Spin-fluctuation description of chemisorption

It is argued that the picture of an Anderson-Edwards-Newns adatom-substrate electronic resonance, coupled to the collective fluctuations of the local spin density, provides a qualitatively correct, physically sensible, and calculationally practical description of electronic states in the vicinity of a chemisorbed atom. At the simplest level, we observe that it is exactly correct to second order in the Coulomb interaction parameter $U$, and that it reproduces the proper two-level structure in the adatom density of states in the opposite limit of weak adatom-substrate electron transfer energy $V$. We investigate the question of how to parametrize a local spinfluctuation (LSF) model of the dynamics of a system described by an Anderson Hamiltonian, with particular attention focused on the strong-coupling (Kondo) limit. We find that to the extent that bonding or antiboding states or resonances appear far from the Fermi level, a larger $\frac{U}{V}$ is necessary to reach the Kondo limit. Most cases of chemisorption on transition metals are therefore in weak- or intermediate-coupling regimes. To illustrate the scheme, we have calculated the effect of spin fluctuations on the adatom density of states for several model substrate surface densities of states. One sees arising naturally a doubling of structure in the spectrum and possible resonances near the Fermi level. Finally, we generalize the LSF theory to the important case where electrons interact in the substrate as well as on the adatom, and calculate the resulting substrate and adatom densities of states.