Photoelectron spectroscopy of transition metal oxide interfaces

Transcription

1 Eur. Phys. J. Appl. Phys. (2015) 70: DOI: /epjap/ Review Article THE EUROPEAN PHYSICAL JOURNAL APPLIED PHYSICS Photoelectron spectroscopy of transition metal oxide interfaces Jörg Zegenhagen a Diamond Light Source Ltd., Harwell Science and Innovation Campus, Didcot OX11 0DE, UK Received: 5 February 2015 / Received in final form: 23 March 2015 / Accepted: 14 April 2015 Published online: 21 May 2015 c EDP Sciences 2015 Abstract. In the present paper we review applications of the photoelectron spectroscopy (PES) technique to the investigation of transition metal oxide (TMO) interfaces. We summarize very briefly some of the principle, specific characteristics of TMOs. Because of the buried nature of the interfaces, the photoelectrons must penetrate certain thicknesses of material, which is easier with higher kinetic energies in the kev range. Thus, we also briefly summarize some of the hallmarks of hard X-ray photoelectron spectroscopy (HAXPES), before presenting four explicit samples of the analysis of TMO interfaces: The LaAlO 3/SrTiO 3(0 0 1) interface, which had attracted attention because of the discovery of a sheet of high mobility electrons below the thin layer of LaAlO 3; superlattices of the two insulators CaCuO 2 and SrTiO 3, which can be prepared to become superconducting at about 40 K; epitaxial films of the 90 K superconductor GdBa 2Cu 3O 7 δ on NdGaO 3; and finally the analysis of the nucleation of the 90 K superconductor YBa 2Cu 3O 7 δ on SrTiO 3(0 0 1). The latter two cases of the investigation of the superconducting films are examples of photoelectron spectroscopy by X-ray standing wave (XSW) excitation. Because of this, the bare essential features of the XSW technique are also briefly reviewed. 1 Introduction With the discovery of superconductivity in the oxygen deficient Ba-La-Cu-O compound significantly above previously established transition temperatures in 1986 by Bednorz and Müller [1], the community interested in transition metal oxides (TMOs) suddenly widened considerably. This community was further enlarged by surface scientists when it was realized that high temperature superconductors (HTSs), epitaxially grown on transition metal perovskites, represent a rich playground for fundamental and applied research [2]. It was realized early on that epitaxy allowed to tailor the properties of the HTSs such as their critical current density [3]. More recently, with the discovery of a conducting sheet of electrons at the interface of two insulating TMOs, i.e., SrTiO 3 (STO) and LaAlO 3 (LAO) [4], TMO interfaces have attracted further strong interest and became subject of intense research activities. Even 10 years after this discovery, and despite a wealth of studies, all aspects of the mechanism which gives rise to the (almost) twodimensional (2D) electron system of high mobility are not fully understood. The fact that this 2D electron gas develops only when the LAO is deposited on one type of the STO(0 0 1) termination and only if the LAO layer exceeds a critical thickness of about four unit cells suggested from the beginning an intrinsic, non-trivial mechanism. Electron transfer to the interface is thought to be a jorg.zegenhagen@diamond.ac.uk driven by the electrostatic field, building up in the polar LAO with increasing thickness and which would be diverging unless the interfacial charges of the LAO are compensated. Other possible extrinsic origins of the conductivity that had been discussed such as oxygen vacancies in the STO, or intermixing at the interface have meanwhile been excluded of being the principle culprits for the formation of the sheet conductance. In the course of the research activities stimulated by the 2D gas at the STO/LAO interface, rich physics in terms of novel electronic reconstructions and phases have been observed at TMO interfaces (for a review, see e.g., [5]). Thus, there are legitimate expectations that TMO interfaces may represent a fertile basis for novel electronic devices, intimately connected, however, to the already rich physics of TMOs. For a short review on this matter the reader is referred to a recently published paper by Mannhart and Schlom [6]. As semiconductors, mostly of the p-block i.e., column III VI of the periodic system, have been undoubtedly the electronic materials of the 20th century, transition metal oxides (TMOs) and their interfaces may well become the electronic materials of the 21st century [7]. Different from semiconductors, where the bulk electronic properties are meanwhile well understood and well described by existing theoretical tools based on single particle models, transition metal oxides display a wealth of properties, many of which are not well described by any theory and thus not well understood in their complexity at present. The bonding character of TMOs is neither covalent nor ionic. Most of the transition metals are characterised by several p1

2 The European Physical Journal Applied Physics Ti AOs 6 4p 4s 3d 8- *, * * 3a 4t * 3 * 2 nb nb u 1g e g t 2g t 1g t 2u t 1u t 2g t 1u e g a 1g e g t 1u a 1g 2p 2p 2s a wide band-gap of 4.3 ev [11]. Reasonable agreement with the experimental finding was reported in 1990 by Svane and Gunnarsson [12] using a self-consistent version of the self-interacting-corrected local spin density approximation of the spin density functional formalism. For some early and some recent discussion on the issue of NiO see Merlin [13] and Yin et al. [14], respectively. While refined calculations are meanwhile able to describe quite well the bandstructure for TMOs such as NiO [15], they still fail predicting or even explaining many of their properties. High temperature superconductivity of the various copper oxide compounds is a prominent case in this respect [16]. With the discovery of the pn junction in the 1930ies at Bell Labs by Russell Shoemaker Ohl [17], in semiconductor research the focus had shifted very early from the bulk material to interfaces. In fact, as Herbert Kroemer phrased it almost a quarter of a century later in his Nobel lecture on December 8, 2000:... it may be said that the interface is the device. A similar development appears to be happening in the field of TMOs. With the bulk properties of the parent compounds not even fully understood, interest is shifting to the interfaces of TMO materials. 2 Hard X-ray photoelectron spectroscopy a 1g( ) e( g ) t ( ) t ( ) 2g 3 t (nb) 1g 3 t (nb) 2u 3 1u 3 t ( ) 1u Fig. 1. Molecular orbitals for an octahedral oxygen coordinated tetravalent transition metal atom such as Ti. The corresponding symmetry of the orbitals is indicated in the lower part of the figure. AOs means atomic orbitals; nb means nonbonding. After [9] and[10] (taken from Ref. [18]). oxidation states and TMOs thus display a very complex defect chemistry. Since the s-electrons of the transition metal are typically transferred to the oxygen, the electronic properties of TMOs are largely determined by the narrow band of highly correlated d-electrons of the metal [8]. It is the high degree of correlation that leads to the break-down of classical mean-field theories and density functional calculations. The scheme of the molecular orbitals [9] and their symmetry [10] for a sixfold coordinated oxygen coordinated Ti 4+ such as in TiO 2 or STO is shown in Figure 1. Already the comparably simple TMO NiO is a prominent illustrative case for which classical band-structure calculations spectacularly failed. With its partly filled d-band, NiO is by classical theory predicted to be a conductor, but experiment shows it to be an insulator with 3 Photoelectron spectroscopy (PES) is a widely used key technique for the characterization of chemical and electronic structure of materials. However, the pronounced surface sensitivity, due to the strong scattering of electrons, limits its applicability and the study of bulk samples and buried interfaces is not straight forward. Bulk sensitivity can be achieved by detecting photoelectrons with kinetic energies either in the electron-volt or in the several kilo-electron-volt range (cf. Fig. 2). For low energy PES, the excellent features of optical lasers can be employed as excitation source. Because of the low excitation energy, the spectroscopy is limited to the conduction and valence band whereas core levels cannot be accessed. Thus chemical information cannot simply be obtained. How to use hard X-rays in order to obtain chemical information from larger depth was demonstrated by Dallera et al. [19] 10 years ago, coining this technical approach hard X-ray photoelectron spectroscopy (HAXPES). The main results of this study in terms of the measured effective attenuation length (EAL) 1 of electrons is shown in Figure 2. Because the photoelectrons are scattered inelastically largely by valence and conduction electrons, the EAL is only weakly material dependent. Utilizing hard X-rays for PES is not really new. Already Siegbahn and coworkers used initially hard 1 Instead of using the older term electron mean free path, it is more meaningful to use the term effective attenuation length, which takes into account that the path of electrons from the source to the detection point is increased by elastic range straggling. It is important to note in this context that the cross section for elastic scattering, responsible for the range straggling, decreases with increasing electron kinetic energy ( kev), thus helping to increase the EAL p2

3 J. Zegenhagen: Photoelectron spectroscopy of transition metal oxide interfaces Fig. 2. Symbols: the measured effective attenuation length of photoelectrons in GaAs, yielding EAL GaAs Ee-kin 0.85 (taken from Ref. [19]). X-rays from classical anodes for their pioneering ESCA (electron spectroscopy for chemical analysis) studies in the 60s [20]. They used sources such as Mo K α Cu K α simply because they were available, before switching to lower excitation energies and profiting from higher cross sections and better energy resolution. Later, Lindau and coworkers demonstrated that it is realistic to use harder excitation energies when employing synchrotron radiation [21]. Nevertheless, routine bulk-sensitive PES experiments using hard X-rays were long deemed unattractive because photo cross sections σ decrease very strongly ( Eγ 3 ) with increasing X-ray energy E γ, whereas the EAL increases only slowly with electron kinetic energy, roughly proportional to the kinetic energy E e-kin of the photoelectrons as shown in Figure 2. As a rule of thumb, above E e-kin 0.5 kev, the EAL is roughly proportional to E e-kin with EAL/E e-kin 1 nm/kev. Meanwhile, the brilliant flux of modern synchrotron radiation sources renders HAXPES experiments routinely possible and very attractive for investigating properties of bulk materials, buried objects and interfaces. It has been demonstrated that chemical information can be retrieved from depths exceeding 50 nm when exciting photoelectrons with a kinetic energy of about 15 kev [22]. The geometry for photoelectron spectroscopy measurements discussed in this review is shown schematically in Figure 3. To maximize the dipole matrix element M D and thus the electron emission, the photoelectrons are detected along the polarization direction of the X-ray beam. The directional dependence of the emission is governed by the dipole matrix element M D 1+βP 2 (e 0 n p ), where β is the dipole asymmetry parameter, 2 P 2 (z) = (3z 2 1)/2 is the second order Legendre polynomial, and e 0 and n p are unit vectors characterizing the polarization of the X-ray beam and the direction of the emitted 2 Typically, β 2. Fig. 3. Schematic geometry for a hard X-ray photoelectron spectroscopy (HAXPES) experiment. The angle of incidence of the X-ray beam with the surface is α γ and the emission angle of the photoelectron with the surface is θ. To maximize the dipole matrix element and thus the electron emission, the analyzer is pointing into the polarization direction of the X-ray beam. photoelectron, respectively. It should be noted that matrix element effects are rendering the photo cross section not only strongly dependent on energy but also on the specific shape of the electron wavefunction. This was for instance demonstrated by Thiess et al. [23] as shown in Figure 4. Thus, by choice of excitation energy, the emission from certain electronic levels can be enhanced or suppressed. In the HAXPES regime, when necessary, the probed depth in the material z = d sin(θ), where d is the distance the photoelectron travels, can conveniently be adjusted (decreased) by the escape angle θ of the photoelectrons. The hard X-ray excited inelastically scattered photoelectrons travel large distances before being completely thermalized. Thus they are coming from large depth exceeding 1 μm, and it may be beneficial to decrease the angle of incidence of the X-rays α γ and thus the X-ray penetration depth below the escape depth of these electrons to minimize this inelastic background. In order not to increase the footprint of the X-ray beam beyond the acceptance of the analyzer optics (typically 50 μm) this requires a very small, focused beam, nowadays available from third generation synchrotron radiation sources using state-of-the-art focusing optical components such as mirrors or lenses. 3 Last but not least because of the advancement of nanoscience and the increasing interest in interfaces, HAXPES has matured rapidly to an established technique. Following the proceedings of the first HAXPES conference [24], meanwhile several reviews have been published [25, 26], including a short review addressing HAXPES investigations of TMO thin films and interfaces [27]. It is also worthwhile to mention that very high energy resolution can be obtained by using hard X-rays owing to the fact that the energy resolution, achievable with Bragg monochromators, can be in the mev range or better at X-ray pass energies of the order of kev [28,29]. 3 Optical components in the X-ray regime work largely analogous to known optical components, keeping in mind that the refractive index of materials is smaller than unity for X-rays in contrast to the situation in the visible range p3

4 The European Physical Journal Applied Physics LaAlO 3 /SrTiO 3 (001) interface Formal valence: La : 3+ Sr: 2+ Al : 3+ Ti : 4+ O : 2- O : 2- SrO : 0 TiO 2 : 0 LaO : 1+ AlO 2 : 1- Fig. 5. Model of the STO/LAO interfaces and formal charges oftheoxidelayers. Fig. 4. Shallow core levels and valence band of Au recorded at 8.0 kev and 14.5 kev with energy resolutions of ΔE =1.0 ev and ΔE = 1.9 ev, respectively. Higher angular momentum states are suppressed at higher excitation energies (taken from Ref. [23]). 3 The STO/LAO interface As mentioned in the introduction, the quasi twodimensional electron gas (2DEG) [4] at the(001) interface of the band insulators STO (bandgap 3.3 ev) and LAO (bandgap 5.6 ev) has attracted considerable attention [30, 31]. Both materials are cubic perovskites above phase transition temperatures of T 105 K [32] and T 800 K [33] for STO and LAO, respectively. At the transition temperature they undergo a second order phase transition to a low temperature tetragonal and rhombohedral phase, respectively, which can both be viewed as slightly distorted cubic structures. Their cubic lattice constants match closely with 0.39 and 0.38 nm for STO and LAO, respectively and LAO grows with good epitaxially quality on STO. There are two possible structures of the interface, depending on the termination of the STO crystal. The two possible interfaces for LAO on STO(0 0 1) are schematically shown in Figure 5. Of the SrO/AlO 2 and the TiO 2 / LaO interfaces, only the latter gives rise to the 2D electron gas. In the (0 0 1) direction, LAO is strongly polar with the LaO and AlO 2 planes formally carrying a charge of +e 0 and e 0, respectively, with e 0 being the elementary charge while the SrO and TiO 2 planes of STO are formally Fig. 6. The Ti 2p Ti 4+ peak and the Ti 3+ satellite structure for LAO films on STO(0 0 1) of different thickness in terms of the number of LAO unit cells (UC) (taken from Ref. [40]). uncharged. 4 To fulfil charge neutrality, the LAO film will be terminated by LaO and AlO2 layers on its opposing faces with opposite charges. Equivalent to a plane-plate capacitor with a dielectric medium between the charged plates, the electrical potential is diverging as the LAO thickness increases. A scenario referred to as polar catastrophe [35]. Once the potential drop across the LAO layer reaches the value of the effective band gap, which happens for a LAO thickness of about four unit cells (UC), intrinsic doping of the interface is thought to be realized [4]. 4 Calculations suggest that the real valence charges of Sr, Ti,andOare+1.58e 0, +2.18e 0 and 1.26e 0, respectively rendering the STO(0 0 1) direction weakly polar with the TiO 2 and SrO planes carrying a charge of 0.32e 0 and +0.32e 0, respectively [34] p4

5 J. Zegenhagen: Photoelectron spectroscopy of transition metal oxide interfaces (a) (b) (c) Fig. 7. Schematic of CCO/STO SLs. The STO is non-polar in the (0 0 1) direction whereas the Ca and CuO 2 layers in CaCuO 2 carry alternating net charges σ = ±2e 0. The resulting electric field E and electrostatic potential V are indicated. (a) The unreconstructed interfaces which leads to the build-up of electrostatic potential diverging with thickness, (b) the reconstructed interfaces; the built-up of the potential is suppressed by a reconstruction mechanism: on average 1/2 oxygen atom per unit cell is transferred from the SrO plane to the Ca plane, (c) the inclusion of an extra amount δ of oxygen at the interfaces slightly decreases the positive net charge by 2δ at the CaO and SrO planes, which can be compensated by 2δ holes in the adjacent CuO 2 planes (taken from Ref. [59]). Electrons are transferred from oxygen vacancies on the LAO surface. Their formation energy is lowered as the electric potential builds up with increasing LAO film thickness [36,37]. These electrons (1/2 e 0 per surface unit cell i.e., cm 2 ) occupy the empty Ti 3d band states at the Fermi level (cf. Fig. 1). Consequently, the Ti 4+ 3p photoelectron peak should be accompanied by a Ti 3+ satellite at lower binding energy. A combined transmission electron microscopy and electron energy loss spectroscopy study [38] found indeed evidence of the LAO surface being positively charged because of the electron transfer from O 2p at the top of the valence band at the LAO surface to the Ti 3d conduction band at the interface for charge compensation. A field related band bending in the LAO and associated core level shifts have been observed [39], but with a value of 300 mev the magnitude of the shift was about ten-fold smaller than expected. The Ti 3+ states, owing to the partially filled Ti d-band, had been observed earlier by hard X-ray photoelectron spectroscopy (HAXPES) [40] (cf. Fig. 6) and more recently by Chu et al. [41]. Both studies found the Ti 3+ states to be distributed over a few nm in the STO at the interface. As can be seen in Figure 6, some Ti 3+ states are observed even for LAO thickness of less than four unit cells, in which case the interface should not be conducting. These residual Ti 3+ states are suggested to arise from oxygen vacancies close to the interface. At the common growth conditions for the STO/LAO heterostructure, the formation of oxygen vacancies is not expected. Their appearance may be related to the fact that external fields are reported to create charge carriers in STO (with 2D like behavior) [45]. In this context, it should be mentioned that STO can be made n-type conducting by doping e.g. with La, Nb, or oxygen deficiencies [46]. When doped by oxygen deficiencies, besides a coherent (i.e., delocalized) state close to the Fermi level, a broad incoherent (i.e., localized) state is observed, almost filling the whole 3.2 ev band gap [47 49]. The density of states expected to show up at E F was first observed in the form of weak intensity in the band gap, reported in a resonant soft X-ray photo electron spectroscopy study [42, 43]. A very recent polarization controlled, angle resolved photoelectron spectroscopy study of the STO/LAO interface [44] was successful in mapping its electronic structure and concluded that different Ti 3d orbitals are involved in the 2DEG and the mobile carriers observed in transport measurement. The formation of such a 2DEG could be shown for other polar oxide overlayer than LAO [50, 51], but only reports were STO has been used as substrate have been published so far. Hence there is the suspicion that the 2DEG at the interface is intrinsically related to some specific properties of STO. Depending on preparation conditions, the observed carrier densities differ and are in most cases much smaller than the theoretically predicted 1/2 e 0 per surface unit cell (= /cm 2 )[52], which suggests some other charge compensation mechanism being active. 4 Superconducting TMO interfaces The LAO/STO interface is reported to become superconducting at about 200 mk [31], which is the temperature at which superconductivity is also observed in oxygen deficient STO [53,54]. However, in tailored cuprate superlattice (SL) heterostructures of CaCuO 2 (CCO) and BaCuO 2 [55,56], superconductivity is observed with transition temperature beyond liquid nitrogen temperature p5

6 The European Physical Journal Applied Physics Both materials on its own are not superconducting but insulating and metallic, respectively. Superconductivity has also been reported in bilayers consisting of the insulator La 2 CuO 4 and the metal La 1.55 Sr 0.45 CuO 4 [57] with at c of about 50 K and most recently, superconductivity was observed in SL heterostructures of the two insulators CaCuO 2 and SrTiO 3, grown on STO(0 0 1), with a maximum critical temperature of about 40 K [58]. The superconductivity is thought to be associated with a reconstruction at the CCO/STO interface, by a mechanism that shows strong similarities with the LAO/STO case. The CCO/STO SL and the corresponding electronic situation is schematically depicted in Figure 7. For CCO on STO when epitaxially grown on STO(0 0 1), and in the (0 0 1) direction of the STO, the CCO exhibits alternatingcaandcuo 2 layers with formal charges of +2e 0 and 2e 0, respectively. As shown schematically in Figure 7 this leads to an increasing, i.e., diverging electrostatic potential, analogous to the LAO/STO situation. Since the individual Ca and CuO 2 layers carry twice the charge (±2e 0 ) compared to the charge of the LaO and AlO 2 layers in LAO (±1e 0 ), the potential is even more rapidly diverging. Depending on the details of the growth procedure, specifically the oxidizing conditions, superconducting or non-superconducting CCO/STO superlattices can be produced. Both types of CCO/STO SL heterostructures were investigated recently by HAXPES by Aruta et al. [59]. We will reproduce here in the following the HAXPES results obtained for the two different kinds of CCO/STO SLs, each with 20 supercells consisting of three to four unit cells (UC) of CCO and two UC of STO. We will only summarize the main conclusions of that study. For more details the reader is referred to the original publication, were the HAXPES results are comprehensively discussed. Similar to the widely favored scenario for the LAO/ STO interface, the suppression mechanism of the polarity induced electrostatic potential for CCO on STO is thought to be based on an electronic reconstruction involving charge transfer between the CCO and STO blocks, accompanied, however, by a much more pronounced structural rearrangement. Figure 8 shows resistivity and HAXPES data for non-superconducting and superconducting STO/CCO as well as HAXPES data for STO and CCO. In the spectrum of the superconducting SL a shift of about 1.5 to 2.0 ev for the Ti (and Sr, not obvious in the inset) core levels toward lower binding energies with respect to STO can be observed, as emphasized in Figure 8b. In contrast, no core level shifts are observed for the Ca (and Cu, not shown) core levels of the CCO film with respect to the Ca and Cu core levels of the two SLs. The valence band spectra of STO and CCO and the deduced band alignment of STO and CCO is shown in Figures 9a and9b. With the information from valence band and the core level from Figures 8, 9a and9b, the band alignment for the SLs shown in Figures 9cand9dcan be calculated based on the assumption that any change in (a) (c) Fig. 8. (a) Resistivity as a function of temperature of the superconducting CCO/STO SL (closed red circles) and non-superconducting SL (open black circles), for which the spectra below are taken, (b) HAXPES spectra showing on a magnified energy scale the indicated region from the survey spectra below. The Ti 3p core-levels of the nonsuperconducting and superconducting CCO/STO SLs show characteristic shifts with respect to STO, (c) survey spectra of a non-superconducting CCO/STO SL (open black circles), a superconducting CCO/STO SL (closed red circles), an STO(0 0 1) substrate crystal (closed blue triangles), and a thick CCO film (open green triangles) grown on STO(0 0 1) under the same conditions as the superconducting CCO/STO SL recorded for an X-ray excitation energy E γ =2.8 kev and electron emission angle θ =70 (from Ref. [59]). the valence band alignment must be accompanied by a rigid shift of the core levels 5 [60]. The authors conclude that the observed band alignment does not allow any direct charge transfer between the bands of CCO and STO [61] and that the build-up of the polar potential must be inhibited by other means. They argue that the build-up of the electrostatic potential is prevented by a oxygen redistribution, as in the case of the interface between SrO-terminated STO and LAO, where the SrO interface plane is assumed to be depleted of one oxygen ion per two unit cells [35]. 5 The valence band offset ΔE v = (Ev CCO Ev STO ) SL = (Ev CCO Ev STO )+[(Ecl CCO Ecl STO ) SL (Ecl CCO ECL STO )], where E cl and E v denote the binding energy of the core level and the valence band edge, respectively. The subscript SL indicates the corresponding binding energies in the superlattice. For more details the reader is referred to the original publication by Aruta et al. (b) p6

7 J. Zegenhagen: Photoelectron spectroscopy of transition metal oxide interfaces (a) (b) (a) (b) (c) (d) (e) (d) Fig. 9. (a) Valence-band spectra for CCO and STO. The edge of the valence band spectra is obtained by linear extrapolation, (b) (d) schematic of the alignment of the valence and conducting bands: for the CCO and STO reference samples shown in (b), for the non-superconducting SL shown in (c), and for the superconducting SL shown in (d). Each additional oxygen at the interface of the superconducting SL creates two holes in the valence band of CCO, which is shown schematically in (d). The bulk band gap values for STO and CCO are taken as 3.3 ev [85] and 1.5 ev[86]), respectively (taken from Ref. [59]). The Ca and Sr interfacial planes are supposed to accommodate a variable quantity of oxygen ions, although not necessarily the same amount, i.e., TiO 2 Ca x CuO 2 and CuO 2 SrO y TiO 2 at the lower and upper interface, respectively. With x y 0.5 the interfaces would be charge neutral and the build-up of the electrostatic potential would be suppressed. This is schematically shown in Figure 7b. At this stage, the SL is not superconducting yet. This happens only when the film is deposited under highly oxidizing conditions and additional oxygen is accommodated at the interfaces in the CaO x and SrO y layer. In this case, charge neutrality is preserved by additional electron transfer, leaving two holes in the valence band of CCO for each extra oxygen, rendering the SL conducting and ultimately superconducting. This doping mechanism sketched in Figure 7c would prevent the anew build-up of the electrostatic potential. The oxygen redistribution for the CCO/STO SLs is schematically shown in Fig. 10. (a) Sketch of the experimental geometry. The escape probabilities I/I 0 =exp(z/l EAL sin θ) are superimposed onto the schematic of the CCO/STO SL, where l EAL is the effective attenuation length: lower red curve for Θ =15 ; upper black curve for Θ =70, (b) Sr 3d core-level spectra for the superconducting SL measured at an emission angle of 15, (c) Sr 3d core-level spectra of the superconducting SL measured at an emission angle of 70, (d) Sr 3d core-level spectra of the non-superconducting SL measured at an emission angle of 70. For all measurements the excitation energy was 2.8 kev; (b) (d): open circles are experimental data and solid black line is a fit to the data with shown two doublets (taken from Ref. [59]). Figures 7b and 7c. In the following we reproduce and discuss the HAXPES results which the authors provided as evidence for this structural and electronic rearrangement. That there is indeed no potential gradient in the SL is demonstrated by the width of the Sr 3d core-level peak, which is independent of the probed depth as shown in Figures 10b and 10c. Thus, the average electrostatic potential of the CCO block is supposed to be similar for the superconducting and the non-superconducting samples. However, locally charges may differ, internally accompanied by local electric fields. Besides being influenced by the electrostatic potential, binding energies are a function of local chemistry p7

8 The European Physical Journal Applied Physics (a) (c) (a) (b) (d) (b) (c) Fig. 11. Ti 2p3/2 core-level spectra for (a) superconducting and (b) non-superconducting SL. Ca 2p core-level spectra for (c) superconducting and (d) non-superconducting SLs recorded at 2.8 kev and an emission angle of θ =70.Opencircles: experimental data; solid black line: fit to the data based on the shown multi-peak structure (taken from Ref. [59]). and structure that can cause the core-level shifts shown in Figure 8b. Thus, it needs to be considered that the oxygen redistribution reduces the overall oxygen coordination for Sr and Ti at the CuO 2 /SrO interface (cf. Fig. 7). These sites with lower oxygen coordination are suggested to give rise to the components at lower binding energy in the Sr 3d and the Ti 2p core levels as shown in Figures 10b and10c and Figures 11a and 11b, respectively. Consequently, the oxygen coordination of Ca at the TiO 2 /Ca interface (cf. Fig. 7) will be (on average) increased, which is supported by the observed components at higher binding energy in the Ca 2p core levels as shown in Figures 11c and 11d. The HAXPES spectra in Figures 10 and 11 show that for the Ca2p, Sr3d, and Ti 2p core levels the spectral weight of the components at higher binding energy increases when the SLs films are grown at high oxygen/ozone pressure, thereby acquiring superconducting properties. TheCu2p 3/2 core-level HAXPES spectra are in agreement with the hole doping, which was found earlier by Hall effect and X-ray absorption spectroscopy measurements [58]. A signature characteristic of apical oxygen, i.e., the screening feature marked by C in Figure 12, is observed in the Cu 2p core levels [62,63]. In the Cu 2p core-level spectra shown in Figure 12, the relative contribution of the feature C increases when the film is grown under stronger oxidizing conditions. The increase of the intensity of the high-binding-energy components, including the feature C of Cu 2p, further supports the idea that those components are associated with sites with higher oxygen coordination, in agreement with the assignment that extra oxygen is accommodated at the interfaces. Fig. 12. Cu 2p3/2 spectra of a superconducting sample recorded for two different excitation energies, 5.95 kev and 2.8 kev, (a) and (b), respectively and of a non-superconducting sample at an excitation energy of 3.2 kev (c). Measurements were performed at an emission angle of θ = 70.Opencircles are experimental data and solid black lines are fits to the data, based on the three components A, B, and C (taken from Ref. [59]). 5 PES investigation of TMO interfaces with standing wave excitation The X-ray standing wave (XSW) technique adds picometer spatial resolution to the photoelectron spectroscopy technique. Instead of being emitted by a transient X-ray wave, electrons are ejected by a stationary X-ray standing wave. That is, the X-ray intensity is modulated in space and the intensity maxima lead to a maximum in the photoelectron emission. The foundations of this approach were laid 50 years ago (actually at about the same time that the ESCA technique was conceived by Siegbahn and coworkers) by Boris Batterman at Bell laboratories in Murray Hill, New Jersey [64, 65]. In his pioneering experiments, Batterman did not detect the photoelectrons but the secondary X-ray fluorescence emitted from the Ge substrate and As dopant atoms in silicon. The first studies employing photoelectrons as the detection channel were published in the 80s [66,67]. For readers not at all familiar with the XSW technique, we will present here in the following a very brief description of the principles of the method. A comprehensive overview of the XSW technique was published recently [68] and a more concise description is given in a Surface Science Report from 1993 [69] p8

9 J. Zegenhagen: Photoelectron spectroscopy of transition metal oxide interfaces e ћω h (111) P (111) 0 Fig. 13. Reflection of a plane X-ray wave with momentum k 0 and electric field E 0 when tuned to the Bragg angle θ B.Inthe overlap region with the reflected wave characterised by k h and E h a standing wave (interference field) is formed. 5.1 Photoelectrons emitted from an X-ray standing wave The XSW is formed by the superposition of two coherent plane X-ray waves. (Almost) Plane X-ray waves are now available by the monochromatized, highly collimated X-rays emitted from insertion devices at third generation storage ring light sources. A second coherent plane wave is commonly created by Bragg reflection from a single crystal and the standing wave is formed in the overlap region of the incoming and reflected wave, as shown in Figure 13. Also employed for the generation of the second coherent wave is reflection from a mirror surface [70] or Bragg reflection from an artificial multilayer [71]. Importantly, when tuning the crystal through the range of reflection by adjusting the reflection angle or the incident energy of the X-rays, the planar standing wave moves. In case of Bragg reflection, it moves by half its spacing when passing through the reflection range. This allows thus scanning the XSW in a controlled way, sampling the position of PES selected atomic species. Since the planar standing wave is periodic with the diffracting planes, high resolution structural information is obtained, however, only on the length scale of the diffraction plane spacing. The probability for the photo excitation process [72] is in the dipole approximation 6 directly proportional to the intensity of the X-ray wavefield at the center of the atom. (No angular momentum is transferred to the electron by the photon.) Monitoring the intensity of the photoelectron signal as a function of the movement of the XSW provides information of the structural arrangement of the emitting atoms. This is schematically shown in Figure 14. The X-ray wavefield is created by the superposition of the incident and Bragg reflected X-ray waves (cf. Fig. 13) 6 In the dipole approximation, the exponential e 2πikr of the E-field operator in the matrix element, describing the photoelectric process, is simply replaced by 1, the first term of its corresponding representation in a Taylor series (1 + 2πikr π 2 (kr) 2...). Fig. 14. Schematic of an XSW produced by a Bragg reflection from a crystal in side view. The diffraction planes, here denoted as (1 1 1), are indicated. An atom, photoexcited by the maxima of the field intensity, emits a photoelectron. The (simplified) meaning of the coherent position P h is indicated. In case of a single atom or a delta-function like distribution of atoms, it measures the distance of the particular atom from the corresponding diffraction plane, normalized by the diffraction plane spacing. with the electric field vectors E 0 = e 0 E 0 and E h = e h E h, respectively with: E 0 = e 0 E 0 e i(ω0t k0r) and E h = e h E h e i(ω ht k h r), as shown schematically in Figure 13. Here e 0 and e h are polarization vectors. The two E-fields are coherent. Thus they have the same energy, i.e., and ω 0 = ω h = ω, k 0 = k h = k = k =2πλ 1, where k 0 and k h are the X-ray propagation vectors and λ is the X-ray wavelength. Being coherent, the two complex E-field amplitudes can be related by an amplitude and a phase factor via: with R being the reflectivity E h = RE 0 exp (iυ), R = I h I 0 = E h 2 E 0 2, in case of Bragg diffraction, which is considered here. The propagation vectors of the two plane waves are related by the diffraction, i.e., reciprocal space vector h via: as shown in Figure 13. k h = k 0 + h, (1) p9

10 The European Physical Journal Applied Physics The XSW E = E 0 + E h in the overlap region is given by: E = e iωt [ e 0 E 0 e ( ik0r) + e h RE0 e iυ e ( ik hr) ], and the wavefield intensity I = E 2 canbecastintothe simple form: I h = E 2 = I 0 (1+R + e 0 e h 2 ) R cos(υ hr). The dimensionless scalar product hr describes the spatial modulation of the wavefield intensity along h with the (wavefield) spacing d h =2πh 1. Normal to h the wavefield intensity is constant (cf. Figs. 13 and 14). Calculating reflectivity, wavefield intensities, the phase υ and other parameters requires the dynamical theory of X-ray diffraction [73 75] formulated in its most elegant form by Laue. The wavefield exists above the surface to a distance limited by the temporal coherence length. 7 For brevity we assume that the dipole approximation is valid. 8 In the so-called σ-geometry, when e 0 and e h are parallel, the probability of a photoelectron being emitted from an atom within the range of the standing wave and thus the expected photoelectron yield YA h from an atom A is exactly proportional to the field intensity at the location of the atom, i.e., YA h I 0 (1+R +2 ) R cos(υ hr A ). In reality a very large number N A of specific atoms A will be excited by the XSW. The total recorded yield Y h A,T is the summation of the yields of all individual atoms. Normalized the yield is given by: Y h A,T = N 1 A N A j { 1+R +2 } R cos(υ hr Aj ). (2) Because of the large number of atoms and the specific sites r Aj, which the atoms A may occupy, the summation can be substituted by an integral. Introducing the distribution function G(r), we can write: YA,T h = N 1 A r { drg A (r) 1+R +2 } R cos(υ hr A ). (3) Obviously, the summation/integration of cosine functions yields again a cosine function with an amplitude F h A 1 and an average phase P h A i.e., Y h A,T =1+R +2 RF h A cos(υ 2πP h A), (4) 7 The temporal or longitudinal coherence length L s is given by L s = λ 2 Δλ, whereδλ is the wavelength spread i.e., monochromaticity of the X-ray beam. 8 It should be noted that multipole cross sections generally cannot be neglected when performing photoelectron spectroscopy with standing wave excitation [68, 76]. with 2πPA h = hz Ah =2πz Ah /d h = hr A. (5) Casting the yield function in this form, 9 it becomes evident that the two parameters fa h and P A h (coherent fraction and coherent position) in equation (4) represent in fact the amplitude and phase of one of the coefficients of the Fourier transform of G A (r) [69,77], i.e., G A (h) =G h A = F h Ae ihra = F h Ae i2πp h A. (6) Thus, an image of the analyzed distribution of atoms A can be created by a simple Fourier back transformation [78, 79]. G A,XSW (r) =1+2 h F h A cos[2πp h A hr]. (7) Since the available h-values and thus Fourier coefficients are restricted (to the allowed Bragg reflections), length scales beyond the dimensions of the unit cell are not accessible and thus the image represents the real distribution of the selected atomic species folded on to the length scale of the (largest) diffraction plane spacing. Thus, the maximum measured length scale cannot exceed the size of the unit cell of the crystal which produces the XSW. However, just one XSW measurement can produce valuable information since the determined coherent position PA h gives an average position of the specific atomic species A. This is particular valuable in case of surface adsorbates or epitaxial layers, where the distance from the substrate surface is a very important parameter. The coherent fraction FA h provides information of the distribution around this mean position. The closer FA h is to unity, the narrower the distribution. A meaning of the coherent position is given a simple explanation in Figure Epitaxial HTS on NdGaO 3 Relatively early after the discovery of high temperature superconductivity, epitaxial films of these materials were found to offer new ways to study their properties and a rich playground to modify their properties. For instance, by tailoring the substrate surface structure, extremely high critical currents of A/cm 2, 10 have been achieved in thin films of the 90 K high temperature superconductor YBa 2 Cu 3 O 7 δ grown on SrTiO 3 (0 0 1) [80,81]. Consequently, the interface structure of HTS films on substrates, typically with perovskite structure, became of interest. In 2005 Lee et al. applied photoelectron spectroscopy using the XSW technique for investigating the interface structure and epitaxial properties of such films [82]. 9 The electron yield Y h A detected in an XSW experiment is given by Y h A = I A,0Y h A,T where I A,0, called off-bragg yield, is proportional to the number of sampled atoms and other parameter such as photoelectric cross section, solid angle, beam intensity and more. 10 This is 30% of the theoretical depairing limit p10

11 J. Zegenhagen: Photoelectron spectroscopy of transition metal oxide interfaces Reflectivity & normalized yield Intensity (arb. units) (a) (b) (2) (1) (2) (1) Binding energy (ev) O 1s f = 0.15 P = 0.40 f = 0.39 P = E γ - E B (ev) O 1s NGO(004) Fig. 15. O 1s results of HAXPES/XSW investigations for a 3.5-nm GdBCO film on NdGaO 3(0 0 1). (a) The photoelectron spectrum shows two components with the smaller component (1) originating from the intrinsic part of the film whereas the larger one (2) is from contaminated parts of the film due to air exposure, (b) the XSW data (symbols) and best fits to the data (lines) for the NdGaO 3(0 0 4) reflection (taken from Ref. [82]). They studied ultra thin films (3.5 nm and 17.5 nm) of the HTS GdBa 2 Cu 3 O 7 δ (GBCO, orthorhombic, a = nm, b = nm, c = nm) grown on the (0 0 1) surface of NdGaO 3 (NGO). With the chemical sensitivity and enhanced probing depth of HAXPES, the internal structures of the film and the interface with the NGO (for the 3.5 nm sample) could be characterized. NGO has also an orthorhombic structure (a = nm, b = nm, c = nm), but one can define an almost cubic ( pseudo-cubic ) perovskite unit cell with a = b = nm and c = nm. The a/b-plane of YBCO matches well the a/b-plane of the pseudo cubic NGO unit cell, which will be referred to in the following. The GBCO thin films were grown ex-situ on singlecrystal NGO( 0 01) substrates by pulsed laser deposition (PLD). The samples were transferred without any particular precautions to the analysis chamber. XSW measurements were performed using the (almost) plane wave from a Si(1 1 1) monochromator and scanning the energy through the (0 0 4) ( pseudo cubic ) Bragg reflection at 3.23 kev ± ΔE of the NGO substrate at a Bragg angle close to 90. At Bragg angles close to backreflection, the reflection curves become wide in angle and some crystal mosaicity can be tolerated [68, 83, 84]. Different chemical components could be identified in the photoelectron spectra of the O 1s and Cu 2p corelevel. The components that exhibited high coherent fractions, were attributed to well-ordered (intrinsic) parts of the oxide films. It was concluded that components of the O1s andcu2p spectra that showed low coherent fractions were arising from degraded (extrinsic) parts of the films. Figure 15 showstheo1s spectrum and the results of the corresponding XSW analysis for both components. The large mean-free paths of the photoelectrons permitted to access the interface. The analysis of the signal of the Nd and Ga atoms coming from the buried interface below the 3.5-nm film suggested an essentially bulkterminated NGO surface. Together with the information about the lattice constant of the GBCO, the X-ray standing wave results allowed to conclude that the CuO plane of the GBCO is attached to the NdO plane, with a distance between the CuO and NdO plane of 0.18 nm. 11 Interestingly enough, for YBa 2 Cu 3 O 7 δ on SrTiO 3 (0 0 1), it had been found that the superconductor binds with the BaO plane to the TiO 2 plane of the substrate [87]. The XSW results also showed that the 3.5 nm GBCO film was adapting the in-plane lattice constant of the NGO surface and was thus pseudomorphically strained. The thicker, 17.5-nm film, was found to be already partially relaxed. Lee et al. also analyzed the GBCO valence band, and the results are reproduced in Figure 16. The values of coherent fraction F H and coherent position P H of the integral of the valence band were nearly identical to the values for the intrinsic part of the Cu 2p core level. This allowed to conclude that the dominant contribution to the valence band yield in this high-temperature superconducting cuprate originates from the Cu (Cu 3d) in the hard X-ray regime. 12 Readers interested in more detailed information are referred to the original publication [82]. 5.3 XSW imaging of the nucleation of a complex TMO The orthorhombic YBa 2 Cu 3 O 7 δ (YBCO) (a =0.383 nm, b =0.388 nm, c =1.168 nm) is in the a/b-plane well lattice matched with STO (a =0.391 nm) and thin films grow on STO(0 0 1) in-plane lattice matched in a pseudotetragonal form (a = b) with very good epitaxial quality [88]. However, it is a interesting question how such a 11 This surface distance, which is the projection of the distance between Cu and Nd on to the surface normal, should not the confused with bond length. 12 Note that the valence band yield is proportional to the density of states and the X-ray cross section which is for any element a strong function of X-ray energy p11

12 The European Physical Journal Applied Physics Intensity (arb. units) (a) Gd 4f VB Binding energy (ev) 2.0 Reflectivity & normalized yield (b) f = 0.34 P = 0.12 VB NGO(004) Fig. 17. Growth of YBa 2Cu 3O 7 δ on SrTiO 3(0 0 1). Left hand: a photograph of the PLD process and the 5 10 mm substrates with YBa 2Cu 3O 7 δ films. Right hand: an STM image and cross section (taken from Ref. [90]) E γ - E B (ev) Fig. 16. O1s HAXPES/XSW results of the valence band of a 3.5-nm GdBCO film on NdGaO 3(0 0 1). (a) The photoelectron spectrum of the GBCO valence band region including the Gd 4f photoelectron lines, (b) the XSW data (symbols) and best fits to the data (lines) of the GBCO valence band (hashed area) for the NdGaO 3(0 0 4) reflection (taken from Ref. [82]). complex TMO with a large unit cell starts to nucleate on a substrate surface. This question had been addressed by scanning tunnelling microscopy (STM) [89, 90]. Films of YBCO with less than a monolayer (ML) coverage had been deposited on STO(0 0 1) by pulsed laser deposition (PLD). The STM image of a 0.5 ML film revealed that YBCO does not grow in the 1.2 nm tall orthorhombic structure but in the form of smaller, 0.4 and 0.8 nm high sub-units [90] (cf. Fig. 17). Scanning tunnelling spectroscopy (STS) showed that these nuclei were not metallic. These smaller building blocks exhibited a pronounced band-gap, which was not the case for the 1.2 nm high islands, which were observed when the coverage was about one monolayer. It was concluded that the nuclei were perovskite-like material [90]. The internal structure of such a complex, incomplete layer is not easy to determine. In his PhD work, Thiess investigated the early stages of growth of YBCO on STO(0 0 1) by XSW imaging [18]. Half a monolayer of Fig. 18. Results of the XSW HAXPES measurements of 0.5 ML YBa 2Cu 3O 7 δ on SrTiO 3(0 0 1); symbols: data, lines: fits to the data. The corresponding reflectivity curves are shown on the bottom. The geometry of the used reflections is indicated on the top. YBCO was deposited by PLD in 50 Pa oxygen atmosphere on atomically clean STO(0 0 1), which was held at 760 C. The sample was slowly cooled down and then kept at 450 in 80 hpa oxygen for 20 min, which is the standard procedure for producing superconducting YBCO films. After cooling to room temperature, the sample was transferred under vacuum to the ultra high vacuum analysis chamber at the ID32 beamline at the ESRF [91]. To obtain sufficient spatial resolution, XSW measurements of several SrTiO 3 reflections in the energy range from 2.75 to 5.49 kev were carried out at Bragg angles close to 90. Photoelectron spectra of core levels of all six elements were recorded at each point of the XSW scans. The net counts were extracted from the spectra for all elements. The results of the XSW scans are shown in Figure 18.Afit to each of the yield curves using equation (4) provided the amplitude FA h and phase P A h of the Fourier coefficient p12

13 J. Zegenhagen: Photoelectron spectroscopy of transition metal oxide interfaces Fig. 19. Results of the image reconstruction for 0.5 ML YBa 2Cu 3O 7 δ on SrTiO 3(0 0 1) shown with respect to the nm cubic SrTiO 3 unit cell. of the distribution function for each reciprocal space vector h and element A. The images of all six elements could be constructed with the help of equation (7) 13 as shown in Figure 19. The largest length scale available is the unit cell size of the SrTiO 3, since STO diffraction planes were used for generating the reflections. However, this is perfectly appropriate for resolving the internal structure of the 0.4 and 0.8 nm nuclei. As suggested by the earlier STM/STS study [90], the deposited YBCO material adopts a perovskite structure. The position of the oxygen atoms are clearly resolved in the structure. However, the position of the oxygen of the substrate and the film are convoluted in the image since they could not be distinguished in the spectra. 6 Summary Transition metal oxide interfaces represent fascinating subjects of research. After 10 years of theoretical and experimental work, much progress has been made in the analysis and the understanding of the electronic properties of the STO/LAO interface last not least owing to photoelectron spectroscopy. Still, many questions remain unanswered regarding the formation of the 2D electron gas. The filling of the Ti d-band is undisputed, but the field in the LAO, which is supposed to cause the charge transfer, is an order of magnitude weaker than expected. Furthermore, the carrier density remains consistently below the expected value of 1/2e 0 per surface unit cell. Last not least, the origin and the role of the oxygen vacancies at the interface remains to be explained. As second example for TMO interfaces we presented a concise report of the HAXPES investigation of the interfacial reconstruction of superconducting CCO/STO SLs grown on STO(0 0 1) by Aruta et al. [59]. The authors suggest a complex interfacial reconstruction mechanism ultimately leading to the observed superconductivity in the heterostructure. Because of the polarity of the CCO, the CCO interfaces with the STO would be oppositely 13 Utilizing additionally the cubic symmetry of the SrTiO 3. charged, giving rise to a built-in, strongly diverging electrostatic potential. The result of the HAXPES study provides evidence that such a potential is suppressed by a combination of oxygen redistribution and charge transfer in the alkaline-earth-metal interface planes. Direct charge transfer at the interface between CCO and STO, i.e., band doping appears to be prohibited as the observed band alignment suggests. However, extra oxygen, which is incorporated at the interface under strongly oxidizing growth conditions, preserves the interface charge neutrality by leaving holes in the CuO 2 planes, rendering the cuprate block superconducting. Photoelectron spectroscopy can advantageously be combined with the XSW technique, thus adding structural resolution to its already strong features. For this, we provided two examples. We reported briefly the main results of the XSW/ HAXPES study of two GdBCO thin films that had been grown by PLD on NGO(0 0 1) substrates and studied by XSW using photoemission and X-ray fluorescence. We showed that the valence band and the Cu 2p core level of the film exhibit identical XSW modulations, confirming that the Cu 3d states dominate the GdBCO valence band at the used hard X-ray energy. It should be noted that in particular at (hard) X-ray energies not only the density of states but also strongly the X-ray cross section determine the photoelectron yield. Because of space constraints we did not explicitly show all results of that study, but highlighted the fact that the structure of different chemically modified components can be analysed. The films had been prepared ex-situ and HAXPES obviously allows to study buried interfaces of such samples. Structure and electronic properties are intimately connected and XSW combined with HAXPES allows to resolve structural details of complex interfaces of TMO materials which are very difficult (or even impossible) to obtain by other means. The determination of the structure of an ultra-thin, sub-monolayer HTS film on STO, which we briefly described here, is a nice example in this respect. References 1. J.G. Bednorz, K.A. Müller, Z. Phys. B 64, 189 (1986) 2. N. Klein, Rep. Prog. Phys. 65, 1387 (2002) 3. P. Chaudhari, R.H. Koch, R.B. Laibowitz, T.R. McGuire, R.J. Gambino, Phys. Rev. Lett. 58, 2684 (1987) 4. A. Ohtomo, H.Y. Hwang, Nature 427, 423 (2004) 5. H.Y. Hwang, Y. Iwasa, M. Kawasaki, B. Keimer, N. Nagaosa, Y. Tokura, Nat. Mater. 11, 103 (2012) 6. J. Mannhart, D.G. Schlom, Science 327, 1607 (2010) 7. J. Heber, Nature 459, 28 (2009) 8. J. Hubbard, Proc. Roy. Soc. (London) Ser. A 276, 238 (1963) 9. D.W. Fischer, Phys. Rev. B 5, 4219 (1975) 10. C.N.R. Rao, B. Raveau, Transition Metal Oxides, 2nd edn. (Wiley-VCH, Weinheim, 1995) 11. G.A. Sawatzky, J.W. Allen, Phys. Rev. Lett. 53, 2339 (1984) p13

14 The European Physical Journal Applied Physics 12. A. Svane, O. Gunnarsson, Phys. Rev. Lett. 65, 1148 (1990) 13. R. Merlin, Phys. Rev. Lett. 54, 2727 (1985) 14. Q. Yin, A. Gordienko, X. Wan, S.Y. Savrasov, Phys. Rev. Lett. 100, (2008) 15. V.I. Anisimov, I.V. Solovyev, M.A. Korotin, M.T. Czyżyk, G.A. Sawatzky, Phys. Rev. B 48, (1993) 16. P. Fulde, Correlated Electrons in Quantum Matter (World Scientific, Singapore, 2012), Chap W. Shockley, Bell System Technical Journal, 28, 435 (1949) 18. S. Thiess, PhD thesis, University Hamburg, C. Dallera, G. Panaccione, G. Paolicelli, B. Cowie, J. Zegenhagen, L. Braicovich, Appl. Phys. Lett. 85, 4532 (2004) 20. K. Siegbahn, Nucl. Instrum. Methods Phys. Res. A 547, 1 (2005) 21. I. Lindau, P. Pianetta, S. Doniach, W.E. Spicer, Nature 250, 214 (1974) 22. P. Risterucci, O. Renault, E. Martinez, B. Detlefs, J. Zegenhagen, G. Grenet, S. Tougaard, Surf. Interface Anal. 46, 906 (2014) 23. S. Thiess, C. Kunz, B.C.C. Cowie, T.-L. Lee, M. Renier, J. Zegenhagen, Solid State Commun. 132, 589 (2004) 24. HAXPES, Proceedings of the Workshop on Hard X-ray Photoelectron Spectroscopy, Grenoble 2003, edited by J. Zegenhagen, C. Kunz, Nucl. Instrum. Methods Phys. Res. A 547, 661 (2005) 25. L. Kövér, J. Electron Spectrosc. Related Phenom , 221 (2010) 26. Recent Advances in Hard X-ray Photoelectron Spectroscopy, edited by W. Drube, J. Eletron Spectrosc. Related Phenom. 190, Part B, 126 (2013) 27. H. Wadati, A. Fujimori, J. Eletron Spectrosc. Related Phenom. 190, 222 (2013) 28. Y. Shvyd ko, X-Ray Optics: High-Energy-Resolution Application (Springer-Verlag Berlin and Heidelberg, 2004) 29. T. Ishikawa, Nucl. Instrum. Methods A 547, 42 (2005) 30. A. Brinkman, M. Huijben, M. van Zalk, J. Huijben, U. Zeitler, J.C. Maan, W.G. van der Wiel, G. Rijnders, D.H.A. Blank, H. Hilgenkamp, Nat. Mater. 6, 493 (2007) 31. N. Reyren, S. Thiel, A.D. Caviglia, L. Fitting Kourkoutis, G. Hammerl, C. Richter, C.W. Schneider, T. Kopp, A.-S.Retschi,D.Jaccard,M.Gabay,D.A.Muller,J.-M. Triscone, J. Mannhart, Science 317, 1196 (2007) 32. L.X. Cao, E. Sozontov, J. Zegenhagen, Phys. Stat. Sol. A 181, 387 (2000) 33. C.J. Howard, B.J. Kennedy, B.C. Chakoumakos, J. Phys. Condens. Matter 12, 349 (2000) 34. J. Goniakowski, C. Noguera, Surf. Sci. 365, L657 (1996) 35. N. Nakagawa, H.Y. Wang, D.A. Müller, Nat. Mater. 5, 204 (2006) 36. Z. Zhong, P.X. Xu, P.J. Kelly, Phys. Rev. B 82, (2010) 37. N. Pavlenko, T. Kopp, E.Y. Tsymbal, J. Mannhart, G.A. Sawatzky, Phys. Rev. B 86, (2012) 38. C. Cantoni et al., Adv. Mat. 24, 3952 (2012) 39. E. Slooten, Z. Zhong, H.J.A. Molegraaf, P.D. Eerkes, S. de Jong, F. Massee, E. van Heumen, M.K. Kruize, S. Wenderich, J.E. Kleibeuker, M. Gorgoi, H. Hilgenkamp, A. Brinkman, M. Huijben, G. Rijnders, D.H.A. Blank, G. Koster, P.J. Kelly, M.S. Golden, Phys. Rev. B 87, (2013) 40. 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) 41. Y.Y. Chu, Y.F. Liao, V.T. Tra, J.C. Yang, W.Z. Liu, Y.H. Chu, J.Y. Lin, J.H. Huang, J. Weinen, S. Agrestini, K.-D. Tsuei, D.J. Huang, Appl. Phys. Lett. 99, (2011) 42. G. Drera, F. Banfi, F. Federici Canova, P. Borghetti, L. Sangaletti, F. Bondino, E. Magnano, J. Huijben, M. Huijben, G. Rijnders, D.H.A. Blank, H. Hilgenkamp, A. Brinkman, Phys. Lett. 98, (2011) 43. A. Koitzsch, J. Ocker, M. Knupfer, M.C. Dekker, K. Dörr, B. Büchner, P. Hoffmann, Phys. Rev. B 84, (2011) 44. C. Cancellieri, M.L. Reinle-Schmitt, M. Kobayashi, V.N. Strocov, P.R. Willmott, D. Fontaine, Ph. Ghosez, A. Filippetti, P. Delugas, V. Fiorentini, Phys. Rev. B 89, (2014) 45. Y. Lee, C. Clement, J. Hellerstedt, J. Kinney, L. Kinnischtzke, X. Leng, S.D. Snyder, A.M. Goldman, Phys. Rev. Lett. 106, (2011) 46. G. Perluzzo, J. Destry, Can. J. Phys. 56, 453 (1978) 47. A. Fujimori, I. Hase, M. Nakamura, H. Namatame, Y. Fujishima, Y. Tokura, M. Abbate, F.M.F. de Groot, M.T. Czyzyk, J.C. Fuggle, O. Strebel, F. Lopez, M. Domke, G. Kaindl, Phys. Rev. B 46, 9841 (1992) 48. D.D. Sarma, S.R. Barman, H. Kajueter, G. Kotliar, Europhys. Lett. 36, 307 (1996) 49. Y. Ishida, R. Eguchi, M. Matsunami, K. Horiba, M.Taguchi,A.Chainani,Y.Senba,H.Ohashi,H.Ohta, S. Shin, Phys. Rev. Lett. 100, (2008) 50. P. Perna, D. Maccariello, M. Radovic, U. Scotti di Uccio, I. Pallecchi, M. Codda, D. Marr, C. Cantoni, J. Gazquez, M. Varela, S.J. Pennycook, F. Miletto Granozio, Phys. Lett. 97, (2010) 51. C. Aruta, S. Amoruso, R. Bruzzese, X. Wang, D. Maccariello, F. Miletto Granozio, U. Scotti di Uccio, Appl. Phys. Lett. 97, (2010) 52. H. Chen, A.M. Kolpak, S. Ismail-Beigi, Adv. Mater. 22, 2881 (2010) 53. J.F. Schooley, W.R. Hosler, M.L. Cohen, Phys. Rev. Lett. 12, 474 (1964) 54. J.F. Schooley, W.R. Hosler, E. Ambler, J.H. Becker, M.L. Cohen, S.C. Koonce, Phys. Rev. Lett. 14, 305 (1965) 55. G. Balestrino, P.G. Medaglia, P. Orgiani, A. Tebano, C. Aruta, S. Lavanga, A.A. Varlamov, Phys. Rev. Lett. 89, (2002) 56. P. Orgiani, C. Aruta, G. Balestrino, D. Born, L. Maritato, P.G. Medaglia, D. Stornaiuolo, F. Tafuri, A. Tebano, Phys. Rev. Lett. 98, (2007) 57. A. Gozar, G. Logvenov, L. Fitting Kourkoutis, A.T. Bollinger, L.A. Giannuzzi, D.A. Muller, I. Bozovic, Nature 455, 782 (2008) 58. D. Di Castro, M. Salvato, A. Tebano, D. Innocenti, C. Aruta, W. Prellier, O.I. Lebedev, I. Ottaviani, N.B. Brookes, M. Minola, M. Moretti Sala, C. Mazzoli, P.G. Medaglia, G. Ghiringhelli, L. Braicovich, M. Cirillo, G. Balestrino, Phys. Rev. B 86, (2012) 59. C. Aruta, C. Schlueter, T.-L. Lee, D. Di Castro, D. Innocenti, A. Tebano, J. Zegenhagen, G. Balestrino, Phys. Rev. B 87, (2013) 60. S.A. Chambers, T. Droubay, T.C. Kaspar, M. Gutowski, J. Vac. Sci. Technol. B 22, 2205 (2004) p14

15 J. Zegenhagen: Photoelectron spectroscopy of transition metal oxide interfaces 61. S. Yunoki, A. Moreo, E. Dagotto, S. Okamoto, S.S. Kancharia, A. Fujimori, Phys. Rev. B 76, (2007) 62. T. Boske, 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, 3438 (1997) 63. T. Boske, K. Maiti, O. Knauff, K. Ruck, M.S. Golden, G. Krabbes, J. Fink, T. Osafune, N. Motoyama, H. Eisaki, S. Uchida, Phys. Rev. B 57, 138 (1998) 64. B.W. Batterman, Phys. Rev. 133, A759 (1964) 65. B.W. Batterman, Phys. Rev. Lett. 22, 703 (1969) 66. S.M. Durbin, L.E. Berman, B.W. Batterman, J.M. Blakeley, J. Vac. Sci. Technol. A 3, 973 (1985) 67. T. Ohta, Y. Kitajima, H. Kuroda, T. Takahashi, S. Kikuta, Nucl. Instrum. Methods A 246, 760 (1986) 68. J. Zegenhagen, A.Y. Kazimirov (eds.), The X-Ray Standing Waves Technique: Principles and Applications (World Scientific, Singapore, 2013) 69. J. Zegenhagen, Surf. Sci. Rep. 18, 199 (1993) 70. M.J. Bedzyk, D.H. Bilderback, G.M. Bommarito, M. Caffrey, J.S. Schildkraut, Science 241, 1788 (1988) 71. T.W. Barbee, W.K. Warburton, Mater. Lett. 3, 17 (1984) 72. A. Einstein, Ann. Phys. 322, 132 (1905) 73. M.v. Laue, Röntgenstrahl-Interferenzen (Akademische Verlagsgesellschaft, Becker & ErlerKolm.-Ges., Leipzig, 1941; Akademische Verlagsgesellschaft, Frankfurt, 1960) 74. B. Batterman, H. Cole, Rev. Mod. Phys. 36, 681 (1964) 75. A. Authier, Dynamical Theory of X-ray Diffraction (IUCr Monographs on Crystallography, Oxford Science Publications, 2001) 76. I. Vartanyants, J. Zegenhagen, Solid State Commun. 113, 299 (1999) 77. N. Hertel, G. Materlik, J. Zegenhagen, Z. Phys. B 58, 199 (1985) 78. Z. Zhang, P. Fenter, L. Cheng, N.C. Sturchio, M.J. Bedzyk, M.L. Machesky, D.J. Wesolowski, Surf. Sci. 554, L95 (2004) 79. J. Zegenhagen, Surf. Sci. 554, 77 (2004) 80. T. Haage, J. Zegenhagen, J.Q. Li, H.-U. Habermeier, M. Cardona, Ch. Jooss, R. Warthmann, A. Forkl, H. Kronmüller, Phys. Rev. B 56, 8404 (1997) 81. Ch. Jooss, R. Warthmann, H. Kronmüller, T. Haage, H.-U. Habermeier, J. Zegenhagen, Phys. Rev. Lett. 82, 632 (1999) 82. T.-L. Lee, S. Warren, L. Cao, S. Thiess, B.C.C. Cowie, J. Zegenhagen, Nucl. Instrum. Methods A 547, 216 (2005) 83. T. Ohta, H. Sekiyama, Y. Kitayima, H. Kuroda, T. Takahashi, S. Kikuta, Jpn J. Appl. Phys. 24, L475 (1985) 84. D.P. Woodruff, D.L. Seymour, C.F. McConville, C.E. Riley, M.D. Crapper, N.P. Prince, Phys. Rev. Lett. 58, 1460 (1987) 85. D. Goldschmidt, H.L. Tuller, Phys. Rev. B 35, 4360 (1987) 86. Y. Tokura, S. Koshihara, T. Arima, H. Takagi, S. Ishibashi, T. Ido, S. Uchida, Phys. Rev. B 41, (1990) 87. J. Zegenhagen, T. Siegrist, E. Fontes, L.E. Berman, J.R. Patel, Solid State Commun. 93, 763 (1995) 88. Q.D. Jiang, D.M. Smilgies, R. Feidenhans l, M. Cardona, J. Zegenhagen, Solid State Commun. 98, 157 (1996) 89. T. Haage, Q.D. Jiang, M. Cardona, H.-U. Habermeier, J. Zegenhagen, Appl. Phys. Lett. 68, 2427 (1996) 90. T. Haage, J. Zegenhagen, H.-U. Habermeier, M. Cardona, Phys. Rev. Lett. 80, 4225 (1998) 91. J. Zegenhagen, B. Detlefs, T.-L. Lee, S. Thiess, H. Isern, L. Petit, L. André,J.Roy,Y.Mi,I.Joumard,J.Electron. Spectrosc. Related Phenom , 258 (2010) Jörg Zegenhagen received his PhD (University Hamburg) in 1984 and his Habilitation (Dortmund University) in He was staff meber at SUNY, Albany and AT&T Bell Labs from , at Max-Planck-Institut-FKF Stuttgart from and the ESRF Grenoble from Since 2013 he is Science coordinator at Diamond Light Source Ltd. JZ lectured in France and Germany and edited several books and special journal issues. He (co)authored more than 250 publications and is a fellow of APS and member of DPG and MRS. His research is mainly focused on surfaces and interfaces using synchrotron X-ray scattering and spectroscopy techniques. Jörg Zegenhagen p15