Figure 1

Different STS measurements [5, 6] on ${\mathrm{C}}_{60}$ molecules docked onto Si(100) $2\times 1$ surface (left panel). We attribute different bonding geometries (middle panel) for each experiment, leading to a theoretical $G\mathrm{\text{−}}V$ (right panel) that agrees quite well with the corresponding measurement. The arrow represents the dimer direction of the reconstructed surface (cross going into the page). The upper geometry corresponds to ${\mathrm{C}}_{60}$ physisorbed on four surface dimers, the middle one represents the buckyball chemically bonded with a single surface dimer, while the bottom one has the molecule lowered into the trough caused by a missing dimer.Reuse & Permissions

Figure 2

Calculated surface band structure of asymmetric dimer (solid lines) and symmetric dimer (dashed lines) reconstructions along Si(100)-$2\times 1$ using EHT bulk Si parameters [14] and optimized surface geometries [11].Reuse & Permissions

Figure 3

Energy level alignment between ${\mathrm{C}}_{60}$ and silicon. The left panel shows the energy levels of isolated ${\mathrm{C}}_{60}$, with the calculated Fermi energy of the doped silicon shown as a dashed line. The ${\mathrm{C}}_{60}$ energy levels are shifted by 1.3 eV due to self-consistent charging driven by the work function difference between the molecule and the substrate, and the corresponding charge transfer from Si to ${\mathrm{C}}_{60}$. Chemisorption creates new levels due to SiC bonds, easily seen by adding a silicon dimer to the molecule and passivating the cluster at the bottom. Finally, including the silicon self-energy amounts to adding the entire silicon substrate and generates a continuum molecular band. The right panel shows the density of states corresponding to the geometry in the middle of Fig. 1.Reuse & Permissions

Figure 4

Introducing tunnel barriers between the STM and the sample or the sample and the substrate (far left) deemphasizes the HOMO levels relative to the LUMO ones due to their relatively larger barrier heights, accounting for measurements (left) with little or no signature of HOMO levels [7, 8]. Although the unnormalized conductance shows this effect with increasing tip-sample separation $d$ (right), normalizing the conductance (center) as in Fig. 1 restores these peaks.Reuse & Permissions