3.1 Experimental details

All STM experiments were performed using a commercial variable-temperature STM (78–300 K) from Createc GmbH in an in an ultra-high-vacuum (UHV) system consisting of an analysis chamber (with a base pressure of 2 × 10-11 mbar and a preparation chamber (5 × 10-11 mbar). The STM tips were fabricated from [001]-oriented 0.3 × 0.3 × 10 mm3 single-crystalline tungsten bars by electrochemical etching in 2 M NaOH and cleaned in the UHV chamber by Ar+ ion bombardment [73, 74]. The voltage V sample corresponds to the sample bias with respect to the tip. No drift corrections have been applied to any of the STM images presented in this paper, and all STM images were recorded in the constant-current mode. The experiments were performed in the temperature range of 78–300 K, while the switching phenomenon, over the time scale of the experiment, was observed in the temperature range of 220–260 K. The acquisition time of 10 ms per point was used to measure the time-evolution of the distance between the STM tip and the surface. All STM images were analysed using the software package WSxM [56].

A W(110) single crystal, prepared at the Institute of Solid State Physics, Russian Academy of Sciences, was used as the substrate. An atomically-clean W(110) surface was prepared by in situ annealing at 1900 K in an oxygen atmosphere of 1 × 10-7 mbar, followed by a series of high temperature flashes at 2200 K. The sample was heated by electron beam bombardment and temperatures were measured using an optical pyrometer (Ircon UX20P, emissivity 0.35). The clean W(110) surface was verified by LEED and STM before oxidation. Once a clean surface was obtained, the sample was oxidised at 1600 K in an oxygen atmosphere of 1 × 10-6 mbar for 60 minutes. The quality of the resulting oxide structure was verified by LEED and STM before the deposition of C60 molecules.

C60 (Aldrich Chemicals) was evaporated in the preparation chamber isolated from the STM chamber at a rate of about 0.2 ML (monolayer) per minute from a deposition cell operated at a temperature of approximately 700 K. Before evaporation, the C60 powder was degassed for about 8 hours to remove water vapour. The total pressure during C60 deposition was in the 1 × 10-9 mbar range and the substrate was kept at room temperature [46].

The WO2/W(110) surface was covered with monatomic steps and terraces up to 50 nm in width. Oxide rows and monatomic steps were used to calibrate the STM scanner in the variable temperature experiments. The height from the substrate surface to the molecules’ centers was 5 Ċ, as measured by STM of the fullerene film thickness at the edges of the closed C60 layer.

In order to determine which part of the static and spinning C60 molecules face the WO2/W(110) surface, density functional theory (DFT) simulations of the partial charge distribution of electron states were performed in collaboration with Dr Olaf Lübben, whose work and expertise are gratefully appreciated and acknowledged. Density of states (DOS) calculations were performed using the Vienna Ab initio Simulation Package (VASP) program. VASP implements a projected augmented basis set (PAW) [70] and periodic boundary conditions. The electron exchange and correlation was simulated by local density approximation (DFT-LDA) pseudopotentials with a Ceperley-Alder exchange-correlation density functional [75]. A Γ-centred (2 × 2 × 1) k-point grid was used for all calculations to sample the Brillouin zone. The applied energy cut-off was 400 eV. The global break condition for the electronic self-consistent loops was set to a total energy change of less than 1 × 10-4 eV. For the DOS a smearing of 0.2 eV was applied using the Methfessel-Paxton method [72].