n. ..

.. 1: 0A .. .

.5. yam...
2 it .1 32"
I1. 1:1. .‘Itn\,!.-
muiujirég

3.3.3

J... 1...:
.22 r $ . . n .
.9 h:

.

(a... .
3 {if E}!
Jpn-3.,
. 8.?

 

. “aura”. 5;!
:I‘P’n...‘ nvllul -
.

323511,;
9333!...

II. .1: u.)
.0) J .

1.3.1.5... v .y.
. I“... S ‘ , . 1 y. 1..
$33.3 .

1.... . ..J..1 .nih v.91.
. v.3. . .3 , , wfimxmg infirm
....: r... an , .

 

 

msls LI ":3 HA HY
3 Michigan State
5‘ OM University

 

This is to certify that the
thesis entitled

SUBSTITUTION STUDIES IN CstmmaTesfln AND NanmeTem+2
FOR THERMOELECTRIC APPLICATIONS

presented by

AURELIE GUEGUEN

has been accepted towards fulfillment
of the requirements for the

doctoral degree in Chemistry

 

 

Wfiw

Major Professor’s Signa

42/30/ /02

Date

 

MSU is an Afinnative ActiorVEqual Opportunity Employer

 

 

PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Aoc&Pres/ClRC/DateDue.indd

 

SUBSTITUTION STUDIES IN CstmBi3Tes+m AND NanmeTem+2 FOR
THERMOELECTRIC APPLICATIONS

By

Aurelie Guéguen

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Chemistry

2009

ABSTRACT

SUBSTITUTION STUDIES IN CstmBi3,Te5+m AND NanmeTem+2 FOR

THERMOELECTRIC APPLICATIONS
By
Aurélie Guéguen

Therrnoeletric devices convert thermal energy into electrical power or vice versa.
If materials with thermoelectric figure of merit 2 3 can be synthesized, thermoelectric
devices will not be restricted to niche applications such as spacecrafi power generation
and will have a significant impact on the economy. Power generation from waste heat
recovery would be an alternative energy resource whereas electronic cooling would
significantly increase computing processor speed. Good thermoelectric materials require

high electrical conductivity, large therrnopower and low thermal conductivity. Complex
chalcogenides materials, such as CsBi4Te6, exhibit promising thermoelectric properties.
It consists of Cs+ cations weakly bound to [Bi4Te6]’ layers. The Cs+ cations act as rattlers,
hence contributing to a reduction of the lattice thermal conductivity. Partial substitution
of Bi by Pb resulted in the discovery of the homologous family CstmBi3Te5+m. Further
substitution studies both on the Cs and Bi sites lead to the synthesis of the new

compounds CSo.76Ko.74Bi3.5T66 (1), CsNao.9sBi4.01Te7 (2), €50.69C30.6SBi3.34T36 (3),

Rbo.82Pb0.szBi3.18T66 (4), Rbo.l9K1.3iBi3.50T66 (5), RbSnBisTeé (6). Rbo.94Cao.94Bi3.06T€6

(7), RbeBi3Te6 (8) and KSnSb3Te6 (9). However synthesis conditions could not yet be
optimized to prepared pure phase with good quality crystal.

Doped PbTe is used in commercial applications. Recent studies on preparation of
bulk nanostructured PbTe have shown significant improvement in the thermoelectric
figure of merit, achieved mainly through reduction of lattice thermal conductivity. The
observation by TEM of nanoprecipitats embedded in the PbTe matrix is believed to be

the origin of their low thermal conductivity. A figure of merit ZT ~ 1.6 at 650 K was

reported for example for the p-type Nal-bemeyTem+2 (SALT) system. Partial

substitution of Pb by Sn was studied through the synthesis of the Nan13-xSnbeTe20

series and increases the carrier concentration. As a result, an increase in electrical
conductivity and a decrease in therrnopower were observed with increasing amount of
Sn. Such substitution did not affect the formation of nanoprecipitates, which were
observed with TEM along with lamellar features. The substitutions did not result in

higher figure of merit. Replacing Sb by Bi produced weaker samples. The K analog
series, KPb13-xSnbeTe20, was prepared as well and resulted in weaker and more water-
sensitive specimens. These different studies show that he best p-type materials so far are

Ag(Pb1-ySny)meTe2+m (LASTT) and Na|_bemeyTem+2 (SALT) systems. Substituting

Ag by Cu in the n-type AngmeTem+2 (LAST) system resulted in phase segregation of

CuzTe and szTe3 in the PbTe matrix. The figure of merit of the specimens was lower

than that of LAST.

 

ACKNOWLEDGMENTS

First, I would like to thank my advisor, Prof. Mercouri G. Kanatzidis, for giving
me the opportunity to join his lab and learn about solid state chemistry and thermoelectric
materials.

Then, I would like to thank he Kanatzidis group, previous and current members,
for their help, support and helpful discussions. I would like to thank particularly Joseph
Sootsman and Melanie Francesco for their help and friendship.

I would like also to thank our collaborators. The Hall effect measurements were
done at University of Michigan in collaboration with Prof. Ctirad Uher, Huijun Kong,
Steven Moses and Chang-Pen Li. SEM studies were carried out using the facilitities at
the Center of Advanced Microscopy at Michigan State University and at the NUANCE
center at Northwestern University. Part of the TEM studies was done by Robert Pcionek t
the Center of Advanced Microscopy at Michigan State University. The rest of the TEM
work was done by Dr. Jiaqing He and Prof. Vinayak Dravid at the NUANCE center at
Northwestern University.

Finally, I would like to thank my family and friends for their support. I met great

friends during my stay at the United-States.

iv

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... xi
LIST OF FIGURES ....................................................................................................... xiii
CHAPTER 1. Introduction to T hermoelectrics ................................................................ 1
1.1 Thermoelectric concepts .................................................................................. 1

1.2 Optimization of the figure of merit ................................................................. 5

1.3. Bulk thermoelectric materials ......................................................................... 8

1.4. Complex chalcogenide systems as potential thermoelectric materials ......... 11

1.5. Low-dimensional and nanostructured materials ........................................... 16
CHAPTER 2. Substitutions in the CstmBi 3 Te5+,,, System ........................................... 27
2.1 Introduction ................................................................................................... 27

2.2 Experimental section ..................................................................................... 29

2.2.1 Synthesis ........................................................................................... 29

2.2.2 Characterization techniques .............................................................. 33

2.3. Results and discussion .................................................................................. 35

2.3.1 Structure description ......................................................................... 35

2.3.2 Characterizations ............................................................................... 44

2.3.3 Preliminary thermoelectric results .................................................... 53

2.4. Conclusions .................................................................................................. 54

CHAPTER 3. Preparation and Characterization of Members of the Series

NanmeTem.) with m=6, 8, 12 57
3.1 Introduction ................................................................................................... 57
3.2 Experimental section ..................................................................................... 59

3.2.1 Synthesis ........................................................................................... 59
0 Method 1 ..................................................................................... 59
0 Method 2 ..................................................................................... 60
0 Method 3 ..................................................................................... 60
0 Method 4 ..................................................................................... 6O
0 Method 5 ..................................................................................... 6O
0 Method 6 ..................................................................................... 61
0 Method 7 ..................................................................................... 61

0 Method 8 ..................................................................................... 61

0 Method 9 ..................................................................................... 61

3.2.2 Characterization techniques .............................................................. 62

3.3. Results and discussion .................................................................................. 65
3.3.1 Method 1 ........................................................................................... 65

3.3.2 Method 2 ........................................................................................... 81

3.3.3 Method 3 ........................................................................................... 83

3.3.3 Method 4 ........................................................................................... 84

3.3.3 Method 5 ........................................................................................... 85

3.3.3 Method 6 ........................................................................................... 86

3.3.3 Method 7 ........................................................................................... 89

3.3.3 Method 8 ........................................................................................... 96

3.3.3 Method 9 ........................................................................................... 97

3.4. Concluding remarks ...................................................................................... 99

CHAPTER 4. Thermoelectric Properties of the Nanostructured Materials Nan13.

xSanTezo (M=Sb, Bi) .................................................................................................... 103
4.1 Introduction ................................................................................................. 103
4.2 Experimental section ................................................................................... 105

4.2.1 Synthesis ......................................................................................... 105

4.2.2 Characterization techniques ............................................................ 106

4.3. Results and discussion ................................................................................ 109
0 Structure and characterization ................................................... 109

0 Scanning electron microscopy .................................................. 115

0 Electronic transport properties .................................................. 119

0 Optical spectroscopy ................................................................. 128

0 Thermal transport properties ..................................................... 131

0 High resolution transmission electron microscopy ................... 137

4.4. Conclusions ................................................................................................ 144

CHAPTER 5. Synthesis and Characterization of the KPb13.xSnbeTe20 (x=0, 2, 5, 9,

I3, 16 and 18) series ........................................................................................................ 149
5.1 Introduction 149
5.2 Experimental section ................................................................................... 152

5.2.1 Synthesis ......................................................................................... 152

5.2.2 Characterization techniques ............................................................ 154

5.3. Results and discussion ................................................................................ 156
0 Structure and characterization ................................................... 156

0 Scanning electron microscopy .................................................. 158

0 Optical band gap measurements ............................................... 163

0 Electronic transport properties .................................................. 163

0 Thermal transport properties ..................................................... 171

5.4. Conclusions ................................................................................................ 174

vi

CHAPTER 6. The systems CuPbmeTem+2 (m=8, 12, 18, 20, 22, 30, 40, 50), CuPblg.

xSnbeTezo (x=0, 5, 9, 13, 18), CuzTe/szTe3 and CuzTe/PbTe ................................... 177
6.1 . Introduction ................................................................................................ 177
6.2. CuPbmeTem+2 (m = 8, 12, 18, 20, 22, 30, 40 and 50) .............................. 179

6.2.1. Experimental section ...................................................................... 179
6.2.1.1. Synthesis ......................................................................... 179

6.2. 1 .2. Characterization techniques ............................................ 180

6.2.2. Results and discussion ................................................................... 182

0 Structure and characterization ................................................... 182

0 Scanning electron microscopy .................................................. 183

0 Thermal analysis ....................................................................... 185

0 Electronic transport properties .................................................. 188

0 Thermal transport properties ..................................................... 189

6.3. CuPb13-xSnbeTe20 (x=0, 5, 9, 13, 18) ....................................................... 192
6.3.1. Experimental section ...................................................................... 192
6.3.2. Results and discussion ................................................................... 193

0 Structure and characterization ................................................... 193

0 Scanning Electron Microscopy ................................................. 193

0 Electronic transport properties .................................................. 198

0 Thermal transport properties ..................................................... 199

6.4. Investigation of the system Cu/Sb/Te ......................................................... 201
6.4.1. Motivation ...................................................................................... 201
6.4.2. Experimental section ...................................................................... 202
6.4.3. Results and discussion ................................................................... 202

0 Structure and characterization ................................................... 202

0 Scanning electron microscopy .................................................. 203

0 DTA analysis ............................................................................. 208

6.5. Doping studies of PbTe with CuzTe ........................................................... 208
6.5.1. Motivation ...................................................................................... 208

6.5.2. Experimental section ...................................................................... 209
6.5.3. Results and discussion ................................................. _. ................. 209

0 Structure and characterization ................................................... 209

0 Scanning electron microscopy .................................................. 209

0 DTA analysis ............................................................................. 213

0 Electronic transport properties .................................................. 214

6.6. Conclusions ................................................................................................ 21 5
CHAPTER 6. Conclusions and Future Directions ...................................................... 219

vii

Table 2-1

Table 2-2

Table 2-3

Table 2-4

Table 2-5

Table 2-6

Table 2-7

Table 2-8

Table 2-9

Table 2-10

Table 3-1

Table 3-2

Table 3-3

Table 3-4

LIST OF TABLES

Ratio and quantity of elements mixed for (1), (2), (3), (4), (5), (6),
(7), (8) and (9). ........................................................................................... 30

Summary of the crystallographic data for (l), (2) (3), (4) and (5) ............. 38

Summary of the crystallographic data for RbSnBi3Te6 (6),
Rb0_94Ca0_94Bi3_06Te6 (7), RbeBi3Te6 (8) and KSnSb3Te6 (9) ............... 39

Metal-tellurium and alkali site-tellurium bond distances (A) in
compounds (1), (3), (4), (5), (6), (7) and (8) .............................................. 40

Metal-Te and alkali metal-Tel bond distances (A) in compound (2) ........ 41
Metal-Te and alkali metal-Te bond distances (A) in compound (9) .......... 41

Atomic coordinates and equivalent isotropic displacement

parameters (A2 x 103) of the metal and alkali metal positions for
(1), (3), (4), (5), (6), (7) and (8). U(eq) is defined as one third of
the traces of the orthogonalized Uij tensors. .............................................. 42

Atomic coordinates and equivalent isotropic displacement
parameters (A2 x 103) for (2). U(eq) is defined as one third of the
traces of the orthogonalized Uij tensors. .................................................... 43

Atomic coordinates and equivalent isotropic displacement
parameters (A2 x 103) for (9). U(eq) is defined as one third of the
traces of the orthogonalized Uij tensors. .................................................... 43

Preliminary thermoelectric measurements on the compounds (1)
and (4) ........................................................................................................ 53

Concentrations of the standards used for the ICP-AES calibration ........... 63
Melting and crystallization points for Nangstem+2 (m=6, 8, 12) .......... 65

Summary of the elemental concentrations of Na and Sb obtained
with ICP-AES ............................................................................................ 70

Comparison between experimental and theoretical mass A
percentages of Na and Sb in the powders analyzed with ICP-AES ........... 71

viii

 

 

Table 4—1

Table 4-2

Table 4-3

Table 4-4

Table 4-5
Table 4-6

Table 5-1

Table 5-2

Table 5-3

Table 5-4

Table 6-1

Table 6-2

Amounts of elements used to prepare Nan]g_xSnbeTe20 ..................... 106

Amounts of elements used to prepare Nanlg-xSaniTe20 ...................... 106

Summary of the physical and electronic properties of the materials

Nan l 8-xsnbeTezo .................................................................................. 1 30
Summary of the physical and electronic properties of the materials
Nan18-xsaniTezo .................................................................................. 1 30
Total and lattice thermal conductivities of Nanlg-xSnbeTe20
(x=0, 3, 5, 9, l3 and 16) at 300 and 550 K .............................................. 136
Total and lattice thermal conductivities of Nan13-xSaniTe20
(x=0, 3, 5, 9 and 16) at 300 and 550 K .................................................... 136
Amounts of elements used to prepare KPb18_xSnbeTe20 ....................... 153
Amounts of elements used to prepare KPb18-xSaniTe20 ........................ 154

Summary of the physical and electronic properties of the materials

KPb13_xSnbeTe20 .................................................................................... 170
Summary of the physical and electronic properties of the materials

KPb13-xSaniTezo .................................................................................... I70
Amounts of elements used to prepare the series CuPbmeTem+2 ............ 180

Amount of elements used for to prepare the series CuPb13_
xSnbeTego ............................................................................................... I 92

ix

LIST OF FIGURES

Figure 1-1 Scheme of (a) a cooling device and (b) a generator ..................................... 2

Figure 1-2 Figure of merit ZT shown as a function of temperature for several
bulk materials ............................................................................................... 5

Figure 1-3 Evolution of the Seebeck coefficient and the electrical conductivity
as a function of the carrier concentration ..................................................... 6

Figure 1-4 Density of states for electrons in bulk semiconductors (3D),
quantum wells (2D), quantum wires (1D) and quantum dots (0D) ............. 7

Figure 1-5 Structure of a skutterutide antimonide ......................................................... 9
Figure 1-6 Various building blocks (shaded) based on different “cuts” of the
NaCl-type structure. The diagram is view down with [011] plane.

Black and white circles are bismuth and chalcogen atoms,
respectively ................................................................................................ 1 3

Figure 1—7 Perspective view of the CsBi4Te6 along the b-axis ................................... 14

Figure 1-8 The structures of (a) CstBi3Te6, (b) CstzBi3Te7, (c)
CSPb3BI3TCg and (d) Cst4Bi3Teg in projection down the c axis

for (a) and (c) and the 3 axis for (b) and (d) ............................................... 16
Figure 2-1 Structure type of (a) A]-XM4Te6 along the c axis and (b) of

CsNa0_9gBi4,01Te7 along the a-axis ............................................................ 36
Figure 2-2 Crystal structure of KSnSb3Te6 viewed along the b-axis .......................... 37

Figure 2-3 (a) Comparison between the calculated and experimental powder X-
ray diffraction patterns of compound (1) and (b) DTA results for
compound (1) ............................................................................................. 44

Figure 2-4 Comparison between the powder X-ray diffraction patterns of
ground samples taken from top and bottom of the ingot with

nominal composition €50.76K0_74Bi3_5Te6 prepared in the vertical
furnace, the calculated powder pattern and that of CsBi4Te6 .................... 46

Figure 2-5 DTA results on powders taken from (a) the top and (b) the bottom
of the material Cso.6Ko,9Bi3_5Te6 prepared in a vertical furnace ................ 47

Figure 2-6

Figure 2-7

Figure 2-8

Figure 2-9

Figure 2-10

Figure 3-1

Figure 3-2

Figure 3-3

Figure 3-4

Figure 3-5

DTA results of the batch from which the compound

Cso_69Ca0,65Bi3.34Te6 was synthesized ....................................................... 48

Picture of the ingot with composition RbagszquBillgTeG prepared

with the Bridgman technique ..................................................................... 49

(3) X-ray diffraction patterns of powders from top, middle and
bottom of the ingot with composition (4) prepared by Bridgman
technique and (b) DTA analysis of the powder from the middle of

the ingot ...................................................................................................

(a) Comparison between the experimental and calculated powder
X-ray diffraction patterns for compound (6) and that of BiTe and

(b) DTA analysis of the powder from the middle of the ingot ................

SEM image of the needle with composition KSnSbgTeé.

Powder X-ray diffraction from samples from top and bottom part
of the ingots with composition (a) Nan6SbTe3, (b) the area
between 20 = 80 and 110 deg is enlarged to show peak splitting;
(c) NanSSbTelo, the area between 20 = 60 and 80 deg is enlarged
to show peak splitting; and (e) NannSbTeM, the area between 20
= 80 and 110 deg is enlarged to show peak splitting. All three

specimens were prepared with method 1 .................................................

Thermal analysis for (a) Nan6SbTe8, (b) Nangstelo and(c)

NanlzsteM prepared with method 1 ...................................................

BSE inages of one of the surface of the sample NannSbTeM used
for thermoelectric characterization. EDS analysis on the inclusions
indicated the regions to contain ~ 67 % Sb, ~ 22 Te and ~ 1% Pb.

The matrix is pure PbTe ..........................................................................

Diffuse reflectance absorption spectra of (a) Nan6SbTe3, (b)
Nangstem, (c) NanIZSbTeM and (d) Nanlgstezo prepared

with method 1 .........................................................................................

Temperature dependence of (a) the electrical conductivity, (b) the
therrnopower and (c) the power factor the compounds

NanmeTem+2 (m=6, 8, 12 and 18) and Nao.95PbZOSbTe22 prepared

with method 1 .........................................................................................

xi

..50

..51

52

..66

..67

...69

...72

...74

Figure 3-6

Figure 3-7

Figure 3-8

Figure 3-9

Figure 3-10

Figure 3-11

Figure 3-12

Figure 3-13

Figure 3-14

Figure 3-15
Figure 3-16

Figure 3-17

Figure 3-18

Comparison of (a) electrical conductivity and (b) therrnopower

collected during the first and second measurement for Nangstem
prepared with method 1 ............................................................................. 75

Temperature dependence of (a) electrical conductivity and (b)
therrnopower measured for 2 different samples with composition

NanngTelo prepared with method 1 ....................................................... 76

Temperature dependence of (a) the total and (b) lattice thermal
conductivity for the materials NanmeTem+2 (m=6, 8, 12, 18) and
Na0,95Pb208bTe22 prepared with method 1 ................................................ 78

Figure of merit ZT for NanmeTem+2 (m=6, 8, 12, 18) and

Na0_95szoSbTe22 prepared with method 1 ................................................ 79
Typical HRTEM images obtained for Nangstelo prepared with

method 1 ..................................................................................................... 80
Powder X-ray diffraction of Nangstem prepared by method 2 ............ 81

Comparison of (a) the electrical conductivity, (b) the therrnopower

and (c) the power factor of the compounds NanSSbTem prepared
by methods 1 and 2 .................................................................................... 82

Comparison of the (a) total and (b) lattice thermal conductivities for
samples prepared by methods 1 and 2 ....................................................... 83

Powder X-ray diffraction of Nangstelo prepared with method 3.
The area between 20 = 35 and 60 deg is enlarged to show the
presence of extra peaks can be assigned to NaSbTe2 ................................ 84

Powder X-ray diffraction of Nangstelo prepared with method 4 ......... 85
Powder X-ray diffraction of Nangstelo prepared with method 5 ......... 86

Powder X-ray diffraction patterns of (a) Nan7,6Sn0.4SbTe10, (b)
Nan6SnZSbTe10, (c) Nan4Sn4$bTe10 and (d) NanZSn6SbTe10 ............ 87

BSE images from areas close to (a) the top and (b) the bottom of the
ingot Nan4SmeTelo. (c) BSE image from 3 sample from the
middle of the sample NanZSn6SbTe10 ..................................................... 88

xii

Figure 3-19

Figure 3-20

Figure 3-21

Figure 3-22

Figure 3-23

Figure 3-24

Figure 3-25

Figure 3-26

Figure 3-27

Figure 4-1

Figure 4-2

Figure 4-3

Powder X-ray diffraction of powders from (a) Nao_5Pb88bTe10, (b)
Nao.5Pb85bo.75Tero, (C) Nao.5Pb8$bo.5Tero and (d)
Na0_5Pb38b0,25Tem ..................................................................................... 89

BSE images of areas from a part close to the top of the ingot
Nag-5Pngbo,75Te10 ..................................................................................... 91

BSE images of sample from (a) bottom and (b) middle of the ingot
Na0_5Pb3$bo.5Te10 ...................................................................................... 92

Temperature dependence of the (a) electrical conductivity, (b)
therrnopower and (0) power factor of Na0,5PbBSb0.25Te10 and
Nangste 10 .............................................................................................. 94

First and second sets of measurements for the (a) electrical
conductivity and (b) therrnopower of the compound

Na0_5Pb38b0_25Te 1 0 ..................................................................................... 95

Temperature dependence of the (a) electrical cconductivity and (b)
therrnopower of two different ingots with composition

Na0,5Pb38b0,25Telo ..................................................................................... 96

Powder X-ray diffraction patterns from powders from the
composition NanngTe10 using NazTe as a source of Na ....................... 96

Comparison between the X-ray diffraction patterns from powders
from the original ingot and the two pellets prepared from that ingot ........ 97

SEM images of the surface of (a) HP] and (2) HP2 .................................. 98

Powder X-ray diffraction patterns of Nanlg-xSanTe20 with x=5,
9, 13, 16 for (a) M=Sb and (b) M=Bi; variation of the unit cell
parameter as a function of x for (c) M=Sb and (d) M=Bi ........................ 1 ll

X-ray powder diffraction patterns of (a) Nm,ng13Sn5$bTe20, (b)

Nao_8Pb13Sn58bo_4Te20. For each composition, powder from both
top and bottom of the ingot was analyzed to check homogeneity
along the ingot. The small arrows indicate diffraction peaks that do
not belong to the PbTe structure-type and indicate the presence of

szTe3 as a minor phase .......................................................................... l 12

(a) Typical DTA results of the composition Nan5Sn13SbTe20, (b)
variation of the melting and crystallization points of Nan13.
xSnbeTezo as a function of x .................................................................. 1 l4

xiii

Figure 4-4

Figure 4-5

Figure 4-6

Figure 4-7

Figure 4—8

Figure 4-9

Figure 4-10

Figure 4-11

Figure 4-12

Infrared absorption spectra near the band edge of samples Nanlg-
xSnbeTezo with x=9, 13, 16 and 18 ........................................................ 115

BSE imaging of inclusions observed from (a) part close to the top
and (b) part close to the bottom of the ingot Nan9$n9SbTe20, (0)
middle part of Nan13Sn5SbTe20 ............................................................. 1 17

BSE imaging of samples from middle part of the ingots (a)
Nan13Sn5BiTe20 and (b)-(C) NanZSntiTezo ..................................... I 18

Electronic transport properties of the Nan13-xSnbeTe20 samples:
(a) electrical conductivity, (b) variation of the power law
dependence of the electrical conductivity (T' ) as a function of x
(the 1» parameters were extracted from the data of panel (a)), (c)
therrnopower, ((1) power factor. The point marks in the inset of
panel (a) identify all samples and apply to panels (c) and (d) ................. 121

Temperature dependence of (a) the carrier concentration and (b)
carrier mobility for Nan13Sn5MTe20 (M=Sb, Bi) .................................. 122

(a) Temperature dependence of the thermoelectric properties of the

Nan13Sn5SbTezo and Nag,ng13Sn5SbyTe20 (y=0.4, 0.6, 0.8, I)
compositions: (a) electrical conductivity, (b) thennopower, (c)
power factor. The point marks in the inset of panel (b) apply to plots
(3) and (c) ................................................................................................. 124

T Electronic transport properties of the Nan13_xSaniTe20 samples:
(a) electrical conductivity, (b) variation of the power law

dependence of the electrical conductivity (T')‘) as a function of x
(the 1» parameters were extracted from the data of panel (a)), (c)
therrnopower, (d) power factor. The point marks in the inset of
panel (a) identify all samples and apply to all panels (c) and (d) ............ 127

Specular reflectance spectra of (a) Nan13-xSnbeTe20 (x=3, 5, 9)
and (b) Nan18-xSaniTe20 (x=3, 5, 9) .................................................... 129

Temperature dependence of (a) the total thermal conductivity of

the Nan18_xSnbeTe20 compositions and (b) their lattice thermal
conductivity component (the electronic thermal conductivity was

estimated using the Wiedemann-Franz law with Lo=2.45.10'8
WQK'I); (c) the total thermal conductivity of the Nan13.

xSaniTezo compositions and ((1) their lattice thermal conductivity
component. 1 32

xiv

Figure 4-13

Figure 4-14

Figure 4-15

Figure 4-16

Figure 4-17

Figure 4-18

Figure 4-19

Figure 5-1

Temperature dependence of the thermal conductivity for the
compositions (a) Nan13SbTe20, (b) Nan9Sn9SbTe20 and (c)
Nan58n13SbTe20 .................................................................................... I 33

Temperature dependence of the figure of merit ZT for (a) Nanlg-
xSn,‘SbTe20 (x=0, 3, 5, 9, 13, 16) and (b) Nan13-xSaniTezo (x=0,
3, 5, 9 and 16) .......................................................................................... 135

Typical HRTEM images of (a) and (b) Nanl3SnsBiTe20 and (c)
and (d) Nan] 3Sn5SbTe20 ........................................................................ 138

(a) A multitude of precipitates were observed in low magnification

Z-contrast image for Nan13Sn5BiTe20; (b) high magnification
STEM image showing the weak contrast of some small
precipitates; (c) Line scanning profile of the selected atomic array
in (b), the positions that the dark arrows point out are Te, and those
marked by gray arrows mainly come from Pb ......................................... 139

(a) High magnification high resolution TEM image for

Nan138n5BITe20 and (b) enlarged part show the lattice parameters
of the precipitate and that of the matrix are different; (c) and (d)
FF T images of the two regions C and D in (a) ........................................ 140

(a) Different precipitate from Nan13Sn5BiTe20 with different
contrast lamella structure labeled as ABAB... (b-f) are strain
mapping for this precipitate; (b) power spectrum image in (a) with
g1: 002 and g2= 220. (c) and ((1) 4.2.1. show the strain map profile
along the 001 direction (8”,) and the shear direction(exy),
respectively .............................................................................................. 142

(a)-(d) Different magnification images for Nan13Sn5BiTe20 with
another type of lamellar structure. (a) only shows the saw shape at
the edge of sample; (b) further ion milling for sample, the lamella
profile appear in the region close to edge, (0) strong contrast was
observed in the area far from the edge; ((1) From the high resolution
image, the lamella was indexed as 112 direction, which is different
from the first type lamella in Figure 4-16 ................................................ 143

X-ray powder diffraction patterns of KPb13-xSanTe20 with x=0,
2, 5, 9, 13, 16 and 18 for (a) M=Sb and (c) M = Bi; variation ofthe
unit cell parameter as a function of x for (b) M = Sb and (d) M =
Bi .............................................................................................................. 157

XV

Figure 5-2 Typical DTA results of the composition KPblngTezo, (b) variation

of the melting and crystallization points of KPb18-xSnbeTe20 as a
function of x ............................................................................................. 158

Figure 5-3 BSE images of the surface of (a) KPb13Sn5SbTe20 showing a
precipitate with the atomic percentages Pb:2.73, K: 4.59, Sb:27.59
and Te:65.09 and (b) KPngngstezo showing an inclusion with
atomic percentages Pb:2.32, K: 14.87, Sb:23.38 and Te:59.43.
EDS analysis on the matrices indicated Pb, Sn and Te with ratios
consistent with the nominal composition ................................................. 160

Figure 5-4 BSE image of (a) a Sb-rich inclusion and (b) a mixed-phase
inclusion in KPbSSn13SbTe20. EDS on area 1 showed the presence

of szTe3 and EDS on area 2 gave the atomic percentages Pb: 6.32,
K: 3.80, Sb:13.51 and Te:76.37 ............................................................... 161

Figure 5-5 BSE images of inclusions (a) and (b) observed for KPngbgBiTezo.
Different phases are visible in the inclusion (a): area 1 is pure Bi
whereas area 2 contains Pb, Sn, Bi and Te with the following atomic
percentages: 8.14, 15.02, 32.17 and 44.67. EDS analysis on
inclusion (b) gave the atomic percentages Pb: 1.36, K: 4.77, Sn:
22.90 and Te: 70.95 ................................................................................. 162

Figure 5-6 Electronic transport properties of the KPb13-xSnbeTe20 samples:
(a) electrical conductivity, (b) therrnopower, (c) power factor. The
point marks in the inset of panel (b) identify all samples and apply
to panels (a) and (c) .................................................................................. 165

Figure 5-7 Electronic transport properties of the KPb13-xSaniTe20 samples:
(a) electrical conductivity, (b) therrnopower, (c) power factor. The
point marks in the inset of panel (a) identify all samples and apply
to all panels (b) and (c) ............................................................................ 167

Figure 5-8 Band-structure of PbTe doped with (a) (Ag, Sb), (b) (Ag, Bi). Ag
and Sb/Bi atoms in the pair are the second-nearest neighbors of one
another in a 64-atom supercell ................................................................. 169

Figure 5-9 Temperature of the (a) total thermal conductivity, (b) lattice thermal
conductivity and (c) electronic thermal conductivity for KPblg-
xSnbeTezo (x=5, 9, 13 and 16) ............................................................... 172

Figure 5-10 Temperature dependence of the figure of merit ZT for KPblg-
xSnbeTezo (x=0, 3, 5, 9, 13, 16) ............................................................. 173

xvi

Figure 6-1

Figure 6-2

Figure 6-3

Figure 6-4

Figure 6-5

Figure 6-6

Figure 6-7

Figure 6-8

Figure 6-9

Figure 6-10

Figure 6-11

Powder X-ray diffraction of the series CuPbmeTem+2 (m=5, 12,
18, 20, 22, 30, 40 and 50), the area between 20 = 25 and 50 deg is
enlarged to show the presence of extra peaks that do not belong to
the NaCl structure-type. Two small peaks at 28 and 38 deg are

characteristic of szTe3. Small peaks ~ 42-43 deg belong to CuzTe ...... 184

BSE images of a sample with composition CuPb13SbTe20 showing

regions rich in CuzTe and szTe3 embedded in the PbTe matrix.
The two phases form dentritic ribbons as long as 3 mm in all
directions .................................................................................................. 1 85

BSE images of the compound CuPb40SbTe42. More dispersed

regions composed of CuzTe and szTe3 are visible in the PbTe
matrix ....................................................................................................... 186

DTA results: (a) first cycle and (b) second cycle for CuPb3SbTelo;
(0) first and ((1) second cycle of CuPb30SbTe32; (e) first cycle and (f)
second cycle of CuPbsoSbTesz ................................................................ 187

Temperature dependence of (a) the electrical conductivity, (b) the
therrnopower and (c) the power factor for the series

CuPbmeTem+2 (m=8, 12, 18, 20, 22, 30, 40, 50); ((1) carrier
concentration as a function of temperature for CuPblgstezo and
CuPb22SbTe24 .......................................................................................... 1 89

Temperature dependence of the (a) total and (b) lattice thermal
conductivity for the compounds CuPbmeTem+2 (m=8, 18, 20, 30,
40) ............................................................................................................ 190

Figure of merit ZT for the compounds CuPbmeTem+2 (m= 8, 18,
20, 30, 40) ................................................................................................ 191

Powder X-ray diffraction for (a) CuPbrgstezo, (b)
CuPbgsngstezo and (c) CuSnlngTezo ................................................. 194

BSE images of CuPb13Sn58bTe20. Complex regions rich in CuzTe
and szTe3 are clearly visible in the Pb1-xSnxTe matrix ......................... 195

BSE images of (a) CuPbgsngstezo and (b) CuPb58n13SnTe20 ............. 196

BSE images obtained for CuSnlngTezo. Only CuzTe regions
segregated in the SnTe matrix .................................................................. 197

xvii

Figure 6-12

Figure 6-13
Figure 6-14
Figure 6-15
Figure 6-16
Figure 6-17
Figure 6-18

Figure 6-19

Figure 6-20

Figure 6—21

Figure 6-22

Figure 6-23

Figure 6-24

Temperature dependence of the (a) electrical conductivity, (b)

therrnopower and (c) power factor of the series CuPb13-xSnbeTe20
(x=0, 5, 9, 13, 18) ..................................................................................... 200

Total, electronic and lattice thermal conductivities of CuSnlgstezo ....20]

Powder X-ray diffraction patterns for samples with nominal

composition CquTez ratios prepared by water-quenching, air-

quenching and slow cooling. 203

BSE images of the sample with nominal composition CquTez
quenched in water .................................................................................... 204

BSE images of the sample with nominal composition CquTez
quenched in air ......................................................................................... 205

BSE images of the sample with nominal composition CquTez
prepared by slow cooling ......................................................................... 206

BSE imaging of a sample prepared with nominal composition
CquTez with the furnace being rocked during the reaction ................... 207

DTA analysis of the sample prepared by mixing Cu, Sb and Te
with the ratio l:1:l:2 by rocking the fiirnace: (a) first cycle, (b) 2“‘1
cycle ......................................................................................................... 208

Powder X-ray diffraction of PbTe doped with 1, 2 and 5 % CuzTe.
The small stars indicate diffraction peaks that do not belong to the

PbTe structure-type and indicate the presence of CuzTe as a minor
phase ....................................... 210

BSE images of PbTe doped with 5 % CuzTe .......................................... 211

BSE images of PbTe samples doped with (a) 2 % and (b) 1 %
CuzTe ....................................................................................................... 212

Differential thermal analysis results for (a) the first run and (b)
second for the PbTe doped with 1% CuzTe; (c) the first run and (d)
second run of PbTe doped with 2 % CuzTe ............................................. 213

Temperature dependence of (a) the electrical conductivity and (b)

the therrnopower of PbTe doped with l % CuzTe, temperature
dependence of (c) the electrical conductivity and (d) the

therrnopower of PbTe doped with 2 % CuzTe ......................................... 214

xviii

CHAPTER 1

Introduction to Thermoelectrics

1.1. Thermoelectric concepts

Thermoelectric devices are solid state systems able to convert heat into electricity
(power generation) or vice versa electrical energy to cooling (refrigeration). Typical
applications of thermoelectric devices include supplying power to NASA spacecraft,
power generation in remote areas, electronic equipment along fuel pipelines, and cooling

for electronic devices.

Thermoelectric devices are based on thermoelectric phenomena, the Seebeckl and

Peltier2 effects. The former was discovered by Thomas Seebeck in 1823. An electrical

potential is generated within any isolated conducting material that is subjected to a
temperature gradient. In thermoelectric devices, a junction is formed between two
different conducting materials, one containing positive charge carriers and the other
negative charge carriers. A heat source at the junction will make the carriers flow away
from the junction, making an electrical generator (Figure l-1b). The Seebeck coefficient
(or therrnopower) S is defined as follows:

5 =11”. (1)
dT
With dV the potential difference generated and dT the temperature difference.

The Peltier effect is the reverse process. Heat is absorbed or liberated when a current

crosses an interface between two different conductors. This is the basis for cooling

devices: when an electric current is passed in the appropriate direction through the
junction, both types of caniers move away from the junction and convey heat away

(Figure 1-1a).

 

 

 

(a) Cooled Surface 03) , Heat source

i" gt i" g?
—-» 1—— ——+ 1—

A Dissipated heat It A Cool Side V

+li- “—WV—

Figure 1-1. Scheme of (a) a cooling device and (b) a power generator.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Thermoelectric devices (power generators and coolers) present several advantages
compared to other systems. They are light, small, reliable and vibrationless. For example,
radioisotopes thermoelectric generators (called RTGs) produce electrical power by
converting the nuclear decay of radioactive isotopes (typically plutonium-238) into

electricity and have been successfully used to power a number of long-time space
missions (tens of years).3 There are many economic and environmental benefits of

thermoelectric devices as well. Waste heat can be recovered and used as a source for
power generation. Thermoelectric coolers can replace environmentally harmful
chlorofluorocarbons in. refrigeration compressors. The automotive industry is also
interested in the development of thermoelectrics. Only 25 % of the fuel energy is used for

vehicle mobility, the remainder is lost in the form of waste heat in the exhaust and

coolant as well as friction and parasitic losses. Applications such as electrical power
generation and refrigeration (seat coolers, electronic component cooling) are currently
being investigated. Refiigeration of electronics is an important technology that
thermoelectric could greatly impact. More efficient coolers are predicted to produce
speed gains of 30 % to 200 % in some computer processors based on complementary
metal oxide semiconductor (CMOS) technology. However, so far the market of

thermoelectric systems has been restricted to small niches because of their low
efficiencies.4’ 5 Typically, for cooling applications, the coefficient of performance of
thermoelectric devices is about 10% of the Carnot efficiency whereas kitchen
refrigerators for example operate at about 30% of the Carnot efficiency.5

The efficiency of a thermoelectric material is based on the so-called figure of merit ZT

defined by the expression:

2
=. ”S T (2)
K

ZT

 

where S is the therrnopower (or Seebeck coefficient) of the compound, 0 its

electrical conductivity, it its thermal conductivity and T the absolute temperature. The
numerator 0-82 is often referred as the power factor. The thermal conductivity has two
contributions: one from the lattice vibrations (called lattice thermal conductivity) and the

other from the charge carriers (called electronic thermal conductivity). The electronic

component is related to the electrical conductivity through the Wiedemann-Franz law:

where L is the Lorenz factor, 6 the electrical conductivity and T the temperature. L is

typically taken as 245108 WQ/K for degenerate semiconductors.6

For a thermoelectric device to be competitive with compressor-based
refrigerators, materials with ZT ~ 3 - 4 have to be discovered.5 So far, the materials used

for thermoelectric applications exhibit ZT not higher than 1. Today’s commercial

materials are alloys discovered decades ago. For example, PbTe with ZT~0.8 at 700K is

used for temperature range 600 — 700 K and BigTe3/Sb2Te3/Bi28e3 alloys are used for
electronic refrigerators (ZT=0.9 at 400 K). The RTG systems are based on (AngTe2)1-x

(GeTe)x (TAGS) materials and SiGe alloys. Figure 1-2 shows the figure of merit as a

function of temperature for several bulk materials. Several systems clearly exhibit ZT
higher than unity at temperatures above 600 K and hold promise for technological
applications. Since 1990, the field of therrnoelectrics is experiencing a rebirth because of
the interest of the US Department of Energy in the potential of thermoelectrics. Progress
in solid state science such as the discovery of new complex materials, more reliable band-
structure calculations, and the progress in structure determination by X-ray diffraction
have contributed to the renewal of the field. A new research direction for high ZT

materials is to prepare known compounds in low dimensional scale such as quantum

10,11

wells,7’ 8 quantum wires,9 superlattices and quantum dots. These materials exhibit

higher ZT than their bulk counterpart.

 

     
  
     
     

l I I I ' ' '
1.8 .. NanzosteZZI SALT (p) Ang18+beTe20' LAST (n) -
1.6- Ag(PbSn)meTem,2, LASTT'

PbTe-PbS (n)

1_4- AngTez-GeTe, TAGS (p) '
1.2: . Zn.Sb3 (p) Yb14MnSb11 (p')'
1 O- BlzTe3 (n. p) CeFe3CoSb12 -

I— . ‘ A/

N 0.8- CsBi4Te6 (p
0.6 - '
0.4 - -
0.2 - '
0.0 - -

 

 

 

6 1 250 ' 460 I 600 ' 800 '10'00'12'0021400

Temperature, K

Figure 1-2. Figure of merit ZT shown as a function of temperature for several bulk

materials.

1.2. Optimization of the figure of merit

According to equation (2), good thermoelectric materials should have large
therrnopower, high electrical conductivity and low thermal conductivity. A good
description of such a material was given by Slack and is known as the “PGEC” concept:
“Phonon Glass Electron Crystal”.12 Good thermoelectric materials would conduct
electricity like a crystalline solid but conduct heat like a glass. The problem is that the

electrical conductivity, the therrnopower and the electronic thermal conductivity are

determined by the details of the electronic and crystal structures and the scattering of
charge carriers. Therefore they are not independent from each other. This makes the
design of materials with high ZT a challenge. Two approaches can be considered to
improve the figure of merit: the first consists of increasing the power factor while the
second focuses on reducing the lattice thermal conductivity.

As metals have high electrical conductivity and low therrnopower and insulators

low electrical conductivity and high therrnopower, optimum power factors are achieved

in degenerate semiconductors with carrier concentrations ~ 1020 /cm3 (Figure 1-3).

 

     
   

   

   
 

 

 

1400 v r v I v ‘IOOOO [5?-

o

“1200 ‘ :1“

§ ‘8000 §

31000 : _8_

2;: 800 ‘ 6000 a

o I I 2 c

8' 600 ,’ Heavily Doped ‘ ‘ 4000 2..

E 400 Semiconductor, Semiconductor E

a, " 2000 A
.c I \

'— 200 \ Metal ‘ (RD

\ I
O A A n e - r r 1 r . r . ~ _ O 3
17 18 19 20 21 22 V

Log (Carrier Concentration)

Figure 1-3. Evolution of the Seebeck coefficient and the electrical conductivity as a

. . . 12
function of the carrier concentration.

The Boltzmann transport theory provides a general understanding of the

thermopower using the Mott formula”:

_ 2r_2 sz dln our)

S 3 e dE |E=Ef (4)

 

The conductivity 0 (E) is determined as a function of band filling. If the carrier scattering
is independent of energy, then 6 (E) is just proportional to the density of states at E. The
therrnopower is a measure of the difference in o (E) above and below the Fermi surface.
Therefore thermoelectric materials with high therrnopower require high compositional
and structural complexity contributing to complex electronic structure since the
therrnopower is a measure of the asymmetry in electronic structure near the Fermi level.

7‘ 14 predicted dramatic changes in the density

Theoretical studies pursued by Hicks et al.
of states as the system size decreases and approaches nanometer length scales (Figure 1-

4). As a result, an increase in therrnopower is expected in reduced dimensionality

 

 

 

 

 

 

 

 

 

 

systems.
1E+21 - f/BBKBD)
E+2fl
' SEW I / Quantum Well
§ TE+2O - W
(T; BE+20 (
‘5 5E+2D j /
'33: 4E+20-l /| (QILDintum Vtfire
:
o 3E+20 - /
n 2520 - \
15,20 . Quantum dot (0D)
0 E.

 

 

010 20 so 40 so so 70 so so 100110120
Energy (meV)
Figure 14. Density of states for electrons in bulk semiconductors (3D), quantum wells

(2D), quantum wires (1 D) and quantum dots (OD).

One way to optimize ZT is to minimize the lattice thermal conductivity by
increasing phonon scattering. In bulk materials, phonon scattering can be enhanced by

introducing heavy atoms and mass fluctuation in the lattice. Large unit cells will also tend

to exhibit low lattice thermal conductivity.15 The presence of a weakly bound atom that

rattles inside a cavity will also reduce the thermal conductivity of the material without
severely affecting the electronic conduction. Different systems based on this concept are
under investigation: skutterudites, clathrates, complex chalcogenides. They are described

in the following section. Another strategy is to scatter phonons at interfaces, leading to

the use of multiphase composites mixed on the nanoscale.16

1.3. Bulk Thermoelectric Materials

Different bulk systems are under investigation for thermoelectric applications:
skutterutides, clathrates, Half-Heusler, zintl phases, oxides, complex chalcogenides. A
brief overview of these different systems is presented in this section. Further details are

. . . 17-19
available in the references and several revrew papers.

The skutterudite-type structure (CoAS3-type) is a cubic structure with the space
group composed of eight comer-shared TX6 (T=Co, Rh, 1r, X=P, As, Sb) octahedra. The

linked octahedra produce a void or vacancy site at the center of the (TX6)8 cluster. The

metal atoms occupy the corners of the eight “cubes” with six Sb rings inside the cube and
two voids in the remaining cubes. Figure 1—5 shows the structure of a skutterudite
antimonide compound. Many different elements have been introduced into the voids of

skutterudites, including lanthanides, actinides, alkaline-earth, alkali, thallium and group

IV elements, resulting in significant reduction of the lattice thermal conductivityzo’ 2'

Skutterudite antimonides possess the largest voids. ZT ~ 1.4 above 1173 K was reported

for LaFe3CoSb12 and CeFe3CoSb12.22

 

Figure 1-5. Structure of a skutterutide antimonide.

Clathrates, like skutterudites, exhibit cage-like structures with rattling
mechanisms. They can be thought of as periodic solids in which tetrahedrally bonded

atoms form a framework of cages. Their crystal structures are closely related to those of
type-I and type-II clathrates hydrides such as (C12)g(H20)4g and (C02)24(H20)136. A wide
variety of different elements have been encapsulated inside these polyhedra, including

alkali-metal, alkaline earth and rare earth atoms. As a result, low thermal conductivity

3

was measured for type I clathrates.2 For example, the thermal conductivity of

SrgGalgGe3o shows a similar magnitude and temperature dependence to that of

. 24-26
amorphous materials.

Half-Heusler alloys, with general formula MNiSn (with M a group IV transition
metal), have the MgAgAs crystal structure (space group F4-3m) consisting of three

interpenetrating fcc sublattices with one Ni sublattice vacant. They are small band gap

27-29

semiconductors with Eg ~ 0.1-0.5 eV. The Half-Heusler alloys exhibit a high

negative thennopower (-40 to -250 pV/K) and low electrical resistivity (0.1 to 8
mSZ/cm).18 However the thermal conductivity is relatively, high, ~ 10 W/m'K. The
chemistry of the three sublattices can be tuned independently. For example, in TiNiSn,

doping the Sn site provides the charge carriers while partial substitution of the Ti and Ni

sites causes mass fluctuations that can lead to the reduction of thermal conductivity. A

power factor as high as 4.5 W/m'K2 was reported at 650 K for TiNiSn0.958b0,5.18 Partial
substitution of Ni by Pd in the system ZrNiSn resulted in KL~ 2 W/m-K at 800 K for

. 30
Zf0.5Hfo.5Nlo.5Pbo.53n0.993bo.0r.
The discovery of high ZT at high temperature ( ZT ~ 1 at 1223 K) in the p-type

Yb14MnSbH material demonstrates the successful combination of electron—crystal

31, 32

phonon glass-properties diplayed by Zintl phases. The structure is isostuctural to

C314AISbH33 and consists of [MnSb4]9' tetrahedra, polyatomic [Sb3]7' anions and

isolated Sb3‘ anions. X-ray magnetic circular dichroism (XMCD) and XPS (X-ray
Photoelectron Spectroscopy) measurements show that the Yb is in the 2+ oxidation state
replacing Ca2+ but the Mn was found to be Mn2+, supplying one less electron than AP”.
The material is p-type and Hall effect measurement showed carrier concentration ~
131021 holes/cm3. The total thermal conductivity is remarkably low ranging between ~
0.7 - 0.9 W/m-K for the temperature range 300-1275 K. Subtracting the electronic
contribution results in low lattice thermal conductivity, comparable to that of a glass. The

reason of such low values can be attributed to the complexity of the structure and the

heavy atomic mass of the crystal.

10

Complex oxide materials are a good example of hybrid materials. Layered cobalt

oxide materials such as NaxCon434 and Ca3Co40935 are composed of C002 nanosheets

and sodium ion layers or calcium cobalt oxide misfit layers. The C002 nanosheets

possess a strong correlated electron system that serves as electronic transport layers while

the sodium ion nanoblock layers or calcium cobalt oxide misfit layers serve as phonon-

scattering boundaries that reduce the thermal conductivity.36’ 37

1.4. Complex chalcogenide systems as potential thermoelectric

materials

Exploratory synthesis of temary and quaternary chalcogenides has led to the
discovery of a wide range of compounds. Some of them, notably thallium chalcogenides,
show promising thermoelectric properties. Ternary and quaternary bismuth chalcogenides
exhibit a wide compositional and structural diversity. Some of them are good examples of

the PGEC concept with alkali metal rattlers.

Thallium chalcogenides, particularly TlgBiTe638 and TIZSnTes,39 tend to possess
very low thermal conductivitiy. TlgBiTe6 is derived from the isostructural T15Te340 and
optimized compositions exhibit ZT ~ 1.2 at 500 K mainly because of its extremely low
lattice thermal conductivity (0.39 W/m-K at 300 K). In T12SnTe5, chains of (SnTe5)2' run
parallel to each other and charge-balancing Tl+. The long Te-Tl bonds (~ 3.49-3.66 A)

produce low-frequency phonons which results in a very low lattice thermal conductivity
(0.5 W/m-K over 50-300 K). However these compounds are unlikely to be used for

practical use because of toxicity issues.

11

The ternary and quaternary bismuth sulfide and selenide systems have been
extensively studied in our laboratory.41'47 The systems exhibit a wide compositional and
structural diversity because of the presence of bismuth. The element possesses a lone pair
of 6s2 electrons. The lone pair can be stereochemically expressed as a distortion the Bi
coordination or can be suppressed by hybridization with adjacent p- or d-orbitals

resulting in symmetrical octahedral coordination geometry. Bismuth atoms adopt

different coordination environment from 3 to 9 near neighbors. Among the various Bi-Q
(Q=S, Se, Te) coordination geometries, Bng octahedral coordination is the most

abundant. Furthermore, octahedral and square pyramidal geometry when combined can

100

produce several common building fragments such as NaCl-(NaCl ), Sb28e3-(NaCllOO),

BizTe3-(NaCll 1), Cdlz-(NaCll l I), and galena types (NaCl3l 1) (Figure 1-6).

4

Some of these materials, particularly B-KzBigSe1343 and CsBi4Te6,48’ 9 show

promising thermoelectric properties. These ternary chalcogenides materials are made of
layers separated by channels filled with the alkali metal. The presence of the alkali metals

rattling inside the channels leads to low thermal conductivity.
B-KzBi38e13 is made up of BizTe3, NaCl and Cdlz-type infinite rod-shaped
blocks. The CdIz and BizTe3-type blocks are arranged side by side to form layers. NaCl-

type blocks serve as connectors between these layers to build a 3D framework with
channels filled with potassium cations. As a result, thermal conductivity of ~ 1.28 W/m'K

was measured at room temperature for a polycrystalline sample.

12

 

 

 

NaClll
-type

NaCl311
43138
:>

 

Figure 1—6. Various building blocks (shaded) based on different “cuts” of the NaCl-
type structure. The diagram is view down with [011] plane. Black and white circles

are bismuth and chalcogen atoms, respectively.

Attempts to prepare telluride analogs of these sulfide and selenide compounds
resulted in the formation of CsBi4Te6 instead of CszBigTe13. The compound is highly

anisotropic with rod-shaped crystals. The crystal structure viewed along the b-axis is

displayed on Figure 1-7.

o go... {.01 oogoofigo

0“
C

 

r ‘ ,«. k

 

 

 

 

 

 

- ,’ _ I z: r .‘ g 2 ._

o.,.£0‘_07,o or. 0 Om.v.°‘_.Cs
. ‘ s " " fii- : 3‘ A Te

., . 3 BI

12}. .
..p “- '» 7 ._, '
‘6 0"00‘00’00 Q‘. Q'Qi'o
‘———>
b 3 23A Bi-Bi bond

Figure 1-7. Perspective View of the CsBi4Te6 along the b-axis.

CsBi4Te6 crystallizes in the space group C2/m and is composed of anionic
infinitely long [Bi4Te6]- blocks and Cs+ ions residing in the interlayer space. The Bi/T e

layers consist only of Na-Cl—type Bi/Te blocks linked by Bi-Bi bonds. The presence of

Bi-Bi bonds (3.2383 (10) A) in the Bi/Te blocks is remarkable. The addition of one
electron per 2 equivalent of BizTe3 does not give a formal intercalation compound but
causes a dramatic reorganization of the B12T63 framework and reveal an inability to
delocalize such electrons. Isolated Bi-Bi bonds in solids are rare with the only other

example reported for BizGang (Q=S, Se).50

CsBi4Te6 is a promising candidate for cooling applications, with a thermoelectric

figure of merit of about 0.8 at 225 K. This is the highest ZT reported for a material at
such low temperature. Undoped samples have room temperature electrical conductivity

between 900-2500 S/cm and therrnopower values between 90 and 120 uV/K. Doping

studies on CsBi4Te6 showed that the system is significantly affected by low doping
levels. The highest ZT of 0.8 at 225 K was achieved by dOping CsBi4Te6 with 0.05%

Sbl3. In that case, a power factor as high as 51.5 uW/cm-K2 was attained at 184 K. The
total thermal conductivity measured along the direction parallel to the needles are in the
range 1.25-1.85 W/m-K at room temperature.

Attempts to introduce Pb in CsBi4Te6 led to the discovery of the homologous

51, 52

family of materials CstmBi3,Te5+,m (m=1, 2, 3, 4) (Figure 1-8). The expression

“homologous family” was introduced by Magneli53 to characterize chemical series that

are expressed by a general formula and built on common structural principles. The only
difference between members of the family is the dimensions of these buildings units. The

concept of phase homologies has become a useful tool in solid state chemistry to design

and target new compounds. 54’ 55 The exploration of the quaternary system A/M’/M”/Se

(A=K, Rb, Cs, Sr, Ba; M’=Sn, Pb, Eu; M”=Sb, Bi) led to numerous compounds.42’ 43

This includes members of the homologous megaseries

56-58
Amer+1362+rl2m[M21+nS€2+3r+n]-

The four members of the family were prepared by introducing various equivalents

of PbTe in the layered framework of CsBi4Te6. As CsBi4Te6, these compounds are made

of layered Bi/Te frameworks separated by Cs+ cations. However the partial substitution

15

of Bi by Pb causes the loss of the Bi-Bi bond. As can be observed in Figure 1-8, the
difference between the four members is the thickness of the anionic layers. This thickness

increases with increasing Pb content. All four members of the family show reduced

thermal conductivity compared to that of BizTe3 and CsBi4Te6 (1.8 and 1.5 W/m'K for

CstBi3Te6 and Cst4Bi3Te9 at room temperature).52

C O i C ‘ 0 ‘
D O ‘ C O ~ 0 0.}:1—3.1_I 1‘:
,o‘ . , . . titrtrtrI trrr {LIN
- . o . - . o . o If}? iii: rim rtrt r‘r": ‘ r“:
I‘.I-‘r“_I:r.‘I-tit; Slip: II 3;: ; r3; t 3;», f it; 3 z
:3"? "Iii"??? ‘. - .‘ . - .‘ -‘ I. If. ;‘ 3.1" :‘I 1+1- I? If I '1
~ 0 j o ‘ o ‘ o ‘ Trix". I #1., 11.93-17,...” l- 4-123.111
3913914391 '1‘ I! III‘TI. LIT—1 iii-"t P'I‘Iiiii 51 I771???
filllllilli f;f;f‘;f.} o}ofo:o‘ o IWIIITI‘TEI
0:0:o:o:o , __ ~--~-°~v$~r-f~ ~ ,--,
rt 1,}, A, Lgfigjri trtrrstr-‘ Ir: 1 :‘rt 1 r
_.‘,‘ 1-; r,“ if,“ E;i;~t.;*;‘ ritrtit-..‘ rrrrtr I r:
l;i;i;i;i i;r;i;i;i r333} rn‘r‘i‘r
. O C O

(a) CSPbBiaTe6 (b) CstzBiaTe7 (C) CSPbaBiaTe8 (d) Cst4Bi3Te9
Figure 1-8. The structures of (a) CstBi3Te6, (b) CstzBi3Te7, (c) Cst3Bi3Teg and (d)

Cst4Bi3Te3 in projection down the c axis for (a) and (c) and the a axis for (b) and (d).

1.5. Low-dimensional and nanostructured materials
Till now, this chapter focused mainly on bulk materials. However a different
approach is under investigation by several groups. Many experimental and theoretical

. . . . 59-61 .
studies on low dimensronal nanostructured materials and mixed-phase

7, I4, 62, 63

nanocomposites provide encouraging results in getting high ZT by reducing

lattice thermal conductivity while maintaining high power factors. The idea of using

16

reduced dimensionality to improve ZT was first discussed by M. S. Dresselhaus at al.7’ 14

The quantum confinement effect could be used to enhance the therrnopower whereas the
lattice thermal conductivity could be reduced through the presence of numerous

interfaces that scatter phonons more effectively than electrons. The superlattice thin-film

structures of BizTe3/Sb2Te3 grown from chemical vapor deposition64 and of

59, 60

PbSeaggTeuoz/PbTe formed by molecular beam epitaxy claimed ZT values greater

than 2 (at~ 300 and 550 K respectively). Studies have shown that significant thermal
conductivity reduction occurs regardless of the orientation of the superlattices or
nanostructures to the measurement direction. 64

Large scale production and utilization of such materials is significantly
jeopardized by their low thermal stability and the complex and costly synthesis process.
Bulk analogues of such systems with similar figures of merit emerge as alternative
candidate materials for thermoelectric energy conversion. Different synthetic routes can

be considered to prepare such materials. For example, nanostructured PbTe bulk

materials could be achieved through matrix encapsulation,65’ 66 spinodal
decomposition,67’ 68 nucleation and growth.69'7l Recently our group reported on the n-
type systems AngmeTem+2 (LAST)69, and Pb1.xSnxTe-PbS67 and the p-type systems
Na1-bemeyTem+2 (SALT)71 and Ag(Pb1-ySny)meTe2+m (LASTT)7O, that exhibit high
thermoelectric figure of merit. The compositions Angjgstezo and Nao_95Pb20SbTe22

reach ZT of ~ 1.8 and 1.6 at 700 and 650 K respectively. These outstanding values are
due essentially to their very low thermal conductivity. The total thermal conductivity

drops very rapidly with rising temperatures and reaches a minimum value of 0.85 W/m-K

17

at 700 K for Na0.95Pb20SbTe22. Careful analysis of high resolution transmission electron

microscopy images of these materials revealed the presence of nanostructuring coherently
embedded in what is essentially a PbTe matrix. The precise composition of these

nanostructures could not be determined due to instrumental limitation. These clusters are
believed to be rich in monovalent cations (Ag+, Na+) and trivalent cations Sb3+ because
of the Coulombic forces that do not allow charge imbalances around the atoms. Ab initio
calculations within density functional theory were pursued for PbTe with and without
monovalent impurities (Na, K, Rb, Cs, Cu and Ag).72 The result showed an increase of
the density of states near the top of the valence band for K, Rb, Cs, Cu and Ag dopants.
Further studies on these systems have been investigated: the partial substitution of Pb by

Sn in NanlSSbTezo and KPblngTezo, the preparation of members with low m values
for NanmeTem+2, the substitution of Ag by Cu in AngmeTem+2. The systems

NaSanTcm+2 (M=Sb, Bi) were prepared as well and their properties are compared to
that of the Pb analogs. These studies are detailed in chapters 3, 4, 5, 6 and 7.

Tl-doped PbTe materials also hold promise to improve the thermoelectric figure
of merit of PbTe. Heremans et al.73 reported ZT ~ 1.5 at 773 K for Pb0_93Tlo,2Te.
Contrary to the nanostructured systems described above where an increase in ZT is

achieved mainly through a reduction in lattice thermal conductivity, the enhancement of

ZT in Tl-doped PbTe is attributed to an increase in the thermopower. Calculations

indicated that T1 creates resonant energy levels located in the valence band of Tl-PbTe.72’

74

18

A A recent study has revealed that co-nanostructuring PbTe with two different
phases (Pb and Sb in that case) could significantly impact the temperature dependence of

the electrical conductivity compared to that of PbTe or PbTe nanostructured either with

Pb or Sb nanodots.75 For PbTe —Pb(2 %)-Sb ( 3 %), the electrical conductivity increases

with temperature. Thermopower measurements and Hall effect data support the theory of
a slower rate of decrease in mobility of the charge carriers. As a result, for that
composition, an increase in power factor with temperature was observed. This is a
significant improvement towards increasing the power factor of thermoelectric materials
because so far improved thermoelectric figures of merit were achieved through reduction

of the lattice thermal conductivity.
In this dissertation, substitution studies on CSPbBi3TC6 and Na1-bemeyTem+2

were pursued in order to characterize the resulting materials and particularly to evaluate

their thermoelectric materials. Chapter 2 describes the studies of partial substitution of Pb

and Cs by other elements such as Sn, Na, K, Ca, Yb in the compound CstBi3Te6, the
first member of the homologous family CstmBi3,Te5+m described in section 4. Single-
crystal X-ray diffraction data, collected on several needle-shape crystals, indicate the
COHIPOSitiOHS CS0.76Ko.7413i3.5T€6 (1), CSNao.9sBi4.orTe7 (2), CSo.69Cao.6sBi3.34T66 (3),
Rbo.82Pbo.8zBi3.18T66 (4), Rbo.19K1.3rBi3.5oT66 (5), RbSnBi3T66 (6), Rbo.94Cao.94Bi3.06Tes
(7), RbeBi3Te6 (8) and KSnSb3Te6 (9). However optimum synthesis conditions have

not yet been found to prepare a pure phase of the compounds.
Chapters 3, 4 and 5 concern the preparation and characterization of bulk

nanostructured PbTe materials. In section 5, the promising thermoelectric properties of

19

the p-type system Na1-bemeyTem+2 (SALT)71 were discussed. However only high m

values (m higher than 19) were studied. In Chapter 3, the synthesis and characterization

of members with lower m values are discussed. Addition of Sn to the LAST system

resulted in the promising p-type system Ag(Pb1_ySny)meTe2+m (LASTT).70 Similar

studies with the SALT system were performed, resulting in the preparation of the series
of samples with composition Nantg_xSnbeTe20. The detailed synthesis procedures and
characterizations are described in Chapter 4. To assess the role of the alkali metal on the
properties of the bulk nanostructured PbTe, potassium analogs of the Nan13-xSnbeTe20

were prepared. This study is reported in Chapter 5.
Substitution of Ag by Na or K did not affect the presence of nanostructures inside

the PbTe matrix. However, replacing Ag by Cu in the LAST system resulted in the
precipitation of CuzTe and szTe3 phases in PbTe. Details about the characterization and

the thermoelectric properties of the cast ingots are reported in Chapter 6.

20

10.

11.
29, L1.

12.

13.

References

Seebeck, T. J ., Abh. K. Akad. Wiss. 1823, 265.

Peltier, J. C., Ann. Chem. 1834, LV 1, 371.

Yang, J .; Caillat, T., Mater. Res. Bull. 2006, 31, 224.

Mahan, G.; Sales, 8.; Sharp, J ., Physics Today 1997, 42.

Di Salvo, F. J., Science 1999, 285, 703.

Kittel, C., Introduction to Solid State Physics. Wiley: 2005.

Hicks, L. D.; Dresselhaus, M. 8., Phys. Rev. B 1993, 47, (19), 12727.

Hicks, L. D.; Harman, T. C.; Dresselhaus, M. 8., Appl. Phys. Lett. 1993, 63, 3230.

Sofo, J. 0.; D., M. 0., Appl. Phys. Lett. 1994, 65, 2690.

Harman, T. C.; Spears, P. J .; Manfra, M. J ., J. Electron. Mater. 1996, 25, 1121.

Harman, T. C.; Taylor, P. J .; Spears, P. J .; Walsh, M. P., J. Electron. Mater. 2000,

Slack, G. A., CRC Handbook of T hermoelectrics. 1995.

Mott, N. F .; Jones, H., The Theory of the Principles of Metals and Alloys. Dover

Publications, NY: 1958.

14.

15.

Hicks, L. D.; Dresselhaus, M. 8., Phys. Rev. B 1993, 47, 16631.

Kanatzidis, M. G., Semicond. Semimet. 2000, 69, 51.

21

16. Dresselhaus, M. S.; Chen, G.; Tang, M. Y.; Yang, R.; Lee, H.; Wang, D.; Ren, Z.;
Fleurial, J.-P.; Gogna, P., Adv. Mater. 2007, 19, 1043.

17. Snyder, G. J .; Toberer, E. 8., Nat. Mater. 2008, 7, 105.

18. Tritt, T. M., Mater. Res. Soc. Symp. Proc. 2002, 691, Gl.1.l.

19. Chen, G.; Dresselhaus, M. S.; Dresselhaus, G.; Fleurial, J.-P.; Caillat, T., Int.
Mater. Rev. 2003, 48, 1.

20. Nolas, G. S.; Morelli, D. T.; Tritt, T. M., Annu. Rev. Mater. Sci. 1999, 29, 89.

21. Uher, C., Semicond. Semimet. 2000, 69, 139.

22. Sales, B.; Mandrus, D.; Williams, R. K., Science 1996, 272, 1325.

23. Nolas, G. S.; Slack, G. A.; Schujman, S. B., Semicond. Semimet. 2001, 69, 255.

24. Cohn, J. L.; Nolas, G. S.; Fessatidis, V.; Metcalf, T. H.; Slack, G. A., Phys Rev.
Lett. 1999, 82, 779.

25. Nolas, G. S.; Weakley, T. J. R.; Cohn, J. L., Chem. Mater. 1999, 11, 2470.

26. Sales, B.; Chakamoukos, B. C.; Jin, R.; Thompson, J. R.; Mandrus, D., Phys Rev.
B 2001, 63, 245113.

27. Aliev, F. G., Z. Phys. B 1989, 75, 167.

28. Aliev, F. G., Z. Phys. B 1990, 80, 353.

29. Ogut, S.; Rabe, K. M., Phys. Rev. B 1995, 51, 10443.

30. Shen, Q.; Chen, L.; Goto, T.; Hirai, T.; Yang, J.; Meissner, G. P.; Uher, C., Appl.
Phys. Lett. 2002, 79, 4165.

22

31. Brown, S. R.; Kauzlarich, S. M.; Gascoin, F.; Snyder, G. J ., Chem. Mater. 2006,
18, 1873.

32. Kauzlarich, S. M.; Brown, S. R.; Snyder, G. J ., Dalton Trans. 2007, 2099.

33. Cordier, G.; Schaefer, H.; Stelter, M., Z. Anorg. Allg. Chem. 1984, 519, 183.

34. Terasaki, I.; Sasago, Y.; Uchinokura, K., Phys. Rev. B 1997, 56, R12685.

35. Funahashi, R.; Matsubara, I.; Ikuta, H.; Takeuchi, U. ; Mizutani, U.; Sodeoka, 8.,
Jpn J. Appl. Phys. 2000, 39, L1127.

36. Satake, A.; Tanaka, H.; T., O.; Fujii, T.; Terasaki, I., J. Appl. Phys. 2004, 96, 931.

37. Shikano, M.; Funahashi, R., Appl. Phys. Lett. 2003,82, 1851.

38. Wolfing, B.; Kloc, C.; Teubner, J .; Buchner, E., Phys. Rev. Lett. 2001, 86, 4350.

39. Sharp, J. W.; Sales, B. C.; Mandrus, D. G., Appl. Phys. Lett. 1999, 74, 3794.

40. Babanly, M. B.; Gotuk, A. A.; Kuliev, A. A., Inorg. Mater. 1979, 15, (7), 1292.

41. Chung, D. Y.; Iordanidis, L.; Choi, K. S.; Kanatzidis, M. G., Bull. Korean Chem.
Soc. 1998, 19, 1283.

42. Kanatzidis, M. G.; McCarthy, T. J.; Tanzer, T. A.; Chen, L.-H.; Iordanidis, L.;
Hogan, T.; Kannewurf, C. R.; Uher, C.; Chen, 3., Chem. Mater. 1996, 8, 1465.

43. Chung, D. Y.; Choi, K. S.; Iordanidis, L.; Schindler, J. L.; Brazis, P. W.;
Kannewurf, C. R.; Chen, 8.; Hu, S.; Uher, C.; Kanatzidis, M. G., Chem. Mater. 1997, 9,
3060.

44. Chung, D. Y.; Iordanidis, L.; Rangan, K. K.; Brazis, P. W.; Kannewurf, C. R.;
Kanatzidis, M. G., Chem. Mater. 1999, 11, 1352.

23

45. Iordanidis, L.; Brazis, P. W.; Kyratsi, T.; Ireland, J. R.; Lane, M.; Kannewurf, C.
R.; Chen, W.; Dyck, J. S.; Uher, C.; Ghelani, N. A.; Hogan, T.; Kanatzidis, M. G., Chem.
Mater. 2001, 13, 622.

46. Kim, J. H.; Chung, D. Y.; Bilc, D.; Loo, S.; Short, J.; Mahanti, S. D.; Hogan, T.;
Kanatzidis, M. G., Chem. Mater. 2005, 17, 3606.

47. Choi, K. S.; Chung, D. Y.; Mrotzek, A.; Brazis, P. W.; Kannewurf, C. R.; Uher,
C.; Chen, W.; Hogan, T.; Kanatzidis, M. G., Chem. Mater. 2001, 13, 756.

48. Chung, D.-Y.; Hogan, T.; Brazis, P.; Rocci-Lane, M.; Kannewurf, C.; Bastea, M.;
Uher, C.; Kanatzidis, M. G., Science 2000, 287, 1024.

49. Chung, D.-Y.; Hogan, T. P.; Rocci-Lane, M.; Brazis, P.; Ireland, J. R.;
Kannewurf, C. R.; Bastea, M.; Uher, C.; Kanatzidis, M. G., J. Am. Chem. Soc. 2004, 126,
6414.

50. Kalpen, H.; Hoenle, W.; Somer, M.; Schwartz, U.; Peters, K.; Von Schnering, H.
G.; Blachnik, R. Z., Anorg. Allg. Chem. 1998, 624, 1137.

51. Hsu, K. F.; Chung, D. Y.; Lal, S.; Mrotzek, A.; Kyratsi, T.; Hogan, T.;
Kanatzidis, M. G., J. Am. Chem. Soc. 2002, 124, 2410.

52. Hsu, K. F.; Lal, S.; Hogan, T.; Kanatzidis, M. G., Chem. Commun. 2002, 13,
1380.

53. Magneli, A., Acta Crystallogr. 1953, 6, 495.

54. Mrotzek, A.; Kanatzidis, M. G., Acc. Chem. Res. 2003, 36, l l 1.

55. Kanatzidis, M. G., Acc. Chem. Res. 2005, 38, 359.

56. Mrotzek, A.; Chung, D. Y.; Hogan, T.; Kanatzidis, M. G., J. Mater. Chem. 2000,
10, 1667.

57. Mrotzek, A.; Kanatzidis, M. G., J. Solid. State Chem. 2002, 167, 299.

24

58. Mrotzek, A.; Kanatzidis, M. G., Chem. Commun. 2001, 17, 1648.

59. Harman, T. C.; Taylor, P. J.; Walsh, M. P.; LaForge, B. E., Science 2002, 297,
2229.

60. Harman, T. C.; Taylor, P. J.; Walsh, M. P.; IaForge, B. E., J. Electron. Mater.
2005, 34, L19.

61. Venkatasubramanian, R.; Siivola, E.; Colpitts, T.; O'Quinn, B., Nature 2001, 413,
597.

62. Kim, W.; Singer, K. L.; Majumdar, A.; Vashaee, D.; Bian, Z.; Shakouri, A.; G.,
2.; Bowers, E. J .; Zide, J. M. O.; Gossard, C., Appl. Phys. Lett. 2006, 88, 242107.

63. Caylor, J. C.; Coonley, K.; Stuart, J .; Colpitts, T.; Venkatasubramanian, R., Appl.
Phys. Lett. 2005, 87, (2), 023105.

64. Vankatasubramanian, R.; Siivola, E.; Colpitts, T.; O'Quinn, B., Nature 2001, 413,
597.

65. Sootsman, J. R.; Pcionek, R. J.; Kong, H.; Uher, C.; Kanatzidis, M. G., Chem.
Mater. 2006, 18, 4993.

66. Heremans, J. P.; Thrush, C. M.; Morelli, D. T., J. Appl. Phys. 2005, 98, 063703.

67. Androulakis, J.; Lin, C. H.; Kong, H. J.; Uher, C.; Wu, C. I.; T., H.; Cook, B. A.;
T., C.; Paraskevopoulos, M.; Kanatzidis, M. G., J. Am. Chem. Soc. 2007, 129, 9780.

68. Ikeda, T.; Collins, L. A.; Ravi, V. A.; Gascoin, F. S.; Haile, S. M.; Snyder, G. M.,
Chem. Mater. 2007, 19, 763.

69. Hsu, K. F.; L00, S.; Guo, F.; Chen, W.; Dyck, J. S.; Uher, C.; Hogan, T.;
Polychroniadis, E. K.; Kanatzidis, M. G., Science 2004, 303, 818.

70. Androulakis, J.; Hsu, K. F.; Pcionek, R.; Kong, H. J.; Uher, C.; D'Angelo, J.;
Downey, A. D.; Hogan, T.; Kanatzidis, M. G., Adv. Mater. 2006, 18, 1170.

25

71. Poudeu, P. F. P.; D'Angelo, J.; Downey, A. D.; Short, J. L.; Hogan, T.;
Kanatzidis, M. G., Angew. Chem, Int. Ed. 2006, 45, 3835.

72. Ahmad, S.; Mahanti, S. D., Phys. Rev. B 2006, 74, 155205.

73. Heremans, J. P.; Jovovic, V.; Toberer, E. S.; Saramat, A.; Kurosaki, K.;
Charoenphakdee, A.; Yamanaka, S.; Snyder, G. J ., Science 2008, 321, 554.

74. Nemov, S. A.; Ravich, Y. I., Uspekhi F izicheskikh Nauk 1998, 168, 735.

75. Sootsman, J. R.; Kong, H.; Uher, C.; D'Angelo, J.; Wu, C. 1.; Hogan, T.; Caillat,
T.; Kanatzidis, M. G., Angew. Chem, Int. Ed. 2008, 47, 8618.

26

Chapter 2

Substitutions in the Cst,,,Bi;r,Te5+m system

2.1. Introduction

Good thermoelectric materials should combine a particular set of properties. High
electrical conductivity, large therrnopower and low thermal conductivity are required to
reach a high thermoelectric figure of merit ZT (ZT=oS‘2T/rc). Thermal conductivity 1c is
comprised of two contributions, one from the lattice, the other one from the charge
carriers. Consequently both constituents are usually considered separately. Because the
electrical conductivity, the therrnopower and the electronic thermal conductivity both
depend on the electronic structure and the carrier transport properties of the compound,

optimizing ZT remains a big challenge. Two approaches can be followed to improve the
figure of merit, one is to maximize the so-called power factor (defined as 032), the other
to minimize the lattice thermal conductivity.

The best known and used materials for room temperature application are BizTe3

alloys with ZT ~ 1.1-3 The incorporation of alkali metals in the BizTe3 and Bi28e3

10, ll

structures resulted in the discovery of B-K2B188e134'9 and of CsBi4Te6. Both

compounds, anisotropic, show promising thermoelectric properties. Particularly,

CsBi4Te6 exhibits the highest reported figure of merit at low temperature (~ 0.8 at 225

K). The structure consists of [Bi4Te6]' layers separated by Cs+ cations. The presence of

Bi-Bi bonds inside the layers contributes to the complexity of the electronic band

27

structure and gives the largest contribution to the states near the conduction band
minimum.'2 The thermal conductivity is also low (~ 1.48 W/m-K in the direction parallel
to the crystal growth, i.e. along the b direction), in part due to the large unit cell and the
mass fluctuation between Bi and Te that favors phonon scattering.13 The alkali cations

are loosely bound to the Te atoms via ionic interactions and tend to rattle under the

influence of the temperature, thus reducing further the lattice thermal conductivity. In an

effort to produce materials that resemble CsBi4Te6, Pb metal was introduced to the

layered material. A new homologous family, CstmBi3Te5+m (m=l, 2, 3, 4),”’ 15 was

discovered.
The expression “homologous family” was introduced by Magnelil6 to

characterize chemical series that are expressed by a general formula and built on common

structural principles. This concept has been very useful in predicting the existence and

design of new compounds such as the megaseries Am[M1+1Se2+1]2m[M21+nSe2+3l+n] (A=K,
Rb, Cs, Sr; M=Sn, Pb, Eu, Bi, Sb)l7’ ‘8 and A2[M5+,,Se9+n].19 The four members have a

structure similar to that of CsBi4Te6. The main difference is the loss of Bi-Bi bonds as

some Bi sites are now occupied by Pb atoms. Their moderate therrnopower and metallic-

like temperature dependence of the electrical conductivity indicate heavily-doped
materials. The Sn analogs CsSnBi3Te6 and CsSnzBi3Te7 could be synthesized as well.14

Substitutions of Pb by other elements and Cs by other alkali metals could be a tool to

fine-tune the electronic properties of the system without affecting its low dimensional

structure. Here we report the results of such substitution experiments on CstBi3Te6.

Nine new compounds were identified: €50,76K074Bi35Te6 (l), CsNaoggBi4mTe7 (2),

28

CSo.69Cao.6sBi3.34Te6 (3), Rbo.82Pbo.8zBi3.18T€6 (4), Rb0.191(1.31Bi3.50TC6 (5), RbSnBi3Te6
(6), Rbo_94Cao.94Bi3_06Te6 (7), RbeBi3Te6 (8) and KSnSb3T66 (9). Single crystal
diffraction studies indicated that all these compounds except CsNa0.9gBi4.01Te7 and
KSnSb3Te6 are isostructural to CstBi3Te6. CsNaoggBirmTey is isostructural to
CstzBi3Te7, the second member of the family. KSnSb3Te6 exhibits a different crystal

structure found for the selenides CsAgo,5Bi3,5Se6 and CstBi3Se6.19

2.2. Experimental section

2.2.1. Synthesis

For each compound, all the elements were loaded in a silica tube under a dry
nitrogen atmosphere in a Vacuum Atmosphere Dry-Lab glove-box. The alkali metal was
first weighed in the tube using a Pasteur pipette. Few small chunks of telluirum were then
added to the molten alkali metal in order to prereact the two elements. The other metals
were then added inside the tube. The tubes were sealed under residual pressure of ~ 10‘1
torr. For (1), (2), (3), (4), (5), (6), (7) and (9), the tubes were then heated in a flame at a
temperature above 600 °C in order to prereact the elements before introducing the tube in
a fumace. The products were washed with degased dimethylformamide in order to

remove residual cesium/rubidium telluride formed during the reaction. For
CsNao_98Bi4.01Te7, tellurium and bismuth were introduced in one side of a H-shape silica

tube. Then sodium and cesium metal was introduced in the other part of the H-tube.

The stoichiometries used to prepare the materials are summarized in Table 2-1.

29

Table 2-1. Ratio and quantity of elements mixed for (1), (2), (3), (4), (5), (6), (7), (8) and

(9).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

30

Compounds Elements
C50.74KQ.76BI3.5T96 (1) CS K Bi Te
Ratio 1.44 0.72 7.28 12
Amount, g (mol) 0.086, 0.013, 0.684, 0.688,
(0.65) (0.32) (3.27) (5.39)
CSNaoosBierets (2) CS Na Bi Te
Ratio 0.74 1 3.5 6
Amount, g (mol) 0.150, 0.035, 1.116, 1.168,
(1.13) (1.53) (5.34L (9.15)
Carspcaossmssfles (3) CS CaTe Bi Te
Ratio 1 2 3 4
Amount, g (mol) 0.093, 0.235, 0.439, 0.357,
(0.70) (1 .40) (2.10) (2.80)
Rbo.82Pbo.8zBi3.rsTe6 (4) RP Pb Bi Te
Ratio 1 1 3 6
Amount, g (mol) 0.090, 0.218, 0.660, 0.806,
(1.05) (1.05) (3.16) (6.32)
RbQ 12K, 31Bi3 mTeg (5) Rb K Bi Te
Ratio 1 g 1 3 6
Amount, g (mol) 0.183, 0.084, 1.342, 1.639,
(2.14) (2.14) (6.42) (12.84)
RbSnBi3Te6 (6) Rb Sn Bi Te
Ratio 2 l 3 6
Amount, g (mol) 0.201, 0.140, 0.737, 0.900,
(2.35) (1.18) (3.53) (7.06)
Rb0.94C30.94Bi3.06Te6 (7) RP CaTe Bi Te
Ratio 1 2 3 4
Amount, g (mol) 0.165, 0.647, 1.210, 0.985,
(1.93) (3.86) (5.79) Q72)
RbeBi3Te6.(8) Rb Yb Bi Te
Ratio 1.5 1.5 3 6
Amount, g (mol) 0.105, 0.213, 0.513, 0.627,
(1.23) (1.23) (2.36) (4.91)
KSnSb3Te6 K Sn Sb Te
Ratio 1 1 3 6
Amount, g (mol) 0.030, 0.091, 0.280, 0.587,
(0.77) (0.77) (2.30) (4.60)

 

Temperature profiles

€50,74Ko,76Bi3,5Te6. The tube was placed in a furnace set at 700 °C for an hour, then
cooled down to 450 °C at a rate of 125 °C/h, kept at 450 °C for 12 hours and cooled
down to 50 °C at a rate of 200 °C/hour. After removal of the excess of cesium telluride
with DMF, the material consisted of silvery, thin needles. These needles were a mixture
of CsBi4Te6 and Cso_74K0,74Bi3_5Te6.

A large scale ingot was prepared for preliminary evaluation of the thermoelectric
properties of the compound. The tube was placed vertically in a regular furnace in order
to grow oriented needles. Another good technique to grow oriented needles is the
Bridgman method. In this technique, the elements are loaded in a silica tube terminated

by a tip and are placed in a temperature gradient. This technique has been successfully

applied for needle-shape selenide compounds. However it was not successful for
Cso,74K0,76Bi3,5Te(, because cesium reacts with the silica tube. Examination under

microscope of the ingot obtained with the tube in the vertical position reveals bunches of

oriented needles. However they are not well oriented with each other.

CsNa0.9gBi4,01Te-,. The tube was placed overnight in a furnace set at 300 °C to allow a
progressive vaporization and transfer of cesium and sodium to the other side of the H-
tube where they will react with tellurium and bismuth. Once all cesium and sodium
vaporized, the tube was heated to 700 °C at a rate of 100 °C/hour, held at 700 °C for one
hour and cooled down to 50 °C at a rate of 32 °C/hour. When the tube was removed, both
sides of the tube contained materials, plate material (BizTe3) in the tellurium/bismuth side

and silvery, thin needle-shape material in the cesium/ sodium side.

31

Cso,69Ca0,65Bi3,34Te6. The tube was placed in a furnace set at 900 °C for 10 hours, then

cooled down to 450 °C at a rate of 10 °C/hour, then kept at 450 °C for twelve hours and
cooled down to 50 °C at a 200 °C/hour. After isolation with DMF, silvery thin needles

were obtained. EDS analysis on several needles gave the average composition
C8060C30223i335T66-

Rb0.82Pb0.82Bi3.18Te6o The tube was placed in a furnace set at 650 °C for six hours, then
cooled down to 50 °C at a rate of 20 oC/hour. Alter isolation with DMF, silvery, thin
needles were obtained. EDS analysis gave the average composition Rb1_31Pbo_5lBi2_96Te6.
Rbo,19K1,31Bi3.5oTe6. The tube was placed into a furnace set at 700 °C for three hours,

cooled down to 50 °C at a rate of 125 °C/hour. After isolation with DMF, silvery, thin

needles were obtained. EDS analysis on several needles gave an average composition
RbK0.7713i4.00”1‘€35.74-

RbSnBi3Te6. The tube was then heated in a flame and placed into a furnace set at 700 °C
for six hours, cooled down to 450 °C at a rate of 83 °C/hour, kept at 450 °C twelve hours
and cooled down to 50 °C at a rate of 200 °C/hour. After isolation with DMF, silvery,
thin needles were obtained. The needles consisted of a mixture of RbBi3_66Te6 and
RbSnBi3Te6.

Rbo,94Ca0.94Bi3,o6Te6. The tube was placed into a furnace set at 900 °C for ten hours,

cooled down to 450 °C at a rate of 125 °C, kept at 450 °C for twelve hours and cooled

down to 50 °C at a rate of 100 °C/hour. Alter isolation with DMF, silvery, a mixture of

32

silvery, thin needle (EDS analysis gave the average composition Rb0.83Ca0,41Bi3.9Te6)
and Chunks 0f B12T63.

RbeBi3Te6. The tube was placed into a furnace set at 900 °C for ten hours, cooled

down to 450 °C at a rate of 45 °C/hour, kept at 450 °C for twelve hours and cooled down
to 50 °C at a rate of 100 °C/hour. The resulting product after washing in degassed DMF
consisted of silvery, thin needle-shape crystals. EDS analysis on needles gave the average

composition Rbe0J6Bi3. 14Te537.

KSnSb3Te6. The tube was placed in a furnace set at 700 °C for 6 hours and cooled down
to 50 °C at a rate of 22 °C/hour. The majority of the material consisted of plate-like

crystals (szTe3) but a few needle-shape crystals were also found.

2.2.2. Physical measurements

Powder X-ray diffraction. Powder X-ray patterns of the grinded materials were
recorded using Ni—filtered Cu K0 radiation on a CPS-120 lnel X-ray powder

diffractometer operating at 40 kV and 20 mA equipped with a position sensitive detector.
Energy Dispersive Spectroscopy analysis (EDS). Semiquantitative microprobe analysis
of the crystals were performed with a JEOL J SM-35C scanning electron microscope
equipped with a Tracer Northern energy dispersive spectrometer. Data were acquired

using an acceleration voltage of 25 kV and a 303 accumulation time.

Single crystal X—ray diffraction. For the single crystal of €50.74Ko,76Bi3.5Te6,

C8069C30653i3s4T66, Rbo.82Pbo.8zBi3.18T66, Rbo.19K1.31Bi3.50T66 and KSnSbsTes,

intensity data were collected at room temperature on a Bruker SMART Platform CCD.

33

The single crystal of CsNa0_9gBi4,01Te7 was collected on a similar machine at 173 K. The

SMART software was used for the data acquisition and SAINT for data extraction and

reduction.
The intensity data for the crystals of RbSnBi3Te6, Rb0.94Cao,94Bi3,06Te6 and KSnSb3Te6
were collected at 100 K on a STOE IPDS-II diffractometer using graphite-

monochromatized Mo K, radiation. A numerical absorption correction was applied with
the program X-RED20 based on a crystal shape description. For the structure refinement,

the SHELXTL package of programs21 was used. To refine the occupancy of the mixed

metal sites, linear functions constraining the Bi and metal atoms on the various sites were
used. The first type of function was to constrain the sum of site occupancy factors (SOF)
in the metal sites to be equal to 1 (i.e., metals sites fiilly occupied). The second type of
constraint was to consider the sum of the formal charge of metal atoms multiplied by
their own SOF equal to 2 (i.e., average charge = +2).

Thermal analysis. In order to check the purity and the melting point of the compounds,
differential thermal analysis (DTA) was performed using a computer-controlled
Schirnadzu DTA-50 thermal analysis. A small quantity (~30 mg) of the crushed material
was loaded in a silica ampoule and flame-sealed under vacuum. Another silica tube
containing a similar amount of a-alumina was placed on the reference side of the
detector. The samples were heated to 750 °C at a rate of 10 °C/min, held at 750 °C for
one minute and cooled to 50 °C at a rate of -10 °C/min.

Electrical property measurements. The electrical conductivity of the samples was
measured using a 4-probe apparatus. The sample size was about 3 x 3 x 8 mm. Copper

wires and silver paste were used for the contacts. A MMR Technologies system was used

34

to measure the thermopower of the compounds. Samples were measured with reference

to a constantan wire (~ 1 x l x 4 mm).

2.3. Results and discussion

2.3.1. Structure description

CS0.741(0.76Bi3.5T€6, €50.69C30.6SBI3.34Te6i Rb0.82Pb0.8ZBi3.18Te6i
Rbo, 19K1_31Bi3,50Te6, RbSnBi3Te6 and Rbo_94Cao,94Bi3,06Te6 crystallize in the
orthorhombic space group Cmcm and are isostructural to CstBi3Te6. These compounds

can be expressed with the general formula A1-XM4Te6: (Cso,74K0.26)(K0,5Bi3,5)Te6,

CSO.69(C30.6SBi3-34)Te6a Rb0.82(Pb0.82Bi3.18)Te6i Rbo.19K0.81(K0.soBi3.50)T66,
Rb(SnBi3)Te6, Rb0.94(Ca0_94Bi3.06)Te6, Rb(YbBi3)Te6. The compounds exhibit a layered

structure with infinite anionic [M4Te6]' slabs separated by alkali metal cations (Figure 2-

1a). Each slab contains two crystallographically distinct metal sites M1 and M2, mixed
sites of bismuth and the other metal, and three crystallographically independent tellurium
sites. Both metal sites are octahedrally coordinated with Te atoms. The Te sites have
several coordination environment with Tel, Te2 and Te3 being respectively 2, 4 and 6-
coordinated with the metal atoms. The slabs can be viewed as a fragment excised from

the NaCl—structure type along the [011] directions with a thickness of four monolayers.

CsNao.9gBi4,01Te-; also crystallizes in the space group Cmcm but is isostructural to

CstzBi3Te7, the second member of the family (Figure 2-1b). In this case, the slabs are

one monolayer {NaCl} thicker. Each slab contains four crystallographically distinct

35

metal sites M1, M2, M3, and M4, all octahedrally coordinated by the tellurium atoms. In
both structures, the alkali metal sites are surrounded by nine Te atoms.

KSnSb3Te6 crystallizes in the orthorhombic space group ana (Figure 2-2). In

the structure, K+ cations separate [SnSb3Te6]' layers. Their arrangement differs from that
of the previously mentioned compounds in the orientation of the layers. For the crystals

crystallizing in the Cmcm space-group, the building blocks are NaCllOOtype whereas for

KSnSb3Te6 the layers are made of NaClm. This reorganization is probably due to the

smaller size of the K+ cation.

Figure 2-1. Structure type of (a) A1-XM4Te6 along the c axis and (b) of CsNa0,93Bi4,01Te7

along the a-axis.

36

. tr . . ‘ 1. Art”, 03b
1 VA -r’ “(21):- ""13- """ 1 osriSn
fl" {”1 x, t“ L .K
L s

Figure 2-2. Crystal structure of KSnSb3Te6 viewed along the b-axis.

The refinement data for all crystals are summarized in Tables 2-2 and 2-3. It must be
noticed that the position of Tel is close to a special position but the atom does not sit
exactly at this position. It is a split site that alternates between 2 positions close to the

special position

37

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

35% as; . - . . - . . - . A -213
- eee NNNa - ea. 32 Se N as as M a m use :o m as N as. am e 22 e5 omen Se 3.693
ANS N385 5 £83 ANS : Sod 5 8885 NNV N83 226568 8.853
825 1 ea? 286 u so? w: 3 u 29.3 :86 1 e23 NNNS 1 ea?
8665 m sea
mead nei $8.0 ”am $85 1 am 335 u xx 235 rum
8 2 S: N: 3: SN; music 26880
N n no 0333232 fiber: 25m @0508 Eofionmom
Read :23 88.0 685 eNmoe am
as E owe. 22 a3 888:2 82885
eN.wN 2 e: NQN a a: RN 3 a: MEN 9 SN SN 9 a: 88:8 smwvé ewes e
Ne: 39% ENS MES 3on p-85
E20508 E58022
33 on: NE SE SE memes bacon
N N N e N N
ANS GR 5 on: ANV o. as ANS 23: NC NE»: m6 85:5
E 33. ANV e83. g Q? g NE.N_ ANV 3:1.
5 SN GO erN E :.wN ANS 332 E eNSN
GS 23 NNV e23 Q 33 ANS 3 NE. amine 2v :8 :5
£050 98% 8QO
moo Essa 232m seesaw 23688885
$23 9: 582335
8N 8N 8N m: SN 00 assuage
3.82 3.22 $.82 Sen: 2 «NS amass 2:88
E 8 5 S 8

 

 

 

 

 

 

 

.3 ea. As .5 E .8 é 5% 632282328 6e cc .8583 .N-N 6:;

38

Table 2-3.

Summary of

the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

crystallographic data for RbSnBi3Te6 (6),
Rho-94Cao_94Bi3_06TC6 (7), RbeBi3Te6 (8) and KSnSb3T€6 (9).
(6) (7) (8) (9)
Formula weight 1523.10 1596.70 1651.05 1288.64
Temperature (K) 100 100 293 100
Wavelength (A) 0.71073 0.71073 0.71073 0.71073
Diffractometer STOE IPDS-II Siemens SMART STOE IPDS-II
Platform CCD
Space gmup Cmcm ana
Unit cell (A) 6.2722 (2) 6.2382 (2) 6.3470 (3) 26.610 (5)
28.151 (6) 28.284 (6) 28.280 (6) 4.2480 (8)
4.3562 (9) 4.3429 (9) 4.3990 (9) 12.845 (3)
Volume (A?) 769.2 (3) 766.3 (3) 789.96 (3) 1452.0 (5)
z 2 2 2 4
Density (mg/m7) 6.576 6.920 6.944 5.895
“5.0mm“ --1 49.294 50.263 53.081 19.278
coefticrent (mm )
9 range.” Ea“ 2.89 to 33.22 2.88 to 31.85 1.44 to 27.96 1.76 to 29.42
collecron( )
“1696‘?de 882 794 562 2224
reflections
Rt... 0.1441 0.1317 0.0473 0.1596
Refinement method Full matrix least-square on F2
Goodness of fitness 1.218 1.122 1.339 0.732
. . . R13 = 0.0498 R13 = 0.0560 Rf: 0.0356 R1a= 0.0419
F 1nal R mdrces b b b 1,
wk, = 0.1212 sz = 0.1431 sz = 0.1350 sz = 0.0942
Extinction
coefficient 0.000642 (2) 0.000395 (2) 0.00086 (2) none
Large“ d‘ff' pffk 4.905 and -6.490 4.457 and -7.118 3.325 and -2039 2.512 and —2.499
and hole (e.A )

 

 

 

 

 

 

 

aR1=Z

 

FL

 

 

IF.

 

 

/Z

 

 

F

0

 

 

wa. = {leuij ..ng /le(1~":>211%

39

The metal-Te and alkali metal-Te bond distances are summarized in Tables 2-4, 2-5 and

2-6 -

Table 2-4. Metal-tellurium and alkali site-tellurium bond distances (A) in compounds
(1), (3), (4), (5), (6), (7) and (8).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

Compound (1) (3) (4) (5)
M5521}? (M= Bi/K) (M=Bi/Ca) (M=Bi/Pb) (M=Bi/K)
Ml-Tel 2.9993 (17) 2.977(3) 2.9881 (15) 2.980(3)
M 1-Te2 3.2014(5) 3.180(4) 3.1701 (7) 3.166(3)
M 1-Te3 3.3818 (13) 3.355 (3) 3.3978 (12) 3.371 (3)
M2-Te2 3.1067 (15) 3.085 (3) 3.1076 (13) 3.095(3)
M2_Te3 3.1925 (5)- 3.172 94)- 3.1616 (6)- 3.159 (3)-
3.2231 (13) 3.196(3) 3.2428 (1 l L 3.208(3)
Alkali metal- _ = = =
Te bonds A—Cs A Cs A Rb A Rb
3.196(3)x2 3.168 (5)x2 3.212(5)x2 3.167 (6)x2
A-Tel 3.703 (2) x 2 3.683 (5) x 2 3.620 (4) x 2 3.628 (6) x 2
x 3.941 (2) x 3 3.914 (4) x 4 3.938 (4) x 4 3.912 (6) x 3
Compound (6) (7) (8)
M55223“ (M= Bi/Sn) (M=Bi/Ca) (M=Bi/Yb)
Ml-Tel 2.9542 (16) 2.9713 (7) 2.9869 (18)
M1-Te2 3.1414(7) 3.1473(6) 3.1846(7)
M1-Te3 3.3817 (13) 3.3754(7) 3.3951 (14)
N2-Te2 3.0691 (14) 3.0809(7) 3.1040 (14)
3.1288 (7)- 3.1390 (6)- 3.1762 (7)-
A1 M2363 3.1994 (13) 3.2037(7) 3.2216 (13)
kali metal- _ _ _
Te bonds A—Rb A—Rb A Rb
3.208 (6) x 2 3.228 (3) x 2 3.234 (8) x 2
A-Tel 3.571 (6) x 2 3.556 (2) x 2 3.620 (8) x 2
3.909 (6) x 4 3.927 (2)x 4 3.9mm) x 4

 

 

 

 

 

 

 

1“ all the compounds isostructural to CstBi3Te6, the distance between the metal and the

tellurium sites are comparable to those in CstBi3Te6 (average M-Te distance 3.173 (1)

4O

A), 14 The bond distances in CsNaoggBiamTe7 are in the same range as those in

CstzBi3Te7.

Table 2-5. Metal—Te and alkali metal-Tel bond distances (A) in compound (2)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Metal-Te bond Distance (A)

Bil-Tel 2.9890 (10)

Bil-Te3 3.1754 (11)

Bil-Te2 3.2280 (12)

Bil-Te4 3.3831 (11)

BiZ-TCZ 3.0829 (14)

Bi2-Te4 3.2174 (13)

Bi2-Te5 3.2905 (15)

Bi3-Te3 3.0607 (12)

Bi3-Te4 3.1708 (13)

Bi3-Te5 3.2593 (13)

_¥ Bi4-Te4 3.1549 (10)

\ Bi4—Te5 3.1888 (11)
3.7270 (15) x 2
CS‘Tel 3.9295 (1 2) x 4

 

 

Table 2-6. Metal-Te and alkali metal-Te bond distances (A) in compound (9).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Metal-Te bond Distance (A)

Sbl-Te3 3.0149 (16)
Sbl-TeS 3.037 (2)

Sb] -Te2 3.0946 (16)

Sb2-Te4 3.0132 (17)
Sb2-Te2 3.148 (2)

Sb2-Te5 3.1573 (17)

Ml-Te6 2.8981 (15)

i\ M1-Te2 3.3280 (16)
i\ M1-Te3 3.35342)

\ M2-Te1 2.9776 (16)

\ M2-Te5 3.1824 (17)
\ K-Tel 3.842 (6)
\ K-Te3 3.396 (6)
K K-Te6 3.899 (6)

 

41

 

 

Atomic coordinates, equivalent isotropic displacement parameters and occupancy

refinements for the metal sites for all compounds are reported on Tables 2-7, 2-8 and 2-9.

Table 2-7. Atomic coordinates and equivalent isotropic displacement parameters (A2 x

103) of the metal and alkali metal positions for (l), (3), (4), (5), (6), (7) and (8). U(eq) is

defined as one third of the traces of the orthogonalized Uij tensors.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Position x y z U(eq) Occupancy
M1
(1), Bi/K 0.5 0.3800(1) 0.75 23 (1) 0904/0096
(3), Bi/Ca 0.5 0.3798 (D 0.75 21 (1) 0855/0145
(4) Bi/Pb 0.5 0.3811 (10 0.75 24(1) 0826/0174
(5), Bi/K 0.5 0.3808 (1) 0.75 21 (10 0905/0095
(6), Bi/Sn 0.5 0.3814(1) 0.75 15 (1) 0751/0249
_ (7), Bi/Ca 0.5 0.3810(1) 0.75 5 (1) 0766/0234
‘L8), Bi/Yb 0.5 0.3808 (1) 0.75 24(1) 0.833/0.167
M2
4), Bi/K 0 0.2934(1) 0.25 24 (1) 0847/0153
\Q), Bi/Ca 0 0.2933 (1) 0.25 23 Q) 0813/0187
4) Bi/Pb 0 0.2943 91) 0.25 27(1) 0763/0237
5 , Bi/K 0 0.2942 (1) 0.25 24(1) 0846/0154
6 , Bi/Sn 0 0.2937 (1) 0.25 17 (1) 0755/0245
‘7 , Bi/Ca 0 0.2941 (1) 0.25 8 (1) 0765/0235
A 8 , Bi/Yb 0 0.2939(1) 0.25 25 (1) 0667/0333
1 , Cs/K 0 0.4971 (1) 0.25 40(1) 0740/0260
(3), Cs 0 0.4969 (2) 0.25 27 (1) 0.693
(4), Rb 0 0.4977 (3) 0.25 112 (4) 0.821
5), Rb/K 0 0.4070 (4) 0.25 40 (3) 0191/0809
(6), Rb 0 0.4983 (4) 0.25 101 (7) 0.791
(7), Rb 0 0.4983 (2) 0.25 120(4) 0.791
(8), Rb 0 0.4975 (4) 0.25 114(4) 1

 

 

 

 

 

 

42

 

Table 2-8. Atomic coordinates and equivalent isotropic displacement parameters (A2 x

103) for (2). U(eq) is defined as one third of the traces of the orthogonalized Uij tensors.

 

 

 

 

 

 

 

Position x y z U(eq) Occupaniy
M1, Bi/Na 0.5 0.1465 (1) 0.4967(1) 15 (1) 0945/0055
M2, Bi/Na 0 0.787(1) 0.75 18 (1) 0.588/0.412
M3, Bi/Na 0 0.684 (1) 0.25 17 (1) 0827/0173
M4, Bi/Na 0.5 0 0.5 15 (1) 0702/0298
Cs 0.5 0.2523 (1) 0.25 22 (I) l

 

 

 

 

 

 

Table 2-9. Atomic coordinates and equivalent isotropic displacement parameters (A2 x

103) for (9). U(eq) is defined as one third of the traces of the orthogonalized Uij tensors.

 

 

 

 

 

 

 

 

 

 

 

 

Position x y z U(eq) Occupancy
Sbl 0.4801 (1) -0.75 0.8601 (1) 14(1) 1
Sb2 0.3978(1) 0.25 0.3114(1) 24(1) 1
M1, Sn/Sb 0.3461 (1) -0.25 0.389 (1) 12 (1) 0513/0487
M2, Sn/Sb 0.4322 (10 -0.25 0.5874(1) 16(1) 0.557/0.443
K 0.2822(1) -O.25 0.7271 (5) 41 (2) 1

 

For compounds (1), (3), (4), (5), (6), (7) and (8), refinement data indicated all sites to be

mixed occupied. In the case of (1) and (5), the alkali sites are also mixed-occupied with

K. K has a larger ionic radius than the other elements inserted in the structure; its radius

is close to that of Cs and Rb. However, in the system Rb/K/Bi/T e, there is less Rb and

more K in the formula. The inverse situation was found in the system Cst/Bi/Te. This

can be due to the fact that the size of K is closer to that of Rb than that of Cs. K can more

easily replace Rb than Cs. The alkali metal sites are not fillly occupied in (3), (4) and (6).

The substitutions show the flexibility of CstBi3Te6 and CstzBi3Te7 structure types

towards chemical modifications and that the divalent Pb atom can be replaced by atoms

with slightly smaller (Na, Ca, Yb ) and larger (K) atomic radii. The alkali metals located

 

 

between the layers show very large thermal displacement parameters indicating that they

may be rattling in their local environment.

2.3.2. Characterization
Cso,74Ko_76Bi3_5Te6. The comparison between the experimental and calculated X-ray
powder diffraction patterns showed a good agreement (Figure 2-3a). EDS measurements
on five needles gave the average formula C8052K022Bi320Te6. However the DTA
analysis (Figure 2-3b) showed two melting peaks at 821 and 831 K. Upon cooling, only
one recrystallization peak appears at 807 K. The second run showed two melting points
(543 and 841 K) and two crystallization points (807 and 834 K). This is probably due to a
melting/decomposition process happening during the heating process. The melting point
at 821 K indicates the presence of CsBi4Te6 as a second phase. This can be expected as
the ternary and quaternary phases have similar crystal structure. The X-ray patterns taken

afier DTA are similar to those taken before.

 

 

   

 

 

 

 

 

 

200 —(1)calculated patterns for(1) 100

180JI—(2)expen'mentalp emsfor(1)(a) J —1Sl run (b)
g: 160- —— (3)CstBi3Te6 80 2nd run
2 140-——(4)CsBi4Tee
g 120-
" 100-
a)
%i 80-
E 60-
a: 40-

20- 0- 4/ "

0 I I If I I ‘r l I ' I I I
0 10 20 30 40 50 400 500 600 700 800 900
2 theta (deg) Temperature (K)

Figure 2-3. (a) Comparison between the calculated and experimental powder X-ray

diffraction patterns of compound (1) and and (b) DTA results for compound (1).

44

A large scale ingot was prepared for preliminary evaluation of the thermoelectric
properties of the compound. The tube was placed vertically in a tube furnace in order to
grow oriented needles. Another good technique to grow oriented needles is the Bridgman
method. In this technique, the elements are loaded in a silica tube terminated by a tip and

are placed in a temperature gradient. This technique has been successfully applied for

needle-shaped selenide compounds. However it was not successful for Cs0_74K0,76

Bi3.5Te6 because cesium reacted with the silica tube. Examination under microscope of
the ingot obtained with the tube in the vertical position revealed bunches of oriented
needles. However they were not well oriented with each other. Powder patterns were
taken from the top and bottom of the ingot. The theoretical diffraction pattern calculated
from the single crystal refinement, the experimental pattern from the first ingot, and that

of the larger ingot are similar. An extra peak at 17° was observed from the powder from

the bottom of the tube and indicates the presence of BizTe3 as a side product.

45

 

140
120-
100-

 

i .
. (4) 4
, (3) i 1 1 - "I

« u!
1(1)

-5 O 5 1015202530 35404550
2 theta (deg)

C) (D
O O
I l

Relative intensity
.5
O

 

 

N
O O
J

 

Figure 24. Comparison between the powder X-ray diffraction patterns of ground

samples taken from bottom (1) and top (2) of the ingot with nominal composition

Cs0,76K0.74Bi3,5Te6 prepared in the vertical furnace, the calculated powder pattern (4) and

that of CsBi4Te6 (3).

Single crystal diffraction refinement on a needle from the large ingot indicated the
composition Cso_6Ko.9Bi3_5Te6. A single melting and single crystallization peaks (at 850

and 828 K respectively) were observed on DTA data from the top of the ingot (Figure 2-

5a). However, DTA analysis from the bottom showed two melting points (815 and 839

K), the first corresponding to CsBi4Te6 (Figure 2-5a).

46

120

 

 

 

 

 

 

 

 

 

 

 

. —1 t (a) | . ——1strun (b)
80 , 2:16:31 . 100 - 2nd run
60 - top of the ingot aoj bottom Of the ingot
S . 9 .
3; 4O - i exo :5 60 . I
< . 40 _ exo
1- l endo . '- . , »-"
o 20: /)F o 20. (endo /N/
0- 0 - ... ._---*
400 ' 560 ' 560 T760 f 3E) 79% f1ooo 400 r500 . 600 . 700 . 800 I 900 V1000
Temperature (K) Temperature (K)

Figure 2-5. DTA results on powders taken from (a) the top and (b) the bottom of the

material Cso,6Ko,9Bi3,5Te6 prepared in a vertical furnace.

CsNao,9gBi4,01Te7. Attempts to prepare the compound in a silica tube were not

successful; the reactions resulted in a mixture of needles (CsBi4Te6), plates (BizTe3), and

chunks containing all four elements. From EDS analysis, some needles showed
quaternary composition, but no suitable needle was found for single crystal diffraction. A
suitable needle was found from the Cs/Na side of the H-tube for single crystal diffraction.
Comparison between the X-ray powder patterns from the Cs/Na side and calculated
patterns from data refinement show good agreement. Materials prepared with cesium in
one side and the other elements in the other side also contained needles randomly
oriented on the top of the material. The. X-ray powder patterns from the two preparations

(Cs/Na or Cs only in one side) are similar.
€50,69Ca0,65Bi3_34Te6. Attack of the silica tube by reagents was observed after the

reaction. The material was not well-formed and had needles grown on the walls of the

tube. The experimental and calculated X-ray powder diffraction agreed well. EDS

47

analysis on several needles (Cso,6oCa0,zzBi3,35Te6) confirmed the presence of all four

elements.

 

 

 

 

 

‘20 ‘ ' l j I ' r ' I r I ' I
400 500 600 700 800 9001000
Temperature (K)

 

Figure 2—6. DTA results of the batch from which the compound Cso_69Ca0.65Bi3.34Te6

was synthesized.

The first run of DTA (Figure 2-6) showed a melting point at 820 K (CsBi4Te6) and a

crystallization point at 842 K. The bump ~ 700 K indicates the presence of impurities.
The fact that the melting point is lower than the crystallization point indicates that the
impurities reacted with the main product. The second run shows a melting point at 863 K

and a crystallization point at 846 K. The powder patterns after DTA corresponded to
BizTe3.

Rbo,32Pbo.82Bi3_13Te6. Comparison between experimental and calculated X-ray powder
diffraction showed the presence of an extra peak on the experimental data at 29 = 10° that

may be attributed to the presence of the BizTe3 phase. EDS analysis confirmed the

48

presence of all four elements in the needles but the average composition
Rb1_gle0_51Bi2,96Te6 shows a higher Rb/Pb ratio.

In order to have a first evaluation of the thermoelectric properties, an experiment using
the Bridgman technique resulted in an ingot with well-oriented needles. A very thin layer
of impurity about 1 mm thick could be seen on the top of the ingot. The X-ray powder
diffraction patterns indicated that the layer is a mixture of Rbo_32Pbo,82Bi3_lgTe6 and
Pb1_71Bi6_36Te1222 (Figure 2-7a). Comparison between the experimental X-ray powder
patterns of other parts of the ingot and the calculated ones showed good agreement but
the presence of three melting points at 828, 839 and at 846 K on the DTA results reveals

the presence of side-products (Figure 2-7b).

 

Figure 2—7. Picture of the ingot with composition Rbo_82Pb0_82Bi3.lgTe6 prepared with the

Bridgman technique.

49

_L
N
O

 

 

   

1001 —1strun
9 2nd run
2 80-

9

E 601

a) ( Ma

1% 404 .4 -"\

E 20- 1 exo \
0 i endo

 

 

 

 

 

 

I I I A I H I “fl I» "7 —20 I I . I . I v I r
0 10 20 30 40 50 60 400 500 600 700 800 9001000
2 theta (deg) Temperature (K)

Figure 2-8. (a) Comparison between X-ray difiraction patterns of powders from bottom
(1), middle (2) and top (3) of the ingot with composition (4) prepared by Bridgman
technique and the calculated X-ray diffraction patterns of Pb1_77Bi6.36Te12 (4), BizTe3 (5)

and compound (4) (6); (b) DTA analysis of the powder from the middle of the ingot.

Rb0_19K1,31Bi3_50Te6. The experimental and calculated XRPD patterns of this
composition showed good agreement. However EDS analysis on several needles show
higher Rb/K ratio with an average composition of RbKo,77Bi4.00Te5_-;4.

Rbo,94Cao_94Big_06Te5. The experimental and calculated powder X-ray diffraction
patterns agreed well. EDS analysis on several needles showed the composition
Rb0_83Cao,41Bi3_9Te6. DTA analysis indicates the presence of impurities. Four melting
points are observed during the first run of a DTA measurement: one at 813 K (close to
the melting point of RbBi3.66Te613), a shoulder at 834 K, a large peak at 849 K and a very

small peak at 861 K (B12T63) but only one crystallization point is visible at 840 K. The

50

second run shows only one melting point at 867 K and one crystallization point at 849 K.
X-ray powder diffraction on the sample indicated only the presence of BizTe3.

RbSnBi3Te6. Comparison between calculated and experimental XRPD showed a good
agreement but two extra peaks were visible in the experimental spectra at 103° and
187°. DTA analysis confirmed the presence of two phases. EDS analysis on the needle
gave the composition Rb1_44Sn0.55Bi3_27Te6. Two melting points at 822 K (close to the
melting point of RbBi3l66Te6) and at 834 K and two crystallization points at 822 and 835

K are observed during the first ramp of the DTA measurements. A second run gave

similar results. Powder patterns afier DTA are similar to that before DTA.

 

 

100 ' 50

 

 

 

 

 

 

 

 

(a) —1st run (b)
80- 40_ 2nd run .
o - . . ”'
3.5 60 g 30 fill.
:23 40- 3) E 20- I
8 20 1(2) 1 I i
. . . exo
0: J1) 1 l. i 10 l
0 . . . r . ! endo
0 10 20 30 40 50 60 400 500 600 700 800 9001000
2 theta (deg) Temperature (K)

Figure 2-9. (a) Comparison between the experimental (3) and calculated (2) powder X-
ray diffraction patterns for compound (6) and that of BiTe (3) and (b) DTA analysis of

the powder from the middle of the ingot.

RbeBi3Te6. The comparison between calculated and experimental powder X-ray

diffraction presents a good agreement. EDS analysis on the needles gave the average

51

composition Rbe0.16Bi3.14Te5,27. DTA analysis showed a single melting point at 861 K
followed by a single crystallization point at 846 K.

KSnSb3Te6. Figure 2-10 shows a SEM picture of the needle used fro single-crystal X-ray
diffraction. EDS analysis on the needle gave the composition KSHO27Sb3TCS_]9. The

content of tin is low. There is an uncertainty on the relative ratios of Sn/Sb/T e because all

three elements have close L X-ray emission lines.

 

Figure 2-10. SEM image of the needle with composition KSnSbgTe6.

The attempts to substitute lead by silver, strontium and barium have not been
successful. The ionic radius of octahedral Ag is 1.15A, which is close to that of lead
(1.19A). It should be possible to replace some Bi by Ag atoms in the metal sites. When

Pb was replaced by Ag, only a mixture of BizTe3 and RbAggTe223 was obtained.

Substituting Pb with Ba gave shiny BaBiTe3 rods.”26 EDS analysis from needles

52

obtained fi'om a mixture of Rb, Sr, Bi and Te indicated only the ternary phase

RbBl3_65Te6.

2.3.3. Preliminary Thermoelectric Results

This family of compounds combines numerous structural features that make them
interesting for thermoelectric investigations. These are low dimensionality, preferential
site occupancy of metal positions and isomorphic substitutions of Pb by a wide range of
elements with different charges and atomic radii. In addition, the alkali metals located.
between the layers show very large thermal displacement parameters indicating that they
may be rattling in their local environment, which may contribute in suppressing the

thermal conductivity of the compounds.

Preliminary charge transport properties have been measured on Cso_74K0.76Bi3,5Te6 (1)

and Rb0,32Pbo,32Bi3_18Te6 (4) and are reported in Table 2-10. These two compositions

were selected because of the better quality of the crystals.

Table 2-10. Preliminary thermoelectric measurements on the compounds (1) and (4).

 

 

 

 

 

 

 

 

 

 

 

 

1) (4 ** CstBi3Te6
top bottom top bottom
Electrical
conductivity 601 986 1437 1811 830
o (S/cm)
Thermopower S
-26 -37 -20 -22 ~40
(pV/K) 2
Power factor (= S o)
0.4 . . 0. 7 1. 2
(uW/K.cm2) 06 1350 0575 8 7 3 8

 

*ingot prepared with a regular furnace in vertical position

“ingot prepared with the Bridgman method

53

2.4. Conclusions

This study showed it is possible to substitute Pb by other elements such as Na, K, Ca, Sn,

Yb in CstBi3Te6, the first member of the homologous family CstmBi3Te5+m.
€50.74KO.76Bi3.5Te6a €50.69C30.6SBi3.34Te6a Rb0.82Pb0.8zBi3.18Te6, Rb0.191(1.3113i3.50'1'€=6,
RbSnBi3Te6 and Rb0_94Ca0,94Bi3,06Te6 are isostructural to CstBi3Te6 whereas

CsNa0_93Bi4,01Te7 is isostructural to CstzBi3Te7, the second member of the family.
Substitution of Cs by K resulted in a reorganization of the layers around the alkali metal.

The crystal structure of KSnSb3Te6 is similar to that of the selenides CsAg0.5Bi3_5Se6 and
CstBi3Se6.19 However these compounds can not be prepared as pure phases so far.
Alkali tellurides and the ternary phases CsBi4Te6 and RbBi3.66Te6 were identified as side

products. The quality of the needles also was quite poor. Using AzTe as starting materials

did not help improving the yield of the reactions.

54

References

1. Jeon, H.-W.; Ha, H.-P.; Hyun, D.-B.; Shim, J.-D., J. Phys. Chem. Solids 1991, 52,
579.

2. Champness, C. H.; Chiang, P. T.; Parekh, P., Can. J. Phys. 1965, 43, 653.

3. Champness, C. H.; Chiang, P. T.; Parekh, P., Can. J. Phys. 1967, 45, 3611.

4. Chung, D. Y.; Choi, K. S.; Iordanidis, L.; Schindler, J. L.; Brazis, P. W.;
Kannewurf, C. R.; Chen, B.; Hu, 8.; Uher, C.; Kanatzidis, M. G., Chem. Mater. 1997, 9,
3060.

5. Kyratsi, T.; Chung, D.-Y.; Kanatzidis, M. G., J. Alloys Compd. 2002, 338, (1-2),
36-42.

6. Kyratsi, T.; Dyck, J. S. ; Chen, W.; Chung, D.-Y.; Uher, C.; Paraskevopoulos, K.
M.; Kanatzidis, M. G., Journal of Applied Physics 2002, 92, (2), 965-975.

7. Kyratsi, T.; Chung, D.-Y.; Ireland, J. R.; Kannewurf, C. R.; Kanatzidis, M. G.,
Chemistry of Materials 2003, 15, (15), 3035-3040.

8. Kyratsi, T.; Kanatzidis, M. G., Z. Anorg. Allg. Chem. 2003, 629, (12-13), 2222-
2228.

9. Kyratsi, T. ; Hatzikraniotis, E.; Paraskevopoulos, K. M.; Malliakas, C. D.; Dyck, J.
S.; Uher, C.; Kanatzidis, M. G., Journal of Applied Physics 2006, 100, (12), 123704/1-
123704/11.

10. Chung, D.-Y.; Hogan, T.; Brazis, P.; Rocci-Lane, M.; Kannewurf, C.; Bastea, M.;
Uher, C.; Kanatzidis, M. G., Science 2000, 287, 1024.

ll. Chung, D.-Y.; Hogan, T. P.; Rocci-Lane, M.; Brazis, P.; Ireland, J. R.;
Kannewurf, C. R.; Bastea, M.; Uher, C.; Kanatzidis, M. G., J. Am. Chem. Soc. 2004, 126,
6414.

12. Larson, P.; Mahanti, S. D.; Chung, D.-Y.; Kanatzidis, M. G., Physical Review B
2002, 65, 045205.

55

13. Kanatzidis, M. G., Semicond. Semimet. 2000, 69, 51.

14. Hsu, K. F .; Chung, D. Y.; Lal, S.; Mrotzek, A.; Kyratsi, T.; Hogan, T.;
Kanatzidis, M. G., J. Am. Chem. Soc. 2002, 124, 2410.

15. Hsu, K. F.; Lal, S.; Hogan, T.; Kanatzidis, M. G., Chem. Commun. 2002, 13,
1380.

16. Magneli, A., Acta Crystallogr. 1953, 6, 495.

17. Mrotzek, A.; Kanatzidis, M. G., Acc. Chem. Res. 2003, 36, 111.

18. Kanatzidis, M. G., Acc. Chem. Res. 2005, 38, 359.

19. Kim, J. H.; Chung, D. Y.; Kanatzidis, M. G., Chem. Commun. 2006, 15, 1628.

20. Darmstadt, Germany, 2001.

21. Sheldrick, G. M. SHELXTL NT, 5.1; Bruker: 1998.

22. Zhukova, T. B.; Zaslavskii, A. I., Kristallografiya 1971, 16, 918.

23. Eanes, M. E.; Schimek, G. L.; Kolis, J. W., J. Chem. Crystallogr. 2000, 30, (4),
223-226.

24. Cook, R.; Schaefer, H., Stud. Inorg. Chem. 1983, 3, (Solid State Chem), 757-60.

25. Chung, D.-Y.; Jobic, S.; Hogan, T.; Kannewurf, C. R.; Brec, R.; Rouxel, J .;
Kanatzidis, M. G., Journal of the American Chemical Society 1997, 119, (10), 2505-
2515.

26. Chung, D.-Y.; Jobic, S.; Hogan, T.; Kannewurf, C. R.; Brec, R.; Rouxel, J .;
Kanatzidis, M. G., Mater. Res. Soc. Symp. Proc. 1997, 453, (Solid-State Chemistry of
Inorganic Materials), 1522.

56

Chapter3

Preparation and Characterization of Members of the

Series N anmeTem+2 with m=6, 8, 12

3.1. Introduction

Doped PbTe is used as a thermoelectric material in power generation devices

running in temperature conditions ~ 700 K. Both p-type and n-type thermoelements can
be produced by d0ping with acceptors (e.g., NazTe, KzTe or Ange)l'4 or donors (e.g.,
Pblz, PbBrz).5 The ZT value of PbTe solid solutions is low near room temperature but
rises to ~ 0.8 at 700 K. Several studies on low dimensional nanostructured systems have
shown significant improvement in the thermoelectric figure of merit, mainly through a
reduction of the lattice thermal conductivity while maintaining a high power factor. For

example, high ZT values, ranging from ZT ~ 1.6 at 300 K to ZT ~ 3 at 550 K, were

reported by Harman and coworkers for Bi-doped n-type PbSe-PbTe quantum-dot
superlattices grown by molecular beam epitaxy.6 However, the cost of such fabrication

processes is quite high for mass production. Nanostructured bulk analogs of such systems
would be more sustainable. Excess Pb in PbTe results in Pb precipitates with sizes on the

order 30—40 nm in the PbTe matrix and an increase in the thermpower compared to that of

. . 7
pure PbTe for the same earner concentratlon was observed.

57

Our group is investigating bulk analogs of such devices, meaning bulk materials

containing nanostructures. We reported first on the n-type system AngmeTem+2 (so-
called LAST).8 Several members of that system exhibit a figure of merit above 1 at high

temperature. ZT ~ 1.7 at 700 K was obtained for Anglgstezo. This value is achieved

mainly through low total thermal conductivity (~ 1.1 W/m-K at 700 K). Careful TEM
studies on the samples indicated the presence of nanostructures embedded in the PbTe
matrix. These nanostructures scatter phonons more efficiently than charge carriers, thus
reducing the lattice thermal conductivity without any significant impact on the electronic

properties. We used the LAST system as a platform for optimization. Substituting Ag by
the alkali metal Na resulted in p-type materials.9 To understand the reason for the change
of charge carriers, ab initio calculations within the density functional theory were carried
out.Io The study showed an increase in the density of states near the top of the valence
band for PbTe doped with Ag whereas no change in the DOS within 0.5 eV of the band

maxima was observed for Na-doped PbTe. This new system, Na1-bemeyTem+2, also

holds promise as thermoelectric materials. ZT ~ 1.6 was reported for Na0,95Pb20SbTe22 at

650 K, which is nearly twice that of p-type PbTe. Low total thermal conductivity ~ 0.85
W/m-K at 700 K was obtained for that composition. TEM analysis on specimens also
indicated the presence of nanocrystals embedded in the PbTe matrix. In this chapter, we
describe the synthesis of members with low m values (m=6, 8, 12), members which were
not studied in the reference 9. Structural and thermoelectric characterization results are

also reported.

58

3.2. Experimental section

3.2.1. Synthesis

All samples were prepared as polycrystalline ingots by mixing high purity
elements in the appropriate stoichiometric ratio in 10 mm outer diameter fused silica
tubes. Prior to use, the tubes were carbon-coated in order to avoid reaction between the
sodium metal and the glass. All components (except Na) were loaded into the tubes under
ambient atmosphere and the corresponding amount of Na was later added under nitrogen

atmosphere in a dry nitrogen glove box. The silica tubes were then flame-sealed under a
residual pressure of ~10—4 Torr and placed into a tube furnace (mounted on a rocking

table). Cast ingots were first prepared following the same temperature profile as the
parents system Na1-bemeyTem+2 (method 1).9 The ingots were found quite brittle. As
the m=8 composition exhibited the more promising thermoelectric properties, other
synthesis conditions (methods 2-9) were investigated to try to improve its mechanical
properties.

Method 1. Ingots with nominal compositions Nan6SbTe3, Nan3SbTe10, NanlzsteM

were prepared using the same cooling profile as the parent system Na1_bemeyTem+2.

Namely, the furnace was heated to 1250 K for 4 h to allow complete melting of all
components. While molten, the furnace was rocked for 2 h to facilitate complete mixing
and homogeneity of the liquid phase. The furnace was finally immobilized at a vertical
position and was cooled from 1250 to 820 K over 43 h followed by a fast cool to room

temperature. The resulting ingots were silvery-metallic in color with a smooth surface.

59

The samples were quite brittle. Different strategies, described below, were attempted to
improve the mechanical strength of the samples.

Method 2. Slow cooling profile. Stoichiometric amounts of the elements were -loaded in
a carbon-coated tube. After being sealed, the tube was introduced in a fumace mounted
on a rocking table. The furnace was heated overnight to 1253 K and kept at that
temperature 24 hours. During that period, the furnace was rocked for 4 hours. Then the
tube was slowly cooled down to 773 K over 99 hours and kept at that temperature 12
hours. Finally the furnace was cooled down to 323 K fast.

Method 3. Slow cooling profile. Stoichiometric amounts of the elements were -loaded in
a carbon-coated tube. After sealing, the tube was introduced in a furnace mounted on a
rocking table. The fumace was heated overnight to 1253 K and kept at that temperature 4
hours. During that period, the furnace was rocked for 2 hours. Then the tube was slowly
cooled down to 973 K over 140 hours. The furnace was then cooled to 323 K in 12 hours.
Method 4. Faster cooling profile. Stoichiometric amounts of the elements were -loaded
in a carbon-coated tube. After sealing, the tube was introduced in a furnace mounted on a
rocking table. The furnace was heated overnight to 1273 K and kept at that temperature 6
hours. During that period, the furnace was rocked for 2 hours. Then the tube was cooled
down to 323 K over 48 hours.

Method 5. Faster cooling followed by slower cooling step. Stoichiometric amounts of the
elements were loaded in a carbon coated tube. The fiirnace was heated in 12 hours to
1253 K and kept at that temperature 4 hours. During that period, the furnace was rocked
for 2 hours. The furnace was then cooled down to 1023 K within an hour and down to

323 K within 48 hours.

60

Method 6. Addition of Sn to the system. Alloying Pb with Sn helped improving the
mechanical properties of the LAST material. In a first experiment, 5% of Pb was
substituted by Sn. The cooling profile was as described in method 1. In a different set of
experiments, 25, 50 and 75 % of Pb was substituted by Sn. The ingots were prepared
using the temperature profile described in method 5.

Method 7. Preparation of off-stoichiometric compositions. One reason of the low
mechanical strength of the samples may be a high concentration of nanostructures inside

the matrix. Several compositions with deficiency in Na and Sb were prepared:
Na0_5Pb3SbTe10, N30.5Pb83b0.75T€10. N305Pb83b05T€10 and Nao.5Pb8Sb0.25T€lo- The
temperature profile used was as described in method 1.

Method 8. Use of NazTe as starting material. After reactions, peelings in the carbon

coating are often visible along the top/middle part of the tube. For this experiment, NazTe
was used as starting material: as Na metal already reacted with Te, less glass attack issue
is expected. The temperature profile was the one described in method 1.

Method 9. Hot-press experiments. First, ingots with nominal composition Nangstelo

were synthesized using the method described in part I. The ingots were then grinded in
air using mortar and pestle. Particles with size S 125 um were collected using sieves. A
first pellet was prepared using a pressure ~ 10, 000 pounds. House vacuum was applied
to the system for 15 min before supplying heat to the press. The chamber was heated to
373 K in 8 min and kept at that temperature 30 min. The furnace was then turned off. The
resulting pellet (named HPl) was 1.98 mm thick with a diameter of 12.77 mm. A second
pellet was prepared using the same method but the press was kept at 423 K. The diameter

of the final pellet (HP2) was 12.78 mm and its thickness 1.71 mm. For the third pellet

61

1le

111

P1

161'

sen
D1

D1

113

(HP3), the ingot was grinded into a fine powder in a glove-box under nitrogen
atmosphere. Particles with size S 46 um were collecting using sieves. House vacuum was
applied to the system for 15 min before starting heating the press. The chamber was
heated to 473 K in 15 min and kept at that temperature for 30 min. The furnace was then
turned off. The final pellet had a diameter of 12.78 mm and was 1.86 mm thick. A

rectangular sample was cut from the pellet using a wire saw and was annealed in a quartz

tube at 400 °C for 2 hrs.

3.2.2. Characterization techniques
Powder X-ray Diffraction. Powder X-ray patterns of the grinded materials were
recorded using Cu K, radiation (A = 1.54056 A) in reflection geometry on a CPS-120 Inel
X-ray powder diffractometer operating at 40 kV and 20 mA equipped with a position
sensitive detector.
DTA Analysis. Differential thermal analysis (DTA) data were collected with a Shimadzu

DTA-50 thermal analyzer. Approximately 35 mg of finely ground powder of material

Was sealed in a carbon-coated quartz ampoule under residual pressure of ~ 10"4 torr.

Another ampoule containing similar amount of alumina and prepared the same way was
used as a reference. The samples were heated to 1273 K at a rate of 10 K/min, held at
1 273 K for 2 min and cooled to 323 K at a rate of -10 K/min.

Scanning Electron Microscopy (SEM). The surface of several samples was polished
Very carefully using silica suspension solution (0.05 um) in order to get a smooth, mirror-

like surface. The samples were then studied with a scanning electron microscope (Hitachi

62

S3400N—II) with 25 kV acceleration voltage using both energy-dispersive spectroscopy
(EDS) and back-scattered electron imaging (BSE).
Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES). The

amounts of Na and Sb from five powdered samples from the top, middle and bottom part
of an ingot with composition Nangstelo were studied by ICP-ABS using a Varian

model ICP spectrometer. ~ 20 mg of the samples were dissolved in 8.4 mL of freshly
prepared aqua regia. Then 1.6 mL of ultra pure water was added to the solution. The final
solution was diluted 10 times with ultra pure water. Five standards were prepared for the
calibration. The concentrations of the standards are reported in Table 3-1. The
wavelength used for Na was 589.592 nm and for Sb 206.834 and 217.582 nm. The
correlation factors were respectively 0.995690, 0.999292 and 0.999264. A blank

containing the same amount of acid was measured as well.

Table 3-1. Concentrations of the standards used for the ICP-AES calibration.

 

 

 

 

 

 

 

Standard Concentration of Na (ppm) Concentration of Sb (ppm)
1 1.007 5
2 1.24868 6
3 1.49036 7
4 1.73204 8
5 2.014 9

 

 

 

 

Electrical Transport Properties. Thermopower and electrical conductivity properties
were measured simultaneously under helium atmosphere using a ZEM-3 Seebeck
coefficient/electrical resistivity measurement system (ULVAC-RIKO, Japan). Samples
for transport measurement were cut to size 10 x 3 x 3 m using a diamond saw (Buehler

isomet 1000), a wire saw (South Bay Technology) and a polishing machine (Buehler

63

ecomet 3000). Rectangular shape samples with approximately 3 x 3 mm cross-section
were sandwiched vertically by two nickel electrodes (current injection) with two Pt/PtRh
thermocouples (for temperature difference and voltage measurements) attached on one
side. The sample and measurement probes were covered by a nickel can to maintain a
constant temperature during the measurement and the base temperature was measured by
a thermocouple attached to the outside of the can. The sample, electrodes, and nickel can
were placed in a vacuum chamber then evacuated and refilled with He gas (0.1 atrn) to
provide necessary heat transfer. Properties were measured from room temperature to 670
K under helium atmosphere. A

Thermal Conductivity. The thermal conductivity was determined as a function of
temperature using the flash diffusivity method on a LFA 457/2/G Microflash NETZSCH.
The front face of a small disc-shaped sample (diameter ~ 8 mm; thickness ~ 2mm) coated
with a thin layer of graphite is irradiated with a short laser burst, and the resulting rear
face temperature rise is recorded and analyzed. The experiments were carried out under

nitrogen atmosphere. Thermal conductivity values were calculated using the equation x:
a Cp d, where a is the thermal diffusivity, Cp the specific heat and d the bulk density of

the material calculated from the sample’s geometry and mass. A pyroceram reference
was used to determine the heat capacity of the sample. The thermal diffusivities were
measured typically over the temperature range 300-670 K. The electronic component of

the thermal conductivity was quantified through the Wiedemann-Franz law according to

which to, = o.T.L0 (L0 being the Lorenz number, L.,=2.45.10'8 WQK").“ The lattice

contribution was then derived by subtracting the electronic component from the total

thermal conductivity.

64

3.3. Results and discussion

3.3.1. Method 1
Structure and characterization. Figures 3-1a, 3-1c and 3-1e shows the X-ray
diffraction patterns from powders prepared from the compositions NanmeTem+2 (m=6,

8, 12). All the peaks observed can be indexed to the PbTe-structure. However some of the

peaks at high angle appear to be split (Figures 3-1b, 3-1d and 3-11). There may be two
reasons for that splitting. The splitting may result from the differentiation of the K011 and
K012 rays. It may be also the result of phase segregation between two phases with similar
lattice constants.

Thermal analysis. Figure 3-2 shows the DTA results for NanmeTem+2. The first and

second sets of measurements for NanmeTem+2 are similar and a single melting upon
heating and a single crystallization point upon cooling were observed. Table 3-2

summarizes the melting and crystallization temperature for NanmeTem+2 (m=6, 8, 12).

Table 3-2. Melting and crystallization points for NanngTem+2 (m=6, 8, 12).

 

 

 

 

 

 

 

 

m 6 8 12
T.,, (K) 1165 1180 1178
T,(K) 1111 1144 1149

 

65

60

J>
n

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—(1)top " _
50‘ (a) —(2) bottom 40-1 (b) ——:):ttom
Q Nanestea g. 35 [4an6st98
tn m ‘
C C
o .
E 40 g 30.
3 30- .“2’ 25‘
g 20‘ (2) I g 20-
m (1) I I I a: 151
10 I I I 1O I I I l I
20 40 60 80 100 80 85 90 95 100105110
2 theta (deg) 2 theta (deg)
60 60
(C) — lg trivii’tidle : :fddle
a. 50. N PbaSbTew a 50-(Nan88bTe10
.17) (7)
if: 40- 93 40-
.E E
2 30- Li g 30-
E (2) I E
a 20- IJ 32 20-
(1)
10 r " L._ r * ‘ 10 . . .
20 4O 60 80 100 60 65 70 75 80
2 theta (deg) 2 theta (deg)
60
4 (e) —(1)top 4O :Igopm
3‘ 504 —~(2)bottom 35_(
.6 Nantzsteu %‘ NanIZSbTe“
a 1 s 30-
g 404 ‘E 25
.2 30- I 3 l
a E 20-
g; 20_ (2’ . l l AA g
I 15%
(0 I
10 . 1 . H.“ lc—A—M‘J“
20 40 60 80100 80 85 90 95 100105110
2 theta (deg) 2 theta (deg)

Figure 3—1. Powder X-ray diffraction from samples from top and bottom part of the

ingots with composition (a) Nan6SbTeg, (b) the area between 20 = 80 and 110 deg is
enlarged to show peak splitting; (c) NanngTeIo, the area between 20 = 60 and 80 deg is
enlarged to show peak splitting; and (e) NanIZSbTeM, the area between 20 = 80 and 1 10

deg is enlarged to show peak splitting. All three specimens were prepared with method 1.

66

 

-200 -

 

 

— 1 st cycle
2nd cyoe
NaPDGSbTeg

 

 

 

400 ' 600 ' 800 '10'00'1200

Temperature (K)

 

 

   

 

 

 

 

 

0 1—K A
'20: l
a '40: .
< '50 '1 .
'5 ~-80: '
-100-I —1st cycle ’3
_ 2nd cycle \
133‘ NangsteIO \
400 600 800 1000 1200
Temperature (K)
0 41...,
3m: ‘11:“ (C)
-50 . \
S ' ‘ I x.
j -100 - i
1— . .
Q J
-150 - —1st cycle
2nd cycle \\
Nan SbTe
'200 f 1T2 f 14' - I ' j
400 600 800 1000 1200

Figure 3-2. Thermal analysis for (a) Nan6SbTeg,

Temperature (K)

Nan12SbTeI4 prepared with method 1.

67

 

 

(b) Nan3SbTeIo and(c)

Scanning electron microscopy. Finely-polished surfaces were analyzed by SEM. Very
disperse inclusions could be observed in the case of NanIszTeM. Figure 3-3 shows
pictures of two such inclusions observed on the sample used for thermoelectric
characterization. For a surface ~ 10 x 3 m2, only three round-shape inclusions with
diameter ~ 30 um were observed far apart from each other. For the sample used for
thermal diffusivity measurements, a single inclusion was observed. EDS analysis on the

different inclusions gave similar results: Sb~ 67 at. %, Te ~ 22 at. % and Pb ~ 1 at. %.
EDS analysis on the matrix indicated the presence of Pb and Te in ratio close to 1:1. BSE
analysis on samples Nangstelo and Nan6SbTe3 did not show evidence of a second
phase. EDS analysis on several areas indicated the presence of Na in the matrix (~ 5 at.
%) but no reliable information about Sb could be obtained because of overlapping
between Sb and Te peaks.

ICP-AES. As accurate amount of Na and Sb elements in the sample is not possible by
EDS, several powders from the ingot with composition NangsteIo were analyzed by
ICP-ABS. Tables 3-3 summarizes the elemental concentrations of Na and Sb in the
different samples and gives the Na/Sb ratio obtained. It can be observed that for the
samples from the bottom of the ingot, the Na/Sb ratio is lower than 1, indicating that Sb
is in excess compared to the nominal composition. For the samples from middle and top

parts of the ingot, the Na/Sb is ~ 1, close to the nominal composition.

68

 

Figure 3-3. BSE inages of one of the surface of the sample NanIgsteM used for

thermoelectric characterization. EDS analysis on the inclusions indicated the regions to

contain ~ 67 °/o Sb, ~ 22 Te and ~ 1% Pb. The matrix is pure PbTe.

69

In Table 3-4, the experimental and theoretical mass percentages of Na and Sb are

compared. For samples from the bottom and the middle part of the ingot, the

experimental percentages of Na and Sb are lower than the theoretical values. The

experimental percentages of the powders from the top of the ingot are higher than that of

the powders fi'om the other parts of the ingots and are closer to the theoretical values.

There is a gradient in Sb and Na concentrations along the ingot.

Table 3-3. Summary of the elemental concentrations of Na and Sb obtained with ICP-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

AES.
Concentration of Na Concentration of Sb
Na/Sb
(mmol) @mol)
Bottom part of the
ingot
Sample A 3.02 5.29 0.57
Sample B 2.92 5.17 0.56
Sample C 2.70 5.46 0.49
Sample D 2.58 5.07 0.51
Sample E 2.65 5.15 0.51
Middle part of the
ingot
Sample F 4.62 4.52 1.02
Sample G 4.75 4.66 1.02
Sample H 5.10 4.93 1.03
Sample I 5.17 4.83 1.07
Sample J 4.77 4.62 1.03
Top part of the ingot
Sample K 7.36 6.82 1.08
Sample L 8.45 7.89 1.07
Sample M 7.04 6.83 1.03
Sample N 7.55 6.92 1.09
Sample 0 6.04 5.95 1.02

 

 

 

70

 

Table 3-4. Comparison between experimental and theoretical mass percentages of Na

and Sb in the powders analyzed with ICP-AES.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mass % Na Mass % Na Mass % Sb Mass % Sb
(experimental) (theoretical) (experimental) (theoretical)
Bottom part of
the ingot
Sample A 0.74 0.75 2.82 3.96
Sample B 0.60 0.75 3.13 3.96
Sample C 0.58 0.75 3.31 3.96
Sample D 0.53 0.75 2.93 3.96
Sample E 0.54 0.75 2.96 3.96
Middle part of ‘
the ingot
Sample F 0.48 0.75 2.62 3.96
Sample G 0.50 0.75 2.61 3.96
Sample H 0.57 0.75 2.94 3.96
Sample I 0.54 0.75 2.66 3.96
Sample J 0.52 0.75 2.65 3.96
Top part of the
ingot
Sample K 0.75 0.75 3.67 3.96
Sample L 0.80 0.75 3.93 3.96
Sample M 0.66 0.75 3.69 3.96
Sample N 0.81 0.75 3.92 3.96
Sample 0 0.66 0.75 3.46 3.96

 

 

 

 

 

Optical band gap measurements. Infrared spectroscopy of NanmeTem+ 2 (m=6, 8,
12) shows that the three members are narrow gap semiconductors (Figure 3-4). The band

gap slightly decreases with increasing m value: Nan6SbTeg, NangsteIo and

NanIzsteM have band gaps ~ 0.43, 0.39 and 0.26 eV respectively.

71

 

 

 

  
  

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

48
16.0- (a)
. 471
15.5- 1
3’; 15.0- f; 45'
a) . a 1
3 14.5- 3 45.
140+ '1 E9 = 0.43 eV 44 _ E9 = 0.39 eV
13'5; NanGSbTea l NaPDBSbT81o
f I V 1 i I r I 43 r I f T ' I ' I
0'35 °°4° 0'45 0-50 0.30 0.35 0.40 0.45 0.50
Energy (eV) Energy (eV)
25- c d
( ) 2.025. ( )
A 1 I?
' 3
3. ...- 2.000-
m v
V 20- w
(I) B
3 1.975-
E9 = 0'26 W N Ergo: 0sallreV
NanIZSbTe“ 5' 18 020
15e.a.-..,,1.950.,-rr,,
o 25 0.30 o 35 0 40 0.30 0.35 0.40 0.45 0.50
Energy (eV) Energy (eV)

Figure 3-4. Diffuse reflectance absorption spectra of (a) Nan6SbTeg, (b) NanBSbTeIo,

(c) NanIZSbTeM and (d) NanIngTezo prepared with method 1.

Electronic Transport Properties. The temperature dependence of the electrical

conductivity for the compounds NanmeTem+2 (m=6, 8, 12, 14) is plotted on Figure 3-
5a. For comparison, the data collected previously for NanIgstezo and

Na0_95szoSbTe22 were also added to the graph. At room temperature, there is a
significant difference between the different members: the value ~1540 S/cm for

N%.95szoSbTe22 is almost three times the value for Nan6SbTe8 (~540 S/cm). At high

72

temperature, the electrical conductivity varies between 90 S/cm for Nan6SbTeg and

NanIZSbTeM and 170 S/cm for NanostzoSbTezz. Compositions with high m values tend

to have higher electrical conductivity with the exception of the m=8. This might be the
results of the quality of the samples. The m=6 and m=12 samples were more brittle than
the m=8.

All compositions showed p-type behavior (Figure 3-5b). Room temperature

thermopower varies between 96 uV/K for Na0_o5Pb20SbTe22 and 170 uV/K for
Nan6SbTeg. At temperatures above 650 K, all compositions have therrnopower values
above 300 uV/K, the highest value being ~ 360 uV/K at 670 K for Nan6SbTe3.

In order to check the values measured for the composition NangsteIO, the data were

recollected on the same sample. A good agreement between the two data sets was
observed (Figure 3-6).

Using the above data, the power factor values were calculated and are reported in
Figure 3-5c. Clearly the composition NaoostzoSbTegz exhibits the highest power factors
over the temperature range 300-675K with a maximum ~ 24 uW/m-K2 at 450 K.

NangstCIo and NanIgstezo have similar power factors with maximum values ~ 19

uW/m-K2 at 460 K.

73

 

 

 

 

 

E1600- (a)

(7) 1400-

@1200 - 'v

.2 -

g. 1000 ..

.0 800 - o I 1‘

C 600- . ° . <

8 . ;

E—g 400 " ‘ ‘ .
g 2081 a II ‘0‘ 'l: 0
g I I I I I
LU 300 400 500 600 700

Temperature (K)
350- (b) '1 I 3 r

E? 300- 1 I '
>3 I I .4. 0

a3- 250 'I' I ‘ a, . ‘

g 200- . A ' I Nansstea
8- l‘ 0 o NaPDBSbTOIO
E 150- A . ‘ A NanIZSbTe“
(T) .° v NanIBSbTozo
E 100- ‘ Na0.951’1920'54’T'322

 

 

 

300 ' 400 ' 500 ' 600 ' 700
Temperature (K)

>4 22- .4

a \
E120- —— ”:4 4
:18- /f r{ \\'.o\

O

616' ‘ kl \
IS A \v
812- \I‘

300 400 ' 500 ' 600 700
Temperature (K)

 

 

 

 

Figure 3-5. Temperature dependence of (a) the electrical conductivity, (b) the
therrnopower and (c) the power factor the compounds NanmeTem+2 (m=6, 8, 12 and

18) and Nao_95Pb2()SbTe22 prepared with method 1.

74

 

 

i
' s
'I
. 1st measurement '11 ,
. 2nd measurement

 

 

300 ' 400 ' 500 ' 600 ' 700 ' 800
Temperature (K)

 

 

(b) . . °
I
O I
. I
I I
I
I
i
. I
I
.I ' . lst measurement
' . 2nd measurement

 

 

300 ' 400 ' 500 ' 600 ' 700 ' 800

Temperature (K)

Figure 3-6. Comparison of (a) electrical conductivity and (b) therrnopower collected

during the first and second measurement for Nan38bTe10 prepared with method 1.

In order to check the reproducibility of the properties, other ingots with
composition NanSSbTeIO were prepared using the method. Many of the samples were

found brittle during the polishing step. A second sample could be measured and the data

75

set was compared with that measured for the first sample (Figure 3-7). The room
temperature electrical conductivity was ~ 650 S/cm for the second sample, much lower
than that of the first sample (1000 S/cm) (Figure 3-7a). At high temperature, the
difference in electrical conductivity decreases: (the electrical conductivity of the first
sample was only 70 S/cm higher than that of the second sample). Over the range 300-700
K, the second sample exhibits higher therrnopower than the first sample, with an average
difference ~ 3- 4O uV/K with the highest value ~ 360 uV/K at 715 K. The reason for such
difference in electrical properties is probably a difference in carrier concentration as both

electrical conductivity and therrnopower are affected.

 

 

 

 

 

 

 

 

A I
$310003 - (a) . other ingot A 350, (b) . . . .
V ' 0 first ingot ¥ 4 _ . o
’E 8001 . >3 300‘ . . .
g 600- ' gzso- - . °

1 I Q 1 I
g 400- . ézoo- _ .

. I O E 4 0
.3 200. I ' I .- . ' o . 2 150. .0 I other ingot
{3 ' ' ' - l—100‘ . firstingot
a) I ' I '71 ' I ' I ' ' r I ' ' ' ' ' I '
E 300 400 500 600 700 800 300 400 500 600 700 800

Temperature (K) Temperature (K)

Figure 3—7. Temperature dependence of (a) electrical conductivity and (b) therrnopower
measured for 2 different samples with composition Nangstelo prepared with method

1.

Thermal Transport Properties. Yhermal diffusivity measurements were collected for
the compositions NanmeTem+2 (m=6, 8, 12). The resulting thermal conductivity data

are plotted in Figure 8a. For all compositions, the thermal conductivity decreases as a

76

function of temperature. NanISSbTezo and Na0,95PbZOSbTe22 have the highest thermal
conductivity over the temperature range 300-670 K with values varying between 1.85
W/m°K at room temperature and 0.9 W/m-K at 675 K. Nangstelo and NanIZSbTeM

have lower thermal conductivity, with values ~ 1.3 W/m-K at 300 K and ~ 0.8 W/m°K at
670 K. The m=6 compound exhibits the lowest thermal conductivity, with values
between 1.05 W/m-K at 300 K and 0.65 W/m°K at 670 K. Figure 3-8b shows the lattice
thermal conductivity obtained by substracting the electronic component from the total

thermal conductivity. The lowest lattice components are found for the compositions

NanngTem and Nanésteg with the lowest lattice thermal conductivity ~ 0.45
W/m°K at 650 K for Nangstelo. This value is slightly lower than that reported for

NaagsprOSbTezz (0.55 W/m-K at 600 K).

77

 

2.0
r I (a) - Nan68bTe3
1 . 8 ‘ I o NanngTe10
1 6 - v A Nan128bTe14
1 4 ' I v Nan18$bTezo
' '. :A ' , < NaoesF’bzoSbTezz
12- ot‘ '.
‘ . ° : ' .
1.0". I. . . . V ‘ v
I I I . . . V V ‘
II I I - . . ‘ . .
' I I I I I I I

 

 

 

360 1 460 ' 560 ' 600 f 700
Temperature (K)

 

h rmal conductivity (W/mK) Total thermal conductivity (W/mK)
.0 .0
C) CD

 

 

 

1.3
1 2; (b) - NanSSbTeg
1'1 : o Nangste1o
' - A Nan128bTe14
10.1 v Nan13$bT620
' +
0'9? * < NaoespbzoSbTezz
0'81 l. x X
071 ‘4 ‘ xxxxx
05- : xxxlt
. I. I - 4 ‘ 4
(”0.5' . . ' : I ' '
4—0 . g l ' I I I I
804 . . . . . 3 f'?
'3 300 400 500 600 700
3 Temperature (K)

Figure 3-8. Temperature dependence of (a) the total and (b) lattice thermal conductivity
for the materials NanmeTem+2 (m=6, 8, 12, 18) and NaOgSszoSbTezz prepared with

method 1.

The estimated ZT values are given in Figure 3-9. It must first be observed that for

all compositions ZT 2 0.9 were observed at high temperature. The composition

78

Na0,95Pb208bTe22 exhibits the highest figure of merit with a maximum ZT ~ 1.7 at 650 K.

For Nan3SbTe10, a maximum of ~ 1.5 was observed at 650 K.

 

 

 

men
1.6- 4 ‘ ‘ ‘
. 11..
1.4“ ‘ ‘ ' . . .
1.21 1 . . : v
i {l- V'Vv
1.0' .3 x x A ‘ A A A
I- 0 8- . ' I x X
N ' - o It. i - NanBSbTem
0.5 - ‘ ' t
, . ‘4 o NanGSbTeB
0.4"I .a ‘ Nan128bTe14
0 2: 54 V Nan1BSbT820
0.0 “ ‘ Na0.95szoSbTe22

 

360 '460 '560 ‘600 700
Temperature (K)
Figure 3-9. Figure of merit ZT for NanmeTem+2 (m=6, 8, 12, 18) and

Na0,95Pb20SbTe22 prepared with method 1.

High Resolution Transmission Electron Microscopy (HRTEM). Figure 3-10a and 3-
10b-d show low and high magnification HRTEM images of the specimen Nangstelo.

Low magnification images (Figure 10a) revealed the presence of nanometer size
inhomogeneities embedded inside the matrix. Although no accurate elemental analysis

could be performed, we anticipate these regions to be rich in Na and Sb, based on the
chemistry of the system. Because low m members of the family NanmeTem+2 are rich

in Na and Sb, we can think that such specimen have higher densities of nanoregions

79

embedded in the matrix, thus contributing to lower the lattice thermal conductivity

compared to higher rn members.

 

Figure 3-10. Typical HRTEM images obtained for NanngTem prepared with method

As the samples were quite brittle, samples with composition NanngTelo were prepared

with different synthesis conditions.

80

3.3.2. Method 2

A sample for electronic transport measurements could be prepared but very
careful polishing work was necessary as the edges of the samples were brittle. Figure 3-
11 shows X-ray diffraction patterns taken for powder from both top and bottom of the

sample. All peaks could be indexed as belonging to the PbTe-structure.

 

 

 

 

 

 

 

 

 

60
. —(1)top
3‘ 50? -—-(2)bottom
'Z’
.9 40i
E i
(D
%i 30 . JJ (2)
a I i i
a: 20-
'..LJ d I (1)
10 LTJ - r - . - .
20 40 60 80 100
2theta (deg)

Figure 3-11. Powder X-ray diffraction of NanngTem prepared by method 2.

Temperature dependence of the electrical conductivity of the compound
NanngTem prepared by method 2 is presented in Figure 3-12a. The electrical
conductivity of the sample prepared with method 2 is much lower (almost half) than that
prepared by method 1. The values for sample obtained by method 2 are ~ 490 and 72
S/cm at 300 and 675 K respectively whereas they were ~ 1008 and 173 S/cm for the
sample prepared by method 1. This difference in electrical conductivity may be the result

of cracks present inside the sample prepared with method 2 or of difference in carrier

81

concentrations. The sample prepared with method 2 has higher therrnopower than that
prepared with method 1 (Figure 3-12b). The difference increases with temperature. The
sample prepared with method 1 has therrnopower ~ 313 uV/K at 685 K whereas the
therrnopower of the sample prepared with method 2 reaches 285 uV/K at the same
temperature. This behavior indicates a difference in carrier concentrations between
samples prepared with methods 1 and 2.When comparing the power factors, the sample
prepared with method 1 has the highest power factor over the range 300-700 K (Figure 3-
12C). The reason is the highest electrical conductivity measured for the sample prepared

with method 1.

 

 

 

 

 

 

 

 

 

E1000 ' (a) - method1 400: (b) .
g), 800: I O methOd 2 g 350. .
3:" 2a ' I '
.2 ‘ - 1 300- o '
*5 600- .6 ‘ . .
é . o I § 2501 . '

4 - I I
8 0°. ’ . . CE) 200- . -
.8 200. I I . I I E) 150‘ .. I method1
g 0‘ _ fi fi' _- _ I- . ' .methodz
Eu‘ 300 400 500 600 700 800 360 460 560 600 760 800

Temperature (K) Temperature (K)

NA 22.: I I I (C)

E 20- ' '

x 18- - -

:16- . ' .

o .

'8 O

“E ‘14-. .

$12 . method1 .

8 10 0 method 2 o

 

 

Figure 3-12. Comparison of (a) the electrical conductivity, (b) the therrnopower and (c)

 

300 400 500 600 '700 800
Temperature (K)

the power factor of the compounds NanngTelo prepared by methods 1 and 2.

82

Thermal diffusivity measurements were carried on the sample prepared with
method 2. The resulting thermal conductivity data are compared to those measured for
the sample prepared with method 1 (Figure 3-13a). The values are similar with ~ 1.25
and 0.75 W/m-K at 300 and 625 K respectively. The sample prepared with method 1 has

lower lattice conductivity with value ~ 0.6-0.45 W/m-K at 300 and 650 K respectively.

 

 

 

 

 

 

 

 

A >2

”E 1 3 s

E ' ‘ : (a) $0-9' - , (b)

E1'2' 3. gos- . .

§ 1.1 '1 . I O .

3 ° I 307- . '

E 1.0' . I C I Q . . . . .
8 0.9- '1: t£20.6— .

T“ _ I I E I I I

$0.8 .method1 ..Il:° l6305'-method“1"".. .
507-. method2 ' $04 . method2 ' "-
Ed 300 400 500 600 700 E ' 300 400 500 600 700
*- Temperature (K) 3 Temperature (K)

Figure 3-13. Comparison of the (a) total and (b) lattice thermal conductivities for

samples prepared by methods 1 and 2.

3.3.3. Method 3

The X-ray diffraction patterns for prepared from top, middle and bottom of the
ingot are showed in Figure 3-14a. The patterns from the different parts of the sample look
similar. The ingot seems to be homogeneous. A zoom of the region 38—60 ° (Figure 3-
14b) showed the presence of a second phase with the cubic PbTe structure. The indexing

of these peaks indicated the position of these peaks to be close to those of NaSbTe2.'2’ 13

BSE imaging of a finely polished sample did not show any indication of the presence of a

second phase in the PbTe matrix.

83

No thermoelectric properties could be measured because of the fragility of the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

sample.
60
W (a) —(1) top
—-—(2) middle
g 50' ——(3) bottom g
2 ‘ E’
2 40- 9!.
C C
._ . (3) I ..
g 30. “—— Lilli. g
z—g 'fliitw J ._i ‘ ii A g
a: 20‘ 1) LL 0:
‘J‘JJLJLJiLJ LUUJL '
1O ' ' ' ' ' ' I ' I 10 I ' I ' I ' I V
20 4O 60 80 100 40 45 50 55 60
2 theta (deg) 2 theta (deg)

Figure 3—14. Powder X-ray diffraction of NanngTelo prepared with method 3. The

area between 29 = 35 and 60 deg is enlarged to show the presence of extra peaks can be

assigned to NaSbTez.

3.3.4. Method 4

The X-ray diffraction patterns for prepared from powders both top and bottom of
the ingot are showed in Figure 3-15. All peaks could be indexed as belonging to the
PbTe-structure. However some peaks at high angles seem to split. No thermoelectric

measurements could be performed because of the low mechanical properties of the

sample.

84

 

O)
O

 

 

 

 

 

» —(1)top
g" 50-: ——(2) bottom
8
m 40-
:‘é’ t
a; 30J
2.: ‘ (2)
E“; 20.. 1 w

10- H
20 ' 4'0 ' 6'0 ' 8'0 '160
2theta (deg)

Figure 3-15. Powder X-ray diffraction of NanngTelo prepared with method 4.

3.3.5. Method 5
The X-ray diffi'action patterns for powders prepared from both top and bottom of
the ingots are showed in Figure 3-16.The patterns look like clean PbTe patterns. No

thermoelectric properties could be measured because of the fragility of the sample.

85

 

45 mt
i _ op
40* _ (2) bottom

35-

J

30-
25- (2) I

20- ‘
15..(1)
10

 

Relative intensity

 

 

 

 

 

 

 

2'0 ' 4'0 ' 6'0 ' 8'0 '160
2theta (deg)

Figure 3-16. Powder X-ray diffraction of Nangstelo prepared with method 5.

3.3.6. Method 6

Figure 3-17 shows the X-ray diffraction patterns from samples from both top and
bottom of the ingots Nan7.6Sn0.4SbTe10, Nan6SHZSbT610, Nan4Sn48bTelo and
NanZSmeTelo. The peaks could be indexed as PbTe phase, but splitting in the peaks
located at high angles is visible.

Careful analysis on fine polished surface was done on the samples
Nan4Sn4SbTe10 and NanZSn6SbTelo with SEM. BSE images from areas from top
(Figure 3-183) and bottom (Figure 3-18b) of the material Nan4Sn48bTelo showed areas

with different composition than that of the matrix. EDS analysis for the sample from the
top of the ingot gave the atomic percentages: Pb: 1.14, Sn: 10.96, Sb: 62.24, Te 25.57. No

accurate information on the presence of Na could be obtained as Na is a light element and

86

was introduced in low amount. EDS on the inclusion in the sample from the bottom part
of the ingot gave similar result: Sn1.l 1, Sb: 63.83, Te: 25.05. EDS analysis on the matrix

gave a ratio consistent with the nominal composition: Pb: 20.23, Sn: 19.94, Sb: 6.94, Te:
51.92. A sample from the middle of the ingot NanZSnéSbTelo was also analyzed,
inclusions were also observed (Figure 3-17c). EDS analysis indicated a composition

similar to that observed for Nan4Sn4$bTe10: Sn: 13.51 at. %, Sb: 63.37 at. % and Te:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

23.01 at. %.
60
60+ (a) —(1) top ‘ (b) —(1) top
— (2) bottom — (2) bottom
2: 50- Nan7'58no_4SbTe1o g 50' Nan58n2$bTe1o
.. a u) .
%40- g 40‘
.‘é - -- «
a,» 30- A g 30- (2)
% 20339-113 ‘ i , __I\.,LLA. g zoi'bLJ
a: .(1) ‘i I A l « (1)
10 I I ' IU'L—JU r I ' I 10 Q: ~_ I -
20 40 60 80 100 20 40 60 80 100
2 theta (deg) 2 theta (deg)
60 , (c) — (1) top 60 , d —m top
—(2) bottom ( ) _(2) bottom
% 50- Nan4Sn4SbTe1o % 50) Nan2$n58bTe1o
‘ l
E 40- g 40-
.E . ._ .
.3 30« (2) 3 30-
4-l ‘ H ‘ 2
i3 207“L § 207‘L H
«mud <1) uLJLJ ‘1’
1O - 1 . . ~ . - . 10 - . - 1 - -1---—r-—‘
20 40 60 80 100 20 40 60 80 100
2 theta (deg) 2 theta (deg)

Figure 3-17. Powder X-ray diffraction patterns of (a) Nan7o6Sno.4SbTe10, (b)

Nan6SHZSbTelo, (c) Nan4Sn4$bTe10 and (d) NanZSn68bTe10.

87

 

Figure 3-18. BSE images from areas close to (a) the top and (b) the bottom of the ingot

Nan4Sn48bTelo. (c) BSE image from 3 sample from the middle of the sample

NanzsmeTelo.

88

3.3.7. Method 7

The X-ray diffraction patterns for powders prepared from both top and bottom of

the ingots are shown in Figure 3-19. Analysis of the patterns for NaoijngTelo,
Nao,5Pngbo,75Te10 showed the presence of szTe3 as a second phase. In the case of
Na0.5Pngbo,75Te10, the presence of small peaks ~ 28 and 39 ° indicates the presence of a

small amount of szTe3. 1n the case of Na05Pb88bOI75Te10, two small peaks close to the

large peak at 40 ° are visible and may be the indication of the presence of Te, which can

be expected as the nominal composition is not charge-balanced with an excess of Te.

45 50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. —(1)top « —(1)top
40- (a) —(2) bottom 45- (b) ——(2) bottom
E 35: Nao.5Pb85bTeto E 40: Nao.5Pt>BSbo.75Te1o
g . g 35-'
.9 30- .93 +
.E . .E 301 (2)
. 2 o
$239.1. ‘1 (AM %25- I ll it
E 20-t " ‘— — '
0) J a.) 20"
a: m ‘ (1)
15-(1)| I “I 15.| IIHH
10‘ r ' 1 ' fiu'ff I 10I I '— ' ' 1 fi— I ' I
20 40 60 80 100 20 40 60 80 100
45 2 theta (deg) 50 2 theta (deg)
. (C) —(1)tOP . (d) —(1)top
40 - — (2) bottom 45 j — (2) bottom
a i Nao.5Pb85bo.5Te10 .é‘ 40- Nao_5Pb88b0'25Te1o
“Z; 35i g 35.
an ‘ tn '
._. 3o. ... .
C C
._ ( ._ 30.
d C) ‘
2m ._ J .1 .1... 2o. 1 1 1 ..t
'5 ‘ “W '6 20- “‘ ‘-‘—* ‘
a: ' 0: 1(1)
15- (1) I 15.
10‘ HdUDUL-M 10‘ . U. - . - . _ .L
20 4O 60 80 100 20 40 60 80 100
2 theta (deg) 2 theta (deg)

Figure 3-19. Powder X-ray diffraction of powders from (a) Na0,5Pb3$bTe10, (b)

Nadsl’bsSbmsTeio, (C) Nao.5Pb83bo.5Telo and (d) N30.5Pb83bo.25T€10-

89

Figure 3-20 shows two BSE images from a sample close to the top of the ingot

Nao_5Pb3Sbo_75Te10. Inclusions were observed embedded in the matrix. EDS on the dark

ribbon indicated the region to be Sb2T€3 with a small percentage of Pb (3.24 at. %).

Analysis on the dark areas contained in the lace part of the inclusion showed only the
presence of Te. Finally analysis on the bright areas in the lace part showed the presence

of Na (7.82 at. %), Pb (11.90 at. %), Sb (10.01 at. %) and Te (70.27 at. %). Samples fi'om
the ingot Na0.5PbSSb0_5Te10 were also analyzed (Figure 3-21). Composite inclusions were

observed in the PbTe matrix. The picture 21a is from a sample close to the bottom of the
sample. EDS analysis on area 1 reveals the presence of the elements Na (18.53 at. %), Sb
(14.19 at. %), Pb (19.94 at. %) and Te (52.34 at. %). Darker areas such as 2 contain only
elemental Te. The image 20b is from an area close to the middle of the ingot. Complex
inclusions are also observed. Area 1 contains mostly Te (80.87 at. %), but also some Na
(17.21 at. %) and Pb (2.12 at. %). Area 2 contains more Te (89.49) and small amount of
Na (10.51 at. %). Finally, Na (16.08 at. %), Pb (18.21 at. %), Sb (14.59 at. %) and Te

(51.11 at. %) were present in area 3.

9O

33400 25.0w 9.9mm x1 20k BSECOMP 9/17/2008 18:18 ' ' ' Lioottn'w

53400 25.0w 100mm x950 BSECOMP 9/17720'08'18'30

 

Figure 3-20. BSE images of areas from a part close to the top of the ingot

Nao.5Pb8$bo.75Tero-

91

SECOMP 9/17/2008' 1%

 

33400 25,0kv 10.4mm X37011. BSECOMP 9/17/2008 {18:08 ' "Ido'tm'n
Figure 3-21. BSE images of sample from (a) bottom and (b) middle of the ingot

Nao.5Pb8$bo.5Tero-

92

Because of the very high brittleness, only a sample from the ingot with nominal
composition Na05Pb38b025Te10 could be prepared for thermoelectric measurements

(Figure 3-22). Surprisingly the electrical conductivity increases from 379 to 440 S/cm
between 300 and 370 K. Then the electrical conductivity decreases down to 210 S/cm at
570 K. The sample was measured a second time to check on this atypical behavior. The

results are plotted on Figure 3-23a. The atypical behavior was confirmed by the second

measurement. At high temperature, the electrical conductivity of Na0.5Pb38b0.25Te10 is
close to that of Nan3SbTelo.

The therrnopower for Na0.5Pngb0_25Te10 is lower than that of Nan3SbTe10 but at

high temperature the values are almost similar for both compositions, ~ 274 uV/K at 575
K (Figure 3-22b).
The values measured during the first run were confirmed by running the same

sample a second time (Figure 3-23a). Over the range 300-570 K, Nangstelo exhibit

higher power factors than Na0_5Pngb0.25Telo (Figure 3-23b).

93

 

10001 o (a) I Nao.5Pb83bo.25Te1o
. . o NanaSbTe1o

800
600 -

400' I I . I .

 

 

200 - '

 

Electrical conductivity (Slcm)

360 'i 460 ' 560 ' 600
Temperature (K)

 

 

 

 

300
. (b) '-
g 250- °.
5 ' 0
33 200- ' '
3 . n
8
o 150- . l
g .
g 100_ - Nao.5Pb88bo.25Te1o
I o Nangste1o

360 ' 400 500 ' 600
Temperature (K)

 

22: (C) . . .

5418-: .

 

 

$9
‘a‘:
3 4; . - Nao.5Pb8$bo.25T61o
8 2‘ - . NaPDBSbTem
. . . . .

300 400 500 600
Temperature (K)

 

Figure 3-22. Temperature dependence of the (a) electrical conductivity, (b) therrnopower

and (c) power factor of Na0,5Pb38b0_25Te10 and NanngTelo.

94

 

 

 

 

300

 

 

 

 

’5‘ 450 _ . (a) i (b) .. -

‘ l A .

g), 4001 . . g 250 ._

.é‘ « I - 3. l

.> 350- . :: 200- °-

8 , d3 . .-

3 . 3

.2 300' o 8 150'

8 250‘ E ' l 1

_ . 1 t . I St run

8 ‘ : Zidnftrin . . g 100. a. 0 2nd run

@200 Na0.5Pb38bo_25Te1o +- 501 Nao.5Pb83bo.25Te1o

m 360 460 560 600 300 400 500 600
Temperature (K) Temperature (K)

Figure 3-23. First and second sets of measurements for the (a) electrical conductivity and

(b) therrnopower of the compound Na0_5Pngb0_25Te10.

In order to check if this behavior was reproducible, a second sample with the
composition Na0,5Pb33b0_25Te10 was prepared in the same conditions. The electronic

transport properties are reported in Figure 3-24. At room temperature the electrical
conductivity is higher for the second ingot but the electrical conductivity at 300 and 320
K are close (590 and 600 S/cm respectively). At temperatures > 425 K, similar values for
the electrical conductivity are found for both samples.

Both ingots have similar therrnopowers at room temperature (Figure 3-24b) but as the
temperature increases the second ingot exhibits higher values with the maximum ~ 326

uV/K at 620 K.

95

 

 

 

 

 

 

 

 

E 700 350
a o
c33:600- .- ( ) ,2 300- (b) . ' '
3 .
t? 500 - e 250- . .
l ' a

= 400‘ - a g 200- ° -
C I Q .
§ 300 ° . E 150- _
g 200- - firstingot ' : pa.) 0 I first ingot
.3 100 0 second ingot o o )— 100 :I . second ingot
a) 17* I I 1' I I l l
E 300 400 500 600 700 300 400 500 600 700

Temperature (K) Temperature (K)

Figure 3-24. Temperature dependence of the (a) electrical cconductivity and (b)

therrnopower of two different ingots with composition NaOI5Pngb035Telo.

3.3.8. Method 8
The X-ray diffraction patterns for powders prepared from both top and bottom of

the ingots are shown in Figure 3— 25. The sample was brittle and no thermoelectric

properties could be measured.

 

6O
— (1) top

—— (2) bottom

01
O
l .

.b
O
. I A

 

 

Relative intensity
0)
‘F’

 

 

<2> . l in.

20- L“

M Lu MM

10 l T' 17 ' I ' l ' l

20 4O 60 80 100
2theta (deg)

 

 

 

 

Figure 3-25. Powder X-ray diffraction patterns from powders from the composition

NanngTelo using NazTe as a source of Na.

96

3.3.9. Method 9

Resistance measurements on HPl and HP2 gave values ~ 0.6- 1 Q for HP] and
1.6 (2. Each coin was cut with the wire saw to prepare a parallelized sample. Both
samples showed high resistance. Figure 3-26 shows the powder X-ray diffraction patterns
of the original ingot before grinding and of the two pellets. All patterns are similar and

can be indexed as PbTe phase.

 

 

 

 

 

 

 

 

2'0 ' 4'0 ' 60
2theta (deg)

— (1) original ingot

60 " — (2) hot press 30 min at 100 °C
b . — (3) hot press 30 min at 150 °C
.7) 5o .
c .
g 40 4
a) ' (3) I
.5 30 -
£9 1 (2)_ 1 i
3" 2‘1... I I

. (1)
1O - . .
80

'160

Figure 3-26. Comparison between the X-ray diffraction patterns from powders from the

original ingot and the two pellets prepared from that ingot.

The surface of the pressed pellets HP] and HP2 were examined through SEM (Figure 3-

27).

97

I \ ‘ “u. ?
. .

\.

83400 25.0kV11 2mm x1_60k SE 10/5/2007'16'51'

 

Figure 3-27. SEM images of the surface of (a) HPl and (2) HP2.

98

A wide distribution of grains size and the presence of a significant amount of voids
between grains could be observed. This explains the high resistivity of the pellets. The
HP3 sample was found broken into two pieces when removed from the tube after
annealing. Figure 3-27 shows SEM pictures of HP] and HP2. Further pressing conditions

need to be investigated.

3.4. Concluding remarks

Thermoelectric characterization of the low rn members NanmeTem+2 shows promising

TE properties. The composition Nangste 10 exhibited particularly low lattice thermal
conductivity (~ 0.45 W/m-K at 650 K). Electron transport properties are also similar to
what was reported to NaoostZOSbTezz. High resolution TEM revealed the presence of
nanostructuring inside the matrix. However this does not give us information about the
density of such regions. The brittleness of these samples is a serious drawback for
industrial application. Using a slower or faster cooling profile did not improve the
mechanical properties. Further hot-press experiments should be pursued. Due to
limitation on our press system, experiments using higher temperature conditions could
not be done. Several groups have reported on hot-press experiment on the LAST system.
For example, Kosuga et al. prepared various Ag1-be|38bTe20 pellets (x=0, 0.1 and 0.2)
by sintering powders under a pressure of 80 MPa at 723 K under nitrogen atmosphere.14
The bulk densities of the samples were ~ 90 % of the theoretical density. Mechanical
alloying and spark plasma sintering techniques were used by Wang et al. to prepare

Agongb13+beTe20.15 Densities were found above 95 % of the theoretical value and SEM

99

studies indicated the average grain size ~ 1 pm. Samples prepared by hot press or spark
plasma sintering have low electrical conductivity at room temperature but the electrical
conductivity increases with temperature. The thermal conductivity is also reduced
because of the fine-grain microstructure. Yang et a1. prepared TAGS materials with
compositions (GeTe)x(AngTe2)1oo.x (x=75, 80, 85 and 90) using hot—press technique
(773 K, 70 MPa pressure).16 It is hoped that the application of these processing

techniques will be helpful in producing strong specimens of SALT-m samples as well.

100

References

1. Rogers, L. M., Brit. J. Appl. Phys. 1967, 18, 1227.

2. Noda, Y.; Orihashi, M.; Nishida, I. A., Mater. T rans., JIM 1998, 39, (5), 602.
3. Crocker, A. J ., J. Phys. Chem. Solids 1967, 28, 1903.

4. Borisova, L. D., Phy. Stat. Sol. A 1979, 53, K19.

5. Orihashi, M.; Noda, Y.; Kaibe, H. T.; Nishida, I. A., Mater. T rans., JIM 1998, 39,
(6), 672.

6. Harman, T. C.; Taylor, P. J .; Walsh, M. P.; Laforge, B. E., Science 2002, 297,
2229.

7. Heremans, J. P.; Thrush, C. M.; Morelli, D. T., J. Appl. Phys. 2005, 98, 063703.

8. Hsu, K. F.; L00, 8.; Guo, F.; Chen, W.; Dyck, J. S.; Uher, C.; Hogan, T.;
Polychroniadis, E. K.; Kanatzidis, M. G., Science 2004, 303, 818.

9. Poudeu, P. F. P.; D'Angelo, J .; Downey, A. D.; Short, J. L.; Hogan, T.;
Kanatzidis, M. G., Angew. Chem, Int. Ed. 2006, 45, 3835.

10. Ahmad, S.; Mahanti, S. D., Phys. Rev. B 2006, 74, 155205.
11. Kittel, C., Introduction to Solid State Physics. Wiley: 2005.
12. Eisemnann, B.; Schaefer, H., Z. Anorg. Allg. Chem. 1979, 456, 87.

13. Zhou, G.-T.; Pol, V. G.; Palchik, 0.; Kemer, R.; Sominski, E.; Koltypin, Y.;
Gedanken, A., J. Solid State Chem. 2004, 177, (1), 361.

14. Kosuga, A.; Uno, M.; Kurosaki, K.; Yamanaka, S., J. Alloys Compd. 2005, 391,
288.

101

15. Wang, H.; Li, J.-F.; Nan, C.-W.; Zhou, M.; Liu, W.; Zhang, B.-P.; Kita, T., Appl.
Phys. Lett. 2006, 88, 092104.

16. Yang, S. H.; Zhu, T. J.; Sun, T.; He, J.; Zhang, S. N.; Zhao, X. B.,
Nanotechnology 2008, 19, (24), 245707.

102

Chapter 4

Thermoelectric Properties of the N anostructured

Materials Nan13_xSanTe20 (M=Sb, Bi)

4.1. Introduction

Thermoelectric-based cooling systems and power generators have limited
applications because of their relatively low efficiency.1 As a consequence, worldwide
research for systems with higher performance is currently underway.2’ 3 The performance
of a thermoelectric material is defined by its figure of merit ZT = (052)T/K; where 0'=
electrical conductivity, S = Seebeck coefficient (also called therrnopower) and K =
thermal conductivity. The numerator (OS?) is called the power factor. Many recent
experimental and theoretical studies on low dimensional nanostructured materials“ and
mixed-phase nanocomposites7'lo point to encouraging results in how to get low lattice

thermal conductivity while maintaining high power factors. The superlattice tlrin-film

structures of BizTe3/Sb2Te3 grown from chemical vapor deposition6 and of

PbSeoggTerz/PbTe formed by molecular beam epitaxy4’ 5 claimed ZT values greater

than 2 (at~ 300 and 550 K respectively). Bulk silicon has poor thermoelectric properties,

however silicon nanowires show thermal conductivity which approaches the amorphous
limit.l 1’ 12 We are interested in investigating bulk analogues of such systems with similar

figures of merit.

103

Recently, we have reported on the bulk n-type systems AngmeTem+2

13’ 14 Pb1_xSnxTe-PbS15 and Pb9_6Sb0.2Te-10.,(Sex16 and the p-type systems Na]-

(LAST),
bemeyTem+2 (SALT)17 and Ag(Pb1_ySny)meTe2+m (LASTT),18 all bulk materials that

exhibit high thermoelectric figure of merit. High resolution transmission electron
microscopy images of these materials revealed the presence of nanoparticles coherently
embedded in what is essentially a PbTe matrix. These features are believed to be the

origin of very low lattice thermal conductivity, which is a common property of all these
systems. For example, the p-type composition Nag-95PbZOSbTe22 reached ZT of ~ 1.6 at

650 K, which is nearly twice that of p-type PbTe, and arising predominantly from a very
low thermal conductivity. Namely, the total thermal conductivity in these materials drops

very rapidly with rising temperature and reaches a minimum value of 0.85 W/m°K at 700
K for Na0,95Pb208bTe22. In a recent study, we showed also that the influence of the
trivalent element in AnglgMTezo (M=Sb, Bi) on the power factor and the lattice thermal
conductivity is significant with the presence Sb of producing more favorable properties
than that of Bi.19

In this chapter, we examine the partial substitution of Pb by Sn in the high ZT
material Nan13SbTe20 to produce the solid solutions Nan13-xSnbeTe20 and study their
physical and thermoelectric properties. To better understand the role of the pnicogen on
the properties of the system, Bi analogs with general formula Nan18-xSaniTe20 were
prepared as well. Partial substitution of Pb by Sn results in a increase of the hole

concentration and a decrease of their mobility. The electrical conductivity of Nan18.

xSanTezo (M=Sb, Bi) increases and the therrnopower decreases with increasing fraction

104

of Sn. All the compositions examined showed p-type behavior and exhibit low lattice
thermal conductivity. That is attributed to a combination of point defect scattering
associated with solid solution behavior and also the presence of nano-sized inclusions in

the crystalline matrix.

4.2. Experimental section

4.2.1. Synthesis.

All samples were prepared as polycrystalline ingots in silica tubes by mixing high
purity elements in the appropriate stoichiometric ratio. To prevent reaction between the
sodium metal and silica, the tubes were carbon-coated prior to use. All components
(except Na) were loaded into the silica tubes under ambient atmosphere and the

corresponding amount of Na was later added under nitrogen atmosphere in a dry glove
box. The silica tubes were then flame-sealed under a residual pressure of ~10—4 Torr,

placed into a tube fiimace (mounted on a rocking table) and heated at 1250 K for 4 h to
allow complete melting of all components. While molten, the fumace was allowed to
rock for 2 h to facilitate complete mixing and homogeneity of the liquid phase. The
firrnace was finally immobilized at the vertical position and was cooled from 1250 to 820
K over 43 h followed by a fast cool to room temperature. The resulting ingots were
silvery-metallic in color with a smooth surface.

Table 4-1 and 4-2 summarize the amount of elements used for each reaction.

105

Table 4-1. Amounts of elements used to prepare Nan13-xSnbeTe20.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Composition $1,531) 5:}301) (stillingol) (Sriiifol) (fuelifol)
NanwaTezo 3.93:5)g 35%??? ° 8368)” 33.9553?
3,966.0; $591353: 25.2.9.4: 25:3,“ gar
Nan13Sn5SbTe20 22015 %g (522528;)g gig)? (02216;; g (54§S5()g)g
Nangsngstezo 2);:ng 212.2359? éfiég $5611.38 3 gigs
3,933 2.39:? are 25:3“ are
N.p..s...s:T... 3,9,ng ”:79.” E 3:22;: $5.99? g 83%?
3.9.3:; 0 5.29.6.1; art 2.3%
Table 4-2. Amounts of elements used to prepare Nan]3_xSaniTe20.

Composition mil) F301) (Suriingol) (Snliingol) ([51:01)
NanlgBiTezo 893$ Z3??? 0 i353; g its???
Nan15Sn3BiT620 $206688 (83.19.1112)g 27.98239)4g 35:15)” gégg
Nan13SnsBiTezo €973? $3932? 22.39513)? g 31.07517)2 g (1331,6396)g
NangsngBiTezo 32°93? €123? 2i:.29665)g (:33 g $.92?
NanSSnlgBiTezo 2&3? (1)2717; g fi39f‘793f $05128; g (3.355?
NanZSn16BiTezo 2353f €42,332 g 2151195373; 31561; g €431?
NaSnlgBiTezo goggg 0 55208611)g 33.6346; g (7607:03g

 

 

 

 

 

 

 

 

4.2.2. Characterization techniques

Powder X-ray Diffraction. Powder X-ray patterns of the ground materials were

recorded using Cu K, radiation (7» = 1.54056 A) in reflection geometry on a CPS-120 Inel

106

X-ray powder diffractometer operating at 40 kV and 20 mA equipped with a position
sensitive detector. The lattice parameters were refined from the X-ray powder diffraction
patterns using the software Jade 6.0 from Materials Data Inc.

Infrared Spectroscopy. Room temperature optical diffuse reflectance measurements
were performed on finely ground powder to probe the optical band gap of the materials.
The spectra were monitored in the mid-IR region (4000-400 cm'l) using a Nicolet 6700
FTIR spectrometer. Absorption (ct/S) data were calculated from reflectance data using the
Kubelka-Munk function.”22 The optical band gaps were derived from a/S versus E (eV)

plots. Specular IR reflectivity measurements were carried out on polished specimens in
the spectral range 100 -2500’ cm", at room temperature, with non-polarized light, using a

Bruker IFS 113V Fourier transform interferometer working under vacuum and equipped
with the special reflectance unit. The angle of incidence was less than 10 deg.

DTA Analysis. Differential thermal analysis (DTA) data were collected with a Shimadzu
DTA-50 thermal analyzer. Approximately 35 mg of finely ground powder of material
was sealed in a carbon-coated quartz ampoule under residual pressure of ~ 10‘4 torr.
Another ampoule containing similar amount of alumina and prepared the same way was
used as a reference. The samples were heated to 1273 K at a rate of 10 K/rrrin, held at
1273 K for two minutes and cooled to 323 K at a rate of -10 K/min.

Scanning Electron Microscopy. The surface of several samples was polished very
carefully using silica suspension solution (0.05 pm) in order to get a smooth, mirror-like
Surface. The samples were then studied with a scanning electron microscopy (Hitachi
S34OON-II) with 25 kV acceleration voltage using both energy-dispersive spectroscopy

(EDS) and back-scattered electron imaging (BSE).

107

Electrical Transport Properties. Thermopower and electrical conductivity properties
were measured simultaneously under helium atmosphere using a ZEM-3 Seebeck
coefficient/electrical resistivity measurement system (ULVAC-RIKO, Japan). Samples
for transport measurement were cut to size 10 x 3 x 3 m using a diamond saw (Buehler

isomet 1000), a wire saw (South Bay Technology) and a polishing machine (Buehler
ecomet 3000). Rectangular shape samples with approximately 3 X 3 mm2 cross-section

were sandwiched vertically by two nickel electrodes (current injection) with two Pt/PtRh
thermocouples (for temperature difference and voltage measurements) attached on one
side. The sample and measurement probes were covered by a nickel can to maintain a
constant temperature during the measurement and the base temperature was measured by
a thermocouple attached to the outside of the can. The sample, electrodes, and nickel can
were placed in a vacuum chamber then evacuated and refilled with He gas (0.1 atrn) to
provide necessary heat transfer. Properties were measured from room temperature to 670
K under helium atmosphere.

Hall Measurements. Above 300 K, Hall measurements were carried out by an in-house
high temperature/high magnetic field Hall apparatus. It consists of a nine Tesla air-bore
superconducting magnet with a water-cooled oven inside the bore of the magnet, and a

Linear Research AC bridge with 16 Hz excitation. Four—wire AC Hall measurements
were performed on parallelepiped samples with the typical size of 1.5 x 3 x 10 mm3 to

temperatures of at least 800 K with the protection of Argon gas.
Thermal Conductivity. The thermal conductivity was determined as a function of
temperature using the flash diffirsivity method on a LFA 457/2/G Microflash Netzsch.

The front face of a small disc-shaped sample (diameter ~ 8 mm; thickness ~ 2mm) coated

108

with a thin layer of graphite is irradiated with a short laser burst, and the resulting rear
face temperature rise is recorded and analyzed. The experiments were carried out under

nitrogen atmosphere. Thermal conductivity values were calculated using the equation K:
a Cp d, where a is the thermal diffusivity, Cp the specific heat and d the bulk density of

the material calculated from the sample’s geometry and mass. A pyroceram reference
was used to determine the heat capacity of the sample. The thermal diffusivities were
measured typically over the temperature range 300-670 K. The electronic component of

the thermal conductivity was quantified through the Wiedemann-Franz law according to
which IQ.) = o.T.L0 (Lo being the Lorenz number, Lo=2.45.10’8 WQK'I).23 The lattice

contribution was then derived by subtracting the electronic component from the total
thermal conductivity.

Transmission Electron Microscopy. The microstructures of several pieces cut from
different locations of the ingots were examined by high resolution electron microscopy
(HRTEM). Specimens for the investigation were prepared by conventional standard

methods which was described elsewhere”. The HRTEM images were recorded at 200 kV
using a JEOL 2200FS and JEOL 2100FS for Nan13Sn5SbTe20 and Nan13Sn5BiTe20
respectively. Composition analysis was done by Energy Dispersive X-ray Spectroscopy

(EDS) using a spot size ~ 1 nm in scanning transmission electronic microscopy (STEM)

mode.

4.3. Results and discussion

Structure and Characterization. During the course of this study, we observed that

ingots containing antimony tended to be mechanically more robust than their bismuth

109

analogs. The Bi containing samples were extremely brittle and ofien cleaved into lamellar
pieces. Nevertheless all ingots were found to be brittle and great care was taken to select
specimens that were free of cracks.

Figure 4-la shows the X-ray diffraction patterns for samples with general formula
Nan18_xSnbeTe20 (x = 5, 9, 13 and 16). The patterns can be indexed to a NaCl-type

structure. No extra peaks of a potential second phase were observed. The Bragg peaks

shift towards higher diffraction angle with increasing Sn/Pb ratio, indicating contraction

of the lattice constant. This result is consistent with the expected substitution of Pb2+
cations by the smaller Sn2+ cations in the lattice. Diffraction patterns of the bismuth

analogs (general composition Nan18_xSaniTe20) are depicted in Figure 4-lb. Again, the
patterns indicated NaCl-type structure and with higher fraction of Sn displayed a similar
shifting of the diffraction peaks towards higher angles. Lattice parameters for Nan13.
xSnbeTezo and Nan13_xSaniTe20 samples refined from the diffraction patterns are
plotted as a function of x in Figures 4-1c and 4-1d. For example, the lattice constant
varies from 6.444 (1) A for Nan158n3BiTe20, which is closer to PbTe (6.459 A)24 to
6.342 (1)A for Nanzsnl6BiTeZO, which is closer to SnTe (6.328 A).25

To achieve doping in these systems, stoichiometric deficiency was created on Na
as it was done on the parent composition Na1-bemeyTem+2.l7 X-ray powder diffraction

was taken on powder from both top and bottom of the ingot to assess homogeneity along

the samples. Analysis of the powder diffraction patterns of Na-deficient compositions

revealed the presence of szTe3 as a minor impurity. For example, in the diffraction

patterns of compositions Na0.8Pbl3Sn58bTe20 (Figure 4-2a) and Nm,ng13Sn5Sb0,4Te20

110

(Figure 4-2b), extra peaks attributed to szTe3 were clearly observed. This suggests that
vacancies of Na+ cations are not created in the NaCl-type lattice. Instead szTe3 is

expelled according to equation (1 ):

Na1-bemeTem+2 —* Na1-beme1-yTe(2+m-3/2 y) + y/2 Sb2TC3 Eq (1)

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

80 (a) 6.50:. (c)
Q .2 gig \
an 50. . 5 - '.
5;. I t I § 6.44- I _
5 ‘ illli (l. 036.42: I
g 40- r ~~~l‘~tfl-‘-'W~ F15 § 6.4o-j _
3 1M9 __,'L_JI.__,i-__,-._,.:_}__,\_ x:13 g 232: .
0 A t . .w" U — .— . 'i 7
I: 20- “4‘" "’ X‘9 33 ‘ ,

‘hl-L,” LJLJCJW x:5 3 23:? .5.
2'0 70 '30 ' 8T0 '1601120'140 6 ' 2 '21 '6 ' é '110'1213'123'13'20
2 theta (deg) x value
80 6 46
(b) ' ' (d)
g i l .3: 6.44- i ‘
cc: 60' l ( l l . 26.42- *_
,5 .Nwfi7sfw'J4J- x=16 €6.40. .
g 4OWMleaaLtw-Jca. x=13 3 6.38- '
é ‘Wtwmi' ——al~-—-ti‘~—-—«'~-4"-—J"~—"‘u x=9 g g 3:: - .
‘ i — ' - VIX‘
2N. J: E w: x-é \.
20 40 60 80 100 120 140 6 2 4 e é 1'01'2121'612320
2 theta (deg) x value

Figure 4-1. Powder X-ray diffraction patterns of Nan18_xSanTe20 with x=5, 9, 13, 16

for (a) M=Sb and (b) M=Bi; variation of the unit cell parameter as a function of x for (c)

M=Sb and (d) M=Bi.

111

 

40- (a) —top
. — bottom
30- —PbTe

J

 

 

 

W

..s
0

Relative intensity
N
‘?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0‘ . 1i ll.
2'0 ' 4'0 ' 6'0 ' 3'0 '160'
2theta (deg)

40-(b) —top
5‘ - —bottom
g 30. —PbTe
.‘é’ 20.:— . A . L quot...
Q,
.5 . \
2 10- flame
0‘3 - J

0. _ _ . i [J l

 

 

 

20 4o 60 ' 80 '160'
2theta (deg)

Figure 4-2. X-ray powder diffi'action patterns of (a) Na0_ng13Sn58bTe20, (b)

Na0.3Pb13Sn58bo.4Te20. For each composition, powder from both top and bottom of the

ingot was analyzed to check homogeneity along the ingot. The small arrows indicate

diffraction peaks that do not belong to the PbTe structure-type and indicate the presence

0f szTe3 as a minor phase.

112

Figure 4-3a shows DTA data for Nan58n13SbTe20. The plot shows only one

endothermic peak of melting during the heating cycle and one exothermic peak of

crystallization on the cooling cycle. Similar curves were obtained for the other

compositions Nan18-xSnbeTe20 and their Bi analogs. Melting and crystallization points
for the Nan13-xSanTe20 series are summarized in Tables 4-3 and 4-4. The melting
point decreases with increasing Sn fraction and varies from 1160 K for Nan158n3SbTe20

to 1081 K for NanZSanbTezo (Figure 4-3b). This trend is consistent with the expected

solid solution alloying between the high melting PbTe (1196 K) and the lower melting

SnTe (1063 K) compounds.

The band-gap energy values obtained from electronic absorption spectra for
Nan13-xSanTe20 are reported in Tables 4-3 and 4-4. For Nan18_xSnbeTe20, the
lowest band gap was observed for x=9, the band gap decreases in going from x=18 to
x=9, Figure 4-4. The same behavior is observed for the Bi analogs with band gaps below

0.1 eV for x=3, 5, 9, 13 and 16. The band gap increases again then for x=18. This is

consistent with trends reported previously: for example, several investigations have
shown that the band gap of the Pb1_xSnxTe system decreases as the tin content increases,

goes through zero and then increases again with further increase in tin content.26

113

 

 

    
 

 

 

 

 

 

 

‘ a
o. ( )
-20.
a .40.
E 450:
-80- i exo
'100j1 endo
~120 . . . - . . . . .
400 600- 800 1000 1200
Temperature (K)
1200' b
1180-‘\\A ()
-11601 \A
x Iv\ \\
v 1140' V”"'\ \
g 1120: \\v\\\\A
E11004 \ \.
§10801 \ \,
'1’ 10601 —A— melting point ' '
1040'. —v— crystallization point V
1020I'Ifil'l'rfi'l'l'I'l
024681012141618
xvalue

Figure 4-3. (a) Typical DTA results of the composition Nan58n13SbTe20, (b) variation

of the melting and crystallization points of Nanlg-xSnbeTe20 as a function of x.

114

14

 

 

12-
A 10-
3' .
3i 8-
(L) .
:3 5-

4-

 

 

 

2 . . . . . . . . . ,
0.0 0.1 0.2 0.3 0.4 0.5
Energy (eV)
Figure 4-4. Infrared absorption spectra near the band edge of samples Nan13_

xSnbeTezo with x=9, 13, 16 and 18.

Scanning Electron Microscopy. The surface of several samples was polished very
carefully and analyzed using both secondary and back-scattered electron microscopies. In
the case of Sb analogs, micros-size inclusions were observed for the different
compositions. Such inclusions are randomly dispersed in the Pb1_ySnyTe matrix and
typical distance between inclusions is in the millimeter range. For example, Figure 4-5
shows BSE imaging of inclusions for different parts of the ingots Nangsngstezo and
Nan13Sn58bTe20. EDS analysis on the inclusions observed in Figure 4-5a gave the
following atomic percentages: Te: 20.21, Sb: 67.63 and Sn 10.50. Similar ratios were
obtained for the precipitate in Figure 4-5b. The inclusion observed in Figure 4-5c had less

Sb, the atomic percentages of Pb, Te, Sb and Sn were 5.22, 40.05, 46.42 and 8.31

115

respectively. The formation of such precipitates rich in Sb could be expected because of
the low solubility of Sb in PbTe.27
Similar SEM studies were done for the Bi samples. Micro-size precipitates could

be observed as well. Figure 4-6a shows a Bi precipitate ~ 23 mm diameter embedded in
the Pb1-ySnyTe matrix for the composition Nan13SnsBiTe20. Some of these precipitate

are actually multi-phase. Examples of such inclusions are visible in Figure 4-6b and 4-6c.

Both inclusions were observed for the same sample from the middle area for
NanZSntiTezo. Only Bi was present in the bright area 1 of the precipitate shown in

Figure 4-6b but EDS analysis on the darker zone 2 gave the following atomic
percentages: Pb2.35, Sn 22.74, Bi 25.79, Te 48.50. The composition of the matrix was
analyzed with EDS and the respective atomic percentages for Pb, Sn, Bi and Te were
6.14, 37.07, 4.20 and 52.66 respectively. EDS analysis on the bright area lof the
precipitate on Figure 4-6c indicated only the presence of Bi whereas the darker zone 2
contained Pb, Sn and Bi as well. The atomic percentages were Pb 1.58, Sn 20.88, Bi

32.59 and Te 43.84. The variety of the shapes and dimensions of the inclusions suggest

that they form randomly in the Pb1_ySnyTe matrix. As these phases have lower melting

point that Pb1_ySnyTe, they are trapped in the matrix which starts to crystallize before

them upon cooling.

116

 

Figure 4—5. BSE imaging of inclusions observed fi'om (a) part close to the top and (b)

part close to the bottom of the ingot Nan9Sn9SbTe20, (c) middle part of

Nan13Sn5SbTe20.

117

(a)

 

Figure 4-6. BSE imaging of samples from middle part of the ingots (a) Nan13SnsBiTe20

and (b)-(c) Nanzsn16BiT620.

118

Electronic Transport Properties.
Nan13_xSnbeTe20. The electronic transport properties of the Nan13-xSanTe20
systems can be tuned by varying the Sn:Pb ratio. Figure 4-7a shows the evolution of the

electrical conductivity as a function of temperature for the Sb analogs Nanlg-

xSnbeTezo (x=0, 3, 5, 9, l3, l6 and 18). Regardless of the composition, the electrical

conductivity monotonically decreases with increasing temperature, indicating degenerate
conduction for the whole temperature range examined. The electrical conductivity

increases with increasing Sn concentration. The room temperature value of the electrical

conductivity of NaSnlngTezo was ~ 4220 S/cm, which is about three times that of

Nanlgstezo. This dramatic increase in the electrical conductivity with addition of Sn is

consistent with the high electrical conductivity of SnTe, which is the result of massive,

naturally occuring Sn vacancies in the lattice that drastically increases the carrier
concentration.25 Presumably, in the Nan13_xSnbeTe20 system, these vacancies become

more and more dominant with increasing fraction of SnTe (i.e. x). This hypothesis was
confirmed by Hall effect measurements.

From measurements of the Hall Effect, carrier concentrations of ~ 1.2-1020 and
8-1019/cm3 were obtained at room temperature for Nan13Sn58bTe20 and
Nan15Sn3SbTe20 respectively. The carrier concentration was extracted from the hall

coefficient assuming an extrinsic regime, thus using the formula for a single carrier

transport 11 = l/eR (with e being the electron charge and R the Hall coefficient). For

comparison, the room temperature hole concentration of NaogstzoSbTezz was ~

6.3-1019/cm3. This indicates an increase in hole concentration with increasing Sn content.

119

Nan13Sn5BiTe20 has similar hole concentration than its Sb analog (Figure 4-8a). For

both materials, the carrier concentration increases up to ~ 2.41020 /cm3 at 675 K. This

increase in the carrier concentration is due probably to the very low band gap of the
material. Higher temperature allows more valence electrons enough energy to be
promoted to higher energy levels. The higher carrier concentrations of the Sn-rich
compositions were independently confirmed by the spectroscopic infrared reflectivity
studies of these systems (see below).The carrier mobility was calculated using the

electrical conductivity data from Figure 4-7a. The values of the mobility vary from 54
cmZ/V-s at room temperature to 7 cmzN-s at 675 K for Nan13Sn5SbTezo and fiom 38 to
5 cmZN-s for its Bi analog (Figure 4-8b). Such relatively low mobility values are a result

of the very high carrier concentration and the cation disorder in the NaCl-type lattice

which scatters carriers more strongly.
The temperature dependence of the electrical conductivity of Nan13-xSnbeTe20
samples is in accordance with the power law T'l, where the power exponent /1 ranges

from 2.64 (for x=0) to 0.99 (for x=18) (Figure 4-7b). The power law dependence of the
electrical conductivity (also found in PbTe) results from the phonon scattering of charge

carriers.28 The value of 7» decreases with increasing content of Sn. The electrical

conductivity of the composition NaSn18SbTe20 decreases as VT (1» = l), which is similar
to SnTe and more typical of a metal. The electrical conductivity of NanlSSbTezo
decreases with increasing temperature according to the power law T'2'64. This is

consistent with the trends observed in the Na1-bemeyTem+z system17 and similar to p-

type PbTe.

120

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E 4500 _ 3_o
g 4000 ’3‘, (a) 1;: 2.8-_ (b)
S 3500- "I +x=5 33 '
3 3000' ‘\ "N +x=9 22:
=5 2500: 4\ ‘k’. —n—X=13 - I
g ‘\ ’ +X=16 20'
u 2000- .,,\ ‘ +x=18 .< 1.8-
§ 1500. ”V‘K“ ‘\‘\‘ 1.6- - .
a 1000- ANN»: \,, V 13 ' _
*3 . TEFFFF'Lr 0'8 '
5 300 400 500 600 700 o 2 4 6 81012141618
Temperature (K) Y vallm
24
350- (c) . N. 22 - (d)
2 300- ./-/ ¥. 20-
S l/ E 18'
a 250- 49 2 16-
; 200- / ./"‘ 331%:
0 /° // v' '- 10-
3150- ‘ A,A v/v/‘_‘ *3 31
E 100. . ./"/ /</ 8 64
a A/ V/ / ‘ h
.c ‘ 1'5. ./ " a 4'
._ 50- if.» 3 g.
360 460 500 600 760 °' 360 460 560 600 760

Temperature (K) Temperature (K)

Figure 4-7. Electronic transport properties of the Nan13_xSnbeTe20 samples: (a)
electrical conductivity, (b) variation of the power law dependence of the electrical
conductivity (TJ‘) as a function of x (the )V parameters were extracted from the data of

panel (a)), (c) therrnopower, ((1) power factor. The point marks in the inset of panel (a)

identify all samples and apply to panels (c) and (d).

121

 

 

 

 

Temperature (K)

 

 

 

 

m.
g 261 (a) A “60‘

93 241 . gso- ' . (b)

52 22. : - '

E 20 .1 A Nan15Sn3BiT620 , NE 40. ‘ v v

C ' v N Pb S SbT ' ‘ V ' ‘ v

“@181 a 13 n5 e20 ' ‘ 3330' ‘ ‘ v

"" 16' ' ‘ 5 ‘ A '

C ' v A ‘ v

g 14.: v I ‘ g 2041 ‘ ‘ A '

8 12: ' 'VVVII‘ E10. . Nan15Sn38iTe20“::v,v

'5, 10- ‘ “ ‘ ‘ :5 0' v Nan138n5SbTezo ‘ H

'- I ' I ' I ' I ffi 0 1 f . v T . I . I
go 300 400 500 600 700 300 400 500 600 700

Temperature (K)

Figure 4-8. Temperature dependence of (a) the carrier concentration and (b) carrier

mobility for Nan13Sn5MTe20 (M=Sb, Bi).

All compositions showed positive therrnopower indicating p-type transport. The

therrnopower trend as a function of Pb/Sn ratio is opposite to that of the electrical
conductivity, Figure 4-7c. The therrnopower of NanlngTezo starts at ~105 uV/K at 300
K increases rapidly with rising temperature reaching the high value of 329 uV/K at 700
K. The corresponding Sn analog (NaSnlgstezo) showed only ~ 44 uV/K at room
temperature and the therrnopower did not exceed 75 uV/K at 560 K. One interesting

feature of the Nan13-xSnbeTe20 system is the high rising slope of the therrnopower with

increasing temperature. For example, in the case of Nan158n3SbTe20, the value at 560 K

is 2.5 times larger than that at room temperature (101 uV/K at 302 K and 246 uV/K at

560 K).

The power factors for each composition are shown as a function of temperature in

Figure 4-7d. Clearly, the Sn-free composition NanlgstCzo has a much higher power

factor reaching 20 uW/cm-K2 at ~ 400 K. For Pb-Sn mixed compositions, the graph can

122

be divided into two temperature regions, the first ranging from 300 to 500 K and the

second above 500 K. Within the first region, Nan13Sn5SbTe20 exhibits the highest
power factor at 460 K whereas in the second region, NanZSnmeTezo surpasses the
other compositions, reaching a power factor of 20 iiW/cm-K2 at 650 K, see Figure 4-7d.

Doping studies on the previously reported Na1_bemeyTem+2 system17 showed a

slight increase of electrical conductivity with decreasing Sb content and an increase of

therrnopower with decreasing Na/Sb ratio. It was found that the power factor of the
NaangzoSbyTezz materials increases with decreasing Na/Sb ratio.17 In order to
investigate the role of these ratios on the thermoelectric properties of the title materials
we chose the composition Nan13Sn58bTe20. Figure 4-9a shows the temperature
dependence of the electrical conductivity for the compositions Nm,ngl3Sn58bTe20,
Naoing13SnSSb03Te20, Na0,ngl3SnSSb0_6Te20 and Nm,ng13Sn5Sb0,4Te20. Other
compositions Na0.9Pb13Sn5SbTe20, Na0_9Pb13SnSSb0_gTe20, Na0,9Pb13Sn58bo_6Te20 and

Na0,9Pb13SnSSb0_4Te20 were also prepared but the ingots were too brittle to allow proper

electrical characterization. The transport data show that for similar Na concentration, the

electrical conductivity decreases and the therrnopower increases with increasing amount
of Sb (Figure 4-9b). The composition Na0_3Pb13Sn58bTe20 reaches a maximum ~ 257
uV/K at 650 K compared to 201 uVK for Nan13Sn5$bTe20. In the range of 450 — 500 K,

similar power factors were obtained for samples with various Sb fractions (Figure 4-9c).

At higher temperature, the power factor plots diverge. The composition

123

Na0_ng13Sn5SbTe20 reaches a maximum of 11.5 uW/cm-K2 at 550 K whereas

Naolng13Sn5Sbo_4Te20 reaches ~16 )iW/cm-K2 at 675 K.

 

_L
0'!
O
o

1250: \
1000: EL'\'_"-\— -
750i '?:.\'\\.
500- g_ ”x. -s,;\

250 . 2,“): T- ‘

 

 

 

3607 460 360T 6% 760

Electrical eonductlvlty (Slcm)

 

 

 

 

 

 

 

 

Temperature (K)
275
2 2501 (b)
S 225‘: ‘2. ,/Q: ,.
3 200‘. . 4m,“
3 175-: / / ...V’"
g 150-: ”I "I éfjfix=5
g 1001 32-5% -;.-y=0.8
g 751 {116 %y=0.6
1- 50- ’f - . _ . . '—.\:y=9.4
300 400 500 600 700
Temperature (K)
-15. t.
i 12 . é;/ ”\T}
..A. [
:1101 , ___/ "
o ‘
3.3: 81 g «1/
'5 5j gr/
3 4. 3:; ' V ' I ' I t I
m 300 400 500 600 700
Temperature (K)

Figure 4-9. (a) Temperature dependence of the thermoelectric properties of the
Nan13Sn5SbTe20 and Na0,3Pb13Sn5$byTe20 (y=0.4, 0.6, 0.8, 1) compositions: (a)
electrical conductivity, (b) therrnopower, (c) power factor. The point marks in the inset of

panel (b) apply to plots (a) and (c).

124

Nan13.xSaniTe20. Similar general trends as above were observed for the Bi
analog series (Nan13-xSaniTe20). Increasing the Sn concentration also resulted in an
increase in conductivity and decrease in therrnopower. Electrical conductivity data for
Nan13-XSaniTe20 (x=0, 3, 5, 9, 13, 16, 18) are shown in Figure 4-10a. First it must be
noticed that Nan5$n13BiTe20 has similar electrical conductivity than Nan9Sn9BiTe20,

this is probably due the presence of cracks in the sample NanSSn13BiTe20. At room
temperature, the compositions x=3 and x=5 have lower electrical conductivities than
NanlgBiTezo but similar values are observed at higher temperature. Nan9Sn9BiTe20

and NanlgBiTezo have similar electrical conductivities at 300 K but at higher

temperature the electrical conductivity decreases more rapidly for NanlgBiTezo. As the
Sn:Pb ratio increases, the conductivity is progressively dominated by the properties of
SnTe. Thus, the electrical conductivity of the Sn-rich NanZSntiTezo and

NaSnlgBiTego is much higher than that of N‘anlgBiTezo. Regarding the power law

dependence of the electrical conductivity (T'X), the parameter 2 decreases with increasing
tin content with value ~ 1 for x = 13, 16 and 18 (Figure 4-10b).

The therrnopower values for the Nan13-xSaniTezo series are shown in Figure 4-
100. The composition NanlgBiTezo clearly exhibits the highest therrnopower over the
temperature range 350-650 K. The other compositions can be classified into two groups.
The first group includes the compositions Nan153n3BiTe20 and Nan13Sn5BiTe20 with

therrnopower values ranging from 80 to 270 uV/K between 300 and 675 K. The second

125

group comprises the compositions NangsngBiTezo, Nan5Sn13BiTe20 and

NanzSn16BiTe20 with values < 170 uV/K at 675 K, much lower that of NanlgBiTezo.

Similar to the Sb analogs described above, the Sn-free composition exhibits the

highest power factor over the range 300-600 K, Figure 4-10d. For Pb-Sn mixed

compositions, NanlsSn3BiTe20 exhibits the highest power factor in the range 300-475 K
with a maximum of 11 uW/cm-K2 at 510 K. There is a markedly different behavior
between NanlgBlTezo and the other compositions. For NanlgBiTezo, the power factor

reaches a maximum of ~ 22 uW/cm-K2 at 470K and then decreases at higher

temperatures whereas the power factors for the other compositions rise almost linearly

with temperature (Figure 4-9d). The trend suggests that the maximum values lie well

above 570 K. The Pb-free composition NaSnlgBiTezo exhibits higher power factor than

the mixed Pb-Sn compositions reaching a value ~ 16.8 uW/cm-K2 at 618 K. The higher

power factor in this material arises from the very high electrical conductivity.

In order to evaluate the influence of the pnicogen metal M on the electronic
transport properties of Nan18-xSanTe20, the properties of corresponding compositions
need to be compared. Such comparison can be done by looking at Tables 4-3 and 4-4
where the electronic properties data for of Nanlg_xSnbeTe20 and of Nan13_xSaniTe20
are summarized. No particular trend could be observed with the nature of the pnicogen.
At room temperature NanlgBiTezo, Nan158n3BiTe20 and NaSnlgBiTezo had higher

electrical conductivity than their Sb counterparts but the values converged at high
temperature. The therrnopowers were similar at room temperature but at high temperature

the Bi analog exhibited a slightly higher value. For the x=5 stoichiometry, the Sb analog

126

exhibited higher electrical conductivity than the Bi couterpart but a lower therrnopower.

The electrical conductivity for Nan9Sn9SbTe20 and NanZSnmeTezo were higher than

their Bi analogs over the temperature range 300-670 K.

 

5000 -
4000 -
000 -

NO)

000 -

_\
O
O
O
I

 

 

 

Electrical conductivity (Slcm)

300 400 500 600 700
Temperature (K)

 

350 -
300 -

M
01
O

200-
150-
100-

 

Thermopower (WM)

01
O
l

 

 

 

300 '460' 560' 660' 760

Temperature (K)

 

 

(b)

 

 

6 2 A 6

é1b121h1s1a2b
x value

 

 

(d)

   

 

 

360' 460 560 600 760
Temperature (K)

Figure 4-10. Electronic transport properties of the Nan13_xSaniTe20 samples: (a)

electrical conductivity, (b) variation of the power law dependence of the electrical

conductivity (T4) as a firnction of x (the 7. parameters were extracted from the data of

panel (a)), (c) therrnopower, ((1) power factor. The point marks in the inset of panel (a)

identify all samples and apply to all panels (c) and (d).

127

Optical spectroscopy. Infrared specular reflectance spectroscopy was performed on
polished single crystal surfaces for further characterization of the transport parameters of
these heavily doped semiconductors. This technique allows a direct estimation of the
carrier effective mass by analyzing the infrared reflectivity data within the context of

single band Drude model. Figure 4-11 shows the room temperature specular reflectance
spectra of the compounds Nanlg_xSanTe20 (M=Sb, Bi; x=3, 5, 9). The numerical

analysis of the experimental data was done using a form of the dielectric function that
takes into account the existence of a plasmon-LO phonon interaction. The average
effective mass of carriers m* was obtained from the relation:

2 V2
wp=[ "8 .] MO)

808 m

on

 

where cop, n and a... are the plasma frequency, carrier concentration and the dielectric
permittivity at high frequency, respectively. The value of n was determined from the Hall
effect experiments and cap, and a... were extracted from fitting the reflectance data. From
the relative position of the plasma frequency (0,, in the spectra for each series, it can be
concluded that the number of carriers increases with the amount of tin as expected. This

is in agreement with the Hall effect experiments described above. The spectral analyses

gave room temperature effective mass of ~ 0.295 and 0.224 mg for Nan158n3SbTe20 and
Nanl3Sn5SbTe20. These results are in agreement with values reported for hole effective

mass in Pb1_xSnxTe by Kushev and coworkers.29

128

1.1

 

 

 

 

 

1.01 (a)

., 0.9-'

g 03:

a ' '

g 0.6: ng
0'5: \ 2”; Xx:9
0.4-

0?0 071 02 0.3 074 075 0.6
Energy(eV)
1.1
1.01 (b)

0 091+

g 0.8:

‘g 0.7-1

a 0-61 X:

II 0.5: X;3
0.4-

03

 

 

 

0:0 '011 02 ' 073 ' 0T4 ' 075 ' 0.6
Energy (eV)

Figure 4-11. Specular reflectance spectra of (a) Nan13-xSnbeTe20 (x=3, 5, 9) and (b)

Nan13-xSaniTe20 (x=3, 5, 9).

129

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

on R Sam 3% 2 .o £2 8 $8 aotmwamaz
8 am ES 3 mm iv 82 5 $8 oatmeamsmaz
w: 8 £8 £2 3v 2 : 5 meme oatmeammflaz
42 :. Sh :2 3v 2: Q 2% aobmocmsnaz
we ; wmm 3m 3v 8: 6 3% $9553.32
SN 2: «2 £0 2v m: H E 03% oatmmamm 5.32
gm 3 8m ti and ma _ S mace £0th :52
v. Waive: v. $92me 2:” Q mm a 5 5582.55

 

 

 

 

Sotmxcmxéfimwz 2&5me 05 mo $35qu 088820 98 3293a 05 mo E885 41v 033—.

.uoaomofiuofi Hm use 33:26:00 285020 Mb dam 6:3 ”mm 588 wage”: ”EH .Efimnoo 852 Na 55?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E 9. 3m 05. 2.3 .o 4:: 3 a8 avenmwamaz

g 3 t: 8% S v 82 6 $3 omoenmacmaaaz
a: 3 8w #2 S v z: 5 :8 omoenmeammflaz
5 a. 2K 22 3 v S: S 83 805253.32
m: E Se o3 £6 a: E was omoepmamefiaz
3m 8 m: oz. Ed on: Q :46 afipmamafiaz
%N 2: 3m 82 and mm: 8 23 omoenmafiaz
MfiNfiWSm Mcwwiamm 8m 93 "a mw 9a a .558..an

 

 

ouohpmxcmg Hafiz Eureka 2: mo moEomoa 3:982». 23 32932“ 05 .3 bagm .n... 03:.

130

 

 

C0

Thermal Transport Properties.

Nanlg.ISnbeTe20. Thermal diffusivity measurements were carried on the

Nan13-xSnbeTe20 series to determine the thermal conductivity and respective trends

with composition. The data could not be measured for NaSnlngTezo as the samples

tended to break during sample preparation. The resulting total thermal conductivities are
given in Table 4-5 and Figure 4-12a. For all compositions, the total thermal conductivity
decreases with increasing temperature. Compositions rich in Sn showed clearly higher

total thermal conductivity along the temperature range 300-700 K. For example, the room

temperature thermal conductivity for NanZSanbTezo is ~ 3.9 W/m-K, which is 2.7
times the value for Nan158n3SbTe20 (1.37 W/m-K). The room temperature thermal
conductivity of Nan.8_xSnbeTe20 (x=0, 3, 5), (1.85, 1.37 and 1.58 W/m'K respectively)
have values close to what was reported for Na0.95PbZOSbTe22 (1.8 W/m-K) (Table 4-5).

The lowest total thermal conductivity was ~ 0.84 W/m-K at 670 K for NanlngTezo. For
all compositions studied, the total thermal conductivities are systematically lower than
those reported for Pb1_xSnxTe samples.30

In order to understand the influence of the Pb:Sn ratio on the thermal transport,
the electronic and lattice components of the thermal conductivity were evaluated. The
electronic contribution was estimated using the Wiedemann-Franz law and the lattice
thermal conductivity was determined by substracting the electronic component to the
total thermal conductivity. It is clear that Sn—rich compositions have the highest
electronic thermal contributions because of their high electrical conductivity. A

comparison between the lattice components of the different compositions is shown on

131

Figure 4-12b. Over the temperature range 300-570 K, the compositions

NanZSn16SbTe20 and Nanssnlgstezo have a higher lattice thermal conductivity (1.6
and 1.4 W/m-K respectively at room temperature) than the compositions richer in Pb.
These values however are still lower than the value reported for PbTe (2.2 W/m-K).30 The
reduction is a consequence of the contribution of solid solution alloying to the reduction

in thermal conductivity. Above 500 K, the compositions Nan13$n5SbTe20 and

Nangsngstezo exhibit the lowest lattice component.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

55 S?

g 4.0- « E 3.0- - x=°

E. 35. ““ (a) 52.8" « ‘ ‘ (C) o x=3

E . 4‘ £26. 4 ‘ A x=5

,2 3.0. 4‘ I F0 2 2.4‘ ‘ ‘ v F9

‘6 .a, 4‘ . F3 1322- ‘4 ‘ F16

1:: 2‘5. w" A x=5 ”320.:VVV “‘4

§2.o‘ I “v vx§9 §]-gj,'._"v,

E15. ‘A'I "'v "i :13 'oi1.4-0“A! V'Vvvvv

E . ...:-.‘ vv" < x 16 512‘ o. . l.-

h ' I

31.0 .. “unlu- £10- .....:......I
. . r r I ~—. 0.8 . . . . .

3 300 400 500 600 700 800 E 300 400 500 600 700

.9 Temperature (K) ,2 Temperature (K)

9 :2

E .

~ . 4 x=0 E -

E 1-6 1 ‘ (b) : x=3 E . 4 (d) I x-O

a. 1.4' “ s x=5 E 12 “ . F:

I: » ‘ —9 - ‘ F

E12. 4“ ' :13 g 10. IV ‘4 v F5196

1”$1.0. " o .14 4 x=16 = = XI 4 ‘ x=

g 1'" o... 1‘ E 0.8- ' I“, ...O

3 0.8- : ff... 0.5131 g :llhnx

EO-S‘ “xi In“; T. 0.5. 4“

g 0.4- . E 04 “14

0 fl r r . . ' -

‘E 300 400 500 600 700 g 360 460 560 660 no

3 Temperature (K) 3 Temperature (K)

Figure 4-12. Temperature dependence of (a) the total thermal conductivity of the Nan13_
xSnbeTezo compositions and (b) their lattice thermal conductivity component (the
electronic thermal conductivity was estimated using the Wiedemann-Franz law with
Lo=2.45.10'8 WQK‘I); (c) the total thermal conductivity of the Nan18_xSaniTe20

compositions and ((1) their lattice thermal conductivity component.

132

 

 

 

 

 

 

 

 

 

 

 

 

2 2.0
E 1-3' : , (a) _ total
313' ’11 : electronic
5 1.2‘ , Zr a lattice
§1'0. él‘. . :74 77 ‘
20.8- ~ ,v 7 ., f
g 0.6- ; , -—_.-
a 0.4- _,_7
g 0.2- ' , ~
g 300 400 500 600 700
Temperature (K)

52 . , .
E- lg ._ g , (b) a total _
\ - 7 2 electronic
S 1.4+ 3 p .

1.3 i if. lattlce
g 121 ‘1
5 1.1: U f
S 1I0- r
‘53:- ~
7‘; 027- 7 .- ,.
E 0.6~ N’ a L- ,7. .
¢ 0-5 T T I I I JI"
is 300 350 400 450 500 550

Temperature (K)

2 3.0
g 2.8- a ‘ ‘7 _ (c) L total
E 3.3- , 1 electronic
.5 22: L ’1 r1}. lattice
,2 2.0- J ‘L
‘6 1.81 4fl
3 1.6- L
c 1.4- g
3 1.2-
3 1.0- .-
E 8.2- T ‘ ,
g 'V 300 400 560 600 700

Temperature (K)

Figure 4-13. Temperature dependence of the thermal conductivity for the compositions

(a) Nanlgstezo, (b) Nangsngstezo and (C) Nan5Sn13SbTe20.

133

The lattice thermal conductivity of Nan13Sn58bTe20 at room temperature is estimated ~
0.86 W/m-K, which is 40 % that of PbTe (2.2 W/m-K ).30 This value drops to ~ 0.48

W/m-K at 625 K. For Nan13Sn5SbTe20 and more Pb-rich compositions, the values of the
lattice thermal conductivity are quite low.and comparable to those of Nal-

,(PbmeyTemfl17 and Ag(Pb1.ySny)meTe2+m.l8 The relative contributions of electronic
and lattice thermal components vary for different x values. For example, the lattice and
electronic components are almost equivalent for Nan13Sn5SbTe20 (Figure 4-l3b). For

more Sn-rich compositions, the electronic contribution dominates (Figure 4-13a) whereas

for more Pb-rich compositions, the lattice component dominates (Figure 4-13c).

Nan 13.xSani T e 20. Figure 4-12c shows the total thermal conductivity of Nan13_

xSaniTezo (x=0.3, 5, 9 andl6). The compositions x=l3 and 18 could not be measured

due to the quality issue of the samples. The thermal conductivity decreases with

temperature for all compositions and decreases with decreasing Sn:Pb ratio. The lowest
thermal conductivity at 300 K was found for NanISSn3BiTe20 with value ~ 1.4 W/m-K.
At high temperature, the compounds x=O, 3 and 5 have similar thermal conductivity ~ 1

W/m°K. This value is as almost low as that observed for Nan13_xSnbeTe20 above. The

lattice thermal conductivity for the system Nan13-xSaniTezo was estimated using the

same method as described above for the Sb analogs (Figure 4-12d). The materials x=0, 3,
5 and 9 have close lattice thermal conductivities with value ~ 0.9-1.1 W/m-K at 300 K. At

675 K, the thermal conductivity falls down ~ 0.7-0.8 W/m-K. The decrease in thermal
conductivity is sharper for NanZSntiTezo with values ~ 1.3 and 0.4 W/m-K at 300 and

675K respectively.

134

When comparing the thermal conductivities of Bi and Sb analogs (Table 4-5 and
4-6), similar values were found for the x=0, 3 and 5 stoichiometries over the temperature
range 300-600 K. Nangsngst620 had lower thermal conductivity than its Bi counter
part. Only for the composition NanZSanTezO, the Bi analog exhibited lower thermal
conductivity.

In Figure 4-13, the figures of merit of the Nanlg_xSanTe20 materials calculated
from the above data are plotted. The Sn-free compositions NanlgMTezo exhibit the
highest ZT over the temperature range 300-675 K with a ZT as high as ~1.2 at 675 K. ZT
does not exceed 0.8 for the other Sb anaologs (Nanl3Sn5SbTe20) and 0.7 for the Bi
analogs (Nan13Sn5BiTe20). Even though Nan13Sn5MTe20 materials have low thermal
conductivities, its power factor is not as high as that of Na1_bemeyTem+2. This is

attributed to the presence of tin in the composition, which results in lower therrnopower
values. The temperature dependence of ZT varies with Sn fraction. For Sn—rich

compositions the slope around room temperature is less steep but it increases rapidly at

 

 

 

 

 

 

 

 

higher temperature.
1.4 0
. -0 I X:
1'2 :23 (a) I... 12 ox=3 (D) II.-
1.0 A x=5 .' 10 A x=5 I
v x=9 -' v x=9 .-
0'8 x=13 .' A“ 0.8 < x=16 .
AA
“05— {F16-II ‘A g. “06 I .2::."‘
0.4- I 7 04 I. 0'.A v'x
.OQQ .‘ 0 AA 'V‘d
0.2- " ‘3;"§<‘ 0.2- {.0293114‘
00- ”his!“ 00- ,;..x!<
300 400 500 600 700 360 430 500 600 7 )0
Temperature (K) Temperature (K)

Figure 4-14. Temperature dependence of the figure of merit ZT for (a) Nan13_

xSn,‘SbTe20 (x=0, 3, 5, 9, 13, 16) and (b) Nan18_xSaniTe20 (x=0, 3, 5, 9 and 16).

135

Table 4-5. Total and lattice thermal conductivities of Nan18_xSnbeTe20 (x=0, 3, 5, 9,

 

 

 

 

 

 

 

 

13 and 16) at 300 and 550 K.

Ktotal 1ctotal Klatt 1clatt

Composition (W/m'K) (W/m-K) (W/m-K) (W/m'K)

300 K 550 K 300 K 550 K

NanlngTezo 1.85 0.99 0.93 0.66
Nan15Sn3SbTe20 1.37 0.97 1.07 0.80
Nan13Sn5SbTe20 1.58 1.08 0.86 0.54
NangSngste20 2.13 1.47 1.28 0.64
Nan5Sn13SbTe20 2.80 2.07 1.38 0.89
Nanzsn16SbT620 3.88 3.22 1.60 0.94

 

 

 

 

 

Table 4-6. Total and lattice thermal conductivities of Nan13-xSaniTe20 (x=0, 3, 5, 9

 

 

 

 

 

 

 

and 16) at 300 and 550 K.

lctotal Ktotal Klatt Klatt

Composition (W/m-K) (W/m-K) (W/m-K) (w/qu

300 K 550 K 300 K 550 K

NanlgBiTezo 1.91 1.04 0.86 0.72
Nan15Sn3BiTe20 1.40 0.97 0.92 0.75
Nan13Sn5BiTe20 1.63 1.06 0.98 0.72
NangsngBiTezo 1.57 1.15 1.05 0.72
Nan28n16BiTe20 2.97 2.25 1.25 0.55

 

 

 

 

 

136

 

 

High Resolution Transmission Electron Microscopy. Based on a wide variety of
experimental results, a consensus is emerging that nanostructures are capable of causing
large reductions in the lattice thermal conductivity of materials. Because of the very low

lattice thermal conductivity observed for the compositions Nan18-xSnbeTezo series, we

have investigated the presence of any nanostructures in the bulk samples using HRTEM.

Figures 4-15a and 4-15b show HRTEM images of a sample with composition

Nan13Sn5BiTe20. Similar features were also observed for the Sb analog. Figures 4-15c

and 4-15d show HRTEM images of a sample with composition Nan13Sn5SbTe20.

Evenly dispersed nanoscale features ~ 10-20 mm in size can be observed in the images.

In order to learn about the composition and the lattice distortion of the inclusions
and the strain distribution at the boundaries, the sample Nan13Sn5BiTe20 was studied

using STEM, HRTEM and EDS. Figure 4-16a, a low magnification STEM image, clearly
shows varied size dark inclusions, which means the inclusions are rich in light elements.
(Figure 4-16b is a high magnification STEM image recorded along the [110] direction
still with inhomogeneous contrast due to the contribution of very small inclusions. We
also confirmed the ordered composition by line scanning profile of a selected array of
atoms (Figure 4-16c). The high intensity columns marked by blue arrows are Te atoms,

those pointed by red arrows mainly come from Pb.

137

 

4-15. Typical HRTEM images of (a) and (b) Nan13SnsBiTe20 and (c) and (d)

Figure

NannSHSSbTezo.

138

 

x10"4

        

1.0.21.4 1.8 2.0 2.2 2m26

nm

0 0.8

Figure 4-16. (a) A multitude of precipitates were observed in low magnification Z-
contrast image for Nan|3Sn5BiTezo; (b) high magnification STEM image showing the

weak contrast of some small precipitates; (c) Line scamiing profile of the selected atomic
array in (b), the positions that the dark arrows point out are Te, and those marked by gray

arrows mainly come from Pb.

Figure 4-l7a present a particular HRTEM image of Nan13Sn5BiTe20 showing a
large size inclusion within the matrix. The lattice parameter of the precipitate was found

to be twice that of the matrix (Figure 4-17a—b). Fast Fourier Transform images of the two

139

selected regions in Figure 4-17a are shown in Figure 4-1 7c and 4-l7d and confirmed the
difference in lattice parameter between the precipitate and the matrix. The nature and the
elemental compositions of these inclusions were analyzed by EDS. But accurate

quantitative analysis was not possible due to instrumental limitation. The inclusions were

found to be richer in Sn and poorer in Pb compared to the matrix.

 

Figure 4-17. (a) High magnification high resolution TEM image for Nan13Sn5BlTe20

and (b) enlarged part show the lattice parameters of the precipitate and that of the matrix

are different; (c) and (d) FFT images of the two regions C and D in (a).

140

Interestingly, aside from the discrete nanoprecipitates, nanoscale lamellar features also
were observed for Nan13SnsBiTe20. Figure 4-18a shows such lamellar stripes oriented

in the [001] direction with their boundaries along the [110] direction. Strain field

analysis3 1’ 32

performed with a geometric phase analysis (GPA) program was used to
investigate the variation in lattice parameter along these layers. Figure 4-18b shows the

power spectrum of the image 15(a) with g1=002 and g2: 220 chosen as the Fourier

vectors. The strain map profiles along the [001] direction (8”,) and the shear direction

(exy) are reported in Figures 4-18c, 4-18d, 4-18e and 4-1 8f. The line scan profiles indicate

that the lattice parameter along the [001] direction of the lamella B differs from that of
the lamella A by ~ 6%. This means that the lamellae A and B have a different lattice
constant. EDS analysis could not give any information about compositional fluctuation
because of the thin thickness of the layers. This lamella features may unlikely be due to
local segregation between PbTe and SnTe phases as they lattice constant differ only ~ 2

%. To our surprise, other lamellar features were observed for the sample
Nan13Sn5BiTe20 (Figure 4-19). Saw shape features were observed at the edge of the

sample. These stripes are oriented along the [112] direction, which is a different
orientation than that described above. This may the result of local ordering between PbTe
and SnTe phases. The Sn-Te bond is stronger than the Pb-Te bond (bond enthalpies ~ 369
and 251 kJ/mol respectively). Hence during the sample milling, the SnTe layer may be
more easily destroyed than the PbTe layer. The presence of both disperse precipitates and
lamellar features with nanometer size results in an increase in interfaces, which will

enhance phonon scattering and result in a reduction of the lattice thermal conductivity.

141

 

 

 

 

 

 

Figure 4-18. (a) Different precipitate from Nan13Sn5BiTe20 with different contrast

lamella structure labeled as ABAB... (b-i) are strain mapping for this precipitate; (b)
power spectrum image in (a) with g1= 002 and g¢= 220. (c) and ((1) 4.2.1. show the strain

map profile along the 001 direction (8”) and the shear direction(exy), respectively.

142

 

Figure 4—19. (a)-(d) Different magnification images for Nan13Sn5BiTe20 with another
type of lamellar structure. (a) only shows the saw shape at the edge of sample; (b) further
ion milling for sample, the lamella profile appear in the region close to edge, (e) strong
contrast was observed in the area far from the edge; (d) From the high resolution image,
the lamella was indexed as 112 direction, which is different from the first type lamella in

Figure 4-16.

143

4.4. Conclusions

The partial substitution of Pb by Sn in the Na1_bemeyTem+2 system causes

significant changes in the transport properties of these materials. High fractions of Sn
result in higher electrical conductivity and lower thermOpower. This increase in electrical

conductivity is mainly caused by an increase in the carrier concentration. As a
consequence the mobility of these carriers is relatively low (S 54 cmz/V-s). The increase
in electrical conductivity does not compensate for the decrease in therrnopower, thus
resulting in lower power factor compared to the parent system. The addition of Sn in the
system has an impact on the thermal properties. Sn—rich compositions have higher
electrical conductivities resulting in high electronic thermal conductivities. From all
compositions, NanlgMTezo, Nan13Sn5SbTe20 and Nan9Sn98bTe20 exhibit the lowest
lattice thermal conductivity values over the temperature range 300-650 K because of the
effect of solid solution alloying combined with the presence of nanostructuring. TEM
analysis on Nan13Sn5BiTe20 revealed the existence of different types of nanostructuring
in the sample. In addition to disperse inclusions, lamella features with thickness ~ 10 nm
were observed. Further reduction in the lattice thermal conductivity is achieved through
the combination of these nano-size features. The highest ZT~ 1.2 were obtained for the
Sn-free NanlgMTezo. Doping studies on the Nanl3Sn5SbTe20 showed an increase in
therrnopower with decreasing Na:Sb ratio. Further experiments on Na-deficient
compositions are in progress. The trivalent element i.e. Sb vs. Bi seems to affect the
mechanical properties of the materials with Sb analogs being less brittle. No clear trend

about the effect of the pnicogen on the electrical properties could be observed. Lattice

144

thermal conductivities are similar at room temperature but at high temperature lower

values are achieved for Sb analogs.

145

References

1. Di Salvo, F. J ., Science 1999, 285, 703.
2. Tritt, T. M.; Subramanian, M. A., MRS Bulletin 2006, 31, 188.
3. Snyder, G. J .; Toberer, E. S., Nat. Mater. 2008, 7, 105.

4. Harman, T. C.; Taylor, P. J.; Walsh, M. P.; LaForge, B. E., Science 2002, 297,
2229.

5. Harman, T. C.; Taylor, P. J.; Walsh, M. P.; LaForge, B. E., J. Electron. Mater.
2005, 34, L19.

6. Venkatasubramanian, R.; Siivola, E.; Colpitts, T.; O'Quinn, B., Nature 2001, 413,
597.

7. Caylor, J. C.; Coonley, K.; Stuart, J .; Colpitts, T.; Venkatasubramanian, R., Appl.
Phys. Lett. 2005, 87, (2), 023105.

8. Hicks, L. D.; Dresselhaus, G., Phys. Rev. B 1993, 47, 16631.
9. Hicks, L. D.; Dresselhaus, M. S., Phys. Rev. B 1993, 47, (19), 12727.

10. Kim, W.; Singer, K. L.; Majumdar, A.; Vashaee, D.; Bian, Z.; Shakouri, A.; G.,
2.; Bowers, E. J .; Zide, J. M. 0.; Gossard, C., Appl. Phys. Lett. 2006, 88, 242107.

11. Hochbaum, A. 1.; Chen, R.; Delgado, R. D.; Liang, W.; Garnett, E. C.; Najarian,
M.; Majumdar, A.; Yang, P., Nature 2008, 451, 163.

12. Boukai, A. 1.; Bunimovich, Y.; Tahir-Kheli, J .; Yu, J .-K.; Goddard, W. A.; Heath,
J. R., Nature 2008, 451, 168.

13. Hsu, K. F.; Loo, S.; Guo, F.; Chen, W.; Dyck, J. S.; Uher, C.; Hogan, T.;
Polychroniadis, E. K.; Kanatzidis, M. G., Science 2004, 303, 818.

146

14. Quarez, E.; Hsu, K. F.; Pcionek, R.; Frangis, N.; Polychroniadis, E. K.;
Kanatzidis, M. G., J. Am. Chem. Soc. 2005, 127, 9177.

15. Androulakis, J.; Lin, C. H.; Kong, H. J.; Uher, C.; Wu, C. 1.; T., H.; Cook, B. A.;
T., C.; Paraskevopoulos, M.; Kanatzidis, M. G., J. Am. Chem. Soc. 2007, 129, 9780.

16. Poudeu, P. F. P.; D'Angelo, J.; Kong, H. J.; Short, J. L.; Pcionek, R.; Hogan, T.;
Uher, C.; Kanatzidis, M. G., J. Am. Chem. Soc. 2006, 128, 14347.

17. Poudeu, P. F. P.; D'Angelo, J.; Downey, A. D.; Short, J. L.; Hogan, T.;
Kanatzidis, M. G., Angew. Chem, Int. Ed. 2006, 45, 3835.

18. Androulakis, J.; Hsu, K. F.; Pcionek, R.; Kong, H. J.; Uher, C.; D'Angelo, J.;
Downey, A. D.; Hogan, T.; Kanatzidis, M. G., Adv. Mater. 2006, 18, 1170.

19. Han, M.; Hoang, K.; Kong, H. J.; Pcionek, R.; Uher, C.; Paraskevopoulos, M.;
Mahanti, S. D.; Kanatzidis, M. G., Chem. Mater. 2008, 20, 3512.

20. Wendlandt, W.; Hecht, H. G., Reflectance Spectroscopy. New York, 1966.
21. Kotuem, G., Reflectance Spectroscopy. New York, 1969.

22. Tandon, S. P.; Gupta, J. P., Status Solidi 1970, 38, 363.

23. Kittel, C., Introduction to Solid State Physics. Wiley: 2005.

24. Bouad, N.; Chapon, L.; Marin-Ayral, R. M.; Bouree-Vigneron, F.; Tadenac, J. C.,
J. Solid State Chem. 2003, 173, 189.

25. Rogacheva, E. 1.; Gorne, G. V.; Laptev, S. A.; Arinkin, A. V.; Vesiene, T. B., J.
Solid State Chem. 1982, 43, 364.

26. Orihashi, M.; Noda, Y.; Chen, L.; Hirai, T., Mater. T rans., JIM 2000, 41, 1196.

27. Henger, G.; Peretti, E. A., Journal of the Less-Common Metals 1965, 8, 124.

147

28. Ravich, Y. 1.; Efimova, B. A.; Smimov, I. A., Semiconducting Lead
Chalcogenides. New York, London, 1970.

29. Kushev, D. 8., Phys. Stat. Sol. (b) 1980, 100, 731.

30. Orihashi, M.; Noda, Y.; Chen, L. D.; Goto, T.; Hirai, T., J. Phys. Chem. Solids
2000, 61, 919.

31. Williams, D. B.; Carter, C. _B., Transmission Electron Microscopy. Springer:
1996.

32. Hytch, M. J .; Snoeck, E.; Kilaas, R., Ultramicroscopy 1998, 74, 131.

148

Chapter 5
Synthesis and Characterization of the KPb13_xSnbeTe20

(x=0, 2, 5, 9, 13, 16 and 18) series

5.1. Introduction

Development of efficient thermoelectric devices could have a significant impact
on our daily lives: Cooling devices may help in faster computing processes whereas
devices that generate power could reuse waste heat from a vehicle’s exhaust pipe to
supply additional energy to the engine. For such devices to be economically viable,

materials with enhanced figure of merit need to be developed. The figure of merit ZT is
defined as ZT = OSzT/K with o, S and K being respectively the electrical conductivity, the

therrnopower and the thermal conductivity of the compound. The thermal conductivity

has two components: electronic and thermal contributions. Current devices use materials

with maximum ZT ~ 1 at their operating temperature.1 Competitive devices should have
ZT ~3-4.2 Over the last decade, significant improvement of the figure of merit has been

reported for low dimensional materials}5 and mixed-phase nanocomposites.6'9
Theoretical studies first predicted enhancement of the power factor in quantum confined
system.8’ 10 The nano-components also play an important role in scattering mid-to-long

wavelength phonons, thus reducing the lattice thermal conductivity.

The proof-of-principle of these concepts was confirmed experimentally. In

uniform BizTe3/BizTe2.33Se0_17 superlattices of ~ 6 nm period, Venkatasubramanian et al.

149

were able to reduce the lattice thermal conductivity at room temperature in the cross-
plane direction to ~ 0.58 W/m°K compared to ~ 1.7 for bulk BizTe3.5 Harman et a1.
prepared superlattices made of PbSe quantum dots sandwiched between layers of PbTe
over thousands of arrays.3’ 4’ 11 Using Bi as a n-type dopant, ZT ~ 1.6 and 3.5 were
reported for these quantum dot superlattices at 300 and 570 K respectively. Our group

has reported on the n-type system AngmeTem+2 (so-called LAST system)12 with ZT ~

1.8 at 725 K for Ang13SbTe20. The reason for such good thermoelectric properties of

these materials comes from their extremely low thermal conductivity (~ 1 W/m-K at 780
K). Careful TEM analysis revealed the material not to be a true solid solution but to

12, 13

contain nano-size inclusions dispersed in the PbTe matrix. The presence of such

inclusions is believed to scatter phonons more efficiently than electrons, thus a significant
reduction in lattice thermal conductivity without much impact on the electronic
properties. This system is a good illustration of the concept of bulk nanostructured
materials, which is more viable for high scale production than low dimensional systems.
We have extended that research area by preparing alkali metal analogs of the

LAST system. It has been reported that the combination of alkali metal with Sb tends to

form stable clusters of the type AnSb4 (A=Na, K).14 As a consequence, alkali metal

based, Sb-doped PbTe systems have higher potential to form nanostructures when

compared to the Ag-based analog system. Moreover, the formation of such clusters is

favored by the low solubility of Sb in PbTe.15 The Na1-bemeyTem+2 system (so-called

SALT)16 exhibits promising p-type thermoelectric properties with ZT value as high as

150

1.7 at 650 K for the composition Nao_95Pb20SbTe22. Such a high figure of merit derives

mainly from its low total thermal conductivity (~ 0.85 W/m°K at 675 K). Nano-size
inclusions embedded in the PbTe matrix were observed by TEM but no compositional

analysis could be performed due to instrumental limitation.
Surprisingly, the potassium analogs K1-bem+5Sb1+7Tem+2 (l-x = 0.85, 0.9, 0.95;

m = 19, 20, 21; 5 = 0, 0.2, y = 0, 0.2)17 are n-type materials with electrons as major

charge carriers. The therrnopower has values as high as ~ -250 uV/K at 700 K for all
compositions. This result agrees with ab initio calculations. Calculations within density
fiinctional theory (DFT) of PbTe with and without monovalent impurities showed an

increase in the density of states (DOS) near the top of the valence band for Ag and K

18

dopant whereas Na does not change the DOS within 0.5 eV of the band maxima. A

maximum ZT of 1.6 was achieved at 750 K for the n-type composition
K0,95Pb20Sb1,2Te22. As for the LAST and SALT systems, the specimens exhibited low
lattice thermal conductivity. The room temperature value for samples with m = 19 range

from 1 W/m-K for the composition K0,3Pb19Sb1_2Te21 to 0.6 W/m~K for the sample
K0,3Pb19,2SbL2Te21. These values are two to four times lower than the typical lattice

thermal conductivity of undoped PbTe (2.2 W/m.K).19 At 550 K the values are between

0.4 and 0.7 W/m-K. Similar values of the lattice thermal conductivity were obtained for
samples with m = 20 and 21. TEM studies again confirmed the presence of nanocrystals
randomly oriented in the matrix. The presence of such nano inclusions in PbTe is a key

feature to improve the figure of merit of PbTe.

151

Partial substitution of Pb by Sn into the LAST system appeared to change

significantly the transport properties of the system. ZT ~ 1.45 at 627 K was reported for

the p-type composition Ago,5Pb6Sn2Sb0,2Te10.20 In order to evaluate the impact of partial
substitution of Pb by Sn on the properties of K1-bem+5Sb1+7Tem+2, the compositions

KPb13-xSanTe20 (x=0, 2, 5, 9, 13, 16 and 18, Bi) were prepared. In order. to study the
effect of the pnicogen, Bi analogs were synthesized as well. The homogeneity of the cast
ingots were investigated with powder X-ray diffraction, DTA and SEM. Thermoelectric
measurements on Sb and Bi analogs over the temperature range 300-600 K indicated a
switch from n-to-p type for different Pb/Sn ratios. In the case of Sb-doped materials, all

compositions with x 2 5 were p-type. For Bi analogs, the change in charge carrier type

type appeared for the composition KPngn9SbTe20.

5.2. Experimental

5.2.1. Synthesis

All samples were prepared by mixing appropriate ratios of high purity elemental
starting materials in silica tubes. Before loading, the tubes were carbon-coated to avoid
any reaction of the potassium metal with the silica. All components (except K) were
loaded into the silica tubes under ambient atmosphere and the corresponding amount of K

was later added under inert atmosphere in a dry glove box. The tubes were then sealed

under residual vacuum of ~ 10.4 Torr and heated at 980 °C for 4 hours in a furnace

mounted on a rocking table. During that period, the furnace was allowed to rock for two

hours in order to get a homogeneous melt. After, the furnace was maintained in vertical

152

position for crystal growth. The tubes were slow cooled to 550 °C (10 °C per hour) and
then cooled rapidly to room temperature. The resulting ingots were silvery-metallic in
color with a smooth dark surface. The samples were found quite brittle. During polishing,
leaching issues were observed when the samples were in contact with water. After drying
with acetone, the samples were stable.

In order to improve the mechanical strength of the ingots, the Sb-analogs were prepared
with a slower cooling rate (5 °C per hour) but the samples were even more brittle.

Table 5-1 and 5-2 summarize the amount of elements used for each reaction.

Table 5-1. Amounts of elements used to prepare KPb13-xSnbeTe20.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Composition :1: 01) 31:11:01) Enliingol) 3:501) {13:01)
K. , 21.1.31 .1.- 11131 1 2.111.711.2121
K... . 222,1 2.1.7.11 3.1.1.311 9.11.71 1.31.31
WW... ,2, 221.131 21.31.312.331 2.1.1.31
KPb98n9SbTezo $.15? 32'??? is??? 31' T35; g 358%
KPbSSnnSbTeZO gzlglgg flilggf (13364;?)‘3 £12.3g1136g 251672979)g
KPbZSnmeTezo 22191sz $528188; g 34578866)g $395313 (755808632)g
Ksmgstezo 232%? O 8(65'3.52852)g $5873; g is???

 

153

 

Table 5-2. Amounts of elements used to prepare KPb13_xSaniTe20.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

COHIPOSitiC’n 51,1301 friiingol) (Srlriingol) (Sriiingol) €301)
KPblgBiTezo 2,932; $3119??? 0 315;? g 34mg
K1b..Sn.Bi-re. 25,2651 3.31.131 231.35 g €376.33 g 221?.) ,
K.....B... 3,832; 5.53:? if??? 1 252.38 g 54:2?
....... 2.1221 21.-31.3.31 3.2.3.31 2.1.1.311 1.13.31
K......... 2.33.1 3.329931 2.23.? 35533.5 , 6.55.75?
mm. 2.3131 28.8311 2213.12.129123s
KSnlgBiTezo 22'??? 0 (542,68? 32.55335 g (655121751)g

 

5.2.2. Characterization techniques

 

 

Powder X-ray Diffraction. Powder X-ray patterns of the grinded materials were
recorded using Cu Kn radiation on a CPS-120 Inel X-ray powder diffractometer operating
at 40 kV and 20 mA equipped with a position sensitive detector.

DTA Analysis. Differential thermal analysis (DTA) data were collected with a Shimadzu
DTA-50 thermal analyzer. Approximately 35 mg of finely ground powder of material
was sealed in a carbon-coated quartz ampoule under residual pressure of ~ 10'4 torr.
Another ampoule containing similar amount of alumina and prepared the same way was
used as a reference. The samples were heated to 1273 K at a rate of 10 K/min, held at
1273 K for two minutes and cooled down to 323 K at a rate of -10 K/min.

Scanning Electron Microscopy. The surface of several samples was polished very
carefully using silica suspension solution (0.05 pm) in order to get a smooth, mirror-like

surface. The samples were then studied with a seaming electron microscopy (Hitachi

154

 

S34OON-II) with 25 kV acceleration voltage using both energy-dispersive spectroscopy
(EDS) and back-scattered electron imaging (BSE).

Electrical Transport Properties. Thermopower and electrical conductivity properties
were measured simultaneously under helium atmosphere using a ZEM-3 Seebeck

coefficient/electrical resistivity measurement system (ULVAC-RIKO, Japan). Samples
for transport measurement were cut to size 10 x 3 x 3 mm3 using a diamond saw (Buehler
isomet 1000), a wire saw (South Bay Technology) and a polishing machine (Buehler
ecomet 3000). Rectangular shape samples with approximately 3 x 3 mm2 cross-section

were sandwiched vertically by two nickel electrodes (current injection) with two Pt/PtRh
thermocouples (for temperature difference and voltage measurements) attached on one
side. The sample and measurement probes were covered by a nickel can to maintain a
constant temperature during the measurement and the base temperature was measured by
a thermocouple attached to the outside of the can. The sample, electrodes, and nickel can
were placed in a vacuum chamber then evacuated and refilled with He gas (0.1 atm) to
provide necessary heat transfer. Properties were measured from room temperature to 670
K under helium atmosphere.

Thermal Conductivity. The thermal conductivity was determined as a function of
temperature using the flash diffusivity method on a LFA 457/2/G Microflash Netzsch.
The front face of a small disc-shaped sample (diameter ~ 8 mm; thickness ~ 2mm) coated
with a thin layer of graphite is irradiated with a short laser burst, and the resulting rear
face temperature rise is recorded and analyzed. The experiments were carried out under

nitrogen atmosphere. Thermal conductivity values were calculated using the equation K:

a Cp d, where a is the thermal diffusivity, Cp the specific heat and d the bulk density of

155

 

 

the material calculated from the sample’s geometry and mass. A pyroceram reference
was used to determine the heat capacity of the sample. The thermal diffusivities were
measured typically over the temperature range 300-670 K. The electronic component of

the thermal conductivity was quantified through the Wiedemann-Franz law according to
which Ire, = a.T.L0 (Lo being the Lorenz number, L0=2.45.10'8 WQK'1 .2! The lattice

contribution was then derived by subtracting the electronic component from the total

thermal conductivity.

5.3. Results and discussion

Structural Characterization. Figure 5-1 shows the X-ray powder diffraction for the
compositions KPblg_xSanTe20 (x=0, 2, 5, 9, 13, 16 and 18, M=Sb, Bi). These patterns
can be indexed as NaCl-type structure. No additional peaks were observed. The Braggs

peaks shift towards higher diffraction angle with increasing Sn/Pb ratio, consistent with a

decreasing lattice constant. The variation of the lattice constant as a function of x is

plotted in Figure 6-lb and 6-ld. This trend is expected as the radius of Pb2+ is larger than
that of Sn2+. For example, the lattice constant varies from 6.448 (1) A for
KPbléanSbTezo, which is closer to PbTe (6.459 A)22 to 6.340 (2) A for
KPbZSaniTezo, which is closer to SnTe (6.328 A)23.

Figure 5-2a shows DTA data for KPblgstezo. The plot shows only one

endothermic peak of melting during the heating cycle and one exothermic peak of

crystallization on the cooling cycle. Similar curves were obtained for the other

compositions KPb18_xSnbeTe20 and their Bi analogs. Melting and crystallization points

156

 

 

for the KPb18_xSanTego series are summarized in Tables 5-3 and 5-4. The melting point
decreases with increasing Sn fraction. It varies from 1175 K for KPb16Sn3SbTe20 to 1071
K for KszSHléSbTezo (Figure 5—2b). This trend is consistent with the expected solid

solution alloying between the high melting PbTe (1196 K) and the lower melting SnTe

(1063 K) compounds.

 

(a) 6-48' - (b)

Relative intensity

3 6.34- -

 

 

 

 

20 4o 60 so 6 i Z1 ('5 é1'o1'21'41'61'8
2theta (deg) x

 

(c) 6.46- ‘ - (d)

Relative intensity

9
5
2
0

xxxx
II" II II

A ' '

 

 

 

 

20 4o 60 80 c3 i i1 6 1'3 1'0 1'2 1'4 1'6 1'8
2 theta (deg) x
Figure 5-1. X-ray powder diffraction patterns of KPblg_xSanTe20 with x=0, 2, 5, 9, 13,

16 and 18 for (a) M=Sb and (c) M = Bi; variation of the unit cell parameter as a function

ofx for(b) M= Sb and (d) M = Bi.

157

 

 

 

 

 

 

 

 

 

01 (a)
-501
1162K
E4001
I-<- 4501
O 1 lexo
-200-
. l endo 1191 K
‘250 l v I - I . I r 1
400 600 800 1000 1200
Temperature (K)
1200: _ (b)
1180: .
$11601: '
9311401 . :
1.311201
“E1100:
£10801 . Melting point . -
10601 . Crystallization point . 2
1040 fiI—' I ' I ' I ' I ' I ' I ' I fj ' I
0 2 4 6 81012141618
x

Figure 5-2. Typical DTA results of the composition KPblngTezo, 0)) variation of the

melting and crystallization points of KPb13-xSnbeTe20 as a function of x.

Scanning Electron Microscopy. The surface of several samples was polished very
carefully and analyzed using both secondary and back-scattered electron microscopies. In
the case of Sb analogs, inclusions of ~ 60-70 mm diameter, randomly dispersed inside the
matrix, were observed. In average, these inclusions are ~ 500 um far away from each

other. Figure 5-3a (KPb13Sn5SbTe20) and 5-3b (KPngflngTCzo) show typical images of

such inclusions. It must be noticed that these inclusions are widely disperse in the matrix.

 

EDS analysis of a large area of the inclusion in Figure 5-3a gave the resulting atomic

158

percentages: Pb: 2.73, K: 4.59, Sb: 27.59, Te: 65.09. A peak for oxygen was also
observed. Probably a phase containing K formed upon cooling and tends to oxidize as it
is water-sensitive. EDS analysis on the matrix indicated the presence of Sn, Pb and Te in
ratio consistent with the nominal composition. EDS analysis on the inclusion observed

for KPngnngTe20 gave a higher amount of K: Pb: 2.32, K: 14.87, Sb: 23.38, Te: 59.43.

Few areas rich in Sb were observed. Figure 5-4a for example shows an inclusion of Sb in

KPb5Sn13SbTe20. Figure5-4b shows a different inclusion observed in the same sample.

Different phases are visible: EDS analysis on area 1 indicated the phase to be Sb2TC3 (Sb:

37.95 at %, Te: 62.05 at. %). EDS analysis on the area 2 gave the results Pb: 6.32, K:
3.80, Sb: 13.51, Te 76.37 (at. %).

Similar analysis was carried on Bi analogs. Figures 5-5a and 5-5b show different
inclusions observed for KPngngBiTezo. Two different phases are distinguishable in the

inclusion in Figure 6-5a. The bright area 1 contains only Bi whereas the area 2 is a
quaternary phase. The EDS analysis gave the following atomic percentages: Pb: 8.14, Sn:
15.02, Bi: 32.17 and Te 44.67. This is not charge-balanced. But as Bi and Pb have close
L emission lines, there is a larger uncertainty on the respective Pb and Bi amounts than
for the other elements. EDS analysis on a large area on the inclusion. in Figure 6-5b
indicated the area to be rich in Te: Pb: 1.36 at.%, K: 4.77 at.%, Sn: 22.90 at.% and Te :

70.95 at.%.

159

 

 

53403 :5 CW 10 1mm ~1:1 20k BSECCIMP 1313212008 1141 i 401131.111

133400 25 CW 10 4mn~ x470 BSECOMP 101373126012 {5:159 '

 

Figure 5-3. BSE images of the surface of (a) KPb13Sn5SbTe20 showing a precipitate with
the atomic percentages Pb:2.73, K: 4.59, Sb:27.59 and Te:65.09 and (b) KPb9sngst€20
showing an inclusion with atomic percentages Pb:2.32, K: 14.87, Sb:23.38 and Te:59.43.
EDS analysis on the matrices indicated Pb, Sn and Te with ratios consistent with the

nominal composition.

160

1 25 011V 11 8mm xi 1301. BSECOMP1 "1120120118 22 28 '30 01.1111

.1

25 01v 11 8m 11 x1 30k BSECOMP '101'301’200822 '38 ' 1'10 011111

 

Figure 5-4. BSE image of (a) a Sb-rich inclusion and (b) a mixed—phase inclusion in
KPb5Sn13SbTe20. EDS on area 1 showed the presence of szTe3 and EDS on area 2 gave

the atomic percentages Pb:6.32, K: 3.80, Sb: 13.51 and Te:76.37.

161

 

Figure 5—5. BSE images of inclusions (a) and (b) observed for KPngbgBiTezo. Different
phases are visible in the inclusion (a): area 1 is pure Bi whereas area 2 contains Pb, Sn, Bi
and Te with the following atomic percentages: 8.14, 15.02, 32.17 and 44.67. EDS
analysis on inclusion (b) gave the atomic percentages Pb: 1.36, K: 4.77, Sn: 22.90 and

Te: 70.95.

162

Optical band gap measurements. Infrared spectroscopy of KPb18_xSanTe20 (x=0, 2,

5, 9, 13, 16 and 18, M=Sb, Bi) indicate the samples to be narrow gap semiconductors.

The band gaps of the different samples are summarized in Table 5-3 and 5-4. For the Sb
series, a minimum band gap is observed for the x=9 composition. For Nan18_
xSnbeTezo, the lowest band gap was observed for x=9. For the Bi analogs, the
compositions x=5 and x=9 showed a lowest band gap 5 0.05 eV. These observations
agree with trends reported previously: for example, several investigation524’ 25 have
shown that the band gap of the Pb1-xSnxTe system decreases as the tin content increases,

goes through zero and then increases again with further increase in tin content.

Electronic Properties.
KPb18_xSnbeTe20. The electronic transport properties of the KPb13-xSanTe20
systems can be tuned by varying the Sn:Pb ratio. Figure 5-6a shows the evolution of the

electrical conductivity as a fiinction of temperature for KPb16SHZSbTCZO,
KPb 13Sn58bTe20, KPngnQSbTBZO, KP b5 SD 1 3 SbTCZO, 1(sz Sn 1 68bTe20 and

NaSnlgstezo. The thermoelectric pr0perties of KPblgstezo could not be measured due

to the high brittleness of the sample. Regardless of the composition, the electrical
conductivity decreases with increasing temperature, indicating degenerate conduction for

the whole temperature range. The electrical conductivity increases with increasing Sn

concentration with the exception of KPb13Sn5SbTe20. The room temperature value of the
electrical conductivity of KSn18SbTe20 was ~ 3875 S/cm, which is about ten times that of

KPb16Sn28bTe20. This trend is the same as that observed in the case of Nan18.

163

xSn,.SbTe20. Presumably, in the KPb13-xSnbeTe20 system, vacancies become more and
more dominant with increasing fraction of SnTe (i.e. x).
The composition KPb16Sn2SbTe20 exhibited n-type behavior whereas positive

therrnopower values were measured for KPbl3Sn5SbTezo, KPngn9ste20,

KPb5Sn13SbTe20, KszsnmeTezo and NaSmngTezo (Figure 5-6b). By introducing Sn
into the system, more holes are added and for x between 2 and 5, the concentration of
holes surpasses that of electrons. For KPblGSn2SbTe20, the therrnopower varies between -
159 uV/K at room temperature and ——269 uV/K at 559 K. For p—type samples, the
therrnopower increases with decreasing amount of Sn. KPb13Sn5$bTe20 exhibits the
highest thermOpower among the p-type compositions over the temperature range 300-550
K. KPb13Sn5SbTe20 has therrnopower ~ 161 uV/K at room temperature and reaches a
maximum of ~ 275 uV/K at 462 K. At higher temperatures, the therrnopower
KPngn9SbTe20 becomes higher than that of than KPb13Sn5SbTe20.

The power factors calculated from the above data are plotted in Figure 5-6c.
Clearly, different trends are observed. The n-type composition KPb16Sn2SbTe20 exhibits
the highest power factor over the temperature range 300-550 K. Its power factor is almost
constant, ~ 12 uW/cm-KZ. The power factor for KPb13Sn5SbTe20 decreases with
temperature from 10 uW/cm-K2 at room temperature to l at 600 K. For the other
members, the power factor increases rapidly with temperature. For example, the power

factor of KPbZSnmeTezo increases from 0.9 uW/cm-K2 at room temperature to 12.5

uW/cm-KZ at 600 K.

164

 

 

 

4000 - a

.— 1 31a
\.
g 3500 - K,
‘ 1 ‘3\
U) _
.. 3000 - K1
‘3; is .
2 1 1., \n
.2 2500 - .. ,_ -,
W 1.1 . .\
0 ' E] 1. Q}
g 2000 - .,
c a V b \ m ‘G
8 1500 - ~ \\ ,
V“ “B
- ‘ \1 \ \\$\\
8 1000 I A‘C‘.\‘ ’ \V‘sz\"\ EX‘E
': ““1..- ‘\\,7\
.6 1 '\-"~~\A . -‘\_\v\
0 500 . U'D‘AC H‘\L:‘.~‘_LX r \T‘? _ ‘HV
- 0‘ “C‘D-e- “1hr. .
I O‘\ CROWD—fl _ _‘A‘A
m o O‘““*>—~~-o~cJ El 0‘ ~14
fi f I u I '

 

300 350 400 450 500 550 600

 

 

 

 

 

 

 

 

Temperature (K)
300 ..
‘ , 1/ \2 (b)
200 -1 / /- ‘o§\
A I C’: . A’/\’ \
a: .. «./~ fi/,_/ , .
51 1001 {.221 2%: :2 17;? ._- a
: *H’fi (:1 :'7‘_”._'éT—'—-':’ r ‘J r r -,
g 01 ' 1L- , x:
O - 10 x=5
:- —~1—x=9
E -100- fi x=13
° 1 .— 113 x=16
.3 "1:1,, _ _. =
.- ‘200- L"‘3~.-U_\U-‘Uh-fl L X 18
I D\D\U\D
-30°1-.v----....,
300 350 400 450 500 550 600
Temperature (K)
16
14.: F "\"n‘~r:~—C \\\\\ L; (C)
I I “.U‘
A .. :i ‘n\ g:
N! 12 ‘ b“~c~\_3 B/
E _ ’g'
2 1O ‘ 040 7- in“; ,/ A)
g 6' .1’ygi“‘ei";‘5/
' or” raj..1xi', '
E 4 I a.“ #:,’1/':/,3 \‘Qxx
I N, H 4 1/1. \\
300 350 400 450 500 550 600
Temperature (K)

Figure 5-6. Electronic transport properties of the KPblg-xSnbeTe20 samples: (a)

electrical conductivity, (b) therrnopower, (c) power factor. The point marks in the inset of

panel (b) identify all samples and apply to panels (a) and (c).

165

Nan18.xSaniTe20. Because of the high brittleness of the ingots, only the

thermoelectric properties of KPb13-xSaniTe20 (x= 5, 9, l3 and 18) were measured. The

temperature dependence of the electrical conductivity of the compositions x=5, 9, 13 and
18 are plotted in Figure 5-7a. All compositions are degenerate semiconductors with
electrical conductivity decreasing with temperature. For the Sb analogs, the electrical

conductivity increases with increasing amount of Sn. At room temperature, the electrical

conductivity of KSnlgBiTezo is ~ 2347 S/cm, which is 8 times larger than that of
KPb13Sn5BiTe20 (296 S/cm).

The composition KPb13Sn5BiTe20 showed n-type behavior (Figure 5—7b). At room
temperature the therrnopower is ~ -100 uV/K and a maximum of ~ -180 uV/K at 585 K.
In the case of KPb13Sn5MTe20, the difference in the nature of charge carriers between the
Sb and Bi analogs emphasizes the importance of the pnicogen atom. The compositions
KPb13-xSaniTe20 (x=9, 13 and 18) have positive therrnopower values which increase

linearly with temperature. The x=9 sample has the highest therrnopower over the range

300-600 K with values between 61 uV/K at 300 K and 230 uV/K at 606 K.

166

 

 

 

 

 

 

 

 

 

 

€25“) :1 (a) +x=5
o ' " =
1712000- \1 +:='1'3
E . \7,_ \ d:— -18
E 1500 ' "'\“1'\. \\7‘1 W'— x-
g ‘ "\7\. \G'
g 500 - K‘vam '\v‘v‘v
é O-O-oio-0-0-O.C::132%§1%1%-010
m 300 400 500 600 700
Temperature (K)
250 ”A
2001 (b) WM,”—
1.1 150-} M’s" 1"
E 100' 1.’13"""\ ,1—v-‘1;'{/3
1 53' 1.11 ’ "
$2-501
2400" 0-- 1-..‘
l—-;03 1 O‘O‘O‘O-O-o-o-o—O'O'O
300 400 500 600 700
Temperature (K)
141
<3“ .
E 12 -
$101
1:1 1-
13 1-
11 4-
g :21
o .
0.

 

 

 

300 ' 400 ' 500 ' 600 ' 700
Temperature (K)

Figure 5-7. Electronic transport properties of the KPb13-xSaniTe20 samples: (a)

electrical conductivity, (b) thermopower, (c) power factor. The point marks in the inset of

panel (a) identify all samples and apply to all panels (b) and (c).

167

A recent study on the influence of the pnicogen on the properties of the n-type

Ag1_be13MTezo materials indicated larger therrnopower values for Sb analogs compared

to Bi samples.26 Band structure calculations for Ag1-be13MTe20 were performed using

the lower energy configuration (i.e. Ag and M pairs are the second nearest neighbors of
one another). There are band splittings due to the presence of the (Ag,M) pairs because the
symmetry is lowered compared to the undoped PbTe. The top of the valence is perturbed
predominantly by Ag with an impurity-derived band formed predominantly out of Te p
and Ag (1 states. The bottom of the conduction is, on the other hand, perturbed
predominantly by Sb or Bi. In the case of the (Ag, Sb) pair, a group of three nearly
degenerate bands forms the lowest conduction band with a nondegenerate band (at ~0.06
eV above that group) at the F point (Figure 5-8a). In contrast, in the case of the (Ag, Bi)
pair, the nondegenerate band forms the lowest conduction band and the group of three
nearly degenerate bands is ~0.04 eV above the nondegenerate band at the F point (Figure
5-8b). Due to this basic difference in the arrangement of the bands and lower carrier
concentration in the Sb analog, one expects to see larger negative Seebeck coefficient in Sb
compounds compared to the Bi compounds.

Combining the above data, the resulting power factors are plotted in Figure 5—7c.

For the composition x=5, the power factor is constant over the temperature range but
does not exceed 3 uW/cm-Kz. The composition x=9 reaches a maximum ~ 9.5 uW/cm-K2
at 509 K. For the other members, the poser factors increases linearly with temperature
with a maximum power factor ~ 14 iiW/cm-K2 at 660 K. Such values derive mainly

through a high electrical conductivity.

168

 

 

   

  

 

 

PbTe:(Ag,Sb)

      

  

 

  

L\\L:

 

 

3 0. 3.
0 0 0

33 >9on

 

 

0 0 O.

93 35cm.

Figure 5-8. Band-structure of PbTe doped with (a) (Ag, Sb), (b) (Ag, Bi). Ag and Sb/Bi

atoms in the pair are the second-nearest neighbors of one another in a 64-atom supercell.

169

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m: on S: 9.8 So 2 E: S 23 oatmeamx
3c 2 ME: S 5.8 8.952230

42 8 ms ow: 3o 1 8: 5 £3 sotmmamfimm
SN S a. £2 35 2 mm: 3 83 oabmamsmv.
8:. NS- 2: 3m 26 S: 5 was sotmmameém
43 me: Q 8% oatmamsfim

mac .2: _ E $4.0 seaming.

v. ammuéxmsm v. “meme: 2% 93 um mm as a 838.55

 

 

Omar—._mxcmg REM 235:2: 2: mo moEoQoa Baotou—o can _mofibfi 05 mo bagsm .Tm 935.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

K m _E 3% 3o 62 3 36 52%.st

No 2 me: 2% 85 :2 $33 52223.9
42 8 NE $2 mod 8: 5%8 20:28me
a: m4 .8 N2: 85v G: 353 soimamsfi
88 m2 am Rm .2 S: €248 csimmameflm
wow- .2. 5 8m 02 E. 3.23% 2.833%.an
one a: S £48 5:536.

v— .Wwiwmgm v. ewhemvev. 8m 93.”.— mm a: a 55.8an

 

 

a .
V K1- ~ o
%N&% m EM 23588 05.3 moanoa 068620 new 62m?“ 05 .8 #5885 m.m 95:.

170

Thermal Transport Properties.

KPb18-,SnbeTezo. Thermal diffusivity measurements for the samples KPb13_

xSnbeTezo (x=5, 9, 13 and 16) were used to calculate their total thermal conductivity

(Figure 5-9a). For all compositions, the total thermal conductivity decreases with

temperature. At similar temperature, the thermal conductivity decreases with Pb content.

For example at 300 K the total thermal conductivity of KPbZSanbTezo is ~ 3.6 W/m-K,

which is three times higher than that of KPb13Sn5SbTe20 (1.1 W/m-K). This trend is

similar to that observed in the case of Na analogs. Samples with high amount in Sn have
high electrical conductivity; as a result, their electronic contribution to the total thermal
conductivity is higher than the lattice contribution. The electronic component of the
sample was estimated using Wiedemann-Franz law and the lattice contribution was
evaluated by subtracting the electronic contribution to the total thermal conductivity. The
results are plotted in Figure 5-9b and 5-9c. For Sn—rich composition such as
KszSanbTezo, the electronic contribution at room temperature (2.4 W/m-K) is slightly
higher than the lattice component (1.9 W/m-K). For Pb-rich composition, the electronic
component at room temperature is low (0.3 W/m-K), three times smaller than the lattice
conductivity (0.9 W/m-K). At room temperature, for all compositions, the lattice
component is lower than that of pure PbTe (2.3 W/m-K).27 At room temperature, the x=5
and x=9 compositions have the lowest lattice thermal conductivity (~ 0.9 W/m-K). At

high temperature, lattice thermal conductivity as low as 0.6 W/m-K was found for

KPngIlngTezo. These values are close to that reported for the Na analogs and for the

K1_bem+5Sb1+yTem+2. The reason of such low lattice thermal conductivity is probably

171

the existence of nanostructures embedded in the Pb1-ySnyTe matrix as was observed in

the case of Nan18_xSnbeTe20 (see Chapter 4).

 

 

 

 

 

 

 

 

 

 

 

 

2
E E x=5
\ . II .. a O
3 35‘ a! * E ( ) A x=9
v v b = v x=13
g 30‘ fl - a 33 x=16
o 2.5« .7 ._.
1:3 ‘ ' f: if]
a m 3
g 2.0-
O . ‘3 V
a 1.5. r,
o f. L
g 1.0' J O O O O O O O 8 e a 8
'- Temperature (K)
5?.
g2.0
3" I v V (b)
3‘
:E1.5- 7 h
8 T
3 W
U E ' a
c b j
8 10‘ W a a 7 f“
— ,h i l’: . H (- g '1
Lo 4) A Z: R @
gOS‘ B L) g
g 300 400 500 600
3 Temperature (K)
9.
g 2.5 g a P
E ‘ l A a E m (C)
_3‘ 2.0' pg 9?”:
.5 , a E
3, 1.5.
c ‘ v w
§ 1m n
m r _ a . v v W
E ‘ Q 2. <7 ‘ <7
.2 0.5fi A u a
0 o O ,
.S 00W 0 _ o r.) o O. O o O 0' Q 0
g 300 400 500 500
'71 Temperature (K)

Figure 5-9. Temperature of the (a) total thermal conductivity, (b) lattice thermal

conductivity and (0) electronic thermal conductivity for KPb18-xSnbeTe20 (x=5, 9, 13

and 16).

172

 

The temperature dependence of the figures of merit (ZT) of the KPb13_xSnbeTe20 (x=5,

9, 13 and 16) calculated from the above data are compared in Figure 5-10. For all
compositions except x=5, the figure of merit increases with temperature. The figure of

merit of the x=5 compound reaches a maximum ~ 0.33 at 422 K. The figure of merit of
KPngngste20 increases more rapidly than that of KPb58n13SbTe20 and
KPbZSnmeTezo with a maximum ZT ~ 0.57 at 572 K. For compositions more rich in Sn,

the highest ZT are shifted to higher temperature with ZT ~ 0.59 at 673 K.

 

0.6- o A A
u A x= A A
_ A =13
0'5' EB :=16 A A
0.4- A E
. o A A
no.3? 00000 ::00 Am
A
0.2j A ego
0.1-l AA ée O
A Q 0
‘ A gé 0
oc- rem Go

 

 

 

360 ' 460 ' 560 ' 600 ' 760
Temperature (K)

Figure 5-10. Temperature dependence of the figure of merit ZT for KPblg-xSnbeTe20

(x=0,3,5,9,13,16)

173

5.4. Conclusions

The series KPb13-xSanTezo (x=0, 2, 5, 9, l3, l6 and 18, M=Sb, Bi) were
prepared via high temperature solid state synthesis. From powder X-ray diffraction

studies, the ingots appeared homogeneous with patterns similar to that of PbTe. However,

SEM and BSE studies indicated the presence of micro-size inclusions. Some of these
inclusions consist of szTe3 precipitates were but the majority of these inclusions contain
more than one phase. The roughness of their surfaces and the presence of potassium and
oxygen peaks are indications that unstable phases containing K precipitate in the Pb]-
xSnxTe matrix upon cooling but decompose upon contact with water. This stability issue

is a serious drawback for industrial application. Partial substitution of Pb by Sn increases

the hole concentration, resulting in a switch from n—to-p type. However the absolute
therrnopower values measured for the two series is still lower to that of the K1-
bem+58b1+7Tem+2. As a result, the highest power factor (~ 13-14 pW/cm-Kz) are

obtained at high temperature for Sn-rich compositions because of their high electrical

conductivity. The low lattice thermal conductivity of KPb13Sn5SbTe20 and
KPngngstezo (0.9 W/m-K at 300 K), similar to the values reported for the K1-

bem+5Sbl+7Tem+2 system, is an indication of the existence of nanostructures inside the

matrix.

174

 

 

References

1 - MRS Bulletin 2006, 31, (3).

2 - J., D. F., Science 1999, 285, 703.
3. Harman, T. C.; Taylor, P. J.; Walsh, M. P.; LaForge, B. 13., Science 2002, 297,
2229.

4. Harman, T. C.; Taylor, P. J.; Walsh, M. P.; LaForge, B. E., J. Electron. Mater.
2005, 34, L19.

5. Venkatasubramanian, R.; Siivola, E.; Colpitts, T.; O'Quinn, 3., Nature 2001, 413,
597.

6. Caylor, J. C.; Coonley, K.; Stuart, J .; Colpitts, T.; Venkatasubramanian, R., Appl.
Phys. Lett. 2005, 87, (2), 023105.

7. Hicks, L. D.; Dresselhaus, G., Phys. Rev. B 1993, 47, 16631.

8. Hicks, L. D.; Dresselhaus, M. 8., Phys. Rev. B 1993, 47, (19), 12727.

9. Kim, W.; Singer, K. L.; Majumdar, A.; Vashaee, D.; Bian, Z.; Shakouri, A.; G.,
2.; Bowers, E. J .; Zide, J. M. 0.; Gossard, C., Appl. Phys. Lett. 2006, 88, 242107.

1 0. Hicks, L. D.; Dresselhaus, M. 8., Phys. Rev. B 1993, 47, 16631.

1 l. Harman, T. C.; Taylor, P. J.; Spears, P. J .; Walsh, M. P., J. Electron. Mater. 2000,
29, L1.

12. Hsu, K. F.; Loo, S.; Guo, F.; Chen, W.; Dyck, J. S.; Uher, C.; Hogan, T.;
Polychroniadis, E. K.; Kanatzidis, M. G., Science 2004, 303, 818.

13. Quarez, E.; Hsu, K. F.; Pcionek, R.; Frangis, N.; Polychroniadis, E. K.;
Kanatzidis, M. G., J. Am. Chem. Soc. 2005, 127, 9177.

175

1 4- Hagelberg, F .; Neeser, S.; Sahoo, N.; Das, P. T.; Weil, K. G., Phys. Rev. A 1994,
5 Q, 557.

l 5. Henger, G.; Peretti, E. A., Journal of the Less-Common Metals 1965, 8, 124.

16. Poudeu, P. F. P.; D'Angelo, J.; Downey, A. D.; Short, J. L.; Hogan, T.;
Kanatzidis, M. G., Angew. Chem, Int. Ed. 2006, 45, 3835.

17. Poudeu, P. F. P.; Kong, H.; Gueguen, A.; Wu, C. 1.; Pcionek, R.; Shi, X.; Uher,
C.; Hogan, T.; Kanatzidis, M. G., manuscript in preparation.

18. Ahmad, S.; Mahanti, S. D., Phys. Rev. B 2006, 74, 155205.

19. Orihashi, M.; Noda, Y.; Chen, L. D.; Goto, T.; Hirai, T., J. Phys. Chem. Solids
2000, 61, 919.

20. Androulakis, J.; Hsu, K. F.; Pcionek, R.; Kong, H. J.; Uher, C.; D'Angelo, J.;
Downey, A. D.; Hogan, T.; Kanatzidis, M. G., Adv. Mater. 2006, 18, 1170.

2 1. Kittel, C., Introduction to Solid State Physics. Wiley: 2005.

2 2. Bouad, N.; Chapon, L.; Marin-Ayral, R. M.; Bouree-Vigneron, F.; Tadenac, J. C.,
-J. Solid State Chem. 2003, 173, 189.

2 3. Rogacheva, E. 1.; Come, G. V.; Laptev, S. A.; Arinkin, A. V.; Vesiene, T. B., J.
Solid State Chem. 1982, 43, 364.

24. Dimmock, J. 0.; Melngailis, 1.; Strauss, A. J., Physical Review Letters 1966, 16,
(26), 1193-6.

25. Abramof, E.; Ferreira, S. 0.; Rappl, P. H. 0.; Closs, H.; Bandeira, I. N., Journal
of Applied Physics 1997, 82, (5), 2405-2410.

26. Han, M.; Hoang, K.; Kong, H. J.; Pcionek, R.; Uher, C.; Paraskevopoulos, M.;
Mahanti, S. D.; Kanatzidis, M. G., Chem. Mater. 2008, 20, 3512.

27. Ioffe, A. F., Can. J. Phys. 1956, 34, 1342.

176

Chapter 6

The systems CuPbmeTem+2 (m=8, 12, 18, 20, 22, 30, 40,
50), CUPb13-xSleSbT620 (x=0, 5, 9, 13, 18), CuzTe/szTeg,

and CuzTe/PbTe

6.1. Introduction
Improved figures of merit have been reported for materials prepared in low
dimensional form such as quantum well (QW) two dimensional multilayered structures“
2, one dimensional quantum wire array structures3 and even lower dimensionality such as
quantum dot arrays and superlattices structures. 4’ 5 A significant reduction in the lattice
thermal conductivity was reported for these systems as a result of the increase in phonon-
boundary scattering. Our group reported on the bulk n-type system AngmeTem+2 (so-
called LAST). 6 A figure of merit ~ 1.7 at 700 K was obtained for the composition
Anglgstezo. The electrical conductivity and the therrnopower were respectively ~ 200
S/cm and — 335 uV/K, resulting in a power factor ~ 28 pW/cm-Kz. In addition to a high

power factor, the thermal conductivity obtained from thermal diffusivity measurements
Was ~ 1.1 W/m-K at 700 K. To understand the reason of such low thermal conductivity,
detailed structural analysis was performed on the samples. Powder X-ray diffraction

indicated the material to be single phase with NaCl structure-type. However the presence

177

0f nanocrystals embedded in the PbTe matrix was observed with high resolution TEM
studies. 6’ 7 Because of instrumental limitation, the precise composition of these

nanocrystals could not be determined. We can suppose these regions to be rich in Sb and
Ag in order to maintain locally the electroneutrality of the system. Using the LAST

system as an exploration platform, other systems derived from this family of compounds

were investigated: the n-type systems Pb1_xSnxTe-PbSS, Pb9,6Sb0,2Te10_xSex9 and the p-
type systems Na1-bemeyTem+2 (SALT)IO and Ag(Pb1_ySny)meTe2+m (LASTT).ll Low
thermal conductivity was reported for the bulk p-type system Nal-bemeyTem+2

resulting in ZT ~ 1.7 at 650 K for Na0,95szoSbTe22.lO High resolution TEM studies

again indicated the presence of nanostructures embedded in the PbTe matrix.

In the two first parts of this chapter, we study the impact of substituting Ag by Cu
in AngmeTem+2 and Ang13_xSnbeTe20 on the structural and electronic properties of
these two systems. Powder X-ray diffraction indicated the materials to be not single

phase but to contain CuzTe and szTe3 as well as PbTe. The results were confirmed by
back-scattered electron imaging. In order to understand why the two phases CuzTe and
Sb2TC3 segregate together, mixtures of Cu, Sb and Te with ratio 1:1;2 were prepared in
part 111. These attempts did not result in the hypothetical compound CquTe2 but in a
mixture of CuzTe and szTe3. Finally, in part IV, some doping studies on PbTe using
CuzTe only resulted in the segregation of CuzTe in the PbTe matrix and provides some

insight about the role of szTe3 in the properties of CuPbmeTem+2 and CuPblg_

178

6.2. CuPbmeTem+2 (m = 8, 12, 18, 20, 22, 30, 40 and 50)

6.2.1 Experimental section

6.2.1.1. Synthesis.
CuPbmeTem+2. All samples were prepared as polycrystalline ingots in silica by mixing

high purity elements in the appropriate stoichiometric ratio in 10 mm outer diameter
fused silica tubes. The fused silica tubes were carbon-coated prior to use as glass attack

was observed during preliminary reactions. The tubes were then flame-sealed under a
residual pressure of ~10—4 Torr, placed into a tube furnace (mounted on a rocking table)

and heated at 1373 K for 4 h to allow complete melting of all components. While molten
the furnace was allowed to rock for 2 h to facilitate complete mixing and homogeneity of
the liquid phase. The furnace was finally immobilized at the vertical position and was
cooled from 1373 to 820 K over 55 h followed by a fast cooling to room temperature.
"The resulting ingots were silvery-metallic in color with a smooth surface. Table 6-1

summarizes the amount of elements used for each reaction.

179

Table 6-1. Amounts of elements used to prepare the series CuPbmeTem+2.

 

 

 

 

 

 

 

 

 

 

 

 

 

L Composition 8:31) 52:01) (81:30]) (2:501)
l cupbgste‘O 33.3235; g (651.7213?)g 397:; g 213.2%?
. are are 2:23.” a???
, 3. 2’5” $.- 19.33,), 23:2.” 239$
2.- ,0... as 2.- 29:.” 22le
2.92;.” 5.27:3; 2. 28:.” a???
CuPbsoSbTesz $32737? g (73.2%, $533? g €33,535),
CuPb4OSbTe4z $0852; g 25295931; 3.213242 g (3289?
CuprOSbTeSZ €69;st g 82?? 23:35 g 335;?

 

 

 

6.2.1.2. Characterization techniques

 

Powder X-ray Diffraction. Powder X-ray patterns of the grinded materials were
recorded using Cu K“ radiation (A. = 1.54056 A) in reflection geometry on a CPS-120 Inel
J-ray powder diffractometer operating at 40 kV and 20 mA equipped with a position
sensitive detector.

DTA Analysis. Differential thermal analysis (DTA) data were collected with a Shimadzu

DTA-50 thermal analyzer. Approximately 35 mg of finely ground powder of material

was sealed in a carbon-coated quartz ampoule under residual pressure of ~ 10‘4 torr.

Another ampoule containing similar amount of alumina and prepared the same way was

used as a reference. The samples were heated to 1273 K at a rate of 10 K/min, held at

1273 K for two minutes and cooled down to 323 K at a rate of -10 K/min.

180

Scanning Electron Microscopy. The surface of several samples was polished very
carefully using silica suspension solution (0.05 um) in order to get a smooth, mirror-like
surface. The samples were then studied with a scanning electron microscopy (Hitachi
S3400N-II) with 25 kV acceleration voltage using both energy-dispersive spectroscopy
(EDS) and back-scattered electron imaging (BSE).

Electrical Transport Properties. Thermopower and electrical conductivity properties
were measured simultaneously under helium atmosphere using a ZEM—3 Seebeck

coefficient/electrical resistivity measurement system (ULVAC-RIKO, Japan). Samples
for transport measurement were cut to size 10 x 3 x 3 mm3 using a diamond saw (Buehler
isomet 1000), a wire saw (South Bay Technology) and a polishing machine (Buehler
ecomet 3000). Rectangular shape samples with approximately 3 x 3 mm2 cross-section

were sandwiched vertically by two nickel electrodes (current injection) with two Pt/PtRh
thermocouples (for temperature difference and voltage measurements) attached on one
side. The sample and measurement probes were covered by a nickel can to maintain a
constant temperature during the measurement and the base temperature was measured by
a thermocouple attached to the outside of the can. The sample, electrodes, and nickel can
were placed in a vacuum chamber then evacuated and refilled with He gas (0.1 atm) to
provide necessary heat transfer. Properties were measured from room temperature to 670
K under helium atmosphere.

Hall Measurements. Above 300 K, Hall measurements were carried out by an in-house
high temperature/high magnetic field Hall apparatus. It consists of a nine Tesla air-bore
superconducting magnet with a water-cooled oven inside the bore of the magnet, and a

Linear Research AC bridge with 16 Hz excitation. Four-wire AC Hall measurements

181

were performed on parallelepiped samples with the typical size of 1.5 x 3 x 10 mm3 to

temperatures of at least 800K with the protection of Argon gas.

Thermal Conductivity. The thermal conductivity was determined as a fimction of
temperature using the flash diffusivity method on a LF A 457/2/G Microflash NETZSCH.
The front face of a small disc-shaped sample (diameter ~ 8 mm; thickness ~ 2mm) coated
with a thin layer of graphite is irradiated with a short laser burst, and the resulting rear
face temperature rise is recorded and analyzed. The experiments were carried out under

nitrogen atmosphere. Thermal conductivity values were calculated using the equation K:
a CI) d, where a is the thermal diffusivity, C1) the specific heat and d the bulk density of

the material calculated from the sample’s geometry and mass. A pyroceram reference
was used to determine the heat capacity of the sample. The thermal diffusivities were
measured typically over the temperature range 300-670 K. The electronic component of

the thermal conductivity was quantified through the Wiedemann—Franz law according to
which Kg, = o.T.L0 (Lo being the Lorenz number, Lo=2.45.10'8 WSIK'I).12 The lattice

contribution was then derived by subtracting the electronic component from the total

thermal conductivity.

6.2.2 Results and discussion

Structure and Characterization. Figure 6-1 shows the X-ray diffraction patterns from
powders taken from top and bottom of the ingot for the compositions CuPbmeTem+2

(m=5, 12, 18, 20, 22, 30, 40 and 50). Extra peaks which do not belong to the cubic

structure PbTe are clearly visible. The two small peaks at 28 and 38 deg could be

identified to the phase Sb2TC3 and the peaks present ~ 42-43 to CuzTe. The presence of

182

such phases can be expected due to the low solubility of Sbl3 and Cu in PbTe. As the m

value increases, the intensity of the peaks belonging to these minor phases decreases.

Scanning Electron Microscopy. To obtain precise information about the repartition of
the CuzTe and szTe3 phases in the PbTe matrix, detailed back-scattered electron (BSE)
analysis on the fine-polished surface of samples was carried out. Figure 6-2 shows a

typical BSE-image observed in the case of CuPblngTezo. Regions rich in CuzTe and
szTe3 are randomly dispersed in the PbTe matrix. These features are observed for all
CuPbmeTem+2 compounds with more dense networks observed for low m values. For
example, BSE imaging on CuPb40SbTe42 (Figure 6-3) shows smaller regions rich in
CuzTe-szTe3 embedded in the PbTe matrix. It can be noticed that these CuzTe and
szTe3 always segregate in the same regions forming a microcomposite network inside
the matrix. This is because according to the CuzTe-szTe3 phase diagram there is a

eutectic composition very close to the 50:50 atomic percent ratio.14

183

..s
O
O

 

 

(I)
O

 

 

 

.5
O

 

'\
Q

N
0

Relative intensity
8

 

 

l—

- fi- - u
30 35 40 45 50
2theta (deg)

 

 

_x
O
C)
\
N
01

(D
0

Relative lntenslty
A O)
O O

m=1

N
O

 

20 40 60 80 100120
2theta (deg)

Figure 6-1. Powder X-ray diffraction of the series CuPbmeTem+2 (m=5, 12, 18, 20, 22,

30, 40 and 50), the area between 29 = 25 and 50 deg is enlarged to show the presence of

extra peaks that do not belong to the NaCl structure-type. Two small peaks at 28 and 38

deg are characteristic of Sb2TC3. Small peaks ~ 42-43 deg belong to CuzTe.

184

 

Figure 6-2. BSE images of a sample with composition CuPblngTezo showing regions

rich in CuzTe and szTe3 embedded in the PbTe matrix. The two phases form dentritic

ribbons as long as 3 mm in all directions.

Thermal Analysis. Figure 6-4a and 6-4b shows the DTA results for CuPbgsbelo. A
single melting point ~ 1185 K and a single crystallization point ~ 1165 K were observed
upon heating and cooling. A single melting point ~ 1193 K and a single crystallization
point ~ 1155 K were also visible in the case of CPb30SbTe32 (Figure 6-4c and 6-4d). For
the composition m=50, the spectrum showed a single melting point ~ 1198 K upon

cooling but two crystallization points were observed during cooling at ~ 1165 and 1 139 K

185

during the first cycle (Figure 6-4e). During the second cycle, two melting points at 1194
and 1204 K are visible upon heating and three crystallization points ~ 1104, 1119 and

1149 K upon cooling (Figure 6-41).

V V V t I I | | 1
v '35 010." 1C 3mm X75 ESECCH‘Jl: 512.":‘3'35 17 5‘3 SCOUH‘

l l I I l I I
"J 10 5mm .6130 BSECCPIP 9» 132308 17 3'? 5C Slum

 

Figure 6-3. BSE images of the compound CuPb4oSbTe42. More dispersed regions

composed of CuzTe and szTe3 are visible in the PbTe matrix.

186

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-20.: (a) of (b) A
401 -25':
A -50: -50:
21,. -80: a -75-
<-1oo- V -1oo-'
,_ , < .
0.120: 5 425.
'140‘ 1st cycle '150:
- .‘_ '—2nd cycle
-123 , CuPngbTem - fi :33: CuPDBSbTem
400 600 800 1000 1200 400 ' 600 - 800 T1300 12'00
Temperature (K) Temperature (K)
20 40
J (c) . (d)
01 20-
, .
0..
s '20“ s
5 ‘ 3 -2o
:5 40' :5
D -60‘ D 40':
‘—1st cycle K -50. —2nd cycle
'80' CuPb3oSbTe32 80‘ CuPb308bTe32
400 ' 600 800 10% 12'00 400 ' E30 800 1000 1200
Temperature (K) Temperature (K)
0. (e) 0- (D
-20. -20
5 . A
a .40. :40-
< . <
[_-
D -60- '5 -60-
' ——1st cycle ‘ —2nd cycle \
-80d CuPb5oSbTe52 -30- CuPb5oSbTe52
400 T660 ' 800 '10'00'12'00 400 ' 600 800 1000 1200
Temperature (K) Temperature (K)

Figure 6-4. DTA results: (a) first cycle and (b) second cycle for CuPngbTelo; (c) first
and ((1) second cycle of CuPb30SbTe32; (e) first cycle and (f) second cycle of

CuPb5OSbTe52.

187

Electronic Transport Properties. Figure 6-5a shows the temperature dependence of the
electrical conductivity for the series of compounds CuPbmeTem+2. For all compositions,

the electrical conductivity decreases with increasing temperature. Except for m=50, other
compositions exhibit similar electrical conductivity values for the range 300-725 K.

Room temperature values are typically in the range 1600-1800 S/cm and decrease to ~
400-500 S/cm. CuPb5OSbTe52 shows higher electrical conductivity at room temperature (

~ 2680 S/cm) but at higher temperature no significant difference with the other
compositions was observed.

The trends in therrnopower are reported in Figure 6-5b. All compositions showed
n-type behavior. No significant trend as a function of m can be determined. Values vary
between -70 and — 100 uV/K at room temperature and ~ -150 and -205 uV/K at 675 K.

The power factor calculated from the electrical conductivity and thermopower are plotted

in Figure 6-50. In the range 300 — 500 K, the material CuPblgstezo exhibits the highest
power factor with a maximum value of ~ 23 uW/m-K2 at 345 K. At higher temperatures,
CuPb5OSbTe52 has a higher power factor.

Hall coefficient measurements were carried on the samples CuPblngTezo and

Cuszzstez4. The resulting carrier concentrations are reported on Figure 5d. For both

compositions, the Hall coefficients are negative, indicated electrons as the major charge

carriers. This confirms the negative sign observed for the therrnopower. Room

temperature carrier concentrations are respectively 1.6 and 1.7 10'9 /cm3. These values

are lower than that measured for Nag-95PbZOSbTe22 (~ 6 1019 /cm3).

188

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-75- s; (b)
§ -100- o_i‘f*73-im
3‘ ‘0; NT} \ 7*?1K,
5..) '125" \;§:?: :1? \:.-\
'x\ ‘ ‘V‘r";,i\ l
E -175- ”'5 of: \31
g 25 ‘3?»
:5 FE -200‘ (i
2 250 T l l l l fir ' ' ' '
LIJ 300 400 500 600 700 300 400 500 600 700
Temperature (K) Temperature (K)
«g 3.0 ,

N‘ 24‘ AA/A, (C) a? 2.3 (d) . -
¥ 22: \A‘ ‘ 32.6 - CuPb1aSbTe20 . .
E 20‘ o feta g. . E2 4 . CuPb22$bTe24 -

.516- 3 .4 ""9: E22 '
o 14- Y4: 4 :45" ‘ E20 ' 0
£12— “ ' v VT {3‘14 §18 I n .
5 al- ,: ’aI-.ru«;:i ’ _ ° 8 - . - .
310- -‘ 1,4 "I .
O I Ch) 1.6 I ‘
a o l r I r r -E I I l I
300 400 500 600 700 8 300 400 500 600 700
Temperature (K) Temperature (K)

Figure 6-5. Temperature dependence of (a) the electrical conductivity, (b) the
therrnopower and (c) the power factor for the series CuPbmeTem+2 (m=8, 12, 18, 20, 22,
30, 40, 50); (d) carrier concentration as a function of temperature for CuPblgstego and

CUPbZQSbTCM.

Thermal Transport Properties. Thermal diffusivity measurements were collected for
the compounds m= 8, 18, 20, 30 and 40. The thermal conductivity values calculated from
thermal diffusivity data are reported in Figure 6-6a. For all samples, the thermal

conductivity decreases with temperature. The composition CuPngbTelo exhibits higher

189

thermal conductivity than the other materials, with a value of ~ 3.2 W/mK at room
temperature and decreasing to ~ 2.2 W/m-K at 675 K. The compositions with m=18, 20,
40 have similar thermal conductivity values, starting ~ 2.5 - 2.2 W/m-K at room
temperature and falling down to ~ 1.4 — 1.6 W/m-K at 675K. The sample CuPbgoSbTe32
has slightly higher room temperature thermal conductivity but at high temperature the

thermal conductivity is similar to that of the compounds with m=18, 30 and 40.

 

 

 

 

2
E 3.2
s,ao '- (a)
328 ' "..
g 2.6 ‘ 'v I I

2.4 ’ ‘ 4 ' I I
§ 2.2 ‘1 ii I I
8 20 I m= 1’ ’ I
To 1'8 o m=18 xixg
g 1.6 . m=20 in
ac) 1_4 v m=30 H
: 1.2 ‘ =4'0 l l l I
g 300 400 500 600 700
I— Temperature (K)

 

 

 

 

360 460 560 600 760
Temperature (K)

9.

§20 I

S,18j -- (b)

:g1e- . --__

31.4” V I... .I
D . ‘ v

212.. ‘1‘ vv I
§1‘0:- m=§‘z:"v
203-omfi8 itlil.

L -A =20

EOE-v 2:30 ‘li‘:
80 '4 m=40

g

_I

Figure 6-6. Temperature dependence of the (a) total and (b) lattice thermal conductivity

for the compounds CuPbmeTem+2 (m=8, 18, 20, 30, 40).

190

The electronic contribution was estimated using the Wiedemann—Franz law and the lattice
thermal conductivity was quantified by substracting the electronic component to the total

thermal conductivity. The resulting lattice components are plotted in Figure 6-6b. The

highest lattice contribution was found for CuPngbTelo with values varying between 1.9

W/m-K at room temperature and 1.4 W/m-K at 675 K. The composition CuPb3oSbTe32
showed the sharpest decrease in lattice thermal conductivity with values at 675 K as low

as those of CuPblngTezo, CuPbZOSbTezz and CuPb40SbTe42. For these three

compositions, the lattice thermal conductivity varies between ~ 1.2 W/m-K at room
temperature and ~ 0.6-0.7 W/m-K at 675K. The lattice thermal contribution for these
compositions at 675 K is slightly higher than the value reported for Na0.95Pb20SbTe22
(miminum lattice thermal conductivity ~ 0.55 W/m'K at 675 K).

From the above data, the figure of merit ZT was evaluated for the compositions

CuPbmeTem+2 (m=8, 18, 20, 30 and 40) (Figure 6-7).

 

0.9.: I m=8 '

0.8“ o m=18 V
. v 4

Q74 A m=20 "40

0.61 v m=30 43"“
- < =40 . 'n

0.5- 3

:1 ‘ o..' A“

8.31 0°:':‘A‘ 'I'

.1 °1:‘A III.-

o.21 “

0.1' I.-

 

 

 

360 ' 460 ' 560 T660 ' 760
Temperature (K)

Figure 6-7. Figure of merit ZT for the compounds CuPbmeTem+2 (m= 8, 18, 20, 30, 40).

191

As expected the lowest figure of merit is found for CuPb3SbTe10. For the other

compositions, the figure of merit increases more rapidly with temperature. The maximum

ZT ~ 0.85 is achieved at 675 K for CuPb30SbTe32.

6.3. CuPb13-xSnbeTe20 (x=0, 5, 9, 13, 18)

6.3.1. Experimental section

Synthesis. The samples were prepared following the same procedure as that described

above for the CuPbmeTem+2 series. Table 6-2 summarizes the amount of material used

for each composition.

Table 6—2. Amount of elements used for to prepare the series CuPb13_xSnbeTe20.

 

 

 

 

 

 

 

 

 

 

 

 

Composition Cu, g Pb, g Sn, g Sb, g Te, g
(mmol) (mmol) (mmol) (mmol) (mmol)
CuPblgstezo 0.1046 g 6.1391 g 0 0.2004 g 4.2007 g
(1.65) (29.63) (1.65) (32.92)
CuPb13Sn5SbTe20 0.1246 g 5.2816 g 1.1638 g 0.2387 g 5.0039 g
(1.96) (25.49) (9.80) (1.96) (39.22)
CUPngDngTezo 0.1223 g 3.5890 g 2.0562 g 0.2343 g 4.9116 g
(1.92) (17.32) (17.32) (1.92) (38.49)
CuPb5Sn13SbTe20 0.1323 g 2.1569 g 3.2129 g 0.2535 g 5.3132 g
(2.08) (10.41) (27.07) (2.08) (41.64)
Cusnisstezo 0.1781 g 0 5.9887 g 0.3413 g 7.1525 g
(2.80) (50.45) (2.80) (56.05)

 

The synthesized compounds were characterized by powder X-ray diffraction, scanning

electron microscopy. Electronic transport properties and thermal diffusivity data of the

samples were also measured.

192

 

6.3.2. Results and discussion

Structure and Characterization. Figure 6-8 shows the powder X-ray diffraction

patterns of (a) CuPblngTezo, (b) CuPbgsngstezo and (c) CuSmngTezo. It can be
observed that the peaks corresponding to szTe3 get weaker as the content of Sn

increases. The peaks corresponding to CuzTe are present for all compositions.

Scanning Electron Microscopy. In order to have complementary information about the

presence of other phases in the PbTe matrix, BSE imaging was performed on different
samples. Figure 6-9 shows two pictures obtained for the sample CuPblgsnsstezo.

Regions with different compositions are visible in the PbTe matrix. EDS analysis on area

1 gave the results: Cu: 2.5 %, Pb: 6.20 %, Sb 28.40% and Te: 62.90%. EDS on area 2

indicated Cu: 63.42% and Te 36.88 %. As we saw for CuPbmeTem+2, phases rich in
CuzTe and szTe3 precipitate in the PbTe matrix during the cooling process. In the case
of CuPbl3Sn5$bTe20, the regions containing CuzTe and szTe3 form more complex
composite areas. Similar features were observed for CuPbgsngstezo (Figure 6-10a) and

CuPb58n13SbTe20 (Figure 6-10b). BSE imaging on the compound CuSmngTezo is
different from what was obtained for the compositions discussed above (Figure 6-11).

Only regions with compositions CuzTe embedded in the SnTe matrix were visible. No

szTe3 phase was observed in the SnTe matrix, confirming the powder X-ray data. Sb

and Sn have close radii; as a consequence, Sb atoms can easily substitute Sn atoms.

193

60

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

<a> 13a
—— om
g. 50- CuPb1BSbTe20
U)
93 40-
.E
g 30-
E 2)
3:, 20_ Li
MU
10 l T '7 l I I
20 40 60 80 100
2theta (deg)
60
(b) —<1)top
-—(2) bottom
g 50‘ CuPbQSnQSbTezo
U)
E 40-
E
jg 30- l
L“ (2)
g 205;“ ‘ ‘ -
MLJLLU
10 l ' I I I I
20 40 60 80 100
2theta (deg)
60 (1)1
— o
' (C) -——-(2)b:ttom
50 - CuSn188bTezo

 

Relative intensity

 

 

 

 

 

 

 

2'0 ' 4'0 ' 6'0 ' 8'0 '160
2theta (deg)

Figure 6—8. Powder X-ray diffraction for (a) CuPblngTezo, (b) CuPbgsngste20 and (c)

CuSnlgstezo.

194

(a) a“

  
     
     
 

 

‘.
Area 2:
Cu 63.42 at. %,
Te 36.88

 

 

 

Cu 2.5 at. %, Pb 6.20,
Sb 28.30, Te 62.90

8.5400 25.0KV 10.3mm ' ,LCOMP 8 ,» ' 50.011111

 

Figure 6-9. BSE images of CuPb13Sn5SbTezo. Complex regions rich in CuzTe and

szTe3 are clearly visible in the Pb1_xSnxTe matrix.

195

Cu 2.74 at. %, Pb
4.16, Sb 26.77, Te

Cu: 62.53, Te: 37.47 at. %

S3400 25.0kV10.4mm x350 BSECOMP 8/10/20081'7:42' ' ' ' ' io'ou'm'

 

S3400 25.0kV 10.3mm x320 BSECOMP 8/10/2008 17'45 ' ' ' iobu'm'

Figure 6-10. BSE images of (a) CuPbgsngste20 and (b) CuPb5Sn13SnTezo.

196

Cu: 62.44, Te: 54.73 at. %

$3400 25.0w 10.3mm x650 BSECOMP 8/10/20081823 ' ' '

$3400 25.0w 10.2mm x1.40k BSECOlVlP end/2008 [18:34 I ' 14000.13

 

Figure 6-11. BSE images obtained for CuSnlngTezo. Only CuzTe regions segregated in

the SnTe matrix.

197

Electronic Transport Properties. Figure 6-12a shows the temperature dependence of
the electrical conductivity for the series CuPb13_xSnbeTezo (x=0, 5, 9, 13, 18). The
electrical conductivity of CuPb13Sn5SbT020 is low (below 100 S/cm) over the

temperature range 300-675 K. For the other compositions, the electrical conductivity
decreases with increasing temperature. For the composition x= 9, 13 and 18, the electrical

conductivity increases with decreasing Pb:Sn ratio. The room temperature electrical

conductivity of CuSnlngTezo is 4100 s/cm, which is three times the value measured for
CuPbgsngste20 (~ 1150 S/cm). This trend is similar to what was observed for the series
Nan18_xSnbeTe20. This dramatic increase in the electrical conductivity with addition to
Sn is consistent with the high electrical conductivity of SnTe, which drastically increases
the carrier concentration. The behavior of the electrical conductivity of CuPblgstezo is
similar to that of CuPbSSn13SbTe20.

The sign of the therrnopower is negative for CuPblgstezo and CuPb13Sn5SbTe20

and positive for CuPbgsnngTezo, CuPb58n13SbTe20 and CuSnlgstezo (Figure 6-12b).

The therrnopower is almost equal to zero for CuPb13Sn5SbTezo, indicating that for

compositions close to that particular Pb:Sn ratio the therrnopower switches from n- to p-
type. In other words, for compositions x < 5, electrons are the main charge carriers. For
x=5, there is almost the same number of positive and negative charge carriers. As a result,

the therrnopower for that particular composition is almost zero. For n-type behavior, a

maximum therrnopower of ~ - 225 uV/K was obtained for CuPb13Sn58bTe20 at 475 K.

198

For p-type samples, the highest therrnopower was found for CuPbgsngste20 (~ 200

uV/K at 550 K).

The power factors obtained from the above data are plotted in Figure 6-12c. In the

temperature range 300-625 K, the sample CuPblSSbTezo exhibits the highest power

factors with room temperature values ~ 23 uW/cm-KZ. For the compositions x=0 and

x=9, the power factor reaches a maximum and then slowly goes down whereas for x=13
and x=18, the power factor starts from very low values at room temperature and

constantly increases till 675 K. At 75 K, the compounds x=0 and x=18 have similar

power factor, ~ 14-15 uW/cm-KZ. Such high values derive from a relatively high
therrnopower for CuPblgstezo and a relatively high electrical conductivity for

CuSnlgstezo.

Thermal Transport Properties. Thermal diffusivity measurements were collected for
CuSnlngTezo as that composition exhibited high power factor at high temperature.
Figure 6—13 shows the total, electronic and lattice thermal conductivity of the sample.
The room temperature total thermal conductivity is high compared to CuPblngTezo, the

main reason being the high electrical conductivity of the sample. As a result, the
electronic component of the thermal conductivity is high (~ 3 W/m-K). From 300-675 K,
the electronic contribution dominates the thermal conductivity. The lattice thermal
conductivity ranges from 3 W/m-K at 300 K to ~ 1.2 W/m-K at 675 K. As a result, the

figure of merit ZT does not exceed 0.3 at 675 K.

199

A4500

 

 

 

 

 

 

 

 

 

 

 

 

E4000“, : :g
(1)/3500' { A x=9
Q3000- 4 ' Xi”
34500- < ‘ “8
ézooo- 'v ‘ ,
V
813(68- ‘AIII': r ‘ ‘
8 500- “ "' "~-
‘3 “Huh“ A IV
(D e. n IQ O I "0 ‘0
E 300 400 500 600 700
Temperature (K)
2001(b) ‘A‘AA“AA"
2 150] ‘A v V
S 100' ‘A‘A v v ‘ ‘
3 50' “ I {V 4 4 4
8 '50: .
O 400" I.
€450. Illlll..l-.-
: . '
E -200: . O , o
-250 . . ’ . .
300 400 500 600 700
Temperature (K)
25
A (C) I...
“>220—
e
§15- .:
E v I '
810' ‘AAAAI ‘
.59 ‘ 4 ‘A
5 5‘ ‘4 I :V 0 ’ .““A.
6 o. I: '
a I I I
300 400 500 600 700

Temperature (K)

Figure 6-12. Temperature dependence of the (a) electrical conductivity, (b) therrnopower

and (0) power factor of the series CuPb18-xSnbeTe20 (x=0, 5, 9, 13, 18).

200

 

 

 

 

fgg. I I I total

g 5'5; ' ' . 0 electronic
S 5'0: - I . A lattice
€4.56 II I

-- 4.0- I

‘3 3.5-j ' ' .
23.0' ..x::..00...

§ 33: “A .'o
21.5: A “A
E1IO' I '7‘. ' I f t‘t‘
l— 300 400 500 600 700

Temperature (K)

Figure 6-13. Total, electronic and lattice thermal conductivities of CuSnlngTezo.

6.4. Investigation of the system Cu/Sb/Te.

6.4.]. Motivation

BSE studies on the compounds CuPbmeTem+2 and CuPb13-xSnbeTe20 showed
the presence of CuzTe and szTe3 as second phases inside the cubic matrix. No ternary
phase such as CquTez was observed. In the system Ag/Sb/Te, AngTe2 is a compound
known since 1957.15 AngT02 crystallizes in the NaCl structure with similar lattice
constant that PbTe. 16 Rosi and a1. recognized the potential of AngTez as a
thermoelectric material. 1 7

It has a large therrnopower ~ 200 uV/Klg’ ‘9

and low lattice thermal conductivity
(~ 0.6 W/m-K).19’ 20 Experiments indicated that the Ag and Sb atoms in the structure may

not be randomly occupying the Na-sites of the NaCl lattice.7 Recent electronic structure

201

calculations showed that the lowest energy structure is one where Ag and Sb atoms are
ordered.21 This could be the explanation of the low lattice thermal conductivity reported
for the material. In the Cu system, contradictory reports on the possible existence of the
phase CquTez can be found in the litterature.22'24 To check on the existence of such a

phase, several reactions were prepared using different cooling profiles.

6.4.2. Experimental section
Synthesis. Mixtures of elemental Cu, Sb and Te with ratio 1:1 :2 were loaded in three 10
mm outer diameter, carbon-coated silica tubes in ambient atmosphere. The tubes were
heated up to 1373 K and kept at that temperature 6 hours. Three different cooling profiles
were studies. In the first case, the tube was quenched in water; the second case, the tube
was quenched in air. The last tube was slow cooled to 550 °C in 55 hours followed by a
fast cooling to 323 K. The fumaces were kept in vertical position during the reaction.
Another tube was prepared using similar slow cooling profile but during the time spent at
1373 K, the furnace was rocked during 2 hours to ensure good homogeneity of the liquid
phase.
Powder X—ray Diffraction. Same procedure as described in part 2.
Scanning Electron Microscopy. Same procedure as described in part 2.

Electrical Transport Properties. Same procedure as described in part 2.

6.4.3. Results and discussion

Structure and Characterization. The samples had good mechanical properties and were

cut into halves with a diamond saw. The presence of two phases could be observed by

202

eye on the cut surfaces. Figure 6-14 shows the X-ray powder diffraction patterns from the
samples prepared without rocking the furnace during the synthesis. Analysis of the
patterns indicated the materials to be a mixture of CuzTe and szTe3. The same result

was observed for the sample prepared by rocking the fumace.

 

    

 

80 , —(1)quenching in air -
70 _ —— (2) quenching in water
3‘ . (3) slow cooling
'7) 60-
0C) 4
.E 50'.
.‘é’ 40'.
CU -
T) 30 .
m 20-
10

 

 

 

0 '2'0 ' 4'0 ' 6'0 ' 8'0 '100
2theta (deg)

Figure 6-14. Powder X-ray diffraction patterns for samples with nominal composition

CquT02 prepared by water-quenching, air-quenching and slow cooling.

Scanning Electron Microscopy. The BSE study of the samples confirmed the presence
of CuzTe and szTe3. BSE imaging from the sample quenched in water are shown in
Figure 6-15. Sb2TC3 strips (clear areas) are embedded in a CuzTe matrix (darker area).
For the sample cooled in air (Figure 6—16), the features of szTe3 regions are less regular
and more complex compared to the case of quenching in water. In the case of slow

cooling (Figure 6-17), szTe3 strips run in all directions. Figure 18 shows BSE images of

203

the samples for which the fumace was rocked during the reaction. CuzTe regions (dark

areas, oriented in different directions) are embedded in the Sb2T€3 (clear areas).

SbgTe3

$3400 25.0kV11.1mm x140 BSECOMPl8/1I/20IO8 '1906 ' ' ' '400'um'

 

83400 25 OkV 11.2mm x210 BSECOMP 8/1/2008 1865' ' ' '200Clm'

Figure 6-15. BSE images of the sample with nominal composition CquTez quenched in

water.

204

'500'un'1

S3400 25.0kV10.8mm x800 BSECOMP 8/1/2008’101'4 '

 

83400 250101 10.7mm x1.60k BSECOMP 8/i/2boé 101's ' ' 30.0001
Figure 6-16. BSE images of the sample with nominal composition CquTeg quenched in

air.

205

33400 25.0w 10.7mm x750 BSECOMP 811/2008 i'8:'50' I50.0lunli

S3400 25.0kV 10.3mm x1.30k BSECOMP '8/1720'08 1182.39' '

 

Figure 6-17. BSE images of the sample with nominal composition CquTe2 prepared by

slow cooling.

206

  
  

       
 

I 0).:

r' o .-

"""If§ID"//

il'a
1H.

-.}o*;'cv v:.-
on... '20:..." ’8 .
433".':"': v.0
0 ga'o'o'n
0 . '5'.
‘ o f e '
.g : .‘l.
o a 0' ‘1
Q 00%..
C O. .3
a I 00
1: t I
C
‘ o
I o‘
C
‘ z
I
C
. .

." .
04?:
0.0
“y":
..g
‘0
.0

I.
o

o

0:00
0.:
2

I
O

C

o

I

I

9

Figure 6-18. BSE imaging of a sample prepared with nominal composition CquTez

with the furnace being rocked during the reaction.

207

 

DTA Analysis. An amount of 35 mg of a sample prepared with a rocking step during the
synthesis was analyzed. The results are shown in Figure 6-19. Similar results were
observed during the two cycles. One major melting point was observed at~ 785 K. A
small melting peak ~ 806 K is hardly distinguishable from the background. About 623 K,
two other small peaks are hardly visible. Upon cooling, crystallization points close to the

melting peaks are visible.

 

 

673K -_755K (a) 20. 61915 626 2.. 767K (b)
,.

1794K

      
 

  

 

 

 

 

 

 

0 _ r 795 K O _
5:: _20_ >2; -20- ear/k 616 K
ii: .40 - 785K g 40 d
O _60 _ O ‘60 "
-80 .- ‘80 '
— 1st cycle _ _ —2nd cycle
-1 00 I I I fit 1 00 I I l I
400 500 800 - 1000 1200 400 600 800 . 1000 1200
Temperature (K) Temperature (K)

Figure 6-19. DTA analysis of the sample prepared by mixing Cu, Sb and Te with the

ratio 121:] :2 by rocking the furnace: (a) first cycle, (b) 2nd cycle.

6.5. Doping studies of PbTe with CuzTe

6.5.1. Motivation

To better understand the role of Sb in the system CuPbmeTem+2, doping studies of PbTe

with CuzTe were performed.

208

6.5.2. Experimental section

Synthesis. Direct combinations of elemental Cu, Pb and Te were mixed with appropriate
ratios in order to dope PbTe with 1, 2 and 5 % CuzTe. The tubes (10mm outer diameter,

carbon-coated) were sealed under residual pressure. Then they were heated up to 1373 K
and kept at that temperature for four hours. While molten, the furnace was rocked for 2
hours to facilitate complete mixing and homogeneity of the liquid phase. The furnace was
finally immobilized in the vertical position and cooled down to 820 K in 55 hours
followed by a fast cooling to room temperature.

Powder X—ray Diffraction. Same procedure as described in part 1.

Scanning Electron Microscopy. Same procedure as described in part 1.

Electrical transport properties. Same procedure as described in part 1.

6.5.3. Results and discussion
Structure and Characterization. Additional peaks than those characteristic of the PbTe
phase are clearly visible in the powder X-ray diffraction patterns for the 5% doping
sample (Figure 6-20). For the 1 and 2 % doping compositions, these peaks are hardly
distinguishable from the background.
Scanning Electron Microscopyln order examine the composition of these materials,

BSE imaging on polished samples were performed. Figure 6-21 shows results for the 5%

doping sample. Darker areas correspond to the CuzTe phase. As the doping level

decreases, the CuzTe phases get more dispersed and are hardly visible in the case of l %

, doping composition (Figure 6-22b).

209

I

 

 

 

 

 

‘ *CuzTe
3. 50-
'Z’ '
,9 404 i
C .
'5 ' 0:1. LLJ LL5%
.2 30-
a - l
g 20-

 

 

 

 

 

A;

 

 

\lJQLil Li DUCT 1%
20 40 60 80 100 120
2theta (deg)

 

_x
0

Figure 6-20. Powder X-ray diffraction of PbTe doped with 1, 2 and 5 % CuzTe. The
small stars indicate diffraction peaks that do not belong to the PbTe structure-type and

indicate the presence of CuzTe as a minor phase.

210

 

$3400 25.0w 11.0mm x1.00k BSECOMP 8'1’8/20081'6335 ' ' ' 5.0.0011)

 

S3400 25.0w 10.7mm x80 BSECOMP 8/8/2008 1'6;'41' ' '5001'1m'

Figure 6-21. BSE images of PbTe doped with 5 % CuzTe.

211

 

S3400 25.0kV 11.5mm x210 BSECOMP 8/8/2008 16:58' ' '200i1m'

 

'2000m'

S3400 25.0kV 11.4mm x200 BSECOMP 8/8/2008l17IL1'2 I

Figure 6-22. BSE images of PbTe samples doped with (a) 2 % and (b) l % CuzTe.

212

 

DTA Analysis. Figure 6-23a shows the result obtained during the first run for PbTe
doped with 1% CuzTe. One single melting point and one single crystallization point are
visible at 1208 and 1170 K. However during the second run, two melting points and two
crystallization points were observed (Figure 6-23b). For the sample doped with 2 %
CuzTe, one melting point and two crystallization peaks are present (Figure 6-23c). Three
melting points and three crystallization points are visible during the second run. The
melting point of PbTe is known to be ~ 1196 K. CuzTe is known to exist under different

structures depending on the temperature.

 

 

    

 

 

 

 

 

 

 

 

  

   

 

     

 

 

 

20 I (a) 20 ' (b) 1166 K 7.-
o. 0_ rev»: :11“
S. -20 - S -20 -
3' -40 - 3 .40 _
<
E -60 - 15 -50 .
-80 - -80 .
— 1st cycle 7 2nd cycle
'100 ' PbTe doped 1 '36 CuzTe 1208 -100 - PbTe doped 1 % CuzTe
400 600 800 1000 1200 400 600 800 1000 1200
Temperature (K) Temperature (K)
20 '
(d)
0 - .— 1 185 K
-20 .
>3. -4o-
1‘5 -60 -
c1
-80 - d
—- 1SI e _ _.-
-100 . PbTe docpyed 2 as CuzTe 1196 K/ 100 W333; % CuzTe1196 K 1200 K
400 500 800 1000 1200 900 1000 1100 1200 1300
Temperature (K) Temperature (K)

Figure 6-23. Differential thermal analysis results for (a) the first run and (b) second for

the PbTe doped with 1% CuzTe; (c) the first run and (d) second run of PbTe doped with 2

% CuzTe.

213

Electronic Transport Properties. Data were collected for the 1 and 2 % CuzTe doping

compositions (Figure 6-24). It must be noticed that after the measurements, the aspect of
the surface changed and looked dark blue. EDS analysis on the surface after measurement
gave the following atomic percentages: Pb: 46.05, Cu: 32.59, Te: 35.71. As a

consequence, measurements upon heating and cooling showed a hysteresis due to thermal

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

instability.
E2500 100
l ' (a) 990/ PbT 1°/ c T ’ ' ' b
O o e- 0 U 9 1 oo 00
@2000. _ 2 Q _120: . ( ) 99/ PbTe-1/ Cu2Te
g i >=. ‘140' I
51500- - Z’ 160‘
o . a) ' ' '
3 I ‘ -
21000- ' @1801 -
8 - g-zoo- -
E 500' I . I 0 . I. oh) -220. . . .
§ 0 - - - - 15 -240- ' '
m 300 400 500 600 700 300 400 .500 600 700
Temperature (K) Temperature (K)
A3500 -120
E - -
(‘7; 3000- (C) 98 % PbTe- 2 % Cu2Te 2 4401, ' _(d) 98 % PbTe- 2 % CuzTe
gzsoo- ' S "1501 _
:2 ' E" -130.
U 2000' m . .
131500j ' @2231 _ - . . -
81000- ' g ' T . '
E 1 I I - ' m -240. I I
'C 500' ' . . - . I .C ' I
8 0 ' ' l— -260' ' '
Q) I ' I ' I ' I ' ' j ' ' ' ' ' r '
E 300 400 500 600 700 300 400 500 600 700
Temperature (K) Temperature (K)

Figure 6-24. Temperature dependence of (a) the electrical conductivity and (b) the

therrnopower of PbTe doped with 1 % CuzTe, temperature dependence of (c) the

electrical conductivity and (d) the thennopower of PbTe doped with 2 % CuzTe.

214

6.6. Conclusions

Powder X-ray diffraction and BSE characterizations on samples CuPbmeTem+2

and CuPb13-xSnbeTe20 indicate that CuzTe and Sb2TC3 phases precipitate together in the
PbTe and Pb1-xSnxTe matrices. These regions are micro-size and run along all directions.

No ternary phase in the system Cu/Sb/T e was observed. Attempts to prepare the

compound CquTez using either fast or slow cooling resulted in a mixture of CuzTe ands
szTe3. CquTez does not seem to be a stable nor a metastable phase. Electrical

conductivity measurements on CuPbmeTem+2 (m=8, 18, 20, 22, 30, 40) indicated values
between 1800 S/cm at 300 K and ~ 400 S/cm at 675 K. This is similar to the values
reported for AnglSSbTezo. The presence of the micro-size inclusions in the cubic matrix
does not seem to affect the electrical conductivity. The therrnopower at high temperatures
has values ~ -150, -180 uV/K, which are not as high as those of Anglgstezo. Further
doping studies may result in materials with higher therrnopowers. The highest power
factor was obtained CuPblngTezo. The lattice thermal conductivities for m= 18, 20 and

40 are ~ 1.2 W/m'K at room temperature. At 675 K, the lattice thermal conductivity falls

down to ~ 0.6 W/m'K. Partial substitution of Pb by Sn results in an increase in hole
concentration in the system. For compositions close to CuPbl3Sn5SbTezo, holes become
the major charge carriers and the therrnopower changes from n— to p-type. The electrical

conductivity also increases with increasing amount of Sn. The presence of Sb2TC3 seems

to improve the thermal stability of the CuPbmeTem+2 and CuPb13-xSnbeTe20. PbTe

215

samples doped only with CuzTe exhibited thermal instability issues associated with Cu

migrating towards the surface of the sample.

216

References

l. Hicks, L. D.; Dresselhaus, M. 8., Phys. Rev. B 1993, 47, (19), 12727.

2. Hicks, L. D.; Harman, T. C.; Dresselhaus, M. 8., Appl. Phys. Lett. 1993, 63, 3230.

3. Sofo, J. o.; D., M. G., Appl. Phys. Lett. 1994, 65,2690.

4. Harman, T. C.; Taylor, P. J .; Spears, P. J .; Walsh, M. P., J. Electron. Mater. 2000,
29, L1.

5. Harman, T. C.; Spears, P. J .; Manfra, M. J ., J. Electron. Mater. 1996, 25, 1121.

6. Hsu, K. F.; Loo, S.; Guo, F.; Chen, W.; Dyck, J. S.; Uher, C.; Hogan, T.;
Polychroniadis, E. K.; Kanatzidis, M. G., Science 2004, 303, 818.

7. Quarez, E.; Hsu, K. F.; Pcionek, R.; Frangis, N.; Polychroniadis, E. K.;
Kanatzidis, M. G., J. Am. Chem. Soc. 2005, 127, 9177.

8. Androulakis, J.; Lin, C. H.; Kong, H. J.; Uher, C.; Wu, C. 1.; T., H.; Cook, B. A.;
T., C.; Paraskevopoulos, M.; Kanatzidis, M. G., J. Am. Chem. Soc. 2007, 129, 9780.

9. Poudeu, P. F. P.; D'Angelo, J .; Kong, H. J .; Short, J. L.; Pcionek, R.; Hogan, T.;
Uher, C.; Kanatzidis, M. G., J. Am. Chem. Soc. 2006, 128, 14347.

10. Poudeu, P. F. P.; D'Angelo, J .; Downey, A. D.; Short, J. L.; Hogan, T.;
Kanatzidis, M. G., Angew. Chem, Int. Ed. 2006, 45, 3835.

ll. Androulakis, J .; Hsu, K. F.; Pcionek, R.; Kong, H. J.; Uher, C.; D'Angelo, J .;
Downey, A. D.; Hogan, T.; Kanatzidis, M. G., Adv. Mater. 2006, 18, 1170.

12. Kittel, C., Introduction to Solid State Physics. Wiley: 2005.

13. Henger, G.; Peretti, E. A., Journal of the Less-Common Metals 1965, 8, 124.

14. Kuliev, R. A.; Krestovnikov, A. N.; Glazov, V. M., Zhurnal F izicheskoi Khimii
1969, 43, (12), 3063-6.

217

15. Wernick, J. H.; Benson, K. E., Physics and Chemistry of solids 1957, 3, 157.

16. Geller, S.; Wernick, J. H., Acta Crystallographic 1959, 12, 46.

17. Rosi, F. D.; Hockings, E. F .; Lindenblad, N. E., RCA Rev. 1961, 22, 121.

18. Irie, T.; Takahama, T.; Ono, T., Jpn. J. Appl. Phys. 1963, 2, 72.

19. Wolfe, R.; Wernick, J. H.; Haszko, S. E., J. Appl. Phys. 1960, 31, 1959.

20. Hockings, E. F., J. Phys. Chem. Solids 1959, 10, 341.

21. Ye, L.-H.; Hoang, K.; Freeman, A. J.; Mahanti, S. D.; He, J .; Tritt, T. M.;
Kanatzidis, M. G., Phys. Rev. B 2008, 7, 245203.

22. Zhuse, V. P.; Sergeeva, V. M.; Shtrum, E. L., Zhurnal T eknicheskoi F iziki 1958,
3, 1925.

23. Kuliev, R. A.; Krestonikov, A. N.; Glazov, V. M., Russian Journal of Physical
Chemistry 1969, 43, 12.

24. Sharaf, K. A. ; Abdel Mohsen, N.; Naser, S.; Abou El-Ela, A. F. H., F izika
(Zagreb) 1991, 23, (4), 317-23.

218

Chapter 7

Conclusion and Future Directions

For low temperature thermoelectric applications, the highest figure of merit
reported in the past decade is for CSBI4T66 (ZT ~ 0.8 at 225 K).l’ 2 The material exhibits
features suitable for thermoelectric properties. The Cs atoms weakly bonded to the
[Bi4Te6]' framework act as rattlers and contribute to the reduction of the lattice thermal
conductivity. Moreover the existence of Bi-Bi bonds within the framework results in a
very narrow band gap. Using CsBi4Te6 as a platform for exploratory synthesis, the
introduction of Pb in the anionic framework resulted in the discovery of the homologous
family CstmBi3Te5+m (m=1, 2, 3, 4).3’ 4 The crystal structure of these compounds is
similar to that of CsBi4Te6, the main difference being the loss of Bi-Bi bonds due to the
mixed occupancy of the Bi sites with Pb in the layered framework. The electronic
properties of the four members are not as promising as those of CsBi4Te6. Substituting Pb

and Cs by other elements may be a tool to tune the electronic properties. In the first

chapter of this dissertation, attempts to substitute Pb and Cs by other elements in

CstBi3Te6, the first member of the homologous family CstmBi3Te5+m, were reported.
The reactions lead to the discovery of nine new compounds Cso,76K0.74Bi3.5Te6 (1),
CsNao.9sBi4.orTe7 (2), C5069C30.6SBi3.34Te6 (3), Rbo.82Pbo.8zBi3.18T66 (4),

Rbo.19K1.3lBi3.50T66 (5), RbSnBisTee (6), Rbo.94Cao.94Bi3.06Ted (7). RbeBi3Teé (8) and

219

KSnSb3Te6 (9). Refinements of single crystal X-ray diffraction data indicated that
C5074K0.7613i3.5T66, C5069C30.6SBi3.34Te6s RboszpboszBisisTeé, Rbo.19K1.3iBi3.50Tes,
RbSnBi3Te6 and Rb0_94Ca0_94Bi3,06Te6 crystallize in the orthorhombic space group Cmcm
and are isostructural to CstBi3Te6. CsNa0,9gBi4,01Te7 is isostructural to CstzBi3Te7, the
second member of the family. KSnSb3Te6 exhibits a different crystal structure found for the

selenides CsAg0_5Bi3,5Se6 and CstBi3Se6.5 The thermoelectric properties of these

compounds could not be assessed because these compounds can not be prepared as pure

phases so far. Alkali tellurides and the ternary phases CsBi4Te6 and RbBi3,66Te6 were

identified as side products. The quality of the needles also was quite poor.
For high temperature applications, the bulk nanostructured-systems based on

PbTe showed promising thermoelectric properties. These are both n-type
(AngmeTem+2 (LAST),6 Pb1-xSnxTe-PbS7) and p—type (Na1-bemeyTem+2 (SALT),8
Ag(Pbl-ySny)meTe2+m (LASTT)9) thermoelectric materials. The work done in this
dissertation was related to the p-type systems NanmeTem+2 (Chapter 3), Nan13.

xSanTezo (Chapter 4) and KPb18_xSanTe20 (Chapter 5).

Among these systems, NanmeTem+2 exhibited the most promising

thermoelectric properties. Characterization of members with m=6, 8 and 12 showed

thermal conductivity values varying between 1.05 W/m-K at 300 K and 0.65 W/m°K at
670 K for m=6. This is slightly lower than that reported for NaogstzoSbTezz (0.85
W/m°K at 670 K).8 This decrease in thermal conductivity derives mainly through their

low lattice thermal conductivity (~ 0.45 W/m-K at 650 K for Nangstelo). TEM studies

220

on low and high m members of NanmeTem+2 revealed the presence of nanostructures

embedded in the PbTe matrix. This is believed to be the reason for the low lattice thermal
conductivity. High and low m value members exhibit large positive therrnopower at high
temperature (2 300 uV/K at 650 K) with electrical conductivity ~ 100-200 S/cm at 650
K.

Addition of Sn to NanlSSbTezo and KPblgstezo improved the mechanical
strength of the specimens but did not improve the electrical properties of the parent
materials. Nanlgstezo is a p-type material. Partial substitution of Pb by Sn resulted in
an increase in the hole carriers and the electrical conductivity. However, this increase in
electrical conductivity did not compensate for the decrease in therrnopower and increase

of the electronic thermal conductivity. As a consequence, the power factor of the Nan13.

xSnbeTezo compounds is not as high as that of Nanlgstezo. SEM and TEM analysis
indicated the cast ingots to be not true solid solutions. Micro-size inclusions rich in Sb
were observed by SEM and nano-size precipitates were observed by TEM. The accurate
chemical composition of these nano-size phases could not be determined due to
instrumental limitation. However their dark aspect suggests that these areas are rich in
light elements.

The KPb13-xSanTe20 series has weaker mechanical properties than its Na
analog. They are also not so water-stable. Different types of inclusions were observed by
BSE microscopy. Some of them are rich in Sb whereas others contain Pb, K, Sb, Te and

0. Most likely a phase containing K forms upon cooling and tends to oxidize upon

contact with water. KPblgstezo is a n—type semiconductor. Addition of Sn to the system

221

introduced more holes in the system and a compensation between positive and negative
charge carriers occur for a composition close to x=5. All compositions with x 2 5 showed
p-type behavior. The power factors of these compositions do not exceed that of

KPblgstezo because of the loss in therrnopower.

Substituting Sb by Bi in Nan18_xSanTe20 and KPb13_xSanTe20 resulted in
more brittle samples. BSE imaging revealed the presence of micro-size inclusions inside

the Pb1-xSnxTe matrix. They are mixtures of pure Bi precipitates and quaternary

Pb/Sn/Bi/Te phases. TEM studies on Nan13SnsBiTezo showed the presence of nano-size

precipitates and lamellar features, which may be an indication of local ordering between
the PbTe and SnTe phases. No TEM analysis was done for the K anaolgs, but the low

lattice thermal conductivity of some samples suggests the presence of nanostructures in
the matrix. Such features were observed for the parent system K1-bem+5Sb1+7Tem+2.

The various systems studied showed low lattice thermal conductivity. Different
characterization techniques were used to analyze the samples. SEM analysis on samples
indicated the presence of micro-size inclusions rich in Sb/Bi. Because of their size and
random distribution, these inclusions could not be observed by X-ray diffraction.
HRTEM studies confirmed the presence of nano-size features. Due to experimental
limitation, the exact composition of these features is unknown. Because of Coulombic
interactions, one could assume that the regions would be rich in alkali metal and
pnicogen. However, the existence of micro-size inclusions rich in pnicogen has an impact
in the amount of pnicogen in the nano-size precipitates.

Some issues still need to be addressed. The samples are quite weak mechanically,

which jeopardizes their potential industrial applications. The more robust samples are

222

those prepared by slow cooling. Quenched samples are much more brittle. As a result,
some gradient of concentration may exist along the ingot for samples prepared by slow
cooling. SEM and TEM analysis clearly indicate the existence of micro-and nano-size
inclusions. The exact chemical composition of the nanostructures embedded in the PbTe
matrix is still unknown as TEM analysis can not give accurate information about the

nature of these nano-size inclusions. Hence the understanding of the formation of these

phases is not complete. Zhu et al have studied in detail quenched AnglngHTezo

samples.10 Detailed TEM analysis on the samples did not indicate any presence of

nanostructures. Their results prove that the formation of nanoprecipitates is favored by
slow cooling conditions, which suggest they form through nucleation and growth.

To improve the strength of the samples, more hot press and spark plasma sintering
experiments need to be explored. The technique has the advantage of resulting in more

homogeneous samples as the powders are mechanically milled for several hours before

sintering. Several studies on known thermoelectric systems such as (Bi/Sb)2Te3,”’ 12

Ge/Si13 and TAGS ((GeTe),((AngTe2)100.x)14 alloys have shown significant

improvement in the figure of merit compared to their bulk counterpart.

In the case of bismuth antimony alloys, nanopowders from the elemental chunks”
and from grinded cast ingots12 were pressed by direct-current hot-press. A peak ZT value
~ 1.3 and 1.4 at 473 K was Iobtained for the pellets prepared from elemental chunksll and

ingots12 respectively (for comparison, commercial bulk (Bi/Sb)2Te3 have maximum ZT ~

1 at 350 K). Such improvement in ZT derives from a reduction in the thermal

conductivity. TEM analysis showed that grains have good crystallinity and large angles

223

between them, which results in isotropic properties. Nanosized features within the grains
were also observed. According to EDS, these features are pure Sb phase in the case of
powders from the elements and Te phase in the case of powders from cast ingots.

A maximum ZT ~ 1.3 was reported at 1173 K for n—type nanostructured bulk
SiGe (bulk commercial materials have maxi ZT ~ 0.9 at 1173 K).13 The SiGe
nanopowders, prepared by mechanical alloying, were hot-pressed by the direct current
induced hot pressing at temperatures ~ 1273-1473 K. A thermal conductivity ~ 2.5

W/m-K was reported for the nanostructured SiGe, much lower than that reported for the
bulk alloy (4.6 W/m°K).

In the case of TAGS, ingots with composition (GeTe)x(AngTe2)100_x (x=75, 80,
85 and 90) were prepared by temperature synthesis. The ingots were then crushed, ball-
milled and hot-pressed 773 K under 70 MPa for 30 min. A low lattice thermal
conductivity ~ 0.8 W/m~K at room temperature was measured for all compositions.l4 As a
result, ZT ~ 1.5 at 725 K was reported for x = 75, 80 and 85. TEM studies showed
nanoscale inhomogeneities embedded in crystallites with sizes varying between tens of
nanometers to several microns.

Hot—press experiments have also been carried out on the LAST system.15 Both
stoichiometric, Pb-rich and Te-rich compositions were studied. Maximum ZT ~ 1.1 was
reported for Ag0.3Pb18.3Sb0_3Te20 but no TEM study was mentioned. Recent studies have
reported on the thermoelectric properties of Pb-rich LAST pellets prepared by combining

16,17

mechanical alloying and spark-plasma sintering (SPS). The pellets were annealed for

different periods. The advantage of using SPS compared to hot-pressing is that the heat is

generated internally, which facilitates a very high heating or cooling rate. ZT ~ 1.5 was

224

 

reported for the pellet with composition Ag0,ng22,5SbTe20 annealed for 30 days at 700

K. Nanoscopic black regions embedded in the PbTe matrix were observed by HRTEM.
The number and size of these precipitates increase with annealing time. The presence of
these inclusions further scatters phonons. As a result, the thermal conductivity of

annealed samples is lower than that of unannealed samples.

Substitution of Ag by Cu in the AngmeTem+2 (LAST) system resulted in

stronger samples. SEM analysis clearly showed precipitates of CuzTe and Sb2TC3 inside

the PbTe matrix. The electrical conductivity of the samples was similar to that of the
LAST compounds. All samples showed n-type behavior but the absolute value of the
therrnopower was not as high as that of LAST. The thermal conductivity of samples with

m 2 18 is slightly higher than that of LAST. As a result ZT no higher than 0.9 was
obtained for m=30. PbTe samples doped only with CuzTe exhibited thermal instability

issues associated with Cu migrating towards the surface of the sample.

We can see from these studies that partial substitution of Pb by Sn in

Nanlgstezo and KPBlngTezo does not improve the figure of merit ZT of the parent
materials. The best p—type materials for thermoelectric materials are the Ag(Pb1-

ySny),,,srarl=:2.,,, (LASTT)9 and Nal-bemeyTem+2 (SALT)8 systems.

More complete experiments should be carried out to better understand the factors
influencing the size and dispersion of the nanoparticles embedded in the PbTe matrix for
the SALT system. Such experiments would require TEM studies of samples prepared
with different cooling profiles and annealed under different conditions (temperature,

time). More importantly, hot press and spark plasma sintering experiments should be

225

pursued. When optimal conditions for processing are determined, more robust specimen
can be obtained. Further reduction in lattice thermal conductivity can be expected as a
result of grain scattering. These synthesis techniques could be a useful tool to investigate
the influence of the m value on the thermoelectric properties. Specimen with m values
higher than 20 were too brittle to be characterized. More studies about the influence of

the Na/Sb ratio may also help to optimize the power factor of the system.

226

References

l. Chung, D.-Y.; Hogan, T.; Brazis, P.; Rocci-Lane, M.; Kannewurf, C.; Bastea, M.;
Uher, C.; Kanatzidis, M. G., Science 2000, 287, 1024.

2. Chung, D.-Y.; Hogan, T. P.; Rocci-Lane, M.; Brazis, P.; Ireland, J. R.;
Kannewurf, C. R.; Bastea, M.; Uher, C.; Kanatzidis, M. G., J. Am. Chem. Soc. 2004, 126,
6414.

3. Hsu, K. F.; Chung, D. Y.; La], 8.; Mrotzek, A.; Kyratsi, T.; Hogan, T.;
Kanatzidis, M. G., J. Am. Chem. Soc. 2002, 124, 2410.

4. Hsu, K. F.; Lal, S.; Hogan, T.; Kanatzidis, M. G., Chem. Commun. 2002, 13,
1380.

5. Kim, J. H.; Chung, D. Y.; Kanatzidis, M. G., Chem. Commun. 2006, 15, 1628.

6. Hsu, K. F.; Loo, S.; Guo, F.; Chen, W.; Dyck, J. S.; Uher, C.; Hogan, T.;
Polychroniadis, E. K.; Kanatzidis, M. G., Science 2004, 303, 818.

7. Androulakis, J.; Lin, C. H.; Kong, H. J.; Uher, C.; Wu, C. 1.; T., H.; Cook, B. A.;
T., C.; Paraskevopoulos, M.; Kanatzidis, M. G., J. Am. Chem. Soc. 2007, 129, 9780.

8. Poudeu, P. F. P.; D'Angelo, J.; Downey, A. D.; Short, J. L.; Hogan, T.;
Kanatzidis, M. G., Angew. Chem, Int. Ed. 2006, 45, 3835.

9. Androulakis, J.; Hsu, K. F.; Pcionek, R.; Kong, H. J.; Uher, C.; D'Angelo, J.;
Downey, A. D.; Hogan, T.; Kanatzidis, M. G., Adv. Mater. 2006, 18, 1170.

10. Zhu, T. J.; Yan, F.; Zhang, S. N.; Zhao, X. B., J. Phys. D: Appl. Phys. 2007, 40,
(1 l), 3537.

11. Ma, Y.; Hao, Q.; Poudel, 8.; Ian, Y.; Yu, B.; Wang, D.; Chen, G.; Ren, Z., Nano
Lett. 2008, 8, (8), 2580.

227

12. Poudel, B.; Hao, Q.; Ma, Y.; Lan, Y. C.; Minnich, A.; Yu, 8.; Yang, J.; Wang, D.
Z.; Muto, A. J.; Vashaee, D.; Chen, X. Y.; Liu, J. M.; Dresselhaus, M. S.; Chen, G.; Ren,
Z. F., Science 2008, 320, 634.

13. Wang, X. W.; Lee, H.; Lan, Y. C.; Zhu, G. H.; Joshi, G.; Wang, D. Z.; Yang, J.;
Muto, A. J.; Tang, M. Y.; Klatsky, J.; Song, 8.; Dresselhaus, M. S.; Chen, G.; Ren, Z. F.,
Appl. Phys. Lett. 2008, 93, (19), 193121.

14. Yang, S. H.; Zhu, T. J.; Sun, T.; He, J.; Zhang, S. N.; Zhao, X. B.,
Nanotechnology 2008, 19, (24), 245707.

15. Kosuga, A.; Uno, M.; Kurosaki, K.; Yamanaka, S., J. Alloys Compd. 2005, 391,
288.

16. Wang, H.; Li, J.-F.; Nan, C.-W.; Zhou, M.; Liu, W.; Zhang, B.-P.; Kita, T., Appl.
Phys. Lett. 2006, 88, 092104.

17. Zhou, M.; Li, J.-F.; Kita, T., J. Am. Chem. Soc. 2008, 130, (13), 4527.

228