Olefinic-π Interaction inside a Carbon Nanocage | Journal of the

Article Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

pubs.acs.org/JACS

H2O/Olefinic‑π Interaction inside a Carbon Nanocage Yoshifumi Hashikawa and Yasujiro Murata* Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan

Downloaded via UNIV OF SOUTHERN INDIANA on July 25, 2019 at 15:17:36 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: The H2O/CH2CH2-type hydrogen-bonding (H-bonding) model was experimentally constructed using a water complex of an open-cage C60 derivative, in which an olefinic double bond and a single molecule of H2O are geometrically confined. To investigate OH/π-type H-bonding, that is, H2O···(CC) interaction, we performed 1H NMR spectroscopic studies that demonstrated the monotonic downfield shift of the proton signal corresponding to H2O with remarkable rotational perturbation by lowering the temperature. From the temperature dependence of the angular momentum correlation time (τJ), the interaction energy was quantitatively estimated to be ca. 0.3 kcal/mol. The computational studies were thoroughly conducted to clarify its inherent nature. As a consequence, the orientation of H2O was found to play a prominent role in varying the bonding strength as well as contribution from the electrostatic attraction and orbital−orbital interaction significantly driven by the favorable orbital overlap identified as π(CC) → σ*(OH) interaction.

■

INTRODUCTION Single-molecule studies have been recognized as the center of basic science and have received growing attention because they can construct fundamental principles and theory for understanding essential properties of molecules or higher-order clusters.1 In the biological field, the single-molecule techniques have been developed by means of fluorescence spectroscopy, tethered particle microscopy, optical and magnetic tweezers, atomic force microscopy, and so forth.2 From the aspect of electron transport properties and conductivities, physicists focus on a single-molecule junction, where a single molecule bridges a gap between metal electrodes.3 In many cases, however, it is difficult to “isolate” a single molecule in macroscopic quantities under ambient conditions. Fullerenes, being in the shape of caged structures, can selectively accommodate only one or two molecules inside their cavities. As representative examples, H2@C60,4 He@C60,5 H2O@C60,6 HF@C60,7 and CH4@C608 have been prepared using organic synthetic methods.9 Despite encapsulation of these neutral species inside the inner space of C60, there is negligible interaction with the carbon framework.6,10 This undoubtedly indicates that the entrapped molecule is chemically and thermally stable, providing precious opportunities to investigate the nature of a single molecule at the molecular level. So far, using these molecular systems, it has been elucidated that single molecules inside nanoconfined environments exhibit intriguing properties largely different from those in bulk, i.e., nuclear spin relaxation,11 rotational behavior,12 ortho−para conversion,13 dielectric constant,14 rotational and vibrational transitions,7,13f electron densities,15a and chemical reactivity.15b For the sake of investigation on single-molecule properties, we have shown the availability of H2O@C60, where the © XXXX American Chemical Society

encapsulated H2O molecule works as a magnetic probe (Figure 1). Whereas the H2O molecule behaves like a free rotator inside pristine C60 even at cryogenic temperatures,13d its rotational motion becomes slowed down due to van der Waals

Figure 1. Single-molecule studies using endohedral fullerenes: (a) amine−H2O interaction in H2O@RC59N (previous work) and (b) olefin−H2O interaction in H2O@open-C60 (this work). Received: June 26, 2019

A

DOI: 10.1021/jacs.9b06759 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

responsible for the magnitude of the attraction and directionalities in the order OH/π > NH/π > CH/π.23c Thus, it is hard to make a conclusion only from hitherto reported crystallographic data since it requires consideration of effects from packing forces, steric factors, solvation, and conformation. An ideal system to examine a pure OH/π interaction should be a molecular complex of water and ethene (H2O/CH2 CH2) even though this complex is unlikely to be formed under ambient conditions24 and has not been studied well even by computational methods.25 To the best of our knowledge, there is only one report for investigating the ROH−olefin interaction using an isolable molecular complex,26 but this interaction is substantially overestimated since it includes a direct O(H2O)δ−···C(olefin)δ+ interaction as a static form from the aspect of the C···O distance (2.866(2), 3.037(2) Å), which is shorter than the sum of their van der Waals radii (3.22 Å).27 To realize a pure H2O−olefin interaction, we considered that H2O@C60 is a suitable platform since the H2O molecule is geometrically isolated, and thus it is not necessary to consider any factor from the surroundings. It should be mentioned that there is little interaction between the H2O molecule and the C60 cage. In other words, the H2O molecule is regarded as electrostatically isolated.10 In addition to this feature on H2O@ C60, it still demands several structural requirements: (i) fixing an isolated olefinic bond instead of CH2CH2, (ii) a decrease in the rotational freedom of H2O, and (iii) negligible translational diffusivity of H2O. Under these circumstances, open-cage C60 derivatives would be potential systems for modeling the H2O−olefin interaction since they possess olefinic double bond(s) on their openings with high HOMO coefficients. Furthermore, the rotational motion of the entrapped H2O molecule is able to be perturbed by addition of sp3 character on some of the carbon atoms in C60 due to attractive H2Oδ−···C(sp3)δ+ and/or repulsive HO− Hδ+···C(sp3)δ+ interactions, which cause restriction of the rotation.17 As a feasible model system that may fulfill all requirements, we selected a naphthalene-fused open-cage C60 derivative enclosing a water molecule inside its cavity, i.e., H2O@3 (Figure 3). In this compound, an olefinic double bond

interaction with the carbon wall induced by chemical modification of the C60 cage. This phenomenon is rationally explainable by 1H NMR relaxation time, which is closely related to the rotation. For instance, by embedding an NR3 moiety in the form of C59N, the concave surface of the amine was demonstrated to be positively charged and interacts with the H2O molecule via attractive C59Nδ+···Oδ−H2 interaction (Figure 1a).16 As an expanded study, it was possible to rate electronegativities of heteroatoms (E = C, N, and O) by introducing them onto the H2O@C60 cage, resulting in the restricted rotation of the H2O molecule via an electrostatic Eδ−−C60δ+···Oδ−H2 interaction.17 Additionally, a single but H-bonded H2O molecule was realized using a hydroxy open-cage C60 derivative (R− Oδ−H···Hδ+−OH), revealing unusual acid−base character.18 As exemplified above, H2O@C60 enables examining the electronic states of chemical bondings and single-molecule properties of H2O inside the confined space, as well as weak interactions between the H2O molecule and chemically modified carbon cages. Herein, we focused on the OH/π interaction19 as one of the nonclassical hydrogen-bonding (Hbonding) modes.20 As is obvious, H-bondings play crucial roles in numerous fields involving chemistry, biology, and physics.21 Even though the definition of the term “H-bonding” was stated by the IUPAC recommendation in 2011,22 it has been a great topic of interest and is still under exploration at the present time. Hence, from experimental and theoretical viewpoints, we discuss details on the olefin−H2O interaction both qualitatively and quantitatively, using a cage-opened system of H2O@ C60 as a model, in which olefin and H2O are geometrically confined (Figure 1b).

■

RESULTS AND DISCUSSION Molecular Design Concept for the Evaluation of an OH/π-Type H-Bonding. The H-bonding interaction can be expressed as D−H···A, where D and A represent a H-donor and acceptor, respectively. The OH/π-type H-bonding is a noncovalent interaction and categorized as one of the unconventional H-bondings involving atoms with π-bonding molecular orbitals as H-acceptors. As a typical π-type Hacceptor, aromatic rings such as benzene have been recognized to enable the interaction in the solid state.23 Recently, it was revealed that both orbital and electrostatic interactions contribute to form an OH/aomatic-π interaction.23f According to the crystallographic analyses, the OH/aomatic-π interaction dominantly shows close contacts with the π-edge, suggesting an intense orbital−orbital interaction, which stands in contrast to the interaction with the π-centroid, e.g., electrostatic cation/ π interaction (Figure 2). On the grounds of theoretical studies, however, the electrostatic interaction is predicted to be mainly

Figure 3. 1H NMR chemical shifts (800 MHz, ODCB-d4, 300 K) of endohedral derivatives studied herein (H2O@C60, H2O@1, H2O@2, and H2O@3).

Figure 2. Different types of H-bonding modes: (a) interaction with πedge and (b) interaction with π-centroid. B

DOI: 10.1021/jacs.9b06759 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

underestimated when compared with the reported value obtained from a 13C self-diffusion coefficient (1 mM in benzene at 298 K, D = 8.3 × 10−10 m2/s, rH = 4.05 Å).30 If the diffusivities of the H2O molecule are high enough, the molecular diffusion coefficients would become larger than those for empty ones (rH-H2O/rH-empty >1). However, there is little difference between D-H2O and D-empty, resulting in the ratio of rH-H2O/rH-empty close to a unity (rH-H2O/rHempty ≈ 1). These results indicate that the translational diffusivity of H2O inside each cage is negligibly small. As a consequence, H2O@3 fully satisfies all the requirements on the aforementioned molecular design and the other three compounds should be suitable references to examine the pure OH/π interaction. X-ray Crystallographic Analyses. The single crystals for H2O@2 (CCDC 1895269) and H2O@3 (CCDC 1583657)28 were obtained from CS2 solutions by slow evaporation. As shown in Figure 4, the bond lengths of C(1)C(2) are

is embedded in the opening and a water molecule is located at the center of the cage, where the H2O molecule potentially adopts a better orientation to maximize the OH/π interaction. Furthermore, it has sp3-carbon atoms, nearby the opening, introduced by naphthalene fusion. As reference compounds, H2O@C60, naphthalene-fused C60 (H2O@1), and a nonfused open-cage C60 derivative (H2O@2) were also comprehensively studied from the viewpoints of self-diffusion coefficients, X-ray crystallographic analyses, temperature-dependent 1H NMR chemical shifts, and relaxation times, as well as theoretical calculations including electrostatic potential (ESP), natural charge based on natural population analysis (NPA), atoms-inmolecule (AIM), natural bond orbital (NBO), and Wiberg bond order (WBO). Preparation of Endohedral C60 Derivatives. H2O@1 was obtained in 32% yield from H2O@C606 and 1-bromo-4-nbutylnaphthalene in the presence of a Pd catalyst by heating a 1-methylnaphthalene solution at 200 °C for 3 h. The product was purified by HPLC equipped with a Buckyprep column. H2O@2 was synthesized according to the literature,28 and H2O@3 was obtained from H2O@2 in a similar way to H2O@ 1. All compounds studied herein have the guarantee of at least 60% encapsulation ratio of H2O, determined by 1H NMR. Self-Diffusion Coefficients of Encapsulated H2O Molecules. For all compounds, we measured self-diffusion coefficients of the encapsulated H2O molecules (D-H2O) to make sure whether or not these molecular systems meet requirement (iii): negligible translational diffusivity of H2O inside the cavity. We prepared 1.5 mM solutions in ODCB-d4 using 2:3 mixtures of empty 1−3 and H2O@1−3. The signals observed in the aromatic region were partly separated between the empty and the encapsulated ones. Using these signals corresponding to the empty ones (Href, displayed in Figure 3), D-empty values were determined. The D-H2O values were determined using the signal corresponding to the entrapped H2O molecule. The measurements were conducted at 300 K under a field strength of 800 MHz. These coefficients were calibrated using the precisely determined value of D2O.29 Table 1 summarizes self-diffusion coefficients D and effective Table 1. Self-Diffusion Coefficients D and Effective Hydrodynamic Radii rHa compd H2O@C60 H2O@1 H2O@2 H2O@3

D (10−10 m2/s)b 5.25 4.32 3.28 3.26

(−) (4.32) (3.25) (3.21)

rH (Å)b,c 3.17 3.86 5.08 5.11

(−) (3.85) (5.13) (5.18)

Figure 4. Single-crystal X-ray structures of (a) H2O@2 and (b) H2O@3. Thermal ellipsoids are shown at 50% probability. Solvent molecules are omitted for clarity.

rH-H2O/rH-empty 1.00 0.991 0.987

refined to be 1.384(3) Å for H2O@2 and 1.363(8) Å for H2O@3, being in good agreement with those for typical olefinic double bonds. Whereas H(1) and H(2) for H2O@2 cannot be placed without DFIX instructions,31 the orientation of O(1)−H(1) for H2O@3 was able to be determined since a residual Q-peak corresponding to H(1) was still observed after placing the O(1) atom.28 The distances for C(1)C(2)··· O(1) of H2O@2 (3.650(5) Å) are comparable to those for H2O@3 (3.685(6) and 3.643(6) Å), which are adequately longer than the sum of van der Waals radii (3.22 Å) with no direct C···O interaction. Importantly, these distances are shorter than that for an H2O molecule inside a hydroxy open-cage C60 derivative that exhibits intramolecular Hbonding between the H2O molecule and OH groups on the opening (HOδ−···Hδ+−OH: 3.81(6) Å).18 These results suggest the intramolecular olefin−H2O interaction in H2O@ 3 and possibly in H2O@2.

a

Measured using 1.5 mM solutions in ODCB-d4 at 300 K (800 MHz), calibrated with D2O. bValues in parentheses are obtained from empty ones. cCalculated using the Stokes−Einstein equation.

hydrodynamic radii rH calculated using the Stokes−Einstein equation: rH =

kBT 6πηD

where kB is the Boltzmann constant, T is temperature, and η is viscosity. It was shown that the D values become smaller in the order H2O@C60 > H2O@1 > H2O@2 > H2O@3 with increasing the molecular size (rH). From the rH value of H2O@ C60, the diameter was calculated to be 6.34 Å, which seems to be the same as the size of C60 (ca. 1 nm) but slightly C

DOI: 10.1021/jacs.9b06759 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society Temperature-Dependent 1H NMR Chemical Shifts. As shown in Figure 3, the chemical shifts corresponding to the entrapped H2O molecules inside the fullerene cages appear at the high-field region from −4 to −10 ppm, reflected by their different shielding environment.28 The change in NMR chemical shift at variable temperature (VT) is usually utilized as a diagnostic tool to judge the presence or absence of Hbondings. As we previously reported, the chemical shift of the H2O molecule in C60 slightly moves to the upfield region with decreasing the temperature (Figure 5) presumably due to the

spin−lattice and spin−spin relaxation times (T1 and T2). The sample concentration was set to 1.5 mM in ODCB-d4. Prior to measurements, all the solutions were degassed by three argon− vacuum cycles. The measurements were conducted on an 800 MHz NMR instrument by applying the inversion−recovery method for T1 and the CPMG method for T2.32 The 90° pulse width was determined at each temperature. With regard to H2O@C60, the T1 and T2 values faintly increase with lowering the temperature (Figure 6a).16a This indicates that the

Figure 5. VT 1H NMR chemical shifts of H2O@C60, H2O@1, H2O@ 2, H2O@3, and H2O@4 (800 MHz, 1.5 mM in ODCB-d4, 260−340 K).

intense shielding effect originated from its slower rotation.16a With respect to H2O@1, the same trend was observed with a Δδ/T of −0.68 ppb/K. These temperature dependences entirely differ from H-bonding character. Contrastingly, the downfield shifts were observed for both H2O@2 and H2O@3 with positive Δδ/T values, explicitly suggesting the existence of intramolecular H-bonding on account of possible positive and negative charges on H(1) and C(1)/C(2) atoms, respectively. In general, the chemical shift corresponding to protons engaged in the strong static interaction is less affected by the change in temperature. The temperature dependence observed for H2O@2 and H2O@3 indicates the weak interaction associated with the transiently occurring intramolecular hydration and dehydration that should be induced by the rotational reorientation of the entrapped H2O molecule. It is worth mentioning that the reorientation of the H2O molecule is rapid enough on the NMR time-scale because the signals appeared as a singlet over the measured range. H2O@3 has roughly 2 times larger Δδ/T values (+0.52 ppb/K) with a monotonic shift relative to H2O@2 (+0.29 ppb/K), indicative of stronger interaction in H2O@3. These results clearly demonstrated that the naphthalene fusion promotes the olefin−H2O interaction inside 3 likely due to the decrease in the rotational freedom of H2O and stronger olefinic character in 3 ascribed to partial isolation of the double bond from the conjugated system on the C60 cage as well as the planarization of the olefinic C(1) atom. This leads to the better orientation of the O(1)−H(1) group facing the isolated olefinic C(1) C(2) bond. Relaxation Times of Encapsulated H2O Molecules. To get further insights into the rotational behavior of the entrapped H2O molecules inside C60, 1, 2, and 3, we measured

Figure 6. (a−c) Relaxation times T1 and T2 of H2O@C60, H2O@1, H2O@2, and H2O@3 and (d) T1/T2 ratio (800 MHz, 1.5 mM in ODCB-d4, 260−340 K). Measuring errors are less than 3%.

relaxation process is predominantly governed by the spinrotation mechanism in which the collisions with the C60 wall give rise to perturbation in both direction and magnitude of the angular momentum vector followed by a fluctuation in the magnetic field, leading to rapid relaxation at higher temperatures.32 Since H2O@C60 displays negligible gap between two parameters with the T1/T2 values being nearly unity within the measured range, the H2O molecule is considered to isotropically rotate inside the C60 cavity. From the temperature dependence of T1, the spin-rotation mechanism is dominant for relaxation in all the compounds studied herein. Once the naphthalene ring is introduced on the C60 cage, the encapsulated H2O molecule inside 1 showed larger T1 values with a similar temperature dependence, whereas the T2 values became smaller with decreasing the temperature (Figure 6a). This causes a slightly larger T1−T2 gap at lower temperatures and further suggests the restricted rotation that is reminiscent of H2O@C60E having two sp3 carbon atoms (E: heteroatom).17 In terms of a plotting feature, H2O@2 resembles H2O@1 with a much larger T1−T2 gap, implying the decreased rotational freedom of the H2O molecule (Figure 6b). It is notable that an exceptionally large T1−T2 gap was seen in H2O@3 (Figure 6c) accompanying a downfield shift of the proton signal (Figure 5). The T1/T2 values thereby followed the order H2O@3 > H2O@2 > H2O@1 > H2O@C60 (Figure 6d). Compared with H2O@1, the T1/T2 curve of H2O@3 is D

DOI: 10.1021/jacs.9b06759 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society shifted by a factor of +60 K and the T1/T2 values exceed the simple summation of those for H2O@1 and H2O@2. By employing the semiclassical form of the density operator theory of relaxation, if the spin-rotation interaction is the only mechanism of the relaxation under the extreme narrowing conditions, the rate of relaxation (1/T1,sr) can be given by33,11a,c 2 i μ y 2IkBT 1 (2C⊥2 + C 2)τJ = jjj 0 zzz 2 T1,sr k 4π { 3ℏ

where μ0 is the vacuum permittivity, I is the moment of inertia, kB is the Boltzmann constant, ℏ is the Dirac constant, and C is the spin-rotation coupling constant at the perpendicular (⊥) and parallel (||) axes. Angular momentum correlation time τJ is a measure of fluctuation in the angular velocity which determines the efficiency of the relaxation and is related to the average lifetime of the rotational states.32 For all the compounds, the τJ values were calculated and found to be shorter in the order H2O@3 (0.44 ps) < H2O@2 (0.55 ps) < H2O@1 (1.10 ps) < H2O@C60 (2.23 ps) at 300 K, being in excellent agreement with the prediction from the T1/T2 values. Importantly, the τJ values except for H2O@C60 became larger with increasing temperature (Figure S6), as typically seen in (1H)2,11a,34 C(1H)4,35 (13C)S2,36 CH(19F)3,37 and CH(19F)Cl2 . 38 This provides assurance that the spin-rotation mechanism almost solely contributes to the relaxation in H2O@1−3 even though this situation has been observed only for supercritical water39 or water vapor under reduced pressure.40 This indicates that the H2O molecules inside C60 cages are considered to show gaseous-like behavior. In general, τJ obeys the Hubbard relation τJτc = I/6kBT,33 where τc is the rotational correlation time, which is a measure of fluctuation in the orientation and shows an exponential temperature dependence.32 Thus, the plots of the natural logarithm of TτJ against reciprocal temperature exhibit Arrhenius behavior that is characteristic of the thermally activated process (see details in the Supporting Information). From the slopes, the rotational activation energies were estimated to be Ea 0.42 ± 0.03 kcal/mol for H2O@3, 0.23 ± 0.06 kcal/mol for H2O@2, and 0.13 ± 0.02 kcal/mol for H2O@1, respectively. By comparison with the activation energy for H2O@C60 (Ea + 0.07 kcal/mol) obtained from the measurement of the heat capacity,12b it is conclusive that the installation of the isolated olefinic bond considerably prevents the rotational motion of the H2O molecule in 2 and 3, likely arising from the H2O− olefin interaction (Figure 6d). If this increased activation energy in H2O@3 originates only from the stabilization of the ground state, the interaction energy could be roughly estimated by subtracting Ea(H2O@1) from Ea(H2O@3) in consideration of the naphthalene fusion’s influence on the energy of the rotational transition state. Thereby, the interaction energy was obtained to be ca. 0.3 kcal/mol for H2O@3. Electrostatic Potential Maps. For a better understanding of the rotational behavior of H2O, we conducted single-point calculations41 for drawing ESP maps of all host cages at the MP2/6-31G(d,p) level of theory using geometries optimized at the M06-2X/6-31G(d,p) level of theory. All calculations were performed using simplified models where tert-butyl and n-butyl groups are replaced with methyl groups and a hydrogen atom, respectively. As displayed in Figure 7, C60 has an electrostatically homogeneous inner potential,16 whereas 1′ shows a

Figure 7. ESP maps of (a) C60, (b) 1′, (c) 2′, and (d) 3′ (MP2/631G(d,p)//M06-2X/6-31G(d,p)).

positive potential at the backside of the two sp3-carbon atoms bound to the naphthalene ring. This should affect the rotational motion of H2O via attractive H2Oδ−···C(sp3)δ+ and/or repulsive C(sp3)δ+···Hδ+OH interactions. In contrast, the negative potential with weak intensity can be seen under the olefinic double bond in 2′, possibly leading to the HO− Hδ+···(CC)δ− interaction. The negative potential becomes more intense in 3′ under the olefinic double bond with positive charge at the backside of the C(sp3) atoms. This ambipolar character would bring significant perturbation to the rotational behavior of H2O and be preponderant for the formation of a H2O−olefin interaction. Optimized Structures and Orientation of the H2O Molecules. Subsequently, we turned our focus back to the entrapped H2O molecule. Using the optimized structures in Figure 7, the H2O molecules were placed at the center of the cages and further optimized at the same level of theory. The optimized structures of H2O@1′, H2O@2′, and H2O@3′ are summarized in Figure 8. The number of possible conformations declines in the order H2O@1′ > H2O@2′ > H2O@3′, suggesting a loss of rotational freedom of H2O. This trend is in good accordance with the relaxation time (Figure 6c). The rotational barrier of H2O inside 1′ was predicted to be small enough (Figure S10), as is observed experimentally (Ea +0.13 kcal/mol for H2O@1), while the orientation of A is preferable among the possible conformations owing to the electrostatic H2Oδ−···C(sp3)δ+ interaction (Figure 8a). In contrast, H2O@ 2′ and H2O@3′ can adopt both conformations (A and B), without considerable difference in ΔG, where one of the OH groups in H2O points to the olefinic double bond, supporting the observed H2O···(CC) interaction. Interestingly, in conformations A, the C(i)···H(1) (i = 1, 2) distances (3.07 Å for H2O-A@2′, 3.02 and 3.13 Å for H2O-A@3′) are more than the sum of van der Waals radii (H and C atoms: 2.90 Å)27 with keeping the directional preference of the O(1)−H(1) bond as the most stable orientation (Figures S11 and S12). This is suggestive of OH/π-type H-bonding dominated by the electrostatic interaction in this longer regime for A. In contrast, the C(i)···H(1) distances in B are substantially shorter than 2.90 Å (2.76 Å for H2O-B@2′ and 2.83, 2.72 Å for H2O-B@ E

DOI: 10.1021/jacs.9b06759 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

Table 2. Natural Charges (q) of H2O, (H2O)2, and Olefinic Double Bond C(1)C(2) (M06-2X/6-31G(d,p))a compd H2O (free) (H2O)2 H2O-A@2′’ H2O-B@2′’ H2O-A@3′’ H2O-B@3′’ a

q[H(1)], q[H(2)] 0.480 0.496 +0.491, +0.492, +0.491, +0.492,

+0.488 +0.487 +0.489 +0.487

Δq[O(1)−H(1)]

q[C(1)], q[C(2)]

−1.440 −1.492 −1.466 −1.471 −1.470 −1.471

+0.002, +0.002 −0.002, −0.002 +0.062, −0.027 +0.059, −0.031

The labels of H, O, and C are shown in Figures 4 and 8.

point of view, the strength of the H2O···(CC) interaction would lie in the order H2O-B@3′ (q[C(2)] −0.031) > H2OA@3′ (−0.027) > H2O-B@2′ (−0.002) > H2O-A@2′ (+0.002). In addition, the C(2) atom in H2O@3 is considered to be dominantly engaged in the H2O−olefin interaction, which can be expressed as HO−H(1)δ+···C(2)δ−C(1). Atoms-in-Molecules. The quantum theory of atoms in molecules (QTAIM) provides a faithful mapping in terms of topological properties for electron densities. The AIM analysis was conducted using the AIM2000 program package.43 The wfn files were created with Gaussian 09 (M06-2X/6-31G(d,p)). The electron density maps for H2O@2′ and H2O@3′ were drawn in Figure 9. The contours displayed with black bold curves are the van der Waals surfaces. The bond critical Figure 8. Optimized structures of (a) H2O@1′’, (b) H2O@2′, and (c) H2O@3′ with values of ΔG at 298 K and ratio under thermal equilibrium in parentheses (M06-2X/6-31G(d,p)).

3′), implying the possible orbital−orbital interaction as an impelling force to construct the OH/π-type H-bondings. So far, it has been recognized that electrostatic attraction has a greater contribution to the OH/π, NH/π, and cation/π interactions, whereas orbital−orbital and/or dispersion interactions are predominant in CH/π and SH/π-bondings.23 Our results, however, strongly suggest that the orientation of the H2O molecule engaged in the OH/π interaction exclusively controls the contribution from the electrostatic attraction and orbital−orbital overlap. Natural Population Analysis. To examine the bonding character of the observed H2O···(CC) interaction in H2O@ 2 and H2O@3, we conducted NPA, which provides a clue to understand the electron distribution of the atomic charge based on occupancies for orthonormal natural atomic orbitals of constituent atoms. For H2O@2′ and H2O@3′, the sum of natural charges corresponding to the H2O molecule were calculated to be 0.000 to +0.001 (M06-2X/6-31G(d,p)), indicating negligible electronic interaction between the H2O and the fullerene cages. As shown in Table 2, the H2O molecules entrapped inside 2′ and 3′ exhibit a larger polarization in the O(1)−H(1) bond with larger q[H(1)] and smaller Δq[O(1)−H(1)] values when compared with non-H-bonded free H2O but slightly smaller than that for Hbonded water dimer (H2O)2 with a trans-linear conformation.42 While H2O@2′ has the same natural charges on C(1) and C(2) atoms (q ± 0.002), H2O@3′ possesses q[C(1)] +0.06 and q[C(2)] −0.03 in both conformations, originated from the electron-donating character of the sp3 carbon attached to the naphthalene ring. From the electrostatic

Figure 9. Cross-sectional images of electron density maps for (a) H2O-A@2′, (b) H2O-B@2′, (c) H2O-A@3′, and (d) H2O-B@3′ with (e) positioning maps of C(1), C(2), H(1), BCP(#1), and BCP(#2) (black dot: H, C, or O atoms, red dot: BCP, red curve: bond path, black bold curve: van der Waals surface). F

DOI: 10.1021/jacs.9b06759 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

positive HBCP values. This indicates that the observed BCPs(#1) are associated with a pure closed-shell interaction without covalent nature. For other BCPs involving BCP[H(1)···C], BCP[H(2)···C], and BCP[O(1)···C], the HBCP values are larger than those for BCPs(#1), implying the less favorable interaction due to destabilization of electron densities at these BCPs. Natural Bond Orbital and Wiberg Bond Order Analyses. The calculations were performed using the NBO 3.1 program package46 at the M06-2X/6-31G(d,p) level of theory. The atom−atom overlap-weighted Wiberg bond orders were analyzed for HO−H(1)···(CC) interactions in H2O@ 2′ and H2O@3′. As summarized in Table 4, the bonding

points (BCPs) were calculated together with their electron densities (ρBCP), Laplacians (∇2ρBCP), and total electron energy densities (HBCP).44 The BCP denotes a saddle point where the gradient of the electron density between two atoms becomes zero, demonstrating the through-space interaction existing between those two atoms. For all structures, several BCPs were observed besides BCP(#1) corresponding to the H2O···(CC) interaction. In conformations A for both H2O@2′ and H2O@3′, the orbital−orbital overlap is predicted to be negligible because BCPs(#1) were located outside the contours at the threshold of the electron densities on the H(1) and C(1)/C(2) atoms (Figure 9a,c). This stands in sharp contrast to conformers B, in which isosurfaces of the electron density are merged between H(1) and C(1)/C(2) atoms (Figure 9b,d). Upon seeing bond paths in H2O@2′ (Figure 9a,b,e), BCP(#1) connects the H(1) atom with BCP(#2), which exists at the center of the C(1)C(2) bond, indicating that the H2O molecule interacts equally with C(1) and C(2) atoms. According to the directionality, the H2O molecule adopts the orientation for the electrostatic interaction to be maximized inside 2′, as is the case with Figure 2b. Contrastingly, BCP(#1) in H2O-B@3′ directly connects H(1) and C(2) atoms, suggesting the stronger orbital−orbital interaction. From these electron density maps, it seems fairly obvious that the orientation of the H2O molecule sensitively affects the electrostatic and orbital−orbital interactions. According to the generalized criteria concerning Hbondings, proposed by Koch and Popelier, the ρBCP values should lie in the range of 0.002−0.035 au (medium to strong interaction),45 which can be seen in the all BCPs(#1) for H2O@2′ and H2O@3′ (0.0041−0.0073 au) (Table 3). The

Table 4. Bond Order of the H2O···(CC) Bondings in H2O@2′ and H2O@3′ (M06-2X/6-31G(d,p))

compd

BCP

ρBCP

∇ ρBCP

HBCP

H2O-A@2′

BCP(#1) BCP[H(1)···C] BCP[H(2)···C] BCP[O(1)···C] BCP(#1) BCP[H(2)···C] BCP[O(1)···C] BCP(#1) BCP[H(2)···C] BCP[O(1)···C] BCP(#1) BCP[H(2)···C] BCP[O(1)···C]

0.0041 0.0041 0.0066 0.0070 0.0073 0.0054 0.0074 0.0042 0.0073 0.0073 0.0073 0.0070 0.0066

0.0126 0.0174 0.0247 0.0280 0.0219 0.0208 0.0291 0.0129 0.0264 0.0288 0.0213 0.0273 0.0240

0.00064 0.00083 0.00085 0.00081 0.0007 0.00079 0.0008 0.00067 0.00087 0.00078 0.00069 0.00078 0.00083

H2O-B@2′

H2O-A@3′

H2O-B@3′

bonding type

bond order

C(1)···H(1)a C(1)···H(1)a C(1)···H(1) C(2)···H(1) C(1)···H(1) C(2)···H(1)

0.0048 0.0080 0.0056 0.0015 0.0078 0.0091

H2O-B@3′ a

The bond order for C(2)···H(1) is exactly the same as that of C(1)···H(1) due to the symmetrical structure of H2O@2′.

strength in conformers B is overwhelmingly larger than those for A. From the absolute values, H2O-B@3′ has the strongest interaction, which is in line with the AIM analysis. The donor− acceptor interaction energies (E(2)) were also computed using the following equation based on the second-order perturbation theory:

Table 3. Electron Densities (ρBCP), Laplacians (∇2ρBCP), and Total Electron Energy Densities (HBCP) of H2O@2′ and H2O@3′ Obtained by AIM Analysis (M06-2X/631G(d,p))a 2

compd H2O-A@2′ H2O-B@2′ H2O-A@3′

E

(2)

ϕD|F |̂ ϕA* 2 = −2 εA* − εD

where the numerator is the Fock matrix element and the denominator is the energy difference between donor (D) and acceptor (A) NBOs. These interaction energies can be regarded as charge transfer energy or stabilization energy associated with delocalization of electron densities. Thus, the larger value offers the greater contribution and strength of the interaction. As listed in Table 5, donor−acceptor interactions are driven by BD (bonding orbital) or LP (lone pair) as donor and BD* (non-Lewis antibonding orbital) or RY* (non-Lewis Rydberg-type orbital) as acceptor. For both compounds, the H2O···(CC) interactions originated from the orbital−orbital overlap between BD[C(1)C(2)] and BD*[H(1)−O(1)] NBOs with the largest E(2) values among those including BD[C(i)C(j)] → BD*[H(k)−O(1)] and LP[O(1)] → RY*[C(i)] interactions (1 < i < j, k = 1, 2). Even though the AIM analysis also predicted these interactions, which were identified with BCP[H(1)···C], BCP[H(2)···C], and BCP[O(1)···C] (Figure 9 and Table 3), the interaction energies are less significant according to the E(2) values, especially in conformers A. The NBOs engaged in the H2O···(CC) interaction for H2O-B@2′ and H2O-B@3′ were depicted in Figure 10. This indicates the significant orbital overlap between BD[C(1)C(2)] and BD*[H(1)− O(1)] NBOs with substantial energetic stabilization. From further analysis, the BD[C(1)C(2)] NBO is identified as a π orbital (2p: 100%) and BD*[H(1)−O(1)] NBO is a σ* orbital

a

Units in au.

∇2ρBCP values enable us to identify whether the electronic charge (ρBCP) is locally concentrated (∇2ρBCP < 0) or depleted (∇2ρBCP > 0) at the corresponding BCPs. The energy density HBCP is an index for judging covalency of weak interactions on the energy basis. For instance, if HBCP < 0, the interaction should be covalent due to stabilization of electrons at BCP.44b Accordingly, larger HBCP values (HBCP > 0) describe the noncovalent nature accompanied by destabilization at the corresponding BCPs. As summarized in Table 3, the local depletion of the electron density was verified at BCPs(#1) with G

DOI: 10.1021/jacs.9b06759 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society Table 5. Donor−Acceptor Interaction Energies E(2) of H2O@2′ and H2O@3′ (M06-2X/6-31G(d,p))a compd

donor NBO

acceptor NBO

E(2)b

H2O-A@2′

BD[C(1)C(2)] BD[C(i)C(j)] LP[O(1)] BD[C(1)C(2)] BD[C(i)C(j)] LP[O(1)] BD[C(1)C(2)] BD[C(i)C(j)] LP[O(1)] BD[C(1)C(2)] BD[C(i)C(j)] LP[O(1)]

BD*[H(1)−O(1)] BD*[H(k)−O(1)] RY*[C(i)] BD*[H(1)−O(1)] BD*[H(k)−O(1)] RY*[C(i)] BD*[H(1)−O(1)] BD*[H(k)−O(1)] RY*[C(i)] BD*[H(1)−O(1)] BD*[H(k)−O(1)] RY*[C(i)]

0.42