Buckyball Crystals Made to Superconduct at 117 K

The discovery in the early 1990s that C₆₀ crystals could superconduct at temperatures of several tens of kelvins raised hopes that such "buckyballs" might rival the copper-oxide superconductors, whose T_c's go as high as 150 K under pressure. Those hopes have now been fulfilled.

The first C₆₀ superconductors were made by incorporating alkali atoms into the spaces between the closely packed C₆₀ spheres. These atoms acted as electron donors because their outer electrons were quickly stripped away by the buckyballs, which have a high affinity for electrons. Unfortunately, T_c seemed to reach a maximum of 40 K, obtained by doping C₆₀ with rubidium atoms. Last year, while working at Bell Labs, Lucent Technologies, Jan Hendrik Schön (also of the University of Konstanz, Germany), Christian Kloc, and Bertram Batlogg (now at ETH Zürich) took T_c up to 52 K by finding a way to inject holes instead of electrons (see Physics Today, January 2001, page 15). Nice, but still far from the T_c region reached by the cuprates.

The achievement of 52 K was only a taste of things to come, however. The same Bell Labs-ETH team noted at the time that the T_c for electron-doped C₆₀ increases with the size of the dopants: Larger dopants spread the C₆₀ molecules farther apart and lead to higher values of T_c, by as much as a factor of 3 for a small percentage change in lattice spacing: When the C₆₀ molecules are farther apart, the energy bands grow narrower, so that the density of states--and concomitantly the T_c--gets higher. Perhaps, the trio reasoned, the same trend would also govern hole-doped crystals.

There was certainly room to expand the C₆₀ crystal. Unlike in electron doping, in which electrons are brought in by the intercalated alkali atoms, the holes are added to the crystal electronically. The T_c of 52 K had been measured on a C₆₀ crystal with excess holes but no intercalated atoms. For the new study, the researchers added neutral molecules to expand the crystal, hoping to realize the promised gain in T_c.

They have now succeeded in their quest.¹ By intercalating molecules of tribromomethane (CHBr₃) into the C₆₀ lattice, the experimenters changed the cubic lattice constant from 14.16 Å to 14.43 Å and thereby raised T_c to 117 K. As shown in the top panel of the figure, the higher the hole concentration, the higher the T_c, up to around 3-3.5 holes per C₆₀ molecule. At this doping level, the critical temperature increases from 52 K to about 80 K and 117 K with insertion of trichloromethane (CHCl₃) and CHBr₃, respectively, as seen in the bottom panel of the figure.

Onset of superconductivity

The method for injecting charges is shown in the inset in the figure. When a voltage is applied to the gate, charges of opposite sign accumulate on the interface between the insulating aluminum oxide layer and the single buckyball crystal. In this way, the experimenters can dope the crystal with either electrons or holes, depending on the sign of the gate voltage. The resistivity is measured by monitoring the current that flows from the source to the drain electrode. The experimenters must have crystals of exceptional purity; otherwise, the charge carriers can be trapped locally.

The magnitude of the gate voltage controls the degree of doping, and the doping determines the onset of superconductivity, as seen in the figure. There is considerable evidence that the charge carriers are all confined to a single layer of molecules at the surface of the crystal. So what's seen is superconductivity in two dimensions.

In searching for a compound that would expand the C₆₀ crystal by just the right amount, the Bell Labs-ETH team drew on numerous studies of intercalated compounds, including previous measurements² of the lattice parameters of C₆₀ intercalated with CHBr₃ and CHCl₃. Kloc had had experience slipping those molecules from solution into a C₆₀ lattice.³ Unlike the case of alkali metal atoms, in which the transfer of electrons results in ionic bonding between the intercalated atoms and the C₆₀ lattice, the CHCl₃ and CHBr₃ molecules remain neutral; with them in place, the crystal barely holds together. Still, the experimenters hope to find another molecule that will expand the lattice spacing by another percent because just that small amount can increase T_c to 150 K. If they do succeed, the results will be of interest for a limited range of applications: "You can't make wires out of this stuff," says Batlogg, "but the equivalent of thin-film electronics will be feasible."

"The implications of this work are gigantic," said Robert Cava of Princeton University. "They show us that the only thing that's stood in the way of finding more 100-K superconductors is thermodynamics." According to a prevailing view, he explained, the same interactions that give rise to conventional superconductivity can also break crystal structures apart or cause structural distortions that change the electronic states at the Fermi energy and thus kill the potential for superconductivity. Normally, compounds are made by heating the components together at high temperatures; that process gives the system the energy needed to overcome the activation barriers and reach its thermodynamic ground state, where distortions set in. But Batlogg and company have foiled thermodynamics by forming their compound, with the correct size and electron count, at low temperature. The same trick can be tried on other materials, Cava suggested.

Schön, Kloc, and Batlogg are now looking for superconductivity in both electron- and hole-doped C₇₀. If an electron-phonon mechanism is responsible for superconductivity in these materials, one would anticipate a lower T_c for C₇₀ because of the weaker electron-phonon coupling expected in those molecules. The next step would then be to turn to smaller molecules such as C₃₆, where the coupling, and hence the T_c, is expected to be even higher than in C₆₀.

The new results on C₆₀ will no doubt stimulate a renewed look at superconductivity in the fullerenes. Although an electron-phonon mechanism has been a leading candidate to explain electron pairing in these materials, the question has never been fully settled. What's the role, for example, of the very high-frequency intramolecular vibrations of the buckyballs, or of the electron-electron interactions? For superconductivity governed strictly by electron-phonon interactions, some theorists predicted in the 1970s that T_c would be limited to about 30 K. That assertion is clearly challenged by the 117-K result.