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

112 List of Figures 3.12 The relative magnetoconductivity ∆σ(B) − ∆σ(B = 0) in units of 2e 2 /2π, with the same parameters and number of modes as in Fig.3.11. . . . . . . . 53 3.13 Width of wire WQ SO at which there is a crossover from negative to positive magnetoconductivity as function of the lower cutoff 1/D e Q 2 τ SO ϕ. . . . . . . 54 3.14 Spectrum in the case of adiabatic boundaries. Except for states which are polarized in z direction, the relaxation rates for all states diverge at small wire widths, Q SO W ≪ 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.15 (a) Real and (b) imaginary parts of the spectrum of the 2D Cooperon with Zeeman term of strength gµ B B/D e Q 2 = 0.4. E SO B,2D,1 (black), E B,2D,2 (red dashed), E B,2D,3 (green), E B,2D,4 (blue dashed). Dashed vertical lines are located at K x = ±1/ √ 2, the wave vector where the triplet mode E T− and the singlet mode E S are crossing each other (without loss of generality K y = 0). 59 3.16 (a) Real and (b) imaginary parts of the spectrum of the 2D Cooperon with Zeeman term of the strength gµ B B/D e Q 2 SO = 0...2 in steps of 0.25 (w.l.o.g K y = 0). The B independent mode E T0 is not shown.. . . . . . . . . . . . . 61 3.17 (a) Real and (b) imaginary parts of the spectrum of the Cooperon with Zeeman term of the strength gµ B B/D e Q 2 SO = 0...0.8 in steps of 0.1 in a finite wire of the width Q SO W = 0.5. The B independent mode E t0 is not shown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 3.18 The magnetoconductance ∆σ(B) in a magnetic field perpendicular to the quantum well, where its coupling via the Zeeman term is considered by exact diagonalization while its effect on the orbital motion is considered effectively by the magnetic phase-shifting rate 1/τ B (W) for wire width Q SO W = 1, dephasingrate 1/τ ϕ = 0.06D e Q 2 SO, andelastic-scattering rate 1/τ = 4D e Q 2 SO. ThestrengthofthecontributionoftheZeemantermisvariedbythematerialdependent factor ˜g = gµ B H s /D e Q 2 SO in the range ˜g = 0...1.5 in steps of 0.5: The system changes from positive magnetoconductivity (˜g = 0, green) to negative one (˜g = 0.5...1.5, continuous decrease of the absolute minimum). 64 3.19 (a) Change of crossover width W c with g factor: The magnetic field is included as an effective field 1/τ B and in the Zeeman term. The strength of the contribution of the Zeeman term is varied by the material dependent factor ˜g = gµ B H s /D e Q 2 . (b) Change of crossover width W SO c with Zeeman field: To calculate W c , we fix the Zeeman field to a certain value, horizontal axis, while we vary the effective field independently and calculate if negative or positive magnetoconductivity is present. For different Zeeman fields B Z /H s , we find thereby a different width W c . Here, we set g/8m e D e = 1. In (a) and (b), the cutoff due to dephasing is varied: 1/D e Q 2 SOτ ϕ ∈ = 0.04,0.06,0.08 (lowest first). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.1 Dependence of the W 2 coefficient in Eq.(4.27) on the lateral rotation (θ). The absolute minimum is found for α x1 (θ = 0) = −q 2 (here: q 1 /q 2 = 1.63) and for different SO strength we find the minimum at θ = (1/4+n)π, n if q 1 < (q s3 / √ 2) (dashed line: q 1 = (q s3 / √ 2)) and at θ = (3/4+n)π, n else. Here we set q s3 = 0.9. The scaling is arbitrary. . . . . . . . . . . . . . 73

List of Figures 113 4.2 The spin dephasing time T 2 of a spin initially oriented along the [001] direction in units of (D e q2 2 ) for the special case of equal Rashba and lin. Dresselhaus SOC. The different curves show different strength of cubic Dresselhaus in units of q s3 /q 2 . In the case of finite cubic Dresselhaus SOC we set W = 0.4/q 2 . If q s3 = 0: T 2 diverges at θ = (1/4 + n)π, n ∈(dashed vertical lines). The horizontal dashed line indicated the 2D spin dephasing time, T 2 = 1/(4q2 2D e). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 4.3 Example of an experiment done by Kunihashi et al.[KKN09]. The width dependence of the spin relaxation length l 1D SO of different carrier density. Solid lines and dashed lines in (b) show the l 1D SO calculated from the theory presented in this work, with neglecting cubic Dresselhaus term and taking into account full SOIs, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . 79 4.4 The lowest eigenvalues of the confined Cooperon Hamiltonian Eq.(4.49), equivalent to the lowest spin relaxation rate, are shown for Q = 0 for different number of modes N = k F W/π. Different curves correspond to different values of α 2 /q s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 4.5 The lowest eigenvalues of the confined Cooperon Hamiltonian Eq.(4.49), equivalent to the lowest spin relaxation rate, are shown for Q = 0 for different number of modes N = k F W/π. Different curves correspond to different values of α 1 /q s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 5.1 Energy band E − (k), Eq.(5.10) is plotted for pure Rashba SOC as function of wave vector k. The contour lines indicate the energy at which one finds a Van Hove singularity in the DOS (below half-filling). . . . . . . . . . . . . . 87 5.2 (a) SHC σ SH as a function of Fermi energy E F , in a clean system of size L 2 = 170×170 with both Rashba and linear Dresselhaus SOC with α 2 > α 1 (blue curve) and α 2 < α 1 (red curve). (b) SHC σ SH as a function of Fermi energy E F , in a clean system of size V = 150×150 with only Rashba SOC of strength α 2 = 0.8t (blue/solid), α 2 = 1.4t (red/dotted) and α 2 = 2t (yellow/dashed). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 5.3 DOS, as a function of Fermi energy in units of t, in presence of impurities of binary type with a concentration of 10% calculated using exact diagonalization. The system size is L 2 = 32 2 , and the SOC is Rashba type with α 2 = 1.2t with cutoff η = 0.02t. . . . . . . . . . . . . . . . . . . . . . . . . . 93 5.4 (a) SHC, as a function of Fermi energy in units of t, in presence of impurities of binary type calculated using exact diagonalization. The impurity strength is V = −2.8t with a concentration of 10%. The system size is L 2 = 32 2 , and the SOC is Rashba type with α 2 = 1.2t with cutoff η = 0.06. (b) Comparison of a) with DOS (blue curve). . . . . . . . . . . . . . . . . . . . . . . . . . . 94 5.5 SHC σ SH as function of filling n in presence of impurities of binary type calculated using exact diagonalization. The impurity concentration is set to 10% for all plots and the average is performed over 200 impurity configurations. The system size is L 2 = 32 2 , and the SOC is Rashba type with α 2 = 1.2t with cutoff η = 0.06 (a) V = −0.2t...−2.8t (b) V = −2.8t...−5t. . . . . 95

112 List <strong>of</strong> Figures<br />

3.12 The relative magnetoconductivity ∆σ(B) − ∆σ(B = 0) <strong>in</strong> units <strong>of</strong> 2e 2 /2π,<br />

with the same parameters and number <strong>of</strong> modes as <strong>in</strong> Fig.3.11. . . . . . . . 53<br />

3.13 Width <strong>of</strong> wire WQ SO at which there is a crossover from negative to positive<br />

magnetoconductivity as function <strong>of</strong> the lower cut<strong>of</strong>f 1/D e Q 2 τ SO ϕ. . . . . . . 54<br />

3.14 Spectrum <strong>in</strong> the case <strong>of</strong> adiabatic boundaries. Except for states which are<br />

polarized <strong>in</strong> z direction, the relaxation rates for all states diverge at small<br />

wire widths, Q SO W ≪ 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55<br />

3.15 (a) Real and (b) imag<strong>in</strong>ary parts <strong>of</strong> the spectrum <strong>of</strong> the 2D Cooperon with<br />

Zeeman term <strong>of</strong> strength gµ B B/D e Q 2 = 0.4. E SO B,2D,1 (black), E B,2D,2 (red<br />

dashed), E B,2D,3 (green), E B,2D,4 (blue dashed). Dashed vertical l<strong>in</strong>es are<br />

located at K x = ±1/ √ 2, the wave vector where the triplet mode E T− and<br />

the s<strong>in</strong>glet mode E S are cross<strong>in</strong>g each other (without loss <strong>of</strong> generality K y = 0). 59<br />

3.16 (a) Real and (b) imag<strong>in</strong>ary parts <strong>of</strong> the spectrum <strong>of</strong> the 2D Cooperon with<br />

Zeeman term <strong>of</strong> the strength gµ B B/D e Q 2 SO<br />

= 0...2 <strong>in</strong> steps <strong>of</strong> 0.25 (w.l.o.g<br />

K y = 0). The B <strong>in</strong>dependent mode E T0 is not shown.. . . . . . . . . . . . . 61<br />

3.17 (a) Real and (b) imag<strong>in</strong>ary parts <strong>of</strong> the spectrum <strong>of</strong> the Cooperon with<br />

Zeeman term <strong>of</strong> the strength gµ B B/D e Q 2 SO<br />

= 0...0.8 <strong>in</strong> steps <strong>of</strong> 0.1 <strong>in</strong> a<br />

f<strong>in</strong>ite wire <strong>of</strong> the width Q SO W = 0.5. The B <strong>in</strong>dependent mode E t0 is not<br />

shown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62<br />

3.18 The magnetoconductance ∆σ(B) <strong>in</strong> a magnetic field perpendicular to the<br />

quantum well, where its coupl<strong>in</strong>g via the Zeeman term is considered by exact<br />

diagonalization while its effect on the orbital motion is considered effectively<br />

by the magnetic phase-shift<strong>in</strong>g rate 1/τ B (W) for wire width Q SO W = 1,<br />

dephas<strong>in</strong>grate 1/τ ϕ = 0.06D e Q 2 SO, andelastic-scatter<strong>in</strong>g rate 1/τ = 4D e Q 2 SO.<br />

Thestrength<strong>of</strong>thecontribution<strong>of</strong>theZeemantermisvariedbythematerialdependent<br />

factor ˜g = gµ B H s /D e Q 2 SO<br />

<strong>in</strong> the range ˜g = 0...1.5 <strong>in</strong> steps <strong>of</strong> 0.5:<br />

The system changes from positive magnetoconductivity (˜g = 0, green) to<br />

negative one (˜g = 0.5...1.5, cont<strong>in</strong>uous decrease <strong>of</strong> the absolute m<strong>in</strong>imum). 64<br />

3.19 (a) Change <strong>of</strong> crossover width W c with g factor: The magnetic field is <strong>in</strong>cluded<br />

as an effective field 1/τ B and <strong>in</strong> the Zeeman term. The strength <strong>of</strong> the<br />

contribution <strong>of</strong> the Zeeman term is varied by the material dependent factor<br />

˜g = gµ B H s /D e Q 2 . (b) Change <strong>of</strong> crossover width W SO c with Zeeman field:<br />

To calculate W c , we fix the Zeeman field to a certa<strong>in</strong> value, horizontal axis,<br />

while we vary the effective field <strong>in</strong>dependently and calculate if negative or<br />

positive magnetoconductivity is present. For different Zeeman fields B Z /H s ,<br />

we f<strong>in</strong>d thereby a different width W c . Here, we set g/8m e D e = 1. In (a)<br />

and (b), the cut<strong>of</strong>f due to dephas<strong>in</strong>g is varied: 1/D e Q 2 SOτ ϕ ∈<br />

= 0.04,0.06,0.08<br />

(lowest first). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65<br />

4.1 Dependence <strong>of</strong> the W 2 coefficient <strong>in</strong> Eq.(4.27) on the lateral rotation (θ).<br />

The absolute m<strong>in</strong>imum is found for α x1 (θ = 0) = −q 2 (here: q 1 /q 2 = 1.63)<br />

and for different SO strength we f<strong>in</strong>d the m<strong>in</strong>imum at θ = (1/4+n)π, n<br />

if q 1 < (q s3 / √ 2) (dashed l<strong>in</strong>e: q 1 = (q s3 / √ 2)) and at θ = (3/4+n)π, n<br />

else. Here we set q s3 = 0.9. The scal<strong>in</strong>g is arbitrary. . . . . . . . . . . . . . 73

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

Saved successfully!

Ooh no, something went wrong!