Expos de soutenance pour le titre de Docteur de lcole Polytechnique - - PowerPoint PPT Presentation

expos de soutenance pour le titre de docteur de l cole
SMART_READER_LITE
LIVE PREVIEW

Expos de soutenance pour le titre de Docteur de lcole Polytechnique - - PowerPoint PPT Presentation

Expos de soutenance pour le titre de Docteur de lcole Polytechnique Spcialit: Physique Iurii Timrov 27 March 2013, cole Polytechnique 1/57 Outline 1. Introduction 1.1 Motivation 1.2 Material: Bismuth 1.3 State of the art


slide-1
SLIDE 1

Exposé de soutenance pour le titre de Docteur de l’École Polytechnique Spécialité: Physique Iurii Timrov 27 March 2013, École Polytechnique

1/57

slide-2
SLIDE 2

Outline

  • 1. Introduction

1.1 Motivation 1.2 Material: Bismuth 1.3 State of the art methods

  • 2. Results

2.1 High-energy response: new approach for EELS 2.2 Low-energy response: free-carrier response

  • 3. Conclusions

2/57

slide-3
SLIDE 3

Outline

  • 1. Introduction

1.1 Motivation 1.2 Material: Bismuth 1.3 State of the art methods

  • 2. Results

2.1 High-energy response: new approach for EELS 2.2 Low-energy response: free-carrier response

  • 3. Conclusions

3/57

slide-4
SLIDE 4

Motivation

How to understand the nature of materials? Perturb them and see what happens!

4/57

slide-5
SLIDE 5

Motivation

Optics: q → 0, ω → 0 Drude model: ǫ(ω) = 1 − ω2

p

ω(ω + iγ) EELS: q = 0, ω = 0 Loss function −Im[ǫ−1(q, ω)] 5/57

slide-6
SLIDE 6

Motivation

Ab initio description of the full charge-carrier response of bismuth to external perturbations: low-energy and high-energy response. 6/57

slide-7
SLIDE 7

Why do we need a new method for EELS?

  • 1. Bridging the valence-loss and the core-loss EELS.

It is computationally ex- pensive for state-of-the- art methods to describe EEL spectra of complex systems in the energy range up to 100 eV.

  • C. Wehenkel et al., Solid State Comm. 15, 555 (1974)

7/57

slide-8
SLIDE 8

Why do we need a new method for EELS?

  • 2. Calculation of EEL spectra of large systems (hundreds of atoms).

Example: Calculation of surface plasmons = ⇒ Simulation of the surface is needed Figure: View of a 5-layer slab model

  • f a surface, as used in periodic

calculation. ⇓ Large number of atoms ⇓ Computationally demanding task for state-of-the-art methods

  • D. Scholl and J. Steckel, “DFT: A practical introduction” (2009).

8/57

slide-9
SLIDE 9

Low-energy response: photoexcited bismuth

Photoexcitation of Bi ⇐ ⇒ Pump-probe THz expt. (L. Perfetti, J. Faure.) Theoretical model is needed in order to explain the evolution of the Drude plasma frequency ωp after the photoexcitation of Bi. 9/57

slide-10
SLIDE 10

Outline

  • 1. Introduction

1.1 Motivation 1.2 Material: Bismuth 1.3 State of the art methods

  • 2. Results

2.1 High-energy response: new approach for EELS 2.2 Low-energy response: free-carrier response

  • 3. Conclusions

10/57

slide-11
SLIDE 11

Material: Semimetal Bismuth

J.-P . Issi, Aus. J. Phys. 32, 585 (1979)

  • M. Z. Hasan and C. L. Kane, Rev. Mod. Phys. 82, 3045 (2010).

11/57

slide-12
SLIDE 12

Crystal and Electronic Structure

A7 rhombohedral structure: Peierls distortion of sc lattice Semimetallicity is due to the Peierls distor- tion: Overlap between valence and conduc- tion bands. The Fermi surface consists of 1 hole pocket and 3 electron pockets.

  • Y. Liu et al., Phys. Rev. B 52, 1566 (1995).

J.-P . Issi, Aus. J. Phys. 32, 585 (1979) 12/57

slide-13
SLIDE 13

Crystal and Electronic Structure

A7 rhombohedral structure: Peierls distortion of sc lattice Semimetallicity is due to the Peierls distor- tion: Overlap between valence and conduc- tion bands. The Fermi surface consists of 1 hole pocket and 3 electron pockets.

  • Y. Liu et al., Phys. Rev. B 52, 1566 (1995).

J.-P . Issi, Aus. J. Phys. 32, 585 (1979) 12/57

slide-14
SLIDE 14

Crystal and Electronic Structure

A7 rhombohedral structure: Peierls distortion of sc lattice Semimetallicity is due to the Peierls distor- tion: Overlap between valence and conduc- tion bands. The Fermi surface consists of 1 hole pocket and 3 electron pockets.

  • Y. Liu et al., Phys. Rev. B 52, 1566 (1995).

J.-P . Issi, Aus. J. Phys. 32, 585 (1979) 13/57

slide-15
SLIDE 15

Spin-orbit coupling (SOC)

Spin-orbit coupling is a coupling of electron’s spin S with its orbital motion L. The SOC Hamiltonian reads: HSOC ∝ ∇V (L · σ), where V is the potential, and σ are Pauli spin-matrices: S = 2 σ σ

  • .

material SOC-assisted split- ting of levels at Γ (eV) Si 0.04 GaAs 0.3 InSb 0.8 As 0.3 Sb 0.6 Pb 1.0 Bi 1.5 In bismuth the spin-orbit coupling is very strong!

  • A. Dal Corso, J. Phys. Condens. Matter 20, 445202 (2008).

14/57

slide-16
SLIDE 16

Spin-orbit coupling (SOC)

Spin-orbit coupling is a coupling of electron’s spin S with its orbital motion L. The SOC Hamiltonian reads: HSOC ∝ ∇V (L · σ), where V is the potential, and σ are Pauli spin-matrices: S = 2 σ σ

  • .

material SOC-assisted split- ting of levels at Γ (eV) Si 0.04 GaAs 0.3 InSb 0.8 As 0.3 Sb 0.6 Pb 1.0 Bi 1.5 In bismuth the spin-orbit coupling is very strong!

  • A. Dal Corso, J. Phys. Condens. Matter 20, 445202 (2008).

14/57

slide-17
SLIDE 17

Kohn-Sham band structure of bismuth

  • X. Gonze et al., Phys. Rev. B 41, 11827 (1990)
  • A. B. Shick et al., Phys. Rev. B 60, 15484 (1999)
  • I. Timrov, J. Faure, N. Vast, L. Perfetti et al., Phys. Rev. B 85, 155139 (2012)

15/57

slide-18
SLIDE 18

Kohn-Sham band structure of bismuth

  • X. Gonze et al., Phys. Rev. B 41, 11827 (1990)
  • A. B. Shick et al., Phys. Rev. B 60, 15484 (1999)
  • I. Timrov, J. Faure, N. Vast, L. Perfetti et al., Phys. Rev. B 85, 155139 (2012)

16/57

slide-19
SLIDE 19

Outline

  • 1. Introduction

1.1 Motivation 1.2 Material: Bismuth 1.3 State of the art methods

  • 2. Results

2.1 High-energy response: new approach for EELS 2.2 Low-energy response: free-carrier response

  • 3. Conclusions

17/57

slide-20
SLIDE 20

Density Functional Theory

Ground-state: DFT The Kohn-Sham equation:

  • −1

2∇2 + VKS(r)

  • ϕi(r) = εi ϕi(r).

The Kohn-Sham potential VKS(r):

  • ρ(r′)

|r − r′| dr′ + δExc[ρ(r)] δρ(r) + Vext(r). The charge-density: ρ(r) =

  • cc
  • i

|ϕi(r)|2. The quantum Liouville equation: [ ˆ HKS, ˆ ρ ] = 0.

Hohenberg and Kohn, Phys. Rev. (1964) Kohn and Sham, Phys. Rev. (1965)

18/57

slide-21
SLIDE 21

Historical note

19/57

slide-22
SLIDE 22

Time-Dependent Density Functional Theory

Ground-state: DFT The Kohn-Sham equation:

  • −1

2∇2 + VKS(r)

  • ϕi(r) = εi ϕi(r).

The Kohn-Sham potential VKS(r):

  • ρ(r′)

|r − r′| dr′ + δExc[ρ(r)] δρ(r) + Vext(r). The charge-density: ρ(r) =

  • cc
  • i

|ϕi(r)|2. The quantum Liouville equation: [ ˆ HKS, ˆ ρ ] = 0.

Hohenberg and Kohn, Phys. Rev. (1964) Kohn and Sham, Phys. Rev. (1965)

Excited-state: TDDFT The TD Kohn-Sham equation:

  • −1

2∇2 + VKS(r, t)

  • ϕi(r, t) = i ∂

∂t ϕi(r, t). The TD Kohn-Sham potential VKS(r, t):

  • ρ(r′, t)

|r − r′| dr′ + δExc[ρ(r, t)] δρ(r, t) +Vext(r, t), The TD charge-density: ρ(r, t) =

  • cc
  • i

|ϕi(r, t)|2. The TD quantum Liouville equation: [ ˆ HKS(t), ˆ ρ(t) ] = i ∂ ∂t ˆ ρ(t).

Runge and Gross, PRL (1984) Onida, Reining, Rubio, RMP (2002)

20/57

slide-23
SLIDE 23

Time-Dependent Density Functional Theory

Ground-state: DFT The Kohn-Sham equation:

  • −1

2∇2 + VKS(r)

  • ϕi(r) = εi ϕi(r).

The Kohn-Sham potential VKS(r):

  • ρ(r′)

|r − r′| dr′ + δExc[ρ(r)] δρ(r) + Vext(r). The charge-density: ρ(r) =

  • cc
  • i

|ϕi(r)|2. The quantum Liouville equation: [ ˆ HKS, ˆ ρ ] = 0.

Hohenberg and Kohn, Phys. Rev. (1964) Kohn and Sham, Phys. Rev. (1965)

Excited-state: TDDFT The TD Kohn-Sham equation:

  • −1

2∇2 + VKS(r, t)

  • ϕi(r, t) = i ∂

∂t ϕi(r, t). The TD Kohn-Sham potential VKS(r, t):

  • ρ(r′, t)

|r − r′| dr′ + δExc[ρ(r, t)] δρ(r, t) +Vext(r, t), The TD charge-density: ρ(r, t) =

  • cc
  • i

|ϕi(r, t)|2. The TD quantum Liouville equation: [ ˆ HKS(t), ˆ ρ(t) ] = i ∂ ∂t ˆ ρ(t).

Runge and Gross, PRL (1984) Onida, Reining, Rubio, RMP (2002)

20/57

slide-24
SLIDE 24

Fluctuation-dissipation theorem

Optical absorption Perturbation: electric field ⇓ Polarization of the dipole: d(ω) = χ(ω) Eext(ω) χ is the polarization-polarization correlation function Im ǫ(ω) ∝ S(ω) S(ω) = 2 π ω Im χ(ω) S is the oscillator strength

◮ Im ǫ: Measured experimentally ◮ S: Fluctuation of polarization ◮ Im χ: Dissipation of energy

21/57

slide-25
SLIDE 25

Two implementations of linear-response TDDFPT

Optical absorption spectra of finite systems

Conventional TDDFT approach Independent-transition polarizability χ0 χ0(ω) =

  • v,c

(fv−fc)ϕc(r)ϕ∗

v(r)ϕv(r′)ϕc(r′)

ω − (εc − εv) + i η Dyson-like equation: χ = χ0 + χ0 (vCoul + fxc) χ

Onida, Reining, Rubio, RMP (2002)

Liouville-Lanczos approach Definition: χ(ω) ≡ Tr

  • ˜

V ′

ext(r, ω) ˆ

ρ′(ω)

  • ˆ

ρ′(ω) =? Quantum Liouville equation: [ ˆ HKS(t), ˆ ρ(t) ] = i ∂ ∂t ˆ ρ(t) Linearization + Fourier transform: (ω − ˆ L) · ˆ ρ′(ω) = [˜ V ′

ext(ω), ˆ

ρ0] ˆ L · ˆ ρ′ ≡ [ˆ H0

KS, ˆ

ρ′] + [ˆ VHXC, ˆ ρ0] χ(ω) = ˜ V ′

ext(ω)|(ω − ˆ

L)−1[˜ V ′

ext(ω), ˆ

ρ0] ⇓ Use of Lanczos recursion method

Rocca, Gebauer, Saad, Baroni, JCP (2008)

22/57

slide-26
SLIDE 26

Two implementations of linear-response TDDFPT

Optical absorption spectra of finite systems

Conventional TDDFT approach Independent-transition polarizability χ0 χ0(ω) =

  • v,c

(fv−fc)ϕc(r)ϕ∗

v(r)ϕv(r′)ϕc(r′)

ω − (εc − εv) + i η Dyson-like equation: χ = χ0 + χ0 (vCoul + fxc) χ

Onida, Reining, Rubio, RMP (2002)

Liouville-Lanczos approach Definition: χ(ω) ≡ Tr

  • ˜

V ′

ext(r, ω) ˆ

ρ′(ω)

  • ˆ

ρ′(ω) =? Quantum Liouville equation: [ ˆ HKS(t), ˆ ρ(t) ] = i ∂ ∂t ˆ ρ(t) Linearization + Fourier transform: (ω − ˆ L) · ˆ ρ′(ω) = [˜ V ′

ext(ω), ˆ

ρ0] ˆ L · ˆ ρ′ ≡ [ˆ H0

KS, ˆ

ρ′] + [ˆ VHXC, ˆ ρ0] χ(ω) = ˜ V ′

ext(ω)|(ω − ˆ

L)−1[˜ V ′

ext(ω), ˆ

ρ0] ⇓ Use of Lanczos recursion method

Rocca, Gebauer, Saad, Baroni, JCP (2008)

22/57

slide-27
SLIDE 27

Outline

  • 1. Introduction

1.1 Motivation 1.2 Material: Bismuth 1.3 State of the art methods

  • 2. Results

2.1 High-energy response: new approach for EELS 2.2 Low-energy response: free-carrier response

  • 3. Conclusions

23/57

slide-28
SLIDE 28

High-energy response

EELS: q = 0, ω = 0

24/57

slide-29
SLIDE 29

Fluctuation-dissipation theorem

Optical absorption Perturbation: electric field ⇓ Polarization of the dipole: d(ω) = χ(ω) Eext(ω) χ is the polarization-polarization correlation function Im ǫ(ω) ∝ S(ω) S(ω) = 2 π ω Im χ(ω) S is the oscillator strength

◮ Im ǫ: Measured experimentally ◮ S: Fluctuation of polarization ◮ Im χ: Dissipation of energy

EELS Perturbation: electron beam ⇓ Double differential cross-section: d2σ dΩdω ∝ S(q, q; ω) S(q, q; ω) = − 1 π Im χ(q, q; ω) S is the dynamic structure factor χ(q, q; t) = ˆ ρq(t)ˆ ρq(0) is the density-density correlation function −Im ǫ−1(q, q; ω) ∝ −Im χ(q, q; ω)

d2σ dΩdω : Measured experiment.

◮ S: Fluctuation of density ◮ Im χ: Dissipation of energy

25/57

slide-30
SLIDE 30

Fluctuation-dissipation theorem

Optical absorption Perturbation: electric field ⇓ Polarization of the dipole: d(ω) = χ(ω) Eext(ω) χ is the polarization-polarization correlation function Im ǫ(ω) ∝ S(ω) S(ω) = 2 π ω Im χ(ω) S is the oscillator strength

◮ Im ǫ: Measured experimentally ◮ S: Fluctuation of polarization ◮ Im χ: Dissipation of energy

EELS Perturbation: electron beam ⇓ Double differential cross-section: d2σ dΩdω ∝ S(q, q; ω) S(q, q; ω) = − 1 π Im χ(q, q; ω) S is the dynamic structure factor χ(q, q; t) = ˆ ρq(t)ˆ ρq(0) is the density-density correlation function −Im ǫ−1(q, q; ω) ∝ −Im χ(q, q; ω)

d2σ dΩdω : Measured experiment.

◮ S: Fluctuation of density ◮ Im χ: Dissipation of energy

25/57

slide-31
SLIDE 31

TDDFPT: Liouville-Lanczos approach

Optical absorption Perturbation: electric field Definition: χ(ω) ≡ Tr

  • ˜

V ′

ext(r, ω) ˆ

ρ′(ω)

  • ˆ

ρ′(ω) =? Quantum Liouville equation: [ ˆ HKS(t), ˆ ρ(t) ] = i ∂ ∂t ˆ ρ(t) Linearization + Fourier transformation: (ω − ˆ L) · ˆ ρ′(ω) = [˜ V ′

ext(ω), ˆ

ρ0] ˆ L · ˆ ρ′ ≡ [ˆ H0

KS, ˆ

ρ′] + [ˆ VHXC, ˆ ρ0] χ(ω) = ˜ V ′

ext(ω)|(ω − ˆ

L)−1[˜ V ′

ext(ω), ˆ

ρ0] ⇓ Use of Lanczos recursion method EELS Perturbation: electron beam Definition: χ(q, q; ω) ≡ Tr

  • ˜

V ′

ext,q(r, ω) ˆ

ρ′

q(ω)

  • ˆ

ρ′

q(ω) =?

Quantum Liouville equation: [ ˆ HKS(t), ˆ ρq(t) ] = i ∂ ∂t ˆ ρq(t) Linearization + Fourier transformation: (ω − ˆ L) · ˆ ρ′

q(ω) = [˜

V ′

ext,q(ω), ˆ

ρ0] ˆ L · ˆ ρ′

q ≡ [ˆ

H0

KS, ˆ

ρ′

q] + [ˆ

VHXC,q, ˆ ρ0] χ(q, q; ω) = ˜ V ′

ext,q(ω)|(ω − ˆ

L)−1[˜ V ′

ext,q(ω), ˆ

ρ0] ⇓ Use of Lanczos recursion method 26/57

slide-32
SLIDE 32

TDDFPT: Liouville-Lanczos approach

Optical absorption Perturbation: electric field Definition: χ(ω) ≡ Tr

  • ˜

V ′

ext(r, ω) ˆ

ρ′(ω)

  • ˆ

ρ′(ω) =? Quantum Liouville equation: [ ˆ HKS(t), ˆ ρ(t) ] = i ∂ ∂t ˆ ρ(t) Linearization + Fourier transformation: (ω − ˆ L) · ˆ ρ′(ω) = [˜ V ′

ext(ω), ˆ

ρ0] ˆ L · ˆ ρ′ ≡ [ˆ H0

KS, ˆ

ρ′] + [ˆ VHXC, ˆ ρ0] χ(ω) = ˜ V ′

ext(ω)|(ω − ˆ

L)−1[˜ V ′

ext(ω), ˆ

ρ0] ⇓ Use of Lanczos recursion method EELS Perturbation: electron beam Definition: χ(q, q; ω) ≡ Tr

  • ˜

V ′

ext,q(r, ω) ˆ

ρ′

q(ω)

  • ˆ

ρ′

q(ω) =?

Quantum Liouville equation: [ ˆ HKS(t), ˆ ρq(t) ] = i ∂ ∂t ˆ ρq(t) Linearization + Fourier transformation: (ω − ˆ L) · ˆ ρ′

q(ω) = [˜

V ′

ext,q(ω), ˆ

ρ0] ˆ L · ˆ ρ′

q ≡ [ˆ

H0

KS, ˆ

ρ′

q] + [ˆ

VHXC,q, ˆ ρ0] χ(q, q; ω) = ˜ V ′

ext,q(ω)|(ω − ˆ

L)−1[˜ V ′

ext,q(ω), ˆ

ρ0] ⇓ Use of Lanczos recursion method 26/57

slide-33
SLIDE 33

Pros & Contras

Conventional TDDFT approach

Numerous empty states

χ0(ω) =

  • v,c

(fv−fc)ϕc(r)ϕ∗

v(r)ϕv(r′)ϕc(r′)

ω − (εc − εv) + i η

Multiplication and inversion of

large matrices χ = χ0 + χ0 (vCoul + fxc) χ

Calculation of χ0 and χ must be

repeated for each frequency

Scaling:

[Nv × Nc × Nk × N2

G + N2.4 G ] × Nω

Approximations beyond the

adiabatic one are possible: fxc(ω) Liouville-Lanczos approach

No empty states (use of DFPT

techniques)

No matrix inversions (use of

Lanczos recursion method)

Lanczos recursion has to be done

  • nce for all frequencies

Scaling: Only a few times larger

than ground-state DFT calculations: α[Nv × Nk × NPW ln NPW] × Niter

Limitation by the adiabatic aproxi-

mation: static fxc 27/57

slide-34
SLIDE 34

Testing of the Liouville-Lanczos approach

28/57

slide-35
SLIDE 35

Testing of the Liouville-Lanczos approach

29/57

slide-36
SLIDE 36

Testing of the Liouville-Lanczos approach

30/57

slide-37
SLIDE 37

Experimental EEL spectrum of Bi for q → 0

Ab initio calculations are needed to understand the origin of 4 features.

  • C. Wehenkel et al., Solid State Comm. 15, 555 (1974)

31/57

slide-38
SLIDE 38

Comparison between experiment and theory

◮ Four features in the EEL spectrum are well reproduced. ◮ Accurate description of the the broad structure in 40 - 100 eV range.

32/57

slide-39
SLIDE 39

Effect of the spin-orbit coupling (SOC)

33/57

slide-40
SLIDE 40

Effect of the spin-orbit coupling (SOC)

◮ Integrated intensity is improved by SOC. ◮ Red-shift of peaks in the range 20 - 30 eV, due to splitting of 5d levels.

34/57

slide-41
SLIDE 41

Origin of the peaks between 20 - 30 eV

Interband transitions from the 5d semicore levels to lowest unoccupied levels. 35/57

slide-42
SLIDE 42

Effect of the 5d semicore levels

Ionization from 5d semicore levels = ⇒ broad structure between 40 - 100 eV. 36/57

slide-43
SLIDE 43

Plasmon dispersion

◮ Increase of q =

⇒ blue-shift of the plasmon peak.

◮ Plasmon enters in electron-hole continuum =

⇒ broadening of spectrum. 37/57

slide-44
SLIDE 44

Conclusions (I)

◮ Developed a new method for EELS - Liouville-Lanczos

approach;

◮ The new method is computationally more efficient then

conventional TDDFT method;

◮ The new method tested successfully on bulk Si and Al; ◮ First ab initio calculations of the EEL spectra in bulk Bi.

38/57

slide-45
SLIDE 45

Outline

  • 1. Introduction

1.1 Motivation 1.2 Material: Bismuth 1.3 State of the art methods

  • 2. Results

2.1 High-energy response: new approach for EELS 2.2 Low-energy response: free-carrier response

  • 3. Conclusions

39/57

slide-46
SLIDE 46

Low-energy response

Optics: q → 0, ω → 0 Drude intraband contribution to the dielectric function: ǫintra(ω) = 1 − ω2

p

ω(ω + iγ)

40/57

slide-47
SLIDE 47

Dielectric properties of Bi in equilibrium: Free carrier response

Time-resolved (pump-probe) terahertz experiment: L. Perfetti, J. Faure,

  • T. Kampfrath, C. R. Ast, C. Frischkorn, M. Wolf.

Circles: Experimental data Solid lines: Fit by Drude model

Drude model:

ǫ(ω) = − ω2

p,eq

ω(ω + iγ) + ǫ∞ ⇓ Fitting: Plasma freq. ωp,eq = 560 meV, Scattering rate γ = 37 meV, and ǫ∞ = 100. The Drude model accurately fits expt. data = ⇒ Free carrier response 41/57

slide-48
SLIDE 48

Photoexcited bismuth (q = 0, ω = 1.6 eV)

42/57

slide-49
SLIDE 49

Photoexcited bismuth (q = 0, ω = 1.6 eV)

∆ǫintra(ω) is the change of the intraband dielectric function due to the photoex- citation of bismuth. ∆ǫintra(ω) displays a free carrier response. Fitting by the Drude model: ∆ǫintra(ω) = 1 − ∆ω2

p

ω(ω + iγ), ω2

p = ω2 p,eq + ∆ω2 p.

43/57

slide-50
SLIDE 50

Photoexcited bismuth (q = 0, ω = 1.6 eV)

44/57

slide-51
SLIDE 51

Photoexcited bismuth (q = 0, ω = 1.6 eV)

44/57

slide-52
SLIDE 52

Hypothesis

Just after the photoexcitation, electrons and holes stick in the true local extrema

  • f the valence and conduction bands.

Drude model: ∆ω2

p = 4πe2∆n

m ⇑ ∆n = ⇒ ⇑ ∆ω2

p

Effective mass approximation for the true local extrema:

  • m∗ −1(k)
  • i j = 1

2 ∂2E(k) ∂ki ∂kj Verification of the hypothesis: compare average effective masses of the true local extrema with optical masses near the T and L points. 45/57

slide-53
SLIDE 53

Optical mass at L and T points

Definition of the optical mass on the basis of the Drude model: ∆ω2

p(T) = 4πe2∆n(T)

mop Semiclassical model: ∆ω2

p(T) = 4πe2

3 v2

F

  • g(E)
  • f ′

FD(E, T0) − f ′ FD(E, T)

  • dE

∆n(T) =

  • g(E) |fFD(E, T) − fFD(E, T0)| dE

where g(E) is the restricted DOS, vF is the Fermi velocity of carriers, fFD is the Fermi-Dirac distribution function. g(E) and vF were calculated from first principles. 46/57

slide-54
SLIDE 54

Photoexcited electrons and holes get stuck in true local extrema

47/57

slide-55
SLIDE 55

Photoexcited electrons and holes get stuck in true local extrema

48/57

slide-56
SLIDE 56

Relaxation of carriers

The relaxation of carriers occurs due to electron-phonon (e-ph) and hole-phonon (h-ph) scattering, and Auger recombination.

49/57

slide-57
SLIDE 57

Two regimes in the evolution of the plasma frequency

Rate equations = ⇒ relaxation times τ Relaxation of electrons and holes, which were stuck in the true local extrema Electron-hole recombination near the Fermi level 50/57

slide-58
SLIDE 58

Ab initio calculation of electron-phonon coupling

Photoemission experiment: L. Perfetti and J. Faure At higher fluence of the photoexcitation (0.6 mJ/cm2), the A1g phonon mode is activated in bismuth. ⇓ Due to electron-phonon interaction, the highest valence bulk band oscillates with the frequency of the A1g phonon mode.

  • E. Papalazarou, I. Timrov, N. Vast, L. Perfetti et al., PRL 108, 256808 (2012).

51/57

slide-59
SLIDE 59

Conclusions (II)

◮ Theoretical description of free carrier response in photoexcited Bi. ◮ Evolution of the plasma frequency displays two regimes due to the

existence of true local extrema in the band structure of Bi.

◮ Relaxation of carriers occurs with a time rate of 0.6 ps, and the

electron-hole recombination occurs with a time rate of 4 ps.

◮ Wavevector-dependence of electron-phonon coupling is in agreement

with experiment. 52/57

slide-60
SLIDE 60

General conclusions (I)

  • 1. Description of the full charge-carrier response in excited bismuth

from low energy to high energy range. Low-energy response: Theoretical model for the de- scription

  • f

the free car- rier dynamics in photoex- cited bismuth. High-energy response: New method for EELS and application to bismuth. 53/57

slide-61
SLIDE 61

General conclusions (II)

  • 2. Importance of the electron-phonon coupling for the interpretation of

photoexcited bismuth. Relaxation times in photoexcited bismuth: 0.6 ps for carrier- phonon scattering, and 4 ps for electron-phonon recombination. Ab initio calculations of wave- vector-dependent electron-pho- non coupling are in good agree- ment with experiment. 54/57

slide-62
SLIDE 62

Perspectives

◮ Spin-orbit coupling: from bulk to surfaces

◮ Surfaces of Bi and of Bi compounds (Bi2Te3, Bi2Se3)

◮ Importance of electron-phonon coupling

◮ Relaxation times in Bi ◮ Thermoelectricity in Bi and Bi compounds (Bi2Te3, Bi2Se3) ◮ Occurrence of charge density waves in some materials

◮ Application of the Liouville-Lanczos approach for large systems

◮ Surface plasmons ◮ Acoustic surface plasmons

55/57

slide-63
SLIDE 63

Collaborations

Laboratoire des Solides Irradiés, École Polytechnique Nathalie Vast, Luca Perfetti, Jelena Sjakste, Jérôme Faure, Paola Gava SISSA – Scuola Internazionale Superiore di Studi Avanzati Stefano Baroni ICTP – The Abdus Salam International Centre for Theoretical Physics Ralph Gebauer

56/57

slide-64
SLIDE 64

Perspectives

◮ Spin-orbit coupling: from bulk to surfaces

◮ Surfaces of Bi and of Bi compounds (Bi2Te3, Bi2Se3)

◮ Importance of electron-phonon coupling

◮ Relaxation times in Bi ◮ Thermoelectricity in Bi and Bi compounds (Bi2Te3, Bi2Se3) ◮ Occurrence of charge density waves in some materials

◮ Application of the Liouville-Lanczos approach for large systems

◮ Surface plasmons ◮ Acoustic surface plasmons

57/57