Ultrahigh-resolution photoemission spectroscopy of superconductors using a VUV laser

Journal of Electron Spectroscopy and Related Phenomena 144–147 (2005) 953–956

Ultrahigh-resolution photoemission spectroscopy of superconductors using a VUV laser T. Kiss a, 1 , T. Shimojima a , F. Kanetaka a , K. Kanai a, 2 , T. Yokoya a, 3 , S. Shin a, b, ∗ , Y. Onuki c , T. Togashi b , C.Q. Zhang d , C.T. Chen d , S. Watanabe a a Institute for Solid State Physics, University of Tokyo, Kashiwa, Chiba 277-8581, Japan The Institute of Physical and Chemical Research (RIKEN), Sayo-gun, Hyogo 679-5143, Japan c Graduate School of Science, Osaka University, Toyonaka, Osaka 560-0043, Japan Beijing Center for Crystal R&D, Chinese Academy of Science, Zhongguancun, Beijing 100080, China b

d

Available online 19 February 2005

Abstract We have developed an ultrahigh-resolution photoemission system using a quasi continuous wave (quasi-CW) VUV laser. The photon energy of the VUV laser is 6.994 eV. It has a line width of 260 ␮eV and a repetition rate of 80 MHz. The photon energy of this laser is the highest for a quasi-CW laser. Using a quasi-CW laser removes the space charge effect and hence does not degrade the energy resolution. For the electron analyser, we use the newly designed ultrahigh-resolution photoelectron analyser GAMMADATA-SCIENTA R4000. The analyser resolution is about 250 ␮eV with a wide collecting angle of about 38◦ . The total energy resolution of this system is measured to be 360 ␮eV for the Fermi edge of gold. Using this system, we have measured several low transition temperature superconductors and succeeded to observe the bulk superconducting state of these materials. © 2005 Elsevier B.V. All rights reserved. Keywords: Ultrahigh-resolution photoemission spectroscopy; Quasi-CW laser; Superconducting gap; Bulk electron structure

1. Introduction It is well known that the variety of properties that conducting materials display stems from the electronic structure near the Fermi level (EF ). In the case of superconductors, the superconducting energy gap is one such fundamental parameter. Photoemission spectroscopy has played important roles in the study of electronic structures, especially the structures near the EF of high temperature superconductors (high-Tc ). For example, the dx2 −y2 symmetry of the superconducting ∗

Corresponding author. Tel.: +81 471 36 3381; fax: +81 471 36 3383. E-mail address: [email protected] (S. Shin). 1 Present address: The Institute of Physical and Chemical Research (RIKEN), Wako, Saitama 351-0198, Japan. 2 Present address: Department of Chemistry, Nagoya University, Nagoya, Aichi 464-8602, Japan. 3 Present address: Japan Synchrotron Radiation Research Institute (JASRI), Sayo-gun, Hyogo 679-5198, Japan. 0368-2048/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.elspec.2005.01.154

gap [1,2] and the anisotropic pseudogap [3] are well known. To observe more detailed and smaller energy scale structures such as the superconducting gap of low-Tc superconductors, we have previously constructed an ultrahigh-resolution photoemission spectrometer based on a He discharge lamp which has an energy resolution of 1.4 meV and can achieve a minimum temperature of 5.3 K [4]. By using this system, we have measured the superconducting gaps of simple metals and phonon induced structure [4,5], the anisotropy of superconducting gap in borocarbides [6], the Fermi sheet-dependent superconducting gaps in 2H–NbSe2 [7] and two superconducting gaps in MgB2 [8] so far. However, the escape depth of photoelectron shows strong photon-energy dependence [9]. The escape depth of photoelectrons using a He discharge ˚ However, such lamp (He I␣: 21.218 eV) is less than 10 A. studies have not been able to provide reliable data for felectron superconductors which exhibit a difference between surface and bulk electronic structure. For example, more bulk sensitive photoemission spectroscopy using soft X-ray exci-

954

T. Kiss et al. / Journal of Electron Spectroscopy and Related Phenomena 144–147 (2005) 953–956

Fig. 1. Schematic diagram of the photoemission spectrometer system using a laser as the photon source (laser–PES) for demonstration of sub-meV energy resolution. The second harmonic of a quasi-CW frequency-tripled Nd:YVO3 laser produced by using an optically-contacted prism-coupled KBe2 BO3 F2 (KBBF) crystal is focused on a sample with a CaF2 lens [23]. The kinetic energies of electrons emitted from the sample are measured with a high-precision hemispherical electron analyser (GAMMADATA-SCIENTA R-4000). The photoemission spectrometer and the laser systems are vacuum-separated by a CaF2 view port, through which the resultant 6.994 eV light can be transmitted.

tation was reported on f-electron superconductor CeRu2 , and confirmed the difference between surface and bulk [10]. But the energy resolution was not enough to observe superconducting electronic structure. To observe such a small energy scale bulk character structure, we have constructed a new photoemission spectrometer, which has higher energy resolution, lower sample temperature and is more bulk sensitive. In this paper, we present ultrahigh-resolution photoemission spectra of low-Tc superconductors obtained with a newly constructed photoemission spectrometer using a quasi-CW VUV laser with the highest energy resolution of 360 ␮eV and the lowest measurement temperature of 2.7 K. Due to the energy resolution and low temperature, we can now study superconducting gaps of the superconductors that have been impossible to observe due to the limitation of energy resolution and temperature so far. And due to enhanced bulk sensitivity at low kinetic energy, we could observe clear opening of the superconducting gap of f-electron superconductor CeRu2 . The present study thus establishes the role of PES as a powerful tool to investigate electronic structures with small energy scales even in the low-Tc superconductors.

2. Experiment Fig. 1 shows the schematic diagram of the newly constructed ultrahigh-resolution photoemission spectrometer using a VUV laser. To get higher energy resolution we use the newly designed GAMMADATA-SCIENTA R4000 hemispherical electron analyser. The total energy resolution is, of course, determined by the energy resolution of electron

analyser and the line width of photon source. Using a He discharge lamp with a line width of about 1 meV, we cannot reach better than 1 meV energy resolution. So we have used a VUV laser that has narrower line width. But if an intense pulsed laser is used, the emitted photoelectrons can lose true density of states information owing to the longrange interaction of space charge effects [11]. To avoid the space charge effect, a quasi-CW laser with high repetition rate is more useful. We obtain the second harmonic light of the frequency tripled Nd:YVO4 laser by using the opticallycontacted prism-coupled non-linear crystal KBBF. This gives the highest photon energy of 6.994 eV for a quasi-CW laser. This photon energy, which in turn yields photoelectrons in the 1–2 eV energy range, corresponds to a typical photoelectron ˚ [9]. For cooling samescape depth of more than ∼100 A ples, we use a pumped liquid He continuous flow cryostat with newly designed triple-walled radiation shields. The lowest temperature of this system is about 2.7 K. Fig. 2 shows an ultrahigh-resolution measurement of a gold Fermi edge spectrum. The solid line shows the convolved Fermi–Dirac (FD) function with the total energy resolution of 360 ␮eV at a temperature of 2.7 K. These values represent the highest energy resolution and the lowest temperature in photoemission spectroscopy of solids that have been measured so far. Using this system, we can measure the bulk and detailed electronic structures such as the superconducting gap of low-Tc superconductors. The sample temperature was measured using silicon-diode sensors. The base pressure of the measurement chamber was less than 2 × 10−11 Torr. Clean surfaces were obtained by scraping in situ with a diamond file for Nb and fracturing

T. Kiss et al. / Journal of Electron Spectroscopy and Related Phenomena 144–147 (2005) 953–956

Fig. 2. Ultrahigh-resolution PES spectrum of an evaporated gold film measured at 2.7 K, together with the FD function at 2.7 K convolved by a Gaussian with full width at half maximum of 360 ␮eV. A total energy resolution of 360 ␮eV is confirmed from the very good match between the experimental and calculated spectra.

in situ for CeRu2 . The EF of samples was referenced to that of a gold film evaporated onto the sample substrate and its accuracy is estimated to be better than ±50 ␮eV.

3. Results and discussion The open circles in Fig. 3 shows the photoemission spectrum of Nb measured by a He discharge lamp system at 5.3 K [5] and the solid line shows the same one measured by laser–PES at 3 K. Both spectra show the shift of leading edge and quasi particle peak owing to the superconductivity. The quasi particle peak is much clearer and sharper in the spectrum measured by laser–PES system. It shows the large progress of energy resolution and low temperature compared with previous system. We have estimated the value of the gap using a Dynes’ function [12] fit to the peak in the density of states (DOS). We obtain a value of ∆(3 K) = 1.64 meV and a Γ = 0.14 meV for the spectra measured by laser–PES system. We thus obtain the value of 2∆(0)/kB Tc for this spectrum of about 4, where ∆ is the superconducting gap value and kB , the Boltzmann constant which is larger than the previous value of 3.7 that was measured by He I [5]. We believe the difference between these values to be the difference between surface and bulk electronic structure. This is possibly the first report for the difference between surface and bulk electronic structure for the superconducting gap size. To further demonstrate the bulk sensitivity of our measurement, the upper solid curve in Fig. 4 shows the photoemission spectra of Nb with a clean surface. The open circle in Fig. 4 shows the same spectrum of Nb but without surface preparation in the vacuum. Both the spectra were measured with the same energy resolution, temperature, and scan time. The measurements were carried out with an energy resolution of 1.2 meV and at the temperature of 3.8 K. Both spectra show similar shapes with the shift of leading edge and quasi particle peak. The signal-to-noise ratio of the spectrum of dirty surface becomes worse compared with the clean sur-

955

Fig. 3. Ultrahigh-resolution PES spectrum of Niobium measured by a He discharge lamp system (open circle) and a laser system (solid line). The electronic structure in the superconducting phase is clear for the measurement using laser. Dashed line is a result of fitting with Dynes’ function. The size of the superconducting gap (∆) is estimated to be 1.64 meV, and Γ is the thermal broadening parameter.

face. However, the overall features of both spectra seem the same. This result clearly shows that photoemission spectra using VUV laser reflect bulk electron state rather than the surface state. Next we measured the f-electron superconductor CeRu2 to study superconducting electron state and demonstrate the advantage of VUV laser photoemission spectroscopy. Most intermetallic materials containing cerium and uranium show anomalous properties originating from interplay of localized and itinerant nature of the f-electrons. Even below the superconducting transition temperature, the f-electron systems show unconventional superconducting properties that cannot be explained by the simple BCS theory assuming an isotropic energy gap, most likely due to the strong correlation between f-electrons. Indeed, anomalous superconducting properties suggestive of anisotropic order parameters have been reported for f-electron superconductors from thermodynamic and magnetic studies [13]. Among the f-electron superconductors, CeRu2 has the highest Tc of 6.2 K [14]. The symmetry of the order parameter as suggested from superconducting properties is controversial. While early specific heat

Fig. 4. Ultrahigh-resolution PES spectrum of Niobium with surface preparation (solid circle) and without surface preparation (open circle). These two spectra seem almost same except for signal to noise ratio owing to the count rate.

956

T. Kiss et al. / Journal of Electron Spectroscopy and Related Phenomena 144–147 (2005) 953–956

superconducting gap in an f-electron superconductor using photoemission spectroscopy. The superconducting spectrum is again analysed with a Dynes’ function. However, we cannot fit this spectrum using this function, suggesting anisotropic superconducting gap of CeRu2 .

4. Conclusion

Fig. 5. Ultrahigh-resolution PES spectra of CeRu2 . (a) T-Dependent ultrahigh-resolution spectra near EF . The spectra were measured using the He discharge lamp system with He I␣ (21.218 eV photon energy). (b) TDependent ultrahigh-resolution spectra near EF . The spectra were measured using the VUV laser with 6.994 eV photon energy.

measurements have suggested an axial symmetry with a line node [15], later studies reported an isotropic s-wave gap [16]. More recently, from specific heat and magnetization measurements, Hedo et al. concluded that CeRu2 is a BCS type superconductor, with magnetic field dependence of the specific heat at 0.5 K exhibiting a H0.5 -like behaviour in low fields [17]. The specific heat behaviour is similar to Ni borocarbide superconductors, but for which existence of a line node has been discussed [18]. From measurements of the nuclear lattice relaxation rate, Matsuda et al. [19] and Ishida et al. [20] have suggested that CeRu2 is an s-wave superconductor with a finite gap having 2∆/kB Tc = 3.8–4.0. However, Mukuda et al. [21] have shown from impurity effects that the superconductivity is described in terms of an anisotropic s-wave gap. Tunnelling studies have provided scattered gap values of 0.8–3.1 meV, corresponding to 2∆/kB Tc = 3.3–6.6 (Refs. [18–21]). This is most likely due to fast degradation of sample surfaces and/or different electronic states in the surface region [22], as shown by soft-X ray PES [10]. Thus, a bulk sensitive technique that directly measures the electronic structure with an extremely high-energy resolution is essential to understand the superconducting electronic structures of CeRu2 . Fig. 5 (a) shows the temperature-dependent photoemission spectra of CeRu2 . High quality single crystals of CeRu2 with the residual resistivity ratio of ∼270 were grown with the Czochralski pulling method in a tetra-arc furnace. The details are described in Ref. [17]. The Tc of 6.2 K was determined from magnetization measurements. Using the system of He discharge lamp and lowest temperature of 4.5 K, the spectra measured at 4.5 K can be supposed to be superconducting state, but we cannot observe a clear shift of the leading edge and the quasiparticle peak. Fig. 5b shows the temperaturedependent ultrahigh-resolution photoemission spectra across Tc using VUV laser PES with a total energy resolution of 800 ␮eV. While the spectrum at 7 K in normal phase has a clear Fermi edge, the spectrum at 3.5 K in the superconducting phase shows a sharp peak at 1.4 meV with a leading-edge shift to higher binding energy, indicative of the opening of the superconducting gap. This is the first observation of a

We have constructed a photoemission system using a VUV laser as a photon source and having a total energy resolution of 360 ␮eV and lowest temperature of 2.7 K. This system has high bulk sensitivity, so samples can be measured without surface cleaning. We have used this system to measure the bulk-superconducting gap of the f-electron superconductor CeRu2 . The present PES study indicates an anisotropic superconducting gap in CeRu2 . These results demonstrate that photoemission spectroscopy using a VUV laser can be a powerful tool for studying the superconducting electronic structures of correlated materials.

Acknowledgements We thank Drs. M. Hedo, A. Chainani, and M. Taguchi for valuable discussion. This work was supported by Grant-inaid from the Ministry of Education, Science, and Culture of Japan. T.K. thanks the Japan Society for the Promotion of Science (JSPS) for financial support.

References [1] Z.-X. Shen, et al., Phys. Rev. Lett. 70 (1993) 1553. [2] T. Yokoya, et al., Phys. Rev. B 51 (1995) 3495. [3] H. Ding, et al., Nature 382 (1996) 51; A.G. Loeser, et al., Science 273 (1996) 325. [4] T. Kiss, et al., J. Electron Spectrosc. Relat. Phenom. 114–116 (2001) 635. [5] A. Chainani, et al., Phys. Rev. Lett. 85 (2000) 1966. [6] T. Yokoya, et al., Phys. Rev. Lett. 85 (2000) 4952. [7] T. Yokoya, et al., Science 294 (2001) 2518; T. Kiss, et al., Physica B 312–313 (2002) 666. [8] S. Tsuda, et al., Phys Rev. Lett. 87 (2001) 17006. [9] M.P. Seah, W.A. Dench, Surf. Interface Anal. 1 (1979) 2. [10] A. Sekiyama, et al., Nature 403 (2000) 369–398. [11] P. Balzers, personal communication. [12] R.C. Dynes, et al., Phys. Rev. Lett. 41 (1978) 1509. [13] M. Sigrist, K. Ueda, Rev. Mod. Phys. 63 (1991) 239–311. [14] S.B. Roy, P. Chaddah, Pramana J. Phys. 53 (1999) 659–684. [15] J.G. Sereni, et al., Mod. Phys. Lett. 3 (1989) 1225–1231. [16] A.D. Huxley, et al., J. Phys: Condens. Matter 5 (1993) 7709– 7718. [17] M. Hedo, et al., J. Phys. Soc. Jpn. 67 (1998) 272–279. [18] M. Nohara, et al., J. Phys. Soc. Jpn. 66 (1997) 1881–1891. [19] K. Matsuda, et al., J. Phys. Soc. Jpn. 64 (1995) 2750–2753. [20] K. Ishida, et al., Z. Naturforsch. 51a (1996) 793–796. [21] H. Mukuda, et al., J. Phys. Soc. Jpn. 67 (1998) 2101–2106. [22] H. Sakata, et al., J. Phys. Soc. Jpn. 69 (2000) 1970–1973. [23] T. Togashi, et al., Opt. Lett. 28 (2003) 254.