Itinerant Spin Dynamics in Structures of ... - Jacobs University

Itinerant Spin Dynamics in Structures of ... - Jacobs University Itinerant Spin Dynamics in Structures of ... - Jacobs University

jacobs.university.de
from jacobs.university.de More from this publisher
11.03.2014 Views

84 Chapter 5: Spin Hall Effect DOS, their average distance and the disorder strength [SBK + 10]. Another way for efficiently injecting spin currents into semiconductors is to make use of the spin Hall effect (SHE). This effect was first proposed by D’yakonov and Perel [DP71a] and describes in today’s terminology the extrinsic SHE which requires spin dependent impurity scattering. It was experimentally confirmed by the angle-resolved optical detection of spin polarization at the edges of a two-dimensional layer [WKSJ05, KMGA04]. The theory of the SHE has been developed in the last 10 years, as reviewed in Refs. [Sch06, ERH07]. The spin transport not only occurs due to the spin precession in the bulk, but is affected by the scattering from nonmagnetic impurities, which can depend on the direction of the spin itself due to the so called skew scattering and the side jump mechanism[Sch06]. In the first part of the following chapter, we will outline the formalism to calculate analytically the SHE which arises even in the absence of impurities, the so called intrinsic SHE, which is due to the bulk SOC. For the clean case we will include both, Rashba and Dresselhaus SOC. In the second part we focus on numerical methods: The first attempt is application of exact diagonalization which has of course strong limitations concerning the system size. To overcome this limitations, we apply the Kernel Polynomial Method (KPM) in the last part of this thesis. It is first applied to treat the metal-insulator transition (MIT) in a symplectic system, finally we calculate the SHE using the KPM. 5.1.1 About the Definition of Spin Current We have learned in Sec.2.3.4 that in presence of the spin-orbit interaction, the spin current components are not conserved[CSS + 04, SZXN06] even if we assume no spin relaxation: In the continuity equation ∂s z ∂t +D e∇J spin = T s − 1 (ˆτ s ) ij s j (5.1) = τ〈∇v F (B SO (k)×S) i 〉− 1 (ˆτ s ) ij s j . (5.2) anadditionaltorquetermT s appearsincontrasttoEq.2.12besidesthetermwhichdescribes the spin relaxation rate 1/τ s . It follows that Noether’s theorem is not applicable to define the spin current. This is a reason why the comparison between results done with Kubo formalism and calculations using Landauer-Büttiker approach is not trivial, where the spin

Chapter 5: Spin Hall Effect 85 current in the latter is defined by[MM05] where I s p,µ = 1 4π ∑ p≠q,ν Tr[Γ µ pG R Γ ν qG A ](V p −V q ), (5.3) Γ µ p = i(Σ µ p −(Σ µ p) † ), (5.4) with the retarded self energy Σ µ p due to coupling of lead p, which has voltage V p , and sample for spin channel µ and advanced and retarded Green’s function G A/R . However, in an experiment the SHE can be measured. This measurement has to be connected to the theoretical description using linear response to a transverse electrical field with the frequency ω which yields, Appendix B, Eq.(B.16), σ µν (ω) = i V ∑ m,n (f(E m )−f(E n ))〈m|j ν |n〉〈n|j µ |m〉 E n −E m E n −E m +ω +iη . (5.5) To calculate the spin Hall conductivity (SHC) the correlation function consists, in contrast to charge conductivity, of the charge current and a current which contains a spin operator. In the simplified model we use the spin current which is given by the anticommutator of the spin and the group velocity v = i[H,r], J z = 4 {σ z,v}. (5.6) This spin current does not differ from the one defined in Sec.2.3.[ESL05, BNnM04] Note that this quantity is dissipationless, because the spin current is even under the time-reversal operation. 5.2 SHE without Impurities: Exact Calculation We consider theHamiltonian for alattice which provides linear Rashba, Eq.(2.15), and linear and cubic Dresselhaus SOC, Eq.(2.14), as introduced in Sec.2.3.3. The confinement to generate the 2D electron gas is in [001] direction. In order to do calculations numerically, one needs to define a tight binding model on a discrete lattice of finite lattice spacing a. It has the following characterization: • square lattice,

Chapter 5: <strong>Sp<strong>in</strong></strong> Hall Effect 85<br />

current <strong>in</strong> the latter is def<strong>in</strong>ed by[MM05]<br />

where<br />

I s p,µ = 1<br />

4π<br />

∑<br />

p≠q,ν<br />

Tr[Γ µ pG R Γ ν qG A ](V p −V q ), (5.3)<br />

Γ µ p = i(Σ µ p −(Σ µ p) † ), (5.4)<br />

with the retarded self energy Σ µ p due to coupl<strong>in</strong>g <strong>of</strong> lead p, which has voltage V p , and<br />

sample for sp<strong>in</strong> channel µ and advanced and retarded Green’s function G A/R . However,<br />

<strong>in</strong> an experiment the SHE can be measured. This measurement has to be connected to<br />

the theoretical description us<strong>in</strong>g l<strong>in</strong>ear response to a transverse electrical field with the<br />

frequency ω which yields, Appendix B, Eq.(B.16),<br />

σ µν (ω) = i V<br />

∑<br />

m,n<br />

(f(E m )−f(E n ))〈m|j ν |n〉〈n|j µ |m〉<br />

E n −E m E n −E m +ω +iη . (5.5)<br />

To calculate the sp<strong>in</strong> Hall conductivity (SHC) the correlation function consists, <strong>in</strong> contrast<br />

to charge conductivity, <strong>of</strong> the charge current and a current which conta<strong>in</strong>s a sp<strong>in</strong> operator.<br />

In the simplified model we use the sp<strong>in</strong> current which is given by the anticommutator <strong>of</strong><br />

the sp<strong>in</strong> and the group velocity v = i[H,r],<br />

J z = 4 {σ z,v}. (5.6)<br />

This sp<strong>in</strong> current does not differ from the one def<strong>in</strong>ed <strong>in</strong> Sec.2.3.[ESL05, BNnM04] Note<br />

that this quantity is dissipationless, because the sp<strong>in</strong> current is even under the time-reversal<br />

operation.<br />

5.2 SHE without Impurities: Exact Calculation<br />

We consider theHamiltonian for alattice which provides l<strong>in</strong>ear Rashba, Eq.(2.15),<br />

and l<strong>in</strong>ear and cubic Dresselhaus SOC, Eq.(2.14), as <strong>in</strong>troduced <strong>in</strong> Sec.2.3.3. The conf<strong>in</strong>ement<br />

to generate the 2D electron gas is <strong>in</strong> [001] direction. In order to do calculations<br />

numerically, one needs to def<strong>in</strong>e a tight b<strong>in</strong>d<strong>in</strong>g model on a discrete lattice <strong>of</strong> f<strong>in</strong>ite lattice<br />

spac<strong>in</strong>g a. It has the follow<strong>in</strong>g characterization:<br />

• square lattice,

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!