Hard X-ray photoelectron spectroscopy: A few recent applications

Journal of Electron Spectroscopy and Related Phenomena 190 (2013) 242–248

Contents lists available at ScienceDirect

Journal of Electron Spectroscopy and Related Phenomena journal homepage: www.elsevier.com/locate/elspec

Hard X-ray photoelectron spectroscopy: A few recent applications M. Taguchi ∗ , Y. Takata 1 , A. Chainani RIKEN SPring-8 Center, Sayo, Hyogo 679-5148, Japan

a r t i c l e

i n f o

Article history: Available online 23 January 2013 Keywords: Recoil effect in photoelectrons 3D transition metal compounds Cluster model calculation

a b s t r a c t In this report, we discuss a few recent applications of hard X-ray photoelectron spectroscopy (HAXPES) carried out at the RIKEN beamline BL29XU in SPring-8. We first provide a brief description of the salient features of the instrument in operation at BL29 XU in SPring-8. HAXPES studies on the recoil effect of photoelectrons in core levels and valence band states are presented. The experiments show remarkable consistency with theoretical results and indicate the role of phonon excitations in the recoil effect of photoelectrons. We then overview HAXPES applied to the study of a series of 3d transition metal (TM) compounds. The HAXPES experimental results often show an additional well-screened feature in bulk sensitive electronic structure of strongly correlated compounds compared to surface sensitive spectra. The extended cluster model developed by us for explaining this well-screened feature is validated for a series of TM compounds. These results show that HAXPES is a valuable tool for the study of doping and temperature dependent electronic structure of solids with tremendous potential for future activities. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Hard X-ray photoelectron spectroscopy (HAXPES) has undergone a major revival in the last ten years. Early photoelectron spectroscopy (PES) studies were also actually hard X-ray based experiments using laboratory based Mo K˛ (h = 17.479 keV), Cr K˛ (h = 5.417 keV) or Cu K˛ (h = 8.047 keV) X-ray sources [1–3]. While the incident photon flux as well as the detection efficiency did not allow valence band studies, pioneering core level studies could be carried out nicely. These studies played an important role in initiating the applications of PES such as atomic species identification and chemical valence state fingerprinting. Subsequently, the laboratory based HAXPES went out of vogue, mainly due to requirements of higher energy resolution which could be attained with lower energy soft X-ray (SX) laboratory sources such as the Mg K˛ (h = 1253.6 eV) and Al K˛ (h = 1486.7 eV) sources [4]. However, the improvements in resolution by reducing the incident energy also meant higher surface sensitivity of the technique. The HAXPES revival of the last ten years is mainly associated with a soto-say, going back to the higher incident photon energies but with the important difference that laboratory sources were replaced with synchrotron radiation sources. This situation allowed us the energy resolution as well as the higher photon fluxes required to do core level and valence band studies with high-resolution and

∗ Corresponding author. Fax: +81 791 58 2934. E-mail address: [email protected] (M. Taguchi). 1 Passed away during this work on October 28, 2011. MT and AC wish to dedicate this article to his memory. 0368-2048/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.elspec.2013.01.005

high-throughput. Today, several synchrotrons in the world have established HAXPES end-stations and this has already resulted in important results and directions [5–22] which could not have been possible in laboratory measurements. In this report, we describe the salient features of the instrument developed by us at RIKEN beamline BL29XU in SPring-8. We then discuss typical examples of HAXPES applications carried out by us on: (i) the recoil effect of photoelectrons in solids, and (ii) the electronic structure of strongly correlated transition metal (TM) compounds from core level as well as valence band studies. The HAXPES experimental results often show an additional well-screened feature in the bulk sensitive core level states of strongly correlated compounds compared to surface sensitive measurements. We have developed an extended cluster model for explaining the well-screened feature and this model is validated for a series of TM compounds. HAXPES can today be considered to be a fairly standard and valuable tool for the study of the intrinsic electronic structure of solids. 2. Experimental The HAXPES instrument development at BL 29XU in SPring-8 started in 2001. The main aim was to achieve high-energy resolution and throughput which would allow us to do valence band studies of solids at about 6–8 keV kinetic energy of the photoelectrons. One of the main features of the instrument was the use of a grazing incidence geometry [23], with the lens axis placed perpendicular to the incident X-ray beam with an incidence angle relative to the sample surface being set to typically 5◦ . The adopted configuration with a linearly polarized X-ray beam results in the lens

Intensity (arb. units)

M. Taguchi et al. / Journal of Electron Spectroscopy and Related Phenomena 190 (2013) 242–248

0 200

150

100

50

0

-50

-100 -150 -200

Binding Energy (meV) Fig. 1. The Fermi edge of gold measured at T = 20 K for determining the spectrometer resolution. Comparison with a fit to the Fermi-Dirac function convoluted with a Gaussian function gives an energy resolution of 55 meV FWHM.

axis being parallel to the polarization vector of the X-rays. This geometry helps us to gain photoelectron intensity, based on the assumption that the known behavior of angular distribution of photoelectrons from atoms is also valid for solids. It is well-known that the angular distribution of photoelectrons emitted from free atoms show a positive asymmetry parameter ˇ with a maximum in the direction parallel to the polarization vector, which is expected to result in higher signal intensity of emitted photoelectrons in the experimental configuration used by us. Based on the above considerations, we have developed a HAXPES apparatus at undulator beamline BL29XU in SPring-8 [23]. The available photon flux at the beamline is about 1 × 1011 photons/sec with a spot size of 55 (vertical) × 35 (horizontal) ␮m2 . The energy bandwidth of the incident X-rays at ∼8 keV is nearly 40 meV (FWHM). The details of the X-ray optics has been described in an earlier study [24]. For the electron energy analyzer, we have used a modified SES R4000-10KV (VG-Scienta) analyzer. The system consists of a main chamber, a preparation chamber and a fast entry chamber. The base pressure of the main and preparation chambers is about 2 × 10−8 Pa. The sample preparation chamber is fitted with a cleaver for sample cleaving. The sample manipulator consists of a motorized XYZ stage with a closed cycle liquid helium cryostat which can cool the sample down to 20 K. As is usual, the resolution is obtained as the value of the Gaussain FWHM convoluted with the Fermi Dirac function at a specific temperature, which matches the measured Fermi edge spectrum of a reference metal. Fig. 1 shows the Fermi edge of gold measured at 20 K with the highest energy resolution of our spectrometer. The fit to the Fermi edge, shown as a line superimposed on the data, gives an energy resolution of 55 meV at a kinetic energy of 7.94 keV for the Fermi edge of gold measured at 20 K. In the following, we discuss a few applications of HAXPES carried out with this spectrometer.

243

This recoil energy E should show up as an apparent increase of the binding energy (BE) of the photoelectron emitted from an atom. For a solid, the recoil energy can be expected to be absorbed by the phonon bath, and thereby result in the excitation of phonons. HAXPES experiments were carried out on graphite to check for the recoil effect. The C 1s core level of graphite was measured in normal emission using several photon energies (h = 340 eV, 870 eV, 5950 eV and 7940 eV) and the experimental data are shown in Fig. 2(a). The C 1s spectrum at h = 870 eV shifts first to lower BE compared to the h = 340 eV spectrum. This is known to be due to a surface component occurring at a higher BE compared to a bulk component at lower BE. This change does not reflect the recoil effect. On increasing the energies to hard X-rays (h = 5950 eV and 7940 eV), the C 1s BEs shift to higher values again and this is due to the recoil effect. Using Eq. (1), the calculated value of E is 360 meV for h = 7940 eV in close agreement with the experimental shift of 340 meV. Theoretical calculations of the recoil effect were carried out in the adiabatic approximation, taking into account an anisotropic Debye model which is known to be valid for explaining the phonon density of states of graphite [26]. The calculated spectra for normal emission are shown in Fig. 2(b) and they reproduce the experimental shift and shape of the core level spectra, including the asymmetric line shape on the high BE side. Further, it could be shown that the emission angle dependence from a shallow angle (30◦ from grazing incidence) to nearly normal emission (85◦ ) also showed a good correspondence between experiment and theory for the shape of the core level spectra [26]. These results showed the recoil effect of photoelectrons in the C 1s core level spectra of graphite. In a subsequent study, the role of recoil was investigated for the electrons constituting the Fermi edge of elemental metals

3. Recoil effect in photoelectrons The recoil effect is a very basic process associated with the law of conservation of momentum. It was predicted very early on that the photoelectrons emitted from an atom should also undergo the recoil effect [25], arising from the recoil energy E imparted to an atom of mass M, when a photoelectron of mass m is emitted from an atom. Based on a simple picture of an atom at rest in vacuum, if the emitted photoelectron has a kinetic energy EK , the recoil energy is estimated to be E = EK

m M

(1)

Fig. 2. (a) Photon energy dependence (h = 340 eV, 870 eV, 5950 eV and 7940 eV) of the normal emission C 1s core level spectra of graphite. (b) Theoretical spectra obtained by taking into account the recoil effect in a Debye model shows a very good match to the experimental data [26].

244

M. Taguchi et al. / Journal of Electron Spectroscopy and Related Phenomena 190 (2013) 242–248

4.1. Theory Here we give a brief description of the theoretical model for photoemission spectroscopy of TM compounds. One of the most widely used and prime theoretical model is a cluster model (or Anderson impurity model) for the analysis of core level and valence band photoelectron spectra. In this model, we consider only a central TM atom with localized 3d state, but at the same time appropriate linear combinations of ligand 2p orbitals on neighboring sites, and the hybridization between local 3d states and ligand states is taken into account. The Hamiltonian of the conventional cluster model for 3d TM compounds is given by HI Fig. 3. (a) The Fermi edge of Al (circles) and Au (squares) measured with a photon energy h = 7940 eV, The zero of the BE scale is chosen as the observed value of the chemical potential of Au. The solid (Al) and dashed (Au) lines show the theoretical curves calculated for a Debye model and indicate a very good match to the experimental data [27].

= +

  

†

ε3d ()d d + †

εp ()a a +





+

Udd



() = /

aluminium (Al) and gold (Au). Fig. 3 shows high resolution and carefully calibrated experimental spectra obtained using HAXPES for the valence band of Al and Au. As is clear from the data, the Fermi edge of Al shows a remarkable shift compared to that of Au. From conventional fits to the Fermi edge, a shift of 120 meV could be obtained. Further, the Al spectrum required a Gaussian broadening of 160 meV while Au required a broadening of 124 meV. Since the experimental resolution is the same for both spectra, the larger broadening in Al is also a sign of the recoil effect coupled to phonon excitations. Precise analysis of the SX PES data also showed that the Al spectrum shifts by 12 meV compared to that of Au even in the SX PES data [27]. Theoretical calculations were carried out based on a simple band of Bloch electrons coupling to an isotropic Debye model for the phonon density of states. The calculated spectra are overlaid on the experimental data as lines. The results show that the observed shifts and broadening in the experimental spectra could be nicely reproduced by theory. Thus, a quantitative match between experiment and theory for the recoil effect in core level spectra of graphite [26], and the Fermi edge of valence band spectra in elemental metals Al and Au could be established [27].

4. Applications to strongly correlated electron systems Conventional laboratory PES techniques use Mg K˛ (h = 1253.6 eV) or Al K˛ (h = 1486.7 eV) sources. In such cases, the technique becomes very surface sensitive especially for core level spectra, owing to short mean free paths of the emitted low kinetic energy electrons. Therefore, this technique sometimes yields contradictory results in relation to bulk physical properties. The importance for separating the surface and bulk electronic structure has been recognized [28]. Because of recent developments in the experimental technique of HAXPES, the bulk sensitivity has been much improved. As a result, a clear satellite peak is additionally observed on the lower BE side of a main peak in the TM 2p core level HAXPES spectra for many TM compounds [29–37]. While earlier studies based on a cluster model (or Anderson impurity model) could account for the SX PES spectra, the same model was found to be insufficient to explain the HAXPES results. Therefore, it was felt necessary to extend the cluster model (or Anderson impurity model) in order to explain those new satellite features. In the next subsection, we first discuss the original cluster model and the extension carried out by us, and then apply the same for consistently explaining the HAXPES data in relation to bulk properties of strongly correlated electron systems.

−

Udc

+

Hmult .



†

ε2p pm pm

m 

†

†

V ()(d a + a d )

 † † d d d   d  

(2)

(   ) †

†

d d (1 − pm  pm  )

m 

Here ε3d (), ε2p and εp () represent the energies of TM 3d, TM 2p and ligand state such as oxygen 2p states, respectively, with the irreducible representation  of the local symmetry around the TM atom. The indices m and  are the orbital and the spin states. V(), Udd , and −Udc are the hybridization between the TM 3d and the ligand states, the on-site repulsive Coulomb interaction between TM 3d states and the attractive 2p core-hole potential, respectively. The Hamiltonian Hmult describes the intra-atomic multiplet coupling between TM 3d states and that between TM 3d and TM 2p states. The spin-orbit interactions for TM 2p and 3d states are also included in Hmult . The Slater integrals and the spin-orbit coupling constants are calculated by a Cowan’s Hartree–Fock program [38] and the Slater integrals are usually scaled down to 80–85%. This approach is well established for explaining SX PES spectra for various TM compounds and f electron systems in general [39–42]. In addition to the usual cluster model (HI term), we have introduced an additional state near Fermi level labeled ‘C’ as a new screening channel described by HII term, HII =

 

†

εc ()c c +



†

†

V ∗ ()(d c + c d ).

(3)



An effective coupling parameter V* () for describing the interaction strength between TM 3d orbitals and the C state is introduced, analogous to the hybridization V(). The charge transfer energy from the additional state to the TM 3d orbital is * , whereas the usual charge transfer energy  (from the ligand state to the TM 3d orbitals) is defined as the energy difference of the configurationaveraged energies E(3dn+1 L) − E(3dn ). The ‘C’ states represent the coherent state for V2 O3 [30,31] and VO2 [34], and the doping-induced states for doped LaMnO3 [29,33] which develop into electronic states at and near the Fermi level. For the late TM compounds, it represents the Zhang-Rice states for NiO [35] and high-Tc cuprates [32]. Note that this extended cluster model can be applied even to insulating system such as NiO. We would like to emphasize that the model is valid, irrespective of the condition whether the ‘C’ state develops into metallic states which cross the Fermi level. The essential point is that it is necessary to include the additional screening state with a different screening energy V* (for C state positioned at * ) compared to the conventional ligand screening energy V (for L state positioned at ).

M. Taguchi et al. / Journal of Electron Spectroscopy and Related Phenomena 190 (2013) 242–248

4.2. Strongly correlated electron systems 4.2.1. Mott–Hubbard type TM compounds In this subsection we show some typical examples of HAXPES results on Mott–Hubbard type TM compounds. As we mentioned before, the cluster model (or impurity Anderson model) gives a quantitative interpretation for satellite intensities and positions in PES spectra, leading to an accurate description of the ground state and the excitations [39–41]. In this approach, the physics of TM compounds can be described mainly in terms of few parameters, namely, the d–d Coulomb repulsion energy Udd , the charge transfer energy , the ligand p-TM hybridization energy V. Zaanen, Sawatzky, and Allen [42] proposed a classification scheme for TM compounds which soon evolved into a paradigm. In this scheme, the early TM compounds with Udd <  are well known Mott–Hubbard type systems. In contrast, the late TM compounds are charge-transfer type systems with Udd > . Let us begin with early TM compounds. The early TM compounds with its alloys plays the role of a classic Mott–Hubbard system displaying a correlation-induced metal–insulator transition. As a typical example, we show in Fig. 4(a) a complementary set of experimental and calculated Ti 2p core level of Ti4 O7 for the three different phases [36]. The Magnéli phase compound Ti4 O7 exhibits two first-order phase transitions in the temperature dependence of the electrical resistivity (T) at Tc1 ∼ 142 K and Tc2 = 154 K [43].

α

(a) Ti 2p PES Ti4O7

Intensity (arb. units)

HT

Ti

3+

IT β

exp.

LT

cal. 480

470

460

Intensity (arb. units)

Binding Energy (eV)

(b) Ti 1s PES TiN exp. cal. 4980

4970

4960

Binding Energy (eV) Fig. 4. (a) Comparison between experimental Ti 2p HAXPES spectra for HT-, IT- and LT-phases of Ti4 O7 [36]. (b) Experimental Ti 1s HAXPES spectra of TiN are compared with calculation.

245

In the low temperature (LT) phase below Tc1 , it is well established that charge ordered chains of Ti3+ are separated from each other by Ti4+ chains. In the high temperature (HT) phase above Tc2 , there is no long-range order among the two types of Ti atom and the Ti valence is believed to be uniform 3.5+. Thus the HT transition at Tc2 is attributed to a delocalization of the 3d electrons. The nature of electronic states in the intervening region between Tc1 and Tc2 has been a subject of debate for several decades and its understanding is far from satisfactory. An extremely unusual T dependence in Ti 2p HAXPES spectra was observed with clear changes across Tc1 and Tc2 . In order to understand the essential mechanism of the observed HAXPES spectra, we carried out calculations by using the extended cluster model mentioned in the previous subsection. We used, as basis states, six configurations: 3d0 , 3d1 L, 3d2 L2 , 3d1 C, 3d2 C 2 and 3d2 CL. The 3d1 C represents the charge transfer between Ti 3d and the coherent state at EF , labeled C. An effective coupling parameter V* , for describing the interaction strength between the Ti 3d and coherent state is introduced, analogous to the Ti 3d−O 2p hybridization V. The theoretical results reproduce the experimental data very satisfactorily for all phases, as shown in Fig. 4(a). For a finer comparison between theory and experiment, let us first consider the sharp peak labeled ˛ of HT phase in Fig. 4(a), where the final state is well screened 2p5 3d1 C state. The screening effect from the coherent state C near EF leads to the formation of the low energy peak ˛. Next, we consider the LT phase spectrum. Because of the complete absence of the coherent state near EF in LT insulating phase, we set V* (eg ) = 0, leading to a complete suppression of the ˛ feature. The corresponding spectrum is shown by dashed line in Fig. 4(a). The calculated spectrum does not account for the lowest BE feature, labeled ˇ, which occurred 0.2 eV below the ˛ feature. The difference matches rather well with the calculated spectrum for Ti3+ , shown by dotted line in Fig. 4(a). This confirms the existence of Ti3+ in LT phase, as is well established from various X-ray diffraction (XRD) studies. The total calculated spectrum is obtained by a linear combination of cluster model results for Ti4+ and Ti3+ states with a relative weight of 50% and 50%, respectively. The agreement is remarkable. Our calculations basically confirmed the standard valence assignment: the mean valence for Ti is approximately 3.5 in HT phase, with clear spectral signatures of Ti3+ and Ti4+ in the LT phase. Finally, we discuss the IT spectrum. Since XRD studies suggest the existence of Ti3+ in IT phase as well as LT phase, the peak ˇ is predominantly due to the Ti3+ derived state. However, there is a definite enhancement of the peak ˇ compared to the LT spectrum. We attribute this enhancement to the screening from the small amount of the coherent state. Good agreement is obtained between the experiment and the calculation with the small V* (eg ) spectra. Another important advantage of HAXPES is that a deep core level can be studied, which is not possible with a conventional SX PES experiment, the Ti 1s core level is located at about 4966 eV BE, and can be probed only using hard X-rays. Furthermore, the 1s state is expected to possess a simpler shape, since the exchange splitting is expected to be negligible in Ti 1s spectra due to the much smaller overlap between 1s and 3d orbitals, as compared to the usually observed exchange splitting in 3s spectra of 3d transition metal compounds using SX PES. Fig. 4(b) shows the experimental Ti 1s PES spectra of TiN compared with the calculation. TiN is a good metal which exhibits a low-Tc superconductivity. The observed Ti 1s spectrum shows a clear double peak structure in the main peak. Note that this splitting is not due to the exchange splitting between 1s and 3d orbitals, but due to a poorly screened and well-screened peak. The broad satellite centered at 4979 eV is the conventional charge transfer satellite. We so far have shown that a clear correlation exists between the low BE satellites observed in the 2p HAXPES spectra and the coherent state near the Fermi level. In fact, such a coherent state was

246

M. Taguchi et al. / Journal of Electron Spectroscopy and Related Phenomena 190 (2013) 242–248

Mn 2p PES

Valence Band

x=0.55

Intensity (arb. units)

270 K

exp. 1

2

cal.

d L 10

dL

2

5

d

0

Intensity (arb. units)

320 K

x=0.4

x=0.2

1

dC 0

x=0

exp.

Binding Energy (eV) cal. Fig. 5. Valence band spectra of VO2 measured across the metal–insulator transition using hard X-ray compared with the cluster model calculations [34].

660

650

640

Binding Energy (eV) observed in valence band HAXPES spectra in some TM compounds [44,34]. As an example, Fig. 5 shows the experimental valence band of VO2 measured using HAXPES at 320 and 270 K [34]. The calculated spectra are also shown in Fig. 5 for the comparison. The valence band structure consists of the dominantly V 3d band near EF (0–2.5 eV BE) and a broad band usually considered as the dominantly O 2p band occurring between 3 and 10 eV BE in previous results. The 320 K spectrum shows a peak centered at 0.3 eV and a clear Fermi edge, indicating the metallic state, and another broad weak feature centered at about 1.5 eV. These features correspond to the coherent state at EF and the incoherent (lower Hubbard) state, respectively. The calculations reproduce the experimental spectra in the metallic and the insulating phases. All the parameters for the calculated spectra for the metallic and insulating phase are equal, except for the hybridization between the central V atom 3d orbitals and the coherent band, V* . V* = 0.48 eV for the metallic phase, and it was set to V* = 0 eV for the insulating phase. This confirms the importance of the coherent screening channel in reproducing the metallic phase of VO2 . In the metallic phase, the coherent state 3d1 C is the dominant near EF . On the other hand, the locally screened 3d0 has large intensity at a binding energy of ∼1 eV with weak contribution from 3d1 L stats. In insulating phase, the coherent screening states vanish and the locally screened 3d0 with 3d1 L admixture states remains around 1 eV. This feature corresponds to the effective lower Hubbard band, nicely matching the incoherent band in PES spectra. Interestingly, the main contribution of 3d1 L state appears around 4 eV in the metallic and insulating phases [34]. Another typical example is LaMnO3 with Sr (or Ba) doping, which shows remarkable difference between HAXPES and SX PES [29,33]. Fig. 6 shows Mn 2p core level spectra of LaMnO3 at 300 K measured with different Sr doping [29]. The Mn 2p3/2 HAXPES spectra show dramatic changes with hole doping. The changes observed in HAXPES Mn 2p3/2 spectra have not been observed in earlier SX PES spectra of bulk polycrystals, single crystals as well as thin films of manganites. The bulk-derived low BE or “well-screened” feature exhibits the following characteristics: the separation between the main peak and the well-screened feature increases with hole doping until x = 0.4, but reduces for x = 0.55. This behavior is similar to the physical properties, that is, with increasing x, the hole doping produces a ferromagnetic phase with increasing Tc and reduced resistivity until x = 0.4. On further hole doping, a magnetic transition

Fig. 6. Comparison between the cluster model calculations and HAXPES spectra of Mn 2p core level of La1−x Srx MnO3 [29].

from the ferromagnetic metal to the antiferromagnetic metal state is induced for x > 0.5. The comparison between the HAXPES spectra and the optimized calculations are also shown in Fig. 6. For all x values, the calculation reproduces well the intensity and position of the well-screened feature of HAXPES spectra. In these calculations, we have introduced new states at EF . These new states represent the doping induced states which develop into a metallic band at EF . The well-screened feature in the calculation is analyzed to originate from the 2p5 3d5 C configuration of the final state, and increases in intensity with increasing V* . 4.2.2. Charge-transfer type TM compounds In the charge-transfer type TM compounds, the O 2p band appears between the lower and upper Hubbard bands, and the insulating gap is usually formed between the O 2p and the upper Hubbard 3d band. In contrast to the early TM compounds, insulating NiO is a typical example of CT insulator while the high-Tc cuprates are CT insulators driven metallic by doping. It is well-known that these two compounds involve the formation of Zhang-Rice states positioned at the top of the valence band. They arise from a competition between O 2p – Ni (or Cu) 3d hybridization and the Ni (or Cu) on-site Coulomb interaction. First, we present experimental Ni 2p core level HAXPES and SX PES of NiO in Fig. 7 [35]. The core level spectra were normalized at the feature C, since it is well known that the feature C is insensitive to the surface preparation. In comparison with SX PES, the intensity of the feature A for bulk sensitive HAXPES was found to be enhanced by a factor of 1.35 with respect to the SX PES (see inset). Our experiments in Fig. 7 clearly show that the feature A was appreciably changed between the HAXPES and SX PES. In order to address the origin of the observed behavior, we used a cluster model and include a bound state to describe the spectra. Extended cluster model described in the previous subsection can be applied to insulating NiO by replacing a coherent band or doping induced state with Zhang-Rice bound state labeled ‘Z’. The parameter values (in eV) are summarized in Fig. 7. The calculated spectrum is shown in Fig. 7. The experimental features are reproduced well by

M. Taguchi et al. / Journal of Electron Spectroscopy and Related Phenomena 190 (2013) 242–248

Δ

∗ Δ V(eg ) V(eg ) 10Dq Udd Udc *

(a) Cu 2p3/2 PES

1.43 0.3 6.5 7.7 Rc Rv Parameter values

4.0 1.5 2.2

853

Ni 2p SX PES (hν = 1.25 keV) C HAXPES(hν =7.93 keV)

B a)

exp.

Intensity (arb. units)

Intensity (arb. units)

855

α'

10

NCCO 5

2p 3d

ZRS

9

β' LCO

b)

cal. 5

9

5

9

5

8

LSCO

2p 3d Z

940

2p 3d L

930

Binding Energy (eV) 880

870

860

850

Binding Energy (eV) Fig. 7. (a) Comparison between experimental Ni 2p core level HAXPES and SX PES. (b) Cluster model calculation for Ni 2p spectrum. Final-state components are also displayed [35].

the calculation over the whole energy range. To clarify the peak assignment, we show in the lower panel of Fig. 7 the various final state components of the core level spectra. It is clear that the dominant spectral weight in the feature A arises from the 2p5 3d9 Z state, while the contribution of the 2p5 3d9 L is dominant for the feature B. The satellite C is manly due to the 2p5 3d8 state. As briefly mentioned earlier, another example in which ZhangRice states play an important role is high Tc cuprates. Theoretical studies have predicted that the Zhang-Rice singlet (ZRS) should show a signature in Cu 2p PES of the cuprates. The Zhang-Rice singlet state is considered very important for superconductivity but has not been observed by conventional SX PES. Furthermore, while many Cu 2p core level SX PES of high Tc cuprates have been performed, the spectra show very little change upon doping. These puzzles bring into question the role of the depth sensitivity of PES. In an attempt to describe all the spectroscopic physics correctly, we carried out core level HAXPES of single crystal electron-doped Nd1.85 Ce0.15 CuO4 (NCCO), hole-doped La1.85 Sr0.15 CuO4 (LSCO) and LaCuO4 (LCO) [32]. Fig. 8(a) shows the Cu 2p3/2 HAXPES spectra and SX PES of electron-doped NCCO, LCO and LSCO. The HAXPES spectra of NCCO, LCO, and LSCO are clearly different and provided the following results: (i) The ZRS feature was observed for LCO. The LSCO spectrum showed clear changes compared to LCO. The ZRS feature is retained on hole-doping, but is weakened compared to LCO, and additional spectral weight is seen at higher BE (feature ˇ ). (ii) The HAXPES for NCCO shows a sharp low BE feature ˛ which is not observed in NCCO SX PES. Its energy position is different from the ZRS feature in LCO. (iii) The ‘˛ ’ and ‘ˇ ’ features arising from electron and hole doping in NCCO and LSCO, respectively, are clearly distinguished in HAXPES to have different BEs. The calculated results are shown in Fig. 8(b) with experiments of LSCO and NCCO. The calculations reproduce well the main peaks and satellite structure. It is known that the intensity ratio of the

Intensity (arb. units)

2p 3d 890

5

2p 3d L

HAXPES SX PES

A

0.9 0.9

857

247

(b) Cu 2p

NCCO LSCO exp.

cal. 960

950

940

930

Binding Energy (eV) Fig. 8. (a) Comparison between experimental Cu 2p HAXPES (solid line) and SX PES (line with symbols) for electron-doped NCCO, undoped LCO and hole-doped LSCO. (b) Model calculations for the Cu 2p core level PES of NCCO and LSCO (lower panel) compared with experiments (upper panel) [35].

well-screened peak to the poorly-screened satellite is determined by the effective  [45]. The sharp peak at low BE in NCCO originates from core hole screening by doping induced states at EF , the 2p5 3d10 Z state. The obtained parameter values show small differences for LSCO and NCCO. The most important parameter is * , which represents the energy difference between the upper Hubbard band and doping induced states. The small values of * for NCCO (=0.25 eV) indicates that the doping induced states lie just below the upper Hubbard band, whereas a large * (=1.35 eV) of LSCO describes the situation for doping induced states lying near the top of the valence band, with the upper Hubbard band separated by * . 5. Concluding remarks In conclusion, we have discussed a few recent applications of HAXPES carried out at the RIKEN beamline BL29XU in SPring-8. HAXPES studies on the recoil effect of photoelectrons in core levels and valence band states are consistently explained with theoretical results indicating the role of phonon excitations. HAXPES studies of a series of 3d TM compounds show an additional well-screened feature in bulk sensitive electronic structure compared to surface sensitive results. The extended cluster model explains this wellscreened feature for a series of TM compounds. These results show that HAXPES is a valuable tool for the study of doping and temperature dependent phase transitions of strongly correlated systems with tremendous potential for future activities.

248

M. Taguchi et al. / Journal of Electron Spectroscopy and Related Phenomena 190 (2013) 242–248

Acknowledgements We sincerely thank all our collaborators of HAXPES studies overviewed in this report. We acknowledge the support of Dr. Ishikawa and the X-ray optics group of RIKEN. We thank Dr. Matsunami, Dr. K. Horiba, Dr. R. Eguchi, Dr. T. Takeuchi, Prof. S. Shin, Prof. H. Takagi, Prof. K. Takenaka, Prof. Y. Kayanuma, and Prof. S. Tanaka, for fruitful collaborations. This work was partially supported by KAKENHI (no. 20540324). References [1] E. Sokolowski, C. Nordling, K. Siegbahn, Phys. Rev. 110 (1958) 776. [2] S. Hagstrom, C. Nordling, K. Siegbahn, Phys. Lett. 9 (1964) 235. [3] C.S. Fadley, S.B. Hagstrom, J.M. Hollander, M.P. Klein, D.A. Shirley, Science 157 (1967) 1571. [4] E.B. Karlsson, H. Siegbahn, Nucl. Instrum. Meth. Phys. Res. A 601 (2009) 1. [5] J. Danger, P. Le Fevre, H. Magnan, D. Chandesris, S. Bourgeois, J. Jupille, T. Eickhoff, W. Drube, Phys. Rev. Lett. 88 (2002) 243001. [6] J.C. Woicik, E.J. Nelson, L. Kronik, M. Jain, J.R. Chelikowsky, D. Heskett, L.E. Berman, G.S. Herman, Phys. Rev. Lett. 89 (2002) 077401. [7] K. Kobayashi, M. Yabashi, Y. Takata, T. Tokushima, S. Shin, K. Tamasaku, D. Miwa, T. Ishikawa, H. Nohira, T. Hattori, Y. Sugita, O. Nakatsuka, A. Sakai, S. Zaima, Appl. Phys. Lett. 83 (2003) 1005. [8] S. Thiess, C. Kunz, B.C.C. Cowie, T.-L. Lee, M. Renier, J. Zegenhagen, Solid State Commun. 132 (2004) 589. [9] C. Dallera, L. Duo, L. Bricovich, G. Panaccione, G. Paolicelli, B. Cowie, J. Zegenhagen, Appl. Phys. Lett. 85 (2004) 4532. [10] E. Holmstrom, W. Olovsson, I.A. Abrikosov, A.M.N. Niklasson, B. Johansson, M. Gorgoi, O. Karis, S. Svensson, F. Schafers, W. Braun, G. Ohrwall, G. Andersson, M. Marcellini, W. Eberhardt, Phys. Rev. Lett. 97 (2006) 266106. [11] G.H. Fecher, A. Gloskovskii, K. Koch, J. Barth, B. Balke, C. Felser, F. Schafers, M. Mertin, W. Eberhardt, S. Mahl, O. Schaff, J. Electron Spectrosc. Relat. Phenom. 156–158 (2007) 97. [12] A. Yamasaki, S. Imada, H. Higashimichi, H. Fujiwara, T. Saita, T. Miyamachi, A. Sekiyama, H. Sugawara, D. Kikuchi, H. Sato, A. Higashiya, M. Yabashi, K. Tamasaku, D. Miwa, T. Ishikawa, S. Suga, Phys. Rev. Lett. 98 (2007) 156402. [13] A. Walsh, J.L. Da Silva, S-H. Wei, C. Korber, A. Klein, L.F. Piper, A. DeMasi, K.E. Smith, G. Panaccione, P. Torelli, D.J. Payne, A. Bourlange, R.G. Egdell, Phys. Rev. Lett. 100 (2008) 167402. [14] E.M.J. Johansson, M. Odelius, M. Gorgoi, O. Karis, R. Ovsyannikov, F. Schafers, S. Svensson, H. Siegbahn, H. Rensmo, Chem. Phys. Lett. 464 (2008) 192. [15] M. Sing, G. Berner, K. Goß, A. Müller, A. Ruff, A. Wetscherek, S. Thiel, J. Mannhart, S.A. Pauli, C.W. Schneider, P.R. Willmott, M. Gorgoi, F. Schäfers, R. Claessen, Phys. Rev. Lett. 102 (2009) 176805. [16] M. Takizawa, Y. Hotta, T. Susaki, Y. Ishida, H. Wadati, Y. Takata, K. Horiba, M. Matsunami, S. Shin, M. Yabashi, K. Tamasaku, Y. Nishino, T. Ishikawa, A. Fujimori, H.Y. Hwang, Phys. Rev. Lett. 102 (102) (2009) 236401. [17] C.S. Fadley, Nucl. Instrum. Meth. Phys. Res. A 601 (2009) 8. [18] S. Granroth, W. Olovsson, E. Holmstrom, R. Knut, M. Gorgoi, S. Svensson, O. Karis, J. Electron Spectrosc. Relat. Phenom. 183 (2011) 80. [19] A.Y. Gray, C. Papp, S. Ueda, B. Balke, Y. Yamashita, L. Plucinski, J. Minar, J. Braun, E.R. Ylvisaker, C.M. Schneider, W. Pickett, H. Ebert, K. Kobayashi, C.S. Fadley, Nat. Mater. 10 (2011) 759. [20] G. Panaccione, K. Kobayashi, Surf. Sci. 606 (2012) 125. [21] A.X. Gray, D.W. Cooke, P. Krüger, C. Bordel, A.M. Kaiser, S. Moyerman, E.E. Fullerton, S. Ueda, Y. Yamashita, A. Gloskovskii, C.M. Schneider, W. Drube, K. Kobayashi, F. Hellman, C.S. Fadley, Phys. Rev. Lett. 108 (2012) 257208.

[22] C. Schlueter, P. Orgiani, T.-L. Lee, A.Yu. Petrov, A. Galdi, B.A. Davidson, J. Zegenhagen, C. Aruta, Phys. Rev. B 86 (2012) 155102. [23] 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, K. Kobayashi, Nucl. Instrum. Meth. Phys. Res. A 547 (2005) 50. [24] T. Ishikawa, K. Tamasaku, M. Yabashi, Nucl. Instrum. Meth. Phys. Res. A 547 (2005) 42. [25] C.P. Flynn, Phys. Rev. Lett. 37 (1976) 1445, W. Domcke, L.S. Cederbaum, J. Electron Spectrosc. Relat. Phenom., 13 (1978) 161. [26] 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, T. Ishikawa, Phys. Rev. B 75 (2007) 233404. [27] 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, T. Ishikawa, Phys. Rev. Lett. 101 (2008) 137601. [28] A. Sekiyama, T. Iwasaki, K. Matsuda, Y. Saitoh, Y. Onuki, S. Suga, Nature (London) 403 (2000) 396. [29] K. Horiba, M. Taguchi, A. Chainani, Y. Takata, E. Ikenaga, D. Miwa, Y. Nishino, K. Tamasaku, M. Awaji, A. Takeuchi, M. Yabashi, H. Namatame, M. Taniguchi, H. Kumigashira, M. Oshima, M. Lippmaa, M. Kawasaki, H. Koinuma, K. Kobayashi, T. Ishikawa, S. Shin, Phys. Rev. Lett. 93 (2004) 236401. [30] N. Kamakura, M. Taguchi, A. Chainani, Y. Takata, K. Horiba, K. Yamamoto, K. Tamasaku, Y. Nishino, D. Miwa, E. Ikenaga, M. Awaji, A. Takeuchi, H. Ohashi, Y. Senba, H. Namatame, M. Taniguchi, T. Ishikawa, K. Kobayashi, S. Shin, Europhys. Lett. 68 (2004) 557. [31] M. Taguchi, A. Chainani, N. Kamakura, K. Horiba, Y. Takata, M. Yabashi, K. Tamasaku, Y. Nishino, D. Miwa, T. Ishikawa, S. Shin, E. Ikenaga, T. Yokoya, K. Kobayashi, T. Mochiku, K. Hirata, K. Motoya, Phys. Rev. B 71 (2005) 155102. [32] M. Taguchi, A. Chainani, K. Horiba, Y. Takata, M. Yabashi, K. Tamasaku, Y. Nishino, D. Miwa, T. Ishikawa, T. Takeuchi, K. Yamamoto, M. Matsunami, S. Shin, T. Yokoya, E. Ikenaga, K. Kobayashi, T. Mochiku, K. Hirata, J. Hori, K. Ishii, F. Nakamura, T. Suzuki, Phys. Rev. Lett. 95 (2005) 177002. [33] H. Tanaka, Y. Takata, K. Horiba, M. Taguchi, A. Chainani, S. Shin, D. Miwa, K. Tamasaku, Y. Nishino, T. Ishikawa, E. Ikenaga, M. Awaji, A. Takeuchi, T. Kawai, K. Kobayashi, Phys. Rev. B 73 (2006) 094403. [34] R. Eguchi, M. Taguchi, M. Matsunami, K. Horiba, K. Yamamoto, Y. Ishida, A. Chainani, Y. Takata, M. Yabashi, D. Miwa, Y. Nishino, K. Tamasaku, T. Ishikawa, Y. Senba, H. Ohashi, Y. Muraoka, Z. Hiroi, S. Shin, Phys. Rev. B 78 (2008) 075115. [35] M. Taguchi, M. Matsunami, Y. Ishida, R. Eguchi, A. Chainani, Y. Takata, M. Yabashi, K. Tamasaku, Y. Nishino, T. Ishikawa, Y. Senba, H. Ohashi, S. Shin, Phys. Rev. Lett. 100 (2008) 206401. [36] M. Taguchi, A. Chainani, M. Matsunami, R. Eguchi, Y. Takata, M. Yabashi, K. Tamasaku, Y. Nishino, T. Ishikawa, S. Tsuda, S. Watanabe, C.-T. Chen, Y. Senba, H. Ohashi, K. Fujiwara, Y. Nakamura, H. Takagi, S. Shin, Phys. Rev. Lett. 104 (2010) 106401. [37] P.A. Bhobe, A. Chainani, M. Taguchi, T. Takeuchi, R. Eguchi, M. Matsunami, K. Ishizaka, Y. Takata, M. Oura, Y. Senba, H. Ohashi, Y. Nishino, M. Yabashi, K. Tamasaku, T. Ishikawa, K. Takenaka, H. Takagi, S. Shin, Phys. Rev. Lett. 104 (2010) 236404. [38] R. Cowan, The Theory of Atomic Structure and Spectra, University of California Press, Berkeley, 1981. [39] G. van der Laan, C. Westra, C. Haas, G.A. Sawatzky, Phys. Rev. B 23 (1981) 4369. [40] O. Gunnarsson, K. Schonhammer, Phys. Rev. B 28 (1983) 4315. [41] A. Fujimori, F. Minami, Phys. Rev. B 30 (1984) 957. [42] J. Zaanen, G.A. Sawatzky, J.W. Allen, Phys. Rev. Lett. 55 (1985) 418. [43] S. Lakkis, C. Schlenker, B.K. Chakraverty, R. Buder, M. Marezio, Phys. Rev. B 14 (1976) 1429. [44] G. Panaccione, M. Altarelli, A. Fondacaro, A. Georges, S. Huotari, P. Lacovig, A. Lichtenstein, P. Metcalf, G. Monaco, F. Offi, L. Paolasini, A. Poteryaev, O. Tjernberg, M. Sacchi, Phys. Rev. Lett. 97 (2006) 116401. [45] T. Böske, O. Knauff, R. Neudert, M. Kielwein, M. Knupfer, M.S. Golden, J. Fink, H. Eisaki, S. Uchida, K. Okada, A. Kotani, Phys. Rev. B 56 (1997) 3438.