Encapsulation of Molecular Hydrogen in Fullerene C60 by Organic Synthesis

See allHide authors and affiliations

Science  14 Jan 2005:
Vol. 307, Issue 5707, pp. 238-240
DOI: 10.1126/science.1106185


In spite of their importance in fundamental and applied studies, the preparation of endohedral fullerenes has relied on difficult-to-control physical methods. We report a four-step organic reaction that completely closes a 13-membered ring orifice of an open-cage fullerene. This process can be used to synthesize a fullerene C60 encapsulating molecular hydrogen, which can be isolated as a pure product. This molecular surgical method should make possible the preparation of a series of C60 fullerenes, encapsulating either small atoms or molecules, that are not accessible by conventional physical methods.

Endohedral fullerenes, the closed-cage carbon molecules that incorporate atoms or a molecule inside the cage (16), are not only of scientific interest but are also expected to be important for their potential use in various fields such as molecular electronics (7), magnetic resonance imaging [as a contrast agent (8)], and nuclear magnetic resonance (NMR) analysis (9, 10). However, development of their applications has been hampered by a severe limitation in their production, which has relied only on physical methods, such as co-vaporization of carbon and metal atoms (2, 3) and high-pressure/high-temperature treatment with gases (914), that are difficult to control and yield only milligram quantities of pure product after laborious isolation procedures.

An alternative approach to synthesizing endohedral fullerenes is “molecular surgery,” in which the cage is opened and then closed in a series of organic reactions (15, 16). For example, an open-cage C60 derivative 1 with a 14-membered ring orifice has been synthesized (17), and the insertion of molecular hydrogen into 1 in 5% yield has also been achieved (15). However, the closure of its orifice was not attempted. A C60 derivative 2, which we synthesized recently (18), has a 13-membered ring orifice with a sulfur atom on its rim and a relatively circular shape compared with the elliptic orifice of 1. This opening has enabled us to insert molecular hydrogen through this orifice in 100% yield (19). When matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry was conducted under enhanced laser power on compound 2 encapsulating hydrogen (H2@2), we observed a molecular ion peak for H2@C60 at a mass-to-charge ratio (m/z) of 722 (19). This result suggested that H2@2 could be a precursor for H2@C60 in an actual chemical transformation. We now report the synthesis of 100% pure H2@C60 from H2@2 (Scheme 1).

Fig. 1.

Size reduction and closure of the orifice of the open-cage fullerene encapsulating hydrogen, in a four-step process. Percentage values are product yields; that shown in parenthesis is that based on the consumed precursor. m-CPBA, r.t., and o-DCB stand for m-chloroperbenzoic acid, room temperature, and o-dichlorobenzene, respectively.

Fig. 2.

Proposed reaction mechanism for the formation of C60 from compound 5 by heating. Only the tops of the molecules are shown. Ph and Py stand for phenyl and 2-pyridyl groups, respectively.

Fig. 3.

Structural characterization of H2@C60. (A) Expanded 13C NMR spectrum (75 MHz, o-DCB-d4) of H2@C60 contaminated by 9% C60. (B) Expanded 13C NMR spectrum (75 MHz, o-DCB-d4) of purified H2@C60. (C) MALDI-TOF mass spectrum (positive ionization mode, dithranol matrix) of purified H2@C60. (D) Predicted isotope distribution pattern for H2@C60.

Scheme 1.

Encapsulated H2 escapes from the cage when the compound H2@2 is heated above 160°C (19), so high temperatures must be avoided if the chemical synthesis of H2@C60 is attempted from H2@2. With such a precaution being taken, we performed a stepwise reduction of the orifice size of H2@2 and completed its closure by a thermal reaction. The application of heat to the last step did not cause a serious loss of H2, because the orifice size was already reduced sufficiently to prevent such loss.

The first step involved the oxidation of the sulfide unit (-S-) in H2@2 to a sulfoxide unit (>S=O) to give H2@3. The resulting >S=O unit was removed by a photochemical reaction to produce H2@4 (Fig. 1, steps A and B) (20). Both reactions proceeded at room temperature with yields of 99% and 42% (68% for step B based on consumed H2@3), respectively. The MALDI-TOF mass spectrum of H2@4 exhibited the molecular ion peak of H2@C60 as a base peak, indicating its enhanced accessibility from H2@4 as compared to H2@2. The spectrum, however, also showed the presence of empty C60 in 20% yield relative to H2@C60 and indicated that further reduction of the orifice size was needed. Thus, in the next step, two carbonyl groups in H2@4 were reductively coupled by the use of Ti(0) (21) at 80°C, to give H2@5 with an eight-membered ring orifice (Fig. 1, step C).

At each process in these three steps, complete retention of encapsulated H2 was confirmed by observing the characteristically upfield-shifted NMR signal of the incorporated H2. The integrated signal intensity exactly corresponded to 2.00 ± 0.05 H for the signals at a chemical shift δ of –6.18 parts per million (ppm) in H2@3, at –5.69 ppm in H2@4, and at –2.93 ppm in H2@5, with reference to the 2.00 H signal for two aromatic protons. The gradual downfield shift of the hydrogen signal observed at steps B and C reflects the formation at each step, within the fullerene cage, of a fully π-conjugated pentagon, which exerts a strong deshielding effect through its paramagnetic ring currents (22).

Finally, complete closure of the orifice was achieved by heating powdery H2@5 in a glass tube at 340°C for 2 hours under vacuum (Fig. 1, step D). The desired product H2@C60 (118 mg, contaminated with 9% empty C60) was obtained in 67% yield by passing a carbon disulfide solution of the crude product through a silica-gel column. Similar results were obtained when H2@5 was heated at 300°C for 24 hours, at 320°C for 8 hours, or at 400°C for 2 min. Thus, H2@C60 was synthesized in a total yield of 22% from H2@2, which can be obtained in 40% yield from consumed C60 (18, 19).

We presume that the closure of the orifice takes place by way of a thermally allowed [π2s + π2s + π2s] electrocyclization reaction that produces two cyclopropane rings (Fig. 2). Sequential radical cleavage and a retro [σ2s + σ2s + σ2s] reaction produce C60 by splitting off 2-cyanopyridine and diphenylacetylene.

The 13C NMR spectrum of the desired product exhibited a signal at δ = 142.844 ppm together with a very small signal at δ = 142.766 ppm (Fig. 3A), the latter corresponding exactly to the signal of empty C60. In an expanded spectrum obtained with 56,576 data points for a 50-ppm spectral width, the integrated peak areas of these signals yield an estimated ratio of H2@C60 and empty C60 of 10:1.

We separated H2@C60 from C60 through recycling high-performance liquid chromatography on a semipreparative Cosmosil Buckyprep column (two directly connected columns, 25 cm by 10 mm inner diameter, with toluene as a mobile phase; flow rate, 4 ml min–1; retention time, 395 min for C60 and 399 min for H2@C60). Isolated H2@C60 was judged to be 100% pure on the basis of a single 13C NMR signal at 142.844 ppm (Fig. 3B), the results of high-resolution fast-atom-bombardment mass spectrometry (calculated molecular weight for C60H2: 722.0157; found: 722.0163), and the agreement of the observed and predicted isotope distribution patterns in the MALDI-TOF mass spectrum (Fig. 3, C and D), in addition to correct elemental analysis for hydrogen (calculated for C60H2: C, 99.72, and H, 0.28%; found: C, 99.04, and H, 0.24%).

The very small downfield shift (0.078 ppm) observed for the 13C NMR signal of H2@C60 (as compared to empty C60) indicates that the electronic property of the fullerene cage is largely unaffected by the encapsulation of H2. The ultraviolet-visible and infrared spectra of H2@C60 are also exactly the same as those of empty C60. This situation contrasts with the cases of Kr@C60 (13) and Xe@C60 (12), in which larger downfield shifts are observed (0.39 ppm and 0.95 ppm, respectively), caused by appreciable electronic and van der Waals interactions between the C60 cage and the encapsulated atoms, which are much larger than H2.

The 1H NMR signal for the encapsulated hydrogen of H2@C60 in o-dichlorobenzene-d4 was observed at δ = –1.44 ppm, which is 5.98 ppm upfield-shifted relative to the signal of dissolved free hydrogen. The extent of this upfield shift is comparable to that observed for 3He@C60 (6.36 ppm) (9, 10) in 3He NMR relative to free 3He. This result shows that the shielding effect of total ring currents of the C60 cage is nearly the same, regardless of the paramagnetic species inside the cage.

The irrelevance of the encapsulated H2 to the electronic character of the outer cage was also demonstrated by cyclic voltammetry (0.5 mM in o-dichlorobenzene with 0.05 M Bu4NBF4 for reduction and 0.5 mM in 1,1,2,2-tetrachloroethane with 0.1 M Bu4NPF6 for oxidation). The voltammogram of H2@C60 exhibited four reversible reduction waves and one irreversible oxidation peak at the same potentials as C60, within an experimental error of ±0.01 V.

In order to clarify the reactivity of H2@C60, the solid-state mechanochemical [2+2] dimerization reaction (23) was conducted. A mixture of H2@C60 and 1 molar equivalent of 4-aminopyridine as the catalyst (24) was vigorously shaken by the use of a high-speed vibration mill for 30 min under N2 according to our previous procedure (23, 24). The 1H NMR spectrum of the product mixture exhibited a signal at δ = –4.04 ppm of the [2+2] dimer, (H2@C60)2, and a signal of unchanged H2@C60 at δ = –1.44 ppm, in an integrated ratio of 3:7. This result indicates that the dumbbell-shaped dimer of H2@C60 is formed in the same yield as that for the reaction of empty C60 (24) (Fig. 4). No effect of the encapsulated H2 was observed upon reactivity of the C60 cage. The extent of the upfield shift of the 1H NMR signal (2.60 ppm) observed for the dimer (H2@C60)2 was similar to that observed upon the same dimerization reaction in 3He NMR (2.52 ppm) (24) for 3He encapsulated in the ratio of ∼0.1% in C60 (9, 10).

Fig. 4.

Mechanochemical solid-state dimerization of H2@C60 by the use of a high-speed vibration milling (HSVM) technique.

The endohedral fullerene H2@C60 is nearly as stable as C60 itself. For example, the encapsulated H2 does not escape even when heated at 500°C for 10 min. Thus, H2@C60 can be viewed as a stable hydrocarbon molecule that has neither C-H covalent bonds nor C···H interactions. It is likely that our method could be used to synthesize endohedral fullerenes such as D2@C60 and HD@C60, as well as the homologous series with C70. Our work here complements the total chemical synthesis of C60 recently achieved by Scott and co-workers (25) and implies that organic synthesis can be a powerful means for the production of yet unknown classes of endohedral fullerenes.

Supporting Online Material

Materials and Methods

Figs. S1 to S9


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article