- Email: [email protected]
Bulk-sensitive high-resolution photoemission study of a temperature-induced valence transition system EuPd2Si2
Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
Bulk-sensitive high-resolution photoemission study of a temperature-induced valence transition system EuPd2 Si2 Kojiro Mimura a,∗ , Yukihiro Taguchi a , Shuichi Fukuda b , Akihiro Mitsuda b , Junji Sakurai b , Kouichi Ichikawa a , Osamu Aita a a
Department of Mathematical Sciences, Graduate School of Engineering, Osaka Prefecture University, Sakai 599-8531, Japan b Department of Physics, Faculty of Science, Toyama University, Toyama 930-8555, Japan Available online 21 March 2004
Abstract Eu 4f electronic structures of a temperature-induced valence transition system EuPd2 Si2 have been investigated by bulk-sensitive highresolution photoemission spectroscopy at temperatures from 20 to 300 K. The bulk Eu2+ 4f component is definitely distinguished from two surface Eu2+ 4f components. The changes in the spectral intensity of the bulk Eu2+ and Eu3+ 4f configurations and in the energy separation between these states are observed in the temperature dependent photoemission spectra. These temperature dependences are related to the valence transition of EuPd2 Si2 . The Eu mean valence is evaluated to be 2.75 ± 0.03 at 20 K and 2.30 ± 0.05 at 300 K. These values are in good agreement with those evaluated from Mössbauer and Eu LIII -edge X-ray absorption measurements. © 2004 Elsevier B.V. All rights reserved. Keywords: EuPd2 Si2 ; Temperature-induced valence transition; Photoemission spectroscopy; Eu 4f state
1. Introduction Rare-earth compounds have attracted much attention because of their interesting properties such as valence transition, dense Kondo effect, heavy Fermion behavior, and so on. Such phenomena are derived from the competition between the localized and itinerant nature of 4f electrons. EuPd2 Si2 with the tetragonal ThCr2 Si2 -type structure is well known to exhibit intermediate valence between Eu2+ (4f7 ) and Eu3+ (4f6 ) [1]. This compound undergoes an abrupt but continuous valence transition at around 160 K. It has been revealed that the Eu mean valence changes from 2.8 below 130 K to 2.3 above 180 K from the results of the Mössbauer and Eu LIII -edge X-ray absorption spectroscopy (XAS) measurements [1–3]. As a Eu2+ ion is larger than a Eu3+ ion, the Eu atoms in the high-temperature phase are stabilized in the Eu2+ state. Furthermore, high-field magnetization at low temperature shows a first-order valence transition between the Eu3+ (J = 0) and Eu2+ (J = 7/2) components [4,5]. Both the temperature- and field-induced valence transitions
∗ Corresponding author. Tel.: +81-72-254-9367; fax: +81-72-254-9916. E-mail address: [email protected] (K. Mimura).
0368-2048/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.elspec.2004.02.098
have been explained on the basis of the interconfigurational fluctuation model [5]. It is important to investigate the electronic structures of the rare-earth compounds in order to understand their electrical and magnetic properties. Photoemission spectroscopy is a powerful technique for directly obtaining information on the occupied electronic states. Mårtensson et al. have investigated the electronic structures of EuPd2 Si2 by measuring photoemission spectra, which were taken at 50 and 300 K with an incident photon energy (hν) of 120 eV [6]. Temperature dependence of the bulk Eu2+ and Eu3+ 4f components has been observed in the spectra. The intensity of the Eu2+ 4f component, however, did not show so large temperature dependence as expected from the change in the Eu mean valence by the Mössbauer and Eu LIII -edge XAS measurements [1–3]. They have explained, as one of the reasons, that the Eu atoms of the second surface Eu layer are divalent at all the temperatures and that the emission from these Eu2+ atoms is not measurably shifted from the deep-bulk Eu2+ emission. It is well known that the surface electronic structures in rare-earth metals and compounds are considerably different from the bulk electronic structures. In Eu compounds, for example, Eu atoms are almost divalent in the surface layers, even if trivalent in the bulk [6,7]. Recently high-resolution photoemission measurements with
K. Mimura et al. / Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
soft X-rays have succeeded in observing clearly the bulk 4f states for Ce compounds [8]. It also enables us to distinguish the bulk Eu 4f states from the surface states. In the present study, we have investigated the temperature dependence of the valence-band electronic structures, in particular, the Eu 4f states of EuPd2 Si2 by means of bulk-sensitive high-resolution photoemission spectroscopy. Bulk Eu2+ 4f states, showing distinctly temperature dependence, have been successfully observed. We observed the changes in the spectral intensity of the bulk Eu2+ and Eu3+ 4f configurations and in the energy separation between the Eu2+ and Eu3+ 4f configurations, through the temperature-induced valence transition of EuPd2 Si2 . We compare the temperature dependence of the Eu mean valence and the transition temperature evaluated from the analysis of the photoemission spectra with those from the Mössbauer and Eu LIII -edge XAS measurements [1–3].
h ν = 700 eV EuPd2Si2 300 K
Intensity (arb. units)
530
200 K 150 K 100 K 50 K 20 K
2. Experimental Polycrystalline EuPd2 Si2 was grown by arc-melting the constituent elements under argon atmosphere and by annealing at 900 ◦ C in an evacuated quartz tube for one week. We have confirmed by X-ray diffraction that the sample has a single phase with the ThCr2 Si2 -type structure. The χ–T curve of the sample exhibited the valence transition at around 160 K [1]. High-resolution photoemission measurements were performed at a twin-helical undulator beam line BL25SU of SPring-8 [9]. Considering the difference of photoionization cross section between the Eu 4f and Pd 4d states [10] and the escape depth of photoelectrons [11], hν = 700 eV was chosen in the present photoemission measurements. The overall energy resolution was estimated to be about 85 meV. The measurements were carried out at temperatures from 20 to 300 K. The sample temperature was controlled by a liquid-He cryostat and a heater. Clean surface of the sample was prepared in situ by repeatedly scraping with a diamond file under the base pressure of 3 × 10−10 Torr. We have checked the sample cleanliness by O 1s signal in the spectra. Binding energy of the photoemission spectra was calibrated with respect to the Fermi edge of Au.
3. Results and discussion Fig. 1 shows the bulk-sensitive high-resolution photoemission spectra of EuPd2 Si2 measured at temperatures from 20 to 300 K with hν = 700 eV. The photoemission spectrum at 200 K of LaPd2 Si2 , which has no 4f electron, is also shown in Fig. 1 as a reference. The spectrum of LaPd2 Si2 exhibits two peaks at 0.7 and 3.3 eV and two shoulders near the Fermi level (EF ) and at 4 eV. By taking account of the photoionization cross section [10], the spectrum of LaPd2 Si2 dominantly reflects Pd 4d states. In addition to Pd 4d, mul-
Eu
3+
Eu
2+
LaPd2Si2 200 K
10
5
0
Binding Energy (eV) Fig. 1. Temperature dependence of bulk-sensitive high-resolution photoemission spectra of EuPd2 Si2 . The spectra were measured from 20 to 300 K with hν = 700 eV. Photoemission spectrum of LaPd2 Si2 at 200 K is also shown as a reference. Intensities of the spectra are normalized at 3.3 eV. Binding energies are referred to the Fermi level (EF ).
tiplet structures of the Eu 4f final states are observed at the binding energy ranges from ∼EF to 2.5 and 6 to 11 eV in the spectra of EuPd2 Si2 . The multiplet structures from ∼EF to 2.5 and 6 to 11 eV are ascribed to the Eu2+ and Eu3+ 4f states, respectively [6]. The Eu2+ 4f states consist of three components around 0.7, 1.1 and 2 eV. One should notice that only the component around 0.7 eV among the Eu2+ 4f states distinctly shows the temperature dependence, corresponding to the change in the Eu mean valence [1–3]. The two components around 1.1 and 2 eV, on the other hand, hardly show the systematic temperature dependence of the spectral shape and intensity. Considering the temperature dependence and the surface core-level shifts, we believe that the feature around 0.7 eV is ascribed to the bulk Eu2+ 4f emission, while those around 2 and 1.1 eV to the Eu2+ 4f states of the first and second surface Eu layers, respectively. The existence of the second Eu surface layer has been expected for the photoemission spectra of the Eu-based valence transition compounds such as EuPd2 Si2 [6] and EuNi2 (Si0.25 Ge0.75 )2 [12]. We have identified the bulk and the two surface Eu2+
K. Mimura et al. / Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
300 K VEu = 2.30
Exp. Fitting
200 K VEu = 2.33
Intensity (arb. units)
4f states of EuPd2 Si2 for the first time by the bulk-sensitive and high-resolution measurements. One also notices that the photoemission intensity of the Eu3+ 4f states decreases with increasing the temperature, in contrast to the change in the bulk Eu2+ 4f states. In particular, the change in the photoemission intensity is remarkable between 100 and 200 K. The temperature dependence of the bulk Eu 4f states corresponds to the valence transition of EuPd2 Si2 [1–3]. We notice, furthermore, that the 7.4 eV peak at 100 K, which is composed of the Eu3+ 4f multiplets, gradually shifts by about 0.2 eV toward the higher binding energy side, as the temperature goes from 100 to 200 K. The energy shift of the other features of the Eu3+ multiplets is hardly observed because the features collapse and/or merge into a tail of the Pd 4d peak at high temperatures. We assume in the later fitting procedure, however, that all the features of the Eu3+ multiplets shift as a whole toward the higher binding energy side, since the intensity and the energy position of the 7.4 eV peak change but the shape does not change with increasing the temperature. On the other hand, the energy position of the Eu2+ 4f multiplet structures is almost unchanged. The energy separation between the center of gravity of the Eu2+ and Eu3+ 4f structures is roughly given by εf + Uff , where εf represents the energy level of the 4f electron and Uff the averaged Coulomb interaction energy between the 4f electrons. That is, the observed change in the energy separation suggests that the εf + Uff value varies through the temperature-induced valence transition of EuPd2 Si2 . In order to see more clearly the temperature dependence of the Eu 4f states, we have extracted only the Eu 4f components from the spectra. The Eu 4f spectrum has been obtained by subtracting the spectrum of LaPd2 Si2 from that of EuPd2 Si2 , on the assumption that the electronic structures of conduction electrons such as Pd 4d, Si 3p and Si 3s in EuPd2 Si2 are the same as those in LaPd2 Si2 . Fig. 2 shows the Eu 4f spectra of EuPd2 Si2 from 20 to 300 K indicated by dots. Both the Eu2+ and Eu3+ 4f states are successfully extracted by this procedure except for the spectral weight between 3 and 4.5 eV in the region of Pd 4d. The Eu 4f spectra also show the changes in the spectral intensity of the bulk Eu 4f states and in the energy separation between the Eu2+ and Eu3+ states, corresponding to the change in the Eu mean valence. The dips at 3 and 4 eV may originate from the hybridization between Eu 4f and Pd 4d and/or the spin-flip satellite of the Eu2+ 4f states as observed in the Eu 3d–4f resonant photoemission spectra of EuNi2 (Si0.25 Ge0.75 )2 [13]. For the Eu 4f spectra, we have performed the line-shape analysis by using a least-square fitting and evaluated the Eu mean valence. For the analysis of both the bulk and surface components, we adopted the line spectra for the Eu2+ and Eu3+ 4f final states calculated by Gerken based on the atomic model [14]. The relative energy position of the calculated Eu2+ and Eu3+ 4f line spectra was determined from the experimental Eu 4f spectra. The calculated line spectra were convoluted by Gaussian and Lorentzian. Full-width at half maximum (FWHM) of the Gaussian was fixed to
531
150 K VEu = 2.39
100 K VEu = 2.59
50 K VEu = 2.74
Eu2+ 20 K VEu = 2.75
10
Eu3+
5
0
Binding Energy (eV) Fig. 2. Temperature-dependent Eu 4f spectra of EuPd2 Si2 (dots). Thick-solid lines indicate the spectra obtained by the least-square fitting. The calculated spectrum is composed of the bulk components, the surface components and the background indicated by thin-solid, dash-and-dotted and broken lines, respectively. Vertical lines in the spectrum at 20 K indicate the position and relative intensity of multiplet structures for the bulk Eu final states [14]. Label of VEu indicates the value of the Eu mean valence. Details of the fitting are explained in the text.
90 meV for the bulk Eu2+ and Eu3+ components, which is nearly the same as the instrumental resolution. For two surface Eu2+ components the phonon broadening affects the Gaussian width together with the inhomogeneity. The phonon broadening is substantially stronger for the photoemission component from the surface layer as compared to the bulk because of enhanced vibrational motion perpendicular to the surface and is included in the Gaussian for the analysis [15]. Then, FWHM’s of the Gaussian for two surface Eu2+ components have been assumed to be larger
K. Mimura et al. / Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
than that of the bulk component and to be fitting parameters. FWHM of the Lorentzian (2Γn ) for all the components is energy-dependent and is taken to be 2Γn = 2Γ0 + β E, where 2Γ0 = 50 meV is the Lorentzian width for the most intense line of the 7 F final states for the bulk Eu2+ multiplets, E the energy separation from the most intense line, and β a constant equal to 0.08. We assumed an asymmetric line shape for all the components [16]. The asymmetric parameters (α’s) were set at 0.18, 0.25 and 0.1 for the bulk Eu2+ , Eu3+ and surface Eu2+ components, respectively. The fitting results are also shown in Fig. 2. The experimental spectra are well reproduced by the calculated spectra within the framework of the atomic model [14] and precisely divided into the bulk and surface Eu 4f components. That is, such a successful calculation indicates that the Eu 4f electrons in EuPd2 Si2 have strongly localized characters and weakly hybridize with the conduction electrons. One recognizes that the second surface Eu layer is clearly resolved in the present spectra. According to the fitting results, it is considered that the existence of the surface layers, especially, the second surface Eu layer makes it difficult to observe the bulk Eu 4f states. We point out that the intensity of the Eu 4f states of the second surface Eu layer depends on the surface condition. Very recently, we have measured the Eu 4d–4f resonant photoemission spectra of EuPd2 Si2 with hν ∼ 140 eV for both the scraped and fractured sample [17]. The spectra for the fractured sample have shown that the Eu2+ 4f states of the second surface Eu layer are strongly suppressed, while the bulk Eu 4f states become more clear and intense, in comparison with the spectra for the scraped sample. This suggests that at the scraped surface the local coordination around Eu atom is disordered and is not well defined, in contrast with the original atomic coordination of tetragonal ThCr2 Si2 -type structure. Thus, for the scraped sample not only two surface Eu layers but also several Eu underlayers may contribute differently to the Eu 4f spectra besides the bulk. In Fig. 2, the fitting results have made clear that the experimental spectra are well represented by using the bulk and two surface Eu components and that the spectral intensity of the second surface Eu layer is stronger than that of the first surface Eu layer. The facts above mentioned consequently support the following consideration. Among the surface Eu layers which contribute unlike the bulk, only the topmost Eu layer contributes to the first surface Eu2+ 4f component, while the other Eu underlayers to the Eu2+ 4f component of what we call the second surface Eu layer in the experimental spectrum. That is, the Eu 4f emissions from the several Eu underlayers, which form the second Eu surface layer in the experimental spectrum, are not shifted each other in energy. Moreover the photoemission–intensity ratio of the second to the first surface Eu layers changes at each sample temperature. This indicates that the sample-surface condition considerably differs at each sample temperature due to the scraping. Finally, the evaluated Eu mean valence at temperatures from 20 to 300 K are shown by closed circles in Fig. 3,
2.3 2.4
Eu valence
532
2.5 2.6 2.7
EuPd2Si2
2.8 0
100
200
300
Temperature (K) Fig. 3. Temperature dependence of the Eu mean valence evaluated by fitting the Eu 4f spectra (closed circles). Solid line is a guide to the eye. Open triangles indicate the Eu mean valence evaluated from the analysis of the Eu LIII -edge XAS spectra [2].
together with that obtained from the Eu LIII -edge XAS measurements indicated by open triangles [2]. The Eu mean valence was evaluated by using the integrated intensities of the bulk Eu2+ and Eu3+ 4f components. We have estimated the Eu mean valence to be 2.75 ± 0.03 at 20 K and 2.30 ± 0.05 at 300 K. These values agree well with the values evaluated from the Mössbauer and Eu LIII -edge XAS measurements [1–3]. The obtained Eu mean valence also exhibits the abrupt decrease with increasing temperature. The estimated transition temperature, however, shifts by about 50 K toward the lower temperature side than those from the Mössbauer and Eu LIII -edge XAS measurements [1–3]. One of the possible reasons is that the existence of the large second surface Eu2+ 4f emission has an influence on the estimation of the Eu mean valence. In order to suppress the second surface Eu2+ 4f emission we will investigate the temperature dependence of the bulk-sensitive high-resolution photoemission spectra for the fractured sample.
4. Conclusions We have investigated the Eu 4f states of EuPd2 Si2 by means of the bulk-sensitive high-resolution photoemission spectroscopy from 20 to 300 K. We succeeded in observing the Eu2+ 4f states of the bulk as well as those of two surface layers in the spectra. The temperature-dependent photoemission spectra show the changes in the spectral intensity of the bulk Eu2+ and Eu3+ 4f components and in the energy separation between the Eu2+ and Eu3+ 4f components, through the temperature-induced valence transition of EuPd2 Si2 . The Eu mean valence has been evaluated by using the integrated intensities of the bulk Eu2+ and Eu3+ 4f contribution. The evaluated Eu mean valence is 2.75 ± 0.03 at 20 K and 2.30 ± 0.05 at 300 K. These values are in good agreement with those from the Mössbauer and Eu LIII -edge XAS measurements [1–3]. The estimated transition tem-
K. Mimura et al. / Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
perature, however, is about 50 K lower than those from the Mössbauer and the Eu LIII -edge XAS [1–3], because of the influence of the strong Eu2+ emission from the second surface Eu layer upon the Eu mean-valence estimation. In order to understand the mechanism for the temperature-induced valence transition of EuPd2 Si2 , we are planning to carry out the temperature-dependent high-resolution photoemission measurements for the fractured sample by using several hν’s and to obtain further information on the electronic structures of the conduction electrons such as Pd 4d, Si 3p and Si 3s as well as those of Eu 4f.
Acknowledgements We are grateful to Prof. S. Kimura for useful discussions. We also thank Messrs K. Mantani, K. Kitamoto and Ms. Y. Watanabe for their assistance. The photoemission measurements were performed at SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal No. 2001B0396-NS-np).
References [1] E.V. Sampathkumaran, L.C. Gupta, R. Vijayaraghavan, K.V. Gopalakrishnan, R.G. Pillay, H.G. Devare, J. Phys. C: Solid State Phys. 14 (1981) L237.
533
[2] E. Kemly, M. Groft, V. Murgai, L.C. Gupta, C. Godart, R.D. Parks, C.U. Segre, J. Magn. Magn. Mater. 47&48 (1985) 403. [3] G. Wortmann, K.H. Prank, E.V. Sampathkumaran, B. Perscheid, G. Schmiester, G. Kaindl, J. Magn. Magn. Mater. 49 (1985) 325. [4] H. Wada, A. Mitsuda, M. Shiga, H. Aruga Katori, T. Goto, J. Phys. Soc. Jpn. 65 (1996) 3471. [5] A. Mitsuda, H. Wada, M. Shiga, H. Aruga Katori, T. Goto, Phys. Rev. B 55 (1997) 12474. [6] N. Mårtensson, B. Reihl, W.-D. Schneider, V. Murgai, L.C. Gupta, R.D. Parks, Phys. Rev. B 25 (1982) 1446. [7] W.D. Schneider, C. Laubschat, G. Kalkowski, J. Haase, A. Puschmann, Phys. Rev. B 28 (1983) 2017. [8] A. Sekiyama, T. Iwasaki, K. Matsuda, Y. Saitoh, Y. Onuki, S. Suga, Nature 403 (2000) 396. [9] Y. Saitoh, H. Kimura, Y. Suzuki, T. Nakatani, T. Matsushita, T. Muro, T. Miyahara, M. Fujisawa, K. Soda, S. Ueda, H. Harada, M. Kotsugi, A. Sekiyama, S. Suga, Rev. Sci. Instrum. 71 (2000) 3254. [10] J.J. Yeh, I. Lindau, Atom. Data Nucl. Data Tables 32 (1985) 1. [11] S. Tanuma, C.J. Powel, D.R. Penn, Surf. Sci. 192 (1987) L849. [12] S. Kimura, M. Okuno, H. Iwata, T. Saitoh, T. Okuda, A. Harasawa, T. Kinoshita, A. Mitsuda, H. Wada, M. Shiga, J. Phys. Soc. Jpn. 71 (2002) 255. [13] T. Kinoshita, H.P.N.J. Gunasekara, Y. Takata, S. Kimura, M. Okuno, Y. Haruyama, N. Kosugi, K.G. Nath, H. Wada, A. Mitsuda, M. Shiga, T. Okuda, A. Harasawa, H. Ogasawara, A. Kotani, J. Phys. Soc. Jpn. 71 (2002) 148. [14] F. Gerken, J. Phys. F: Met. Phys. 13 (1983) 703. [15] E. Weschke, A. Höhr, S. Vandré, C. Schüßler-Langeheine, F. Bødker, G. Kaindl, J. Electron. Spectrosc. Relat. Phenom. 76 (1995) 571. [16] G.D. Mahan, Phys. Rev. B 11 (1975) 4814. [17] K. Mimura et al., unpublished.
Bulk-sensitive high-resolution photoemission study of a temperature-induced valence transition system EuPd2 Si2 Kojiro Mimura a,∗ , Yukihiro Taguchi a , Shuichi Fukuda b , Akihiro Mitsuda b , Junji Sakurai b , Kouichi Ichikawa a , Osamu Aita a a
Department of Mathematical Sciences, Graduate School of Engineering, Osaka Prefecture University, Sakai 599-8531, Japan b Department of Physics, Faculty of Science, Toyama University, Toyama 930-8555, Japan Available online 21 March 2004
Abstract Eu 4f electronic structures of a temperature-induced valence transition system EuPd2 Si2 have been investigated by bulk-sensitive highresolution photoemission spectroscopy at temperatures from 20 to 300 K. The bulk Eu2+ 4f component is definitely distinguished from two surface Eu2+ 4f components. The changes in the spectral intensity of the bulk Eu2+ and Eu3+ 4f configurations and in the energy separation between these states are observed in the temperature dependent photoemission spectra. These temperature dependences are related to the valence transition of EuPd2 Si2 . The Eu mean valence is evaluated to be 2.75 ± 0.03 at 20 K and 2.30 ± 0.05 at 300 K. These values are in good agreement with those evaluated from Mössbauer and Eu LIII -edge X-ray absorption measurements. © 2004 Elsevier B.V. All rights reserved. Keywords: EuPd2 Si2 ; Temperature-induced valence transition; Photoemission spectroscopy; Eu 4f state
1. Introduction Rare-earth compounds have attracted much attention because of their interesting properties such as valence transition, dense Kondo effect, heavy Fermion behavior, and so on. Such phenomena are derived from the competition between the localized and itinerant nature of 4f electrons. EuPd2 Si2 with the tetragonal ThCr2 Si2 -type structure is well known to exhibit intermediate valence between Eu2+ (4f7 ) and Eu3+ (4f6 ) [1]. This compound undergoes an abrupt but continuous valence transition at around 160 K. It has been revealed that the Eu mean valence changes from 2.8 below 130 K to 2.3 above 180 K from the results of the Mössbauer and Eu LIII -edge X-ray absorption spectroscopy (XAS) measurements [1–3]. As a Eu2+ ion is larger than a Eu3+ ion, the Eu atoms in the high-temperature phase are stabilized in the Eu2+ state. Furthermore, high-field magnetization at low temperature shows a first-order valence transition between the Eu3+ (J = 0) and Eu2+ (J = 7/2) components [4,5]. Both the temperature- and field-induced valence transitions
∗ Corresponding author. Tel.: +81-72-254-9367; fax: +81-72-254-9916. E-mail address: [email protected] (K. Mimura).
0368-2048/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.elspec.2004.02.098
have been explained on the basis of the interconfigurational fluctuation model [5]. It is important to investigate the electronic structures of the rare-earth compounds in order to understand their electrical and magnetic properties. Photoemission spectroscopy is a powerful technique for directly obtaining information on the occupied electronic states. Mårtensson et al. have investigated the electronic structures of EuPd2 Si2 by measuring photoemission spectra, which were taken at 50 and 300 K with an incident photon energy (hν) of 120 eV [6]. Temperature dependence of the bulk Eu2+ and Eu3+ 4f components has been observed in the spectra. The intensity of the Eu2+ 4f component, however, did not show so large temperature dependence as expected from the change in the Eu mean valence by the Mössbauer and Eu LIII -edge XAS measurements [1–3]. They have explained, as one of the reasons, that the Eu atoms of the second surface Eu layer are divalent at all the temperatures and that the emission from these Eu2+ atoms is not measurably shifted from the deep-bulk Eu2+ emission. It is well known that the surface electronic structures in rare-earth metals and compounds are considerably different from the bulk electronic structures. In Eu compounds, for example, Eu atoms are almost divalent in the surface layers, even if trivalent in the bulk [6,7]. Recently high-resolution photoemission measurements with
K. Mimura et al. / Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
soft X-rays have succeeded in observing clearly the bulk 4f states for Ce compounds [8]. It also enables us to distinguish the bulk Eu 4f states from the surface states. In the present study, we have investigated the temperature dependence of the valence-band electronic structures, in particular, the Eu 4f states of EuPd2 Si2 by means of bulk-sensitive high-resolution photoemission spectroscopy. Bulk Eu2+ 4f states, showing distinctly temperature dependence, have been successfully observed. We observed the changes in the spectral intensity of the bulk Eu2+ and Eu3+ 4f configurations and in the energy separation between the Eu2+ and Eu3+ 4f configurations, through the temperature-induced valence transition of EuPd2 Si2 . We compare the temperature dependence of the Eu mean valence and the transition temperature evaluated from the analysis of the photoemission spectra with those from the Mössbauer and Eu LIII -edge XAS measurements [1–3].
h ν = 700 eV EuPd2Si2 300 K
Intensity (arb. units)
530
200 K 150 K 100 K 50 K 20 K
2. Experimental Polycrystalline EuPd2 Si2 was grown by arc-melting the constituent elements under argon atmosphere and by annealing at 900 ◦ C in an evacuated quartz tube for one week. We have confirmed by X-ray diffraction that the sample has a single phase with the ThCr2 Si2 -type structure. The χ–T curve of the sample exhibited the valence transition at around 160 K [1]. High-resolution photoemission measurements were performed at a twin-helical undulator beam line BL25SU of SPring-8 [9]. Considering the difference of photoionization cross section between the Eu 4f and Pd 4d states [10] and the escape depth of photoelectrons [11], hν = 700 eV was chosen in the present photoemission measurements. The overall energy resolution was estimated to be about 85 meV. The measurements were carried out at temperatures from 20 to 300 K. The sample temperature was controlled by a liquid-He cryostat and a heater. Clean surface of the sample was prepared in situ by repeatedly scraping with a diamond file under the base pressure of 3 × 10−10 Torr. We have checked the sample cleanliness by O 1s signal in the spectra. Binding energy of the photoemission spectra was calibrated with respect to the Fermi edge of Au.
3. Results and discussion Fig. 1 shows the bulk-sensitive high-resolution photoemission spectra of EuPd2 Si2 measured at temperatures from 20 to 300 K with hν = 700 eV. The photoemission spectrum at 200 K of LaPd2 Si2 , which has no 4f electron, is also shown in Fig. 1 as a reference. The spectrum of LaPd2 Si2 exhibits two peaks at 0.7 and 3.3 eV and two shoulders near the Fermi level (EF ) and at 4 eV. By taking account of the photoionization cross section [10], the spectrum of LaPd2 Si2 dominantly reflects Pd 4d states. In addition to Pd 4d, mul-
Eu
3+
Eu
2+
LaPd2Si2 200 K
10
5
0
Binding Energy (eV) Fig. 1. Temperature dependence of bulk-sensitive high-resolution photoemission spectra of EuPd2 Si2 . The spectra were measured from 20 to 300 K with hν = 700 eV. Photoemission spectrum of LaPd2 Si2 at 200 K is also shown as a reference. Intensities of the spectra are normalized at 3.3 eV. Binding energies are referred to the Fermi level (EF ).
tiplet structures of the Eu 4f final states are observed at the binding energy ranges from ∼EF to 2.5 and 6 to 11 eV in the spectra of EuPd2 Si2 . The multiplet structures from ∼EF to 2.5 and 6 to 11 eV are ascribed to the Eu2+ and Eu3+ 4f states, respectively [6]. The Eu2+ 4f states consist of three components around 0.7, 1.1 and 2 eV. One should notice that only the component around 0.7 eV among the Eu2+ 4f states distinctly shows the temperature dependence, corresponding to the change in the Eu mean valence [1–3]. The two components around 1.1 and 2 eV, on the other hand, hardly show the systematic temperature dependence of the spectral shape and intensity. Considering the temperature dependence and the surface core-level shifts, we believe that the feature around 0.7 eV is ascribed to the bulk Eu2+ 4f emission, while those around 2 and 1.1 eV to the Eu2+ 4f states of the first and second surface Eu layers, respectively. The existence of the second Eu surface layer has been expected for the photoemission spectra of the Eu-based valence transition compounds such as EuPd2 Si2 [6] and EuNi2 (Si0.25 Ge0.75 )2 [12]. We have identified the bulk and the two surface Eu2+
K. Mimura et al. / Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
300 K VEu = 2.30
Exp. Fitting
200 K VEu = 2.33
Intensity (arb. units)
4f states of EuPd2 Si2 for the first time by the bulk-sensitive and high-resolution measurements. One also notices that the photoemission intensity of the Eu3+ 4f states decreases with increasing the temperature, in contrast to the change in the bulk Eu2+ 4f states. In particular, the change in the photoemission intensity is remarkable between 100 and 200 K. The temperature dependence of the bulk Eu 4f states corresponds to the valence transition of EuPd2 Si2 [1–3]. We notice, furthermore, that the 7.4 eV peak at 100 K, which is composed of the Eu3+ 4f multiplets, gradually shifts by about 0.2 eV toward the higher binding energy side, as the temperature goes from 100 to 200 K. The energy shift of the other features of the Eu3+ multiplets is hardly observed because the features collapse and/or merge into a tail of the Pd 4d peak at high temperatures. We assume in the later fitting procedure, however, that all the features of the Eu3+ multiplets shift as a whole toward the higher binding energy side, since the intensity and the energy position of the 7.4 eV peak change but the shape does not change with increasing the temperature. On the other hand, the energy position of the Eu2+ 4f multiplet structures is almost unchanged. The energy separation between the center of gravity of the Eu2+ and Eu3+ 4f structures is roughly given by εf + Uff , where εf represents the energy level of the 4f electron and Uff the averaged Coulomb interaction energy between the 4f electrons. That is, the observed change in the energy separation suggests that the εf + Uff value varies through the temperature-induced valence transition of EuPd2 Si2 . In order to see more clearly the temperature dependence of the Eu 4f states, we have extracted only the Eu 4f components from the spectra. The Eu 4f spectrum has been obtained by subtracting the spectrum of LaPd2 Si2 from that of EuPd2 Si2 , on the assumption that the electronic structures of conduction electrons such as Pd 4d, Si 3p and Si 3s in EuPd2 Si2 are the same as those in LaPd2 Si2 . Fig. 2 shows the Eu 4f spectra of EuPd2 Si2 from 20 to 300 K indicated by dots. Both the Eu2+ and Eu3+ 4f states are successfully extracted by this procedure except for the spectral weight between 3 and 4.5 eV in the region of Pd 4d. The Eu 4f spectra also show the changes in the spectral intensity of the bulk Eu 4f states and in the energy separation between the Eu2+ and Eu3+ states, corresponding to the change in the Eu mean valence. The dips at 3 and 4 eV may originate from the hybridization between Eu 4f and Pd 4d and/or the spin-flip satellite of the Eu2+ 4f states as observed in the Eu 3d–4f resonant photoemission spectra of EuNi2 (Si0.25 Ge0.75 )2 [13]. For the Eu 4f spectra, we have performed the line-shape analysis by using a least-square fitting and evaluated the Eu mean valence. For the analysis of both the bulk and surface components, we adopted the line spectra for the Eu2+ and Eu3+ 4f final states calculated by Gerken based on the atomic model [14]. The relative energy position of the calculated Eu2+ and Eu3+ 4f line spectra was determined from the experimental Eu 4f spectra. The calculated line spectra were convoluted by Gaussian and Lorentzian. Full-width at half maximum (FWHM) of the Gaussian was fixed to
531
150 K VEu = 2.39
100 K VEu = 2.59
50 K VEu = 2.74
Eu2+ 20 K VEu = 2.75
10
Eu3+
5
0
Binding Energy (eV) Fig. 2. Temperature-dependent Eu 4f spectra of EuPd2 Si2 (dots). Thick-solid lines indicate the spectra obtained by the least-square fitting. The calculated spectrum is composed of the bulk components, the surface components and the background indicated by thin-solid, dash-and-dotted and broken lines, respectively. Vertical lines in the spectrum at 20 K indicate the position and relative intensity of multiplet structures for the bulk Eu final states [14]. Label of VEu indicates the value of the Eu mean valence. Details of the fitting are explained in the text.
90 meV for the bulk Eu2+ and Eu3+ components, which is nearly the same as the instrumental resolution. For two surface Eu2+ components the phonon broadening affects the Gaussian width together with the inhomogeneity. The phonon broadening is substantially stronger for the photoemission component from the surface layer as compared to the bulk because of enhanced vibrational motion perpendicular to the surface and is included in the Gaussian for the analysis [15]. Then, FWHM’s of the Gaussian for two surface Eu2+ components have been assumed to be larger
K. Mimura et al. / Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
than that of the bulk component and to be fitting parameters. FWHM of the Lorentzian (2Γn ) for all the components is energy-dependent and is taken to be 2Γn = 2Γ0 + β E, where 2Γ0 = 50 meV is the Lorentzian width for the most intense line of the 7 F final states for the bulk Eu2+ multiplets, E the energy separation from the most intense line, and β a constant equal to 0.08. We assumed an asymmetric line shape for all the components [16]. The asymmetric parameters (α’s) were set at 0.18, 0.25 and 0.1 for the bulk Eu2+ , Eu3+ and surface Eu2+ components, respectively. The fitting results are also shown in Fig. 2. The experimental spectra are well reproduced by the calculated spectra within the framework of the atomic model [14] and precisely divided into the bulk and surface Eu 4f components. That is, such a successful calculation indicates that the Eu 4f electrons in EuPd2 Si2 have strongly localized characters and weakly hybridize with the conduction electrons. One recognizes that the second surface Eu layer is clearly resolved in the present spectra. According to the fitting results, it is considered that the existence of the surface layers, especially, the second surface Eu layer makes it difficult to observe the bulk Eu 4f states. We point out that the intensity of the Eu 4f states of the second surface Eu layer depends on the surface condition. Very recently, we have measured the Eu 4d–4f resonant photoemission spectra of EuPd2 Si2 with hν ∼ 140 eV for both the scraped and fractured sample [17]. The spectra for the fractured sample have shown that the Eu2+ 4f states of the second surface Eu layer are strongly suppressed, while the bulk Eu 4f states become more clear and intense, in comparison with the spectra for the scraped sample. This suggests that at the scraped surface the local coordination around Eu atom is disordered and is not well defined, in contrast with the original atomic coordination of tetragonal ThCr2 Si2 -type structure. Thus, for the scraped sample not only two surface Eu layers but also several Eu underlayers may contribute differently to the Eu 4f spectra besides the bulk. In Fig. 2, the fitting results have made clear that the experimental spectra are well represented by using the bulk and two surface Eu components and that the spectral intensity of the second surface Eu layer is stronger than that of the first surface Eu layer. The facts above mentioned consequently support the following consideration. Among the surface Eu layers which contribute unlike the bulk, only the topmost Eu layer contributes to the first surface Eu2+ 4f component, while the other Eu underlayers to the Eu2+ 4f component of what we call the second surface Eu layer in the experimental spectrum. That is, the Eu 4f emissions from the several Eu underlayers, which form the second Eu surface layer in the experimental spectrum, are not shifted each other in energy. Moreover the photoemission–intensity ratio of the second to the first surface Eu layers changes at each sample temperature. This indicates that the sample-surface condition considerably differs at each sample temperature due to the scraping. Finally, the evaluated Eu mean valence at temperatures from 20 to 300 K are shown by closed circles in Fig. 3,
2.3 2.4
Eu valence
532
2.5 2.6 2.7
EuPd2Si2
2.8 0
100
200
300
Temperature (K) Fig. 3. Temperature dependence of the Eu mean valence evaluated by fitting the Eu 4f spectra (closed circles). Solid line is a guide to the eye. Open triangles indicate the Eu mean valence evaluated from the analysis of the Eu LIII -edge XAS spectra [2].
together with that obtained from the Eu LIII -edge XAS measurements indicated by open triangles [2]. The Eu mean valence was evaluated by using the integrated intensities of the bulk Eu2+ and Eu3+ 4f components. We have estimated the Eu mean valence to be 2.75 ± 0.03 at 20 K and 2.30 ± 0.05 at 300 K. These values agree well with the values evaluated from the Mössbauer and Eu LIII -edge XAS measurements [1–3]. The obtained Eu mean valence also exhibits the abrupt decrease with increasing temperature. The estimated transition temperature, however, shifts by about 50 K toward the lower temperature side than those from the Mössbauer and Eu LIII -edge XAS measurements [1–3]. One of the possible reasons is that the existence of the large second surface Eu2+ 4f emission has an influence on the estimation of the Eu mean valence. In order to suppress the second surface Eu2+ 4f emission we will investigate the temperature dependence of the bulk-sensitive high-resolution photoemission spectra for the fractured sample.
4. Conclusions We have investigated the Eu 4f states of EuPd2 Si2 by means of the bulk-sensitive high-resolution photoemission spectroscopy from 20 to 300 K. We succeeded in observing the Eu2+ 4f states of the bulk as well as those of two surface layers in the spectra. The temperature-dependent photoemission spectra show the changes in the spectral intensity of the bulk Eu2+ and Eu3+ 4f components and in the energy separation between the Eu2+ and Eu3+ 4f components, through the temperature-induced valence transition of EuPd2 Si2 . The Eu mean valence has been evaluated by using the integrated intensities of the bulk Eu2+ and Eu3+ 4f contribution. The evaluated Eu mean valence is 2.75 ± 0.03 at 20 K and 2.30 ± 0.05 at 300 K. These values are in good agreement with those from the Mössbauer and Eu LIII -edge XAS measurements [1–3]. The estimated transition tem-
K. Mimura et al. / Journal of Electron Spectroscopy and Related Phenomena 137–140 (2004) 529–533
perature, however, is about 50 K lower than those from the Mössbauer and the Eu LIII -edge XAS [1–3], because of the influence of the strong Eu2+ emission from the second surface Eu layer upon the Eu mean-valence estimation. In order to understand the mechanism for the temperature-induced valence transition of EuPd2 Si2 , we are planning to carry out the temperature-dependent high-resolution photoemission measurements for the fractured sample by using several hν’s and to obtain further information on the electronic structures of the conduction electrons such as Pd 4d, Si 3p and Si 3s as well as those of Eu 4f.
Acknowledgements We are grateful to Prof. S. Kimura for useful discussions. We also thank Messrs K. Mantani, K. Kitamoto and Ms. Y. Watanabe for their assistance. The photoemission measurements were performed at SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal No. 2001B0396-NS-np).
References [1] E.V. Sampathkumaran, L.C. Gupta, R. Vijayaraghavan, K.V. Gopalakrishnan, R.G. Pillay, H.G. Devare, J. Phys. C: Solid State Phys. 14 (1981) L237.
533
[2] E. Kemly, M. Groft, V. Murgai, L.C. Gupta, C. Godart, R.D. Parks, C.U. Segre, J. Magn. Magn. Mater. 47&48 (1985) 403. [3] G. Wortmann, K.H. Prank, E.V. Sampathkumaran, B. Perscheid, G. Schmiester, G. Kaindl, J. Magn. Magn. Mater. 49 (1985) 325. [4] H. Wada, A. Mitsuda, M. Shiga, H. Aruga Katori, T. Goto, J. Phys. Soc. Jpn. 65 (1996) 3471. [5] A. Mitsuda, H. Wada, M. Shiga, H. Aruga Katori, T. Goto, Phys. Rev. B 55 (1997) 12474. [6] N. Mårtensson, B. Reihl, W.-D. Schneider, V. Murgai, L.C. Gupta, R.D. Parks, Phys. Rev. B 25 (1982) 1446. [7] W.D. Schneider, C. Laubschat, G. Kalkowski, J. Haase, A. Puschmann, Phys. Rev. B 28 (1983) 2017. [8] A. Sekiyama, T. Iwasaki, K. Matsuda, Y. Saitoh, Y. Onuki, S. Suga, Nature 403 (2000) 396. [9] Y. Saitoh, H. Kimura, Y. Suzuki, T. Nakatani, T. Matsushita, T. Muro, T. Miyahara, M. Fujisawa, K. Soda, S. Ueda, H. Harada, M. Kotsugi, A. Sekiyama, S. Suga, Rev. Sci. Instrum. 71 (2000) 3254. [10] J.J. Yeh, I. Lindau, Atom. Data Nucl. Data Tables 32 (1985) 1. [11] S. Tanuma, C.J. Powel, D.R. Penn, Surf. Sci. 192 (1987) L849. [12] S. Kimura, M. Okuno, H. Iwata, T. Saitoh, T. Okuda, A. Harasawa, T. Kinoshita, A. Mitsuda, H. Wada, M. Shiga, J. Phys. Soc. Jpn. 71 (2002) 255. [13] T. Kinoshita, H.P.N.J. Gunasekara, Y. Takata, S. Kimura, M. Okuno, Y. Haruyama, N. Kosugi, K.G. Nath, H. Wada, A. Mitsuda, M. Shiga, T. Okuda, A. Harasawa, H. Ogasawara, A. Kotani, J. Phys. Soc. Jpn. 71 (2002) 148. [14] F. Gerken, J. Phys. F: Met. Phys. 13 (1983) 703. [15] E. Weschke, A. Höhr, S. Vandré, C. Schüßler-Langeheine, F. Bødker, G. Kaindl, J. Electron. Spectrosc. Relat. Phenom. 76 (1995) 571. [16] G.D. Mahan, Phys. Rev. B 11 (1975) 4814. [17] K. Mimura et al., unpublished.