A Single Molecule of Water Encapsulated in Fullerene C60

See allHide authors and affiliations

Science  29 Jul 2011:
Vol. 333, Issue 6042, pp. 613-616
DOI: 10.1126/science.1206376


Water normally exists in hydrogen-bonded environments, but a single molecule of H2O without any hydrogen bonds can be completely isolated within the confined subnano space inside fullerene C60. We isolated bulk quantities of such a molecule by first synthesizing an open-cage C60 derivative whose opening can be enlarged in situ at 120°C that quantitatively encapsulated one water molecule under the high-pressure conditions. The relatively simple method was developed to close the cage and encapsulate water. The structure of H2O@C60 was determined by single-crystal x-ray analysis, along with its physical and spectroscopic properties.

The unusual properties of bulk water, such as its high boiling and melting points, high dielectric constant, and ability to act as both acid and base, arise from hydrogen bonding in bulk. Clusters of H2O molecules, both in the gas phase (1) and confined in nanoscale spaces defined by the hydrophobic interiors of carbon nanotubes (2) or a self-assembled coordination cage (3), also undergo hydrogen bonding. However, a single molecule of H2O without any hydrogen bonding to other organic molecule or coordination to metal is rare so far (4).

The inner space of the fullerene C60, which is spherical with a diameter of 3.7 Å, is suitable to entrap a water molecule. When atoms or molecules are encapsulated in fullerenes, it is often possible to control the properties of the outer carbon cage as well as to study the isolated species. Endohedral fullerenes encapsulating a wide variety of species, including metal ions, rare gases, and nitrogen atoms, have been synthesized with physical methods under harsh conditions such as (i) arc discharge of carbon rods containing metal, (ii) ion implantation to empty C60 and C70, and (iii) high-pressure and high-temperature treatments of empty fullerenes with rare gases (5). However, these methods are not suitable to obtain endohedral fullerenes encapsulating small molecules.

The fourth method for synthesizing endohedral fullerenes is the molecular surgical approach (6), which includes construction of an opening on an empty C60 or C70 cage, insertion of small molecules through it, and then restoration of the original framework, retaining the encapsulated species (7). By using such an approach, hydrogen molecules have been entrapped in C60 and C70 in macroscopic quantities (8, 9). To apply this method to a larger molecule than H2, developments of synthetic methods are needed to create open-cage fullerenes with an opening large enough for the small molecule to pass through (10). Fullerenes with large openings to encapsulate an H2O molecule have been reported (11, 12), but the encapsulated H2O is in an equilibrium, with outer H2O molecules existing in the system in large amounts. Furthermore, attempts to close such large openings have not been reported.

If the size of an opening on fullerenes can be controlled dynamically—that is, a small opening changes into a larger one in situ under specific conditions and regenerates itself again after insertion of a molecule through it—wide varieties of endohedral fullerenes could be synthesized by organic synthesis because restoration of the small opening should be easier than that of the larger one. We report the macroscopic synthesis of H2O@C60 using such a dynamic control of the opening size with a series of organic reactions, as well as the structure and the properties of the molecule in which a molecule of H2O is completely isolated in a confined space.

The synthetic route for open-cage C60 derivatives is outlined in Fig. 1A. Applying our previous synthetic route (13, 14), diketones 3a and 3b were synthesized in 29 and 43% isolated yields, respectively, from the reaction of C60 with pyridazine derivatives 1a and 1b followed by photochemical cleavage of one of the C=C double bonds on the rim of the opening on intermediates 2a and 2b, respectively (15) (figs. S1 to S12). The lowest unoccupied molecular orbital (LUMO) of open-cage C60 with a similar opening motif is mainly located on the conjugated butadiene moiety (13), and sulfur (13, 14) and hydrazine derivatives (11) were reported to react with that moiety. N-Methylmorpholine N-oxide (NMMO) is known as a nucleophilic oxidant (16), and we found that the reaction of 3 with 2.3 equivalents of NMMO in wet tetrahydrofuran (THF) at room temperature led to the synthesis of 5a and 5b, respectively, in good yields after purification with silica gel chromatography (figs. S13 to S18).

Fig. 1

(A) Synthetic route of open-cage C60 5 (1-chloronaphthalene being abbreviated as 1-ClNp) and (B) the x-ray structure of 5a at the 50% probability level. The solvent molecules were omitted for clarity.

The structure of 5a determined by single-crystal x-ray analysis (Fig. 1B) shows that the opening is constructed by the 13-membered ring containing two hemiacetal carbons, in addition to two carbonyl carbons (table S1). This molecule can be considered as the hydrate of tetraketone 4a having the 16-membered-ring opening with four carbonyl carbons. Because wet THF was used in this reaction, a water molecule attacked one of the carbonyl carbons in 4a to give 5a. When the hydrate 5a was heated at reflux temperature in toluene for 30 min, complete transformation of 5a was observed and a new species, probably 4a (fig. S19), was detected with high-performance liquid chromatography (HPLC) (Buckyprep, toluene) analysis. However, because this compound quickly went back to hydrate 5 at room temperature, characterization of 4 was not achieved yet. As a similar reaction is known in the literature (17), the conversion between 5a and 4a would be performed quantitatively. The size of the opening on tetraketone 4 is greater than that of hydrate 5, implying that insertion of an H2O molecule into 4 could be possible.

When hydrate 5b was refluxed in wet toluene (15 μL of water in 10 mL toluene) and the resulting product was analyzed by 1H nuclear magnetic resonance (NMR), an H2O molecule encapsulated inside 5b was detected in 8% yield by comparison of the integrated peak areas. The encapsulation ratio is apparently lower than the reported values by Iwamatsu et al. (75%) (11) and by Gan and co-workers (88%) (12), reflecting the smaller size of the opening on tetraketone 4b than those on their compounds. However, quantitative encapsulation of an H2O molecule inside 5b was achieved when this process was conducted under the high pressure of 9000 atm at 120°C for 36 hours (Fig. 2 and figs. S20 to S22). The high pressure is critical; only 40% encapsulation was observed under 5000 atm under the similar conditions.

Fig. 2

Encapsulation of an H2O molecule to open-cage C60 5b and restoration of the opening for the synthesis of H2O@C60.

The formation of H2O@5b can be explained by the dynamic control of the opening size. Elimination of a water molecule from 5b generates 4b. The insertion of an H2O molecule into 4b takes place through the 16-membered-ring opening generated in situ, followed by addition of a water molecule to regenerate 5b encapsulating an H2O molecule. The 1H NMR spectrum of H2O@5b displayed a strongly shielded sharp signal at –9.87 parts per million (ppm), corresponding to the encapsulated H2O molecule (11, 12). The encapsulation was also supported by fast atom bombardment mass spectrometry analysis displaying the molecular ion peak at a mass-to-charge ratio m/z 1139 (M+H+), corresponding to H2O@5b (15).

Our synthetic pathway for closing the 13-membered-ring opening on H2O@5b is a coupling reaction of two carbonyl groups with a phosphite ester (18). As shown in Fig. 2, the reaction of H2O@5b with excessive amounts of P(Oi-Pr)3 (i-Pr, isopropyl) in refluxing toluene gave H2O@2b in 50% isolated yield without the loss of the encapsulated H2O molecule (figs. S23 to S25). The reaction eliminated one water molecule from H2O@5b to generate H2O@4b, followed by two successive carbonyl couplings on H2O@4b and then on H2O@3b, to give H2O@2b. As a control experiment, empty 3a was converted into 2a in 80% under the same conditions. The first carbonyl coupling from H2O@4b to H2O@3b appeared to proceed with ~60% yield. This one-pot process is very simple compared with our previous method for the synthesis of H2-encapsulating endofullerenes, in which three reactions were needed to reduce the size of the openings from 13-membered rings to 8-membered rings, with purification being necessary at each step (8, 9).

The final step in the synthesis of H2O@C60 is removal of the organic addend from H2O@2b. We mixed H2O@2b (50 mg) with neutral Al2O3 (1 g) in a mortar, and the resulting mixture was heated at 360°C for 1 hour under vacuum to give H2O@C60 as brown powder in 29% yield after purification with silica gel chromatography. This step likely proceeded though a [4+2]cyclization, retro[4+4]reaction (19, 20), and formation of a bicylo[2.2.0]hexene moiety followed by a retro[4+2]reaction, which would be the reverse of the reaction of the addition, or radical cleavage of four C-C single bonds (8). The Al2O3 support might help such reactions as a solid acid, which dilutes the substrate and helps prevent undesirable intermolecular reactions. Without the Al2O3 support, the yield of H2O@C60 decreased to 3%. When a mixture of H2O and D2O, instead of pure H2O, was applied to 5b under the same conditions described above, it was possible to obtain a mixture of H2O@C60, HDO@C60, and D2O@C60 (15).

Direct evidence for the existence of the encapsulated H2O molecule inside C60 was obtained from single-crystal x-ray analysis (tables S2 to S4). Although single crystals of H2O@C60 without any solvated molecules were obtained from o-xylene solution, the x-ray diffraction data analysis showed that this crystal has severe orientation disorder of the C60 cage, even at –173°C. Such problems of orientation disorder can be solved by formation of molecular complexes with porphyrin derivatives developed by Balch and co-workers (21). From the 1:2 mixture of H2O@C60 and nickel (II) octaethylporphyrin (NiOEP) in o-xylene solution, the single crystals with enough size and quality were obtained. The x-ray diffraction data was collected at –173°C and analyzed without any restraints. No such disorder was observed for the molecular complex H2O@C60•(NiOEP)2 with the final R1 value of 3.64% (Fig. 3). The oxygen atom of the encapsulated H2O molecule is located just at the center of the C60 cage. This positioning is similar to that seen for Kr@C60 (21) and H2@C60 (22) but different from that of Li@C60SbCl6, where the encapsulated Li atom is located off-center (23). The position of the hydrogen atom of the encapsulated H2O molecule was refined experimentally, and the OH bonds were found to be directed toward the Ni atoms through the C60 cage (Fig. 3B). The single crystals of empty C60•(NiOEP)2 were also obtained with the same space group and similar cell constants. Comparison of the bond lengths and bond angles between H2O@C60•(NiOEP)2 and empty C60•(NiOEP)2 revealed that the encapsulated H2O molecule does not affect the structure of the outer C60 cage (15).

Fig. 3

X-ray structure of the molecular complex H2O@C60·(NiOEP)2 at the 50% probability level. (A) The viewpoint from the two five-membered rings overlapped with hydrogen atoms omitted for clarity and (B) the viewpoint from the two six-membered rings overlapped with displaying hydrogen atoms determined experimentally for H2O and generated geometrically for NiOEP.

No escape of the encapsulated H2O molecule from H2O@C60 was observed when the powder of H2O@C60 was heated at 420°C for 30 min under vacuum and the resulting material was analyzed by HPLC. Atmospheric-pressure chemical-ionization mass spectrometry analysis of H2O@C60 (fig. S26) predominantly showed peaks at m/z 738 with the isotopic distribution pattern of C60H2O. The ultraviolet-visible spectrum of H2O@C60 in toluene (fig. S27) is almost superimposable with that of empty C60, indicating no detectable electronic interaction present between the encapsulated H2O molecule and the outer C60 cage. The 1H NMR spectrum of H2O@C60 in o-dichlorobenzene-d4 (ODCB-d4) (fig. S28) revealed a sharp singlet signal at –4.81 ppm, reflecting a strong shielding effect of the C60 cage. The signal of dissolved H2O in ODCB-d4 appeared at 1.39 ppm. The value of upfield shift (6.2 ppm) for H2O@C60 is very close to that of H2@C60 (5.9 ppm) (8) and 3He@C60 (6.1 ppm) (24). Interestingly, apparent couplings between the proton and the deuteron of HDO@C60 were detected with JH-D 0.9 Hz in the 1H NMR and JH-D 0.9 Hz in the 2H NMR (figs. S29 to S30), indicating no exchange of the proton and the deuteron inside the C60 cage.

The 13C NMR (ODCB-d4) spectrum (fig. S31) showed a sharp signal at 142.89 ppm, suggesting rapid rotation of the encapsulated H2O molecule inside the C60 cage with respect to the time scale of NMR. In a series of endohedral C60 encapsulating rare gases and an H2 molecule, the differences in 13C NMR chemical shifts compared with that of empty C60 (142.78 ppm in ODCB-d4) reflect the van der Waals radii of the encapsulated species: Δδ 0.02 ppm for He (25), Δδ 0.08 ppm for H2 (8), Δδ 0.17 ppm for Ar (26), Δδ 0.39 ppm for Kr (27), and Δδ 0.95 ppm for Xe (28). The difference in chemical shifts for H2O in this study is 0.11 ppm (fig. S32), which is consistent with the size of an H2O molecule between that of H2 and Ar. No difference in chemical shift of the 13C NMR was observed among H2O@C60, HDO@C60, and D2O@C60.

Separations of He@C60, H2@C60, Ar@C60, Kr@C60, and Xe@C60 from empty C60 were usually difficult and possible only when an HPLC equipped with Buckyprep [3-(1-pyrenyl)propylsilyl] or PYE [2-(1-pyrenyl)ethylsilyl] column(s) was used with many time-consuming recycles. In contrast, H2O@C60 was easily separated from empty C60 by the single-stage HPLC (Buckyprep, toluene), with retention times of 7.93 and 8.28 min for empty C60 and H2O@C60, respectively (fig. S33). It is believed that π-π interaction of the fullerene cage with pyrene moieties attached to the silica gel surface is important for the separation, and the presence of the encapsulated H2O molecule might influence such interactions. One of the characteristic properties of an H2O molecule is its high dipole moment, whereas C60 with the Ih symmetry does not have the dipole. Thus, we expect that H2O@C60 should be a polar molecule. The density functional theory (DFT) calculations at the M06-2X/6-311G(2d,p) level of theory (29) with basis set superposition error correction during structural optimization showed that the dipole moments of H2O, C60, and H2O@C60 are 2.02, 0.00, and 2.03 Debye, respectively.

Infrared spectroscopy is a useful method to study the properties of water (30). However, the vibrational frequency analysis of H2O@C60 by the DFT calculations suggested that the symmetric and asymmetric stretching modes of the H2O molecule inside C60 (3810 and 3894 cm–1) should be very weak (15). It was difficult to see spectral features clearly in the observed Fourier transform infrared spectrum (diffuse reflectance spectroscopy, KBr, fig. S34), probably because of the shielding effect of the dipole (30). The observed peaks for H2O@C60 were 1429.3, 1182.4, 576.7, and 526.6 cm–1, which were exactly the same as those of empty C60.

Upon cyclic voltammetry in ODCB with 0.1 M Bu4NPF6 as a supporting electrolyte, an irreversible oxidation peak and four quasi-reversible reduction waves were observed (fig. S35), and the redox potentials were determined by differential pulse voltammetry (fig. S36) as +1.32, –1.08, –1.46, –1.91, and –2.38 V versus a ferrocene/ferrocenium couple within the potential window. These values were almost the same as those of empty C60 (+1.32, –1.08, –1.47, –1.92, and –2.39 V) under the same conditions. This result indicated that the single molecule of H2O is electrochemically stable under the hydrophobic environment inside C60.

The H2O@C60 molecule, which can be considered as “wet fullerene” or “polar C60,” should allow the study of the intrinsic properties of a single molecule of H2O, such as ortho and para conversion (31). When this synthetic methodology is expanded to other small molecules into C60, C70, and higher fullerenes, research work on the isolated molecule or the control of physical properties of outer fullerene cages will progress in the near future.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S36

Tables S1 to S4

References 32, 33

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Acknowledgments: This work was supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) project on Integrated Research on Chemical Synthesis, Grants-in-Aid for Scientific Research on Innovative Areas (20108003, pi-Space) and for Scientific Research (A) (23241032), and the Global COE Program Integrated Materials Science (B-09) from MEXT, Japan. The structural coordinates have been deposited with the Cambridge Data Bank. The accession numbers for the three x-ray structures are as follows: 5a·(o-xylene)2.5, CCDC 828041; H2O@C60·(NiOEP)2, CCDC 828029; empty C60·(NiOEP)2, CCDC 828028.
View Abstract

Stay Connected to Science

Navigate This Article