Reinvestigation of Co 2p Satellite Peak on the Co Ultrathin Film: Screening Channel at Interface

Spectral features of the core-level photoemission from the thin films of magnetic transition metals have attracted much attention for understanding the electronic structures of both the films themselves and the interfaces to the substrates. In these studies, the binding energies and the intensities of satellite peaks relative to those of the main peaks have important information. Recently, for example, the electron configuration and correlation were successfully elucidated by X-ray photoelectron spectroscopy (XPS) and model calculations on the Mn films on Ag(100). On the other hand for another well-known magnetic thin film, Co on Cu(100) substrate, the origin of a satellite signal in XPS has not yet been settled. Previously, Chen predicted strong satellite signals at 5:5 and 10 eV below the Fermi level in valence photoemission spectra of the two mono-atomic layers (ML) bcc Co thin film without a substrate. He included all the many-body effects due to photoemission process. Contrary to this, no satellite was observed in spin-resolved photoemission spectroscopy for 1 ML fcc Co on the Cu(100) surface. Then, Chen calculated the spectra for the fcc Co thin films on a Cu substrate including the hybridization with Cu 4s. The result indicated that the Cu 4s–Co 3d hybridization makes the satellite signal too broad to be detected. On the other hand, Nath et al. recently reported a satellite in XPS of the fcc Co thin films on the same substrate. Its separation is 3 eV from the Co 2p3=2 main peak. They attributed it to the correlation enhanced by the low dimensionality as suggested by Chen. Actually the intensity of the satellite continuously decreased with increasing the film thickness, which was interpreted as the fading out of low dimensionality in this model. Thus they ascribed the origin of the satellite to the film itself rather than the interface. In the present study, we reinvestigated the origin of the XPS satellite in Co/Cu(100) more carefully to make clear the role of the interface. For this purpose, we made two-fold XPS measurements on the Cu(100) surface: First, photoemission angle dependence of the satellite, and second XPS at low Co coverages less than 1ML. The results indicate that the satellite originates mainly from the interface. A Cu(100) single crystal was cleaned by repeating Arþ ion sputtering and 600 C annealing cycles. Cobalt was deposited on the surface from the surface normal at room temperature (RT) by electron bombardment onto a Co rod (99.999% purity). The deposition rate was 0:33ML/min. The conventional MgK from a twin anode (TA10, VSW Ltd.) was used as an X-ray source. Photoelectrons were detected by a spherical analyzer (SES100, Gamma Data). In most of the measurements, the X-ray incidence angle was 15 and the photoelectron detection angle was 60 from surface normal. The sample was rotated for the measurements in the detection angle of 30 . Figure 1 shows the XPS spectra taken on the Co films grown on Cu(100) with changing the thickness from 0.5 to 9ML. Between the two peaks of Co 2p1=2 and Co 2p3=2, we can see a satellite separated from the main peak (2p3=2) with 3 eV when the film thickness is small. Broad peaks at 805 eV are attributed to the signal from the substrate. As shown in inset (a), the intensity of the satellite decreases gradually with increasing the film thickness and disappears in the 9ML film. These results are consistent with the previous observations. Additionally we noticed the Co 2p3=2 main peak continuously shifts toward the higher binding energy with increasing the film thickness. This implies that the electron occupation of Co 3d in the thin Co films is different from that of bulk Co. In other words, there is an electron transfer from the Cu 4s to the Co 3d at the interface. Nath et al. attributed the obsrved gradual decrease of the satellite signal to the fading out of the low dimensionality of the film, i.e., from 2D to 3D structure. There is, however, another possibility that the satellite signal is made at the Co/ Cu interface, and signal intensity is decreased by inelastic scattering of photoelectron by the Co over layers. In order to ascertain which is true, we made XPS measurements on the 3ML Co film with two photoemission detection angles e 1⁄4 30 and 60 (see Fig. 2). The configuration of e 1⁄4 30 is more interface sensitive than that of e 1⁄4 60 . If the decrease of the satellite intensity is only due to the loss of the low dimensionality, the relative intensity of the satellite to the main peaks should not depend on the photoemission angle. As clearly seen in the inset of Fig. 2, the relative intensity of the satellite peak at e 1⁄4 30 is larger than that at e 1⁄4 60 . This indicates that the satellite is produced mainly at the Co/ Cu interface. -800 -790 -780 -770 -760 -779 -778 -777 0.5 ML 1 ML 3 ML 9 ML