Resolving discrepancies between LEED and STM throughab initiocalculations: Surface and bonding of sulfur on Mo(110)

Abstract

The adsorption of sulfur at 0.5 ML in both $c(2\ifmmode\times\else\texttimes\fi{}2)$ and ${[}_{1}^{2}{}_{1}^{2\ifmmode\bar\else\textasciimacron\fi{}}]$ configurations on the Mo(110) surface is studied using the density-functional, pseudopotential method with a plane-wave basis and a seven-layer slab geometry in conjunction with scanning tunneling microscopy (STM) and low-energy electron diffraction (LEED) experiments. The sulfur adatoms are placed in different possible binding sites in order to determine the most favorable adsorption site. The ${[}_{1}^{2}{}_{1}^{2\ifmmode\bar\else\textasciimacron\fi{}}]$ overlayer is more stable than the $c(2\ifmmode\times\else\texttimes\fi{}2)$ by 0.31 eV, in agreement with experiment. The greater stability of the ${[}_{1}^{2}{}_{1}^{2\ifmmode\bar\else\textasciimacron\fi{}}]$ structure is attributed to differences in metal-metal bonding. Sulfur is predicted to adsorb at a low-symmetry position near the long-bridge site; the long-bridge site is slightly less favorable in energy. Simulated STM images of the sulfur-covered surface are constructed, and found to model well the experimental images. We find that the bright areas in the calculated STM images do not necessarily correspond to the position of the sulfur atoms, which explains the difference between the LEED pattern and the experimentally observed STM images.