Photoelectron spectroscopy, including emission from the Si 2p core level, was used to investigate the orientation dependence of the photoionization threshold ξ(α) and the surface position of the valence-band edge below the Fermi level, $E_{\mathrm{VFS}}$(α), on a cylindrically shaped Si crystal with [11¯0] as its axis. The average ξ values for (001), (113), (111), and (110) are 5.33, 5.32, 5.26, and 5.26 eV, respectively, and thus differ only very little as expected for a covalent, nonpolar crystal. Increasing corrugation on stepped, faceted, defect-rich, and high-index surfaces increases ξ values, in contrast to what is observed on metal surfaces. The orientation dependence of ξ is interpreted in terms of a relaxation- or reconstruction-related surface polarity which is increased at edges. Water adsorption generally reduces ξ. In the range (001)–(113) the shape of ξ(α) is conserved, whereas it is changed in the range (112)–(111)–(110), consistent with fundamentally different adsorption mechanisms as observed earlier. After oxygen adsorption, ξ(α) is strongly changed, indicating a strong distortion of the surface by bond breaking. On the clean surface, the Fermi-level–pinning position $E_{\mathrm{VFS}}$(α), which is given by the surface-state structure, varies strongly. After oxygen exposure it is pinned at a nearly constant position ∼0.1 eV above the midgap, probably by defect-related surface states. For water adsorption, again two ranges can be distinguished with a similar pinning as for oxygen in the range (112)–(111)–(110) and a reduced but still strong variation between (001) and (113).

• Received 9 August 1984

DOI:https://doi.org/10.1103/PhysRevB.31.2246