Tuning fulleride electronic structure and molecular ordering via variable layer index

Abstract

The ability to tune competing interactions in the fullerides arises from advances in our ability to grow well-controlled heterogeneous molecular films. Here we describe measurements on potassium doped C 60 (K x C 60 ) ultra-thin films having variable thickness from one to three layers (layer index i = 1, 2, and 3) for three specific doping concentrations (x = 3, 4, and 5). Fig. 1a displays a scanning tunneling microscope (STM) topograph of a representative K x C 60 multilayer on Au(111), where the color scale highlights the plateau structure. Narrow slivers of C 60 -free voids containing only K atoms 3 (brown) exist between continuous patches of K x C 60 . Islands of second (blue) and third layer (red) K x C 60 can be seen residing on top of the first K x C 60 layer (green). The average layer thickness is ~9.9 Å, greater than the 8 Å spacing found in undoped C 60 films 11 .We begin by describing our results for a multilayer of the x = 3 metallic system.Layer-dependent electronic structure in K 3 C 60 can be seen in Fig. 2a, which shows spatially-averaged dI/dV spectra measured at three different layer levels. Within each layer the spectrum is highly uniform with no sign of spatial inhomogeneity such as that found in the surface of bulk fullerides 12 . The first layer dI/dV displays a wide peak at the Fermi energy (E F ), reflecting the large electronic density of states (DOS) of a metallic LUMO-derived band (LUMO = Lowest Unoccupied Molecular Level). In contrast, the second layer spectrum shows a sharp dip at E F , indicating the emergence of an energy gap that tends to split the band into two halves. A similar gap-like feature persists in the third layer. The width of the gap-like feature (measured between adjacent local maxima) is ~ 0.2 eV, a much larger value than the superconducting gap 2∆ sc ~ 6 meV found in bulk K 3 C 60 13.The spatial arrangement of C 60 molecules also changes dramatically with layer index. The first layer of K 3 C 60 (Fig. 2c) exhibits a complex 3 3 × superstructure of bright molecules having different orientation from their dimmed nearest neighbors 8 . In the second layer ( Fig. 2d), however, C 60 molecules form a very simple hexagonal lattice (lattice constant a ~10.5 Å) with long-range orientational ordering. The tri-star-like topography of each molecule suggests that C 60 in the second layer is oriented with a hexagon pointing up 14 . The third layer topograph is the same as the second layer. 4The insulating x = 4 multilayer system displays a similar trend. Fig. 3a shows dI/dV spectra measured on a K 4 C 60 plateau structure where the number of layers is varied from i = 1 to 3. First layer spectra (i = 1) exhibit an insulating energy gap ∆ ~ 0.2 eV that is induced by molecular Jahn-Teller (JT) distortion 8 . As the layer index increases from i = 1 to 3, the energy gap opens continuously (by layer 3 the gap has well-defined edges and a flat bottom). The gap amplitudes observed here are estimated to be ∆ ~ 0.6 eV and 0.8 eV for layer 2 and 3 respectively. As seen in the metallic x = 3 ...

Surface Hubbard U of alkali fullerides

“…and shape of the features in the spectra of pristine C 60 are in agreement with published work, testifying the good alignment of the Fermi level in our spectra and that our film is thick enough so that screening from the metallic substrate [41] is negligible. The experimental HOMO-LUMO energy separation is 3.6 eV as previously observed [36,52,53].…”

“…The relatively low values of U found in open-shell compounds (RbC 60 , Cs 4 C 60 , K 3 C 60 ) thus entail even smaller correlation energies in the bulk of these systems. Motivated by these findings and by a recent study of potassium fulleride thin films [41], we have investigated to what extent molecular polarization screening can account for our observations. We assume the contribution of molecular polarization screening only, as the atomic polarizability of the alkali counter-ions is negligible in comparison to that of C 60 species.…”

“…However, a recent experimental study on ultra-thin films has shown that the actual value of the Hubbard U is a function of the stoichiometry of the compound, besides being dependent on film thickness [41]. The stoichiometry-dependence is not surprising, as the molecular polarizability varies according to the charge state of the C 60 anion [40] and the crystal structure and hence the coordination are different depending on the compound, and also because screening may be enhanced in open-shell fullerene compounds by multiplet effects and, in metallic and superconducting phases, by itinerant-carrier-like screening [41].…”

“…The stoichiometry-dependence is not surprising, as the molecular polarizability varies according to the charge state of the C 60 anion [40] and the crystal structure and hence the coordination are different depending on the compound, and also because screening may be enhanced in open-shell fullerene compounds by multiplet effects and, in metallic and superconducting phases, by itinerant-carrier-like screening [41]. These possibilities have indeed been taken into consideration in some theoretical work [30,40,42,43], but only very recently explored in experiments.…”

Jahn–Teller Effects in Molecules on Surfaces with Specific Application to C60

“…Of particular merit is the work of Crommie and co-workers who, in a series of papers [30,5,38,39], have recorded some very intriguing images and spectra of a series of doped C 60 molecules. In one experiment [on Ag(111)], these workers were able to use the STM tip to progressively attach/detach potassium atoms to individual C 60 molecules and subsequently record the scanning tunnelling spectroscopy data shown in Fig.…”