Multiplet structure in Pu-based compounds: A photoemission case study of PuSi x „0.5 Ï x Ï 2… films T. Gouder,* R. Eloirdi, J. Rebizant, P. Boulet, and F. Huber European Commission, Joint Research Centre, Institute for Transuranium Elements, Postfach 2340, D-76125 Karlsruhe, Germany sReceived 13 October 2004; published 1 April 2005d Thin films of PuSi x s0.5 l x l 2d were prepared by sputter deposition and studied by x-ray and ultraviolet photoelectron spectroscopy. Photoemission of the Pu 4 f core levels and the Pu 5 f valence band states showed first a breakdown of 5 f band behavior at low Si content, but then an increasing tendency for 5 f hybridization for more diluted PuSi x systems. In valence band spectra, a broad peak at high binding energy is observed for concentrated systems sPuSid. It is replaced by a sharp three-peak structure close to the Fermi level for more diluted systems sPuSi 2 d. Both structures are discussed in terms of final-state multiplets of localized 5 f states with different degrees of hybridization. It is concluded that multiplets and final-state effects occur quite generally in Pu systems and explain many of the spectral features, which often have been ascribed to the ground-state density of state. DOI: 10.1103/PhysRevB.71.165101 PACS numberssd: 71.20.Gj, 71.28.1d I. INTRODUCTION The physics of plutonium and its compounds is dominated by the competition between 5 f itinerancy associated with chemical bonding and 5 f on-site localization. Pu has a par- ticular position within the actinide series. It is the last ele- ment where the 5 f electrons still participate in bonding, while in the next element, Am, they retract from bonding and become localized. The delicate balance between itinerancy and localization is quite sensitive to slight variations in the chemical environment. The existence of six allotropic phases sa to dd in a narrow temperature range, 1 the substantial drop in density sby 20%d between the room temperature a- and the high temperature d-Pu, and the localization of the 5 f states when Pu is confined to thin films 2,4 are direct conse- quences of the weak bonding properties of the 5 f states. Over the last 20 years, the electronic structure of Pu metal sa and dd has been widely investigated by photoemission spectroscopy. 2,4–8,10 Valence band data have been compared to theoretical calculations, and good correspondence between them and the calculated ground-state density of states sDOSd has been taken as an indication for the validity of the theo- retical model fdynamical mean field theory, 11 constrained lo- cal density approximation sLDAd, or a “mixed level model” with four of the five 5 f states forced into localization, 8,12,13 and LDA+ U sRef. 14dg. In this approach, it was generally implied that photoemission itself represents one well-defined final state, which then can be related to the ground state. Competing final states swith good or poor screeningd would not occur in the valence band spectra. However, for the 4 f core-level spectra of Pu, these competing final states exist, and are directly related to the narrow-band nature of Pu. 6 Therefore, similar final effects should occur in the valence band spectra and superimpose on the ground-state DOS fea- tures. These effects should become more important when the 5 f states approach localization. In this paper we want to address this issue and present a photoemission study of the Pu-Si series, where the 5 f itinerancy is modified by diluting Pu with a ligand element. We will argue that the spectra show band features together with localized 5 f states, appear- ing as final-state multiplets. Coexistence of these does not reflect a mixed level ground state, but is due to final-state screening effects as for the 4 f core levels. The discussion will be extended to other Pu systems. The electronic structure of Pu in the silicide series is de- termined by two opposing effects. On the one hand, dilution of Pu in a Si matrix favors 5 f localization, because the direct 5 f -5 f overlap of neighboring Pu atoms is suppressed. Above a critical distance sthe Hill limit, 0.34 nm for Pud the overlap is too weak to enable 5 f band formation. On the other hand, the bonding interaction between Pu and the Si ligand atoms favors hybridization between the Pu 5 f and the Si 3p states, and thus could itself result in 5 f delocalization. Replacement of the pure f conduction band by an f ligand hybrid band is at the base of heavy-fermion materials in the actinides and lanthanides, 15 where the f -atom separation is significantly greater than the Hill limit. It was not clear, a priori, which of the two effects would dominate and whether or not the 5 f states would become localized at high dilution. In this con- text, comparison with the Ce-Si system is instructive. As in Pu, f states in Ce are close to the localization threshold and there are strong similarities between the physics of these two elements. 16 CeSi x s1.6 l x l 2d has been studied by photo- emission and magnetic measurements. 17 The more concen- trated CeSi x was found to be closer to localization than the dilute, suggesting that ligand hybridization effects prevail over dilution effects. High-resolution photoemission data show a sharp emission right at the Fermi level, a peak at 0.3 eV binding energy sBEd, and a large peak at 2.5 eV BE. These data are interpreted either in the framework of the single-impurity model, 17 or in terms of different final-state screening channels. 18 Both models attribute the peak at 2.5 eV to the localized response sthe 4 f 1 → 4 f 0 transitiond, and the structures close to the Fermi level to some type of hy- bridized response, be it the Kondo resonance with its spin- orbit sSOd sideband, or the ordinary SO split 5 f peak, 19,20 which corresponds to the f 1 final-state configuration. Below we will show that a very similar interpretation can be applied to the Pu-Si system. PHYSICAL REVIEW B 71, 165101 s2005d 1098-0121/2005/71s16d/165101s7d/$23.00 ©2005 The American Physical Society 165101-1