High-Temperature Superconductivity in Lattice-Expanded C60

See allHide authors and affiliations

Science  28 Sep 2001:
Vol. 293, Issue 5539, pp. 2432-2434
DOI: 10.1126/science.1064773

This article has been retracted. Please see:


C60 single crystals have been intercalated with CHCl3 and CHBr3 in order to expand the lattice. High densities of electrons and holes have been induced by gate doping in a field-effect transistor geometry. At low temperatures, the material turns superconducting with a maximum transition temperature of 117 K in hole-doped C60/CHBr3. The increasing spacing between the C60 molecules follows the general trend of alkali metal–doped C60 and suggests routes to even higher transition temperatures.

The superconducting properties of various materials can be modulated by the application of an electric field, and a variety of field-effect devices have been studied lately (1–9). We recently demonstrated the switching between insulating and superconducting behavior in single crystals of C60. The gate-induced superconductivity is observed for electron doping (5) as well as hole doping (7). The higher superconducting transition temperature (T c) for hole doping may be ascribed to a larger density of states at the Fermi level and stronger coupling to phonons (8, 10, 11). C60 is a particularly interesting superconductor because the dominant electron-phonon interaction, being an on-site intramolecular property, can be conceptually separated from the electronic density of states, which is given by the distance between adjacent molecules. Expanding the lattice, therefore, increases the density of states, and the resulting increase of T c is well documented in alkali metal–doped bulk samples (A3C60) (12, 13). The observation of gate-induced hole doping of C60 resulting in aT c of 52 K suggests that significantly higher T c's could be anticipated in suitably “expanded” C60 crystals. Indeed, here we report on raising T c to 117 K with such methods.

Undoped C60 single crystals have been grown from the vapor phase in a stream of hydrogen (14). CHCl3 and CHBr3 are intercalated into C60 in solution according to (14). The co-crystals exhibit a hexagonal crystal structure. The expansion of the lattice corresponds to a cubic lattice constant of ∼14.28 and 14.43 Å for CHCl3 and CHBr3, respectively (15). Field-effect devices (5,7) were prepared on growth surfaces of undoped as well as intercalated single crystals (inset of Fig. 1). The observation of ambipolar transport (that is, n- and p-channel activity in intercalated crystals) reveals the absence of severe hole trapping or significant electron doping due to CHBr3 or CHCl3. Nevertheless, the intercalation results in additional disorder as compared with undoped C60 single crystals. This disorder increased the residual resistivity ρ0. In the case of electron doping (about three electrons per C60), ρ0 is in the range of 500 to 650 microhm·cm for intercalated crystals as compared with 250 to 300 microhm·cm for undoped C60. Thus, the resistivity values for intercalated samples are similar to those reported for alkali metal–doped material (7). The doping level of the material is estimated from the gate charge, assuming that only one monolayer takes part in the conduction (5–7).

Figure 1

(A) Channel resistance of a C60/CHBr3 co-crystal field-effect device as a function of temperature for different hole doping (1.3, 1.6, 1.9, 2.25, 2.7, and 3.2 holes per C60 molecule). The maximumT c is at 117 K, and the main drop of the resistance starts at 115 K. The schematic structure of the device is shown in the inset. (B) Comparison of optimum hole-doped C60, as grown and intercalated with CHCl3 and CHBr3, respectively.

When the gate voltage in the field-effect device was increased, the charge carrier concentration increased and the temperature dependence exhibited metallic characteristics. When the hole density p reached approximately one hole per molecule, superconductivity above 1.7 K was observed. More holes led to higher values of T c, and the value ofT c depended on the size of the intercalant molecule and p. The highestT c we observed was for C60/CHBr3 with 3 to 3.5 holes per C60 molecule [with an onset at 117 K and a rapid drop at 115 K (Fig. 1)]. When a magnetic field was applied perpendicular to the metallic layer, the resistive transition shifted to lower temperature, and from the shift, interpreted as the upper critical field H c2, values for the coherence length ξGL were extracted by means of the standard extrapolation (12, 13). A systematic decrease of ξGL with increasingT c was observed (Fig. 2). This general trend is expected because ξ is proportional tov f/T c, where v f is the Fermi velocity. A more detailed analysis would require that parameters such as the electronic mean free path and density-of-states effects have to be taken into account. In any case, the coherence length at highestT c is only approximately two intermolecular distances.

Figure 2

Coherence length as a function ofT c for hole-doped as-grown and intercalated C60 single crystals. The decrease of ξ with increasingT c is in accordance with theoretical considerations.

The shape of the carrier concentration dependence ofT c was similar in samples with or without intercalation (Fig. 3), and this was the case for both electron and hole doping. For electron doping, superconductivity was observed only in the range of ∼2.5 to 3.5 electrons per C60 molecule, which corresponds to a narrow region around half-filling of the conduction band that is derived from three molecular states. In contrast, a smooth increase ofT c was observed for hole doping. The maximum was reached between 3 and 3.5 holes per C60. The upper end of the doping level is given by the electric breakdown strength of the gate oxide. In Fig. 3, all of the data are collected and the scaling of T c among the different variants of C60 is evident. Among the hole metals, the T c ratios are 2.3:1.6:1 for C60/CHBr3–C60/CHCl3–C60. The same ratios are also found for the respective electron metals, which may be a consequence of changes in the bandwidth and density of states upon intercalation. Density-of-states effects had been thought to be the main reason for the changes inT c in chemically doped A3C60 (12, 13). When we compare the present results with those data, we find good agreement (Fig. 4). This shows that the structural change from a face-centered cubic to a hexagonal structure does not significantly influence the superconducting properties of the material. TheT c in the gate-doped materials is slightly lower than expected from the bulk data. This difference might be a consequence of the two-dimensionality of the field-effect structure (5), because the charges at these high densities are confined in the topmost layer of the crystal. An approximately linear increase of T c with increasing lattice constant was observed for hole-doped intercalation compounds, pointing to a vanishing of T c for a lattice parameter of ∼13.9 to 14.0 Å. Recent theoretical considerations are in contrast to this result (16).

Figure 3

Variation in T c as a function of charge carrier density for electron- and hole-doped C60 crystals. Incorporation of CHCl3 and CHBr3 expands the lattice, enhances the electronic density of states, and causes T to increase. Holes are doped into the valence band that holds 10 charge carriers per C60molecule, and electrons are accommodated in the conduction band holding six electrons.

Figure 4

Variation in T c as a function of lattice constant for electron- and hole-doped C60 crystals. Square symbols are the present results using gate doping with electrons (open squares) or holes (solid squares). Open circles are for chemically doped A3C60bulk samples, where the variation of the lattice parameter is due to the different size of the ions (A).

The analysis in Fig. 4 suggests that even higher values of T c should be achievable. If the lattice parameter can be further increased by ∼1%,T c is expected to exceed 150 K. The question is how to further expand the lattice without finally losing the cohesion. Although Coulomb forces between the anions and cations provide some additional stability of the A3C60crystals, the expansion of van der Waals–bonded C60 with neutral molecules poses new challenges. In this context, the field-effect transistor geometry with the charged layer sandwiched in between the gate dielectric and the bulk of the crystal might provide additional advantages. It is generally thought that superconductivity in C60 is mediated by electron-phonon interaction. In light of T c exceeding 100 K, it will be particularly revealing to evaluate the effective Coulomb interaction and the role of nonadiabatic pairing channels, because the Fermi energy is not much greater than the phonon energy (12,13, 17, 18).

  • * To whom correspondence should be addressed. E-mail: hendrik{at}


View Abstract

Stay Connected to Science

Navigate This Article