V_S values ranging between −5.5 and −0.2 kcal mol^–1. The lowest V_S values of 9 were found between the oxygen atoms of the crown ether cavity, varying between −58.6 and −57.5 kcal mol^–1. The MEP surface of 9’s C₆₀ moiety has an additional negative point of V_S with −29.5 kcal mol^–1, perpendicular to the C₆ ring just below the crown ether. This electron-rich site is perfectly prepared for the coordination of K⁺, as evidenced by a computed K⁺···C₆ distance of 2.93 Å in 9·K⁺. The potassium cation binding induces a significant redistribution of the electrostatic potential in 9, with the C₆₀ π-holes’ V_S values now ranging from 30.4 to 47.2 kcal mol^–1. For comparison, typical π-hole donors trifluoro-1,3,5-triazene or hexafluorobenzene,⁷² investigated at the M06-2X/Def2-SVP theory level, respectively display V_S,max values of 40.9 and 21.5 kcal mol^–1. However, in complex 9·K⁺, the V_S,max of 111.8 kcal mol^–1 is found over the metal cation, which explains its strong tendency to form close putative ion-pair contacts with halides, as depicted in Figure S71 with the DFT optimized structures of 9·K·X (X = Cl, Br, or I) complexes. The computed K⁺···X^– distances (Cl^–: 2.85; Br^–: 3.02; I^–: 3.23 Å) mirror the anion’s size, with the shortest contact corresponding to the highest association constant, found for 9·KCl (Table 2).

Figure 9.

DFT-optimized structure of 9·K⁺ (center, top), together with the MEP surfaces calculated on the adduct free of potassium (left) or on the complex (right), in lateral and top views. The MEP surfaces are rendered at the 0.001 electrons Bohr^–3 contour and the π-holes on the surface of C₆₀ are identified as black dots.

The starting geometries of 1 and 2 were generated via crude gas-phase MD simulations of KCl complexes, enforcing halogen bonds through geometric restraints, as detailed in the Supporting Information. Multiple conformations were selected and subsequently subjected to geometry optimizations using DFT. Figure 10 shows the optimized structures of chloride complexes of 2 with potassium hosted within the DB18C6 cation binding domain for the A–D anion binding scenarios, consistent with the binding modes hypothesized based on ¹H NMR titrations, while Figures S72 and S73 show equivalent optimized binding arrangements for bromide and iodide.

Figure 10.

DFT-optimized anion binding modes of fullerene-functionalized halogen-bonding heteroditopic host 2 in the presence of the crown ether complexed potassium cation.

Potassium binding by the DB18C6 moiety can be characterized by K⁺···C₆ distances summarized in Table S7. In scenarios A and C, the average K⁺···C₆ distance is ca. 3 Å, whereas in binding mode B, the ca. 3.5 Å average K⁺···C₆ distance is significantly larger. Due to the close ion-pair contact in B, the anion pulls K⁺ from the crown ether, weakening the potential interaction with the fullerene surface.⁵⁴ Interestingly, in scenario D, the K⁺···C₆ distances have intermediate values between those computed for B and A/C, showing that binding in one domain can influence the behavior on the opposite side.

In binding mode A, the four convergent halogen bonds with chloride are not equivalent (Table S8). Two interactions have an average I···Cl^– distance of 3.35 Å and an average C–I···Cl^– angle of 163°, while the other two interactions are nearly linear with distances of 3.03 Å and angles of 173°, consistent with highly directional σ-hole XB interactions. The XB distances and angles in the bromide (3.58 Å, 162°; 3.21 Å, 174°) and iodide (3.83 Å, 162°; 3.43 Å, 176°) complexes follow a similar pattern, adjusted for the sizes of the ions (Table S8). This asymmetry is not surprising considering the differences in the iodotriazole units of the anion binding domain. One is directly connected to a strong electron-withdrawing 3,5-bis(trifluoromethyl)phenyl group, while the other is connected to an electron-rich alkyl-substituted phenyl. Importantly, the halogen-bonding anions’ recognition is assisted by anion−π interactions with short contacts between the fullerene surface and each anion, leading to the trend of C₆···X^– distances Cl^– (3.13 Å) < Br^– (3.35 Å) < I^– (3.63 Å).

After optimization of the putative binding arrangements for the recognition of halides in binding modes A–D, the MEP maps of 2 and 2·K⁺ were evaluated through single-point DFT calculations. To achieve this, we used the optimized structure of the Cl^– association in scenario D and removed the necessary ions (Figure 11). The most negative region of electrostatic potential on the electronic surface of 2 covers the oxygen atoms of the crown ether (including the V_S,min of −45.8 kcal mol^–1), while the most positive regions are found in front of the iodine binding clefts, with their σ-holes characterized by V_S values of 43.6 and 44.3 kcal mol^–1 and of 48.2 and 48.3 kcal mol^–1 for the iodo-triazole unit activated by neighboring −CF₃ groups. Complexation of K⁺ in the cation binding domain of 2 leads to a significant redistribution of the MEP surface. Naturally, the V_S,max of 130.2 kcal mol^–1 is located over the cation; however, the four XB units display augmented V_S values between 75.3 and 79.7 kcal mol^–1. Notably, an V_S point of 61.5 kcal mol^–1 was found positioned over the C₆ ring in the vicinity of the preorganized XB binding units.

Figure 11.

MEP surfaces calculated on the DFT-optimized geometry of 2 in binding scenario D (Figure 10D): free of ions (left) and in the presence of K⁺ (right). The MEP surfaces are rendered at the 0.001 electrons Bohr^–3 contour and the σ-holes in front of the XB binding units are identified with black dots, while a neighboring π-hole on the surface of the C₆₀ scaffold is identified with a white dot.

The strength of the XB interactions in different binding modes was further evaluated with the natural bond orbital (NBO) analysis using the second-order perturbation theory interaction energies (E², see Table S9). The analysis of the E² values for the interactions between the C–I antibonding orbitals of 2 and the halides’ lone pairs orbitals (n_X → σ*_C–I) revealed that in binding mode A the total energies follow the trend Cl^– (37.5 kcal mol^–1) > Br^– (36.3 kcal mol^–1) > I^– (33.1 kcal mol^–1). A similar analysis was also performed for the interactions between the fullerene scaffold’s π-holes and the three halides in binding modes A and D, revealing that E² values resulting from n_X → σ*_C–C are over an order of magnitude weaker (0.5–1.3 kcal mol^–1) than the XB interactions.

Although the computational analysis suggests that in the gas phase binding mode B is preferred for 2 by 3.0 (Cl^–), 5.1 (Br^–), and 6.2 kcal mol^–1 (I^–) , it is worth noting that in the case of 1·KCl, binding mode A is favored by 24.9 kcal mol^–1 relative to B (Figure S74).⁷³ In this complex, however, the halide does not form a contact with the fullerene surface, and the receptor adopts a conformation that maximizes the strength of the halogen bonding.

Altogether, the computational analysis reveals that the fullerene platform can play a dual role in anion binding by receptors 1 and 2. Indeed, it can be actively involved in the binding events by exploiting π-holes on its surface, but it can also serve as a bulky scaffold to preorganize the potent XB-based binding units into a tight binding cavity.

Conclusions

For the first time, the C₆₀ fullerene motif has been successfully integrated into a heteroditopic ion-pair host design. The combination of highly potent XB donors and a crown ether moiety separated by C₆₀ led to the rationally designed receptors with ion-pair binding properties influenced and modulated by the fullerene motif. Receptors 1 and 2, which differ in length of linkers separating anion binding motifs from the fullerene surface, are capable of strong and more selective binding of halide anions than their non-fullerene analogue. This is achieved by the preorganization of binding units, solvent shielding, and π-hole assistance provided by the bulky and highly polarizable C₆₀ architecture. Remarkable halide anion binding enhancements can be achieved by the complexation of a potassium cation by the rigid crown ether moiety located close to the fullerene surface. Notably, potassium binding by receptors 1 and 2 results in strong augmented iodide binding selectivity, with association constants matching the value of the non-fullerene heteroditopic receptor analogue which is capable of anion binding assisted by the close contact with the crown ether cobound potassium cation.

Fullerene spatial separation of XB donors and the crown ether complexed K⁺ resulted in a rare example of an artificial receptor containing two anion binding sites with opposing preferences for hard and soft halides. Detailed analysis of ¹H NMR titration data allowed for deconvolution of the ion-pair binding modes contributing to the overall 1:1 stoichiometric halide anion binding by 1·K⁺ and 2·K⁺. Soft polarizable iodide was bound almost exclusively (>90%) by the tetradentate XB binding domain in proximity to the fullerene surface. This binding mode also dominated in the case of smaller and harder chloride (ca. 60%); however, a significant portion of chloride (ca. 40%) was associated in a close contact ion-pair with the crown ether bound potassium cation without assistance of the XB binding domains.

Altogether, the presented results demonstrate the unprecedented potential of fullerene surfaces in anion recognition host–guest chemistry. Importantly, polarizing the C₆₀ surface via proximal cation recognition can elicit particularly strong interactions with polarizable soft anion species such as iodide and transfer electrostatic effects over significant distances. Modulating the properties of fullerene-based compounds via reversible noncovalent charged guest recognition may find future applications in molecular sensors, solar cell devices, and photodynamic therapy.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.3c07774.

• Experimental procedures and methods, NMR spectra, titration data, additional data, and figures (PDF)

Author Present Address

EaStChem, University of Edinburgh, Joseph Black Building, David Brewster Road, Edinburgh EH9 3FJ, U.K

The authors declare no competing financial interest.