THE CONNECTION BETWEEN MULTISTATE RESONANT CHARGE-TRANSFER DYNAMICS AND MANY-ELECTRON STATES IN ATOM-METAL SURFACE SCATTERING

For Li+ and Na+ scattering from clean and cesium-covered Cu(001) surfa ces, we have measured probabilities to form different final electronic states in the scattered flux as a function of the Cs-induced work-fun ction shift Delta phi. Specifically, positive and negative ion yields for 400 eV Li+ scattering, and the relative yields of excited neutral Li(2p) and Na(3p) for 400 and 100 eV Li+ and 1320 eV Na+ scattering we re measured. As we lowered the work function from its clean (metal) su rface value, we observed a monotonic decrease in the Li+ yield, a mono tonic increase in the Li- yield, and a peak in the minority Li(2p) cha nnel yield (and the Na(3p) yield). The major trends in the Li/Cu(001) data (and likewise in the Na/Cu(001) data) can be reproduced by use of a multi-state model, developed by Marston and co-workers, of resonant charge transfer. Here we present a new, straightforward explanation o f these trends, based upon an examination of the many-electron states of the atom-metal system. Much of the charge transfer dynamics can be understood through the ground state of the interacting Li/Cu(001) syst em, since the 'atom probabilities' - the probabilities that the Li is a Li+, Li(2s), Li(2p), or Li- - tend to equilibrate towards their grou ndstate values throughout the atom's trajectory. In each of our experi ments, the velocity is low enough that the Li/Cu(001) system electroni cally equilibrates close to the surface, where the atom-metal coupling s are large; at small atom-metal separations z, the Li atom probabilit ies therefore track their ground-state values. As the atom moves along its outgoing trajectory, the couplings decrease exponentially, and ev entually the atom probabilities lose track of their ground-state value s. Qualitative arguments, based upon general principles of quantum mec hanics, allow us to understand the dependence of the ground-state prob abilities on the work function phi and z. To comprehend the origin of the Li(2p) peak we must consider both the ground-state probabilities a nd the conditions under which the dynamical probabilities lose track o f their ground-state values. (C) 1998 Elsevier Science B.V.