Band inversion and topology of the bulk electronic structure in FeSe 0.45 Te 0.55 (original) (raw)

FeSe 0.45 Te 0.55 (FeSeTe) has recently emerged as a promising candidate to host topological superconductivity, with a Dirac surface state and signatures of Majorana bound states in vortex cores. However, correlations strongly renormalize the bands compared to electronic structure calculations, and there is no evidence for the expected bulk band inversion. We present here a comprehensive angle resolved photoemission (ARPES) study of FeSeTe as function of photon energies ranging from 15-100 eV. We find that although the top of bulk valence band shows essentially no k z dispersion, its normalized intensity exhibits a periodic variation with k z. We show, using ARPES selection rules, that the intensity oscillation is a signature of band inversion indicating a change in the parity going from Γ to Z. Thus we provide the first direct evidence for a topologically non-trivial bulk band structure that supports protected surface states. PACS numbers: 74.25.Jb, 74.70.Dd, 71.20.Be Iron-based superconductors (FeSCs) have been intensely investigated since their discovery in 2008 [1] as strongly correlated materials that harbor high temperature superconduc-tivity. Recently, interest in this field has increased greatly due to new experiments that suggest that some of these systems may be topological superconductors [2] that harbor Majorana bound states (MBS) in their vortex cores, which could be potentially important for quantum information processing [3]. Wang et al. [9] first suggested that FeSe 0.5 Te 0.5 (FeSeTe) can host topologically protected Dirac surface states, which were recently observed directly using angle resolved photoe-mission spectroscopy (ARPES) [10]. Soon after, such states were found in other FeSCs [11] and in thin films [7]. In addition , clear zero bias conductance peaks (ZBCP) were observed [8, 9] in the superconducting vortex cores in FeSeTe using scanning tunneling spectroscopy (STS), and identified as the MBS expected in topological superconductors. In fact, the strong correlations in these materials, which leads to surprisingly large ∆/E F ratios [10, 11], helps in separating the ZBCP from (topologically) trivial vortex core bound states. Despite these exciting developments, direct evidence for the topological nature of the bulk band structure-responsible for the topologically protected surface states and MBS-is lacking. Density functional theory (DFT) calculations [9] for FeSeTe find a p z band that is highly dispersive along k z , which mixes with an appropriate linear combination of the d xz,yz bands. As a result, the orbital character and the parity of the band changes as one goes from Γ(0,0,0) to Z(0,0,π/c). However, no such highly dispersive band is observed in the data, as we shall show below, and-at first sight-there seems to be no evidence for the band inversion expected in a topo-logically nontrivial bulk band structure. FeSeTe is known to be the most strongly correlated member of the FeSC family [12, 13], making it difficult to directly compare ARPES measurements with DFT. It offers an exciting opportunity to study the interplay between the topological nature of the band structure and the effect of the strong electronic correlations. In this letter, we present a systematic ARPES study of FeSeTe for a broad range of incident photon energies (15 to 100 eV) to investigate the k z-dispersion of the bulk electronic structure. Using symmetry analysis and dipole selection rules, we present clear evidence for the change in the parity eigenvalue going from Γ to Z, in spite of the absence of any highly dispersive band. We also present a tight-binding model, with reasonable values of renormalization parameters relative to DFT and of spin-orbit coupling, which gives insight into ARPES observations. We thus provide compelling evidence for bulk "band inversion", the hallmark of a topo-logical band structure via the Fu-Kane invariant [14], which leads to a protected Dirac surface state in the energy gap near the Γ point. We used high quality Fe 1.02 Se 0.45 Te 0.55 single crystals for ARPES measurements. Fig. 1(a,b) shows the geometry of our ARPES experiments. We will focus on near-normal emission with (k x , k y) near (0, 0), and light incident in the YZ plane in either LV (linear vertical) or LH (linear horizontal) po-larizations, as shown. This geometry will be crucial in the analysis of the selection rules later in the paper. Our laboratory axes (X,Y, Z) conform with the literature [10, 11], however , we label orbitals with reference to the crystallographic axes (x, y, z), irrespective of sample rotations, consistent with Refs. [5, 6, 15]. We show ARPES data along the Γ-M direction using 22 eV LV photons in Fig. 1 (c), and its second derivative [18] sharpened image in panel (d). This allows us to see in addition to a dispersive bulk band, which we label as α 1 , an intense state at a binding energy (BE) of around 10 meV, that lies between the top of the α 1 band (BE 18 meV) and the chemical potential (BE = 0 meV). This state is similar to the linearly dispersive Dirac surface state (SS), recently been reported by Zhang et. al. [10]. In Fig. 1 (e) we show LH polarization data where in addition to the states seen in LV data of panel (c), we also see another dispersive α 2 band. The ARPES intensity allows a direct mapping of the electronic dispersion for momenta parallel to the sample surface.