Energy-Filtered Scanning Tunneling Microscopy Using a Semiconductor Tip

The use of cleaved, [111]-oriented monocrystalline InAs probe tips enables state-specific imaging in constant-current filled-state scanning tunneling microscopy. On Si(111)-(7x7), the adatom or rest-atom dangling bond states can thus be mapped selectively at different tip-sample bias. This state-selective imaging is made possible by energy gaps in the projected bulk band structure of the semiconductor probe. The lack of extended bulk states in these gaps gives rise to efficient energy filtering of the tunneling current, to which only sample states not aligned with a gap contribute significantly.

Conventional filled-state cc-STM is most sensitive to occupied states near the sample Fermi level. On Si(111)-(7x7) it maps only the dangling bonds of the adatoms over a wide range of tip-sample bias, and is insensitive to the rest-atom state. Filled-state images obtained with our cleaved InAs probes show a much richer bias dependence (Fig. 1). At low voltage (-1.0 V sample bias; Fig. 1 (B)), sharp maxima in apparent height are localized at adatom positions, producing an image that closely matches those obtained with metal probes. With increasing negative voltage, however, these sharp maxima spread out to become shallower and elongated. At -1.7 V, the contrast maxima originally associated with the corner adatoms uniformly cover pairs of adjacent corner adatoms and rest atoms. At -2.0 V (Fig. 1 (C)), the adatoms no longer appear to be protruding highest, but sharp and well-defined peaks are now located at the six rest-atom sites.

Our observations suggest that a semiconductor such as InAs, used as a probe tip in STM, can serve as an energy filter with a narrow 'pass band' between the fundamental gap and a projected conduction gap. STM using such a probe offers exciting possibilities for spectroscopic imaging. If a sample has different dangling bonds with spatially overlapping orbitals, energy-filtered (EF) STM can selectively map each of these orbitals individually. In the more common case of spatially separated dangling bonds, EF-STM will provide contrast between occupied states at different energies. If, for instance, different atomic species at the surface give rise to dangling bonds at different energies, EF-STM will be able to distinguish between these species. We thus foresee the ability to determine the local surface composition of semiconductor alloys, a long-sought capability critical for understanding alloy growth.