Interpretation of satellite structure in the x-ray photoelectron spectra of CO adsorbed on Cu(100)

By employing the Xα scattered-wave method with a Cu₉CO cluster to model the chemisorption of CO on a onefold site of a Cu(100) surface, a simple interpretation of the satellite structure observed in the x-ray photoelectron spectrum in the C $1s$ and O $1s$ regions has been obtained. The physical model obtained by analyzing the results of the Cu₉CO cluster calculations is qualitatively the same as that obtained in a previous study of a Cu₅CO cluster with the CO in a fourfold site [Solid State Commun. 36, 265 (1980)]. The qualitative differences suggest that the present Cu₉CO cluster is the better model, however. Experimentally, a three-peak structure is observed in both the O $1s$ and C $1s$ hole spectra. The "first" peak, at lowest binding energy, is followed by a second peak at 2-3 eV higher binding energy and the third peak is at 7-8 eV higher binding energy with respect to the first peak. The theoretical model derived here suggests that the unoccupied 2π level of isolated CO is split into two levels 2 ̃ \~πₐ and 2 ̃ \~πb on interaction with the Cu metal. In the neutral ground state neither of these levels is occupied. On the introduction of a core hole in the chemisorbed CO (e.g., the C $1s$ hole) the 2 ̃ \~πb and 2 ̃ \~πₐ orbitals change their character quite significantly to become 2 ̃ \~πb^' and 2 ̃ \~πₐ^'. The former is now partially occupied and closely resembles the isolated 2π orbital of CO, and the latter is unoccupied with significant metal character and less CO content. The character of the 1π level of isolated CO is basically the same for the chemisorbed ground state (where it is labeled 1 ̃ \~π). However, it changes rather dramatically (labeled 1 ̃ \~π^') after the removal of the core electron, as it shifts to screen the core hole. A description of the final states which give rise to the three peaks observed in the experimental spectrum can be given in terms of the occupancies of the three orbitals 1 ̃ \~π^', 2 ̃ \~πb^', and 2 ̃ \~πₐ^'; there is of course a $1s$ hole in each of the final states. The assignment of the final-state configuration corresponding to the three observed peaks (in order of increasing binding energy) is as follows: (1) ${(1{{ ̃ \~{}{}}{{π}}}^{{'}})}⁴{(2{{ ̃ \~{}{}}{{π}}}b^{{'}})}¹{(2{{ ̃ \~{}{}}{{π}}}ₐ^{{'}})}⁰$, (2) ${(1{{ ̃ \~{}{}}{{π}}}^{{'}})}⁴{(2{{ ̃ \~{}{}}{{π}}}b^{{'}})}⁰{(2{{ ̃ \~{}{}}{{π}}}ₐ^{{'}})}¹$, and (3) ${(1{{ ̃ \~{}{}}{{π}}}^{{'}})}³{(2{{ ̃ \~{}{}}{{π}}}b^{{'}})}²{(2{{ ̃ \~{}{}}{{π}}}ₐ^{{'}})}⁰$. The last final state corresponds to the final-state configuration found in the isolated CO molecule due to a $1{{π}}^{{'}}{→}2{{π}}^{{'}}$ shake up.