home

100 %

1 / 1
PDF

100 %

prev
1 / 1
next
PDF

whereas a relatively large orbital component remains in CrPt 3 . The spin component always shows a positive moment in spite of the small amplitude on the Cr-rich side, as a result, the total moment increases linearly with the Mn content. In (Cr 0.5 Mn 0.5 )Pt 3 , the total moment is, however, almost cancelled out due to the opposite contribution between the orbital and spin characters, so that this sample has a relatively low Curie temperature. The Pt L edge XMCD provides the evidence that the Pt 5 d magnetic states sensitively depend on the neighboring TM atom and local environment. The magnetic properties of the (Cr 1-X Mn X )Pt 3 system can be explained by taking the average over the volume fractions of ferromagnetic MnPt 3 and ferrimagnetic Hiroshi Maruyama Okayama University E-mail: maruyama@mag.okayama-u.ac.jp References [1] H. Maruyama et al., J. Magn. Magn. Mater. 43 (1995) 140. [2] M. Suzuki et al., Jpn. J. Appl. Phys. 37 (1998) L1488. [3] P. Carra et al., Phys. Rev. Lett. 70 (1993) 694. HIGH RESOLUTION Ce 3 d-4 f RESONANT PHOTOEMISSION STUDY OF CeNiSn and CePdSn Many Ce compounds have attracted interest widely due to their strongly correlated Ce 4 f states hybridized with valence band states. High resolution photoemission spectroscopy is a useful technique to experimentally reveal the electronic states. It is known, however, that rare-earth 4 f electronic states on the surface are remarkably different from those in bulk. Measurements at low h ν (< 200 eV) cannot faithfully probe the bulk Ce 4 f states due to their surface-sensitivity. In order to study such bulk electronic states using high energy excitations ( h ν > 500 eV), a high resolution photoemission station has been constructed at the twin-helical undulator beamline BL25SU [1]. Here, we report high resolution Ce 3 d resonant photoemission spectra of CeNiSn and CePdSn measured at this station. CeNiSn is a so-called "Kondo semiconductor" where the Kondo temperature ( T K ) is of the order of ~40 K and the Ce 4 f electrons are valence-fluctuating [2]. For CePdSn, on the other hand, the Ce 4 f electrons are more localized and T K is thought to be lower than 7 K [3]. A Ce 3 d resonant photoemission study was performed with the total energy resolution of ~0.22 eV, as shown in Figure 1 (b) . The sample temperature was about 20 K. Figure 1 (a) shows Ce 3 d resonant valence band spectra of CeNiSn and CePdSn together with the Ce 4d resonance spectra. The Ce 4 f contributions are obtained by subtracting the Ce 3 d off resonance ( h ν ≤ 875 eV) spectra from the on resonance ( h ν ~882 eV) ones ("3 d-4 f spectra"). In the 3 d-4 f spectra of CeNiSn, one can see a strong peak due to the f 1 final states near E F and the shoulder structure of the f 0 final states at ~2.4 eV. For CePdSn, the f 0 peak at about 1.9 eV and the f 1 peak near E F , which is comparable to slightly stronger than the f 0 peak, are observed. The Ce 4 f contributions obtained by the same procedure in the Ce 4 d resonance region ( h ν ~120 eV, "4 d-4 f spectra") are also displayed. They were measured at beamline BL-3B of the Photon Factory. One can notice that the 4 d-4 f spectral shapes are 44 Fig.4: Orbital, spin and total magnetic moments of Pt atom estimated from the sum rules. These data are subjected to no correction for the saturation magnetization. remarkably different from those of the 3 d-4 f ones. These striking differences are thought to originate from the differences of the c-f mixings in the bulk and surface, where c stands for the valence bands. Details of the resonant spectra near E F are shown in Figure 1 (b) . In the Ce 3 d on resonance spectrum In the Ce 3 d and 4 d resonance spectra of CePdSn, on the other hand, the point at half-intensity of the edge is located fairly close to E F . The point of the maximum intensity in the Ce 3 d on resonance spectrum of CePdSn is positioned at ~0.25 eV, which is deeper than that for CeNiSn. This is considered to be reflected in the differences of the c-f mixings, and therefore, that of T K between weakly hybridized CePdSn and more strongly hybridized CeNiSn. We also show the 4 d-4 f spectra near E F , broadened by the resolution of the Ce 3 d resonant spectra in Figure 1 (b) . The broadened 4 d-4 f spectra do not coincide with the Ce 3 d on resonance spectra for both materials, indicating that the different line shapes cannot be explained by the resolution effect only. In order to understand the bulk Ce 4 f contributions, CeNiSn, the binding energy of the maximum intensity is located at <0.2 eV. A further remarkable finding is that the mid-point of the leading edge in the Ce 3 d and 4 d resonance spectra is unambiguously located above E F , suggesting the existence of a narrow and strong Kondo peak above E F . we have performed spectral calculations using a non- crossing approximation (NCA) based on the single impurity Anderson model (SIAM) [4]. The properly fitted results of the calculations are compared with the bulk-sensitive 3 d-4 f spectra in Figure 2, in which the NCA-calculated spectra semi-quantitatively reproduce the experimental ones for both compounds. The estimated T K ’s from the calculations are comparable to those mentioned above. For reference, we also show a properly broadened Ce 4 f partial density of states (PDOS) obtained by a band-structure calculation for CeNiSn [5] in Figures 2 (a) . According to an improvement of the beamline monochromator, the third resonant photoemission study is nowadays done at a total resolution of 100 meV. 45 Fig. 1: (a) High resolution Ce 3d resonant photoemission spectra: ( on: h ν ~ 882 eV, off: h ν 875 eV ) of CeNiSn and CePdSn compared with the Ce 4f contributions obtained from the Ce 4d resonant photoemission. (b) Spectra near E F . Akira Sekiyama a , Shigemas a Sugaa and Yuji Saitoh b (a) Osaka University (b) SPring-8 / JAERI E-mail: sekiyama@mp.es.osaka-u.ac.jp References [1] Y. Saitoh et al., J. Synchrotron Rad. 5 (1998) 542. [2] T. Takabatake et al., Phys. Rev. B 41 (1990) 9607; Phys. Rev. 45 (1992) 5740. [3] D. T. Adroja et al., Solid State Commun. 66 (1988) 1201. [4] O. Gunnarsson and K. Schönhammer, Phys. Rev. B 28 (1983) 4315. [5] A. Yanase and H. Harima, Prog. Theor. Phys. Suppl. 108 (1992) 19. 46 Fig. 2: Comparison of the 3d-4f spectra with theoretical calculations for (a) CeNiSn and (b) CePdSn. For the experimental 3d-4f spectra, the secondary electron backgrounds of the integration type were properly subtracted from those in Fig. 1 (a) .