Tuning the Transport Properties of Layered

Materials for Thermoelectric Applications using

First-Principles Calculations

Dissertation by
Yasir Saeed

Submitted in Partial Fulfillment of the Requirements for the

degree of
Doctor of Philosophy

King Abdullah University of Science and Technology

Physical Science and Engineering

Materials Science and Engineering

Thuwal, Makkah Province, Kingdom of Saudi Arabia

May, 2014
 2
EXAMINATION COMMITTEE APPROVALS FORM

The dissertation of Yasir Saeed is approved by the examination committee.

Committee Chairperson: Udo Schwingenschlogl

Committee Member: Muhammad M. Hussain

Committee Member: Aurelien C. Manchon

Committee Member:Husam N. Alshareef

Committee Member: Iman Roqan

King Abdullah University of Science and Technology

2014
 3

Copyright May 2014

Yasir Saeed
All Rights Reserved
 4

TUNING THE TRANSPORT PROPERTIES OF

LAYERED MATERIALS FOR THERMOELECTRIC

APPLICATIONS USING FIRST-PRINCIPLES

CALCULATIONS

Yasir Saeed
ABSTRACT

Thermoelectric materials can convert waste heat into electric power and thus
provide a way to reduce the dependence on fossil fuels. Our aim is to model the
underlying materials properties and, in particular, the transport as controlled by
electrons and lattice vibrations. The goal is to develop an understanding of the
thermoelectric properties of selected materials at a fundamental level.
The structural, electronic, optical, and phononic properties are studied in or-
der to tune the transport, focusing on Kx RhO2 , Nax RhO2 , PtSb2 and Bi2 Se3 . The
investigations are based on density functional theory as implemented in the all elec-
tron linearized augmented plane wave plus local orbitals WIEN2k and pseudopo-
tential Quantum-ESPRESSO codes. The thermoelectric properties are derived
from Boltzmann transport theory under the constant relaxation time approxima-
tion, using the BoltzTraP code.
We will discuss first the changes in the electronic band structure under varia-
tion of the cation concentration in layered Kx RhO2 in the 2H phase and Nax RhO2
in the 3R phase. We will also study the hydrated phase. The deformations of
the RhO6 octahedra turn out to govern the thermoelectric properties, where the
 5
high Seebeck coefficient results from pudding mold bands. We investigate the
thermoelectric properties of electron and hole doped PtSb2 , which is not a layered
material but shares pudding mold bands. PtSb2 has a high Seebeck coefficient
at room temperature, which increases significantly under As alloying by bandgap
opening and reduction of the lattice thermal conductivity.
Bi2 Se3 (bulk and thin film) has a larger bandgap then the well known thermo-
electric material Bi2 Te3 , which is important at high temperature. The structural
stability, electronic structure, and transport properties of one to six quintuple
layers of Bi2 Se3 will be discussed. We also address the effect of strain on a sin-
gle quintuple layer by phonon band structures. We will analyze the electronic
and transport properties of Tl-doped Bi2 Se3 under strain, focusing on the giant
Rashba spin splitting (Tl doping breaks the inversion symmetry in Bi2 Se3 ) and its
dependence on biaxial tensile and compressive strain.
 6

ACKNOWLEDGMENTS

In the name of ALLAH, the Most Gracious and the Most Merciful

Alhamdulillah, all praises to Allah for the strengths and His blessing in complet-
ing this dissertation. Thanks to the vision of King Abdullah bin Abdulaziz
(custodian of the two holy mosques), for establishing such a nice graduate-level
research university. Special appreciation goes to my supervisor, Prof. Udo Schwin-
genschlogl, for his kind supervision and constant support, his valuable suggestions
and extreme cooperation throughout my research work.
I would like to express my appreciation to the members of the thesis commit-
tee: Prof. Muhammad M. Hussain, Prof. Aurelien C. Manchon, Prof. Husam N.
Alshareef, and Prof. Iman Roqan, for their inspiration and encouragement dur-
ing the research activities through fruitful discussion and suggestions whenever
needed.
I feel honored to express special thanks to Prof. Nazma Ikram (Ex. Chairper-
son of CSSP, University of Punjab, Pakistan and M. Phil supervisor) and Prof.
Shaukat Ali (Chairman Department of Physics, University of Sargodha, Pakistan),
who have boosted my research carrier with lots of courage and support. Also spe-
cial thanks to postdoctoral fellow Dr. Nirpendra Singh, who helped and guided me
throughout my research. Without his great support I could not have accomplished
this work. Sincere thanks to all my friends, especially to Dr. Safdar Nazir, Dr.
B. Amin, and Dr. T. P. Kaloni, for fruitful discussions. I would like to thank the
KAUST research computing team for providing high performance computational
 7
resources.
Last but not least, my deepest gratitude goes to my beloved parents, Mr. Saeed
Ahmed Awan and Mrs. Mukhtiar Saeed (for their prayers and faith in me), and to
my brothers and sisters, not to forget my wife Fiza and children (Abdul Rehman
and Haleema) for providing me great support, which encouraged me to reach this
stage.
This dissertation is dedicated to my parents.
 8

TABLE OF CONTENTS

LIST OF ILLUSTRATIONS 10

LIST OF TABLES 14

I Introduction 15
1.1 Thermoelectric Phenomena . . . . . . . . . . . . . . . . . . . . . . 17
1.1.1 Figure of Merit and Efficiency . . . . . . . . . . . . . . . . . 18
1.1.2 Seebeck Coefficient . . . . . . . . . . . . . . . . . . . . . . . 21
1.1.3 Electrical Conductivity . . . . . . . . . . . . . . . . . . . . . 21
1.1.4 Thermal Conductivity . . . . . . . . . . . . . . . . . . . . . . 23
1.2 Thermoelectric Applications . . . . . . . . . . . . . . . . . . . . . . 25
1.3 Advantages and Disadvantages . . . . . . . . . . . . . . . . . . . . 26
1.4 Transition Metal Oxide Thermoelectrics . . . . . . . . . . . . . . . 27
1.5 Bismuth Based Binary Thermoelectrics . . . . . . . . . . . . . . . 30
1.6 The pudding mold Bands in PtSb2 . . . . . . . . . . . . . . . . . 33
1.7 Objective and Scope of the Thesis . . . . . . . . . . . . . . . . . . 33

II Method of Calculations 35
2.1 Density Functional Theory . . . . . . . . . . . . . . . . . . . . . . 35
2.1.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.2 Hohenberg-Kohn Theorems . . . . . . . . . . . . . . . . . . . 36
2.1.3 The Kohn-Sham Equations . . . . . . . . . . . . . . . . . . . 37
 9
2.2 Computational Modules . . . . . . . . . . . . . . . . . . . . . . . . 39
2.2.1 WIEN2k Code . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.2.2 Quantum-ESPRESSO Code . . . . . . . . . . . . . . . . . . 39
2.3 Modeling of Thermoelectric Materials . . . . . . . . . . . . . . . . 40
2.3.1 Transport Coefficients . . . . . . . . . . . . . . . . . . . . . . 40
2.3.2 Calculation of the Transport Coefficients using BoltzTraP . . 41

III Kx RhO2 in the 2H Phase 44

3.1 Computational Parameters . . . . . . . . . . . . . . . . . . . . . . 44
3.2 Structural Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3 Electronic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.4 Optical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.5 Transport Coefficients and Figure of Merit . . . . . . . . . . . . . . 49

IV Nax RhO2 in the 3R Phase 53

4.1 Computational Parameters . . . . . . . . . . . . . . . . . . . . . . 53
4.2 Structural Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3 Electronic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4 Seebeck Coefficient and Figure of Merit . . . . . . . . . . . . . . . 59

V Nanostructured Bi2 Se3 62

5.1 Computational Parameters . . . . . . . . . . . . . . . . . . . . . . 62
5.2 Structural Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.3 Electronic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.4 Phonon Band Structure . . . . . . . . . . . . . . . . . . . . . . . . 66
5.5 Transport Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.6 Figure of Merit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

VI Tl-doping in Bulk Bi2 Se3 71

6.1 Computational Parameters . . . . . . . . . . . . . . . . . . . . . . 71
6.2 Structural Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 72
 10
6.3 Electronic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.3.1 Rashba Spin Splitting . . . . . . . . . . . . . . . . . . . . . . 73
6.4 Transport Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.4.1 Variation of S, / , and S 2 / at 300 K . . . . . . . . . . . 77
6.4.2 Variation of S, / , and S 2 / at 500 K . . . . . . . . . . . 78
6.5 Powerfactor and Figure of Merit . . . . . . . . . . . . . . . . . . . 79

VII Alloying PtSb2 with Ir/As 82

7.1 Computational Parameters . . . . . . . . . . . . . . . . . . . . . . 82
7.2 Structural Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.3 Electronic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.4 Transport Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . 84
7.4.1 Variation of S and / . . . . . . . . . . . . . . . . . . . . . 84
7.4.2 Powerfactor and Figure of Merit . . . . . . . . . . . . . . . . 87
7.5 Arsenic Alloying in PtSb2 . . . . . . . . . . . . . . . . . . . . . . . 88

VIII Conclusion 91

REFERENCES 95

PUBLICATIONS 108
 11

LIST OF ILLUSTRATIONS

1.1 Sources of waste heat. . . . . . . . . . . . . . . . . . . . . . . . . . . 16

1.2 Illustration of a TE generator (Seebeck effect) and a TE cooler
(Peltier effect). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.3 S, , S 2 and as a function of the free-charge carrier concentration. 19
1.4 Figure of merit improvement over the years in TE oxide materials. . 27
1.5 Illustration of 2H and 3R structures of prototype Nax CoO2 . . . . . . 29
1.6 Crystal structure of Bi2 Se3 along the c-axis. . . . . . . . . . . . . . . 31

3.1 Crystal structures of Kx RhO2 for x = 1/2 and 7/8. . . . . . . . . . . 45

3.2 Volume optimization of K1/2 RhO2 for different exchange correlation
functionals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.3 Band structures of K1/2 RhO2 and K7/8 RhO2 . . . . . . . . . . . . . . 47
3.4 Density of states obtained for K1/2 RhO2 and K7/8 RhO2 . . . . . . . . 48
3.5 Band structures of hydrated K1/2 RhO2 and K7/8 RhO2 . . . . . . . . . 49
3.6 Band structure of Na1/2 CoO2 with c = 15.59 A. . . . . . . . . . . . . 50
3.7 Optical reflectivity and conductivity of K1/2 RhO2 and K7/8 RhO2
along with experimental data. . . . . . . . . . . . . . . . . . . . . . . 51
3.8 Thermoelectric properties of pristine (solid line) and hydrated (dashed
line) Kx RhO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

4.1 Atomic structures of NaRhO2 , Na0.83 RhO2 , and Na0.75 RhO2 for dif-
ferent vacancy configurations. . . . . . . . . . . . . . . . . . . . . . . 55
 12
4.2 Band structures and Rh 4d density of states of NaRhO2 , Na0.83 RhO2 ,
and Na0.75 RhO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3 Band structures and Rh 4d density of states of hydrated A3-Na0.83 RhO2
and B3-Na0.75 RhO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4 Density of states for non-hydrated Na0.83 RhO2 and Na0.75 RhO2 , com-
pared to pristine NaRhO2 . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.5 Transport properties of non-hydrated and hydrated Na0.83 RhO2 and
Na0.75 RhO2 as functions of the temperature. . . . . . . . . . . . . . . 60
4.6 Transport properties of non-hydrated Na0.83 RhO2 and Na0.75 RhO2
as functions of the temperature. . . . . . . . . . . . . . . . . . . . . . 61

5.1 Structures of one to six quintuple layers of Bi2 Se3 after relaxation. . 63
5.2 Band structures of bulk Bi2 Se3 with and without vdW interaction. . 65
5.3 Band structures of one to six quintuple layers of Bi2 Se3 . . . . . . . . 65
5.4 Bandgap as a function of the number of quintuple layers of Bi2 Se3 . . 66
5.5 Electronic band structures and phonon dispersions of a single quin-
tuple layer of Bi2 Se3 for different strains. . . . . . . . . . . . . . . . . 67
5.6 Bandgap and point phonon frequency of a single quintuple layer
of Bi2 Se3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.7 Thermoelectric properties for systems of different thickness and for
various strains at 300 K. . . . . . . . . . . . . . . . . . . . . . . . . . 69

6.1 Band structure for zero, compressive, and tensile strain. . . . . . . . 73

6.2 Band structures of Tl-doped Bi2 Se3 without and with spin orbit cou-
pling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.3 Spin structure of the Rashba split states in the xy-plane. . . . . . . . 75
6.4 Bandgap at the point and Rashba parameter at the conduction
band minimum for electrons and at the valence band maximum for
holes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
 13
6.5 Room temperature: S, / , and S 2 / under strain as a function of
the carrier concentration. . . . . . . . . . . . . . . . . . . . . . . . . 78
6.6 500 K: S, / , and S 2 / under strain as a function of the carrier
concentration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.7 S and / as a function of the temperature under strain for a fixed
p-doping of 71020 cm3 . . . . . . . . . . . . . . . . . . . . . . . . . 80

7.1 Band structure of doped Pt0.99 Ir0.01 Sb2 . . . . . . . . . . . . . . . . . 83

7.2 Density of states of undoped and doped PtSb2 . . . . . . . . . . . . . 84
7.3 S and / of hole doped PtSb2 . . . . . . . . . . . . . . . . . . . . . . 85
7.4 S and / of electron doped PtSb2 . . . . . . . . . . . . . . . . . . . 86
7.5 S 2 and ZT for electron and hole doped PtSb2 as functions of the
temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.6 Thermopower of PtAs2 for several hole and electron dopings. . . . . 90
 14

LIST OF TABLES

1.1 Thermoelectric materials with high zT . . . . . . . . . . . . . . . . . 19

1.2 Comparison of the TE properties of metal, semiconductor and insu-
lator at room temperature. . . . . . . . . . . . . . . . . . . . . . . . 20

4.1 Structural properties of NaRhO2 , Na0.83 RhO2 , and Na0.75 RhO2 for
different vacancy configurations. . . . . . . . . . . . . . . . . . . . . 56
 15

Chapter I

Introduction

In recent years, energy demand has been increasing as populations continue to

grow and other resources such as oil and gas (which causes environmental issues
like global warming) are been consumed. Solutions to this energy problem must
be found in cost effective and environmental friendly fashion. Renewable energy
sources like wind, geothermal, solar, and tidal, etc are generating interest of the
scientific community to increase their efficiency and make them practically ap-
plicable. The above listed sources are constrained by certain limitations such as
wind-force to drive a windmill, sunlight for full day, tidal for coastal area only, and
earth quick safety for geothermal plant etc. On the other hand, thermoelectric
(TE) generators utilize waste heat to generate electricity. TE refrigerators can
develop a temperature gradient), which can be used for cooling in computers, for
example. TE generators/refrigerators have many notable advantages over existing
technologies like gas compressor based refrigerators and fuel based electric power
generators [17]: They are
1) reliable and silent in operation (no mechanical moving parts and less mainte-
nance),
2) small in size and of light weight,
3) capable of operating at both low and high temperatures,
4) environmentally friendly,
 16
5) able to reuse waste heat, and
6) not location-dependent (as solar and wind technologies).

The major drawback of TE devices is the low conversion efficiency (about 12%)
as compered to conventional devices. Environmentally friendly TE materials will
need to have a better efficiency than state-of-the-art materials like Bi2 Te3 /Sb2 Te3 ,
PbTe/PbSeTe and PbSnTe/PbSnSeTe, and should not include Te (trace element)
and Pb (toxic) [8, 9].

Thermal
Automobile Factory Incinerator
plant

34% 66%
Used energy Energy loss

Waste Heat

Electrical
Thermoelectric Energy
generators

Figure 1.1: Sources of waste heat.

TE devices with high efficiency could make it possible to use waste heat from
currently available sources, see Fig. 1.1. The sources of automobile waste heat, e.g.,
exhaust and engines, can be reused in future automobiles. Glass/Aluminum in-
dustries, thermal plants, and incinerator plants could provide a unique new source
of waste energy. TE devices have key role as an alternate energy source and be-
 17
come one of the solutions to reduce our dependence on fossil energies. Further deep
scientific research in this field will be required to create new, low cost, environmen-
tal friendly, and high efficiency TE materials. Nowadays TE research is divided
into two directions, one searching for new bulk materials with high efficiency and
the other manipulating existing materials (Bi2 Te3 /Sb2 Te3 , PbTe/PbSeTe and Pb-
SnTe/PbSnSeTe) at the nanoscale to reduce the thermal conductivity and thus
achieve a high efficiency. Reducing the dimension (from bulk to nanosheets (thin
films) to nanowires to quantum dots) yields quantum confinement. Introduction
of grain boundaries plays a major role in reducing the thermal conductivity, re-
sulting in enhanced TE properties (reduced thermal conductivity and increased
Seebeck coefficient) [1013].

1.1 Thermoelectric Phenomena

Seebeck effect Peltier effect

Heat source Cooled surface

n-type p-type n-type p-type

Cold end Dissipated heat

Resistance

(a) Power generation (b) Refrigeration

Figure 1.2: Illustration of a TE generator (Seebeck effect) and a TE cooler (Peltier

effect).
 18
Figure 1.2 (a) shows an illustration of a TE generator (Seebeck effect). The
simplest TE generator consists of a thermocouple, comprised of p-type and n-type
materials connected electrically in series and thermally in parallel. Heat is sup-
plied on one side (hot side) of the couple and a potential difference is produced
proportional to the temperature gradient between the hot and cold junctions, as
hot electrons/holes move towards the cold side. Figure 1.2 (b) shows an illustra-
tion of a TE cooler (Peltier effect), which converts the potential difference into
a temperature gradient. If a voltage is applied to the electrodes, it induces an
electrical current, i.e., thermal energy flows in the direction of the charge carriers.

1.1.1 Figure of Merit and Efficiency

The maximum efficiency of a TE material depends on (1) the Carnot efficiency

term (cannot exceed the Carnot efficiency) and (2) a term describing the TE
properties (S, , and ). The figure of merit is defined as

S 2
zT = T, (I.1)

where is the electrical conductivity (1/cm), S is the Seebeck coefficient (VK1 ),

and T is the temperature. Moreover, is the thermal conductivity (W/cmK),
which is further divided in lattice (l ) and electronic (e ) contributions, = l +e
(more details are given in the next section). Some well known TE materials from
Ref. [20] are shown in Table 1.1.
In order to achieve a high figure of merit, the TE material, in general, should
be a good electrical and poor thermal conductor and, at the same time, possess a
high Seebeck coefficient. However, increasing by increasing the carrier concen-
tration usually decreases S and increases e . Therefore, it is a prime task to control
the numerator S 2 (the powerfactor) and the denominator independently. Al-
loying, vacancies, and the application of strain are effective and frequently used
approaches to improve the TE properties, because the phonon scattering is en-
 19

Table 1.1: Thermoelectric materials with high zT .

Compound Chemical formula zT T
Bismuth Telluride Bi2 Te3 0.6 400
Lead Telluride PbTe 0.8 773
TAG-85 (AgSbTe2 )0.15 (GeTe)0.85 1.4 750
Cs-Bismuth Telluride CsBi4 Te6 0.85 255
Clathrate-1 Ba8 Ga16 Ge30 1.35 900
La-Skutterudites LaFe3 CoSb12 1.40 1000
LAST AgPbm SbTe2 1.7 700
SALT NaPbm SbTe2+m 1.6 675
LASTT Ag(Sn,Pb)m SbTe2+m 1.4 700
Zinc Antimonide -Zn4 Sb3 1.3 660
Silicon Germanium Si1x Gex 0.9 900
Mo-Antimonide No0.06 Mo3 Sb5.4 Te1.6 0.93 1023

hanced and the thermal conductivity consequently reduced [1023].

Figure 1.3: S, , S 2 and as a function of the free-charge carrier concentration.

Metals are poor TE materials because they have a low Seebeck coefficient and
large electronic contribution to the thermal conductivity. Insulators have a high
Seebeck coefficient and a small electronic contribution to the thermal conductivity.
Their charge density is low and thus their electrical conductivity is low, leading to
a low TE effect. The best materials are semiconductors with an electronic density
around 1019 cm3 [12], see Fig. 1.3 and Table 1.2. Figure 1.3 illustrates that the
 20
Seebeck coefficient is reduced as the carrier concentration increases from insulator
to semiconductor to metal. The electrical conductivity shows the opposite behav-
ior. In order to reach the optimum figure of merit where both lines intersect has
to be chosen. Also it is clear that a decrease in electronic thermal conductivity
would also result in a decrease in electrical conductivity and hence, decrease in the
figure of merit. Further, the contribution of lattice thermal conductivity is less
than electronic part at optimum carrier concentration of 1019 cm3 . Therefore,
up to a specific carrier concentration doping would have a positive impact on the
figure of merit. By reducing the lattice thermal conductivity, the figure of merit
could be improved further without a negative influence on the other variables.

Table 1.2: Comparison of the TE properties of metal, semiconductor and insulator

at room temperature.
Properties Metal Semiconductor Insulator
S (VK1 ) 5 200 1000
(1/cm) 106 103 1012
Z (K1 ) 3106 2103 51017

The figure of merit of a TE device is given by

2
Sh Se
ZTavg =
r h r e Tavg ,
(I.2)
+
h e

where h and e represent hole and electron doping, respectively. Tavg is the average
Thot + Tcold
system temperature, Tavg = . The device performance relies directly
2
on the temperature difference T = Thot Tcold and the material parameter ZTavg .
The efficiency is defined by the Carnot efficiency [13] as

p
T 1 + ZTavg 1
= (I.3)
Thot p Tcold
1 + ZTavg +
Thot
 21
1.1.2 Seebeck Coefficient

The Seebeck coefficient (or thermopower) S of a material is a measure of the

induced TE voltage in response to a temperature difference across a material. We
have [28]
  
E EF
Z
kB 1 f0 (E)
S= (E) dE, (I.4)
e 0 kB T E

where kB , e, , EF and f0 are the Boltzmann constant, electron charge, electrical

conductivity, Fermi energy and Fermi distribution function, respectively. In the
case of metals and degenerately doped semiconductors the last equation simplifies
to
2 kB
2
 
T d {ln[(E)]}
S= , (I.5)
3e dE E=EF

which is of broad validity.

1.1.3 Electrical Conductivity

The electrical conductivity () represents the materials ability to let charge car-
riers (electrons or holes) pass through its crystal lattice. According to this idea,
it can be calculated from the charge carriers density (n) and carrier mobility ()
by the relation
= ne, (I.6)

where
e
= (I.7)
m

depends on the relaxation/scattering time (time between collisions of carriers)

and the carriers effective mass m , which is directly linked to the band dispersion
[24].
The mobility depends on the drift velocity, which is determined by the relax-
ation/scattering time , i.e., how long the carrier is accelerated by the electric field
until it collides and changes its direction and/or energy. Sources of scattering in
 22
semiconductors are ionized impurities, neutral impurities, electron-phonon inter-
action (acoustic phonon scattering, optical phonon scattering, and piezoelectric
scattering), and electron-electron interaction [29]. The total mobility is calculated
from the contributions of the different scattering effects as

1 1 1 1 1
= + + + (I.8)
T otal Ionized N eutral elph elel

Ionized impurity scattering: Due to Coulomb forces, an electron/hole approaching

an ionized impurity is deflected. The amount of deflection depends on the velocity
of the carrier and its proximity to the ion. In heavily doped materials the colli-
sion probability is higher so that the mean free time between collisions and thus
the mobility are smaller. The mobility in ionized impurity scattering is directly
proportional to the temperature, T 3/2 .
Neutral impurity scattering: Scattering due to unionized donors in semiconduc-
tors becomes dominant at low temperature as carriers freeze out at the impurity
(neutral shallow-level impurities). While the mobility is formally independent of
the temperature, experiments show a weak dependence for many semiconductors.
Neutral impurity scattering is usually strong only at very high carrier concentra-
tions (above 1019 cm3 ).
Electron-phonon scattering: The scattering of electrons by acoustical phonons is
important in lightly doped or intrinsic semiconductors. An acoustical phonon may
cause scattering either by a deformation potential or by piezoelectric scattering.
The change in atomic spacing due to the acoustic wave results in a change of the
energy bandgap on an atomic scale. The mobility due to this deformation poten-
tial is directly proportional to T 3/2 . The electron mobility due to piezoelectric
scattering varies as T 1/2 .
Electron-electron scattering: Electrons can be considered as non-interacting if
their density does not exceed 1017 cm3 . Above this limit the electron-electron
scattering starts to dominate.
 23
1.1.4 Thermal Conductivity

In the thermal conductivity

= e + l , (I.9)

the electronic contribution e = LT represents the heat transport contributed by

the moving charge carriers. This can be approximated by the Wiedemann-Franz
law [25] for metals and narrow bandgap semiconductors [26]. The Wiedemann-
Franz law states that the ratio of the thermal conductivity to the electrical con-
ductivity of a metal is proportional to the temperature, i.e., e = LT . Here L
is the Lorenz number, as theoretically derived by Sommerfeld for metals [27] and
heavily doped semiconductors (2.45108 WK2 ). The Wiedemann-Franz law
gives a suitable approximation of the electronic contribution by only considering
.
1
The lattice contribution l = CV lv is the heat contributed by lattice vibra-
3
tions (phonons). Using the kinetic theory of gases, l depends on the specific heat
of the material (CV ), the mean free path of the phonons (l), and the speed of
sound (v). It can be described by a Debye model as

3 Z /T
x4 e x

kB kB
l = 2 dx, (I.10)
2 v } 0 (ex 1)2

where x = }/kB T , is the phonon frequency, kB is the Boltzmann constant, }

is Plancks constant, and v is the sound velocity. The following scattering mech-
anisms contribute to the thermal conductivity [30]: (a) three-phonon scattering
(or phonon-phonon scattering), normal and umklapp processes, (b) point defects,
(c) electron-phonon scattering, and (d) grain boundary scattering. Therefore, the
total relaxation time is given by

1 = N1 + U1 + P1D + EP
1
+ B1 . (I.11)
 24
Assuming that the lattice thermal conductivity of the undoped compound is dom-
inated by normal and umklapp processes, U is given by

3 3 2/3 2
1 kB kB T n 2
U1 = x, (I.12)
b0 4 2 v } 2 Mave
2 3

where b0 = 3.04 107 s3 K 3 , is the Debye temperature, Mave is the average

mass, is the Gruneisen constant, and is the cube root of the atomic volume.
We have
N 1
N1 = (I.13)
U U

and
4
P1D = 3 , (I.14)
4v 3

where is the scattering parameter. For impurity atoms on a single atomic site
= fi (1 Mi /Mave )2 . Here, fi is the fractional concentration of impurity atom
P

i and Mi is its mass. Moreover,

1 Ed2 m3 v kB T
EP =
4}4 d 1 2
mv
2
1 2

2 2 2

} 1 + exp[( m v EF )/kB T + } /8m v kB T + }/2kB T ]

ln 2 ,
kB T 1 2 2 2 2
1 + exp[( m v EF )/kB T + } /8m v kB T }/2kB T ]


2
(I.15)

where Ed is the deformation potential and d is the density. Finally, the relaxation
time due to the grain boundary scattering is [31]

v
B1 = , (I.16)
l

where l and v are the average grain size length and drift velocity, respectively.
 25
1.2 Thermoelectric Applications

TE materials have applications in a multitude of fields. Space missions, such as

moon and mars, continue for weeks and continuous supply of electric power is
required. A radioisotope TE generator provides a solution and has been used suc-
cessfully over the last four decades [32]. It converts the thermal energy gradient by
radioactive decay into electric energy. Plutonium (238 Pu) is the best radioisotope
due to its long half-life ( 88 years). Solar cells are another effective source of
electrical energy but require sufficient solar flux.
For a typical gasoline-powered vehicle based on combustion engine approxi-
mately 25% of the fuel energy is converted into mobility [33]. Regardless of the
fact that the temperature of exhaust gas is roughly 800 K, the effective tem-
perature for TE conversion is usually lower due to difficulties in extracting the
heat from the exhaust. The cold side is maintained at 400 K. Electricity can
be generated even from a moderate temperature difference between the cold and
hot sides [34]. This electrical energy can be utilized in the air condition system.
TE climate-control seat systems have been developed and currently are used to
provide rapid cooling and heating in the summer and winter, respectively [11].
All electronic devices also generate waste heat which needs to be readily dissi-
pated for optimum performance of the devices. Only TE technology is capable of
providing rapid cooling to the devices with soundless operation than conventional
fan-based cooling systems. For example, TE-powered cooling systems (known as
Peltier coolers), for electronic microprocessors such as central processing unit for
computers have been employed to boost personal-computer performance speeds.
However, due to the low efficiency, current TE devices do not show sufficient per-
formance. Also, thin film based TE devices can be employed to maintain the
temperature of laser diodes at a constant level for stabilizing the wavelength.
TE technology will have a larger commercial market when the efficiency reaches
that of conventional mechanical applications. Commercial solid-state heating and
 26
air-conditioning systems could become practical and would replace the present
systems, which emits a greenhouse gases [11]. The speed and operating life of sev-
eral electronic devices (circuits) decreases gradually with increasing temperature.
Effective thermal management is thus the key factor determining the performance
and operating life of electronic devices. If a figure of merit value of 2 will be
achieved, Peltier coolers could operate more efficiently, offering more cooling with
less input of electric power. Industries like steel, cement, aluminum and glass
manufacturing is reliable source of waste-heat recovery [35]. TE generators could
be install in window glasses (the temperature gradient between outdoor and in-
door of a building) which could be result in generation of electrical energy [3639].
Based on these considerations, success in increasing the efficiency of TE materials
could modernize cooling and electrical energy related markets.

1.3 Advantages and Disadvantages

One of the major positive sides of using TE devices lies in the fact that they can
derive their power from waste heat that would otherwise just dissipate into its
surroundings compared to standard fossil generator. TE devices, possessing no
moving or mechanical parts allows these devices to function for a substantially
long time without the need for repair. Moreover, the lack of moving parts guar-
antees that these devices are extremely quiet. Comparing a TE refrigerator to a
standard refrigerator, the TE device is both quiet and light as it needs no compres-
sor. Another advantage for the environment is that there are no bi-products, such
as chlorofluorocarbon gas, which was used in old refrigerators. The main disad-
vantage of TE devices is their low efficiency. A typical efficiency for a combustion
engine is about 35%, while TE devices based on Bi2 Te3 and PbTe currently
achieve upto 12%.
 27
1.4 Transition Metal Oxide Thermoelectrics

Oxides have narrow conduction bands due to weak orbital overlap, which leads
to localized electrons with low carrier mobilities. Transition metal oxides have
recently been considered as TE materials that can operate at high temperature.
Layered cobalt oxides are of technological interest because of their transport prop-
erties (high S and low ) [4045].
Inspired by the striking TE performance of NaCo2 O4 , a lot of current research
is focused on Co-based layered oxides such as Ca3 Co4 O9 and Bi2 Sr3 Co2 Oy (CoO6
octahedra give rise to conducting layers) that crystallize in misfit layered structures
[46]. Al-doped ZnO (Al0.02 Zn0.98 O) exhibits a high TE performance among the
n-type oxides (zT 0.3 at 1000 K) [47]. Additionally, adjusted layer cobalt oxides
have been found promising as p-type materials. The n-type oxides studied so far,
such as doped ZnO and doped SrTiO3 [48], show rather low zT values, see Fig.
1.4.

4 p-type single crystal

p-type polycrystal
3
n-type single crystal SrTiO/SrO
2 n-type polycrystal

BaSrCoO Ca3Co4O9
1
NaxCoO2
ZTmax

NaxCoO2 (Ga,Al):ZnO

Ga:Ca3Co4O9
Ge:In2O3
La/Nb:SrTiO3
Nb:CaMnO3
Eu:Ca3Co4O9
Nb:SrTiO3
(La,Sr)CoO3
0.1
2000 2002 2004 2006 2008 2010
Year
Figure 1.4: Figure of merit improvement over the years in TE oxide materials.

Among them, the layered cobaltates Nax CoO2 have attracted remarkable at-
 28
tention due to the unexpected discovery of a high S = 100 VK1 (10 times higher
than in a normal metal, see Table 1.2, which results in a high zT 0.8 at 1000
K. In addition, a rich electronic phase diagram has been found as the Na content
is varied [40, 42, 49]. In particularly, the system changes its behavior from metal-
lic to insulating [42, 43]. Strong enhancement of the Seebeck coefficient has been
reported for Na0.88 CoO2 at T = 80 K [44], which is interesting as the CoO6 octa-
hedron is the basic building block also for other cobaltates (such as Ca3 Co4 O9 ).
The peak value of the Seebeck coefficient and the powerfactor are found to be
200 VK1 and 1.8103 K1 , respectively, which is among the highest values for
hole-type materials below 100 K. In addition, the hydrated form of this compound
Nax CoO2 yH2 O (x = 0.35, y = 1.3) is the first cobalt-based system to exhibit
superconductivity [43]. The electronic structures of hydrated and non-hydrated
Nax CoO2 have been studied by first principles calculations in Ref. [50].
In the crystal structure of Nax CoO2 , sodium ions are intercalated between
layers of edge-sharing CoO6 octahedra along the c-axis, see Fig. 1.5. The stacking
sequence of the CoO2 layers determines the number of sheets in a unit cell. Two
layers of CoO6 give rise to the 2H phase and three layers to the 3R phase, as
described in Ref. [51].
RhO2 exhibits a tetragonal rutile structure, which transforms into corundum
type -Rh2 O3 after annealing above 1000 K [52]. Further heating results in a
transformation to orthorhombic type -Rh2 O3 [53]. Rhodium oxides have been
investigated primarily for applications in catalytic chemistry and as conductive
transparent oxides [5456]. Moreover, misfit-layered RhO2 (two-dimensional CdI2 -
type crystal structure, similar to that of cobalt oxides) is a p-type TE material
[60]. Analogous compounds with Rh in place of Co are found to be good TE
materials with reduced correlation effects [5760]. In cobalt oxides, their high
Seebeck coefficient originates from CdI2 -type layers, which favor low spin state of
Rh3+ /Rh4+ with degenerate spin state as the key factors [57].
 29

(a) CoO6 (b)

octahedron

2H phase 3R phase
194-P63/mmc 166-R3 m
c 11.5 c 17

c-axis

CoO2

Na

Figure 1.5: Illustration of 2H and 3R structures of prototype Nax CoO2 .

The first misfit-layered rhodium based metal oxides were synthesized in poly-
crystalline form [61]. Samples of (Bi1x Pbx )1.8 Sr2 Rh1.6 Oy (x = 0, 0.1, 0.2) and
Bi1.8 Ba2 Rh1.9 Oy were investigated and show promising TE characteristics (S =
125 V/K and S 2 = 3.3 W/mK2 ) at room temperature. Rh and Co mixed
solutions with hexagonal (Co, Rh)O2 layers alter the TE properties by decreasing
[62]. Although is reduced in the solid solution, drops due to different ion sizes
in the (Co/Rh)O2 layers, which results in distortions so that the TE performance
does not improve.
Recently, Shibasaki et al. have shown that the substitution of Rh ions in
La0.8 Sr0.2 Co1x Rhx O3 reduces the magnetic moment of Co, where the ther-
mopower is enhanced by a factor of 10 at x = 1/2 as compared to x = 0 and
1 [63]. The electronic structure, optical and TE properties of K0.49 RhO2 is inves-
 30
tigated by Okazaki et al. [64, 65] and found qualitative similarities in the optical
conductivity spectra as compared to Nax CoO2 . An experimental Seebeck coef-
ficient of 40 VK1 (300 K) is reported [64]. The temperature dependence of
transport properties is different from those of Nax CoO2 , which also suggests that
the correlation in the Rh oxides is weaker than in Nax CoO2 . An enhancement
of the transport properties for an increasing concentration of alkali cations has
been reported for other layered oxides as well [66]. Various (Co, Rh, and La) 4d
oxide systems have been investigated experimentally. Yet, their electronic struc-
tures are poorly understood, including the effect of the pudding mold bands
on the transport [67, 68]. Experiments could not verify any magnetic ordering in
Nax RhO2 [69].
In general, the Rh 4d orbitals are spatially more dispersed than the Co 3d
orbitals, which may result in an increased mobility of the charge carriers. Varela
et al. [68] have synthesized Nax RhO2 as solid solutions composed of nanometric
particles of the -NaFeO2 structure type (space group R3m, no. 166) for x = 0.70
and 0.85. The 3R structure of Nax RhO2 contains three layers of RhO6 octahedra
per unit cell. Using the hexagonal setting of the R3m space group, the c-axis of
the unit cell has a length of typically 18 A, while the length of the a-axis strongly
depends on the specific transition metal cation.
It is important to notice that the layered oxides LiRhO2 , NaRhO2 , and KRhO2
can form hydrated phase, i.e., water intercalated phases with an elongated c-
axis [67]. For Nax RhO2 it has been shown that the water intercalation and the
elongated unit cell result in a decrease of the resistivity, while there is no data
available for water intercalation in LiRhO2 [70].

1.5 Bismuth Based Binary Thermoelectrics

Bi2 Te3 and Bi2 Se3 compounds are small bandgap semiconductors belonging to the
three dimensional topological insulators family and are intensively investigated
 31
recently. Physical properties of Bi2 Te3 , Bi2 Se3 , and their alloys (e.g., Bi2 Te2 Se)
are determined to tune the bulk electrical conductivity [7175]. These crystals
tend to contain high concentrations of point defects, bismuth antisites (acceptor)
in Bi2 Te3 and selenium vacancies (donor) in Bi2 Se3 .
Bi2 Te3 , a well-known TE material, has zT 1 in the bulk and 1.5 if nanos-
tructured [76, 77]. However, a good performance is limited to temperatures below
400 K. Also, Te is a trace element, which is a problem for applications on a large
scale. It is necessary to find a substitute with comparable TE performance above
400 K. Bi2 Se3 (Eg = 0.3 eV) has two advantages over Bi2 Te3 (Eg = 0.16 eV): a
larger bandgap and energy-degenerate valence band maxima, resulting in a large
Seebeck coefficient and small bipolar conduction.

Quintuple
layer

Se1
Bi1
Se2
Bi1
Se1

Figure 1.6: Crystal structure of Bi2 Se3 along the c-axis.

Bi2 Se3 has a layered tetradymite-type rhombohedral structure (R3m, no. 167),
often described in a hexagonal representation with 15 atoms per unit cell with
lattice parameters a = 4.143 A and c = 28.636 A [78]. The crystal structure of
Bi2 Se3 consists of five-layer units of Bi and Se hexagonal planes, stacked along
the c-axis with atomic order Se1-Bi1-Se2-Bi1-Se1, forming a rhombohedral lattice
 32
in which the unit cell is composed of three quintuple layers, see Fig. 1.6. These
quintuple layers are stacked on top of each other and coupled by weak van der
Waals forces. For enhancing the electrical conductivity and reducing the thermal
conductivity of bulk Bi2 Se3 , introduction of point defects and doping by Sb, Cu,
and Ca have been put forward [7983]. On the other hand, only few studies have
dealt with the effect of doping by IIIB elements, e.g., Ga [84] and In [85], on the
thermal transport. Recently, TE figure of merit in polycrystalline In and Ga-
doped bulk Bi0.4 Sb1.6 Te3 compounds reduces the thermal conductivity and result
in zT of 1.2 [86]. Luo and coworkers have emphasized the importance of the van
der Waals interaction for the TE properties of bulk Bi2 Se3 and Bi2 Te3 with and
without strain [87]. Recent theoretical work predicted that zT values of bulk
Bi2 Se3 over 1 at 300 K and 1.5 at 500 K may be achieved by substantial reduction
of the lattice thermal conductivity by 70% by alloying or nanostructuring [88].
The recent reports show that the mechanical exfoliation of single quintuple
layer and few quintuple layers of Bi2 Se3 is possible and proposed that a high
TE properties can be achieve in few atomic layer thick film [8993]. In a recent
work of Sun et al. [90] show that TE properties of single quintuple layer of Bi2 Se3
can be enhance by reducing as compared to the bulk value. As the number
of quintuple layers increases the bandgap of Bi2 Se3 decreases and close at the 8
quintuple layers [91]. This variation of bandgap with the thickness can play a vital
role in order to tune its TE properties.
Tl-doping in topological insulators has been found to be a promising approach
to enhance the TE properties. The impact in the case of Bi2 Se3 has been addressed
in Ref. [94] and a linear increase of the powerfactor with the Tl concentration has
been reported. Recently, the TE properties of Tl-doped Bi2 Te3 with Tl concen-
trations of 0 x 0.3 have been studied, showing that an increasing Tl-doping
changes the electrical conduction from p-type to n-type [95].
 33
1.6 The pudding mold Bands in PtSb2

Kuroki et al. [96] have shown that a metal with pudding mold bands (a disper-
sive portion and a flat portion), results in a large Seebeck coefficient. A pudding
mold bands were discovered in the TE oxide Nax CoO2 by angle-resolved photoe-
mission spectroscopy [9799]. Band calculations for PtSb2 have show a pudding
mold valence band [100] and a high S = 250 VK1 at approximately 120 K [101],
suggesting good TE properties. PtSb2 has a melting temperature of 1500 K, which
make it a candidate for a high temperature TE material [102]. The filled Pt4+ t2g
orbital (5d6 low-spin state) is similar to the Co3+ orbital (3d6 low-spin state) of
NaCoO2 , sharing the semiconducting nature.
Bulk PtSb2 can be both metallic and semiconducting for deviations from the
ideal stoichiometry. It can be n- and p-type doped with charge carrier concentra-
tions spanning several orders of magnitude (1016 cm3 to 1020 cm3 ), as observed
for different synthesis methods and parameters [103108]. Recent studies show
that p-type PtSb2 (by Ir doping on the Pt site) gives a high metallic conductiv-
ity and large Seebeck coefficient [101] and that n-type PtSb2 (by Sb deficiency)
gives a large Seebeck coefficient and low thermal conductivity up to room tem-
perature [109]. PtSb2 has a high mobility due to the small difference in the
electronegativities of Pt and Sb (2.28 and 2.05, respectively) [110].

1.7 Objective and Scope of the Thesis

The main objective of this thesis is to address effects on the electronic band struc-
ture by introducing vacancies, hydration, strain, and doping to develop an under-
standing of the TE properties of materials at the quantum mechanics level. We
aim to understand S, , and of Kx RhO2 in the 2H phase, Nax RhO2 in the 3R
phase, doped PtSb2 , and the effects of replacing Te with Se, nanostructuring of
Bi2 Se3 , and the properties of bulk Tl-doped Bi2 Se3 .
 34
The thesis is organized as follows: Chapter II presents the method combining
density functional theory and Boltzmann transport theory. In chapter III the
structural, electronic, optical, and transport properties of Kx RhO2 in the 2H phase
(pristine and hydrated) at different cation concentrations are investigated. The
effect of the 3R phase of Nax RhO2 on transport properties is presented in chapter
IV. In addition, in chapter V, we will discuss the electronic, phononic, and the
transport properties of nanostructured Bi2 Se3 for varying thicknesses and strain
for a single quintuple layer. In chapter VI bulk Bi2 Se3 is doped with Tl and
strain is applied to enhance the figure of merit and Rashba spin splitting by
tuning the bandgap. In chapter VII we study electron and hole doping in PtSb2
using the virtual crystal approximation. The transport coefficients along with the
powerfactor and figure of merit are addressed for various dopings from 300 K to
800 K. The conclusions are given in chapter VIII.
 35

Chapter II

Method of Calculations

2.1 Density Functional Theory

Density functional theory has been developed by Hohenberg and Kohn in 1964
and further modify one year later by Kohn and Sham. Therefore, the description
of the method and an explanation of its features are necessary for the correct
interpretation of the results obtained from the band structure calculations.

2.1.1 Hamiltonian

To solve the Schrodinger equation would require us to solve a system of around

1023 simultaneous differential equations. The Hamiltonian for this many body
system can be described as,

ri ) + TN (R~ ) + Vee (~
Htot = Te (~ ri , r~j ) + VN N (R~ , R~ ) + VeN (~
ri , R~ ) (II.1)

where Te and TN are kinetic energy operators for the electrons and nuclei, re-
spectively. The last three terms describe the electron-electron, nuclei-nuclei, and
electron-nuclei electrostatic coulomb interactions, respectively.
In the light of the Born-Oppenheimer approximation [111] approximation the
nuclei are much heavier, and therefore much slower, than the electrons. As a
result, nuclei can be considered as effectively frozen at fixed positions, while only
 36
electrons being considered as mobile. As a consequence, the kinetic energy term
of the nuclei is zero. The nucei-nuclei potential term also reduces to a constant.
Therefore, the above Hamiltonian equation become simplified as:

Htot = Te + Vee + Vext , (II.2)

where the electron-nuclear interaction represented by Vext . We can further explain

the remaining terms as

1X~ 2 1X 1 X Z
Te = i , Vee = , and Vext =
2 i 2 i6=j |~
ri r~j | ri R~ |
i,j |~

Here, atomic units are used, i.e., m = e = h = 1, ~ri and ~rj are the coordinates
of electrons i and j, respectively, and Z is the charge of the nucleus at position
~ . The Hamiltonian obtained after applying the Born-Oppenheimer approxima-
R
tion is simpler, but still far too difficult to solve. Here the many-body problem
is transferred into a single particle problem by approximating the electronic wave
function by a Slater-determinant of single particle wave functions, which ensures
the antisymmetry of the wave function. In the Hartree-Fock approach the ex-
change between electrons as well as the correlated motion of electrons of like spin
due to the Pauli principle is taken into account. An alternative approach is given
by density functional theory, where the central quantity is not the wave function
but the electron density (~r).

2.1.2 Hohenberg-Kohn Theorems

The main idea of density functional theory is provided by the Hohenberg-Kohn

theorems [112].
Theorem 1: The non-degenerate ground state electron density 0 (~r) deter-
mines the external potential Vext .
Theorem 2: The total energy functional EVext [0 (~r)] reaches its minimal value
(equal to the ground state total energy) at the ground state electron density 0 (~r)
 37
corresponding to Vext ,
EVext [(~r)] EVext [0 (~r)] (II.3)

for every trial electron density (~r). The total energy functional EVext [(~r)] can
be written as
Z
EVext [(~r)] = h|Te + Vee |i + h|Vext |i = FHK [(~r)] + (~r)Vext d~r, (II.4)

where FHK [(~r)] is universal for any many-electron system. Within the scope
of this thesis, we will explain the one-to-one correspondence (~r) Vext , uni-
versality of FHK [(~r)] and the value of the ground state total energy functional
EVext [(~r)].
First, an external potential gives a unique ground-state density. The first the-
orem of Hohenberg and Kohn demonstrates that the density contains as much
information as the wave function. Thus, all observables can be written as func-
tional of the density, i.e., all physical quantities can be recovered from the density
only.
Second, the universality of FHK [(~r)] should be understood. From equation
II.4 one can see that FHK [(~r)] does not contain any information on the nuclei
and their positions. Therefore, it is a universal functional for any many-electron
system. Third, the ground state density corresponding to the external potential
Vext can be obtained by using the second theorem. Indeed, this is possible only
when an appropriate expression is known for FHK [(~r)].

2.1.3 The Kohn-Sham Equations

The total energy functional is given by [113],

E[(~r)] = T + V. (II.5)

Here T and V are the kinetic and electron-electron potential energy functional for
the exact solution. By subtracting equation II.6 from equation II.5, we obtain the
correlation functional as
 38
Vc = T T0 . (II.6)

Thus, the Hartree functional is

EH [(~r)] = T0 + VH . (II.7)

where T0 is for the non-interacting electron gas and VH +Vx = V , VH and Vx are the
Hartree and the exchange contribution, respectively. Form the above information,
we can reformulate the Hohenberg-Kohn functional as

FHK [(~r)] = T0 + V + T T0 = T0 + V + Vc , (II.8)

where Vc is the correlation functional. We write

FHK [(~r)] = T0 + VH + Vc + Vx = T0 + VH + Vxc , (II.9)

where Vxc = Vc + Vx is the exchange-correlation functional. If, we assume that we

know Vxc , then we can write the energy functional as

EVext [(~r)] = T0 + VH + Vxc + Vext . (II.10)

The corresponding Hamiltonian is called the Kohn-Sham Hamiltonian

HKS = T0 + VH + Vxc + Vext . (II.11)

The exchange-correlation potential is given by

Exc [(~r)]
Vxc = . (II.12)
(~r)

As X
(~r) = i i , (II.13)
i=1
the Kohn-Sham equations can be written as

HKS i = i i , (II.14)

where i and i are the single particle wave function and energy, respectively. The
Kohn-Sham method is an exact formulation to describe the ground state properties
 39
of many-electron systems, but the exchange-correlation functional is unknown and
demands further approximation and analysis.

2.2 Computational Modules

2.2.1 WIEN2k Code

The structural, electronic, optical and thermoelectric calculations of Kx RhO2 ,

Nax RhO2 , and PtSb2 in chapters 3, 4, and 7 are carried out using the full po-
tential WIEN2k code [114116]. This package allows us to study the electron
charge density, density of states, magnetism, band structure, and Fermi surface.
We can apply different exchange-correlation potentials including the local density
approximation, various generalized gradient approximation (GGA) [117119], Wu-
Cohen [120], meta-GGA [121, 122] and the LDA+U method [123]. most recently
implemented modified Becke-Johnson potential (mBJ) [124] and hybrid function-
als [125]. The parameter Rmt Kmax controls the size of the plane-wave basis set
(number of plane waves and is defined as the product of the plane-wave cut-off
(Kmax ) and the smallest atomic-sphere radius (Rmt ). We have used an appro-
priate sampling of the full Brillouin zone for electronic structure calculations and
geometry optimization in all cases. The parameters of our calculations depend on
the type of calculation but for any of them we tested convergence with respect to
the ~k-mesh and to Rmt Kmax . The parameters of our calculations are always fully
converged to the required precision.

2.2.2 Quantum-ESPRESSO Code

In chapters 5 and 6 we study nanostructured Bi2 Se3 and Tl-doped Bi2 Se3 , which
have strong van der Waals interactions between adjacent quintuple layers [126]. As
in WIEN2k the van der Waals forces are not implemented, we use the Quantum-
ESPRESSO code [127], which includes the DFT-D and London methods (semi-
 40
empirical dispersion correction). Quantum-ESPRESSO is also based on density
functional theory, using pseudopotentials. We address the strain effect on the
structural relaxation, to study the electronic, phononic, and structural properties
of Bi2 Se3 and to tune the thermoelectric properties. In all calculations relativistic
pseudopotentials [128] are used to incorporate the spin-orbit coupling. The struc-
tures are optimized until an energy convergence of 105 eV and a force convergence
of 0.002 eV/A are reached.

2.3 Modeling of Thermoelectric Materials

2.3.1 Transport Coefficients

In the previous chapter in section 1.2.1 (equation I.1) we have described the fig-
ure of merit, which depends on the transport coefficients of the material. These
coefficients need to be tuned in the search for improved thermoelectric materials.
Now we will explain how these coefficients have been obtained from the results of
our first-principles band structures calculations.
The perturbation of an electric field or a temperature gradient induces electrical
~ or thermal currents (J~Q ) in a material. The relations between the electric field
(E)
~ and the temperature gradient (T ) with the corresponding J~ and J~Q for solid
(E)
can be expressed as [129]:
J~ = E
~ ST (II.15)

J~Q = ST E
~ T (II.16)

where is the electrical conductivity, S the Seebeck coefficient defined as the

voltage gradient produced in a sample by a given temperature gradient when
the electrical current is zero (S = V /T ) and is the thermal conductivity
whose electronic contribution is defined as the heat current produced per unit
temperature gradient when the electrical current is zero.
 41
2.3.2 Calculation of the Transport Coefficients using Boltz-

TraP

To evaluate the transport coefficients in defined above, we need a microscopic

model of the transport process. We will use the semi-classical approach given
by the solution of Boltzmann transport equation in the constant relaxation time
approximation. In general, the electric current of carriers is defined as:

X
J~ = e f~k~v~k (II.17)
~k

where e is the charge of the carriers. f~k is the population of the quantum state
labeled with ~k and ~v~k the group velocity associated to that state defined as

1 ~k
~v~k = . (II.18)
} ~k

f~k is the distribution function which measures the number of carriers in the state
~k in the neighborhood of ~r. The distribution function can change through the

following mechanisms: diffusion, external fields (such as electric and magnetic

fields or a temperature gradient) and scattering of the carriers by phonons or
localized scattering centers, such as impurities or crystal defects. The population
of the state ~k is the solution of Boltzmanns equation that contains the interplay
between all the mechanisms described above:

   
f~k e 1
~ + ~v~ H
~ ~ f~ = f~k
+ ~v~k ~r f~k + E (II.19)
t } c k k k
t scattering

In the absence of fields, the stationary solution of the Boltzmanns transport equa-
tion for fermions is the Fermi-Dirac distribution function f0 (~k). Linearizing the
above equation, using the relaxation time approximation for the scattering term,
and in the absence of magnetic fields and temperature gradients, the population
 42
is !
f0 (~k)
f~k = f0 (~k) + e ~
~k~v~k E (II.20)
~k

The equilibrium distribution of electrons is the Fermi-Dirac distribution

1
f0 (~k) =  , (II.21)
~k

exp +1
kB T

where is the chemical potential that depends strongly on the carrier concentra-
tion and weakly on the temperature.
In the BoltzTraP code ~k will be kept constant in the so-called constant scat-
tering time approximation [130]. The code relies on a Fourier expansion of the
band energies obtained from the WIEN2k calculation which enables the calcula-
tion of the velocities as Fourier sums that can be efficiently evaluated using fast
Fourier transformation. Using the above relation for f~k to calculate the current,
the conductivity tensor can be expressed as

!
X f0 (~k)
= e2 ~k~v~k~v~k . (II.22)
~k
~k

Similar expressions can be derived for the Seebeck coefficient and electronic ther-
mal conductivity. For simplicity, the transport distribution can be defined as

X
= ~k~v~k~v~k . (II.23)
~k

The transport coefficients involved in the figure of merit can directly be calculated
once we have the transport distribution. We have

!
f0 (~k)
Z
= e2 (~k ) d~k (II.24)
~k

!
f0 (~k)
Z
e
S= (~k ) (~k )d~k (II.25)
T ~k
 43

!
f0 (~k)
Z
1
= (~k ) (~k )2 d~k . (II.26)
T ~k

Under the constant relaxation time approximation, the Seebeck coefficient and
dimensionless thermoelectric figure of merit are independent of ~k . This is the
reason why these quantities will be our main focus when analyzing the transport.
For this type of calculations, a very dense ~k-point sampling is required, well be-
yond what is needed for a self-consistent calculation. Boltzmann transport theory
has proven its validity in numerous cases where the transport coefficients can be
compared with experimental results [129, 130].
 44

Chapter III

KxRhO2 in the 2H Phase

In this chapter the structural, electronic, optical and thermoelectric properties of

Kx RhO2 in the 2H phase will be discussed. We study the interconnection between
the enhancement of the thermoelectricity and the cation concentration for Kx RhO2
at x = 1/2 to compare with recent experimental findings for K0.49 RhO2 [64, 65]
and at x = 7/8 to compare with Na0.88 CoO2 [44]. Experimental data of the optical
and transport properties are taken from Refs. [64, 65]. A comparison to Nax CoO2
is given in terms of the chemical nature of the Co 3d and Rh 4d orbitals, where
the optical transitions are explained by the electronic band structure and density
of states. The role of a high cation concentration for the distortions of the RhO6
octahedra due to water intercalation (and thus for the figure of merit) is also
discussed.

3.1 Computational Parameters

We use 66 ~k-points in the irreducible wedge of the Brillouin zone for calculating the
electronic structure and a dense mesh of 480 ~k-points in the optical calculations.
For the transport calculations we use 4592 ~k-points. Self-consistency is assumed
to be achieved for a total energy convergence of 105 Ryd. Various exchange
and correlation functionals (LDA, GGA, GGA-sol, and GGA-PBE0) are used for
 45
x=7/8

Rh

x=1/2

Figure 3.1: Crystal structures of Kx RhO2 for x = 1/2 and 7/8.

optimizing the c/a ratio. Since the differences are small, we will discuss in the
following only the GGA-sol results.

3.2 Structural Properties

Kx RhO2 crystallizes in the -Nax CoO2 structure with space group P63 /mmc and
experimental lattice constants of a = 3.06 A and c = 13.60 A [131], as shown in
Fig. 3.1. The CdI2 -type RhO2 layer and the K layer are stacked alternately along
the c-axis. The experimental lattice constants of K1/2 RhO2 are used as a starting
point of the optimization, obtaining the volume and c/a ratio presented in Fig.
3.2. Our LDA calculation yields a 14% smaller c/a ratio (4.44 experimental, 3.84
calculated), whereas our c/a ratio is close to other layered Co/Rh oxides (about
3.8). The bonding length between Rh and O is reduced from 2.13 A to 2.04 A. To
 46
confirm the large deviation of the c/a ratio from the experimental value, we have
optimized the structure by more involved exchange correlation functionals (GGA,
GGA-sol and PBE0), obtaining lattice constants of a = 3.06 A and c = 11.56 A
with c/a = 3.77 for GGA-sol, see Fig. 3.2 (b).
1 2.4
(a) LDA (b) K1/2RhO2
GGA
GGA-sol 2
0.8 GGA-PBE0

c/a = 3.77 (GGA-sol)

1.6
E-Emin (eV)

0.6
1.2
0.4
0.8

0.2
0.4

0 0
572 616 660 704 748 -36 -24 -12 0 12
3
Volume (aB ) Deviation from exp. c/a ratio (%)

Figure 3.2: Volume optimization of K1/2 RhO2 for different exchange correlation
functionals.

However, we obtain almost the same results for all exchange-correlation func-
tional. In addition, our optimized lattice parameters are confirmed by calculations
of the optical and transport properties, see the discussion below. The calculated
Seebeck coefficient is overestimated for the experimental lattice constants, while
the optimized lattice constants yield a Seebeck coefficient close to the experi-
ment. The calculated optical conductivity of K1/2 RhO2 (with optimized lattice
constants) at zero photon energy is found to be 2500 1 cm1 , which is agrees
excellently with the experiment.
The optimized c value is similar to other isostructural layered oxides such
as Srx RhO2 [132, 133], Nax CoO2 [134], Lix NbO2 [135], and Lax CoO2 [136]. In
addition, LiRhO2 , NaRhO2 , and KRhO2 layered oxides can form a hydrate (water
intercalated) phase [67] with an increased c length [70]. Takada et al. [43] showed
that Nax CoO2 can be readily hydrated to form Nax CoO2 yH2 O, maintaining
the CoO2 layers, but with a considerably expanded c axis, to accommodate the
 47
intercalated water.

3.3 Electronic Properties

In Figs. 3.3 (a) and (b) the electronic band structures of K1/2 RhO2 and K7/8 RhO2
are presented. They are similar to isostructural and isovalent Na1/2 CoO2 , except
for a slightly larger pseudogap between the Rh t2g and eg bands. The less dis-
persive bands (near the Fermi level give a higher carrier density) in Kx RhO2 as
compared to Nax CoO2 promote an enhanced thermoelectricity. The calculated
density of states (Fig. 3.4) reflects a crystal field splitting into eg and t2g states
experienced by the Rh4+ ions that is similar to Nax CoO2 , but with a larger band-
width (more dispersive bands) [65]. The increased bandwidths reflect a reduction
of the electronic correlations. In addition, a weak hybridization between the Rh
4d and O 2p states is seen in density of states at or below the Fermi energy, and
the O 2p states lie deeper in the valence band (below 2 eV) than in Nax CoO2 .
Hence, this reflect more localized Rh d states.

6 6
(a) (b)
Rh d eg

4 4

2 2
E-EF (eV)

0 EF
Rh d t2g

-2 -2
Rh d & O p O p

-4 -4

-6 -6

-8 -8
M K M K

Figure 3.3: Band structures of K1/2 RhO2 and K7/8 RhO2 .

The band structure of the hydrated phase of Kx RhO2 , see Fig. 3.5, is close to
 48
25
K1/2RhO2
20 K7/8RhO2

DOS (States/eV)
EF
15

10

0
-8 -6 -4 -2 0 2 4 6
E-EF (eV)

Figure 3.4: Density of states obtained for K1/2 RhO2 and K7/8 RhO2 .

that of optimized Kx RhO2 , compare Fig. 3.3 and Fig. 3.5, except that the Rh t2g
bands are flatter in a narrow region around the Fermi energy. This fact causes
the resistivity to increase and results in a higher Seebeck coefficient (Seebeck
coefficient and electrical conductivity are inversely proportional to each other, see
section 1.1.2). According to the band structure given in Fig. 3.6, the larger c lattice
parameter of Nax CoO2 (15.59 A as compared to 13.6 A, due to water molecules
in the system causes the Co eg states to shift down in energy to the Fermi level
because the Co-O overlap is reduced. This, in turn, results in a decrease of the
resistivity and, therefore, the Seebeck coefficient.

3.4 Optical Properties

The optical reflectivity and conductivity of Kx RhO2 (x = 1/2 and 7/8) are shown
in Figs. 3.7(a) and 3.7(b), where the experimental values (blue solid line in Fig.
3.7) are taken from Okazaki et al. [65]. The calculated reflectivities are similar to
each other with a maximum value of approximately 90% near zero photon energy.
A Drude-like reflectivity edge is found around 1 eV (experiment: 1.2 eV) in the
reflectivity spectrum of K1/2 RhO2 , while for K7/8 RhO2 this edge appears at 0.5 eV,
 49
6 6
(a)

Rh d eg
4 (a) 4 (b)

2 2
E-EF (eV)

0 EF

Rh d t2g
0

-2 -2

Rh d & O p O p
-4 -4

-6 -6

-8 -8
M K M K

Figure 3.5: Band structures of hydrated K1/2 RhO2 and K7/8 RhO2 .

showing that the bands are flatter in K7/8 RhO2 than in K1/2 RhO2 . At zero photon
energy, the calculated optical conductivity of 2500 1 cm1 for K1/2 RhO2
is obtained in excellent agreement with the experiment (blue and black dots in
Fig. 3.7 (b)). Three well defined peaks are observed: (i) near 1 eV due to the
intra-band transition of Rh t2g -t2g , (ii) at 3 eV due to the inter-band transition
of Rh t2g -eg , and (iii) around 5.5 eV due to the inter-band transition from the O
2p to the Rh 4d eg states. The peaks are also present in the experiment [65], as
well as for Na1/2 CoO2 (0.5 eV, 1.6 eV, and 3 eV) [137], which again reflects the
similarity of these isostructural and isovalent compounds.

3.5 Transport Coefficients and Figure of Merit

In the following, we will address both the experimental (hydrated) and optimized
structure of Kx RhO2 . We have calculated the Seebeck coefficient, thermal con-
ductivity (electronic part), and figure of merit using Boltzmann theory in the
constant relaxation time approximation. The results are plotted in Figs. 3.8 (a),
(b), and (c) as a function of the temperature from 0 to 700 K, and compared with
 50
6

E-EF (eV)
0 EF

-2

-4

-6

-8
M K
Figure 3.6: Band structure of Na1/2 CoO2 with c = 15.59 A.

the experimental figure of merit of Na0.88 CoO2 [44]. According to Fig. 3.8 (a),
at 300 K the Seebeck coefficients of optimized K1/2 RhO2 and K7/8 RhO2 amount
to 50 and 55 VK1 , respectively, in agreement with the experiment [64]. For
hydrated K1/2 RhO2 and K7/8 RhO2 , on the other hand, the values are strongly en-
hanced to 100 and 140 VK1 , respectively. The larger value for K7/8 RhO2 than
for K1/2 RhO2 is due to band filling by increase of cation concentration, which
(rigidly) shifts flatter regions of the bands towards the Fermi energy. For com-
parison, we have calculated the room temperature S value of Na1/2 CoO2 for the
lattice parameters, case (i) a = 2.88 A, c = 15.59 A [51] and case (ii) a = 2.86
A, c = 10.82 A. We obtain 60 and 34 VK1 , respectively, which is in agreement
with the experiment [138] and seconds the above picture that a larger c parameter
can enhance S. We finally have to address the question why hydrated Kx RhO2
has a higher Seebeck coefficient than hydrated Nax CoO2 . According to the band
structure given in Fig. 3.6, the larger c lattice parameter of Nax CoO2 (15.59 A as
compared to 13.6 A causes the Co eg states to shift down in energy to the Fermi
level because the CoO overlap is reduced. This, in turn, results in a decrease
of the resistivity and, therefore, the Seebeck coefficient. Note that the carrier
 51
1
(a) K1/2RhO2
0.8

Reflectivity R()
K0.49RhO2 (exp.)
K7/8RhO2
0.6

0.4

0.2

0
Conductivity ( cm )
-1

8000
(b)
-1

6000

4000

2000
0
0 1 2 3 4 5 6 7 8
Photon Energy (eV)

Figure 3.7: Optical reflectivity and conductivity of K1/2 RhO2 and K7/8 RhO2 along
with experimental data.

concentrations are 1.01021 cm3 in hydrated Na1/2 CoO2 and 1.81021 cm3 in
hydrated K1/2 RhO2 .
For optimized Kx RhO2 the Seebeck coefficient depends not much on the K
concentration up to 300 K, whereas for higher temperatures it increases stronger
for K7/8 RhO2 to reach a value of 80 VK1 at 700 K. In contrast to this behavior,
S remains almost constant above 300 K for hydrated Kx RhO2 . According to
Fig. 3.8 (b) the thermal conductivity is similar for the hydrated phases and for
optimized K7/8 RhO2 , while it is much enhanced for optimized K1/2 RhO2 . In Fig.
3.8 (c), the figure of merit (z) is presented for optimized and hydrated Kx RhO2 .
For hydrated K7/8 RhO2 it reaches a value of z = 3103 K1 at 100 K. This
is much higher than in other hole-type compounds, including Na0.88 CoO2 with a
maximum of z = 1.8103 K1 (at 80 K). At room temperature (300 K) the z of
hydrated Kx RhO2 also clearly exceeds that of Na0.88 CoO2 , whereas for optimized
Kx RhO2 it is only a little higher. The large z of hydrated Kx RhO2 results from
 52
160
120
(a)

S (V/K)
80
40
0
/ (10 W/mKs)

14 K1/2RhO2 (b)
K7/8RhO2
9
14

0
3.0 (c)
z (10 K )

2.5
-1

2.0
-3

1.5
1.0
0.5
0.0
0 100 200 300 400 500 600 700
Temperature (K)

Figure 3.8: Thermoelectric properties of pristine (solid line) and hydrated (dashed
line) Kx RhO2 .

a decrease of the thermal conductivity, increase of the electrical conductivity (as

demonstrated for the hydrated phase of NaRhO2 in Fig. 2 of Park et al. [70]), and
larger Seebeck coefficient as compared to optimized Kx RhO2 .
 53

Chapter IV

NaxRhO2 in the 3R Phase

In this chapter, we study non-hydrated (c = 15.54 A) and hydrated (c = 20.85

A) Nax RhO2 for x = 1, 0.83, and 0.75 in the 3R phase. We analyze the effects
of hydration on the transport properties in the temperature range from 300 K
to 800 K and demonstrate great potential of 3R-Nax RhO2 class of materials as
thermoelectrics.

4.1 Computational Parameters

The exchange-correlation potential is parametrized in the generalized gradient

approximation. We fully optimize the structure of Nax RhO2 for x = 1, 0.83, and
0.75, considering 2 2 1 supercells, until the atomic forces reach values of less
than 10 mRyd/aB . For the wave function expansion inside the atomic spheres a
maximal angular momentum of `max = 12 is employed and a plane-wave cutoff of
Rmt Kmax = 7 with Gmax = 24 is used. Self-consistency is assumed when the total
energy variation is less than 104 Ry. We note that it has been demonstrated for
layered oxides isostructural to the present compound that at elevated temperatures
the lattice part of the thermal conductivity can be neglected as compared to the
electronic part [139]. We use a ~k-mesh of 7 7 2 points for the band structure
and density of states (DOS) and a dense ~k-mesh of 18 18 18 points for the
 54
thermoelectric calculations using BoltzTraP code.

4.2 Structural Properties

We start in each case from the experimental lattice parameters a = 3.09 A and c =
15.54 A of NaRhO2 [68] and construct a 2 2 1 supercell. Afterwards we remove
two Na atoms from different sites to obtain Na0.83 RhO2 . These configurations
are named A1, A2, and A3. Similarly, we remove three Na atoms to obtain
Na0.75 RhO2 and name these configurations B1, B2, and B3. All the structures
under investigation are shown in Fig. 4.1.
The formation energy is calculated for each configuration by the relation

EF ormation = EDef ect EP ristine + 12(1 x)EN a . (IV.1)

Here EP ristine is the total energy of a supercell of pristine NaRhO2 . In addition,

EDef ect and EN a are the total energies of the defective supercell and of the isolated
Na atom, respectively. The energy of an isolated atom is calculated by placing
the atom in a cubic cell of 20 A side length.
The formation energies of configurations A1, A2, and A3 amount to 7.83 eV,
6.73 eV, and 6.65 eV, respectively, and those of configurations B1, B2, and B3
to 11.54 eV, 10.34 eV, and 9.90 eV. These results shows that for Na0.83 RhO2 and
Na0.75 RhO2 , respectively, the A3 and B3 configurations has the lowest energy, see
Table 4.1). The latter structures are characterized by maximal distances between
the defects, which indicates that Na vacancies do not cluster in Nax RhO2 , at
least not in the two systems under investigation. Moreover, a minimum energy
of 6.65 eV is required for creating two Na vacancies in NaRhO2 and an energy
of 9.90 eV for creating three vacancies. In the following, we will discuss only
the lowest energy structures A3 and B3 unless otherwise mentioned. In pristine
NaRhO2 the RhO bond length amounts to 2.07 A and the thickness of the RhO6
 55
layer is 2.11 A, which is larger than the thickness of the CoO6 layer (1.82 A) in
NaCoO2 , reflecting less hybridization between Rh and O. Moreover, the bond
angle RhORh = 96.5 is smaller than the bond angle CoOCo = 98.35 .
Next to vacancies the RhO6 octahedra shrink slightly (the RhO bond lengths
decrease by 0.020.05 A), because the interlayer coupling through the Na atoms
is reduced and the RhO6 octahedra thus become a bit more compact, which also
influences the electronic and thermoelectric properties of the material.
NaRhO 2 Na 0.83 RhO 2
A1 A2 A3

Na 0.75 RhO 2
B1 B2 B3
Na

Na vacancy

Rh

Figure 4.1: Atomic structures of NaRhO2 , Na0.83 RhO2 , and Na0.75 RhO2 for dif-
ferent vacancy configurations.
 Table 4.1: Structural properties of NaRhO2 , Na0.83 RhO2 , and Na0.75 RhO2 for different vacancy configurations.
NaRhO2 Na0.83 RhO2 Na0.75 RhO2
A1 A2 A3 B1 B2 B3
EF ormation (eV) 7.38 6.73 6.65 11.54 10.34 9.90
OOk (A) (top layer) 2.76 2.63 2.66 2.66 2.62 2.67 2.62
OOk (A) (middle layer) 2.76 2.63 2.64 2.62 2.62 2.59 2.62
56

OOk (A) (bottom layer) 2.76 2.96 2.66 2.68 2.68 2.64 2.62
OO (A) 3.09 3.09 3.09 3.09 3.09 3.09 3.09
RhORh (top layer) 96.5 99.3 99.3 98.5 100.7 98.5 98.2
RhORh (middle layer) 96.5 98.1 99.1 99.3 100.7 99.3 99.4
RhORh (bottom layer) 96.5 98.0 97.8 98.1 98.0 98.0 99.5
RhRh (A) 3.09 3.09 3.09 3.09 3.09 3.09 3.09
 57
2
Rh d
Rh d3z2-r2

E-EF (eV)
1 NaRhO2 Rh dx2-y2, dxy
Rh dxz, dyz
0

-1
2
E-EF (eV)

1 Na0.83RhO2

-1
2
E-EF (eV)

1 Na0.75RhO2

-1
K M 0 2 4 6
DOS (states/eV)

Figure 4.2: Band structures and Rh 4d density of states of NaRhO2 , Na0.83 RhO2 ,
and Na0.75 RhO2 .

4.3 Electronic Properties

We determine the Na vacancy effects on the electronic band structure and corre-
sponding density of states (DOS), for NaRhO2 , Na0.83 RhO2 , and Na0.75 RhO2 in
Fig. 4.2. NaRhO2 has an indirect bandgap of 1.33 eV along the M high sym-
metry direction. The valence band mainly is formed by the Rh 4d t2g states and
shows a total width of a bit less than 2 eV. This value is comparable to the band-
width of NaCoO2 , even though 4d orbitals are more extended than 3d orbitals and
thus usually exhibit higher bandwidths. The conduction band minimum belongs
almost purely to the eg states. On the other hand, the valence band maximum,
which is particularly important as it determines the consequences of hole doping,
 58
is due to the d3z2 r2 orbital. The electronic band structures obtained for configu-
rations A3 and B3 are similar to the pristine case, except for the introduction of
holes, which densities 8.81020 cm3 (A3) and 8.71020 cm3 (B3). Besides the
characteristic peak approximately 0.2 eV below the Fermi energy, the density of
states of pristine NaRhO2 shows d3z2 r2 states down to 0.75 eV and between 1
eV and 1.75 eV, which is similar for Na0.83 RhO2 and Na0.75 RhO2 except for the
states at the Fermi level, which have different effective masses and consequently
result in different Seebeck coefficients.
2
E-EF (eV)

1
Na0.83RhO2
0

-1
2
E-EF (eV)

1
Na0.75RhO2
0

-1
K M 0 2 4 6
DOS (states/eV)

Figure 4.3: Band structures and Rh 4d density of states of hydrated A3-

Na0.83 RhO2 and B3-Na0.75 RhO2 .

Moreover, the long c-axis of the hydrated phase causes the Rh 4d states to shift
down in energy towards the Fermi level, because the RhO overlap is reduced
(increased bond length of 2.2 A) as concluded in the last section. The density of
states in Fig. 4.4 highlights the contributions of Rh 4d states.
 59
6 Rh d (middle layer) Rh-dz2 6 Rh d (middle layer) Rh-dz2
Rh-d (bottom layer) Rh-dx2-y2 Rh-d (bottom layer) Rh-dx2-y2
4 Rh d (top layer) Rh-dxy
4 Rh d (top layer) Rh-dxy
2 NaRhO2 Rh-dxz 2 NaRhO2 Rh-dxz
0 Rh-dyz 0 Rh-dyz
6 6
DOS (states / eV)

DOS (states / eV)

4 4
A1-Na0.83RhO2 B1-Na0.75RhO2
2 2
0 0
6 6
4 4
2 2
A2 B2
0 0
6 6
4 4
2 2
A3 B3
0 0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
E-EF (eV) E-EF (eV)

Figure 4.4: Density of states for non-hydrated Na0.83 RhO2 and Na0.75 RhO2 , com-
pared to pristine NaRhO2 .

4.4 Seebeck Coefficient and Figure of Merit

The transport properties of hydrated and non-hydrated Na0.83 RhO2 and Na0.75 RhO2
are addressed in Fig. 4.5, using the equations derived in chapter 2, section 3.2. We
show the Seebeck coefficient and figure of merit for the temperature range from
300 K to 800 K. The Seebeck coefficient is positive in the whole range, indicating
that the carriers are holes. We obtain room temperature values of S = 63 VK1
and S = 37 VK1 for non-hydrated Na0.83 RhO2 and Na0.75 RhO2 , respectively.
For hydrated and non-hydrated Na0.83 RhO2 the Seebeck coefficient behaves rather
similarly as a function of the temperature, except for the high temperature range
where hydration reduces S. For Na0.75 RhO2 the Seebeck coefficient increases at
300 K from 37 VK1 to 51 VK1 under hydration as the density of states at the
Fermi level increases (see Fig. 4.3), where this increment of about 30% remains
essentially constant over the whole temperature range.
Moreover, according to Figs. 4.5(c) and 4.5(d), zT at 300 K amounts to 0.16
and 0.06 for the two systems. The former value is twice that found for Na0.85 CoO2
and Na0.88 CoO2 in Ref. [44]. Starting from almost the same value at 300 K,
hydration enhances zT over the whole temperature range, such that at 800 K
a value of about 0.4 is reached. At 580 K a maximal increment of 34% due to
 60
Na0.83RhO2 Na0.75RhO2
125 125
(a) (b)
100 100
S (V/K)

75 75
non-hydrated
50 hydrated 50

25 25
0.4 (c) 0.4 (d)

0.3 0.3
zT

0.2 0.2

0.1 0.1

0 0
300 400 500 600 700 800 300 400 500 600 700 800
T (K) T (K)

Figure 4.5: Transport properties of non-hydrated and hydrated Na0.83 RhO2 and
Na0.75 RhO2 as functions of the temperature.

hydration is achieved.
The enhancement of Seebeck coefficient under hydration results in higher zT
values. While the gross shape of the band structure changes little under hydration,
see Fig. 4.3, the Rh 4d bands become flatter in a narrow region around the Fermi
level. This fact causes the effective masses and therefore the resistivity to increase
and results in a higher Seebeck coefficient. In order to study a possible disorder
of the vacancies, we have also calculated the Seebeck coefficient, powerfactor,
and figure of merit (see Fig. 4.6) of higher energy configurations A1, A2, B1,
and B2. For A2 and B2 the results are close to the values of the respective
stable configurations, because both have the vacancies in different RhO6 layers. In
contrast, concentration of vacancies in the same RhO6 layer results in a significant
reduction of Seebeck coefficient.
 61

Na0.83RhO2 Na0.75RhO2
120 120
(a) (d)
80 80
S ( V/K)

40 40

0 0
S (10 W/cm K )

(b) (e)
2

20 A1 20
B1
A2
15 A3 15 B2
B3
10 10
14

5 5
2

0 0
0.4 (c) 0.4 (f)

0.3 0.3
zT

0.2 0.2

0.1 0.1

0 0
200 300 400 500 600 700 800 200 300 400 500 600 700 800
T (K) T (K)

Figure 4.6: Transport properties of non-hydrated Na0.83 RhO2 and Na0.75 RhO2 as
functions of the temperature.
 62

Chapter V

Nanostructured Bi2Se3

In this chapter, we investigate the effect of the thickness in terms of the number
of quintuple layers on the electronic structure and transport behavior. We will
demonstrate a high figure of merit for a single quintuple layer of Bi2 Se3 and will
argue that the value can be further enhanced by tensile strain. We calculate the
phonon dispersion under strain and show that the material remains stable.

5.1 Computational Parameters

We have carried out first principles calculations based on density functional theory,
employing the generalized gradient approximation of the exchange correlation po-
tential in the Perdew-Burke-Ernzerhof flavor and spin orbit coupling is included.
The van der Waals (vdW) interaction is taken into account. All calculations
are performed with a plane wave energy cutoff of 544 eV and a Monkhorst-Pack
8 8 1 ~k-mesh for the Brillouin zone integrations. Furthermore, we employ finer
16 16 1 and 40 40 1 ~k-meshes for calculating the density of states (DOS)
and transport properties, respectively. All structures are optimized until an en-
ergy convergence of 105 eV and a force convergence of 0.025 eV/A are reached.
A 15 A vacuum slab is used to prevent artificial interaction with periodic images
due to the employed periodic boundary conditions. The phonon frequencies are
 63
determined by density functional perturbation theory for evaluating the structural
stability [140]. Once self-consistency is achieved in the first principles calculations,
the transport properties are calculated using the semi-classical Boltzmann theory
within the constant relaxation time approximation. This approach yields accurate
results for various types of thermoelectric materials [141145].

Bi Se 5

0
7.336
Energy (Ryd)

-5
3.280
-10

7.352 7.329
-15
4.14 4.16 4.18 4.2 4.22 4.24
Lattice constant ( )
3.323 3.253

7.347 7.325 7.330

3.316 3.232 3.243

7.340 7.348 7.330 7.330

3.329 3.300 3.232 3.253

7.341 7.351 7.348 7.325 7.329

3.238 3.329 3.316 3.323 3.280

7.096 7.341 7.340 7.347 7.352 7.336

1QL 2QL 3QL 4QL 5QL 6QL

Figure 5.1: Structures of one to six quintuple layers of Bi2 Se3 after relaxation.

5.2 Structural Properties

In order to built supercells for one to six quintuple layers of Bi2 Se3 , we start
from a single quintuple layer using the bulk structural parameters and optimize
 64
the in-plane lattice constant, obtaining a slight elongation to a = 4.191 A. The
thickness of the quintuple layer increases from the experimental bulk value of
6.966 A (Ref. [78]) to 7.096 A, similar to the results in Ref. [146]. We then keep
the a lattice constant for all systems fixed at the value obtained for the single
quintuple layer and relax in each case the atomic positions. The resulting values
for the layer thickness (h) and separation between adjacent quintuple layers (d)
are summarized in Fig. 5.1. While effects of the vdW correction expectedly are
negligible for a single quintuple layer, our values for h and d for two and more
quintuple layers are larger than previously reported [146]. We note that we use
an optimized lattice constant, while these authors have employed the bulk value.
The effect of the vdW correction is significant and cannot be ignored in the case
of two and more quintuple layers in order to obtain correct results for h and d.

5.3 Electronic Properties

Band structures of relaxed bulk Bi2 Se3 (with and without vdW interaction) in-
cluding spin orbit coupling are shown in Fig. 5.2. We obtain a bandgap of 0.31
eV (experimental value 0.30 eV [71]) for the structure relaxed with vdW interac-
tion and a bandgap of 0.48 eV for that relaxed without vdW interaction. As a
consequence of including the vdW correction, our calculated energy gaps better
reproduce the experiment [91] than Ref. [146], see the band structures of the re-
laxed systems in Fig. 5.3. This improved agreement is due to the larger d 3.3
A (as compared to 3.0 A). Our results show that the single quintuple layer has
a bandgap of 0.41 eV, which decreases step by step to 0.23 eV for six quintuple
layers, see Fig. 5.4, along with previous experimental results from experiment 1
on Si substrate (Ref. [91]), experiment 2 on SiC substrate (Ref. [92]) and VASP
results from Ref. [146]. The largest Seebeck coefficient and figure of merit thus
are expected for a single quintuple layer of Bi2 Se3 . We will show in the following
that the Seebeck coefficient in fact decreases monotonically when the bandgap
 65
decreases, i.e., when the number of quintuple layers grows.

2
with vdW
without vdW

Energy (eV) 1

-1

-2
Z F L
Figure 5.2: Band structures of bulk Bi2 Se3 with and without vdW interaction.

1QL 2QL 3QL 4QL 5QL 6QL

1

0.5
E-EF (eV)

-0.5

-1

Figure 5.3: Band structures of one to six quintuple layers of Bi2 Se3 .

We next apply both compressive and tensile strains of 2.5% and 5% to explore
the effects on the thermoelectric response assuming that the system is defect free.
In the single quintuple layer case the layer thickness is reduced to 6.725 A and
6.920 A for 5% and 2.5% tensile strain, respectively, and grows to 7.470 A and
7.280 A for 5% and 2.5% compressive strain. Particularly, at 2.5% tensile strain,
the value is almost equal to the layer thickness in the bulk. The band structures
of a single quintuple layer of Bi2 Se3 under strain are depicted in Fig. 5.5 together
with the corresponding phonon dispersions: The bandgap reaches 0.60 eV for 5%
 66
0.8
Present results
0.7 Experiment 1
VASP result
0.6 Experiment 2
Band gap at (eV)
0.5

0.4

0.3

0.2

0.1

0
0 1 2 3 4 5 6
Number of QLs

Figure 5.4: Bandgap as a function of the number of quintuple layers of Bi2 Se3 .

compressive strain and decreases to 0.16 eV for 5% tensile strain. 2.5% tensile
strain applied to a single quintuple layer of Bi2 Se3 has about the same effect
as an increase of the thickness to 4 quintuple layers. This is just an example to
demonstrate that the bandgap can be tune almost independently by the strain and
the number of quintuple layers, which is a great advantage from the technology
point of view.

5.4 Phonon Band Structure

We obtain positive phonon frequencies in the whole range of strain under study
for the single quintuple layer, which confirms that structures are stable. There are
15 phonon modes at the center of the Brillouin zone, 12 optical and 3 acoustical.
The optical modes consist of A11g and A21g out-of-plane modes and E1g and E2g in-
plane modes. Nanostructured Bi2 Se3 is subject to a red shift of the frequencies
of the above modes due to phonon softening. The out-of-plane phonon softening
 67
is explained by the finite size along the c-axis, resulting in a reduction of the
restoring force. In an unstrained single quintuple layer of Bi2 Se3 , the E2g and A21g
frequencies are 129.3 cm1 and 174.8 cm1 , respectively, and thus close to the
experimental values for bulk Bi2 Se3 (131 cm1 and 174 cm1 ) [147].

1 200
150

-5% strain
0.5
0 100
-0.5 50
-11 0

-2.5% strain
0.5 150
0 100
-0.5 50
-11 0
Frequency (cm )
-1
E-EF (eV)

150

0% strain
0.5
0 100
-0.5 50
-11 0

+2.5% strain
0.5 150
0 100
-0.5 50
-11 0
+5% strain
0.5 150
0 100
-0.5 50
-1 0
K M M K

Figure 5.5: Electronic band structures and phonon dispersions of a single quintuple
layer of Bi2 Se3 for different strains.

Strain effects on the A21g mode at -point frequency in single quintuple layer
are addressed in Fig. 5.6 in comparison to the variation of the bandgap, in order to
demonstrate the analogous behavior of these two quantities. Under tensile strain
the frequency decreases, i.e., the out-of-plane phonons soften, while in the case of
compressive strain it increases. The Gruneisen parameter for the A21g mode (i.e.,
 68
800 200

700 190

600
180

(cm )
Eg (meV)

500

-1
170
400
160
300

200 150

100 140
-5 -2.5 0 2.5 5
Strain (%)

Figure 5.6: Bandgap and point phonon frequency of a single quintuple layer of
Bi2 Se3 .

0
the G mode at the point) is given by G = G /2G , where G is the
0
difference of the frequencies with and without strain and G is the frequency of
the unstrained single quintuple layer. We obtain positive values of 0.012 and 0.026
for 2.5% and 5% tensile strain, respectively, because the distance between the Bi
and Se sublattices is reduced. Accordingly, negative values appear for compres-
sive strain. So far no experimental data are available for comparison, though an
experimental confirmation of our predictions by Raman spectroscopy would be
desirable.

5.5 Transport Coefficients

We now turn to the thermoelectric properties of nanostructured Bi2 Se3 with vary-
ing thickness and strain. S and S 2 / are shown in Fig. 5.7 as functions of the
hole doping (in cm3 ). The maximum Seebeck coefficient is found for the single
 69
800 800
1QL -5%
2QL -2.5%
600 3QL 600 1QL
0%
S (VK )
-1

4QL
5QL +2.5%
400 400 +5%
6QL
200 200

0 0
2 2
S / (10 W/mK s)
2

1.5 1.5

1 1
11

0.5 0.5
2

0 0
17 18 19 20 21 17 18 19 20 21
10 10 10 10 10 10 10 10 10 10
-3 -3
(cm ) (cm )

Figure 5.7: Thermoelectric properties for systems of different thickness and for
various strains at 300 K.

quintuple layer around = 2.2 1017 cm3 . At low doping the value progres-
sively reduces when the number of quintuple layers grows, because the bandgap
decreases. Above = 1.21018 cm3 the Seebeck coefficient decreases in each case
for increasing hole doping. The reasons for the particularly high values found for
the single quintuple layer are (1) the maximal bandgap and (2) the qualitatively
different shape of the valence bands.
According to Fig. 5.7, the maximum of S 2 / for the single quintuple layer
is found at a hole doping of = 2.4 1020 cm3 , which is below the degenerate
limit. It shifts to higher when the number of quintuple layers grows, except for
the case of two quintuple layers. Can the value be further improved by strain? We
find that the maximal Seebeck coefficient shifts to higher and lower hole doping,
respectively, for tensile and compressive strain. While S 2 / is reduced by tensile
strain as well as by strong compressive strain, a moderate compressive strain
results in a significant enhancement around = 1020 cm3 .
 70
5.6 Figure of Merit

The BoltzTraP code [130] obtains the electrical conductivity as / . We estimate

for Bi2 Se3 by comparing the calculated values of / and S with the experimental
values of S and at 300 K, which results in = 2.7 1015 sec. This value is used
for all subsequent calculations of the TE properties. A similar methodology has
been used to establish the relaxation time in Bi2 Te3 [129] and has led to reasonable
agreement with the experiment. The assumption that is independent of both the
thickness and the strain is supported by previous theoretical work on Sb2 Te3 [149].
Experimentally, a thermal conductivity of = 0.49 W/Km at a temperature of
300 K has been reported for single quintuple layer Bi2 Se3 in Ref. [90] and is used
in the following.
Using the experimental = 0.49 W/mK and calculated = 2.71015 sec, we
estimate a figure of merit of 0.27 for the single quintuple layer, which is remarkably
more than the bulk value (0.10). A moderate compressive strain (2.5%) further
enhances the value to 0.30. In addition, it can be expected that S and thus the
figure of merit grow with the temperature and that a decreasing for higher
doping also has positive effects.
 71

Chapter VI

Tl-doping in Bulk Bi2Se3

In this chapter, we will investigate the influence of Thallium (Tl) at x = 0.33

on the physical properties of Bi2 Se3 which already show high powerfactor (S 2 )
at x = 0.1 in recent experimental work [94]. In order to enable a systematic
optimization, it is important to gain physical insight into the modifications of the
electronic band structure induced by Tl-doping, which is achieved and explain in
this chapter. In addition, we will study strain as a means of enhancing the figure
of merit, obtaining very promising results.

6.1 Computational Parameters

We model the effect of Tl-doping on Bi2 Se3 by addressing the system Bi1.67 Tl0.33 Se3
by first-principles calculations. We fully optimize the structural parameters and
relax the atomic positions, taking into account the van der Waals (vdW) interac-
tion. The plane wave energy cutoff is set to 544 eV and a Monkhorst-Pack 888
~k-mesh is used for the Brillouin zone integrations. Moreover, the Perdew-Burke-

Ernzerhof parametrization is employed and spin orbit coupling is included. An

energy convergence of 105 eV and a force convergence of 0.025 eV/A provide
sufficient accuracy. The transport behavior is studied under a dense ~k-mesh of
50 50 20 points.
 72
6.2 Structural Properties

For bulk Bi2 Se3 our optimized lattice parameters (a = b = 4.15 A and c =
28.28 A), the distance between adjacent quintuple layers (2.40 A), the thickness
of the quintuple layer (7.03 A), and the bandgap (312 meV) are in agreement with
experiment [78,150]. We replace one Bi atom with Tl to obtain Bi1.67 Tl0.33 Se3 and
then re-optimize the lattice parameters, which results in a considerable volume
reduction (a = b = 4.14 A and c = 27.90 A). Furthermore, we apply in-plane
biaxial strain by modifying the a and b lattice constants, optimizing the c lattice
constant and relaxing the atomic positions. For 2% compressive strain c grows to
29.05 A, whereas it decreases to 26.81 A for 2% tensile strain.

6.3 Electronic Properties

The band structure of unstrained Bi1.67 Tl0.33 Se3 is shown in Fig. 6.1 by black
lines. The indirect bandgap amounts to 158 meV, which is about half of the di-
rect bandgap of bulk Bi2 Se3 (300 meV). The Bi 6p states contribute mainly at
the conduction band minimum and valence band maximum, hybridized with Se
4p states. The Tl 6p and 5s states are located at higher energy in the conduction
band. Comparison to the band structure of Bi2 Se3 reveals that Tl-doping does
not modify the effective masses of the charge carriers at band edges, whereas the
bandgap is substantially reduced. In addition, the doping breaks the inversion
symmetry, which results in a momentum-dependent spin-orbit splitting (Rasbha
spin splitting) at the point (as a consequence of the potential gradient in the
direction perpendicular to the atomic layers), discuss in more detail in next sec-
tion. The band structure of strained Bi1.67 Tl0.33 Se3 is also shown in Fig. 6.1,
demonstrating significant strain effects particularly in the vicinity of the Fermi
energy.
Besides shifting the conduction band minimum and valence band maximum in
 73
0.5
-2%
-1%
0.4 0%
+1%
0.3 +2%

0.2

0.1
Energy (eV)

-0.1

-0.2

-0.3

-0.4

-0.5
M K

Figure 6.1: Band structure for zero, compressive, and tensile strain.

energy, strain also shifts them away from the point (see the inset in Fig. 6.1),
i.e., the Rasbha spin splitting and consequently the thermoelectric properties are
altered. The bandgap increases under compressive strain (195 meV for 1% and 202
meV for 2% strain) and decreases under tensile strain (127 meV for 1% and 104
meV for 2% strain), both monotonously. Under compressive strain the conduction
band at the M point shifts to lower energy, whereas under tensile strain the valence
band at the K point shifts to higher energy. By these variations and S are tuned,
which cannot be achieved by modifying the Tl concentration.

6.3.1 Rashba Spin Splitting

Rashba spin splitting, arising in compounds without structural inversion symmetry

[151], plays a leading role in this field, because its strength can be tuned by an
electric field or by strain. The effect can be quantified by the momentum offset kR ,
Rashba energy ER , and Rashba parameter R = 2ER /kR , which should be as large
as possible from the application point of view, as devices can be scaled, though
 74
this is difficult to achieve [152]. The spin orbit coupling may be exploited for spin
filtering devices and spin transistors (where it generates spin-dependent phase
shifts during transport) [153]. Rashba spin splitting occurs naturally at surfaces,
interfaces, and in doped materials (inversion symmetry broken by the impurity
atoms). We will demonstrate in the following a large Rashba spin splitting in
Tl-doped Bi2 Se3 , at least one order of magnitude larger than in semiconductor
heterostructures [152] and close to the value of BiTeI. We use Tl to break the
inversion symmetry of Bi2 Se3 because of its large spin orbit coupling and the
fact that it modifies only the bandgap and not the effective masses of the charge
carriers at the conduction band minimum and valence band maximum. We also
apply in-plane (biaxial) strain to tune the Rashba spin splitting by modifying the
potential gradient. We will show that the Rashba spin splitting is large/small for
a small/large bandgap.
0.75

0.5
0.2

0.25 0.1
ER
E-EF (eV)

0 0
kR

-0.25 -0.1
ER
-0.2
-0.5

-0.3
-0.75
M K

Figure 6.2: Band structures of Tl-doped Bi2 Se3 without and with spin orbit cou-
pling.

The band structures of Tl-doped Bi2 Se3 without and with spin orbit coupling
are shown in Fig. 6.2. When the spin orbit coupling is turned on, the bands develop
a splitting, so that the local minimum appears no longer at the high symmetry
point. This is a first signature of Rashba spin splitting [154]. The in-plane
 75

0.03 0.03

0.32 0.32

0.02 0.24 0.02 0.24

0.16 0.16
0.01 0.01

0.08 0.08

0.00 0.00 0.00 0.00

0.08 0.08

0.01 0.01

0.16 0.16

0.02 0.24 0.02 0.24

0.32 0.32

0.03 0.03
0.03 0.02 0.01 0.00 0.01 0.02 0.03 0.03 0.02 0.01 0.00 0.01 0.02 0.03

Figure 6.3: Spin structure of the Rashba split states in the xy-plane.

spin components at the conduction band minimum of unstrained Tl-doped Bi2 Se3
are illustrated in Fig. 6.3 and confirm the Rashba spin splitting scenario. We
obtain spin-resolved constant energy contours of almost perfect circular shape for
the inner branch and hexagonal shape for the outer branch. The latter feature is
similar to BiTeI [154], where the hexagonal shape is related to a size able out-of-
plane spin component.
The Rashba parameter as a function of the strain, for an electron at the con-
duction band minimum and a hole at the valence band maximum, is shown in
Fig. 6.4. We observe a significant strain dependence, which demonstrates that the
Rashba spin splitting in Tl-doped Bi2 Se3 can be considerably altered by modifying
the potential gradient. The Rashba parameter R depends on the Rashba energy
ER (a large value indicates that electrons of different spin can be easily stabilized
and controlled [151]) and the momentum offset kR (a large value indicates that a
large phase difference appears in spin transport [153]). Both these parameters are
tuned significantly by strain. We obtain ER = 66 meV at the conduction band
minimum and ER = 197 meV at the valence band maximum without strain and
an enhancement to remarkable values of 95 meV and 227 meV, respectively, under
2% tensile strain. The reduction of ER under compression is probably a conse-
quence of the fact that the atoms begin to shift in z direction. This expansion of
the lattice reduces the potential gradient and thus the Rashba spin splitting. The
 76
225
200

Band gap (meV)

175
150
125
100
4
3.5
3
R (eV)

2.5
2
1.5
1
-2 -1 0 1 2
strain (%)
Figure 6.4: Bandgap at the point and Rashba parameter at the conduction band
minimum for electrons and at the valence band maximum for holes.

value of kR increases monotonically from compressive to tensile strain, whereas

under compression ER decreases sharply as compared to kR , which is unfavorable
for technological aims. Tensile strain, on the other hand, can effectively enhance
ER and thus R .
Very high values of R = 1.57 eVA at the conduction band minimum and
R = 3.34 eVA at the valence band maximum are observed in unstrained Tl-doped
Bi2 Se3 , being about one order of magnitude larger than reported for semiconductor
heterostructures [152]. Together with the tunability of the spin splitting this offers
a promising route to strain engineered materials for spintronics. Electrons and
holes show similar strain dependencies of the Rashba spin splitting. The largest
Rashba parameter of 1.83 eVA for electrons and 3.64 eVA for holes is obtained
for the smallest bandgap, as R is enhanced [155].
 77
6.4 Transport Coefficients

Optimization of the carrier concentration is often useful to enhance the figure of

merit of a thermoelectric material. In the next section, we explore the transport
coefficients S, / , and S 2 / under strain as a function of the carrier concentra-
tion under strain.

6.4.1 Variation of S, / , and S 2 / at 300 K

The room temperature (300 K) S, / , and S 2 / of Bi1.67 Tl0.33 Se3 as a function

of the carrier concentration are shown in Fig. 6.5 for both n and p-doping and
for different values of the strain. An experimental room temperature S = 66
VK1 has been reported for x = 0.1 together with a carrier concentration of
2 1019 cm3 [94]. Our value at the same temperature and carrier concentration
is 114 VK1 for the unstrained system, which is higher than the experimental
value because of the higher Tl-doping of x = 0.33. For 5% compressive strain the
maximal values of S are found at hole/electron concentrations of about 1.5 1018
cm3 . In general, for low carrier concentrations the magnitude of S follows the
variation of the bandgap, whereas the maxima in Fig. 6.5 shift towards higher
carrier concentrations (as expected, since the bandgap decreases). Above carrier
concentrations of 1019 cm3 for p-doping compressive strain has essentially no
effect on S, whereas tensile strain enhances it considerably.
For n-doping strain effects are always negligible for high carrier concentrations.
For low doping / varies with the strain inversely to S, whereas the strain depen-
dence is completely lost above electron/hole concentrations of about 1019 cm3 .
For p-doping tensile strain affects the value of S 2 / more than compressive strain,
while for n-doping any strain effect is small. In general, for n-doping compressive
strain is advantageous and for p-doping tensile strain, where stronger effects are
observed in the latter case.
 78
(a) p-doping (b) n-doping
400 0
-2%
-1%
300 -100 0%
S (V/K)

S (V/K)
+1%
200 -200 +2%

100 -300

0 -400
19 19
10 10
/ (1/ms)

/ (1/ms)
18 18
10 10

17 17
10 10
4 4
S / (10 W/mK s)

S / (10 W/mK s)
2

3 2 3

2 2
17

17

1 1
2

0 0
18 19 20 21 18 19 20 21
10 10 10 10 10 10 10 10
-3 -3
Carrier concentration (cm ) Carrier concentration (cm )

Figure 6.5: Room temperature: S, / , and S 2 / under strain as a function of

the carrier concentration.

6.4.2 Variation of S, / , and S 2 / at 500 K

Fig. 6.6 shows S, / , and S 2 / at an elevated temperature of 500 K for both n

and p-doping as a function of the carrier concentration. In each case we find smaller
values for S and a shift of the maxima towards higher carrier concentrations as
compared to the results at 300 K (as expected, since more carriers are available).
Consequently, S 2 / is enhanced at high doping.
At a hole concentration of 7 1020 cm3 , where the highest values of S 2 /
are obtained, S and / are hardly modified by tensile strain. Hence, we can
conclude that even a moderate tensile strain is sufficient to achieve a high S 2 /
value at 500 K for p-doping. For n-doping, on the other hand, compressive and
tensile strain hardly enhance S 2 / . The temperature dependencies of S and
S 2 / under strain at a fixed p-doping of 71020 cm3 are addressed in Fig. 6.7.
We observe an almost linear increase with the temperature, except for the case of
 79
(a) p-doping (b) n-doping
400 0
-2%
-1%
300 -100 0%
S (V/K)

S (V/K)
+1%
200 -200 +2%

100 -300

0 -400
19 19
10 10
/ (1/ms)

/ (1/ms)
18 18
10 10

17 17
10 10
4 4
S / (10 W/mK s)

S / (10 W/mK s)
2

3 2 3

2 2
17

17

1 1
2

0 0
18 19 20 21 18 19 20 21
10 10 10 10 10 10 10 10
-3 -3
Carrier concentration (cm ) Carrier concentration (cm )

Figure 6.6: 500 K: S, / , and S 2 / under strain as a function of the carrier

concentration.

2% tensile strain where we have a smaller bandgap and reach the metallic regime
above 550 K.

6.5 Powerfactor and Figure of Merit

To estimate the powerfactor and figure of merit, we obtain a relaxation time

of = 1.27 1014 s by comparing the calculated value of / = 2 1017
(ms)1 without strain to the experimental = 2540 (m)1 [94] and = 2.75
W/mK of Bi1.98 Ca0.02 Se3 [83] at 300 K. For n-doping a maximal room temperature
powerfactor of 12.5 W/cmK2 is predicted for unstrained Bi1.67 Tl0.33 Se3 , which
exceeds the experimental values of bulk Bi2 Se3 (7.9 W/cmK2 ) and Bi1.9 Tl0.1 Se3
(11 W/cmK2 ) [94]. We next estimate the powerfactor and figure of merit for
temperatures of 300 K and 500 K under 2% tensile strain for the p-doping level
 80
at that S 2 / is maximal. This results in a powerfactor of 31 W/cmK2 at 300
K, which is slightly smaller than the value of bulk Bi2 Te3 (40 W/cmK2 [156])
and grows to 48 W/cmK2 at 500 K. Finally, a maximal figure of merit of 0.86 is
obtained at 500 K for a hole concentration of 7 1020 cm3 , which is about twice
the value at 300 K. We note that nanostructuring can be employed to further
reduce the thermal conductivity, as reported for Bi2 Se3 [157].

200

150
S (V/K)

100
-2%
-1%
0%
50 +1%
+2%
0
5
S / (10 W/mK s)
2

3
17

2
2

0
300 400 500 600 700 800
Temperature (K)

Figure 6.7: S and / as a function of the temperature under strain for a fixed
p-doping of 71020 cm3 .

Tl-doping modifies the bandgap without changing the effective masses at the
conduction band minimum and valence band maximum (though a strong momentum-
dependent spin-orbit splitting is observed, since Tl is a heavy element). By anal-
ogous calculations, we find that both Pb and Cu-doping of Bi2 Se3 result in a
metallic nature. This can be explain by a larger electronegativity difference be-
tween Bi (2.02) and Tl (1.62) as compared to that between Bi and Pb (2.33) or
 81
Cu (1.90). As a consequence, Tl donates more charge to Bi and Se as compared
to Pb and Cu and therefore does not contribute to the band edges, in contrast to
Pb and Cu. Within the IIIA group of the periodic table, the bandgap is found to
increase from Tl (6p) to In (5p), Ga (4p), and Al (3p) doping, whereas the dis-
persion does not change qualitatively, because the electronegativities are similar.
However, due to the larger bandgaps, strain in these cases is inefficient for tuning
the thermoelectric properties, similar to pristine Bi2 Se3 [87]. Tl-doping therefore
provides a unique combination of effects on the electronic structure that enables
strain engineering.
 82

Chapter VII

Alloying PtSb2 with Ir/As

In this chapter, we address the effect of electron and hole doping on thermoelectric
properties in PtSb2 , which has filled t2g orbital of Pt4+ (5d6 low-spin state) is
similar to the Co3+ (3d6 low-spin state) of NaCoO2 , and have a semiconducting
nature as well. We study the powerfactor and the figure of merit for both types
of doping from 300 K to 800 K. We also address the thermopower of PtAs2 and
the likely effects of alloying PtSb2 with As.

7.1 Computational Parameters

We calculate the band structure of PtSb2 using density function theory as im-
plemented in the WIEN2k package [115]. The popular generalized gradient ap-
proximation [128] is employed to optimize the volume and the internal atomic
coordinates. To simulate doping, we use the virtual crystal approximation [158]
and rigid band approach [159,160]. This approximation is widely employed in cal-
culations of transport properties of doped semiconductors and is accurate when
the doping is not too large. We use 3000 ~k-points in the full Brillouin zone for
calculating the electronic structure and a dense mesh of 3564 ~k-points in the irre-
ducible wedge of the Brillouin zone for the thermoelectric calculations.
 83
7.2 Structural Properties

PtSb2 crystallizes in a cubic pyrite structure, i. e., the Pt atoms constitute a face
centered cubic lattice, while the Sb pairs have their centers on one of the cube
edges or at the body center of the Pt lattice, with space group P a3, see Fig. 7.1
(a). Our optimized lattice constant (a = b = c) is 6.47 A which is close to the
experimental value of 6.44 A [161].
2
(a) (b)

1
E-EF (eV)

0 EF

-1

-2
X M R

Figure 7.1: Band structure of doped Pt0.99 Ir0.01 Sb2 .

7.3 Electronic Properties

The metallic states are reproduced by our band structure calculation for Pt0.99 Ir0.01 Sb2 ,
see Fig. 7.1 (b), while the undoped PtSb2 is an insulator with experimental
bandgap of 110 meV at T10 K [162]. Already a small doping of 0.04 holes
per unit cell results in a metallic state [101]. Similarly, for a small n-doping also
results in a metallic state [109]. The bands near the Fermi energy (EF ) mainly
are due to the Sb 5p orbitals, with some admixtures of the Pt 5d orbitals. The
band that crosses the EF originates completely from the Sb 5p orbitals throughout
the symmetry lines -X-M-. Mori and coworker have shown that the bands at
EF are corrugated and not the pudding mold type bands that establish the high
Seebeck coefficients in Nax CoO2 (Ref. [96]) and Kx RhO2 [163, 164]. Calculated
densities of states (DOSs) of PtSb2 and Pt0.99 Ir0.01 Sb2 are shown in Fig. 7.2. It
 84
15
PtSb2

DOS (states/eV)
10 Pt0.99Ir0.01Sb2

0
-2 -1 0 1 2
E-EF (eV)

Figure 7.2: Density of states of undoped and doped PtSb2 .

is clearly visible that doping does not change the shape of the DOS but only the
position of EF , justifying our usage of the rigid band approximation.

7.4 Transport Coefficients

7.4.1 Variation of S and /

The calculated Seebeck coefficient of PtSb2 in a doping range from 0.04 to 0.4
electrons/holes per unit cell for a temperature range from 300 to 800 K is plotted
in Figs. 7.3(a) and 7.4(a). Note that a doping of 0.04 electrons/holes per unit cell is
equivalent to a carrier concentration of 1.51020 cm3 (as obtained experimentally
for Pt0.99 Ir0.01 Sb2 (Ref. [101])). We find that S shows a maximum a bit above 100
VK1 for each doping level, except for dopings of 0.3 and 0.4 electrons/holes per
unit cell, but at different temperatures. This is due to the fact that the conduction
band close to EF gives a negative contribution to the Seebeck coefficient at different
temperatures and doping levels.
For nh = 0.04 holes per unit cell, our calculation gives at room temperature
a value of S = 122.5 VK1 , which roughly agrees with the experiment (100
VK1 ) [101]. Our maximum S value is 129.7 VK1 , while the experimental
value is 112 VK1 at 400 K. For nh = 0.04, the value of S remains high up to
 85
140
nh= 0.04
120 nh= 0.08
nh= 0.12
100
nh= 0.16
S (V/K) 80 nh=0.20
60 nh=0.30
nh=0.40
40

20 (a)
0
300 400 500 600 700 800
T (K)
400 K
550 K
660 K
18
780 K
/ (1/ cm s)

10 800 K
800 K
800 K

(b)
17
10
0.01 0.1 1
nh

Figure 7.3: S and / of hole doped PtSb2 .

450 K and decreases thereafter. Overall, Fig. 7.3 (a) shows that with increasing
hole doping the Seebeck coefficient decreases at room temperature, which is also
consistent with the experiment. Importantly, the maximal S value can be obtained
at different temperatures by controlling the carrier concentration, which helps
to achieve an optimal performance of the thermoelectric device under different
conditions. In Fig. 7.3 (b), we examine the effect of hole doping on the electrical
conductivity / at the temperatures for which we have obtained the highest S for
each doping. We find variations with the temperature in the case of low doping,
with the highest value at 800 K. For nh & 0.20, the value of / increases rapidly
and becomes virtually identical for the considered temperatures. We note also the
 86
0
ne=0.04
(a)
ne=0.08
-20
ne=0.12
ne=0.16
S (V/K)
-40
ne=0.20
ne=0.30
-60
ne=0.40

-80

-100
300 400 500 600 700 800
T (K)
300 K
(b) 460 K
610 K
18
740 K
/ (1/ cm s)

10 800 K
800 K
800 K

17
10
0.01 0.1 1
ne

Figure 7.4: S and / of electron doped PtSb2 .

unusual temperature dependence (significant increase) of / at low doping. This

is due to bipolar conduction, which increases substantially at higher temperatures.
The high temperature zT predictions are generally in a doping range sufficiently
heavy to preclude bipolar conduction, as this is destructive to the thermoelectric
performance.
As it is experimentally known that both types of doping are possible in PtSb2 ,
we also consider electron doping, which has not been addressed experimentally.
Nishikubo and coworkers have shown that the substitution of Sn at the Sb site
results in a small powerfactor [101]. This suggests that substitution at the Pt site
is more favorable for the transport properties. Hence, we study the substitution
 87
of Au at the Pt site, see Figs. 7.4(a) and 7.4(b). At room temperature, we find for
ne 0.04, 0.12, and 0.4, values of S = 91 VK1 , 62 VK1 , and 34 VK1 ,
respectively. This decreasing trend of S with increasing doping is similar to our
observations for hole doping. The maximal S is always found around 94 VK1 ,
except for ne 0.3 and 0.4, in the considered temperature range. It shifts more and
more to higher temperature with increasing doping. The variation of / with
the electron concentration and temperature is similar to the case of hole doping.

7.4.2 Powerfactor and Figure of Merit

It is fruitful to compare electron and hole doping in PtSb2 for possible applications
in thermoelectric generators with the same host material. In Figs. 7.5 (a)7.5
(d), we present S 2 and zT of doped PtSb2 in a doping range from 0.04 to 0.4
electrons/holes per unit cell and temperature range from 300 K to 800 K. For a
realistic description of the conductivity, we vary the mobility with respect to the
carrier concentration and temperature. Whereas the electronic contribution el is
deduced from the Wiedemann-Franz relation, the phononic contribution is taken
from the experiment and varied as ph T1 . Details of the methodology can be
found in Ref. [165]. According to Fig. 7.5 (a), the calculated room temperature
S 2 for nh = 0.40 is 31 W/cmK2 , which is in excellent agreement with the
experimental value of 35 W/cmK2 . For all other doping levels, S 2 decreases
with temperature, as in the experiment. In the case of electron doped PtSb2 , the
value of S 2 reaches up to 17 W/cmK2 for low doping, which is about half the
value obtained for hole doping, while at higher doping the electron and hole doped
materials exhibit a S 2 of similar magnitude at room temperature. Importantly,
S 2 shows the opposite temperature trend for a doping of 0.04 (both electron
and hole) as compared to the other doping levels. The room temperature zT for a
doping of 0.04 holes per unit cell is about 0.1, which is in good agreement with the
experimental value of 0.12. For doping levels between 0.12 and 0.2 electrons/holes
 88
hole doped electron doped
40 40
(a) (b) n=0.04
n=0.08
30 30
S (W/cmK )

n=0.12
2

n=0.16
n=0.20
20 20 n=0.30
n=0.40
2

10 10

0 300 400 500 600 700 800

0
0.25 (c) 0.25 (d)
0.2 0.2

0.15 0.15
zT

0.1 0.1

0.05 0.05

0 0
300 400 500 600 700 800 300 400 500 600 700 800
T (K) T (K)

Figure 7.5: S 2 and ZT for electron and hole doped PtSb2 as functions of the
temperature.

per unit cell, a large S 2 at 800 K results in a high zT of 0.1/0.2, which is more
than twice the room temperature value. A further enhancement of zT by reduction
of the thermal conductivity may be achieved in nanostructured PtSb2 .

7.5 Arsenic Alloying in PtSb2

The previous discussion demonstrates a relatively low performance of PtSb2 , pri-

marily due to two factors: its comparatively large 300 K lattice thermal conduc-
tivity of approximately 7.5 W/mK and the lack of a bandgap, which severely
limits the Seebeck coefficient. Experimentally, PtSb2 appears to have a small
bandgap of 0.1 eV, which we are unable to reproduce. However, in practice,
the large lattice thermal conductivity will strongly hamper performance at all
but the highest temperatures, where the slight calculational discrepancy will have
 89
little effect. There is, however, an opportunity for significant zT enhancement
via alloying with As. PtAs2 is isoelectronic and isostructural to PtSb2 , with a
somewhat smaller lattice parameter of 5.97 A. More importantly, however, PtAs2
exhibits a significant bandgap of approximately 0.5 eV (our calculations with the
modified Becke-Johnson potential find a bandgap of 0.38 eV). This means that
the Seebeck coefficients in PtAs2 will not be nearly as impacted by bipolar con-
duction as in PtSb2 , where this effect precludes Seebeck coefficients much higher
than 100 VK1 above room temperature. Given this larger bandgap, it is likely
that an alloy PtAs2x Sbx will have a significant bandgap, although the exact value
is unknown (and x-dependent) and must await experimental investigation. Such
alloying would also have a beneficial effect on the lattice thermal conductivity,
although the effect is largest at low temperature, while the best performance of
PtAs2x Sbx is expected at high temperature.
With these effects in mind, we depict in Figs. 7.6(a) and 7.6(b) the calculated
thermopower of PtAs2 within the constant relaxation time approximation (note
that no internal coordinate or lattice parameter optimization was performed) for
hole and electron concentrations comparable to those presented for PtSb2 . For low
temperature (200 to 300 K), the values are comparable to those obtained for the
corresponding dopings (per unit cell) for PtSb2 . At higher temperature, however,
the values continue to rise due to the bandgap, exceeding absolute values of 200
VK1 around 1000 K, for both hole and electron doping. This is a typical value
for a high performance thermoelectric, although it is not clear whether appropriate
optimization techniques in this family of materials can lead to high performance
thermoelectrics (typically described by zT of unity or greater).
Since we are concerned with As alloying, we have included in the figures ther-
mopower results in which the bandgap is taken as 0.2 eV, or half of the calculated
bandgap of PtAs2 , as a rough estimate of the bandgap of PtAsSb (dotted lines
in Fig. 7.6). The p-type thermopower (top panel) still shows values exceeding
 90
300
(a) nh=0.04
250 nh=0.08
nh=0.12
200 nh=0.16
S (V/K)
150

100

50

0
200 400 600 800 1000 1200
T (K)
0
ne=0.04
(b)
ne=0.08
-50
ne=0.12
ne=0.16
S (V/K)

-100
ne=0.02

-150

-200

-250
200 400 600 800 1000 1200
T (K)

Figure 7.6: Thermopower of PtAs2 for several hole and electron dopings.

200 VK1 at temperatures of 800 to 1200 K, whereas the corresponding ther-

mopowers for n-doping do not exceed 160 VK1 . It, thus, is likely that for the
alloying scenario considered here the p-type performance will significantly outstrip
the n-type performance. A second conclusion to be drawn is that for n-doping
higher As concentrations (with likely larger bandgaps) are more probably optimal
than for p-doping. We also point out that the likely dominance of the lattice ther-
mal conductivity in the heat transport means that nanostructuring techniques to
reduce this term can be helpful. Lastly, as a practical matter, we note the high
vapor pressure of the toxic As in the high temperature range, so that great care
in sample preparation and measurement is a necessity.
 91

Chapter VIII

Conclusion

Thermoelectric materials can provide solutions for future power generation and
refrigeration technologies. Layered oxide materials show huge potential for ther-
moelectric devices because of their natural abundance, non-toxicity, and stability
at high temperature. Nevertheless, efficiency is still the major issues for use in
thermoelectric generators. There is a need to discover new layered oxide materials
with enhanced transport properties. In addition, there is the possibility to mod-
ify the existing classic Bismuth based layered materials, by replacing the trace
element to enhance the performance at high temperature to make them more fea-
sible from the application point of view. The goal of this thesis is to develop an
understanding of the electronic properties of selected materials at a fundamental
level and to tune the transport properties.
In chapter III the electronic, optical, and transport properties of layered Kx RhO2
(x = 1/2 and 7/8) for the 2H formation of RhO6 , in the hydrated and non-hydrated
phases were discussed. The optimized structure of K1/2 RhO2 exhibits a remark-
able deviation of the c/a ratio from the experimental result. This indicates that
a hydrated phase exists and that the experimental structure determination refers
to this hydrated phase. The calculated Seebeck coefficient of pristine K1/2 RhO2
amounts to 50 V/K at 300 K, which is close to the experimental value of 40
V/K. Importantly, we find huge values for the Seebeck coefficient and powerfac-
 92
tor for hydrated Kx RhO2 in the whole temperature range from 0 to 700 K. At 100
K we obtain for hydrated K7/8 RhO2 a figure of merit of 0.3, which is the highest
value observed for a hole doped system at this temperature. It exceeds the ex-
ceptionally high value of Na0.88 CoO2 by more than 50%. Our results demonstrate
that hydration is novel and effective approach to modify the lattice parameters
and, as a result, enhance the thermoelectric performance. Suitable doping at the
Rh site and nano-structuring could help to reduce the lattice thermal conductivity
and enhance the figure of merit.
In chapter IV the Nax RhO2 oxides were found to form a new 3R phase with
exciting thermoelectric features. The modified stacking of the atomic layers in
this phase results in a reduced inter-layer coupling and, in turn, in a dramatically
enhanced thermoelectric response in the technologically relevant high temperature
range. We have also demonstrated that Na vacancies in Nax RhO2 avoid clustering
and that the RhO6 octahedra are modified depending on the amount of Na defi-
ciency. Analysis of the induced changes in the density of states close to the Fermi
level indicates that mainly Rh3+ d3z2 r2 states control the transport properties
of the compounds. A high zT value of 0.35 at 580 K is achieved in hydrated
Na0.83 RhO2 due to the enhanced effective mass of the charge carriers. In gen-
eral, zT can be further increased by reduction of the Na vacancy concentration to
increase the resistivity by nanostructuring.
In chapter V we have established the dependence of the thermoelectric prop-
erties of nanostructured Bi2 Se3 on the thickness of the material and on applied
compressive and tensile strain. We have demonstrated that a careful descrip-
tion of the structural properties, mainly the thickness of the quintuple layers and
their separation, as determined by the van der Waals interaction, are essential
for obtaining a quantitatively correct bandgap. We have shown that the bandgap
decreases when the number of quintuple layers increases, which results in smaller
Seebeck coefficients. The bandgap of a single quintuple layer reaches a value of
 93
0.60 eV for 5% compressive strain, whereas we obtain 0.16 eV for 5% tensile strain.
We have confirmed the structural stability of the systems under strain by means
of phonon dispersions. The highest figure of merit is found in the thinnest system
(single quintuple layer) and our results demonstrate that it is enhanced by com-
pressive strain. A value of 0.30 is achieved for a feasible strain of 2.5%, which is
three times the bulk value. Strain engineering of nanostructured Bi2 Se3 therefore
turns out to be an efficient approach for creating high performance TE materials
without trace element.
In chapter VI we have focused on the thermoelectric properties of Tl-doped
Bi2 Se3 . Our results explain the experimental finding that the powerfactor is en-
hanced for increasing Tl-doping [94], because the bandgap is reduced while the
effective mass of the charge carriers at the band edges is maintained. This effect
of Tl is found to be exceptional under the possible dopants. Moderate strain can-
not be used to enhance the powerfactor of pristine Bi2 Se3 , because the bandgap
is 300 meV and therefore very large with respect to the thermal energy. After
Tl-doping, on the contrary, strain turns out to be a very efficient tool for opti-
mizing the powerfactor, because of the reduction of the bandgap to 158 meV. We
have demonstrated that in Tl-doped Bi2 Se3 the bandgap decreases monotonously
from compressive to tensile strain. For high carrier concentrations tensile strain
enhances S more for p-doping than for n-doping due to the different dispersions
of the valence and conduction bands. The maximal powerfactor (48 W/cmK2 )
and figure of the merit (0.86) are obtained for a p-doping of 7 1020 cm3 under
2% tensile strain. Furthermore, Rashba spin splitting is observed. A remarkably
large Rashba energy (66 meV at the conduction band minimum and 197 meV at
the valence band maximum) is obtained, which further grows under tensile strain.
The Rashba parameter R reaches a maximum of 1.83 eVA for electrons and 3.64
eVA for holes at a tensile strain of 2%. These results provide directions to new
materials with large Rashba spin splitting and to strain engineering according to
 94
technological needs.
In chapter VII, we have studied the transport properties of electron and hole
doped PtSb2 over a wide doping range. A doping of 0.04 electrons/holes per unit
cell gives a high powerfactor at room temperature, which decreases with increasing
temperature. Our results show that doping between 0.12 and 0.20 electrons/holes
per unit cell, the zT values, while low, are more than double those achieved at
800 K, due to a high electrical conductivity. Experiments should be performed
at this doping level and temperature range for confirmation. Moreover, it will
be desirable to study in more detail nanostructuring strategies in doped PtSb2 to
further reduce the thermal conductivity. Finally, we have explored the effect of
As alloying, finding that this approach may significantly enhance the performance
due to a likely larger bandgap.
 95

REFERENCES

[1] L. D. Hicks and M. S. Dresselhaus, Effect of quantum-well structures on

the thermoelectric figure of merit Physical Review B 47, 12727 (1993).

[2] L. D. Hicks, T. C. Harman, X. Sun, and M. S. Dresselhaus, Experimental

study of the effect of quantum-well structures on the thermoelectric figure
of merit Physical Review B 53, R10493 (1996).

[3] S. A. Omer and D. G. Infield, Design and thermal analysis of two stage
solar concentrator for combined heat and thermoelectric power generation
Energy Conversion and Management 41, 737 (2000).

[4] R. Venkatasubramanian, E. Siivola, T. Colpitts, and B. OQuinn, Thin-

film thermoelectric devices with high room-temperature figures of merit
Nature 413, 597 (2001).

[5] T. C. Harman, P. J. Taylor, M. P. Walsh, and B. E. LaForge, Quantum

dot superlattice thermoelectric materials and devices Science 297, 2229
(2002).

[6] S. B. Riffat and X. Ma, Thermoelectrics: A review of present and poten-

tial applications Applied Thermal Engineering 23, 913 (2003).

[7] T. Jinushi, M. Okahara, Z. Ishijima, H. Shikata, and M. Kambe, Develop-

ment of the high performance thermoelectric modules for high temperature
heat sources Materials Science Forum 534, 1521 (2007).

[8] M. S. Dresselhaus, G. Chen, M. Y. Tang, R. Yang, H. Lee, D. Wang, Z.

Ren, J.-P. Fleurial, and P. Gogna, New directions for low-dimensional
thermoelectric Materials Advanced Materials 19, 1 (2007).

[9] A. Yadav, K. P. Pipe, and M. Shtein, Fiber-based flexible thermoelectric

power generator Journal of Power Sources 175, 909 (2008).

[10] G. J. Snyder and E. S. Toberer, Complex thermoelectric materials Na-

ture Materials 7, 105 (2008).

[11] L. E. Bell, Cooling, heating, generating power, and recovering waste heat
with thermoelectric systems Science 321, 1457 (2008).
 96
[12] J.-C. Zheng, Recent advances on thermoelectric materials Frontiers of
Physics in China 3, 269 (2008).

[13] J. R. Sootsman, D. Y. Chung, and M. G. Kanatzidis, New and old con-

cepts in thermoelectric materials Angewandte Chemie International Edi-
tion 48, 8616 (2009).

[14] J.-F. Li, W.-S. Liu, L.-D. Zhao, and M. Zhou, High-performance nanos-
tructured thermoelectric materials NPG Asia Materials 2, 152 (2010).

[15] T. Graf, C. Felser, and S. S. P. Parkin, Simple rules for the understanding
of Heusler compounds Progress in Solid State Chemistry 39, 1 (2011).

[16] A. Shakouri, Recent developments in semiconductor thermoelectric

physics and materials Annual Review of Materials Research 41, 399
(2011).

[17] T. M. Tritt, Thermoelectric phenomena, materials, and applications An-

nual Review of Materials Research 41, 433 (2011).

[18] J. R. Szczech, J. M. Higgins and S. Jin, Enhancement of the thermo-

electric properties in nanoscale and nanostructured materials Journal of
Materials Chemistry 21, 4037 (2011).

[19] G. G. Yadav, J. A. Susoreny, G. Zhang, H. Yang, and Y. Wu,

Nanostructure-based thermoelectric conversion: an insight into the fea-
sibility and sustainability for large-scale deployment Nanoscale 3, 3555
(2011).

[20] K. Biswas, J. He, I. D. Blum, C. Wu, T. P. Hogan, D. N. Seidman, V.

P. Dravid, and M. G. Kanatzidis, High-performance bulk thermoelectrics
with all-scale hierarchical architectures Nature 489, 414 (2012).

[21] A. Soni, Y. Shen, M. Yin, Y. Zhao, L. Yu, X. Hu, Z. Dong, K. A. Khor,

M. S. Dresselhaus, and Q. Xiong, Interface driven energy filtering of
thermoelectric power in spark plasma sintered Bi2 Te2.7 Se0.3 nanoplatelet
composites Nano Letter 12, 4305 (2012).

[22] Y. Pei, H. Wang, and G. J. Snyder, Band engineering of thermoelectric

materials Advanced Materials 24, 6125 (2012).

[23] S. Walia, S. Balendhran, H. Nili, S. Zhuiykov, G. Rosengarten, Q. H.

Wang, M. Bhaskaran, S. Sriram, M. S. Strano, and K. Kalantar-zadeh,
Transition metal oxides-Thermoelectric properties Progress in Materials
Science 58, 1443 (2013).

[24] D. Greig, Electrons in metals and semiconductors McGrawHill: London,

UK, 1969.

[25] R. Franz and G. Wiedemann, Ueber die Warme-Leitungsfahigkeit der

metalle Annalen der Physik 165, 497 (1853).
 97
[26] T. M. Tritt, Thermal Conductivity: theory, properties, and applications
Plenum Publishers: New York, (2004).

[27] A. Sommerfeld, Zur Elektronentheorie der Metalle Die Naturwis-

senschaften 15, 825 (1927).

[28] M. Cutler and N. F. Mott, Observation of Anderson localization in an

electron gas Physical Review 181, 1336 (1969).

[29] S. S. Li, Semiconductor physical electronics Chapter 8, Springer (2006).

[30] E. F. Steigmeier and B. Abele, Scattering of phonons by electrons in

Germanium-Silicon alloys Physical Review B 136, A1149 (1964).

[31] D. P. White and P. G. Klemens, Thermal conductivity of thermoelectric

Si0.8 Ge0.2 alloys Journal of Applied Physics 71, 4258 (1992).

[32] D. Emin, Thermoelectrics HandbookMacro to Nano, Taylor & Francis,

(2006).

[33] F. Stabler, DARPA/ONR Program Review and DOE High Efficiency

Thermoelectric Workshop, San Diego, CA, March 24-27, (2002).

[34] X. Shi, W. Zhang, L. D. Chen, and J. Yang, Filling fraction limit for in-
trinsic voids in crystals: Doping in skutterudites Physical Review Letters
95, 185503 (2005).

[35] Industrial Technologies Program, Engineering scoping study of thermo-

electric generator systems for industrial waste heat recovery (U.S. De-
partment of Energy, 2006).

[36] S. B. Inayat, K. R. Rader, and M. M. Hussain, Nano-materials enabled

thermoelectricity from window glasses Scientific Reports 2, 841 (2012).

[37] S. B. Inayat and M. M. Hussain, Power generation from thermoelec-

tric systems embedded plexiglas for green building technology Applied
Nanoscience 3,175 (2013).

[38] S. B. Inayat, K. R. Rader, and M. M. Hussain, Nano-manufacturing of

thermoelectric nanomaterials (Bi0.4 Sb1.6 Te3 /Bi1.75 Te3.25 ) and their integra-
tion into window glasses for thermoelectricity generation Energy Technol-
ogy 2, 221 (2014).

[39] S. B. Inayat and M. Hussain, Nano-micro materials enabled thermoelec-

tricity from window glasses, http://hdl.handle.net/10754/270852, (2013).

[40] I. Terasaki, Y. Sasago, and K. Uchinokura, Large thermoelectric power

in NaCo2 O4 single crystals Physical Review B 56, R12685 (1997).

[41] T. Motohashi, R. Ueda, E. Naujalis, T. Tojo, I. Terasaki, T. Atake, M.

Karppinen, and H. Yamauchi, Unconventional magnetic transition and
transport behavior in Na0.75 CoO2 Physical Review B 67, 064406 (2003).
 98
[42] M. L. Foo, Y. Wang, S. Watauchi, H. W. Zandbergen, T. He, R. J. Cava,
and N. P. Ong, Charge ordering, commensurability, and metallicity in
the phase diagram of the layered Nax CoO2 Physical Review Letters 92,
247001 (2004).

[43] K. Takada, H. Sakurai, E. Takayama-Muromachi, F. Izumi, R. A. Dilanian,

and T. Sasaki, Superconductivity in two-dimensional CoO2 layers Nature
422, 53 (2003).

[44] M. Lee, L. Viciu, L. Li, Y. Wang, M. L. Foo, S. Watauchi, R. A. Pascal,

R. J. Cava, and N. P. Ong, Large enhancement of the thermopower in
Nax CoO2 at high Na doping Nature Materials 5, 537 (2006).

[45] I. Terasaki, M. Iwakawa, T. Nakano, A. Tsukuda, and W. Kobayashi,

Novel thermoelectric properties of complex transition-metal oxides Dal-
ton Transactions 39, 1005 (2010).

[46] G. A. Slack, New materials and performance limits for thermoelectric

cooling Handbook of thermoelectric CRC page 407 (1995).

[47] M. Ohtaki, T. Tsubota, K. Eguchi, and H. Arai, High-temperature ther-

moelectric properties of (Zn12x Alx )O Journal of Applied Physics 79,
1816 (1996).

[48] J. He and Y. Liu, Oxide thermoelectrics: The challenges, progress, and

outlook Journal of Materials Research 26, 1762 (2011).

[49] I. Terasaki and N. Murayama, Oxide thermoelectrics Trivandrum, India:

Res. Signpost, eds. (2002).

[50] M. D. Johannes and D. J. Singh, Comparison of the electronic structures

of hydrated and unhydrated Nax CoO2 : The effect of H2 O Physical Review
B 70, 014507 (2004).

[51] L. Viciu, J. W. G. Bos, H. W. Zandbergen, Q. Huang, M. L. Foo, S.

Ishiwata, A. P. Ramirez, M. Lee, N. P. Ong, and R. J. Cava, Crystal
structure and elementary properties of Nax CoO2 (x = 0.32, 0.51, 0.6, 0.75,
and 0.92) in the three-layer NaCoO2 family Physical Review B 73, 174104
(2006).

[52] M. E. Grillo, Stability of corundum versus rutile-type structures of ruthe-

nium and rhodium oxides Physical Review B 70, 184115 (2004).

[53] S. Music, A. Saric, S. Popovic, and M. Ivanda, Formation and charac-

terization of nanosize -Rh2 O3 particles Journal of Molecular Structure
294, 221 (2009).

[54] P. Watson and G. Somorjai G. The hydrogenation of carbon monoxide

over rhodium oxide surfaces Journal of Catalysis 72, 347 (1981).
 99
[55] W. C. Dautremont-Smith, Transition metal oxide electrochromic materi-
als and displays: a review: part 2: oxides with anodic coloration Displays
3, 67 (1982).

[56] D. A. J. Michel Ligthart, R. A. vanSanten, and E. J. M. Hensen, Sup-

ported rhodium oxide nanoparticles as highly active CO oxidation cata-
lysts Angewandte Chemie International Edition 50, 5306 (2011).

[57] W. Koshibae, K. Tsutsui, and S. Maekawa, Thermopower in cobalt ox-

ides Physical Review B 62, 6869 (2000).

[58] Y. Klein, S. Hebert, D. Pelloquin, V. Hardy, and A. Maignan, Mag-

netoresistance and magnetothermopower in the rhodium misfit oxide
[Bi1.95 Ba1.95 Rh0.1 O4 ][RhO2 ]1.8 Physical Review B 73, 165121 (2006).

[59] S. Shibasaki, W. Kobayashi, and I. Terasaki, Transport properties of the

delafossite Rh oxide Cu1x Agx Rh1y Mgy O2 : Effect of Mg substitution on the
resistivity and Hall coefficient Physical Review B 74, 235110 (2006).

[60] W. Kobayashi, S. Hebert, D. Pelloquin, O. Perez, and A. Maignan, En-

hanced thermoelectric properties in a layered rhodium oxide with a trigonal
symmetry Physical Review B 76, 245102 (2007).

[61] S. Okada and I. Terasaki, Physical properties of Bi-based rhodium oxides

with RhO2 hexagonal layers Japanese Journal of Applied Physics 44,
1834 (2005).

[62] S. Okada, A. Sakai, T. Kanno, S. Yotsuhashi, and H. Adachi, Thermo-

electric properties of bismuth based cobaltrhodium oxides with hexagonal
(Co,Rh)O2 layers Journal of Applied Physics 105, 083711 (2009).

[63] S. Shibasaki, I. Terasaki, E. Nishibori, H. Sawa, J. Lybeck, H. Yamauchi,

and M. Karppinen, Magnetic and transport properties of the spin-state
disordered oxide La0.8 Sr0.2 Co1x Rhx O3 Physical Review B 83, 094405
(2011).

[64] S. Shibasaki, T. Nakano, I. Terasaki, K. Yubuta, and T. Kajitani, Trans-

port properties of the layered Rh oxide K0.49 RhO2 Journal of Physics:
Condensed Matter 22, 115603 (2010).

[65] R. Okazaki, Y. Nishina, Y. Yasui, S. Shibasaki, and I. Terasaki, Opti-

cal study of the electronic structure and correlation effects in K0.49 RhO2
Physical Review B 84, 075110 (2011).

[66] T. Motohashi, Y. Sugimoto, Y. Masubuchi, T. Sasagawa, W. Koshibae,

T. Tohyama, H. Yamauchi, and S. Kikkawa, Impact of lithium compo-
sition on the thermoelectric properties of the layered cobalt oxide system
Lix CoO2 Physics Review B 83, 195128 (2011).
 100
[67] A. Mendiboure, H. Eickenbusch, and R. Schollhorn, Layered alkali
rhodium oxides Ax RhO2 : Topotactic solvation, exchange, and redox re-
actions Journal of Solid State Chemistry 71, 19 (1987).

[68] A. Varela, M. Parras, and J. M. Gonzalez-Calbet, Influence of Na content

on the chemical stability of nanometric layered Nax RhO2 (0.7 x 1.0)
European Journal of Inorganic Chemistry 2005, 4410 (2005).

[69] Y. Krockenberger, M. Reehuis, G. Cristiani, C. Ritter, H. U. Habermeier,

and L. Alff, Neutron scattering and magnetic behavior of Nax RhO2
Physica C 460, 468 (2007).

[70] S. Park, K. Kang, W. Si, W.-S. Yoon, Y. Lee, A. R. Moodenbaugh, L. H.

Lewis, and T. Vogt, Synthesis and characterization of Na0.3 RhO2 0.6H2 O
a semiconductor with a weak ferromagnetic component Solid State Com-
munication 135, 51 (2005).

[71] S. K. Mishra, S. Satpathy, and O. Jepsen, Electronic structure and ther-

moelectric properties of bismuth telluride and bismuth selenide Journal
of Physics: Condense Matter 9, 461 (1997).

[72] Y. Xia, D. Qian, D. Hsieh, L. Wray, A. Pal, H. Lin, A. Bansil, D.

Grauer, Y.S. Hor, R.J. Cava, and M.Z. Hasan, Observation of a large-gap
topological-insulator class with a single Dirac cone on the surface Nature
Physics 5, 398 (2009).

[73] H. Zhang, C.-X. Liu, X.-L. Qi, X. Dai, Z. Fang, and S.-C. Zhang, Topo-
logical insulators in Bi2 Se3 , Bi2 Te3 , and Sb2 Te3 with a single Dirac cone
on the surface Nature Physics 5, 438 (2009).

[74] W. Zhang, R. Yu, H.-J. Zhang, X. Dai, and Z. Fang, First-principles stud-
ies of the three-dimensional strong topological insulators Bi2 Se3 , Bi2 Te3 ,
and Sb2 Te3 New Journal of Physics 12, 065013 (2010).

[75] Z. Ren, A. A. Taskin, S. Sasaki, K. Segawa, and Y. Ando, Large bulk

resistivity and surface quantum oscillations in the topological insulator
Bi2 Te2 Se Physical Review B 82, 241306(R) (2010).

[76] B. Poudel, Q. Hao, Y. Ma, Y. Lan, A. Minnich, B. Yu, X. Yan, D. Wang,

A. Muto, D. Vashaee, X. Chen, J. Liu, M. S. Dresselhaus, G. Chen, and
Z. Ren, High-thermoelectric performance of nanostructured bismuth an-
timony telluride bulk alloys Science 320, 634 (2008).

[77] W. Xie, X. Tang, Y. Yan, Q. Zhang, and T. M. Trit, Unique nanos-

tructures and enhanced thermoelectric performance of melt-spun BiSbTe
alloys Applied Physics Letters 94, 102111 (2009).

[78] S. Nakajima, The crystal structure of Bi2 Te3x Sex Journal of Physics
and Chemistry of Solids 24, 479 (1963).
 101
[79] L. Xue, P. Zhou, C. X. Zhang, C. Y. He, G. L. Hao, L. Z. Sun, and J.
X. Zhong, First-principles study of native point defects in Bi2 Se3 AIP
Advances 3, 052105 (2013).

[80] S. I. Kim, K. Ahn, D.-H. Yeon, S. Hwang, H.-S. Kim, S. M. Lee, and K.-H.
Lee, Enhancement of seebeck coefficient in Bi0.5 Sb1.5 Te3 with high-density
tellurium nanoinclusions Applied Physics Express 4, 09180 (2011).

[81] D. Kong, K. J. Koski, J. J. Cha, S. S. Hong, and Y. Cui, Ambipolar

field effect in Sb-doped Bi2 Se3 nanoplates by solvothermal synthesis Nano
Letters 13, 632 (2013).

[82] C. Martin, V. Craciun, K. H. Miller, B. Uzakbaiuly, S. Buvaev, H. Berger,

A. F. Hebard, and D. B. Tanner, Bulk Fermi surface and electronic prop-
erties of Cu0.07 Bi2 Se3 Physical Review B 87, 201201(R) (2013).

[83] Y. S. Hor, A. Richardella, P. Roushan, Y. Xia, J. G. Checkelsky, A. Yaz-

dani, M. Z. Hasan, N. P. Ong, and R. J. Cava, p-type Bi2 Se3 for topolog-
ical insulator and low-temperature thermoelectric applications Physical
Review B 79, 195208 (2009).

[84] J. Horak, P. Lostak, and A. Montaner, Point defects in Ga-doped Bi2 Se3
single crystals Physica Status Solidi B 119, K17 (1983).

[85] P. Lostak, L. Benes, S. Civis, and H. Sussmann, Preparation and some

physical properties of Bi2x Inx Se3 single crystals Journal of Materials
Science 25, 277 (1990).

[86] K.-H. Lee, S. Hwang, B. Ryu, K. Ahn, J. Roh, D. Yang, S.-M. Lee,
H.Kim, and S.-I. Kim, Enhancement of the thermoelectric performance of
Bi0.4 Sb1.6 Te3 alloys by In and Ga doping Journal of Electronic Materials
42, 1617 (2013).

[87] X. Luo, M. B. Sullivan, and S. Y. Quek, First-principles investigations

of the atomic, electronic, and thermoelectric properties of equilibrium and
strained Bi2 Se3 and Bi2 Te3 including van der Waals interactions Physical
Review B 86, 184111 (2012).

[88] D. Parker and D.J. Singh, Potential thermoelectric performance from

optimization of hole-doped Bi2 Se3 Physical Review X 1, 021005(2011)

[89] M. K. Jana, K. Biswas, and C. N. R. Rao, Ionothermal synthesis of few-

layer nanostructures of Bi2 Se3 and related materials Chemistry-A Euro-
pean Journal 19, 9110 (2013).

[90] Y. Sun, H. Cheng, S. Gao, Q. Liu, Z. Sun, C. Xiao, C. Wu, S. Wei,

and Y. Xie, Atomically thick bismuth selenide freestanding single layers
achieving enhanced thermoelectric energy harvesting Journal of American
Chemical Society 134, 20294 (2012).
 102
[91] Y. Sakamoto, T. Hirahara, H. Miyazaki, S. Kimura, and S. Hasegawa,
Spectroscopic evidence of a topological quantum phase transition in ul-
trathin Bi2 Se3 films Physical Review B 81, 165432 (2010).

[92] Y. Zhang, K. He, C.-Z. Chang, C.-L. Song, L.-L. Wang, X. Chen, J.-F.
Jia, Z. Fang, X. Dai, W.-Y. Shan, S.-Q. Shen, Q. Niu, X.-L. Qi, S.-C.
Zhang, X.-C. Ma, and Q.-K. Xue, Crossover of the three-dimensional
topological insulator Bi2 Se3 to the two-dimensional limit Nature Physics
6, 584 (2010).

[93] G. Zhang, H. Qin, J. Teng, J. Guo, Q. Guo, X. Dai, Z. Fang, and K.

Wu, Quintuple-layer epitaxy of thin films of topological insulator Bi2 Se3
Applied Physics Letters 95, 053114 (2009).

[94] P. Janicek, C. Drasar, L. Benes, and P. Lostak, Thermoelectric properties

of Tl-doped Bi2 Se3 single crystals Crystal Research and Technology 44,
505 (2009).

[95] H. Chi, W. Liu, K. Sun, X. Su, G. Wang, P. Lostak, V. Kucek, C. Drasar,

and C. Uher, Low-temperature transport properties of Tl-doped Bi2 Te3
single crystals Physical Review B 88, 045202 (2013).

[96] K. Kuroki and R. Arita, Pudding Mold band drives large thermopower
in Nax CoO2 Journal of Physical Society of Japan 76, 083707 (2007).

[97] H.-B. Yang, S.-C. Wang, A. K. P. Sekharan, H. Matsui, S. Souma, T.

Sato, T. Takahashi, T. Takeuchi, J. C. Campuzano, R. Jin, B. C. Sales, D.
Mandrus, Z. Wang, and H. Ding, ARPES on Na0.6CoO2: Fermi Surface
and Unusual Band Dispersion Physical Review Letters 92, 246403 (2004).

[98] H.-B. Yang, Z.-H. Pan, A. K. P. Sekharan, T. Sato, S. Souma, T. Taka-

hashi, R. Jin, B. C. Sales, D. Mandrus, A. V. Fedorov, Z. Wang, and
H. Ding,Fermi Surface Evolution and Luttinger Theorem in Nax CoO2 :
A Systematic Photoemission Study Physical Review Letters 95, 146401
(2005).

[99] T. Arakane, T. Sato, T. Takahashi, T. Fujii, and A. Asamitsu,

Angle-resolved photoemission study of the doping evolution of a three-
dimensional Fermi surface in Nax CoO2 New Journal of Physics 13, 043021
(2011).

[100] P. R. Emtage, Band structure of platinum antimonide Physical Review

138, A246 (1965).

[101] Y. Nishikubo, S. Nakano, K. Kudo, and M. Nohara, Enhanced thermo-

electric properties by Ir doping of PtSb2 with pyrite structure Applied
Physics Letters 100, 252104 (2012).

[102] D. H. Damon, R. C. Miller, and A. Sagar, Semiconducting properties of

PtSb2 Physical Review 138, A636 (1965).
 103
[103] D. H. Damon, R. C. Miller, and P. R. Emtage, Valence-band structure of
PtSb2 Physical Review B 5, 2175 (1972).

[104] C. T. Elliott and S. E. R. Hiscocks, On the preparation and electrical

properties of PtSb2 Journal of Materials Science 3, 174 (1968).

[105] A. A. Abdullaev, L. A. Angelova, V. K. Kuznetsov, A. B. Ormont, and Y.

I. Pashintsev, Galvariomagnetic phenomena in Platinum Antimonide
Physica Status Solidi (a) 21, 339 (1974).

[106] A. Dargys and J. Kundrotas, Electrical properties of narrow gap semicon-

ductor PtSb2 Journal of Physics and Chemistry of Solids 44, 261 (1983).

[107] R. A. Laudise, W. A. Sunder, R. L. Barns, G. W. Kammlott, A. F. Witt,

and D. J. Carlson, Czochralski crystal growth in the system PtMnx Sb2x
Journal of Crystal Growth 102, 21 (1990).

[108] P. M. Nikolic, S. S. Vujatovic, D. M. Todorovic, M. B. Miletic, A. Gol-

ubovic, A. I. Bojicic, F. Kermendi, S. Duric, K. T. Radulovic, and J.
Elazar, Thermal and electronic transport properties of single crystal
PtSb2 obtained by the photoacoustic method Japanese Journal of Ap-
plied Physics 36, 1006 (1997).

[109] M. Sndergaard, M. Christensen, L. Bjerg, K. A. Borup, P. Sun, F.

Steglich, and B. B. Iversen, Investigation of the correlation between stoi-
chiometry and thermoelectric properties in a PtSb2 single crystal Dalton
Trans. 41, 1278 (2012).

[110] G. A. Slack, in CRC Handbook of Thermoelectrics, ed. D. M. Rowe, CRC

Press, Boca Raton, page 407 (1995).

[111] S. Cottenier, DFT and the Family of (L)APW-method (2004).

[112] P. Hohenberg and W. Kohn, Inhomogeneous Electron Gas Physical Re-

view 136, 864 (1964).

[113] W. Kohn and L. J. Sham, Self-Consistent Equations Including Exchange

and Correlation Effects Physical Review 140, 1133 (1965).

[114] K. Schwarz and P. Blaha, Solid state calculations using WIEN2k Com-
putational Material Science 28, 259 (2003).

[115] P. Blaha, K. Schwarz, G. Madsen, D. Kvasicka, and J. Luitz, WIEN2k, An

Augmented Plane Wave + Local Orbitals Program for Calculating Crystal
Properties (TU Vienna, Vienna, 2001).

[116] E. Sjostedt, L. Nordstrom, and D. J. Singh, An alternative way of lin-

earizing the augmented plane wave method Solid State Communication
114, 15 (2000).
 104
[117] J. P. Perdew and W. Yue, Accurate and simple density functional for the
electronic exchange energy: Generalized gradient approximation Physical
Review B 33, 8800(R) (1986).

[118] J. P. Perdew and Y. Wang, Accurate and simple analytic representation of

the electron-gas correlation energy Physical Review B 45, 13244 (1992).

[119] J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized gradient approx-

imation made simple Physical Review Letters 77, 3865 (1996).

[120] Z. Wu and R. E. Cohen, More accurate generalized gradient approxima-

tion for solids Physical Review B 73, 235116 (2006).

[121] J. P. Perdew, S. Kurth, J. Zupan, and P. Blaha, Accurate density func-

tional with correct formal properties: A step beyond the generalized gra-
dient approximation Physical Review Letter 82, 2544 (1999).

[122] J. Tao, J. P. Perdew, V. Starovenov, and G. Scuseria, Climbing the den-

sity functional ladder: Nonempirical meta-generalized gradient approxima-
tion designed for molecules and solids Physical Review Letter 91, 146401
(2003).

[123] V. Anisimov, F. Aryasetiawan, and A. Lichtenstein, First-principles cal-

culations of the electronic structure and spectra of strongly correlated sys-
tems: the LDA+U method Journal of Physics: Condensed Matter 9, 767
(1997).

[124] F. Tran and P. Blaha, Accurate bandgaps of semiconductors and insu-

lators with a semilocal exchange-correlation potential Physical Review
Letter 102, 226401 (2009).

[125] F. Tran and P. Blaha, Implementation of screened hybrid functionals

based on the Yukawa potential within the LAPW basis set Physical Re-
view B 83, 235118 (2011).

[126] S. Grimme, Semiempirical GGA-type density functional constructed with

a long-range dispersion correction Journal of Computational Chemistry
27, 1787 (2006).

[127] P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni,

D. Ceresoli, G. L. Chiarotti, M. Cococcioni, I. Dabo, A. Dal Corso, S. de
Gironcoli, S. Fabris, G. Fratesi, R. Gebauer, U. Gerstmann, C. Gougoussis,
A. Kokalj, M. Lazzeri, L. Martin-Samos, N. Marzari, F. Mauri, R. Maz-
zarello, S. Paolini, A. Pasquarello, L. Paulatto, C. Sbraccia, S. Scandolo,
G. Sclauzero, A. P. Seitsonen, A. Smogunov, P. Umari, and R. M. Wentz-
covitch, QUANTUM ESPRESSO: a modular and open-source software
project for quantum simulations of materials Journal of Physics: Con-
densed Matter 21, 395502 (2009).

[128] J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized gradient approx-

imation made simple Physical Review Letters 77, 3865 (1996).
 105
[129] T. J. Scheidemantel, C. Ambrosch-Draxl, T. Thonhauser, J. V. Badding,
and J. O. Sofo, Transport coefficients from first-principles calculations
Physical Review B 68, 125210 (2003).

[130] G. K. H. Madsen and D. J. Singh, BoltzTraP. A code for calculating

band-structure dependent quantities Computer Physics Communications
175, 67 (2006).

[131] K. Yubuta, S. Shibasaki, I. Terasaki, and T. Kajitani, High-resolution

electron microscopy for incommensurate structure of Kx RhO2 crystal
Philosophical Magazine 89, 2813 (2009).

[132] A. L. Hector, W. Levason, and M. T. Weller, Structure of -SrRh2 O4

from X-ray and neutron powder diffraction European Journal of Solid
State and Inorganic Chemistry 35, 679 (1998).

[133] Y. Okamoto, M. Nohara, F. Akai, and H. Takagi, Correlated metallic

phase in a d oped band insulator Sr1x Rh2 O4 Journal of the Physical
Society of Japan 75, 023704 (2006).

[134] J. Sugiyama, H. Nozaki, Y. Ikedo, K. Mukai, J. H. Brewer, E. J. Ansaldo,

G. D. Morris, D. Andreica, A. Amato, T. Fujii, and A. Asamitsu, Static
magnetic order in metallic K0.49 CoO2 Physcal Review Letters 96, 037206
(2006).

[135] K.-W. Lee, J. Kunes, R. T. Scalettar, and W. E. Pickett, Correlation

effects in the triangular lattice single-band system Lix NbO2 Physical Re-
view B 76, 144513 (2007).

[136] K. Knzek, J. Hejtmanek, M. Marysko, E. Santava, Z. Jirak, J. Bursk,

K. Kirakci, P. Beran, Structure and properties of a novel cobaltate
La0.3 CoO2 Journal of Solid State Chemistry 184, 2231 (2011).

[137] M. D. Johannes, I. I. Mazin, and D. J. Singh, Optical properties and

correlation effects in Nax CoO2 Physical Review B 71, 205103 (2005).

[138] N. Kaurav, K. K. Wu, Y. K. Kuo, G. J. Shu, F. C. Chou, Seebeck co-

efficient of Nax CoO2 : Measurements and a narrow-band model Physical
Review B 79, 075105 (2009).

[139] A. Satake, H. Tanaka, T. Ohkawa, T. Fujii, and I. Terasaki, Thermal

conductivity of the thermoelectric layered cobalt oxides measured by the
Harman method Journal of Applied Physics 96, 931 (2004).

[140] S. Baroni, S. de Gironcoli, A. Dal Corso, and P. Giannozzi, Phonons and

related properties of extended systems from density-functional perturba-
tion theory Reviews of Modern Physics 73, 515 (2001).

[141] G. K. H. Madsen, K. Schwarz, P. Blaha, and D. J. Singh, Electronic struc-

ture and transport in type-I and type-VIII clathrates containing strontium,
barium, and europium Physical Review B 68, 125212 (2003).
 106
[142] L. Zhang, M.-H. Du, and D. J. Singh, Zintl-phase compounds with SnSb4
tetrahedral anions: Electronic structure and thermoelectric properties
Physical Review B 81, 075117(2010).

[143] K. P. Ong, D. J. Singh, and P. Wu, Analysis of the thermoelectric prop-

erties of n-type ZnO Physical Review B 83, 115110 (2011).

[144] Y. Saeed, N. Singh, D. Parker, and U. Schwingenschlogl, Thermoelectric

performance of electron and hole doped PtSb2 Journal of Applied Physics
113, 163706 (2013).

[145] B. Amin, N. Singh, T. M. Tritt, H. N. Alshareef, and U. Schwingenschlogl,

Major enhancement of the thermoelectric performance in Pr/Nb-doped
SrTiO3 under strain Applied Physics Letters 103, 031907 (2013).

[146] W. Liu, X. Peng, X. Wei, H. Yang, G. M. Stocks, and J. Zhong, Surface

and substrate induced effects on thin films of the topological insulators
Bi2 Se3 and Bi2 Te3 Physical Review B 87, 205315 (2013).

[147] J. Zhang, Z. Peng, A. Soni, Y. Zhao, Y. Xiong, B. Peng, J. Wang, M.

S. Dresselhaus, Q. Xiong, Raman spectroscopy of few-quintuple layer
topological insulator Bi2 Se3 nanoplatelets Nano Letters 11, 2407 (2011).

[148] J. Maassen and M. Lundstrom, A computational study of the thermoelec-

tric performance of ultrathin Bi2 Te3 films Applied Physics Letters 102,
093103 (2013).

[149] T. Thonhauser, G. S. Jeon, G. D. Mahan, and J. O. Sofo, Stress-induced

defects in Sb2 Te3 Physical Review B 68, 205207 (2003).

[150] J. Black, E. M. Conwell, L. Seigle, and C. W. Spencer, Electrical and

optical properties of some MV2 A NV3 IA semiconductors Journal of Physics
and Chemistry of Solids 2, 240 (1957).

[151] E. I. Rashba, Properties of semiconductors with an extremum loop .1.

Cyclotron and combinational resonance in a magnetic field perpendicular
to the plane of the loop Soviet Physics Solid State 2, 1109 (1960).

[152] C. R. Ast, J. Henk, A. Ernst, L. Moreschini, M. C. Falub, D. Pacile, P.

Bruno, K. Kern, and M. Grioni, Giant spin splitting through surface
alloying Physical Review Letters 98, 186807 (2007).

[153] S. Datta and B. Das, Electronic analog of the electrooptic modulator

Applied Physics Letters 56, 665 (1990).

[154] K. Ishizaka, M. S. Bahramy, H. Murakawa, M. Sakano, T. Shimojima, T.

Sonobe, K. Koizumi, S. Shin, H. Miyahara, A. Kimura, K. Miyamoto, T.
Okuda, H. Namatame, M. Taniguchi, R. Arita, N. Nagaosa, K. Kobayashi,
Y. Murakami, R. Kumai, Y. Kaneko, Y. Onose, and Y. Tokura, Giant
Rashba-type spin splitting in bulk BiTeI Nature Material 10, 521 (2011).
 107
[155] M. Bahramy, B.-J. Yang, R. Arita, and N. Nagaosa, Emergence of non-
centrosymmetric topological insulating phase in BiTeI under pressure Na-
ture Communication 3, 679 (2012).

[156] T. Caillat, M. Carle, P. Pierrat, H. Scherrer, and S. Scherrer, Thermo-

electric properties of (Bix Sb1x )2 Te3 single crystal solid solutions grown by
the T.H.M. method Journal of Physics and Chemistry of Solids 53, 1121
(1992).

[157] K. Kadel, L. Kumari, W.-Z. Li, J. Y. Huang, and P. P. Provencio, Syn-

thesis and Thermoelectric Properties of Bi2 Se3 Nanostructures Nanoscale
Research Letter 6, 57 (2011).

[158] N. J. Ramer and A. M. Rappe, Virtual-crystal approximation that works:

Locating a compositional phase boundary in Pb(Zr1x Tix )O3 Physical
Review B 62, R743 (2000) and references therein.

[159] L. Chaput, P. Pecheur, J. Tobola, and H. Scherrer, Transport in doped

skutterudites: Ab initio electronic structure calculations Physical Review
B 72,085126 (2005).

[160] L. Jodin, J. Tobola, P. Pecheur, H. Scherrer, and S. Kaprzyk, Effect of

substitutions and defects in half-Heusler FeVSb studied by electron trans-
port measurements and KKR-CPA electronic structure calculations Phys-
ical Review B 70, 184207 (2004).

[161] N. E. Brese and H. G. von Schnering, Bonding trends in pyrites and

a reinvestigation of the structures of PdAs2 , PdSb2 , PtSb2 and PtBi2
Zeitschrift fur anorganische und allgemeine Chemie 620, 393 (1994).

[162] R. A. Reynolds, M. J. Brau and R. A. Chapman, Preparation and semi-

conducting properties of platinum antimonide Journal of Physics and
Chemistry of Solids 29, 755 (1968).

[163] K. Mori, H. Usui, H. Sakakibara, and K. Kuroki, Corrugated flat band

as an origin of large thermopower in hole doped PtSb2 AIP Advances 2,
042108 (2012).

[164] Y. Saeed, N. Singh, and U. Schwingenschlogl, Colossal thermoelectric

powerfactor in K7/8 RhO2 Advanced Functional Materials 22, 2792 (2012).

[165] D. Parker and D. J. Singh, High-temperature thermoelectric performance

of heavily doped PbSe Physical Review B 82, 035204 (2010).
 108

PUBLICATIONS

[P1] Y. Saeed, N. Singh, and U. Schwingenschlogl, Thickness and strain effects

on the thermoelectric transport in nanostructured Bi2 Se3 . Applied Physics
Letters 104, 033105 (2014).

[P2] Y. Saeed, N. Singh, and U. Schwingenschlogl, Superior thermoelectric re-

sponse in the 3R phases of hydrated Nax RhO2 Scientific Reports 4, 4390
(2014).

[P3] Y. Saeed, K. Zhao, N. Singh, R. Li, J. E. Anthony, A. Amassian, and

U. Schwingenschlogl, Influence of substitution on the optical properties of
functionalized pentacene monomers and crystals: Experiment and theory
Chemical Physics Letters 585, 95 (2013).

[P4] A. Useinov, Y. Saeed, N. Singh, N. Useinov, and U. Schwingenschlogl, Im-

pact of lattice strain on tunnel magnetoresistance in Fe/Insulator/Fe and
Fe/Insulator/La0.67 Sr0.33 MnO3 magnetic tunnel junctions Physics Review B
88, 060405(R) (2013).

[P5] Y. Saeed, N. Singh, D. Parker, and U. Schwingenschlogl, Thermoelectric

performance in electron and hole doped PtSb2 Journal of Applied Physics
113, 163706 (2013).

[P6] Y. Saeed, N. Singh, and U. Schwingenschlogl, Half-metallicity and giant

magneto-optical Kerr effect in N-doped NaTaO3 Chemical Physics Letters
546, 96 (2012).

[P7] Y. Saeed, N. Singh, and U. Schwingenschlogl, Colossal thermoelectric pow-

erfactor in K7/8 RhO2 Advanced Functional Materials 22, 2792 (2012).

[P8] Y. Saeed, N. Singh, and U. Schwingenschlogl, First principles results on

TbAl3 (BO3 )4 : A promising magneto-optical material Journal of Applied
Physics 110, 103512 (2011).