Refine
Has Fulltext
- yes (3)
Is part of the Bibliography
- yes (3)
Year of publication
- 2014 (3) (remove)
Document Type
- Doctoral Thesis (3)
Language
- English (3) (remove)
Keywords
- Topologischer Isolator (3) (remove)
In this thesis, we investigate aspects of the physics of heavy-fermion systems and correlated topological insulators.
We numerically solve the interacting Hamiltonians that model the physical systems using quantum Monte Carlo algorithms
to access both ground-state and finite-temperature observables.
Initially, we focus on the metamagnetic transition in the Kondo lattice model for heavy fermions.
On the basis of the dynamical mean-field theory and the dynamical cluster approximation,
our calculations point towards a continuous transition, where the signatures of metamagnetism are linked to a Lifshitz transition of heavy-fermion bands.
In the second part of the thesis, we study various aspects of magnetic pi fluxes in the Kane-Mele-Hubbard model of a correlated topological insulator.
We describe a numerical measurement of the topological index, based on the localized mid-gap states that are provided by pi flux insertions.
Furthermore, we take advantage of the intrinsic spin degree of freedom of a pi flux to devise instances of interacting quantum spin systems.
In the third part of the thesis, we introduce and characterize the Kane-Mele-Hubbard model on the pi flux honeycomb lattice.
We place particular emphasis on the correlations effects along the one-dimensional boundary of the lattice and
compare results from a bosonization study with finite-size quantum Monte Carlo simulations.
Numerical Simulations of Heavy Fermion Systems: From He-3 Bilayers to Topological Kondo Insulators
(2014)
Even though heavy fermion systems have been studied for a long time, a strong interest in heavy fermions persists to this day. While the basic principles of local moment formation, Kondo effect and formation of composite quasiparticles leading to a Fermi liquid, are under- stood, there remain many interesting open questions. A number of issues arise due to the interplay of heavy fermion physics with other phenomena like magnetism and superconduc- tivity.
In this regard, experimental and theoretical investigations of He-3 can provide valuable insights. He-3 represents a unique realization of a quantum liquid. The fermionic nature of He-3 atoms, in conjunction with the absence of long-range Coulomb repulsion, makes this material an ideal model system to study Fermi liquid behavior.
Bulk He-3 has been investigated for quite some time. More recently, it became possible to prepare and study layered He-3 systems, in particular single layers and bilayers. The pos- sibility of tuning various physical properties of the system by changing the density of He-3 and using different substrate materials makes layers of He-3 an ideal quantum simulator for investigating two-dimensional Fermi liquid phenomenology.
In particular, bilayers of He-3 have recently been found to exhibit heavy fermion behavior. As a function of temperature, a crossover from an incoherent state with decoupled layers to a coherent Fermi liquid of composite quasiparticles was observed. This behavior has its roots in the hybridization of the two layers. The first is almost completely filled and subject to strong correlation effects, while the second layer is only partially filled and weakly correlated. The quasiparticles are formed due to the Kondo screening of localized moments in the first layer by the second-layer delocalized fermions, which takes place at a characteristic temperature scale, the coherence scale Tcoh.
Tcoh can be tuned by changing the He-3 density. In particular, at a certain critical filling,
the coherence scale is expected to vanish, corresponding to a divergence of the quasiparticle effective mass, and a breakdown of the Kondo effect at a quantum critical point. Beyond the critical point, the layers are decoupled. The first layer is a local moment magnet, while the second layer is an itinerant overlayer.
However, already at a filling smaller than the critical value, preempting the critical point, the onset of a finite sample magnetization was observed. The character of this intervening phase remained unclear.
Motivated by these experimental observations, in this thesis the results of model calcula- tions based on an extended Periodic Anderson Model are presented. The three particle ring exchange, which is the dominant magnetic exchange process in layered He-3, is included in the model. It leads to an effective ferromagnetic interaction between spins on neighboring sites. In addition, the model incorporates the constraint of no double occupancy by taking the limit of large local Coulomb repulsion.
By means of Cellular DMFT, the model is investigated for a range of values of the chemical potential µ and inverse temperature β = 1/T . The method is a cluster extension to the Dy- namical Mean-Field Theory (DMFT), and allows to systematically include non-local correla- tions beyond the DMFT. The auxiliary cluster model is solved by a hybridization expansion CTQMC cluster solver, which provides unbiased, numerically exact results for the Green’s function and other observables of interest.
As a first step, the onset of Fermi liquid coherence is studied. At low enough temperature, the self-energy is found to exhibit a linear dependence on Matsubara frequency. Meanwhile, the spin susceptibility crossed over from a Curie-Weiss law to a Pauli law. Both observations serve as fingerprints of the Fermi liquid state.
The heavy fermion state appears at a characteristic coherence scale Tcoh. This scale depends strongly on the density. While it is rather high for small filling, for larger filling Tcoh is increas- ingly suppressed. This involves a decreasing quasiparticle residue Z ∼ Tcoh and an enhanced mass renormalization m∗/m ∼ Tcoh−1. Extrapolation leads to a critical filling, where the co-
herence scale is expected to vanish at a quantum critical point. At the same time, the effective mass diverges. This corresponds to a breakdown of the Kondo effect, which is responsible for the formation of quasiparticles, due to a vanishing of the effective hybridization between the layers.
Taking only single-site DMFT results into account, the above scenario seems plausible. However, paramagnetic DMFT neglects the ring exchange interaction completely. In or- der to improve on this, Cellular DMFT simulations are conducted for small clusters of size Nc = 2 and 3. The results paint a different physical picture. The ring exchange, by favor- ing a ferromagnetic alignment of spins, competes with the Kondo screening. As a result, strong short-range ferromagnetic fluctuations appear at larger values of µ. By lowering the temperature, these fluctuations are enhanced at first. However, for T < Tcoh they are increas- ingly suppressed, which is consistent with Fermi liquid coherence. However, beyond a certain threshold value of µ, fluctuations persist to the lowest temperatures. At the same time, while not apparent in the DMFT results, the total occupation n increases quite strongly in a very narrow range around the same value of µ. The evolution of n with µ is always continuous, but hints at a discontinuity in the limit Nc → ∞. This first-order transition breaks the Kondo effect. Beyond the transition, a ferromagnetic state in the first layer is established, and the second layer becomes a decoupled overlayer.
These observations provide a quite appealing interpretation of the experimental results. As a function of chemical potential, the Kondo breakdown quantum critical point is preempted by a first-order transition, where the layers decouple and the first layer turns into a ferromagnet. In the experimental situation, where the filling can be tuned directly, the discontinuous transition is mirrored by a phase separation, which interpolates between the Fermi liquid ground state at lower filling and the magnetic state at higher filling. This is precisely the range of the intervening phase found in the experiments, which is characterized by an onset of a finite sample magnetization.
Besides the interplay of heavy fermion physics and magnetic exchange, recently the spin- orbit coupling, which is present in many heavy fermion materials, attracted a lot of interest. In the presence of time-reversal symmetry, due to spin-orbit coupling, there is the possibility of a topological ground state.
It was recently conjectured that the energy scale of spin-orbit coupling can become dom- inant in heavy fermion materials, since the coherence scale and quasiparticle bandwidth are rather small. This can lead to a heavy fermion ground state with a nontrivial band topology; that is, a topological Kondo insulator (TKI). While being subject to strong correlation effects, this state must be adiabatically connected to a non-interacting, topological state.
The idea of the topological ground state realized in prototypical Kondo insulators, in par- ticular SmB6, promises to shed light on some of the peculiarities of these materials, like a residual conductivity at the lowest temperatures, which have remained unresolved so far.
In this work, a simple two-band model for two-dimensional topological Kondo insulators is devised, which is based on a single Kramer’s doublet coupled to a single conduction band. The model is investigated in the presence of a Hubbard interaction as a function of interaction strength U and inverse temperature β. The bulk properties of the model are obtained by DMFT, with a hybridization expansion CTQMC impurity solver. The DMFT approximation of a local self-energy leads to a very simple way of computing the topological invariant.
The results show that with increasing U the system can be driven through a topological phase transition. Interestingly, the transition is between distinct topological insulating states, namely the Γ-phase and M-phase. This appearance of different topological phases is possible due to the symmetry of the underlying square lattice. By adiabatically connecting both in- teracting states with the respective non-interacting state, it is shown that the transition indeed drives the system from the Γ-phase to the M-phase.
A different behavior can be observed by pushing the bare position of the Kramer’s doublet to higher binding energies. In this case, the non-interacting starting point has a trivial band topology. By switching on the interaction, the system can be tuned through a quantum phase transition, with a closing of the band gap. Upon reopening of the band gap, the system is in the Γ-phase, i. e. a topological insulator. By increasing the interaction strength further, the system moves into a strongly correlated regime. In fact, close to the expected transition to the M phase, the mass renormalization becomes quite substantial. While absent in the para- magnetic DMFT simulations conducted, it is conceivable that instead of a topological phase transition, the system undergoes a time-reversal symmetry breaking, magnetic transition.
The regime of strong correlations is studied in more detail as a function of temperature, both in the bulk and with open boundary conditions. A quantity which proved very useful is the bulk topological invariant Ns, which can be generalized to finite interaction strength and temperature. In particular, it can be used to define a temperature scale T ∗ for the onset of the topological state. Rescaling the results for Ns, a nice data collapse of the results for different values of U, from the local moment regime to strongly mixed valence, is obtained. This hints at T ∗ being a universal low energy scale in topological Kondo insulators. Indeed, by comparing T ∗ with the coherence scale extracted from the self-energy mass renormalization, it is found that both scales are equivalent up to a constant prefactor. Hence, the scale T ∗ obtained from the temperature dependence of topological properties, can be used as an independent measure for Fermi liquid coherence. This is particularly useful in the experimentally relevant mixed valence regime, where charge fluctuations cannot be neglected. Here, a separation of the energy scales related to spin and charge fluctuations is not possible.
The importance of charge fluctuations becomes evident in the extent of spectral weight transfer as the temperature is lowered. For mixed valence, while the hybridization gap emerges, a substantial amount of spectral weight is shifted from the vicinity of the Fermi level to the lower Hubbard band. In contrast, this effect is strongly suppressed in the local moment regime.
In addition to the bulk properties, the spectral function for open boundaries is studied as a function of temperature, both in the local moment and mixed valence regime. This allows an investigation of the emergence of topological edge states with temperature. The method used here is the site-dependent DMFT, which is a generalization of the conventional DMFT to inhomogeneous systems. The hybridization expansion CTQMC algorithm is used as impurity solver.
By comparison with the bulk results for the topological quantity Ns, it is found that the
temperature scale for the appearance of the topological edge states is T ∗, both in the mixed valence and local moment regime.
Recently a new state of matter was discovered in which the bulk insulating state in a material is accompanied by conducting surface or edge states. This new state of matter can be distinguished from a conventional insulator phase by the topological properties of its band structure which led to the name "topological insulators". Experimentally, topological insulator states are mostly found in systems characterized by a band inversion compared to conventional systems. In most topological insulator systems, this is caused by a combination of energetically close bands and spin orbit coupling. Such properties are found in systems with heavy elements like Hg and Bi. And indeed, the first experimental discovery of a topological insulator succeeded in HgTe quantum wells and later also in BiSb bulk systems.
Topological insulators are of large interest due to their unique properties: In 2-dimensional topological insulators one dimensional edge states form without the need of an external magnetic field (in contrast to the quantum Hall effect). These edge states feature a linear band dispersion, a so called Dirac dispersion. The quantum spin Hall states are helical edge states, which means they consist of counterpropagating oppositely spin polarized edge channels. They are therefore of great potential for spintronic applications as well as building blocks for new more exotic states like Majorana Fermions. 3-dimensional topological insulators feature 2-dimensional surface states with only one Dirac band (also called Dirac cone) on each surface and an interesting spin texture where spin and momentum are locked perpendicular to each other in the surface plane. This unique surface band structure is predicted to be able to host several exotic states like e.g. Majorana Fermions (in combination with superconductors) and magnetic monopole like excitations.
This PhD thesis will summarize the discovery of topological insulators and highlights the developments on their experimental observations. The work focuses on HgTe which is up to now the only topological insulator material where the expected properties are unambiguously demonstrated in transport experiments. In HgTe, the topological insulator properties arise from the inversion of the Gamma_6 and Gamma_8 bands. The band inversion in HgTe is due to a combination of a high spin orbit splitting in Te and large energy corrections (due to the mass-velocity term) to the energy levels in Hg. Bulk HgTe, however, is a semimetal, which means for the conversion into a topological insulator a band gap has to be opened. In two dimensions (HgTe quantum well structures) this is achieved via quantum confinement, which opens a band gap between the quantum well subbands. In three dimensions, strain is used to lift the degeneracy of the semimetallic Gamma_8 bands opening up a band gap.
The thesis is structured as follows:
- The first chapter of this thesis will give a brief overview on discoveries in the field of topological insulators. It focuses on works relevant to experimental results presented in the following chapters. This includes a short outline of the early predictions and a summary of important results concerning 2-dimensional topological insulators while the final section discusses observations concerning 3-dimensional topological insulators.
- The discovery of the quantum spin Hall effect in HgTe marked the first experimental observation of a topological insulator. Chapter 2 will focus on HgTe quantum wells and the quantum spin Hall effect.
Above a critical thickness, HgTe quantum wells are predicted to host the quantum spin Hall state, the signature of a 2-dimensional topological insulator. HgTe quantum wells exhibiting low carrier concentrations and at the same time high carrier mobilities are required to be able to measure the quantum spin Hall effect. The growth of such high quality HgTe quantum wells was one of the major goals for this work. Continuous optimization of the substrate preparation and growth conditions resulted in controlled carrier densities down to a few 10^10 cm^-2. At the same time, carrier mobilities exceeding 1 x 10^6 cm^2/Vs have been achieved, which provides mean free paths of several micrometers in the material. Thus the first experimental evidence for the existence of the quantum spin Hall edge states succeeded in transport experiments on microstructures: When the Fermi energy was located in the bulk band gap a residual quantized resistance of 2e^2/h was found. Further experiments focused on investigating the nature of transport in this regime. By non-local measurements the edge state character could be established. The measured non-local resistances corresponded well with predictions from the Landauer-Büttiker theory applied to transport in helical edge channels.
In a final set of experiments the spin polarization of the edge channels was investigated. Here, we could make use of the advantage that HgTe quantum well structures exhibit a large Rashba spin orbit splitting. In systems with a large Rashba spin orbit splitting a spin accumulation is expected to occur at the edge of the sample perpendicular to a current flow. This so-called spin Hall effect was then used as a spin injector and detector. Using split gate devices it was possible to bring spin Hall and quantum spin Hall state into direct contact, which enabled an all electrical detection of the spin polarization of the quantum spin Hall edge channels.
- HgTe as a 3-dimensional topological insulator will be presented in chapter 3. Straining the HgTe layer enables the observation of topological insulator behavior. It was found that strain can be easily implemented during growth by using CdTe substrates. CdTe has a slightly larger lattice constant than HgTe and therefore leads to tensile strain in the HgTe layer as long as the growth is pseudomorphic. Magnetotransport studies showed the emergence of quantum Hall transport with characteristic signatures of a Dirac type bandstructure. Thus, this result marks the first observation of the quantum Hall effect in the surface states of a 3-dimensional topological insulator.
Transport experiments on samples fitted with a top gate enabled the identification of contributions from individual surfaces. Furthermore, the surface state quantum Hall effect was found to be surprisingly stable, perturbations due to additional bulk transport could not be found, even at high carrier densities of the system.
- Chapters 4 - 6 serve as in depth overviews of selected works: Chapter 4 presents a detailed overview on the all electrical detection of the spin Hall effect in HgTe quantum wells. The detection of the spin polarization of the quantum spin Hall effect is shown in chapter 5 and chapter 6 gives a detailed overview on the quantum Hall effect originating from the topological surface state in strained bulk HgTe.
The investigations discussed in this thesis pioneered the experimental work on the transport properties of topological insulator systems. The understanding of the fundamental properties of topological insulators enables new experiments in which e.g. the inclusion of magnetic dopants or the interplay between topological insulator and superconductors can be investigated in detail.