Annihilation of positrons trapped at the alkali-metal-covered transition-metal surface.

Abstract

Recent studies of the Cu(100) surface covered with submonolayers of Cs [A. R. Koymen et al., Phys. Rev. Lett. 68, 2378 (1992)] revealed that the normalized intensity of the positron-annihilation-induced Cu ${\mathit{M}}_{2,3}$VV Auger signal remains nearly constant at the clean Cu(100) surface value until the Cs coverage reaches approximately 0.7 physical monolayer, at which coverage the signal intensity drops precipitously. We present a microscopic analysis of this unusual behavior of the Cu ${\mathit{M}}_{2,3}$VV Auger signal based on a treatment of a positron as a single charged particle trapped in a ``correlation well'' in the proximity of the surface atoms. The image-potential-induced positron surface states are calculated using the corrugated-mirror model in a full three-dimensional geometry. These states are studied for the clean Cu(100) surface and for various ordered structures of the Cs adsorbate on the Cu(100) surface below and above the critical alkali-metal coverage of approximately 0.7 physical monolayer. Calculations show that whereas the positron surface state is localized in the region of the Cs/Cu(100) interface for Cs coverages below the critical alkali-metal coverage, at a Cs coverage corresponding to one physical monolayer the positron surface state is localized on the vacuum side of the Cs overlayer. The probabilities for a positron trapped in a surface state to annihilate with relevant Cu and Cs core-level electrons as well as the positron surface-state lifetimes are computed for various alkali-metal structures on the Cu(100) surface and compared with experimental positron-annihilation-induced Auger-electron-spectroscopy data. It is shown that a shift in localization of the positron surface state from the Cs/Cu(100) interface to the vacuum side of the alkali-metal overlayer results in a sharp decrease in the positron-annihilation probabilities with Cu 3s and 3p core-level electrons, in agreement with experiment.