Spectromicroscopy reveals | |
|
A. Goldoni, R. Larciprete, S. Lizzit
In January 2001, Japanese researchers announced [1] the discovery of superconductivity at 39 K in MgB2 (Magnesium Diboride) a material that has been known since the 1950s. Although the observed superconducting transition temperature (Tc) is about 4 times lower than the highest Tc obtained in cuprates, this discovery has caused considerable excitement among the scientists for at least two reasons. First, MgB2 crystallizes in a layered structure formed by alternating graphite-like honeycomb boron layers and hexagonal Mg planes [1] [2], a crystal structure that was generally considered unfavorable for superconductivity (the maximum superconducting transition temperature in similar graphite-like intercalated compounds is, indeed, a few Kelvin degrees). Whereas the high transition temperature observed in MgB2 might imply exotic coupling mechanisms, most indications now suggest that MgB2 is the ultimate strong-coupled electron-phonon superconductor. In such a system, the charge carriers are paired together by lattice vibrations as explained by the BCS theory of superconductivity [3]. This is the fulfillment of a prediction made long ago for boride superconductors: the light mass of boron should bring to high-energy lattice vibrations and therefore high-Tc within the BCS theory framework. Second, being MgB2 a chemically stable bulk compound and being commercially available in great quantities at low cost, there are tremendous promises for everyday applications. Understanding the electronic properties of MgB2 is crucial in order to address the promise of everyday applications and to search for conventional materials with possibly higher Tc. Several electronic structure calculations have already been carried out, and all agree in predicting a dominant contribution of boron states near the Fermi level, while the Mg atoms act as electron reservoir donating their s-electrons to the boron-derived bands [4-8]. The in-plane B sigma-bands retain a localized covalent bonding character, but should exhibit metallic hole-type conductivity, as opposed to the non-bonding pi-bands possessing a typical delocalized character and an electron-type conductivity [4]. Due to their two-dimensional character the sigma-bands should give the highest contribution to the density of states near the Fermi level [4-8]. In addition, because of the small dispersion of the sigma-bands, electron correlation effects should play an important role [4-6]. Finally, it has been argued that the B planes are akin to the Cu-O planes in high-Tc cuprates [6]. The doping of the MgB2 bands with holes and/or electrons (for example by intercalation of alkali metals or halogen atoms) should mirror the behaviour of underdoped and/or overdoped high-Tc cuprates [6]. In spite of these band structure calculations only few reliable experimental investigations have been performed to directly corroborate these theoretical predictions [9-11]. The main reason is the difficulty of having good samples for electron spectroscopies, in particular for photoemission, a uniquely powerful probe of the filled electronic states. This technique is surface sensitive and the surface of MgB2 samples can be very different from the bulk, as well as unstable, due to the tendency of Mg to migrate toward the surface and to form oxides. To the best of our knowledge, apart the high-resolution measurements of the superconducting gap [12] [13], only one photoemission experiment on MgB2 has been published. However, the experimental data obtained do not show a full agreement with the density of states calculations, a contradiction that was attributed to the presence of strong correlation effects [11]. Here we show, for the first time, that by exploiting the capabilities of photoemission microscopy at sub-micron lateral resolution and using pressed pellets of commercial MgB2 powder with single crystalline grains of micrometer size, it is possible to provide a reliable measurement of the density of states (DOS) in MgB2, directly comparable with the theoretical predictions. The photoemission microscopy experiments were performed at the Spectromicroscopy beamline with a lateral resolution of 0.5 µm, therefore, obtaining information on single crystalline grains. The MgB2 pellets were sputtered in UHV by Ar+ ions at 500 eV for hours, in order to remove oxide species, and measured at 90 K after the sputtering. The inset of Fig. 1a shows a 3x3 mm2 image obtained by scanning the sample and collecting the photoemission signal integrated in an energy window of 1.2 eV around the Fermi level with hn=95 eV. This image shows a bright spot of micrometer size (therefore consistent with a single crystalline grain) surrounded by a dark background. The valence band photoemission measured on the dark area has a lineshape similar to that already reported in literature [11] with a low intensity near the Fermi level and no clear Fermi edge. On the contrary, as shown in Fig.1a, the valence band measured on the bright spot shows an intense feature near the Fermi level and a clear Fermi edge. Fig. 1b reports the projected MgB2 DOS calculated in local density approximation [8]. These DOS are similar to and representative of many other mean field DOS calculations published for MgB2 [4-7]. The agreement between the valence band spectrum shown in Fig. 1a and the calculated DOS reported in Fig. 1b is very good considering, in particular, that the calculated B 2s contribution must be multiplied by a factor of ~6 to account for the different photoemission cross-section compared to 2p-states. It is worth noting that the lineshape of the p-derived feature near the Fermi level we measure in photoemission (surface sensitive technique) is in very good agreement also with that obtained using x-ray absorption fluorescence (bulk sensitive technique) [9,10]. Finally, in Fig. 2 we show the effect of surface degradation under the intense synchrotron radiation beam. After about 20 min (2 min. per spectrum) the intensity in the Fermi level region is strongly reduced and the valence band lineshape tends to become similar to that measured on the dark area of the inset of Fig. 1a. In correspondence to this behaviour there is a shift of the valence band features toward higher binding energy and an increase of the Mg 2p core level emission (not shown here). This suggests a migration of Mg atoms toward the surface under the synchrotron beam, with a consequent electron doping of the surface layer, responsible of the Fermi level shift. In conclusion, the close overall agreement between our photoemission valence band spectrum and the calculated DOS in local density approximation over a wide range of binding energies, indicates that the expected electron-electron correlation effects in MgB2 are instead extremely weak. MgB2 looks like a conventional metal and the mechanisms for superconductivity are most likely of conventional (electron-phonon) nature. Moreover, we have shown that the surface is unstable under the synchrotron radiation beam. The changes we observe in the valence band spectra during the beam exposure are likely caused by Mg migration toward the surface. |
|
 |
Fig. 1: valence band photoemission spectrum of MgB2 measured on the bright spot shown in the inset. The inset is a 3x3 µm2 image obtained by scanning the sample and collecting the photoemission signal integrated in an energy window of 1.2 eV around the Fermi level. The bright spot likely corresponds to a single crystalline grain. b) The calculated B 2s and B 2p contribution to the valence band surface-DOS of boron terminated MgB2.The Mg contribution is about a factor of 10 lower and almost flat [8]. |
 |
Fig. 2: effect of the exposure to the synchrotron photon beam on the valence band photoemission spectra of MgB2 measured on the bright spot shown in the inset of Fig. 1a. The total exposure time is about 20 min (2 min per spectrum). The exposure time increases with the spectrum reference numbers. |
|
Correspondence and requests for materials should be addressed to A. Goldoni
|
|