Using the time-dependent local density functional (TDLDF) theory and analytically continuing the resultant response function into the lower half frequency plane, the authors calculate the dispersion relations and Landau damping of the lightly damped surface plasmon modes of a bare jellium aluminium surface, free of foreign overlayers. For the correctly self-consistent zeroth-order surface electron density profile, they find at least one such mode in addition to the familiar 'monopole' surface plasmon.
We present the first accurate values of , the static susceptibility of a jellium surface, including (i) finite surface wave vector ; (ii) Friedel oscillations (from an exact kinetic energy functional); (iii) Kohn-Sham exchange-correlation effects (Wigner form); (iv) self-consistent screening; (v) realistic Lang-Kohn zero-order profile; and (vi) comparison, for , with a useful approximation.
30 July 1986 in Journal of Physics C: Solid State Physics
The electronic susceptibility, chi ( omega , q/sub ///, z, z'), of the jellium half-space is evaluated microscopically for finite frequency omega and surface-parallel wavenumber q/sub ///. Screening is performed both with and without a self-consistent exchange and correlation potential mu xc whose form is derived from static local density functional theory. The resulting chi -functions can be termed local density functional (LDF) and random-phase approximation (RPA) susceptibilities respectively. A q/sub /// integration yields the damping time of an adsorbed oscillating point dipole, which, for separations up to a few angstroms from an aluminium surface, is almost independent of frequency up to around two-thirds of the surface plasmon frequency. The LDF and RPA decay times differ markedly, especially for metals such as sodium with higher rs values, as previously predicted by Liebsch (1985) on the basis of a low-frequency expansion. The RPA lifetime for 'point-dipole' N2 physisorbed on aluminium is in agreement with that deduced from calculations of Eguiluz (1983, 1984). Both RPA and LDF lifetimes are, however, substantially longer than that obtained by extrapolation of previous results valid for large dipole-surface separations (i.e. for small q/sub ///). This in turn means that electron-hole damping of a point dipole is not after all sufficient by itself to explain the lifetime measurements of Avouris, Schmeisser and Demuth (1983). It is nevertheless a substantial contributor to the damping.