Exciton satellites in photoelectron spectra

We propose a new mechanism for satellite peaks in photoelectron spectra. Besides data from the literature new measurements of spectra of early first-row transition-metal compounds are discussed. The observed strong satellites of the metal core peaks of these materials are assigned to the creation of excitons on the anions. In our model, satellite energies can be determined from optical excitation spectra and satellite intensities from anion polarizabilities. It is shown that sharp satellites of this kind are expected if the exciton dispersion is not too large so that exciton states exist over large regions of the Brillouin zone. We also consider competition between charge transfer and exciton satellites. It turns out that only weak charge-transfer satellites are expected for the early $3d$-metal compounds, in agreement with experiment.