Correlated and non-correlated topological insulators

Correlated and non-correlated topological insulators

Katharina Höfer,1 Christoph Becker,1 Diana Rata,1 Deepa Kasinathan,1 Maurits W. Haverkort,1 Peter Thalmeier,1 Steffen Wirth,1 and Liu Hao Tjeng1

1Max-Planck-Institut für Chemische Physik fester Stoffe, 01187 Dresden, Germany

Topological insulators (TI) form a novel state of matter that open up new opportunities to create unique quantum particles. Many exciting experiments have been proposed by theory, but still await their experimental verification, not to mention their implementation into applications. Today, the main obstacle is material quality and cleanliness of the experimental conditions. The charge carrier concentration of the topological surface states (SS) is only about 1012 cm-2. Thus the extrinsic conductivity arising from the presence of tiny amounts of vacancies and defects in the bulk, and the contamination of the surface at ambient atmosphere, will mask the special properties of the TI.

For Bi2Te3 thin films we were able to identify the conditions under which molecular beam epitaxy (MBE) gives stable and reproducible results for fabrication single domain and bulk insulating samples [Höfer2014]. Hereby, the key factors are to ensure a full distillation of the Te while maintaining a Te/Bi ratio of ~8, and the usage of substrates with negligible lattice mismatch, namely highly insulating BaF2 (111). The optimal conditions for MBE growth of Bi2Te3 were found at a substrate temperature of 250C with a Bi flux of ~1 Å/min. Angle resolved photoemission spectroscopy (ARPES) of a 10 QL Bi2Te3 film, recorded in the vicinity of the Fermi level, clearly displays the characteristic linear dispersion of the Dirac states, cf. Fig. 1.

Dirac states, cf. Fig. 1. The ARPES results provide clear evidence for the Fermi level being located well within the band gap and being only intersected by the topological surface states. The absence of a bulk conduction band signal in the ARPES spectra recorded at room temperature ensures that the electrical conductance at low temperatures is solely given by the Dirac surface states.

Special efforts have been made to provide the capability to carry out temperature dependent four-point conductance measurements in situ, i.e. under UHV conditions, in addition to ARPES. Only in this way is the integrity of the topological surface states preserved and reliable results can be achieved. The in situ electrical contacts are realized through a homemade, non-permanent point contact mechanism, using spring-loaded probes. Bi2Te3 films with thicknesses ranging from 10 to 50 QL have been fabricated in an effort to unambiguously clarify whether the conductivity is truly arising from the surface. The results are displayed in Fig. 2. The inset of Fig. 2A shows the voltage-current characteristics at selected temperatures. The linear relation demonstrates that the prepared contacts are ohmic over the entire temperature range measured. All samples display a metallic-like behavior with temperature. Furthermore, no significant changes between the results obtained during cool-down and warm-up of the samples are observable, demonstrating the excellent stability of the film and the fabricated contacts. Comparing the sheet resistance at different film thicknesses (Fig. 2B), a scattering with the tendency of lower resistance with thicker films can be noticed.

However, the resistance is not inversely proportional to the film thickness; instead, it varies only by a factor of 1.3 at 14 K and 1.5 at 295 K, upon investigating film thicknesses from 10 to 50 QL. This provides evidence for the fact that the surface indeed dominates the transport properties of the Bi2Te3 films. The observation of a resistance that is practically constant for thicknesses beyond 20 – 30 QL hints towards the possibly of some minute amounts of defects which may have formed during the initial stages of the growth. The obtained charge carrier concentrations, estimated from the ARPES spectra of each sample, are in the expected range for TI of 2 - 41012 cm-2 corresponding to very high average mobilities of about 3000 cm2/Vs, with the maximum value of 4600 cm2/Vs at 14 K, see Fig. 2C.

Recently, we have also been able to establish a procedure by means of which the Bi2Te3 films can be capped such that they do not suffer from exposure to air [Höfer2015]. This capping can be easily removed, still leaving the properties of the Bi2Te3 films intact. This will allow us future access to many more investigative tools.

For the topological properties of Bi2Te3 to arise, a significant spin-orbit coupling is required while electronic correlations do not play any appreciable role. However, it was recognized recently that also in materials in which strong electronic correlations are present topologically protected surface states may have to form. This applies particularly for the so called Kondo insulators, or strongly correlated semiconductors. In these materials the spin-orbit coupling is typically sizeable, of the order of 0.5 eV. Moreover, the more localized f-states exhibit odd parity while the conduction band is usually of a more d-like character. Some dispersion of the f states is required though in order to open up an energy gap at the Fermi level; in this respect the Kondo insulators are well suited.

The intermediate valence system SmB6 is such a material in which hybridization plays a decisive role for its low-temperature properties—in particular an insulating ground state due to the formation of a gap at the Fermi energy—to evolve.

We conducted scanning tunneling microscopy (STM) and spectroscopy (STS) on SmB6 [Rößler2014]. Here we made use of the unique capabilities of this technique to relate spectroscopic results unambiguously to specific topographies.

This assignment turned out to be of utmost importance for an interpretation of the different spectroscopic results which can not be achieved by other methods like ARPES. The hybridization gap, inter-multiplet transitions and possible crystal field excitations are observed by STS [Rößler2014]. The temperature evolution of these spectra again points toward the Kondo effect being at play. However, a plethora of different surface terminations is found; one example of a disordered, reconstructed surface is presented in Fig. 3. These different surface terminations were found to exhibit profoundly different spectroscopic properties. Here, we take advantage of the unique capability of STM to obtained combined topographic and spectroscopic information. STS conducted on non-reconstructed surfaces gave results in excellent agreement with expectations for a Fano resonance, as presented in Fig. 4 for a Sm and a B terminated surface, respectively. In the simplest description, the tunneling conductance, dI∕dV , is given by

where Γ and q are the resonance width and the asymmetry parameter, respectively. The latter is related to the ratio for tunneling probabilities into the 4f states vs. into the conduction and hence, it is expected to be higher for a Sm terminated surface as indeed observed (fit values of q are given in Fig. 4). In addition, q also depends on the particle-hole asymmetry of the conduction band. Note that additional excitations are observed in case of a Sm-terminated surface, possibly resulting from crystal field or surface states. We also observed electronic inhomogeneities on non-reconstructed surfaces of several tens of nanometers in extent which we interpret as subtle inhomogeneities in the Sm valence at the surface.

As outlined above, the enormous topical interest in SmB6 is driven by the theoretical prediction of topologically protected surface states in this material.

Although our present STS results cannot provide clear-cut evidence in favor of, or against, this prediction we note that all types of surfaces, reconstructed and non-reconstructed, displayed a finite zero-bias conductance of considerable magnitude. The single crystals of SmB6 and, for comparison EuB6, for our measurements were provided by the group of Dr. Zachary Fisk (UC Irvine, USA).

As a next step, we tried to identify other systems with localized f electrons, that not only possess non-trivial topology but also retain many non-polar surfaces such that the Dirac cone properties can be easily investigated. Using density functional theory (DFT) based calculations, we have identified an oxide, SmO, a mixed valent correlated compound to possess a non-trivial band structure [Kasinathan2015].

The electronic structure calculations are performed using the full-potential non-orthogonal local orbital code (FPLO). Collected in Fig. 5 is the FPLO non-spin polarized, full-relativistic band structure of SmO with the inclusion of the strong Coulomb interaction (LDA+SO+U) for the experimental lattice constant. The 4f states are split into lower-lying and filled 4f52 states that can accommodate 6 electrons, and higher-lying (above 5.5 eV), empty 4f72 states. The material is semi-metallic with a “warped gap” such that everywhere in the Brillouin zone (at EF), band number x is always below band number x+1. There are small electron and hole pockets at X and Γ, respectively, but there are no band crossings between the highest occupied 4f52 bands and the lowest unoccupied bands. In contrast to the localized and not very dispersive 4f bands, the 5d bands of samarium are dispersive, broad and dip below the Fermi level, retaining a 100% weight at the X point.

As a consequence, we observe a 4f-5d band inversion at the X point, resulting in a nontrivial topology, and the ensuing topological indices 1;(000) allows us to classify SmO as a 3D strongly topological semi-metal.

To provide additional confirmation of the topological non-triviality in SmO and to explicitly identify the protected surface states, we have calculated the Bloch spectral density (ABl(k)) of the 12 topmost surface layers of a semi-infinite solid with [001] and [110] surfaces (Fig. 6). Consistent with the topological indices, we obtain an odd number of Dirac-cones. For both the surfaces, a single Dirac-cone at Γ, which becomes a surface resonance due to the semi-metallic nature of SmO, is observed, along with two cones each at M and X for the [100] and [110] surfaces, respectively.

Having established the nontrivial topology in SmO for the experimental bulk lattice parameter, we considered the scenario of growing thin films of SmO. In general, the lattice parameter of the substrate plays a decisive role in determining the lattice parameter of a thin film. Then, the relevant question to answer is the robustness of the topological semi-metal state as a function of lattice parameter variation. To this end, we have investigated the topological indices for various lattice parameters (Fig. 7). Treating the strong 4f correlations on a mean-field level, the topological semi-metal state is quite robust and remains so, up to 5.38 Å , a 9% increase in lattice parameter. On an experimental level, this result is very promising, since it provides a large range of lattice constants of the substrate onto which SmO thin films can be grown and reveal topological non-triviality. During the evaluation of the topological indices, we became aware of another interesting feature, a nontrivial topology due to s-f band inversion at Γ for 11% and larger lattice constants (5.5 Å). Albeit no gap will be opened at Γ due to the degenerate quartet in cubic symmetry, thin films of SmO will acquire a tetragonal symmetry induced by strain which lifts the degeneracy and a gap opens, creating a topological insulator.

Thus, our calculations show that the topological nontrivial ground-state of SmO is stable for a large range of Sm-O distances, including both positive and negative strain. The Sm-f band filling is thus tunable by strain, which opens up the possibility to create correlated topological nontrivial bands at different filling. SmO is therefore an ideal candidate for further investigations in bulk and thin-film form.

[1]    K. Höfer, C. Becker, D. Rata, J. Swanson, P. Thalmeier and L. H. Tjeng Intrinsic conduction through topological surface states of insulating Bi2Te3 epitaxial thin films. Proc. Natl. Acad. Sci. USA 111, 14979-14984 (2014)

[2]    K. Höfer, C. Becker, S. Wirth and L. H. Tjeng Protective capping of topological surface states of intrinsically insulating Bi2Te3. AIP Adv. 5, 097139 (2015)

[3]    S. Rößler, T.-H. Jang, D. J. Kim, L. H. Tjeng, Z. Fisk, F. Steglich and S. Wirth Hybridization gap and Fano resonance in SmB6. Proc. Natl. Acad. Sci. USA 111, 4798-4802 (2014)

[4]    D. Kasinathan, K. Koepernik, L. H. Tjeng and M. W. Haverkort SmO thin films: A flexible route to correlated flat bands with nontrivial topology. Phys. Rev. B 91, 195127 (2015)

Go to Editor View