Core-level subsurface shifted component in a 4 d transition metal: Ru„ 101 ¯ 0 … Alessandro Baraldi and Silvano Lizzit Sincrotrone Trieste S.C.p.A., S.S. 14 Km 163.5, 34012 Trieste, Italy Giovanni Comelli Sincrotrone Trieste and Dipartimento di Fisica, Universita ´ di Trieste, 34127 Trieste, Italy and Laboratorio T.A.S.C.-I.N.F.M., Padriciano 99, 34012 Trieste, Italy Andrea Goldoni Sincrotrone Trieste S.C.p.A., S.S. 14 Km 163.5, 34012 Trieste, Italy Philip Hofmann Institute for Storage Ring Facilities, University of Aarhus, Aarhus, Denmark Giorgio Paolucci Sincrotrone Trieste S.C.p.A., S.S. 14 Km 163.5, 34012 Trieste, Italy Received 19 August 1999 Three components have been found in the Ru(101 ¯ 0) 3 d 5/2 core-level spectra. Based on a photoelectron diffraction experiment they have been assigned to emission from the bulk and the first two surface layers. The first layer line is shifted by -480 meV from the bulk peak position and the second layer line by -240 meV. The relative magnitude of the observed shifts is in good agreement with a simple tight-binding model. The same model would, however, predict clearly observable shifts for other open 4 d transition-metal surfaces, where only one shifted component has been reported. Possible reasons for the seemingly anomalous behavior of Ru are discussed. During the past twenty years, the study of surface core- level binding energy shifts SCLS has provided valuable information about the chemical and physical properties of surfaces. 1 The basic idea is that creating a surface changes the environment of the atoms nearby. This is reflected in slightly different binding energies of the core electrons. A quantitative description of the shift turned out to be difficult since it is a complicated interplay between initial state and final state effects. Recently, the interest in such shifts has been renewed because of the possibility to perform reliable ab initio calculations for many systems. 2–4 Such calculations permit the deconvolution of the shifts into initial and final- state contributions such that it is now possible to explain many experimental observations in detail and to obtain im- portant information about surface bonding. On the experi- mental side substantial progress has also been made. The resolution has been improved such that line-shape param- eters can be determined more reliably and, in some cases, new structures can be found in the core-level spectra. On most metals, only the first layer of atoms gives rise to an observable surface core-level shift. This is due to the na- ture of metallic bonding: in the second layer the atoms are highly coordinated and embedded in a charge density similar to the bulk. There are, however, some examples of observ- able shifts from deeper layers. In the 5 d transition metals, second layer shifts have been found in the very narrow 4 f core lines 1,5,6 and in the 1 s core-level spectra of Be0001 and Be(101 ¯ 0) up to four surface related components have been reported and assigned to emission from the first four layers. 7–10 In this paper we report the finding of three peaks in the 3 d 5/2 core-level spectra taken from the Ru(101 ¯ 0). They can be assigned to emission from the bulk and the first two lay- ers. This finding is rather unexpected since no clearly ob- servable second layer core-level shifts have been reported for the other 4 d transition-metal surfaces. It is also of potential importance for further research on the catalytic properties of 4 d transition metals: the second layer core-level shift permits a detailed x-ray photoemission spectra XPS investigation of the role the subsurface atoms play in gas-surface interac- tions and Ru(101 ¯ 0) could be used as a model system. The Ru 3 d 5/2 core-level data has been acquired at the SuperESCA beamline of Elettra. 11 The sample was cleaned by repeated cycles of Ar + sputtering, annealing to 1500 K, oxygen treatment at 900–1100 K and oxygen reduction in hydrogen ( p =1 10 -6 mbar, T =800 K). This procedure resulted in a sharp 11 low-energy electron diffraction LEED pattern and in C, O, and S 1 s XPS intensities below our detection limit of 0.005 ML. The upper part of Fig. 1 shows the Ru 3 d 5/2 spectra mea- sured at different photon energies and at a sample tempera- ture of 120 K. The emission direction was 40° off the surface-normal, the photons were coming in at 80° off nor- mal. The overall experimental resolution was 65 meV. Three distinct structures, named S b , S 1 , and S 2 , are clearly re- solved in the spectra, with relative intensities strongly depen- dent on photoelectron kinetic energy. Saturating the surface with oxygen or carbon monoxide left only the position of the S b peak unchanged, and the latter was therefore assigned to emission from the bulk. The data was analyzed quantitatively by a least-square fit, where the line shape of each peak was described by a convolution of a Doniach-Sunjic function Lorentzian width  , asymmetry  ) and a Gaussian. PHYSICAL REVIEW B 15 FEBRUARY 2000-I VOLUME 61, NUMBER 7 PRB 61 0163-1829/2000/617/45344/$15.00 4534 ©2000 The American Physical Society