J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 Full Papers Journal of the Physical Society of Japan 94, 074703 (2025) https://doi.org/10.7566/JPSJ.94.074703 Polarization-dependent Bulk-sensitive Valence Band Photoemission Spectroscopy and Density Functional Theory Calculations: Part IV. 4f Rare-earths Shigenori Ueda1,2+ and Ikutaro Hamada3 1 Research Center for Electrical and Optical Materials, National Institute for Materials Science (NIMS), Tsukuba, Ibaraki 305-0044, Japan 2 Synchrotron X-ray Station at SPring-8, NIMS, Sayo, Hyogo 679-5148, Japan 3 Department of Precision Engineering, Graduate School of Engineering, The University of Osaka, Suita, Osaka 565-0871, Japan (Received February 4, 2025; accepted May 14, 2025; published online June 11, 2025) The valence band electronic structures of the elemental solids of 4 f rare-earths (REs: La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu) were studied by bulk-sensitive polarization-dependent hard X-ray photoemission spectroscopy (HAXPES). Owing to the localized 4 f electrons in REs, the valence band (VB) region showed complicated multiplet structures reflecting the 4 f photoemission final states, except for La and Ce. The profiles of the 4 f multiplet structures did not depend on the X-ray polarization. In contrast, the VB spectra, which mainly consisted of the 5d and 6s states, near the Fermi-level (EF) weakly depended on the X-ray polarization. The polarization-dependent VB spectra near EF for La and Lu were reproduced by the 5d and 6s partial densities of states, which were obtained from the density functional theory calculations for La and Lu, multiplied by photoionization cross-sections. The complicated multiplet structures were also observed in the 3d and 4d core-level HAXPES spectra via the exchange interaction between the core–hole and 4 f electrons. 1. Introduction In the elemental solids of 4f rare-earths (REs) or so-called lanthanides (La–Lu), the number of electrons in the 4f orbital (n) changes with the atomic number (Z), where n is given by Z  57 (Z ¼ 57 for La), except for Eu and Yb (n ¼ Z  57 þ 1). Many 4f RE elemental solids have a magnetic ground state due to the localized 4f orbital as well as elemental solids of 3d transition metal (TM), but Curie or Neel temperature (TC or TN ) of 4f REs is lower than room temperature (RT) in contrast to that TC or TN of 3d TMs such as Cr, Mn, Fe, Co, and Ni is higher than RT. The localized 4f electrons in REs have attracted in fundamental physics by means of various high-energy spectroscopies such as photoelectron spectroscopy (PES), inverse PES (called as IPES or BIS: bremsstrahlung isochromat spectroscopy), X-ray absorption spectroscopy (XAS), and electron energy loss spectroscopy (EELS), since complicated multiplet structures have been observed in many REs in these methods.1) Valence fluctuation and heavy fermion properties have also attracted in the studies of electronic states of RE alloys and compounds by using PES.2) RE-TM intermetallic compounds have been brought into the research field of advanced permanent magnets.3) In addition, it is known that some of RE alloys and compounds exhibit catalytic properties4) and superconductivity.5,6) Thus, 4f REs are of importance in fundamental physics and applications. This paper is a series of X-ray polarization-dependent valence band (VB) hard X-ray PES (HAXPES) studies of elemental solids using 5.95 keV X-rays. HAXPES is well known as a bulk-sensitive probe of the electronic states of solids,7–10) since the kinetic energy of photoelectron reaches to several-keV. Such high-kinetic energy photoelectrons have a large inelastic mean-free-path (IMFP) in solids according to the TPP-2M equation.11) Therefore, HAXPES enables us to obtain the bulk-sensitive electronic states of solids with less containing surface-related electronic states in a suitable experimental geometry.12) In the previous studies, we have shown the experimental polarization-dependent VB HAXPES spectra for 3d, 4d, and 5d TMs with the simulatied results of VB spectra obtained by using the calculated photoionization cross-sections and partial densities of states (PDOSs) based on the density functional theory (DFT) calculations.13–15) In this work, we have performed the polarization-dependent VB HAXPES measurements for 4f REs of La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu elemental solids. The changes in the VB spectral shapes of the band originated from the 5d, 6s, and 6p states showed a clear X-ray polarization dependence mainly due to the photoionization cross-section ratio of 5d=6s orbitals, which strongly depended on X-ray polarization and experimental geometry.13) In contrast, the spectral shapes of multiplet structures due to the 4f n1 photoemission final states did not depend on X-ray polarization, which would be attributed to the use of polycrystalline 4f RE samples. We have also performed the 3d, 4d, and 4s core-level HAXPES measurements for 4f REs as a reference data set. Note that conventional X-ray photoemission spectroscopy (XPS) using Al-Kα (1486.7 eV) or Mg-Kα (1253.6 eV) cannot access to all the 3d core-levels in 4f REs owing to their high binding energy (830–1650 eV), while all the 3d core-level spectra for 4f REs are observable in HAXPES at the photon energy of 5.95 keV with higher bulk-sensitivity compared to conventional XPS. The 3d and 4d core-level HAXPES spectra for 4f REs have revealed the complicated multiplet structures due to the c4f n photoemission final states (c denotes core– hole) because of high-resolution (HR) measurements. 2. Experimental X-ray polarization-dependent VB HAXPES measurements for 4f RE elemental solids were conducted at the undulator beamline BL15XU 9,16) of SPring-8. The polycrystalline ingots of La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu with purity better than 99.9% were in situ fractured to obtain clean surfaces in a sample preparation chamber with the base pressure of 2:0  107 Pa. Then, the samples were transferred to an analysis chamber, which was equipped with a HR hemispherical electron analyzer (VG 074703-1 Society of Japan ©2025 The Physical ©2025 The Author(s) This article is published by the Physical Society of Japan under the terms of the Creative Commons Attribution 4.0 License. Any further distribution of this work must maintain attribution to the author(s) and the title of the article, journal citation, and DOI. J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 J. Phys. Soc. Jpn. 94, 074703 (2025) Full Papers Scienta, R4000), with the base pressure of 3:0  108 Pa. The temperature of samples was kept at 15 K during the measurements. The photoelectrons were excited by horizontal (H-pol), vertical linear polarized (V-pol) X-rays or left- or right-handed circularly polarized (C-pol) X-rays. A nearnormal emission geometry with a grazing incidence (approximately 2°) of X-rays with respect to a nominal sample surface was applied to the HAXPES experiments. The incident X-rays were firstly monochromatized by a Si 111 double-crystal monochromator and then were further monochromatized by a Si 333 channel-cut monochromator. The Xray polarization was switched by using a diamond phase retarder, while the phase retarder was retracted from the Xray axis for H-pol X-rays. The estimated degree of linear or circular polarization (PL or PC ) was ∼1.00, ∼0.70, and ∼0.95 for H-, V-, and C-pol X-rays, respectively. PL correction to ∼1.00 was done for the VB spectra measured with V-pol Xrays in this work as described elsewhere.13–15) The VB spectra for C-pol X-rays were obtained by the average of the VB spectra for left- and right-handed C-pol X-rays in each element. The 3d, 4d, and 4s core-level HAXPES measurements were performed with H-pol X-rays. Overall energy resolution (E) of the HAXPES measurements was set to 130 meV, and the binding energy (EB ) was calibrated by the Fermi-level (EF ) of a reference Au film. For Ce and Eu, we have done HR HAXPES measurements of VB and Eu 4f states with E ¼ 85 meV. 3. DFT Calculations DFT calculations were performed to obtain the electronic band structures (total DOS and PDOSs) for La and Lu. The projected augmented method17) as implemented in the VASP code18,19) was used. The generalized gradient approximation of Perdew, Burke, and Ernzerhof 20) was applied to the exchange–correlation functional. The experimental structures of La21) and Lu22) were used in the DFT calculations: double hexagonal closed pack (dhcp) structure for La and hcp structure for Lu. A kinetic energy cut-off of 750 eV was used to expand the wave functions in terms of a plane-wave basis set. The tetrahedron method was used for the Brillouin-zone integration with the 36  36  12 and 36  36  24 k-point grids for La and Lu, respectively. We omitted to perform the DFT calculations of the other REs due to the strong final state effects in the partially filled 4f states in PES.23) The spin–orbit coupling was not considered in the DFT calculations for simplicity. 4. Results 4.1 Polarization-dependent VB HAXPES spectra of 4f REs Figures 1(a)–1(n) shows the polarization-dependent VB spectra of the 4f REs of La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu. The VB spectra for H-, V-, and Cpol X-rays were normalized by the incident photon intensity in each element, and subtractions of an integrated-typebackground were performed. The VB intensity was highest (lowest) for H-pol (V-pol) X-rays in each element and the intermediate VB intensity was found for C-pol X-rays, as well as HAXPES for the elemental solids of 3d, 4d, and 5d TMs.13–15) The VB spectra were classified into two parts: 5d6sp hybridized band near EF and multiplet structures due to 4f photoemission final states. For La and Ce, the 4f derived S. Ueda and I. Hamada structures cannot be found in Figs. 1(a) and 1(b). The absence of 4f photoemission in La is natural because of the empty 4f orbital (n ¼ 0). The absence of 4f photoemission in Ce will be mentioned latter. For Yb and Lu, the spin–orbit doublet (4f7=2 and 4f5=2 ) due to the fulfilled 4f orbital (n ¼ 14) can be seen in Figs. 1(m) and 1(n). The 4f multiplet structures strongly depended on n, and were similar to those for 4f REs reported in conventional XPS measurements.23) One sees that the 4f multiplet structures in Eu and Gd are similar each other, which is caused by same n in the initial state of Eu and Gd. The multiplet structure becomes complicate with increasing n up to 13. The assignment of 4f multiplet states was described elsewhere.23,24) Although the profile of multiplet structure in Figs. 1(c)– 1(l), does not depend on the X-ray polarization, the profiles of 5d6sp-derived band near EF in Figs. 1(a)–1(n) change with the X-ray polarization. The changes in the 5d6spderived band profile are weak in peak structures (shoulder structure in the case of La) immediately below EF and are strong in broad peaks located at EB  1:5 eV. According to the previous polarization-dependent HAXPES work for 3d and 4d TMs,13,14) the photoionization cross-sections for s atomic orbitals abruptly decrease in V-pol X-rays compared to H-pol X-rays in our experimental configuration. Therefore, we see that the 5d-derived photoemission intensity is dominant in the observed band profile with a varied contribution of 6s-derived one by the X-ray polarization in the HAXPES spectra of 4f REs. For Eu and Yb, changes in the observed band profile are unclear, since the band is overlapped with 4f photoemission in both cases. A small hump at EB  5 eV shows the X-ray polarization dependence for Eu and Gd in Figs. 1(f ) and 1(g), and the intensity of the hump reduces in the case of V-pol X-rays, suggesting that the small hump mainly originates from the 6s-derived states. Note that the contribution of 6p-derived intensity in 4f REs might be quite small by considering the experimental HAXPES results and calculated photoionization crosssections for 6p atomic orbitals in 5d TMs.15) In Figs. 1(a)–1(n), a metallic Fermi edge was clearly observed in each RE regardless of X-ray polarization. This result indicates that the recoil effect25–27) in 4f RE elemental solids is sufficiently smaller than E ¼ 130 meV. In fact, the calculated recoil energy of photoelectrons with the kinetic energy (EK ) of 5.95 keV for the isolated 4f RE atoms according to Refs. 25–27 ranges from 23.5 to 18.7 meV as listed in Table I. A sufficiently small recoil energy was further validated from the HR VB spectrum for Ce with E ¼ 85 meV, which also showed a clear Fermi edge, as shown in Fig. 1(o). 4.2 3d core-level HAXPES spectra of 4f REs Figure 2 shows the 3d core-level HAXPES spectra of 4f RE elemental solids. The 3d core-level spectra have shown the spin–orbit splitting (3d5=2 and 3d3=2 ) and complicated multiplet structures mainly due to the 3d 9 4f n photoemission final states via the exchange interaction between the core– hole and 4f electrons. In contrast, the spin–orbit splitting is dominant for n ¼ 0 (La) and 14 (Yb and Lu) due to empty and fulfilled 4f orbitals, respectively. The broad structures located at the higher EB side of the 3d5=2 and 3d3=2 main peaks are due to the energy loss satellites for La, Yb, and Lu. 074703-2 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 Full Papers J. Phys. Soc. Jpn. 94, 074703 (2025) Fig. 1. S. Ueda and I. Hamada Continued on next page. Note that conventional XPS measured with Al-Kα (1486.7 eV) or Mg-Kα (1253.6 eV) cannot access to all the 3d corelevels in 4f REs owing to their high EB (> 800 eV). In fact, Ref. 28, which presents the major and highintensity core-levels in elemental solids, shows the limited 3d core-level spectra (La, Ce, Pr, Nd, Sm, and Eu) in 4f REs. The observed La, Ce, Pr, Nd, Sm, and Eu 3d HAXPES spectra are similar to the reported XPS spectra measured with Al-Kα in Ref. 28 and are sharper than them owing to the HR measurements in this work. Note that in the case of light RE elements, it is known that the experimental spectra are understood by the theoretical calculations based on the impurity Anderson model with taking into account the multiplet states and the charge transfer between 4f and conduction electrons and that in the case of heavy RE elements, the ionic model is often applied due to negligible charge transfer effect as described in Ref. 29. In this paper, we do not discuss the charge transfer effects for simplicity, since details of them are described in Ref. 30. The Eu 3d HAXPES spectrum shows several sharp peaks in the 3d5=2 and 3d3=2 main structures. A resemblance between the Gd and Eu 3d HAXPES spectra is observed, because the spectra mainly reflect the 3d 9 4f 7 photoemission final states in both Gd and Eu cases. A higher photoexcitation (5.95 keV in this work) enables the observation of the all 3d core-level photoemission with high intensity and long probing depth in 4f REs. The probing depth (3  IMFP) for the 3d core-level HAXPES in 4f REs ranges from 31.8 (La) to 17.1 nm (Lu), where IMFP is calculated by the TPP2M equation.11) By utilizing the large probing depth (i.e., high bulk-sensitivity), temperature-induced valence transitions in YbInCu4 and EuNi2(Si1x Gex)2 have been analyzed 074703-3 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 Full Papers J. Phys. Soc. Jpn. 94, 074703 (2025) S. Ueda and I. Hamada Fig. 1. (Color online) (a)–(n) X-ray polarization-dependent VB HAXPES spectra of La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu polycrystalline samples for H-pol (red), V-pol (blue), and C-pol (green) X-rays. In each element, the VB spectra were normalized by the incident X-ray intensity. The insets in (m) and (n) show the enlarged view of the VB spectra near EF . (o) HR Ce VB HAXPES spectrum and (p) Eu 4f spectrum in the case of E ¼ 85 meV. Vertical bars (red) in (p) indicate the multiplet states with different J in the final states. Table I. Recoil energy of photoelectrons with EK of 5.95 keV for La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu atoms. Recoil energy (meV) La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 23.5 23.3 23.2 22.6 21.7 21.5 20.7 20.5 20.1 19.8 19.5 19.3 18.9 18.7 from the Yb and Eu 3d core-level HAXPES spectra, respectively.31,32) The observation of the Ce 3d core-level spectrum is useful to determine the crystal phase of Ce metal, because α- and γCe shows the different Ce 3d XPS spectral shapes.33) The observed Ce 3d HAXPES spectrum in Fig. 2(b) shows similarity to the Ce 3d XPS spectrum for α-Ce rather than γ-Ce according to Ref. 33. For the spectrum of α-Ce, characteristic peaks located at EB of ∼894.7 and ∼913.2 eV was found in both the HAXPES and XPS spectra, while these peaks were absent in the 3d XPS spectrum of γ-Ce. It is reasonable to assume that the used Ce sample in this work is in the low-temperature α-phase, since the sample was kept at 15 K during the measurements. For Pr metal, the observed 3d 074703-4 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 Intensity (arb. units) J. Phys. Soc. Jpn. 94, 074703 (2025) Full Papers S. Ueda and I. Hamada a b c d e f g h i j k l m n Fig. 2. (Color online) RE 3d core-level HAXPES spectra measured with H-pol X-rays. 074703-5 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 J. Phys. Soc. Jpn. 94, 074703 (2025) Full Papers S. Ueda and I. Hamada Fig. 3. (Color online) RE 4d core-level HAXPES spectra measured with H-pol X-rays. core-level HAXPES spectrum was similar to the spectra measured with the photon energies of 1253.6, 2450, and 5450 eV in Refs. 34 and 35, suggesting that the EK dependence on the spectral shape is very weak in Pr metal as well as La, Ce, and Nd metals, which show the resemblance between the HAXPES and the conventional XPS spectra in the 3d core-level regions. In contrast, the EK dependence on the spectral shape for Sm metal is large. The weak peaks located at EB of ∼1073.5 and ∼1100.7 eV in the Sm 3d core-level HAXPES spectrum increase in the conventional XPS spectra measured with Al-Kα (1486.7 eV) and Mg-Kα (1253.6 eV). Furthermore, as seen in Ref. 28, these peaks are larger in the spectrum for Mg-Kα than that for AlKα. Lower EK for Mg-Kα than Al-Kα corresponds to shorter IMFP for Mg-Kα than Al-Kα according to Ref. 11. This result indicates that these two peaks originate from electronic states near surface of Sm metal, which has been already reported in the take-off-angle dependent Sm 4d XPS spectra of Sm metal.1) 4.3 4d core-level HAXPES spectra of 4f REs Figure 3 shows the 4d core-level HAXPES spectra of 4f RE elemental solids. Owing to higher energy resolution in HAXPES in this work compared to conventional XPS, the 4d core-level HAXPES spectra shows fine and complicated multiplet structures mainly due to 4d 9 4f n photoemission final states. The observation of these fine structures is due to less core–hole lifetime broadening in the shallower 4d corelevels than the 3d core-levels and relatively larger exchange interactions between the core–hole and 4f electrons in the 4d core-level photoemission process. For La, Yb, and Lu, the spin–orbit doublet (4d5=2 and 4d3=2 ) is clearly seen in Figs. 3(a), 3(m), and 3(n), while it is not clear in the other 4f REs due to their complicated multiplet structures. The broad structure (shoulder) located at the higher EB side of the 4d5=2 and 4d3=2 main peaks is due to the energy loss satellites for La (Yb and Lu). Since EB of 4d core-level region for 4f REs is sufficiently lower than the energies of Al-Kα (1486.7 eV) and Mg-Kα (1253.6 eV), conventional XPS can access the all 4d core-levels for REs and many 4d core-level spectra of the RE elements and oxides have been reported in the early stage of XPS.28,29,36–38) In Ref. 29, the experimental 4d core-level XPS spectra for Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu metals have been compared with the theoretical multiplet calculations based on the ionic model. The observed 4d corelevel HAXPES spectra shown in Fig. 4 are much sharper than the reported experimental 4d core-level XPS spectra in Refs. 29 and 37 due to the HR HAXPES measurements. 074703-6 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 Full Papers Intensity (arb. units) J. Phys. Soc. Jpn. 94, 074703 (2025) S. Ueda and I. Hamada a b c d e f g h i j k l m n Fig. 4. (Color online) RE 4s core-level HAXPES spectra measured with H-pol X-rays. Note that for Sm, the small peak located at EB of ∼123 eV originates from the electronic state near surface, because of strong reduction in the HAXPES spectrum compared to the XPS one37) and absence of the small peak in the theoretical calculation,29) as mentioned above. In the 4d XPS spectra for Ce, Sm, and Eu metals with E ¼ 0:55 eV, fine multiplet structures were observed,37) but our results on Ce, Sm, and Eu metals with E ¼ 130 meV detected further fine structures. For Eu metal, the impact of E in the spectral shape is apparent; the structure located at EB of ∼130 eV was observed as a broad peak with E ¼ 0:7 eV in Ref. 29, that was observed as several peaks with E ¼ 0:55 eV in Ref. 37, and that was clearly resolved as several sharp peaks in HAXPES with E ¼ 130 meV in this work. Due to the 4d 9 4f 7 final states in both Eu and Gd metals, the 4d core-level HAXPES spectral shapes are similar each other as well as the 3d core-level HAXPES cases as shown in Fig. 2. The observed 4d core-level HAXPES spectra with E ¼ 130 meV allow the direct comparison with the theoretical multiplet structures indicated by the vertical bars for Sm, Gd, Tb, Dy, Ho, Er, and Tm in Ref. 29. These results indicate the importance of HR measurements of core-level as well as valence band in HAXPES. 4.4 4s core-level HAXPES spectra of 4f REs Figure 4 shows the 4s core-level HAXPES spectra of 4f RE elemental solids. It is known that the multiplet structures due to photoemission final states are simplified in the s corelevel PES spectra and that the spin exchange splitting appears in the s core-level spectra due to the interaction between the s core–hole and valence electrons.39) Therefore, the complicated multiplet structures like 3d and 4d core-level HAXPES spectra are absent in the 4s core-level HAXPES spectra for 4f REs, and most 4s core-level spectra shown in Fig. 4 show the two peaks due to the configuration of core–hole spin parallel or antiparallel to the spin of 4f electrons.40,41) As mentioned in Sect. 1, many RE elemental solids have a magnetic ground state so that the local spin magnetic moment is non-zero, except for La (empty 4f orbital) and Yb and Lu (fulfilled 4f orbital). Therefore, the peak splitting in the 4s spectra due to the spin exchange interaction between the 4s core–hole and 4f electrons can be clearly seen in Sm, Eu, Gd, Tb, Dy, and Ho metals, while the peak splitting is unclear or small in Ce, Pr, Nd, Er, and Tm metals due to their small spin magnetic moments. One sees that the 4s core-level HAXPES spectra of 4f RE metals in Fig. 4 show clear shoulder structures at the higher EB side of two peaks (Nd, Sm, Eu, Gd, Tb, Dy, Ho, and Er) or main peak (La, Ce, Pr, Tm, Yb, and Lu). Note that 074703-7 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 J. Phys. Soc. Jpn. 94, 074703 (2025) Full Papers Fig. 5. (Color online) Total DOSs and 5d, 6s, 6p, and 4f PDOSs for (a) dhcp La and (b) hcp Lu obtained from the DFT calculations. a small peak at EB of ∼531 eV is oxygen-related surface contaminations for Lu. The shoulder structures in Fig. 4 seem to be absent in the 4s XPS for insulating RE trifluoride.41) Thus, the shoulder structures in the 4s HAXPES spectra for 4f RE metals originate from the energy loss satellites due to plasmon excitations. The relatively large intensity of the energy loss satellite in the 4s HAXPES suggests that the presence of the energy loss satellite at the higher EB side of the main structure in the 3d and 4d HAXPES spectra in the case of 4f RE metals. A series of 3d, 4d, and 4s core-level HAXPES measurements, thus, provides information on the energy loss satellite in the core-level spectra for 4f RE metals and suggests the necessity of considering the energy loss satellite in the analysis of core-level spectra based on the ionic model in metallic RE alloys and compounds. 5. Discussion To understand the electronic structures of 4f RE elemental solids from the VB HAXPES measurements, we have performed the DFT calculations to obtain the total DOS and PDOSs as shown in Fig. 5. Since the DFT calculations generally do not give the photoemission final states, we focused on the RE elements with n ¼ 0 (La) and 14 (Lu) to exclude the 4f contribution to the DOS. As seen in Fig. 5, the 5d PDOS mainly contributes to the DOS in each case. The 4f contribution to the occupied states in La is negligibly small as seen in Fig. 5(a), while the 4f states are localized as a shallow core-level at EB of ∼4.9 eV in Lu as seen in Fig. 5(b). Figure 6 shows the simulated VB HAXPES spectra for various X-ray polarization for La and Lu. In the simulations, the sum of the 5d and 6s PDOSs multiplied by the photoionization cross-sections per electron with taking the orbital-dependent angular distributions with respect to the Xray polarizations into account was used, while the 6p PDOS S. Ueda and I. Hamada Fig. 6. (Color online) Simulated X-ray polarization-dependent VB spectra for (a) La and (b) Lu. The red (blue) [green] solid curves indicate H-pol (Vpol) [C-pol] X-rays. Note that in the simulation for Lu, the peak positions of the Lu 4f7=2 and 4f5=2 were obtained from Fig. 1(n), and the branching ratio was set to 8 : 6 as indicated by black vertical bars in (b). The inset of (b) shows the enlarged view of the VB spectra near EF . was ignored probably due to a quite small cross-section in RE atoms as expected from a quite small 6p cross-section in 5d TMs.15) In addition, the 4f PDOSs for La and Lu obtained by the DFT calculations were not used in the simulations. The simulated VB HAXPES spectra for La and Lu in Figs. 6(a) and 6(b) reproduced well the experimental polarizationdependent VB HAXPES spectra in Figs. 1(a) and 1(n), respectively. Therefore, the X-ray polarization-dependent VB spectral shape can be understood by the changes in the crosssection ratio with considering the angular distribution of 5d orbital to 6s one. The calculated 6s and 4f cross-sections with respect to 5d one for RE atoms for 6 keV photoexcitation42–44) with taking the angular distributions with respect to the X-ray polarizations into account are summarized in Table II. The calculated 5d cross-section ratios of RE atoms with respect to La 5d one are also summarized in Table III. The drastic changes in the 6s cross-section between H- and V-pol in our experimental geometry seen in Table II reflect the changes in the VB profile at EB of ∼1.5 eV (2.2 eV) for La (Lu). This result is consistent with the fact that the PDOS calculations show that the broad 6s states locate at around EB of ∼1.5 eV (∼2.2 eV) for La (Lu) and that the relatively narrow 5d states locate near EF in both La and Lu. These changes in the experimental band profiles derived from the 5d and 6s states for the other REs were also reproduced by tentative simulations as mentioned in Appendix, except for Eu and Yb. Next, we discuss the 4f states in the VB region of Ce metal. In Fig. 1(b), it is hard to find the 4f photoemission signal for Ce in HAXPES, although a small hump at EB of ∼3.3 eV due to the 4f photoemission signal for Pr can be seen 074703-8 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 Full Papers J. Phys. Soc. Jpn. 94, 074703 (2025) S. Ueda and I. Hamada Table II. Polarization-dependent cross-sections per electron of 5d, 4f, 6s orbitals for isolated La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu atoms for H-, V-, and C-pol X-rays at the photon energy of 6 keV. The 5d cross-sections were obtained by the interpolation of those for La, Gd, and Lu. The 6s crosssection for Ce was obtained by the extrapolation of those for the other REs. The values were normalized by that of the 5d orbital in each element for H-pol Xrays. Note that the orbital-dependent angular distributions with respect to the X-ray polarizations were taken into account in the polarization-dependent crosssections in the table. H-pol. Element La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu V-pol. C-pol. 5d 4f 6s 5d 4f 6s 5d 4f 6s 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 — 0.1917 0.2273 0.2670 0.3558 0.4069 0.4942 0.5244 0.5915 0.6645 0.7435 0.8297 0.9219 1.0950 1.4935 1.1898 1.1306 1.0807 0.9868 0.9476 1.0516 0.8799 0.8504 0.8232 0.7980 0.7747 0.7531 0.8898 0.2405 0.2336 0.2247 0.2170 0.2043 0.1990 0.1994 0.1900 0.1861 0.1826 0.1793 0.1764 0.1737 0.1690 — 0.0899 0.1047 0.1201 0.1557 0.1754 0.2097 0.2188 0.2426 0.2680 0.2951 0.3238 0.3546 0.4154 0.01880 0.01549 0.01518 0.01328 0.01443 0.01438 0.01637 0.01415 0.01413 0.01411 0.01421 0.01431 0.01441 0.01762 0.6086 0.6056 0.6017 0.5983 0.5927 0.5904 0.5905 0.5864 0.5847 0.5832 0.5818 0.5805 0.5793 0.5773 — 0.1351 0.1594 0.1858 0.2461 0.2808 0.3397 0.3588 0.4030 0.4508 0.5024 0.5583 0.6182 0.7318 0.7483 0.5962 0.5667 0.5409 0.4948 0.4753 0.5276 0.4416 0.4269 0.4134 0.4009 0.3894 0.3787 0.4477 Table III. Relative per-electron cross-sections of the La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu 5d orbitals to that of the La 5d orbital for H-pol X-rays at the photon energy of 6 keV. Note that the cross-section ratio (Ti 3d=La 5d) is 0.01481. Relative cross-section of 5d orbitals La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 1.0000 1.0997 1.1829 1.2662 1.4328 1.5160 1.6324 1.6826 1.7658 1.8491 1.9324 2.0157 2.0989 2.1658 in Fig. 1(c). If the 4f photoemission signal for Ce is detectable in HAXPES, a sharp peak structure located near EF can be found as observed in Ref. 45. To clarify whether the 4f photoemission signal is detectable, we have performed the HR HAXPES measurement for Ce using H-pol X-rays as shown in Fig. 1(o). The observed spectrum with E ¼ 85 meV was sharper than that with E ¼ 130 meV for H-pol X-rays, but a peak at EB  2 eV due to the 4f 1 -to-4f 0 photoemission process and a sharp peak (a tail of Kondo resonance peak) in the vicinity of EF in α-Ce (lowtemperature phase) reported in Ref. 45 cannot be found. On the other hand, these two structures can be seen in the XPS spectrum for Ce (high-temperature γ-phase), even though their intensities are weak compared to the photoemission intensity derived from the Ce 5d states.23) This fact suspects that the photoionization cross-section ratio of 4f=5d orbitals in HAXPES (5.95 keV) is much smaller than that in XPS (1486.7 eV). The calculated cross-section ratios of 4f=5d orbitals are ∼0.19 and ∼1.16 for HAXPES and XPS in the same experimental geometry, respectively.46) The absence of the 4f related peaks even in HR-HAXPES might lead that the calculated cross-section ratio of 4f=5d orbitals for 6 keV is overestimated, since the 4f=5d ratio of ∼0.19 is not so small to detect the 4f related peaks. These results suggest that the direct observation of Ce 4f photoemission is in a difficult situation for HAXPES due to a quite smaller 4f cross-section compared to 5d one and the condition of n ¼ 1 for Ce. Finally, we mention the impact of the HR measurement in HAXPES. Figure 1(p) shows the Eu 4f HAXPES spectrum with E ¼ 85 meV for H-pol X-rays. The fine multiplet structures due to the 4f 7 -to-4f 6 photoemission process were observed. The multiplet states for J ¼ 6, 5, and 4 deduced from the theoretical calculations24) were found as distinguish- able peaks in the experimental spectrum, while that for J ¼ 3 appeared as a shoulder structure. These peaks were not resolved in the VB spectrum with E ¼ 130 meV as seen in Fig. 1(f ) and the simulated Eu 4f spectrum for E ¼ 130 meV as shown in Fig. A·1(e) for H-pol X-rays. Whether the recoil effects25–27) in high EK photoelectrons exist (e.g., the energy shift as listed in Table I and broadening of spectral shape) in 4f RE elemental solids, the result of the Eu 4f HAXPES spectrum shown in Fig. 1(p) strongly suggests the importance of HR measurements in HAXPES to elucidate the electronic states of materials at least photo-excitation (or EK of photoelectrons) up to 6 keV, which is also supported by the fine and complicated multiplet structures for 4f REs in the 3d and 4d core-level HAXPES spectra as seen in Figs. 2 and 3. 6. Summary We have conducted the VB, 3d, 4d, and 4s core-level measurements of the elemental solids of 4f REs (La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu) by means of bulk-sensitive HAXPES with E ¼ 130 meV. The VB spectra for 4f REs were separated by 5d6sp-derived band located near EF and multiplet structure due to the 4f photoemission final states, while the 4f photoemission intensity was not found for La and Ce. The absence of the 4f signal in La is due to n ¼ 0 (the empty 4f level), but the absence of the 4f signal in Ce with n ¼ 1 even in the HR measurement (E ¼ 85 meV) is due to the quite low photoionization cross-section ratio of 4f=5d orbitals. For Yb and Lu, the spin–obit doublet (4f7=2 and 4f5=2 ) was found due to the fully occupied 4f level in these metals. The profiles of band near EF depended on the X-ray polarization in our experimental geometry, while those of the 4f n1 multiplet 074703-9 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 J. Phys. Soc. Jpn. 94, 074703 (2025) Full Papers S. Ueda and I. Hamada Fig. A·1. (Color online) (a)–(l) Simulated X-ray polarization-dependent VB spectra for Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, and Yb with 4f multiplet final states. For Ce, 4f photoemission was excluded, while for Eu and Yb, 5d6s band photoemission was excluded, for simplicity. structure did not depend on the X-ray polarization. The changes in the observed band profiles were understood by the X-ray polarization-dependent cross-section ratio of 6s=5d orbitals. We have performed the simulations of the polarizationdependent VB spectra near EF for La and Lu by using the 5d and 6s PDOSs, which were obtained from the DFT calculations for La and Lu, multiplied by the calculated photoionization cross-sections with considering the angular distributions with respect to the X-ray polarizations. The simulated spectra including the polarization dependence agreed with the experimental spectra for La and Lu. We 074703-10 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 J. Phys. Soc. Jpn. 94, 074703 (2025) Full Papers have tentatively performed the simulations of the polarization-dependent VB spectra for the other REs by using the PDOSs for La or Lu and photoionization cross-section and the calculated 4f multiplet states as shown in Appendix. The observed 3d and 4d core-level HAXPES spectra for 4f REs showed the complicated multiplet structures via the exchange interaction between the core–hole and 4f electrons. In particular, the 4d core-level spectra showed fine and complicated structures owing to the HR measurements (E ¼ 130 meV), smaller lifetime broadening effects, and relatively large exchange interactions with the core–hole compared to the 3d spectral region. The clear spin exchange splitting in the 4s core-level spectra were also observed. Further HR measurement (E ¼ 85 meV) for the Eu 4f photoemission revealed that the clear separation of J ¼ 6, 5, and 4 states. These results strongly suggest the importance of HR measurements in HAXPES to elucidate the electronic states of materials at least photo-excitation up to 6 keV. Acknowledgements The HAXPES measurements were performed with the approval of NIMS Synchrotron X-ray Station at SPring-8 (Proposal No. 2019B4606). This work was partially supported from Tokodai Institute for Elemental Strategy (TIES: Grant No. JPMXP0112101001) and Data Creation and Utilization Type Material Research and Development Project (Grant No. JPMXP1122683430) funded by MEXT, Japan. Appendix: Tentative Simulations of Polarizationdependent VB HAXPES Spectra of REs In this Appendix, the simulated polarization-dependent VB spectra of Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, and Yb were shown in Fig. A·1. For Ce, Pr, Nd, and Sm, the PDOSs of dhcp La were tentatively used to calculate the VB photoemission intensity for simplicity. For Gd, Tb, Dy, Ho, Er, and Tm, the PDOSs of hcp Lu were tentatively used to calculate the VB intensity for simplicity. The energy separation and relative photoemission intensity of multiplets states in 4f photoemission final states were referred to the theoretical calculation in Ref. 24, except for Ce, Pr, and Yb, and were used in the simulations. Note that the energy separation of the multiplet states was adjusted to the experimental 4f spectra. In the case of Ce, the experimental VB spectra did not show the signature of 4f photoemission due to lower 4f orbital cross-section than 5d one as mentioned in Sect. 5. In addition, we tentatively used the 5d and 6s PDOSs of dhcp La in the simulation for dhcp Ce, while the used Ce sample was in the α-phase. For Pr (Yb), the energy position of single (double) peak(s) due to the 4f photoemission was obtained from the experimental VB spectra. The multiplet states were indicated by vertical bars in Fig. A·1. The VB intensity simulation for the 5d and 6s states was not performed for Eu and Yb by using the 5d and 6s PDOSs of La or Lu, since the band structure calculations for La and Lu with the ð5d6spÞ3 configuration were not suitable for Eu and Yb with the ð5d6spÞ2 configuration. The simulated spectra for the 5d6sp-derived band agreed with the experimental spectra except the spectral width of 5d6spderived band in later REs probably due to the difference in lattice constant. Overall 4f multiplet structures in the experimental spectra agreed with the calculations based on the intermediate coupling model,24) but the deviation of the intensity ratio between the multiplet components in the present experiments and theory remained, which was also S. Ueda and I. Hamada found in later RE elements in Ref. 23, although we did not consider the multiplet-dependent lifetime broadening effects in the simulations. In addition, the agreement between the experiment and calculation for Nd [Figs. 1(d) and A·1(c)] was found to be poor. Since the experimental 4f HAXPES spectra in this work reflected bulk-sensitive 4f multiplet structures owing to the larger IMFP compared to the previous XPS results,23) our HR 4f spectra could be a useful benchmark for the theoretical approach to the 4f photoemission in RE elements. + Corresponding author. E-mail: UEDA.Shigenori@nims.go.jp 1) Handbook of the Physics and Chemistry of Rare Earths, ed. K. A. Gschneidner, Jr., L. Eyring, and S. Hüfner (North-Holland, Amsterdam, 1988) Vol. 10. 2) M. Loewenhaupt and K. H. Fischer, in Handbook of the Physics and Chemistry of Rare Earths, ed. K. A. Gschneidner, Jr. and L. Eyring (North-Holland, Amsterdam, 1993) Vol. 16, Chap. 105. 3) Handbook of Advanced Magnetic Materials, ed. Y. Liu, D. J. Sellmyer, and D. Shindo (Springer, New York, 2006) Vol. 4. 4) Handbook of the Physics and Chemistry of Rare Earths, ed. K. A. Gschneidner, Jr. and L. Eyring (North-Holland, Amsterdam, 2000) Vol. 29. 5) M. B. Maple, in Handbook of the Physics and Chemistry of Rare Earths, ed. K. A. Gschneidner, Jr., L. Eyring, and M. B. Maple (NorthHolland, Amsterdam, 2000) Vol. 30, Chap. 187. 6) P. Thalmeier and G. Zwicknagl, in Handbook of the Physics and Chemistry of Rare Earths, ed. K. A. Gschneidner, Jr., J.-C. G. Bunzli, and V. K. Pecharsky (North-Holland, Amsterdam, 2005) Vol. 34, Chap. 219. 7) Y. Takata, M. Yabashi, K. Tamasaku, Y. Nishino, D. Miwa, T. Ishikawa, E. Ikenaga, K. Horiba, S. Shin, M. Arita, K. Shimada, H. Namatame, M. Taniguchi, H. Nohira, T. Hattori, S. Sodergren, B. Wannberg, and K. Kobayashi, Nucl. Instrum. Methods Phys. Res., Sect. A 547, 50 (2005). 8) K. Kobayashi, M. Yabashi, Y. Takata, T. Tokushima, S. Shin, K. Tamasaku, D. Miwa, T. Ishikawa, H. Nohira, T. Hattori, Y. Sugita, O. Nakatsuka, S. Sakai, and S. Zaima, Appl. Phys. Lett. 83, 1005 (2003). 9) S. Ueda, J. Electron Spectrosc. Relat. Phenom. 190, 235 (2013). 10) C. S. Fadley, J. Electron Spectrosc. Relat. Phenom. 190, 165 (2013). 11) S. Tanuma, C. J. Powell, and D. R. Penn, Surf. Interface Anal. 21, 165 (1994); S. Tanuma, C. J. Powell, and D. R. Penn, Surf. Interface Anal. 43, 689 (2011). 12) S. Ueda, Appl. Phys. Express 11, 105701 (2018). 13) S. Ueda and I. Hamada, J. Phys. Soc. Jpn. 86, 124706 (2017). 14) S. Ueda and I. Hamada, J. Phys. Soc. Jpn. 90, 034706 (2021). 15) S. Ueda and I. Hamada, J. Phys. Soc. Jpn. 91, 024801 (2022). 16) S. Ueda, Y. Katsuya, M. Tanaka, H. Yoshikawa, Y. Yamashita, S. Ishimaru, Y. Matsushita, and K. Kobayashi, AIP Conf. Proc. 1234, 403 (2010). 17) P. E. Blöchl, Phys. Rev. B 50, 17953 (1994). 18) G. Kresse and J. Furthmüller, Phys. Rev. B 54, 11169 (1996). 19) G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999). 20) J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996). 21) D. A. Papaconstantopoulos, Handbook of the Band Structure of Elemental Solids (Springer, New York, 2015) 2nd ed., Chap. 13. 22) F. H. Spedding, A. H. Daane, and K. W. Herrmann, Acta Crystallogr. 9, 559 (1956). 23) J. K. Lang, Y. Bear, and P. A. Cox, J. Phys. F 11, 121 (1981). 24) F. Gerken, J. Phys. F 13, 703 (1983). 25) Y. Takata, Y. Kayanuma, M. Yabashi, K. Tamasaku, Y. Nishino, D. Miwa, Y. Harada, K. Horiba, S. Shin, S. Tanaka, E. Ikenaga, K. Kobayashi, Y. Senba, H. Ohashi, and T. Ishikawa, Phys. Rev. B 75, 233404 (2007). 26) Y. Takata, Y. Kayanuma, S. Oshima, S. Tanaka, M. Yabashi, K. Tamasaku, Y. Nishino, M. Matsunami, R. Eguchi, A. Chainani, M. Oura, T. Takeuchi, Y. Senba, H. Ohashi, S. Shin, and T. Ishikawa, Phys. Rev. Lett. 101, 137601 (2008). 074703-11 Society of Japan ©2025 The Physical ©2025 The Author(s) J.Phys.Soc.Jpn. Downloadedfromjournals.jps.jpbyÿxÿriŒê0ûgPe™xzvj_iËon06/26/25 J. Phys. Soc. Jpn. 94, 074703 (2025) Full Papers S. Ueda and I. Hamada 46) Berlin/Heidelberg, 2007) Lecture Notes in Physics, Vol. 715, Chap. 13. J. C. Fuggle, F. U. Hillebrecht, Z. Zonlnierek, R. Lässer, Ch. Freiburg, O. Gunnarsson, and K. Schönhammer, Phys. Rev. B 27, 7330 (1983). S. P. Kowalczyk, N. Edelstein, F. R. McFeely, L. Ley, and D. A. Shirley, Chem. Phys. Lett. 29, 491 (1974). A. Kotani and H. Ogasawara, J. Electron Spectrosc. Relat. Phenom. 60, 257 (1992). S. Hüfner, Photoelectron Spectroscopy (Springer, Berlin/Heidelberg/ New York, 2003) 3rd ed. F. R. McFeely, S. P. Kowalczyk, L. Ley, and D. A. Shirley, Phys. Lett. A 49, 301 (1974). R. L. Cohen, G. K. Wertheim, A. Rosencwaig, and H. J. Guggenheim, Phys. Rev. B 5, 1037 (1972). M. B. Trzhaskovskkaya, V. I. Nefedov, and V. G. Yarzhemsky, At. Data Nucl. Data Tables 77, 97 (2001). M. B. Trzhaskovskaya, V. K. Nikulin, V. I. Nefedov, and V. G. Yarzhemsky, At. Data Nucl. Data Tables 92, 245 (2006). J. H. Scofield, Lawrence Livermore Laboratory, Tech. Rep. UCRL51326 (1973). K. Shimada, in Very High Resolution Photoelectron Spectroscopy, ed. S. Hüfner (Springer, Berlin/Heidelberg, 2007) Lecture Notes in Physics, Vol. 715, Chap. 4. J. J. Yeh and I. Lindau, At. Data Nucl. Data Tables 32, 1 (1985). 074703-12 Society of Japan ©2025 The Physical ©2025 The Author(s) 27) S. Suga, A. Sekiyama, H. Fujiwara, Y. Nakatsu, T. Miyamachi, S. Imada, P. Baltzer, S. Niitaka, H. Takagi, K. Yoshimura, M. Yabashi, K. Tamasaku, A. Higashiya, and T. Ishikawa, New J. Phys. 11, 073025 (2009). 28) Handbook of X-ray Photoelectron Spectroscopy, ed. J. Chastain and R. C. King, Jr. (Physical Electronics, Chanhassen, MN, 1993). 29) H. Ogasawara, A. Kotani, and B. T. Thole, Phys. Rev. B 50, 12332 (1994). 30) F. de Groot and A. Kotani, Core Level Spectroscopy of Solids (CRC Press, Boca Raton/London/New York, 2008). 31) H. Sato, K. Shimada, M. Arita, K. Hiraoka, K. Kojima, Y. Takeda, K. Yoshikawa, M. Sawada, M. Nakatake, H. Namatame, M. Taniguchi, Y. Takata, E. Ikenaga, S. Shin, K. Kobayashi, K. Tamasaku, Y. Nishino, D. Miwa, M. Yabashi, and T. Ishikawa, Phys. Rev. Lett. 93, 246404 (2004). 32) K. Ichiki, K. Mimura, H. Anzai, T. Uozumi, H. Sato, Y. Utsumi, S. Ueda, A. Mitsuda, H. Wada, Y. Taguchi, K. Shimada, H. Namatame, and M. Taniguchi, Phys. Rev. B 96, 045106 (2017). 33) E. Wuillound, H. R. Moser, W.-D. Schneider, and Y. Bear, Phys. Rev. B 28, 7354 (1983). 34) H. Ishii, K. Obu, M. Shinoda, C. Lee, Y. Takayama, T. Miyahara, T. D. Matsuda, H. Sugawara, and H. Sato, J. Phys. Soc. Jpn. 71, 156 (2002). 35) A. Sekiyama, S. Imada, A. Yamasaki, and S. Suga, in Very High Resolution Photoelectron Spectroscopy, ed. S. Hüfner (Springer, 36) 37) 38) 39) 40) 41) 42) 43) 44) 45)