NOVEL ANTIMONY-BASED HEUSLERS WITH POTENTIAL SPINTRONIC

APPLICATIONS

by
NARIMAN NAGHIBOLASHRAFI
ARUNAVA GUPTA, COMMITTEE CHAIR
PATRICK R. LECLAIR
SUBHADRA GUPTA
MARK L. WEAVER
AMBER GENAU

A DISSERTATION

Submitted in partial fulfillment of the requirements

for the degree of Doctor of Philosophy
in the Materials Science Program
in the Graduate School of
The University of Alabama

TUSCALOOSA, ALABAMA

2017
Copyright Nariman Naghibolashrafi 2017
ALL RIGHTS RESERVED
 ABSTRACT

The Heusler alloys, or compounds to be precise, have become a well-known name among

materials scientists, solid-state physicists and other researchers of the field. In this set of studies,

we examined a ternary system, i.e. Fe-Ti-Sb using experimental methodology comprised of

different microstructural and structural tests alongside theoretical evaluations. The results

showed that there was a compound present in the system, Fe1.5TiSb, which had the structural hall

marks of a Heusler, albeit with vacancies in the tetragonal positions. This meant that the

compound not only was most probably ‘layered’ between half a full Heusler, but also it was

paramagnetic, the compound exhibited a novel Slater-Pauling behavior.

The next system studied was FexCo1-xTiSb, eventually leading to observation of a novel

half Heusler compound, Fe0.5Co0.5TiSb. The compound showed a C1b signature and no sign of

any other phase were witnessed microstructurally. Magnetometry and transport measurements

revealed the nature of the material to be an antiferromagnetic material with possible

semiconductivity. Theoretical evaluations further validated the possibility of an

antiferromagnetic coupling between iron and cobalt in this compound.

Following the discovery of the Fe1.5TiSb, colleagues in our theory group encouraged us

to look into another possible ‘layered’ Heusler, Co1.5TiSn, based on the former compound’s

prototype. The experiments not only found that the Co1.5TiSn has all the hallmarks of a Heusler,

but also magnetometry tests, especially the interesting evaluation of a Rhodes-Wohlfarth model,

led to conclusion that the compound has half-metallic characteristics, found also by theory.

ii
 LIST OF ABBREVIATIONS AND SYMBOLS

A: Area

AFM : Antiferromagnetic

B0 : Flux Density

BSE : Backscattered Electrons

CCD : Charged-Coupled Device

CE : Cluster Expansion

d: Interplanar Spacing

DFT : Density Functional Theory

DOS : Density of States

EBSD : Electron Back Scatter Diffraction

Ef : Fermi Energy Level

ε: Electromotive Force

F: Ferromagnetic

FCC : Face-Centered Cubic

FIB : Focused Ion Beam

FM : Ferromagnetic

FWHM : Full Width at Half Maximum

φ: Magnetic Flux

ICSD : Inorganic Structure Database

iii
L: Camera Length

λ: Wavelength

MOKE : Magneto Optic Kerr Effect

MRAM : Magnetic Random Access Memory

MTJ : Magnetic Tunnel Junction

n: Integer

n : Number of Electrons in the Minority Channel

N: Number of Experimental Data Points

NM : Non-Magnetic

ω: Angular Frequency

pc : Effective Paramagnetic Moment

ps : Saturation Moment

Rexp. : Expected R factor

Rwp : Weight-Profile R factor

r: Distance of the incident beam from the diffracted beam

SC : Superconductivity

SE : Secondary Electron

SEM : Scanning Electron Microscopy

STT : Spin Torque Transfer

TEM : Transmission Electron Microscopy

t: Time

θ: Bragg Angle

iv
VSM : Vibrating Sample Magnetometry

w: Weighting factor

XRD : X-Ray Diffractometer

χ: Chi factor

y: Actual Intensity

v
 ACKNOWLEDGEMENTS

I would like to take this opportunity to express my gratitude to all the remarkable

individuals who assisted me with this research project. First and foremost I wish to thank my

advisor, Professor Arunava Gupta, for his outstanding technical management and organizational

leadership during all the years I was involved in this set of studies. Without Arun’s expert-level

guidance I could have not achieved the present comfort in performing research and obviously in

bringing this dissertation in its present format. My gratitude also goes to Professor William H.

Butler, who as the principal investigator of the DMREF project, agreed to my presence in the

project as a Ph.D. student and supported me both financially and intellectually as a world

renowned expert of the field. I am highly indebted to Bill.

I am also grateful for my dissertation committee members, Professor Patrick R. LeClair,

Professor Subhadra Gupta, Professor Mark L. Weaver and Professor Amber Genau for the time

and effort they put in considering my dissertation and giving me valuable input and

encouragement.

Also I would like to thank our colleagues in the Northwestern University and the

University of Virginia, led by Professor Chris Wolverton and Professor Avik Ghosh respectively,

whose collaboration was indispensable in culmination of our project.

My thanks also go to my many teammates, fellow lab members and technical staff in the

University of Alabama whose help was essential in my progress toward finishing this

dissertation.

vi
 This research would have been impossible without unwavering support of my family. My

father, Mohammad Naghib, my mother, Victoria Kamkar and my sister, Rebecca DesPrez

always encouraged me in the challenging times to take one step further when the path became

even more demanding. They are always in my heart and I am eternally grateful to them. Without

them this dissertation could not have been realized.

Last but not the least I thank the Tri-campus materials science Ph.D. program at the

University of Alabama and the faculty members for the support they gave me all these years.

vii
 CONTENTS

ABSTRACT...........................................................................................................................ii

LIST OF ABBREVATIONS AND SYMBOLS....................................................................iii

ACKNOWLEDGEMENTS ...................................................................................................vi

LIST OF FIGURES ...............................................................................................................xi

CHAPTER 1. INTRODUCTION ..........................................................................................1

1.1 Chemical Bonding and Crystal Structure ....................................................................3

1.2 Electronic Structure and Half Metallicity ...................................................................8

1.3 Research Objectives ....................................................................................................13

CHAPTER 2. EXPERIMENTAL METHODS .....................................................................14

2.1 Arc Melting and Preliminary Sample Preparation ......................................................14

2.2 Metallography .............................................................................................................20

2.3 X-Ray Diffraction Analyses ........................................................................................26

2.4 Electron Microscopy Methods ....................................................................................34

2.5 Magnetic Measurements..............................................................................................41

2.6 Theoretical Calculations ..............................................................................................43

2.7 References ...................................................................................................................43

CHAPTER 3. SYNTHESIS AND CHARACTERIZATION OF Fe-Ti-Sb

INTERMETALLIC COMPOUNDS: DISCOVERY OF A NEW SLATER-PAULING
PHASE ................................................................................................................................46

3.1 Introduction ...................................................................................................................46

viii
 3.2 Electronic Structure of Stoichiometric Heusler Compounds Based on
Fe, Ti and Sb .................................................................................................................47

3.2.1 Full Heusler .............................................................................................................47

3.2.2 Half Heusler ............................................................................................................49

3.3 Experimental .................................................................................................................50

3.4 Results and Discussion .................................................................................................52

3.4.1 Fe2TiSb ...................................................................................................................52

3.4.2 Fe1.5TiSb .................................................................................................................58

3.4.3 FeTiSb .....................................................................................................................61

3.5 Theoretical Cluster Expansion Treatment of the Fe-Ti-Sb Alloy System ....................66

3.6 Conclusion ....................................................................................................................69

3.7 References .....................................................................................................................72

CHAPTER 4. STRUCURAL AND MANGETIC PROPERTIES ANALYSES OF THE

BULK FexCo1-xTiSb COMPOUND AS PROTOTYPICAL HALF HEUSLER ................75

4.1 Introduction ...................................................................................................................75

4.2 Experimental .................................................................................................................76

4.3 Results and Discussion .................................................................................................77

4.3.1 CoTiSb ....................................................................................................................77

4.3.2 Fe0.25Co0.75TiSb .......................................................................................................80

4.3.3 Fe0.5Co0.5TiSb .........................................................................................................82

4.3.4 Fe0.75Co0.25TiSb ......................................................................................................88

4.3.5 FeTiSb .....................................................................................................................93

4.4 First-Principle Calculations of Fe0.5Co0.5TiSb Electronic Bandstructure .....................94

ix
 4.5 Conclusion ....................................................................................................................97

4.6 References .....................................................................................................................97

CHAPTER 5. SYNTHESIS AND CHARACTERIZATION OF Co-Ti-Sn ALLOYS:

Co1.5TiSn AS A HALF METAL BASED ON THE Fe1.5TiSb PROTOTYPE OF
LAYERED HEUSLER ......................................................................................................99

5.1 Introduction ...................................................................................................................99

5.2 Experimental .................................................................................................................101

5.3 Results and Discussion .................................................................................................102

5.3.1 Co1.5TiSn .................................................................................................................102

5.3.2 Co2TiSn ...................................................................................................................108

5.3.3 Co1.75TiSn ...............................................................................................................112

5.3.4 CoTiSn ....................................................................................................................115

5.3.5 Co1.33TiSn ...............................................................................................................118

5.4 Conclusion ....................................................................................................................123

5.5 References .....................................................................................................................124

CHAPTER 6. CONCLUSION .............................................................................................126

REFERENCES ....................................................................................................................132

x
 LIST OF FIGURES

1.1 The unit cell of Cu2MnAl , the prototype for the X2YZ Heusler Structure .................................4

1.2 The unit cell of CuMgSb half Heusler compound .......................................................................6

1.3 The Prototype of quaternary Heusler, the LiMgPdSn unit cell ...................................................7

1.4 (a) The Density of States (DOS) of a metal (b) a non-magnetic metal (c) a ferromagnetic
material (d) half metallic ferromagnet (e) compensated half metallic ferrimagnet .....................9
1.5 The schematic electronic bandstructure of a half metallic Heusler with the relevant orbital
contributions ................................................................................................................................10
1.6 (a) The Slater-Pauling behavior for some of the cobalt-based full Heuslers. (b) The bcc and
fcc sections of the generalized Slater-Pauling diagram with respect to the behavior of some
full Heuslers .................................................................................................................................11
1.7 Schematic configuration of a PCAR setup ..................................................................................12

2.1 A picture of Edmund B𝑢̈ hler GmbH compact arc melter MAM-1 with its main component
defined..........................................................................................................................................15
2.2 The MAM-1 arc melter chamber with electrode and crucible positions .....................................17
2.3 The Co2CrIn sample with an irregular shape resulting from lack of miscibility of chromium....18
2.4 Co2CrGe and Fe2MnGe arc melted samples sealed in quartz tubes ............................................19
2.5 Struers LaboPress-3TM Hot mounting system ..............................................................................21
2.6 The Struers DAP-VTM grinding and polishing system ................................................................22
2.7 Interplanar spacing of different crystal systems ..........................................................................28
2.8 The planar reflections in specific Bravais lattices .......................................................................29
2.9 (a) The schematic diagram of a scanning electron microscope (b) The schematic of a
transmission electron microscope ................................................................................................35

xi
2.10 The schematic of main parts of an EBSD setup ........................................................................39
2.11 Schematic of a Quantum Design VSM system ..........................................................................42
3.1 Calculated density of states for L21 Fe2TiSb ...............................................................................49
3.2 Calculated density of states for C1b FeTiSb ................................................................................50
3.3 (a) the metallography image of a Fe2TiSb sample heat treated at 900 °C for 7 days. (b) SEM
micrograph of the same surface showing the grain structure. .....................................................52

3.4 (a) The XRD pattern measured using Co Kα radiation of the Fe2TiSb sample heat treated at
900 °C for 7 days. (b) The lattice value is also confirmed by a Rietveld refinement analysis
with good agreement with the proposed model ...........................................................................54
3.5 (a) The electron micrograph of the as-cast Fe2TiSb sample heat treated at 900 °C for 7days
taken by EBSD (b) the phase color map scan shows 90% (red color) of the sample has a
homogenous L21 structure with a lattice constant of 0.5967 nm .................................................56
3.6 (a) Magnetization curve of a nominally Fe2TiSb sample at 5 K shows the dominant
ferromagnetic nature of the sample. (b) Temperature dependent magnetization shows no sign
of phase transformation or Curie temperature below 400 K........................................................57
3.7 (a) an optical micrograph of an as-cast Fe1.5TiSb sample heat treated at 900 °C for 7 days.
(b) An SEM micrograph of the same surface showing the grain structure. .................................58

3.8 The XRD pattern measured using Co Kα radiation of the Fe1.5TiSb compound. ......................59

3.9 (a) A SAED ring pattern of Fe2TiSb sample after heat treatment at 900 °C for 7 days. (b) A
SAED pattern of Fe1.5TiSb heat treated at 900 °C for 7 days. .....................................................60
3.10 (a) the magnetization of Fe1.5TiSb shows no indication of saturation for the fields up to
7200 kA/m at T = 5 K, showing the paramagnetic nature of the dominant phase. (b)
Magnetization of the sample as a function of temperature measured in an applied field of H
= 80 kA/m during heating. .........................................................................................................61
3.11 (a) An optical metallography image of FeTiSb exhibiting two distinct granular phases
present. (b) Shows the SEM micrograph of an iron poor grain, whose composition was
found to be Sb1.8Ti1.5Fe. (c) Shows an iron-rich grain, whose composition was Fe1.5TiSb ....62

3.12 The XRD pattern measured using Co-Kα radiation of the as-cast FeTiSb compound..............63

3.13 (a) Magnetization curves for FexTiSb (x = 1, 1.5, 2) specimens. (b) Temperature-dependent
magnetization of these samples measured in the applied field of H = 80 kA/m during the
heating .......................................................................................................................................63

xii
3.14 Field-cooling and zero field cooling of Fe1.5TiSb and Fe2TiSb samples confirm the absence
of superparamagnetism in these samples ...................................................................................65
3.15 Energies of 66 different arrangements of Fe and □ on the Fe sublattice of Fe2TiSb in the L21
structure, calculated using DFT, and those predicted by a binary cluster expansion (CE) fit
to the DFT energies. ...................................................................................................................67
3.16 Structure of the low energy R3m Fe1.5TiSb phase obtained from a cluster expansion
illustrating the layering of the L21 Fe2TiSb and C1b FeTiSb along the [111] direction of the
underlying cubic lattices. ............................................................................................................68
3.17 The density of electronic states of the low-energy Fe1.5TiSb R3m structure predicted by the
cluster expansion. .......................................................................................................................68
4.1 (a) The Metallography image of a CoTiSb Sample heat treated at 900 °C for 7 days. (b)
SEM micrograph of the same surface showing the grain structure and its relevant EDS
spectrum .......................................................................................................................................78
4.2 The XRD diffraction data of the CoTiSb sample heat treated at the 900 °C for 7 days. .............79
4.3 The near zero Kelvin hysteresis loop of the CoTiSb compound heat treated at 900 °C for
7days ............................................................................................................................................79
4.4 (a) The micrograph of the Fe0.25Co0.75TiSb sample heat treated at 900 °C for 7 days and (b)
its EPMA elemental map based on the iron element showing the grain segregation. .................79
4.5 The XRD Pattern of the Fe0.25Co0.75TiSb heat treated for 7 days at 900 °C. ...............................81

4.6 The magnetic behavior of the Fe0.75Co0.25TiSb alloy system at 5˚K and at 300˚K. ..................82
4.7 a) the optical micrograph of a Fe0.5Co0.5TiSb heat treated at 900 °C for 7 days. (b) SEM
micrograph of the same surface showing the grain structure and its relevant EDS
composition. .................................................................................................................................83
4.8 The XRD spectrum of the Fe0.5Co0.5TiSb compound. .................................................................84
4.9 a) The electron backscatter band contrast image of a relatively large area on the surface of the
Fe0.5Co0.5TiSb sample. b) The relevant EBSD phase map scan image of the same area
showing that almost the entirety of the surface corresponds to the hypothetical half-Heusler
C1b phase c) The orientation map of the same location and its inverse pole figure showing the
texture formation of the sample ...................................................................................................85
4.10 A SAED pattern of Fe0.5Co0.5TiSb heat treated at 900 °C for 3 days. .......................................86

xiii
4.11 (a) Moment versus temperature isotherms of the Fe0.5Co0.5TiSb at a range of 5 Kelvin up to
the room temperature. (b) The susceptibility versus temperature diagram of the compound. ...88
4.12 (a) The Resistance-Temperature behavior of the Fe0.5Co0.5TiSb compound at zero field (b)
The Resistance-Temperature behavior of the Fe0.5Co0.5TiSb compound at 1kOe. (c) The
exponential fit at the less than 10 Kelvin .....................................................................................88
4.13 The XRD Pattern measured using a Co Kα radiation of the Fe0.75Co0.25TiSb sample heat
treated at 900 °C for 14 days. .....................................................................................................90
4.14 (a) an optical micrograph of the Fe0.75Co0.25TiSb alloy showing grain formations and
evolution of a secondary phase between them. (b) SEM image exhibiting the grains and the
secondary phase found to have a stoichiometry of Sb1.8Ti1.5Fe through EDS ...........................91
4.15 (a) Electron Backscatter band contrast image of a large section of the Fe0.75Co0.25TiSb alloy.
(b) The phase map taken from the cross section attributes the C1b Fe0.5Co0.TiSb structure to
the grains (colored red). (c) The relevant orientation map and inverse pole figure of the cross
section .........................................................................................................................................92
4.16 The 5 Kelvin and 300 Kelvin hysteresis loops of the Fe0.75Co0.25TiSb alloy ............................92
4.17 (a) An optical metallography image of FeTiSb exhibiting two distinct granular phases
present. (b) The SEM image and the relevant EDS of the intergranular secondary phase
being Sb1.8Ti1.5Fe. (c) The composition of the grains was found to be Fe1.5TiSb ......................94
4.18 The XRD spectrum of the FeTiSb compound. ..........................................................................94
4.19 Calculated electronic density of states (DOS) of Fe0.5Co0.5TiSb with (left) ferromagnetic
(FM) and (right) antiferromagnetic (AFM) order. .....................................................................96
4.20 Schematics showing a few possible antiferromagnetic configurations in a C1b
Fe0.5Co0.5TiSb crystal structure. .................................................................................................96
5.1 Electronic Bandstructure of the Co1.5TiSn in an R3m vacancy stacking.....................................100
5.2 (a) The optical micrograph of a Co1.5TiSn sample homogenized at 900 °C for 3 days. (b) The
SEM image of the same sample. ..................................................................................................103
5.3 The experimentally observed pattern of Co1.5TiSn, the calculated fit and the difference
pattern ..........................................................................................................................................104
5.4 (a) the band contrast image of the Co1.5TiSn alloy microstructure. (b) The phase map of the
same cross section. .......................................................................................................................105
5.5 (a) Paramagnetic behavior of Co1.5TiSn compound at 300 Kelvin. (b) The hysteresis loop of
the material at 5 Kelvin. The saturation moment stands at 0.63 µB/f.u. ......................................106

xiv
5.6 (a) The inverse mass susceptibility behavior of Co1.5TiSn with respect to temperature at 2
Tesla (b) Linear portion of the paramagnetic region of the same graph. .....................................107
5.7 (a) the optical metallography of the Co2TiSn alloy heat treated at 1000 Celsius for 7 days. (b)
The SEM image of the same cross section and EDS of a grain which exhibited a target
stoichiometry of Co49at.%Ti24at.%Sn24at.% .......................................................................................108
5.8 The XRD Pattern of the Co2TiSn using a Kα Cobalt source. .......................................................109
5.9 (a) Band contrast image of a Co2TiSn cross sectional area with relatively large grains and
some intergranular porosity. (b) The relative phase map showing almost the entirety of the
microstructure colored red signifying a high match with the intended L21 phase.......................110
5.10 (a) the hysteresis loop of the homogenized Co2TiSn at 300 Kelvin. (b) The hysteresis loop
at 5 Kelvin. .................................................................................................................................111
5.11 (a) The inverse mass susceptibility behavior of Co2TiSn versus temperature at 2 Tesla (b)
The linear fit to the paramagnetic portion of the previous plot. .................................................112
5.12 (a) Optical metallography image of the homogenized Co1.75TiSn sample. (b) The SEM
image of the same area and an EDS taken from a grain indicate that the composition is
consistent with the intended stoichiometry of Co1.75TiSn ..........................................................113
5.13 The XRD Spectrum of Co1.75TiSn compound. ..........................................................................114
5.14 The hysteresis loop of the Co1.75TiSn compound at 5 Kelvin and 300 Kelvin ..........................114
5.15 (a) Inverse mass susceptibility of Co1.75TiSn with respect to temperature at 3 Tesla (b) The
linear fit to the paramagnetic region .........................................................................................115
5.16 The microstructure of the CoTiSn heat treated for 7 days at 1000 °C. ......................................116
5.17 (a) EDS data taken from a rectangular area of the matrix. (b) The embedded phase in figure
(a) is measured to be Ti53at.%Sn47at%. (c) The smaller-sized features were found to be pure Sn .117
5.18 The XRD spectrum of a CoTiSn sample heat treated at 1000 °C for 7 days. The Co1.5TiSn,
Ti6Sn5 and Sn peaks are assigned with red, green and blue asterisks respectively ....................118
5.19 (a) The optical micrograph of a Co1.33TiSn alloy heat treated at 900 °C for 7 days. (b) The
SEM micrograph of the same sample with respective EDS of the features showing the
dominant phase to be Co1.5TiSn and the intergranular one to be Ti6Sn5 ..................................119
5.20 The XRD Spectrum obtained from a Co1.33TiSn Sample. .........................................................120
5.21 (a) Hysteresis loop of Co1.33TiSn ally at 5 Kelvin. (b) Hysteresis loop of CoTiSn alloy at 5
Kelvin. ........................................................................................................................................121
5.22 The magnetization and phase behavior in the CoxTiSn .............................................................122
xv
 CHAPTER 1

INTRODUCTION

The Heusler alloys, or compounds to be precise, have become a household name in the

field of materials science since the process of their prototype, Cu2MnAl by the German mining

engineer, Friedrich “Fritz” Heusler in 1903. 1 The significance of this discovery was in the fact

that none of the constituting elements of this compound were magnetic at room temperature

which intrigued Heusler himself.

Since that pioneering paper, more than 1000 Heusler compounds based on the first

prototype have been reported in the scientific literature 2. The experimental and theoretical

analyses manifested that these intermetallics not only are not limited to ferromagnets but a bevy

of quantum ground states, or in other words major electronic properties of matter is observed in

these materials.3 Properties such as ferromagnetism in Cu2MnAl 4, ferrimagnetism in Mn2VAl 5 ,

antiferromagnetism in GdPtBi 6, semiconductivity in CoTiSb 7 , superconductivity in Pd2YSn 8,

half metallicity in NiMnSb 9, heavy fermion behavior in Cu2CeIn 10, magneto-optic behavior in

PtMnSb 11 , thermoelectricity in Zr0.25Hf0.25Ti0.5NiSn0.998Sb0.002 12 and most recently the

topological insulator behavior in LuPtSb 13.

It is obvious that there is immense technological potential available based on harvesting

the aforementioned material properties.

1
 In fact, currently there are interesting technological applications such as spintronics and

its relevant role in the development of magnetic random access memories (MRAM), with spin

transfer torque (STT) version of it being under investigation as candidate for the next generation

of nonvolatile data retention.14 Spin polarization of half metallic Heuslers has the pivotal role in

realizing these fascinating technologies Although obtaining results well suited for market-

friendly devices is challenging and yet to be realized, there has been some exciting advances

using Heuslers to achieve the required goals. For instance, Co2MnAl thin films are recently

reported to have been applied in the magnetic tunnel junction (MTJ) with improved room

temperature characteristics 15.

Another major industrial application of Heusler alloys as mentioned previously is their

utilization in the energy harvesting technologies. Reduction of the greenhouse gas emissions has

become a major environmental concern emphatically mentioned in international accords such as

the Kyoto agreements to be reduced in order to avoid the detrimental effects of global warming.
16
The current estimates indicate that at least 60 percent of the entire industrial energy generated

is dissipated in form of the heat. 17 As an important example in a typical commercial gasoline-

fueled internal combustion engine vehicle almost 75 percent of entire fuel energy is wasted in the

form of heat.18 Considerable amount of this heat can be converted back into electricity and thus

harvested by using Thermoelectric devices in effect reducing fuel consumption.

Thermoelectric materials have a dimensionless efficiency figure of merit referred to as ZT,

which is directly related to electronic conductivity and Seebeck coefficient while inversely

related to thermal conductivity. Unfortunately, the industrial value of ZT has remained stagnant

around a value of 2 which is less than the target value of 3.

2
 Adding to that current applied thermoelectric materials are based on the traditional

Bi2Te3 and PbTe alloys which are composed of relatively hazardous elements. The Heusler

alloys do not suffer from such a drawback and might be an alternative to an environment friendly

and mechanically robust thermoelectric system with acceptable efficiency. 19, 20

1.1 Chemical Bonding and Crystal Structure

After the Cu2MnAl compound was synthesized by Heusler in 1903, the structure of the

material remained unknown until the advent of the X-Ray made the atomic diffraction and its

consequent structural analysis feasible in the latter decades of twentieth century. One of the first

researchers to use the X-Ray Diffraction (XRD) method to derive the crystal structure of these

compounds was J.F.T. Young, who in 1923 using a molybdenum source for the beam generation,

found out that there is a face centered cubic (FCC) phase present in some of the Heusler alloys.
21
In the 1930s the progress made in the quantum theory and especially Heisenberg’s treatment of

ferromagnetism with respect to electronic configuration prompted Friedrich Heusler’s son, Otto

Heusler, to investigate the crystal structure of the Heusler alloys.22 around the same time Bradley

and Rodgers in Britain also came to conclusion that these alloys are most possibly constructed of

FCC phases. 23

Since then with advances in the X-Ray diffraction and also in Neutron diffraction, the

crystal structure of the original Heusler compound is well established to be indeed related to

FCC structure. Its space group number is “255” or “Fm3̅m” in Hermann-Mauguin notation or in

its Strukturbericht designation, “L21 “. All these three mentioned designations are used in the

scientific literature.

3
 The chemical formula of a Heusler compound is manifested as X2YZ stoichiometry, in

which X is a transition metal, Y is a less electronegative transition metal or in some cases a

lanthanide or a alkaline metal and Z is the most electronegative component belonging to

occasionally metalloid elements of main groups III, IV and V or boron, carbon and nitrogen

group, respectively. The composition is consequently very adaptive and by permutation as many

as a thousand compounds are possibly attainable.

The unit cell of the prototypical Heusler, Cu2MnAl, is constructed in a way that the X

element (Manganese) sits on the 8c (1/4 , ¼ ,1/4 ) Wyckoff position, Y is on 4b ( ½, ½ ,1/2 ) and

Z is located at the 4a ( 0,0,0 ). By dividing the multiplicity number of the positions by four, the

atomic ratio from 8c:4b:4a turn out to be 2:l:1, which naturally corresponds to the X2YZ

formula. A designation of the unit cell is shown in figure 1.1.

Figure 1.1 The unit cell of Cu2MnAl, the prototype for the X2YZ Heusler Structure.

4
 The stacking of atoms in such a structure can also be represented by assuming 8 unit cells

of the BCC-arranged CsCl lattice superimposed over each other, or as it is evident from figure

1.1, with the X atoms (in this case Cu) occupying the center of each 8 CsCl-type lattice (or the

tetrahedral positions) while Y and Z atoms alternatively fill in the corner of each unit cell. The

octahedral positioning of Y and Z is attributed to the ionic bonding stemming from the

electronegative difference between these two atoms.2

Heuslers compounds are not limited to the above-mentioned stoichiometry and structure:

There is also a major subclass present under the same material category similar to the Heusler

composition in which a 2:1:1 atomic ratio between X,Y and Z elements exist, the ratio between

the same atoms will be 1:1:1 . Therefore, they are known as “half Heuslers”. Additionally, the

crystal structure of the half Heusler compounds is identical to that of the original Heusler (or Full

Heuslers as they are sometimes called to be distinguished from their half Heusler counterparts),

with the exception that the X element no longer fully occupies the tetrahedral positions and half

of the total eight of them are empty. This way the total number of the atoms within the cell is

reduced from 18 in case of full Heusler to 16 in the latter.

The prototype of such a structure is the MgAgAs compound which has the following

designations : Space group 216 in the inorganic structure database (ICSD ), C1b in the

Strukturbericht and F43̅m in the Hermann-Mauguin system with the X element at the 4a (0,0,0),

Y at 4b (1/2,1/2,1/2) and Z at 4c (1/4,1/4,1/4) position. Caution should be taken here that the

electronegativity of elements will directly affect both the nature of bonding within the crystal

structure and the atomic loci.2 As a result, a peculiarity encountered is that the MgAsAs atomic

coordinates are not prevalent among the half Heuslers with most compounds conforming to the

structure of CuMgSb in which X atom fills the 4c and Y and Z at 4b and 4a respectively. 2,24

5
 A representation of the CuMgSb is shown at figure 1.2. Care must be taken that as the

atomic ratio of the half Heusler follows a 1:1:1 ratio, theoretically there are 6 possible

permutations to put atoms in the 4a, 4b and 4c positions without compromising the 1:1:1

stoichiometry. The positioning of the atoms in each permutation will result not only a different

diffraction signature but it also has a profound effect on the electronic bandstructure and the

consequent properties.

Figure 1.2 The unit cell of CuMgSb half Heusler compound.

There is also another possible class of materials under the group of Heuslers which are

not necessary ternary intermetallics as full and half Heuslers are, but in fact are quaternary

compounds. The prototype of this subgroup is LiMgPdSn which is in fact could be considered as

a permutation of space group 216 half Heusler with Li at 4b (3/4,3/4,3/4), Mg at 4b (1/2,1/2,1/2)

, Pd at 4c (1/4,1/4,1/4) and Sn at 4a (0,0,0) . Such an atomic array can accommodate three

transition metals instead of two in case of the ternary Heuslers.

6
 The compound formulation is occasionally also shown as XX́YZ, with X valence being

higher than that of X́ , while valence of Y is lower than the other two.25, 26 A representation of

the quaternary Heusler compound prototype is shown in figure 1.3.

Figure 1.3 The Prototype of a quaternary Heusler, the LiMgPdSn unit cell.

Stacking of the constituting elements in the octahedral or tetrahedral positions alongside

their relative electronegativity is crucial in the resulting ionic or covalent bonds within the

material and their effect in determining the electronic behavior of the compound. There are

different models proposed to explain the electronic structure of the Heusler materials. 2

According to a simple tight bonding model suggested by Jung for the half Heuslers, the least

electronegative element, X, within the Heusler compound exchanges its valence electrons with

more electronegative elements Y and Z present within the formula.27 In this way charge stability

is achieved with Y having a d10 configuration and Z having a s2p6 one, basically summing to a

total of 18 valence electrons that are compensated by the electropositive X element.

7
 As a result, the half Heuslers with a total of 18 valence electrons are stable

semiconductors while a deviation from this amount to values such as 17 or 19 renders a

compound to be magnetic or occasionally a different crystal as well.27

For the full Heusler semiconductors the number of transferring valence electron changes

from 18 to 24 as a similar mechanism of electrons taken from d orbitals by the partially-filled s

and p ones is in effect.

1.2 Electronic Structure and Half Metallicity

In 1982, Van Engen and coworkers studying the PtMnSb Heusler compound in the

K.H.J. Buschow’s group in Philips research laboratories observed an exceptionally high

magneto-optic Kerr effect (MOKE) rotation in the material.28 At the time Buschow group was

exploring a considerable number of alloys and compounds for potential magneto-optic recording

applications among which were Heusler alloys which at the time were considered not to be of

much technological significance ever since their discovery earlier in the century. 29 The

encouraging experimental data on PtMnSb prompted them to delve into the theoretical analysis

of this material in order to investigate its bandstructure and subsequent magnetic behavior.

Adopting a theoretical approach proposed by Kü bler, de Groot and coworkers at Buschow’s

group and research institute for materials in Netherlands came to conclusion that a group of half

Heusler possess a peculiar electronic bandstructure which renders the material remarkably 100

percent spin polarized through a semiconducting behavior for minority spin electrons while the

majority spin electrons were metallic. They heralded this new class of materials as “half

metals”.30 All these pioneering breakthroughs sparked a revival of research on Heusler

compounds.

8
 An instructive schematic representation of the possible electronic bandstructure is shown

at figure 1.4. 2 (a) illustrates a normal metal with limited density of states (DOS) at Fermi level.

(b) Shows a metal in which the DOS at the Fermi level is the same for spin up and spin down

channels resulting in the material being spin resolved. In (c) the Fermi is located in such a

manner that the DOS is not equal anymore for the two spin channels resulting in a magnetization

of the material and so a ferromagnet is generated.

Figure 1.4 (a) The Density of States (DOS) of a metal (b) a non-magnetic metal (c) a
ferromagnetic material (d) half metallic ferromagnet (e) compensated half metallic ferrimagnet.
Reproduced with permission from ref. 2.

The DOS in (d) shows the interesting bandstructure of a half metal in which the Fermi

level (Ef) is located at a gap in the minority spin channel while at the majority channel there are

states available. Such an electronic configuration means that the material acts as a semiconductor

or insulator to the minority spin while it is conductive to the majority one, or in other words the

material is “spin polarized”.

9
 In the case of Heuslers, the minority gap occurs since there is an extensive hybridization

of the d orbitals of the two transition metals. The sp orbitals do not play a crucial role in the

generation of the minority gap as they are considerably beneath the Fermi level. A schematic of

orbital contribution to bandstructure of a half metallic Heusler is also shown at figure 1.5.

Figure 1.5 The schematic electronic bandstructure of a half metallic Heusler with the relevant
orbital contributions. Reproduced with permission from ref 32.

A crucial test for determining half metallicity in the Heusler alloys is based on ruled

derived by Slater and Pauling and named after them. The rule provides a simple relation between

the magnetic moment of the Heusler compound and the total of number of the valence electrons.

This is due to the fact that the magnetic moment is related to the number of electrons within the

minority the channel: 2

m  NV  2n (1.1)

10
 With m being the atomic magnetic moment, Nv valence electron concentration and n the

number of electrons in the minority channel. If s and p electrons are considered irrelevant and

only d orbitals in the minority state are taken into account, there will be nearly 3 electrons in the

d states, resulting in the (1.1) getting simplified as 2 :

m=Nv-6 (1.2)

In case of half Heuslers, as there are three atoms present in the compound, the total moment for

the XYZ stoichiometry will become:

MHalf-Heusler = Nv-18 (1.3)

A similar argument can be adopted for full Heuslers. The compound has 4 atoms per unit cell so

(1.2) can be rewritten as:

MFull-Heusler = Nv-24 (1.4)

The Slater-Pauling behavior for some of the half metallic cobalt-based full Heuslers and also for

some ferromagnetic elements and alloys are shown in the figure 1.6 2

Figure 1.6 (a) The Slater-Pauling behavior for some of the cobalt-based full Heuslers. (b) The
bcc and fcc sections of the generalized Slater-Pauling diagram with respect to the behavior of
some full Heuslers. Reproduced with permission from ref. 2.

11
 It should be noted that although the Slater-Pauling rule is a necessary step in verification

of half metallicity in the Heusler compounds, it is by no means a sufficient one. An essential

experimental method is to detect the spin polarization of the material. There are different

approaches devised to reach this goal with one of the relatively easy methods, referred to as

Point Contact Andreev Reflection or PCAR, widely used to confirm the spin polarized nature of

bulk or thin films. Here a superconducting electrode injects Cooper pairs into the specimen. If

the specimen is half metallic, no conductivity should be detected while the presence of normal

ferromagnet reflects a hole with opposite spin into the metal. This is relevant if bias voltage is

less than the superconductor’s gap.33 Figure 1.7 is a schematic representation of PCAR method.

Figure 1.7 Schematic configuration of a PCAR setup. A superconducting (SC) electrode sends a
cooper pair into a ferromagnet (F) and with respect to the bias voltage a conductivity is detected.
Reproduced with permission from Ref. 33.

Some qualitative methods have also been proposed in order to confirm the half metallic

character of a compound. One such a method is based on Rhodes-Wohlfarth model. Initially the

model used the ratio of the effective paramagnetic moment (pc) to saturation moment (ps) to

distinguish the Heisenberg magnets with localized electrons from weak itinerant

ferromagnets.34,35 For the former the pc/ps ratio, or Rhodes-Wohlfarth ratio, is equal to 1 while

for the latter it could be much higher.35

12
 A quarter of century later after the advent of this model, some members of the same

research group in the Netherlands who discovered the new materials class of half metals,

observed that these materials have an unusually low Rhodes-Wohlfarth ratio not witness before,

with all the values less than unity.36 The explanation that they presented was that the initial

Rhodes-Wohlfarth fails to consider the large contribution of itinerant spin-polarized electrons in

half metals leading to drastic reduction of the pc/ps ratio. Therefore, with an easy magnetometry

measurement from ferromagnetic to paramagnetic regime of the material it is possible to verify

the half metallic behavior of the compound.

1.3 Research Objectives

Prompted by extensive theoretical calculations by W.H.Butler and coworkers on both full

and half Heuslers, our objective was concentrated on evaluating which of the predicted half

metallic Heuslers had not been explored sufficiently. Our efforts pinpointed the Fe-Ti-Sb system

in which there were no substantial evidence in the literature as far as characterization of the

material was concerned. It was also seen that the microstructural analysis of Heuslers, especially

in their bulk form, was not as thoroughly performed as expected from a metallurgy/materials

science standpoint. This all encouraged us to synthesize the promising compounds through arc

melting and eventual heat treatment to evaluate the possibility of a single phase half metallic

alloy emerging.

13
 CHAPTER 2

EXPERIMENTAL METHODS

For our research we adopted the bulk synthesis of Heusler compounds by the arc melting

method, which is the prevalent choice of process for the synthesis of these metallic materials.

The next step, as it is usual with processing any novel material, was to characterize the alloys by

structural and microstructural instrumental techniques available to us in materials science, such

as metallography, Scanning Electron Microscopy (SEM), X-Ray Diffraction (XRD), Electron

Back Scatter Diffraction (EBSD) and Transmission Electron Microscopy (TEM).

Eventually as the crucial property of these alloys, the magnetic behavior was studied by using

Vibrating Sample Magnetometry (VSM).

2.1 Arc Melting and Preliminary Sample Preparation

The first step in processing the Heusler compounds in their bulk form is usually the arc

melting. A compact arc melter MAM-1 system designed by Edmund B𝑢̈ hler GmbHTM was used

for this task. A picture of the instrument is shown in figure 2.1.

14
 Figure 2.1 A picture of Edmund B𝑢̈ hler GmbH compact arc melter MAM-1 with its main
component defined. Reproduced with permission from Ref.1.

Arc melting is a well-established method for processing different metallic compounds

and alloys. As it can produce temperature up to 3500 °C it is also proved useful in synthesis of

ceramic compounds and composites.2 The benefits of the technique lies in its relatively high

speed of sample preparation, high homogeneity and user facility. In the case of intermetallics and

specifically Heuslers, the sample preparation step consists of weighting the constituting elements

in their relative atomic weight ratio with respect to the final compound formula. It is

conventional to use elements at a 3N minimum purity level. For this goal a precise weighting

scale with credible calibration should be used. In our case an electronic balance, A&D

WeightingTM ER-182A with a precision of 0.1 milligrams was utilized.

15
 Caution must be considered when volatile elements such as Manganese and to a lesser

extent Antimony is opted for arc melting, as the relatively high vapor pressure of these materials

results in the loss of matter during the arc melting and consequently in order to compensate for

such loss it is customary to put an excess amount of such elements up to 5 to 10 percent of the

initially calculated mass. After the elements are properly weighted, they are put in a crucible

inside the arc melter chamber and the system is closed for evacuation pumping. The choice of

pump is dependent on sensitivity of the target material to oxidization. In some cases such as

oxides and ceramics or some electronic materials as oxygen might play a detrimental role in final

properties of the material, the usage of high vacuum pump such a turbomolecular one for the arc

melting is recommended. 2 The MAM-1 instrument used for these studies uses a turbomolecular

pump capable of reaching a base pressure of less than 5x10-5 mbar. After such a low vacuum is

reached, 5N purity Argon is first introduced into chamber before the ignition of the arc. The

crucible upon which the elements are located is made out of copper to enhance better conduction

and also is water cooled. A picture of MAM-1 chamber with electrode and the crucible is shown

at figure 2.2. It is also recommendable to first melt an oxygen getter element such as titanium or

zirconium immediately after the ignition of the arc so that any remaining oxygen within the

chamber is eliminated. Then the electrode is moved over the constituting elements and by

gradually increasing the arc intensity and also using the electrode handle to stir the melt a molten

‘button’ composed of elements is formed . The shape of the sample is usually an indication of

how well the constituents are mixed. As mentioned before, in case of thorough mix achieved

within the sample, the shape will be quite round and therefore it is called a button, while lack of

miscibility manifests itself in an irregularly-shaped sample. An example of a properly mixed

button and a sample with clear lack of component miscibility is shown in figure 2.3.

16
 It is recommended that each run does not exceed a time limit of 60 seconds as

continuation of the arc might instigate emission of X-ray which is hazardous to the user.

Sometimes the button is cracked or the some pieces of the elements are left out during the first

run of the arc melting which means that the runs should be repeated until the button is comprised

of all the element shots or pieces and no constituting part is left out.

Figure 2.2 The MAM-1 arc melter chamber with electrode and crucible positions.

After a button in which all the elements’ pieces are amassed, it is required to flip this first

button to its back and at least melt the sample again for 2 more runs to ensure that macro

segregation is minimized. Before the utilization of scanning electron microscopy and electron

dispersive spectrometry became abundant, it was conventional to weight the finalized button and

compare it to the total weight of constituting elements in order to confirm a proper mixing of

elements and lack of macrosegregation. It should be noted in some cases this approach could be

misleading, such as the case of Cr-Mn-Sb alloys in which although the finalized mass of the

sample does not deviate from the initial weight of elements, the sample shows considerable

macro segregation of chromium if cut into pieces.

17
 So it is recommended that the arc melted samples be cut into proper pieces and their

cross sections examined before any claim about the homogeneity of the alloy.

In our studies we used an Allied High Tech Inc. Techcut 4TM precision low speed saw to cut

the samples. If no sign of macrosegregation was seen pieces were cut for heat treatment.

Figure 2.3 The Co2CrIn sample with an irregular shape resulting from lack of miscibility of
chromium. A round-shaped Fe2TiSb signaling a satisfactory constituent miscibility

Heat treatment is a crucial step in preparation of almost every metal and alloy as it is

rarely the case that the as-cast sample is homogenous or exhibits satisfactory materials

properties. So one of the crucial steps in metallic materials processing is homogenization heat

treatment with the aim of reducing segregations, both macro and micro, to a minimum.3,4 The

Heuslers usually require relatively long and high-temperature homogenization cycles as the

diffusion of the atoms progresses rather slowly within these systems.27 So in order to apply the

heat treatment cycle the samples should be detached from the atmosphere to avoid their

oxidization under such extreme conditions. A conventional method is to seal the samples in

quartz tubes under vacuum. The samples are sealed using an oxyhydrogen torch after being

pumped with a rotary-vane pump for at least 6 hours.

18
 Tantalum foils are also sometimes used to wrap the sample to avoid further possible

reactions with ambient especially in case of reactive or volatile nature of the alloy. A picture of

two sealed Heusler samples is shown in figure 2.4.

Figure 2.4 Co2CrGe and Fe2MnGe arc melted samples sealed in quartz tubes.

Heat treatment was performed using a General Signal Lindbergh TM tube furnace of the

type 55332 and also a Paragon Sentry Xpress 4.0TM box furnace. The homogenization cycle

should be adopted so in case of material containing volatile elements such as manganese,

antimony or bismuth the heating ramp is slow enough not to enhance the vaporization of the

volatile elements resulting in pressure accumulation and the consequent cracking of the quartz

tube. For this it is experimentally recommendable that the heating ramp not be adjusted faster

than 9 °C per minute. The dwelling temperatures for homogenization are empirically found but

according to our studies a rule of thumb would be a heat treatment at 900 °C for 3 days as a

starting point.

19
 Once again this is by no means a scientific rule but in absence of calorimetric

measurements it is a valid path to be considered; and if the sample is not molten after

homogenization, it can give us important data through its microstructure by showing which

feature, grains or dendrites, are present and consequently the degree in which the heat treatment

need to be advanced is inferred.

After the dwelling time has run its course, the sample is brought to room temperature.

The cooling ramp, if slow such as in furnace cooling, can result in an equilibrium atomic

structure and also single phase microstructure. But once again this is also experimentally

established and is dependent on the nature of the alloy: There are reports that fast cooling such as

quenching in cool water can be beneficial in some cases, rendering the material pure phase.27

The materials studies in our research work reached single phase structure during slow cooling

ramps of as low as 2 °C per minute. Quenching on the other hand resulted in multiple phases

emerging and material becoming rather an alloy instead of a compound, with latter being our

goal. After the end of the cycle the quartz tubes were broken and the samples were checked for

any abnormalities in shape or in form and if the sample appeared to be intact after the heat

treatment, a chip was taken off using the precision saw for magnetometry studies and the rest of

the sample proceeded to the next step which was metallography.

2.2 Metallography

Metallography is an indispensable technique for revealing the microstructure of metallic

materials. The exposed microstructure can be further analyzed using conventional optical

microscopes or by adopting electron microscopy, scanning and transmission, to elaborately study

the morphology, composition and crystal structure at higher magnifications and resolutions.

20
 The first step in preparing a metallography sample is mounting, in which the metallic

sample is encapsulated in a resin-hardened plastic cylinder basically mount. The diameter of the

mount depends on the mold used. Compression mounting machines have fixed diameters ranging

from 1 inch to 1.5 inches. The height of the sample depends upon how much resin is applied.

Usually for microscopy and X-Ray diffraction purposes it is within a range of 0.5 to 0.75

inch to both enable proper handling of the sample during grinding and polishing process and also

allow it to fit within sample requirements for microscopes and XRD instruments.5 The choice of

the mounting powder or resin relies on the factors that affect mechanical properties of the

sample, such as brittleness or edge retention or especially if scanning electron microscopy is

considered, the conductivity of mount is essential. On the other hand, if a measurement such as

electronic transport is considered, a conductive mount is not favorable so a non-conductive

mount could be considered. The applied pressure and temperature and cooling time differs from

mounting material to other and is provided by the commercial supplier but usually the

temperature is 180 °C with a pressure of 4200 psi is sufficient. The mounting instrument used in

this set of studies was a Struers LaboPress-3TM, a picture of which is shown is at figure 2.5.

Figure 2.5 Struers LaboPress-3TM Hot mounting system.

Reproduced from ref. 6.

21
 The mounting powder adopted was a graphite-based conductive resin produced by Allied

High Tech Products, Inc. As the diameter of the mounting system was 1.25 inches, the

temperature recommended for molding was 180 °C and a pressure of 20 kN was used to get the

optimum results.

After the mount is ready, the next step is grinding it so that any mount residue on sample

surface is eradicated and then the surface is further grinded so that a pristine surface void of any

oxide layer or such is prepared for polishing and eventually etching and revealing the

microstructure of the material. The medium used for grinding are circular abrasive papers made

of silicon carbide or alumina. As the former has superior abrasive performance and is relatively

cheap, it is more popular among metallographers. The diameter of the papers could be different

from 8 to 12 inches depending on different requirements and the accommodating size of the

grinder machine. The machine used in this case was a Struers DAP-V system whose image is

shown at figure 2.6

Figure 2.6 The Struers DAP-VTM grinding and polishing system.

22
 The mesh size of abrasive paper is an indicator of its roughness with lower mesh sizes

being rougher than the higher ones. Commercially there are different sizes available from 60 to

1200 and sometimes higher. A grinding recipe for a smoothly cut sample, such as the ones cut by

precision saws, would start from a mesh size of 180, and proceed to finer papers such as 320,

600 and finally 1200. It should also be noted that for samples with rougher sectioning surfaces,

for example those cut manually by a normal band saw, starting at a lower mesh size paper such

as 60 might be effective in realizing an optimum surface for metallography. For our studies a set

of 180, 320, 600 and 1200 SiC papers supplied by Allied High Tech products, Inc. were used.

The grinding process itself needs sufficient experience and dexterity from the part of

metallographer to result in a sample with an acceptably smooth surface. This is where

metallography becomes more an art than a science. A good practice is to start holding the sample

mount at the outer periphery of the rotating grinding paper with a relatively low rotating speed

such as 100 to 300 rpm and advancing towards the center of the paper. Water should be

constantly applied during the process to wash away the grinding residue and also cool down both

the sample and the paper to avoid any potential detrimental thermal effect.

The amount of manual pressure applied on the mount is another factor which should be

learned through various trials and experiences. An uneven force applied on the sample will result

in the loss of flatness as some edges are grinded more than the others. A skilled metallography

technician usually holds the mount on a certain position for a period of approximately 2 minutes

and then moves about the diameter of the sample forward towards the center of the paper, with

sample rotated about 90 degrees each time to ensure that scratch lines from the previous are

eliminated.

23
 After a certain period on an abrasive paper which is usually not more than 15 minutes and

with certainty that all the scratches on the sample are of the same size and all parallel to one

another, the mount is then moved to the lower mesh size paper.

The next stage in metallographic sample preparation is polishing, which is evidently

using an agent to polish the surface of the sample so that the roughness is minimized and a high

reflectivity is obtained. The absence of the scratch marks from the grinding step and the

deliverance of high metallic shine on the surface enhances the possibility of obtaining most data

for the microstructure of the sample under optical or electron microscopes.

Here a polishing agent which is usually a suspension of diamond particles is applied to

rotating polishing pad. The agent used could be in form of diamond paste in different mesh sizes.

For rough polishing higher sizes e.g. 30 microns could be a proper starting point and then

applying a paste with a lower size and a finer mesh is required. One path to be taken with respect

to polishing is to adopt agents in different order-of-magnitude of sizes. This means starting from

the highest available mesh size (in our case it was 40 micron Diamond water-solution and

diamond paste provided by Allied High Products, Inc. and Beta Diamonds Products, Inc.) and

going one order lower, 10 micron and then one 1 micron and finally the lowest possible mesh

size which is usually 0.1 micron.

As it was the case with grinding the same machine can be used for polishing though here

we used a polishing pad upon which the polishing agent is applied. The pad is made of soft

material such a felt or nylon to retain the agent while the pad is rotating. In case a diamond paste

is used a stream of paste is injected on a whole radius of the pad and then gently diluted with the

tip of gloved finger. If a diamond suspension is used, the whole area of the pad can be wetted.

24
 A metallographer might choose a rotating speed of 100 rpm to 300 rpm, with sample

rotated 90 degrees and advanced towards the center as it was the case with grinding. The applied

pressure on the mount and the time spent on each mesh size pad is of obvious importance to the

quality of the polished sample. It is imperative after each step and before moving to lower mesh

to wash the sample thoroughly to clean any residue, otherwise there might be some artifacts

emerging on the surface misleading the interpretation of the microstructure. At the end it is also

recommendable to use an even finer agent, such a colloidal silica in mesh sizes around 0.02

microns to minimize any possible surface roughness and to enhance the surface quality for those

surface-sensitive measurements such as electron backscatter diffraction or EBSD. Care should be

taken in choosing a different pad for colloidal silica to avoid damage for remainder diamond

particle to the sample surface. Unlike diamond polish step which usually should take about 15

minutes, silica polish of about 5 minutes is sufficient to provide satisfactory results.

After all these steps and stages, the sample should be void of any major scratches and

manifesting good reflectivity with a metallic shine. At this level a chemical agent can be applied

to the surface of the sample to reveal its morphological features. This step, which is actually the

final step in metallography preparations is called etching. The etchants are mixtures of aqueous

acids with some chemical compounds and are diverse. The proper choice of an etchant is

important with respect to the material etched. The reaction is such that the etchant attacks

different morphologies with different rates, thus emerging them by enhancing a topographic

surface contrast discernable under a microscope. Different etchants have different activities and

so different etching rates. Sometimes finding the best etchants to reveal most features with best

optical quality is itself a complicated project, but metallographers have adopted some

conventional etchants during the course of the past century which can be used as a starting point.

25
 For a thorough listing of different etchants for various alloys consult reference 7. In this

case three etchants with different levels of activity was chosen for the task: Nital 3% (3 mL of

Nitric acid in 100 mL alcohol) which is a low activity etchant intended traditionally for carbon

steels and used in our case for sensitive alloys such as manganese rich ones and CoTiSn.

A midrange activity etchant we adopted was Marble (10 grams of CuSO4, 50 mL HCl

and 50 mL of water) and a higher activity one which was Adler (9 grams of CuNH3Cl, 45 grams

of hydrated FeCl2, 150 mL of HCl and 75 mL of DI Water).

The samples can be immersed in the etchants from seconds to minutes depending on the

rate of etching. After immersion the sample are washed with water for a few seconds and then

with an alcohol. For sensitive samples it would be required to dry the sample immediately after

the previously-mentioned steps using a dryer or a hydrogen source.

The following steps would most of time result in a high quality surface if carefully

performed. The resultant sample would then be ready for optical microscopy. The system used in

our case was a Zeiss Axioplan 2TM optical microscope.

2.3 X-Ray Diffraction Analyses

X-Ray Diffraction, or XRD as it is sometimes abbreviated, is one of the fastest and

simplest methods for characterizing crystalline materials. The method is useful in determining

the crystal structure of the compounds, lattice constants and also detection of impurities and

present phases in general.

The impingement of a monochromatic X-Ray with a certain wavelength on a crystalline

material will result in the electromagnetic ray interacting with the electron cloud of the atoms

within the material and eventually scattered by them.

26
 The scattered wave now has amplitude which is called “the atomic scattering factor”

related to the number of the electrons within the atom and consequently differing from element

to element. In some cases the scattered beams could be in phase in one another resulting in

doubling of amplitude. This phenomenon is called ‘diffraction’.

A crucial factor in occurrence of diffraction is that the scattered waves should have a path

difference related to an integer of the wavelength, λ; otherwise the scattered waves will be out of

phase and so cancel out each other. Therefore, in a crystalline solid as the distance between

atoms or interplanar spacing, d, fixed, there will an incident angle in which the scattered beams

will be coherently in phase to one another giving rise to diffraction. Such a characteristic angle is

called a Bragg angle, and the relationship between the beam wavelength and interplanar spacing

is expressed in the general formula below, called the Bragg law:

nλ=2dSinθ (2.1)

The Bragg law is the cornerstone of X-Ray diffraction analysis. n is the number of beam

wavelengths from different atomic plans and usually is considered to be unit value.8,9

The outcome of an X-Ray diffraction analysis on a crystalline compound will be a spectrum

unique to that material which shows a specific pattern composed of intensity peaks emerging at

different θ angles meeting the Bragg law. Therefore every pure crystalline material has a

‘fingerprint’ of X-Ray diffraction for itself which serves as indispensable tool to detect its

presence in solid system. The interplanar spacing is the main factor to this uniqueness, as it is

related to the lattice constant of material and specific plans upon which the Bragg law is allowed

and therefore the diffraction happens. The relationship between the different d spacing in the

crystal systems is shown at figure 2.7.

27
 Figure 2.7 Interplanar spacing of different crystal systems.
Reprinted with permission from Ref. 8.

It is also noteworthy that as the symmetry decreases from cubic to triclinic (from top to

bottom in Figure 2.7) the number of the diffracted planes and therefore observed peaks in the

XRD spectrum of the material also increases.

Knowing the interplanar spacing formula renders the XRD pattern a useful instrument in

deriving the lattice constant of the material considering the locations of the peaks and therefore

the θ values. But it should be noted that due to the interference-based nature of diffraction, not

every atomic plane will diffract the incident beam. In some cases there are destructively scattered

and canceling each other as a consequence and as a result in certain crystal structures only the

planes meeting the Bragg conditions with a non-zero consequent structure factor are reflected

and observed. The condition for this constructive interference can be calculated for different

Bravais lattices and is summarized in the figure 2.8

28
 Figure 2.8 The planar reflections in specific Bravais lattices.
Reprinted with permission form Ref.8

The discussed conditions will be effective in making the X-Ray diffraction a facile

method in deriving the crystal structure of compounds. As the X-Ray spectrum of each material

is unique, its XRD pattern can be used as a fingerprint in distinguishing it from other possible

candidates.

Nowadays there are computer databases and software available which can immediately

scan and designate a certain XRD spectrum for possible candidates and match the pattern to the

best possible compound(s). One example is the database devised by International Centre for

Diffraction Data or ICDD.

If a new material is synthesized, obviously there are no reports on its XRD signature in

any database, so it is customary to hypothetically assign a crystal structure for the material and

then using that structure as a model, generate a related XRD pattern using simulation software.

29
 The adopted model should be consistent with the chemical composition of the material as

a first step in devising the model.

Therefore it is imperative that the atomic Wyckoff positions of the model have

multiplicity ratios in the line with the chemical stoichiometry of the material otherwise the model

will be erroneous even if the final generated pattern matches well with the experimental data.

The next crucial step is adopting the lattice constant(s) so that the generated pattern is in good

agreement with the actual measurements. If the lattice constants chosen are correct, then peak

positions should be located at almost exact positions seen on the experimental spectrum.

Although the lattice constant can be manually calculated from the observed pattern using

the Bragg law, it is conventional to pinpoint the value to three digits after the point. For this

purpose usually a fit of the hypothetical model to the experimental data is done to ensure the

accuracy of lattice constant estimation. A method to achieve this objective is through fitting

refinements. Most of the refinement methodology regime is based on minimizing the sum of

square differences between the proposed model intensity and that of the actual data, which is also

called the least square method. One such method which is also one of the most frequently

applied is called the ‘Rietveld refinement’.9,10 This refinement is based on minimizing the

expected calculated intensity and the actual observed intensity in the data by refining the position

of atoms, the occupation levels and thermal factors. In fact thousands of intensity positions are

considered for this minimization to occur. The corresponding generalized minimum formula is

such46:

S=  wi ( yi (obs.)  yi (calc.) ) 2  Minimum (2.2)

i

y is the actual intensity in many unlimited positions of 2θ, and w is called weighting factor.

30
 Although the Rietveld refinement is performed by software, it is by no means a method

which can be independently ‘done by the machine’ without any astute control from the operator.

There are many factors to be refined which could themselves are quite complex and sometimes

might render the refinement ambiguous and misleading if not carefully monitored.

The main refinement parameters can be categorized into three main groups: 10 the basic

experimental parameters such as profile parameters, halfwidths and asymmetries of the peaks.

The second group includes crystallographic parameters such as space group, and the third group

encompasses structural parameters such as Debye-Waller factors.

Using the approximate value for each parameter, the software attempts the refinement

until after many refinement cycles a certain amount of convergence is achieved.

The success of convergence is manifested in form of goodness-of-fit parameters such as χ2

factor. A good fit usually has a χ2 in single digits and as near the value of 1 as possible. The

further the factor is from 1, the more the fit becomes unreliable. At the initial stages of the

fitting, this factor could emerge as large but eventually as more runs of the software are done ,

the number reduces to lower value with the convergence, which is indicative of the fit gradually

improving. This factor itself is a function of two other weight factors, weight-profile R factor or

Rwp and expected R factor, Rexp :

χ2 = (Rwp/Rexp) 2 (2.3)

The respective definition of the Rwp is :

Rwp2=  w ( y (obs.)  y (calc.)) /  w ( y (calc.))

i
i i i
2

i
i i
2
(2.4)

Which itself is one of the most statistically accessible indexes for showing deviation from

the ideal model.11 The Rexp is the highest possible attainable Rwp and also another important

statistical factor calculated and presented in the Rietveld literature.

31
 The relevant formula is:

Rexp.=N/  wi ( yi (obs.)) 2 (2.5)

i

With N being the number of the experimental data points in the spectrum.

Although with advent of faster computers in 1980s the usage of Rietveld refinement

software has become common practice as an effective method of evaluating the proposed crystal

models to actual data, the Rietveld method is not considered to be immune to controversy among

some scientific cycles. One such problem is with the fitting parameters and their considerable

vulnerability to the effects of the background signal. There have been reports of auspicious R-

factors achieved even without any valid crystal model. 11 Another issue is the paradox of

lengthier XRD measurement, which will lead to higher data points and higher statistical

precision and consequently lower Rexp values. Simultaneously this situation will also increase

Rwp value, meaning longer counting in the measurement will deteriorate the χ2 value and

therefore will infer a bad fit and an unacceptable crystal model.47 These conditions have resulted

in cases that fitting parameters are of low quality while there are minimal discrepancies between

the proposed model and measured data and vice versa. So in order to avoid the confusion it

would be recommendable to compare as many fitting parameters as possible to arrive at a

conclusion regarding the validity of the fit. Also as mentioned before, it is also imperative to

monitor the decrease in the fitting values during the convergence. In case the value is not

decreasing at the course of refinement there might be a problem to consider. Also if the chi-

square factor is disproportionately higher than 1, the systematic problems with the sample and

the measuring instrumented should also be considered.11

32
 The caveats given above call for an elaborate guideline for XRD measurement and

eventual Rietveld fitting: Data collection should be optimized as a first step, with the measuring

instrument precisely calibrated with a standard sample. 10

Also the background signal should be minimized as much as possible as it has

detrimental effect on the fit: For this goal special sample holders could be adopted or the beam

path could be put under vacuum or a crystal analyzer could be located between the sample and

detector on the beam path. 10 Also the size of the sample should be such that that it lies within an

optimum region of 1 to 5 microns. The larger size particles might negatively affect the intensity

of the peaks while the lower sizes result in peak broadening and deterioration in Full Width at

Half Maximum (FWHM) values.

In fact the larger the grains, the smaller number will be diffracted by the beam, resulting

in the intensity of some peaks becoming uncharacteristically higher than expected. In order to

eradicate this problem it is recommendable to rotate the sample around the axis perpendicular to

the surface of the sample, or the φ axis. This increases the possibility of more grains located in

the beam path and therefore diffracted and is especially required when obtaining data from bulk

metallic materials. 12

A preferred procedure of Rietveld refinement is to obtain the most accurate experimental

data, devise the best possible crystal model and start the fine tuning the relevant parameters in

the refinement software. It is advisable that the parameters are refined consecutively and

gradually in specific steps, for example first the scaling factor should be refined for at least three

cycles, then in following order atomic positions, occupancy levels, background signal, remaining

parameters especially the peak distributions and finally Debye-Waller factors.10

33
 Finally it should be mentioned that Rietveld refinement is not be used as a stand-alone

method to confirm the specific structure within a material.

Further diffractive characterizations such as Transmission Electron Microscopy (TEM)

diffraction patterns and Electron Back Scatter Diffraction (EBSD) phase maps might be needed

to approve the validity of the model and even more importantly are chemical analyses

approaches to evaluate the phase model with respect to the actual composition of the material.

In this set of studies we have used a Bruker D8 DiscoverTM system to obtain our XRD spectra.

The instrument had the capability of rotating the sample around the φ axis to reduce the surface

effects stemming from relatively large grains sizes and preferred orientations. The optimal time

of measurement was found to be a total of 9 minutes for the entire spectrum. Also some

oscillation in the X-Y plane was done to ensure the minimization of the surface effects in the

measurement. The source used was cobalt.

The simulation software utilized to generate the XRD spectrum was CaRine 3.1TM which

incorporated the multiplicity factors, structure factors and atomic scattering factors.

The software chosen to perform the Rietveld refinement was Crystal Impact Match! TM based on

the FullProf program platform.13

2.4. Electron Microscopy Methods

The electron microscopy methods were used in these studies to characterize the

microstructure and its features. The instruments used were Scanning Electron Microscope (SEM)

and Transmission Electron Microscope (TEM). The different setup between the two systems is

shown in figure 2.9.

34
Figure 2.9 (a) The schematic diagram of a scanning electron microscope (b) The schematic of a
transmission electron microscope. Reprinted with permission from ref. 14.

The SEM shares similarities with a conventional confocal optical microscope in the sense

that both are actually mapping instead of direct imaging. 15 But unlike the optical microscope,

the much lower electron wavelength used in the SEM will result in a considerably higher spatial

resolution when the beam is focused. 16 Therefore the SEM chamber is held under vacuum to

avoid the electrons getting scattered by atmospheric gas molecules. In the SEM electrons can be

focused into a small electron probe and scanned through the surface of the sample.

35
 The sample could be simply encapsulated within an ordinary metallographic conductive

mount, which in contrast to TEM makes the sample preparation an advantage in the SEM. The

electron primary incident beam can interact with the sample arising two major electron

phenomena : One is called backscattered electrons (BSE) which are primarily incident electrons

reflected from the sample surface at angles between 90 to 180 degrees with respect to the

incident beam.15,16 Another type of electrons resulting from interaction of the beam with the

specimen is secondary electrons (SE) which have lower energy than BSE and therefore penetrate

the sample weakly in about 3 to 10 nanometers.15 The energy of the BSE lies usually between the

energy of the incident beam and 50 eV while for the SE it could be as near to zero as possible.

More than 90 percent of the images reported in the scientific literature are taken in the SE mode

as it is easy to collect these electrons and more importantly they can generate a topographic

image of the sample surface with resolutions of up to 1 nm. The BSE has the difference of being

highly dependent on the atomic number of the element with which it reacts: The higher the

atomic number transpires more backscattered electrons. The BSE enables a chemical contrast of

the image as a result and is a very useful tool in emerging microstructural features such as grain

boundaries that might be difficult to observe under the usual SE mode.

Besides imaging, SEM is also a useful spectrometry tool for deriving the chemical

composition of the morphologies seen under the microscope: The electrons can also have an

entirely different interaction with specimen material than the two phenomena mentioned above.

This interaction is called ‘characteristic X-Ray’ and as its name implies is an X-Ray generated

from material when the incident electron beam excites an inner shell electron of the material to a

higher energy orbit and the resulting vacancy is immediately filled by an electron from a higher

energy shell.

36
 The energy difference in this de-excitation process is manifested in the form of a photon

which is called a X-Ray characteristic photon and its energy is related to the atomic number of

element scattering the incident electron beam. There are detectors capable of producing spectra

from these X-Ray photons which can differentiate the electrons either based on energy or their

wavelength. In case of Energy Dispersive Spectroscopy or EDS, a dispersive semiconductor

diode serves as the detector in which emission of the photon excites the electrons in the n an p

semiconductors and therefore a specific current proportional to the photon energy is generated.16

The EDS has two main limitations one being the resolution limited to 150 eV and consequently

there might be some cases which two peaks belonging to two different elements cannot be

distinguished, and the other is elements with atomic weight lower than sodium cannot be

detected as the energy excited from these elements are too low to be properly detected.14

Care should be taken so that the adopted beam current does not increase the ‘dead time’ of the

detector over its ‘live time’. The dead time is the time in which the detector is not active while

the live time is the time in which it is collecting the data from the sample.

EDS is an important method in quantifying the elements present in the sample based on

their atomic percentage or weight percentage. Still the accuracy level of detection might not be

as high as more precise methods such as electron probe microanalyzer. An acceptable limit of

accuracy is 10 to 15%. 14

Although it is possible to obtain crystallographic and structural data using transmission

electron microscopy, there has been a novel diffractive method incorporated within the SEM to

achieve the same objectives.

37
 The superiority of the method lies in the fact that unlike TEM which needs elaborate

sample preparation techniques such as Focused Ion Beam (FIB) to prepare a very thin sample for

diffraction analysis, the EBSD needs only the metallographic mount with a well prepared

surface. Another advantage of TEM is that diffractive data can be taken from as many as grains

and as much large areas as possible, in contrast to TEM that is limited to one to three grains as

the FIB sample is usually statistically bound to limited grains sampling.

Alongside EDS, EBSD is a very effective tool for identification of phases in the

microstructure. The end product of an EBSD measurement could be a phase map, showing the

presence of the sought crystal structure within the SEM image in the form of a map. Also it can

generate a figure of crystallographic orientations of the grains present in the microstructure

which is indispensable in exposing the texture and local strain within the material. Both of these

tools can be applied up in wide dimensional range of millimeters to nanometers on the sample.
17,18

The mechanism of EBSD, as it is name implies, is diffraction and therefore similar to the

XRD method with the exception of diffractive source being the electron beam instead of an X-

Ray one. The electron beam slowly scans the surface of the specimen and the resulting

backscattered electrons are collected by charge couple device (CCD) camera which is driven

close to the sample surface. The sample itself is tilted at a high angle of 60 to 80 degrees to

enhance the received signal. The pattern formed from each scanned points are in fact some bands

famously called ‘Kikuchi bands’ which follow the Bragg law and their intensity is complexly

dependent on the structure factors of the diffracted material and therefore a unique signature of

it.

38
 The bands are sent to the computer for proper indexing and are almost immediately

corroborated with the target phase to see if a match with it occurs. 17 A Schematic of an EBSD

setup is shown in figure 2.10.

Figure 2.10 The schematic of main parts of an EBSD setup.

Reprinted with permission from ref.19.

It is imperative that the sample preparation for the EBSD result in a very good surface

quality Otherwise the signal received might be spurious. Usage of final polishing agent with low

mesh such as colloidal silica is highly recommended. Also it is required to minimize the surface

charging as much as possible.19 From a sampling statistics standpoint it is also preferred than the

scanning step size be not more than a tenth of the average grain or feature size, so that enough

data is gathered from the surface. Choosing a sampling area in which a normal distribution of

microstructural features is present would be necessary as well. 18

39
 TEM is a versatile approach to analyze materials at low dimensions. Here high electron

beam acceleration voltages of up to 400 kV can bring about wavelengths near the atomic

resolution and therefore it is possible to form a diffraction pattern from the atomic planes of a

very small sampling area, e.g. precipitates, with high accuracy. 16 As the diffractive angle of

electrons is much lower than that of X-Rays, the Bragg law can be simplified as

rd = λL (2.6)

In which r is the distance of the incident beam from the diffracted beam on the image, d is the

interplanar spacing of the material, λ the electron wavelength and L the camera length which is

defined as the distance from the specimen to the camera or its recording film.

The formula 2.6 basically suggests that diffraction from a polycrystalline sample will

result in certain r or rings with this radius emerging on specific diffracted planes, so by

measuring the radii it is possible to derive the interplanar spacing and vice versa, which is an

important tool in distinguishing the crystal structure of material. In case a single crystal is

present, such as a single grain, the pattern will not be in shape of rings but spots each of which

are reciprocal lattice equivalent of individual atomic planes. 14

TEM samples can be prepared using different techniques such as mechanical, chemical

and vacuum techniques.16 In our set studies we used the focused ion beam or FIB to prepare a

sample. Here using Gallium ions a very thin slice of the material is cut from the surface of the

sample on what is basically an SEM instrument giving the element of accuracy in selecting the

specific microstructural features for analysis. The system was a Tescan LyraTM FIB-SEM

machine. Eventually a FEI QuantaTM tool was the instrument used for TEM studies. The SEM

and EBSD images and data was obtained from a JEOL 7000TM FE-SEM.

40
2.5. Magnetic Measurements

Magnetic behavior of materials is exposed through specific methods which are generally

called magnetometry. The main two requirements for observing this behavior are firstly inducing

an external magnetic field to magnetize the material, and secondly measuring and recording the

produced magnetization behavior. 20 The magnetization field can be produced through three

different sources such as permanent magnets, electromagnets and solenoids. 20

One of the widely used techniques for measuring magnetization and hysteresis loops of

magnetic materials is the famous Vibrating Sample Magnetometry or VSM as it is called. The

core of the technique is based on Farday’s law electromotive force, emf or ε :

   Nd / dt (2.7)

Where N is the number of the coil turns in an area of A, and the flux density in the air

will be B0=φ/A , which in the presence of pulsing field will turn the formula 1.11 into 28

   NAB0 sin t (2.8)

Consequently the VSM instrument has a source of magnetic field, which could be an

electromagnet or occasionally a superconductor. 28 The sample is vibrated within the field with

the help of a mechanical vibrator while a non-magnetic rod holds the sample in the field. The

frequency applied is dependent on the type of the driving force used: If a loudspeaker is applied,

the frequency will be approximately 100 Hz with an amplitude of 0.1mm, while a linear motor

has a frequency lower than 40 Hz with amplitudes higher than 1 mm. 20

The mechanical oscillation of the sample will induce an alternating electromotive force in

the detection coils, which itself has a direct relation to the magnetic moment of the material. The

signal is sent to an amplifier which is only capable of detecting signals at the vibrating

frequency.

41
 It is therefore necessary to ensure the stability of the vibration frequency and the

amplitude alongside any potential vibrations in the pickup coils as well. Figure 2.11 shows a

schematic of a Quantum Design VSM system. 20

Figure 2.11. Schematic of a Quantum Design VSM system. Reprinted from Ref. 19.

The VSM can operate in wide range of temperatures such as from near zero Kelvin up to

1000 Kelvin, which is useful in deriving the magnetic thermal behavior of materials such as

Curie temperature or Rhodes-Wohlfarth model.

The accuracy of the system is good and especially for the small amount of the material on

thin films this accuracy is crucially assistive in achieving a correct reading of the moment.

42
 For our set of studies we used a Quantum Design DynacoolTM system which has a good

accuracy level of measuring moments lower than 10-6 emu and apply fields up to 9 Tesla for a

temperature range of 5 to 400 Kelvin.

2.6. Theoretical Calculations

Although the theoretical calculations in this thesis were performed by our colleagues in

Professor William H. Butler and Christopher Wolverton groups, it would be useful to briefly

describe the methodology adopted.

For electronic bandstructure generation, a density functional theory (DFT) approach was

taken using the VASP software.21,22,23 The software receives an input of space group, atoms,

lattice constant, magnetic moment and k points and energy cutoff to minimize the energy

through elaborate nonrelativistic quantum calculations to finally generate the bandstructure.

The cluster expansion (CE) is an expansion of Ising model in which a binary alloy AxB1-x in

which many possible occupation sites or clusters are permuted for a specific property and for

example the end result could be a ground-state evaluation of possible compounds being

stable.24,25,26

2.7 References

1. http://www.edmund-buehler.de/en/materials-science/arc-melting/compact-arc-melter-
mam-1/

2. R. Pöttgen and D. Johrendt, Intermetallics - Synthesis, Structure, Function. Walter De

Gruyter GmbH, 2014.

3. S.H. Avner , Introduction to physical metallurgy, McGraw-Hill , 1974.

4. Y.Lakhtin, Engineering physical metallurgy, CBS Publishers and distributers, 2005.

5. G.F. Vander Voort, Metallography: Principles and practice, ASM International, 1999.

43
6. http://www.struers.com/Search?q=labopress

7. G.F.Vander Voort, ASM Handbook Volume 9: Metallography and Microstructures, ASM

International, 2004.

8. Y. Waseda, E.Matsubara and K. Shinoda in X-Ray Diffraction Crystallography:

Introduction, Examples and Solved Problems, Springer Verlag, Berlin Heidelberg, 2011.

9. R. Guinebretiere in X-Ray Diffraction by Polycrystalline Materials, Iste Ltd., 2007.

10. G. Will in Powder Diffraction: Rietveld method and Two-Stage method, Springer Verlag,
Berlin Heidelberg, 2006.

11. B.H. Toby, Powder Diffr. 21, 67 (2006).

12. V.E.Buhrke, R. Jenkins and D.K. Smith in A practical guide for sample preparation of
specimens for X-Ray Fluorescence and X-Ray diffraction analysis, Wiley-VCH, 1997.

13. J. Rodriguez-Carvajal, Physica B 192, 55 (1993).

14. A.D. Romig in ASM Hanbook Volume 10 : Materials Characterization, Edited by R.E.
Whan, ASM International , 1986.

15. D.C. Joy in Electron Microscopy : Principles and Fundamentals , Edited by S.

Amelinckx, D. Van Dyck, J. Van Landuyt and G. Van Tendeloo , VCH
Verlaggesellschaft mbH, 1997.

16. R.F. Eagerton in Physical Principles of Electron Microscopy : An Introduction to TEM,

SEM and AEM, Springer Science+Business Media, Inc. 2005.

17. A.W. Wikinson and T.B. Britton, Materials Today, 15, 9 , (2012) .

18. V.Randle, Mater. Charact. , 60 , 9, (2009).

19. http://ebsd.com/ebsd-explained/10-ebsd-explained

20. B.D. Cullity and C.D. Graham in Introduction to Magnetic Materials, 2nd. Ed. , The
Institute of Electrical and Electronics Engineers, Inc, 2009.

21. W. Kohn, A. Becke, and R. Parr, J. Phys. Chem. 100, 12974, (1996).

22. G. Kresse and J. Hafner, Phys. Rev. B 47, 558 (1993).

23. G. Kresse and J. Furthmuller, Phys. Rev. B 54, 11169, (1996).

24. J. M. Sanchez, F. Ducastelle, and D. Gratias, Physica A 128,334 (1984).

44
25. M. Asta, C. Wolverton, D. de Fontaine, and H. Dreysse, Phys.Rev. B 44, 4907 (1991).

26 .C. Wolverton, M. Asta, H. Dreyss´e, and D. de Fontaine, Phys.Rev. B 44, 4.

27.K.H.J. Buschow, P.G. van Engen and R. Jongebreuer, J. Magn. Mag. Mat. , 38 , 1 (1983).

28. J.M.D.Coey in Magnetism and Magnetic Materials, Cambridge University Press, 2009.

45
 CHAPTER 3

SYNTHESIS AND CHARACTERIZATION OF Fe-Ti-Sb INTERMETALLIC COMPOUNDS:

DISCOVERY OF A NEW SLATER-PAULING PHASE

Reproduced in part with permission from [N. Naghibolashrafi, S. Keshavarz, Vinay I. Hegde,A.
Gupta, W. H. Butler,J. Romero, K. Munira, P. LeClair, D. Mazumdar, J. Ma, A. W. Ghosh, and
C. Wolverton, PHYSICAL REVIEW B 93, 104424 (2016). ]. Copyright [2016] American
Physical Society.

3.1 Introduction

The ternary intermetallic Heusler compounds with compositions X2YZ (“full” Heusler)

and XYZ (“half” Heusler) have been a subject of great interest in spintronics owing to the fact

that many of them are predicted to be “half-metallic” ferromagnets.1 The term “half-metallic”

refers to the existence of a gap in the density of states at the Fermi level for one spin channel and

not for the other, leading to the possibility of 100% spin polarization of the electron current.

Additionally, many of these Heusler compounds have relatively high Curie temperatures, raising

the possibility of high polarization at or above room temperature. 1-5

We recently completed a first-principles study of over 500 candidate full- and half-

Heusler compounds, with a view to identify unstudied materials that were predicted to be stable

and half-metallic. After screening candidate materials identified by the first-principles study

from an experimental point of view (e.g., toxicity, fabricability, homogeneity, possibility of

decomposition into more stable binary compounds, etc.) and examining the available literature,

we found a number of promising candidate material.6, 7

46
 One of the most intriguing was Fe2TiSb, which was predicted by the calculations to have

a negative formation energy and to be a (near) half-metal in the L21 structure. Perhaps

surprisingly, a literature survey found no reports of the magnetic and spintronic properties of this

or related compounds. This made an inquiry into this metallurgical system potentially interesting,

particularly as a test case for our methods of identifying new compounds. The study reported

here examines the possibility of synthesizing Fe2TiSb and related compounds in phase-pure

form, and a consequent analysis of their structural and magnetic properties. One of the

interesting insights of our study based on a combination of theory and experiment is that Fe2TiSb

may prefer to form as a naturally layered Heusler compound with layers of the full Heusler,

Fe2TiSb, alternating with layers of the half-Heusler, FeTiSb. The predicted lowest energy phase

with composition Fe1.5TiSb is a nonmagnetic semiconductor, consistent with experimental

observations.

3.2 Electronic Structure of Stoichiometric Heusler Compounds Based on Fe, Ti, and Sb

3.2.1 Full Heusler

Our first-principles survey included 180 prospective full- Heusler compounds in the L21

structure (composition X2YZ) and 378 prospective half-Heusler compounds in the C1b structure

(composition XYZ). The full results of this study and the details on the methods and parameters

of the calculations are available on a website. 8 Among the potentially interesting systems that

emerged from these studies are compounds of Fe, Ti, and Sb. Considering first the hypothetical

Fe2TiSb compound in the L21 structure, we found it to be stable against decomposition into its

constituent elements with an energy of formation of −0.264 eV/atom and a lattice constant of a =

0.604 nm. We also found it to be stable with respect to tetragonal distortion.

47
 The most stable magnetic configuration for this phase (considering ferromagnetic,

ferrimagnetic, antiferromagnetic, and nonmagnetic configurations) is a ferrimagnetic ground

state with saturation magnetization of approximately 1.00 µB per formula unit at T = 0 K. The

magnetization is primarily on the Fe sites (0.69 µB) but with a significant opposite (−0.31µB)

moment on the Ti and a small (−0.03µB) moment on the Sb. The quoted moments are those

within spheres of radius 0.145nm centered on each atom. The value of the calculated total

moment (∼1.00 µB) causes one to suspect that this solution to the DFT equations is a “Slater-

Pauling phase.”

We use the term “Slater-Pauling phase” to describe full and half-Heusler compounds that

satisfy a generalization of the rule first enunciated by Slater and Pauling, which states that

compounds based on the bcc crystal structure tend to have three electrons per atom in the

minority-spin channel.9-11,12-14 For the Heusler half-metals, the generalization of the Slater-

Pauling rule is that there tends to be a gap in the density of states for one of the spin-channels

with precisely three valence electron states per atom in the gapped channel below the gap. When

there is a gap like this and when the Fermi energy falls in the gap, we will refer to the phase as a

Slater-Pauling half metal. Often, there is a gap with three electrons per atom, but the Fermi

energy falls just above or just below the gap. We refer to such phases as “Slater-Pauling near

half-metals.” Following this terminology, our DFT calculations predict L21 Fe2TiSb to be a

Slater-Pauling near half-metal. It has a gap in the minority spin channel of 0.31 eV, as shown in

figure 3.1 The Fermi energy falls just below (0.05 eV) the edge of the gap so that the number of

minority electrons is calculated to be 11.99915 rather than 12.

48
 Since the total number of electrons per formula unit is 25, there must be 13.00085

majority spin electrons yielding a moment of 1.0017 µB per formula unit. A moment of exactly

1.0 µB would have produced a Slater-Pauling half-metal.

Figure 3.1. Calculated density of states for L21 Fe2TiSb.

3.2.2 Half-Heusler

Subsequent to our survey of full-Heusler compounds, we completed a survey of the half-

Heusler (C1b) compounds. We found that C1b phase FeTiSb is also predicted to be stable relative

to decomposition into its constituent elements and tetragonal distortions of its cell, with an

energy of formation of −0.382 eV/atom and a lattice constant of 0.595 nm. The density of states

of FeTiSb is shown in figure 3.2 the lowest-energy magnetic configuration is predicted to be a

Slater Pauling near half-metal with a magnetic moment of approximately 1µB per formula unit.

FeTiSb actually has Slater-Pauling gaps in both spin channels. One, however, is quite close to

the Fermi energy.

49
 The calculated Fermi energy is 0.11 eV below the gap in the majority channel, so that

only 8.9775 states are occupied rather than the nine that would have been occupied had the

Fermi energy fallen in the majority gap. FeTiSb has 17 valence electrons, so subtracting

The 8.9775 majority electrons yield 8.0225 minority electrons for a predicted moment per

formula unit of 0.955 µB. This system is also predicted to be ferrimagnetic with moments of 1.3

µB on the Fe and −0.35 µB on the Ti. Half-Heuslers typically have larger gaps than full-Heuslers.

The calculated width of the Slater-Pauling gap is 0.85 eV.

Figure 3.2 Calculated density of states for C1b FeTiSb.

3.3 Experimental

Based on the initial theoretical findings, we attempted to synthesize bulk samples of both

the full Heusler Fe2TiSb and the half-Heusler FeTiSb. Stoichiometric ratios of pure elements Fe

(Alfa Aesar, 99.98%), Ti (Alfa Aesar, 99.9%), and Sb (Alfa Aesar, 99.999%) were mixed

together and melted using an Edmund Buehler Mini MAM-1 compact arc melting system under

an ultrahigh purity Argon (99.999%) pressure of 0.02 Pa. A 10% excess of Sb was added to the

initial melt to compensate for losses in the final ingot due to its higher volatility.

50
 A small piece of Ti was melted before all arc melting runs as an oxygen getter to avoid

oxidization of the sample. Ingots were remelted at least five times to ensure the homogeneity of

the material.

The samples from the arc-melter were cut and mounted for metallographic studies. A

JEOL 7000TM Field emission scanning electron microscope (SEM) was used for energy-

dispersive x-ray spectroscopy (EDS) analysis to ensure that the chemical composition of the as-

cast ingots had not deviated from the intended target stoichiometry. Samples for heat treatment

were wrapped in Ta foil and sealed under vacuum in quartz tubes to avoid oxidation. These

pieces were heat treated at various temperatures (400–1100◦C) and dwell times (1–15 days).

From metallographic and x-ray diffraction analysis of the annealed samples, a treatment at 900

°C for 7 days and cooled in the furnace at approximately 70 degrees per hour appeared to result

in the optimum L21 structure and granular microstructure for all cases.

The crystal structures of the annealed samples were studied using a Bruker D8

DiscoverTM X-Ray Diffraction (XRD) system with a monochromatic Co Kα radiation. In order

to minimize surface effects, the samples were mounted and polished, and rotated around the ϕ

axis during the XRD measurement. Electron backscatter diffraction (EBSD) phase mapping

using the JEOL 7000TM FE-SEM system was additionally performed to affirm the structure

determined from XRD. Simulated XRD patterns were obtained both by using the CARINE

crystallography 4.0TM software as well as by direct calculation to account for dispersive

corrections to the atomic scattering factors while the Rietveld refinement was carried out using

the CRYSTAL IMPACT MATCH! TM Software based on the FULLPROF algorithm.15-17

51
 The magnetic properties were measured using Quantum Design Physical Property

Measurement System (PPMS) Dynacool with vibrating sample magnetometry (VSM) feature.

The Transmission Electron Microscopy (TEM) diffraction pattern was obtained by using a FEI

Tecnai F20TM instrument.

3.4 Results and Discussion

3.4.1 Fe2TiSb

All of the as-cast Fe2TiSb samples heat-treated at 900 °C for soaking times of 3, 5, 7, or 9

days showed a single, mostly granular microstructure. Energy-dispersive x-ray spectroscopy

showed the composition of the grains to be Fe40Ti30Sb30, with an uncertainty of ∼5% in the

abundance of each element, and thus quite significantly different from the intended

stoichiometry - in spite of the fact that the as-cast samples were essentially stoichiometric

Fe2TiSb. Henceforth, we will refer to this nominal composition as Fe1.5TiSb for simplicity.

Figure 3.3 shows an example of such microstructure and its relevant EDS result for a sample

heat treated at 900°C for 7 days. X-ray diffraction results on these samples, discussed further

below, were consistent with a L21 phase alongside a small amount of a Fe-rich precipitate phase.

Figure 3.3 (a) The metallography image of a Fe2TiSb sample heat treated at 900 °C for 7 days.
The sample was etched for 30 seconds using the marble etchant. The microstructure is evidently
single grain, while the voids can be attributed to a Fe-rich precipitate at the grain boundaries,
which was removed by the marble etchant. (b) SEM micrograph of the same surface showing the
grain structure. EDS indicated that the composition of the grains was Fe40Ti30Sb30.

52
 Since the stoichiometry of the grains deviated significantly from the initial as-cast

composition of Fe2TiSb, we hypothesized that the bulk of the material was crystallized in the

form of Fe1.5TiSb grains, while the remaining iron precipitated in the grain boundaries and

disappeared at the etching stage (used to prepare the specimens for metallographic imaging). In

order to test this hypothesis, we performed an EDS analysis of nonetched samples, and indeed

found an overall composition of Fe2TiSb, consistent with the as-cast material and indicating a

loss of iron after metallographic etching. X-ray diffraction analysis of the pre-etched material

also indicates the presence of peaks corresponding to an iron-rich Fe0.975Ti0.025 bcc solid-solution

in addition to the full-Heusler L21 phase peaks originating from the Fe1.5TiSb grains. The etched

sample, on the other hand, showed only a very small (∼1%–2%) presence of the iron rich phase.

Figure 3.4(a) shows the XRD patterns of etched and polished samples measured with Co Kα

radiation. The XRD patterns generated by either CARINE 4.0 simulations or CRYSTAL

IMPACT MATCH are similar to the pattern observed. Rietveld refinement gives good

agreement (χ2 = 0.9 and average RBragg = 0.1) between the data and the calculated pattern for the

L21 structure of the dominant phase with the experimental lattice constant of 0.5967 nm. One can

also clearly see small peaks corresponding to a Fe-rich phase identified as Fe0.975Ti0.025 (e.g., at

approximately ∼52°), which are much clearer in the pre-etched samples. For the pre-etched

sample, we estimate from Rietveld refinement and EBSD that the Fe-rich phase constitutes about

∼10% of the total, and approximately the same amount of bcc Fe for the etched samples was

estimated from magnetometry.

53
 Figure 3.4 (a) The XRD pattern measured using Co Kα radiation of the Fe2TiSb sample heat
treated at 900 °C for 7 days. The pattern is consistent with an L21 structure with a lattice
parameter of 0.5957 nm for the Fe1.5TiSb grains. The stars indicate peaks attributed to a Fe-rich
precipitate. (b) The lattice value is also confirmed by a Rietveld refinement analysis with good
agreement with the proposed model.

The presence of all low-angle L21 peaks, i.e., (111), (200), and (220) effectively rules out

the possibility of significant deviation toward a B2 or A2 disorder. If the (111) peak were

missing or reduced, this would have indicated Ti-Sb mixing and generated a B2 structure,

whereas a missing or reduced (200) peak would have indicated Fe-(Ti, Sb) mixing and generated

the A2 (bcc) structure. 18-20 In fact, the (111) and (200) superstructure peaks are somewhat higher

in intensity than expected compared to the (220) fundamental peak. The higher than expected

values of intensities of the superlattice peaks could be attributed to surface effects and texture or

they could be indicative of the fact that the Fe1.5TiSb grains are not stoichiometric for a L21

structure. If we presume for the moment that we do not have a preferred texture in our samples,

the trend of the x-ray diffraction intensities with Fe content can be readily seen.

54
 If we consider a Fe2TiSb compound in the L21 structure, with Fe at the 8c (1/4, 1/4, 1/4)

Wyckoff positions, the structure factors for the main peaks of interest are: 18-20

F111 = 4|fTi − fSb|, (3.1)

F200 = 4|fTi + fSb − 2fFe|, (3.2)

F220 = 4|fTi + fSb + 2fFe|, (3.3)

Where fi is the total atomic scattering factor for element i. From this we can see that

having less Fe than expected would reduce the (220) x-ray structure factor and increase the (200)

structure factor (leaving the (111) structure factor unchanged), consistent with the XRD results.

A more quantitative analysis, including the dispersive terms in the atomic scattering factors does

give a 111/200 intensity ratio similar to that observed using a nominal Fe content of 1.4–1.6

rather than 2.0.15,16 However, we stress that we cannot rule out a net preferred texture in our

samples that would alter the relative XRD intensities. Indeed, this is likely the case given that the

(220) peak intensity is still smaller than expected compared to the (111) and (200) peaks. Still,

with the assumption that we have no preferred texture, both qualitatively and quantitatively the

XRD intensities do appear to be consistent with a Fe-deficient L21 Fe1.5TiSb compound.

In order to further substantiate the existence of the L21 structure, an EBSD phase map was

performed. The phase mapping showing the L21 structure is shown in figure 3.5 More than 90%

of the microstructure is colored red which means that the material corresponding to an L21 phase

with a lattice constant of 0.5967 nm is homogenous throughout the sample. This is further

evidence that the grains are crystallizing in a cubic L21 structure. The remaining 10% of the

sample is attributed to the Fe-rich phase precipitated at the grain boundaries, which are not

evident in the EBSD maps.

55
 The fact that the dominant phase has a composition of approximately Fe1.5TiSb rather

than Fe2TiSb expected for ideal L21 compound points toward possible Fe vacancies.

Figure 3.5 (a) The electron micrograph of the as-cast Fe2TiSb sample heat treated at 900 °C for
7days taken by EBSD (b) the phase color map scan shows 90% (red color) of the sample has a
homogenous L21 structure with a lattice constant of 0.5967 nm.

Magnetic characterization of the as-cast Fe2TiSb sample heat treated for 7 days at 900 °C

shows the sample has a ferromagnetic component with a small hysteresis, which does not

completely saturate for the magnetic fields of up to 7200 kA/m (9 T) at 5 K, figure 2.6(a). This

behavior is suggestive of small ferromagnetic Fe-rich precipitates in a weakly paramagnetic

matrix. We find the magnetization to be 27 Am2/kg at the maximum applied field of 7200 kA/m,

with a coercivity of 2 kA/m. The ferromagnetic component would correspond to a very low

magnetization for the dominant phase, but does agree well with ∼10% of the sample having

approximately the same magnetization as bcc Fe (or a very Fe-rich phase). We hypothesize,

therefore, that the magnetic behavior comes from a superposition of a small volume fraction

(∼10%) of strongly ferromagnetic Fe-rich precipitates at the Fe1.5TiSb grain boundaries, and a

large volume fraction of Fe1.5TiSb grains that are weakly paramagnetic.

56
 Temperature-dependent magnetization [Figure 3.6(b)], measured at an applied ( field of

H = 80 kA/m during heating, is relatively uniform up to 400 K, consistent with most of the Fe-

rich precipitate material being strongly ferromagnetic. The exception is a small upturn in the

magnetization for temperatures below 10 K, which again could originate from weakly

paramagnetic Fe1.5TiSb.

It would be unusual for a bulk sample to exhibit such a large Curie temperature

(apparently above 400 K) with such a small moment (approximately 10% the mass

magnetization of Fe). This again suggests that the observed ferromagnetic component comes not

from the bulk, but from the Fe-rich (nominally Fe0.975Ti0.025 from the observed XRD peaks)

Precipitate material, from which we would expect a high Curie temperature and low coercivity.

Interestingly, first-principles Calculations (discussed later on) for Fe1.5TiSb, close to the

observed composition for the dominant phase, indicate that a nonferromagnetic ground state is

indeed favored.

Figure 3.6 (a) Magnetization curve of a nominally Fe2TiSb sample at 5 K shows the dominant
ferromagnetic nature of the sample. The inset is the magnetization around the origin, showing the
small coercive field from the Fe-rich precipitates. (b) Temperature dependent magnetization
shows no sign of phase transformation or Curie temperature below 400 K. The inset is the
magnetization curve at 5, 300, and 400Kshowing only a minor change in the magnetization from
5 to 400 K.

57
3.4.2 Fe1.5TiSb

Our observations during the synthesis of Fe2TiSb suggested that Fe1.5TiSb is, in fact, a

more stable phase, a result corroborated by first principles calculations discussed later. Given

this, the Fe1.5TiSb compound seemed worthy of direct synthesis. As-cast samples were prepared

in the same fashion as Fe2TiSb. After the Fe1.5TiSb samples were heat treated at 900 °C for 7

days, a single grain microstructure was observed. There were no signs of precipitates.

Figure 3.7 shows optical and SEM micrographs. Energy-dispersive x-ray spectroscopy

shows that the grains present in the sample have a nominal stoichiometry of Fe1.44TiSb1.12.

Again, for simplicity, we will refer to this nominal composition as Fe1.5TiSb, and it appears that

this particular phase is metallurgically stable in a wide range of iron doping within the Fe1+xTiSb

alloy system. The XRD data, figure 3.8, indicate the presence of a phase in an L21 structure with

a lattice parameter of 0.5967 nm, which is in concordance with our observations for Fe2TiSb.

This shows that there is a possible limit of solubility for iron at the Fe1.5TiSb composition, which

is evidently the stable phase, and additional iron will be precipitated in the grain boundaries of

the dominant L21 phase.

Figure 3.7 (a) An optical micrograph of an as-cast Fe1.5TiSb sample heat treated at
900 °C for 7 days. The sample was etched for 30 seconds using marble etchant. The
microstructure reveals single grains without the presence of any secondary phase in the grain
boundaries. (b) An SEM micrograph of the same surface showing the grain structure. EDS
indicated that the composition of the grains was Fe1.44TiSb1.12.

58
 Figure 3.8 The XRD pattern measured using Co Kα radiation of the Fe1.5TiSb compound. All
peak positions can be attributed to an L21 structure with a lattice constant of 0.5967 nm.

To perform further comparison between the diffraction pattern of Fe2TiSb and Fe1.5TiSb

and verify the crystal structure, selected area electron diffraction (SAED) analysis using a FEI

Tecnai F-20 Transmission Electron Microscope (TEM) was performed. Results are shown in

figure 3.9. The diffraction of the Fe2TiSb sample showed patterns belonging to both bcc α-Fe

and L21 phases, which is in accordance with our findings suggesting Fe precipitation in this

compound.

59
 Figure 3.9 (a) A SAED ring pattern of Fe2TiSb sample after heat
Treatment at 900 °C for 7 days. Both bcc Fe and L21 patterns are
Visible. (b) A SAED pattern of Fe1.5TiSb heat treated at 900 °C for
7 days. There are spots visible belonging to the assumed L21 structure,
but none corresponding to bcc Fe.

The pattern is noticeably different for the Fe1.5TiSb sample - the Fe diffraction spots are

no longer present, while the L21 pattern is still clearly visible. It is notable that the pattern

belonging exclusively to the L21 phase has faded, which could be attributed to the presence of Fe

vacancies in the L21 lattice, which would also be consistent with Fe precipitation.

The magnetic properties of the Fe1.5TiSb material are qualitatively similar to those of the

Fe2TiSb material, with a smaller amount of the Fe-rich phase (∼1.5%). Magnetization versus

field characteristics at T = 5 K, figure 3.10, again suggests paramagnetic behavior of the

dominant L21 phase plus weak ferromagnetism from the small concentration of Fe-rich grains

segregated at the grain boundaries. Coercivity in the low-temperature regime is approximately 2

kA/m for Fe2TiSb sample, as shown in the inset of figure 3.10(a). Magnetic measurements up to

7200 kA/m for this sample shows no sign of saturation of the magnetization at T = 5 K, whereas

at T = 300 and 400 K the magnetization is nearly saturated with only 100 kA/m. Figure 3.10(b)

shows the temperature-dependent magnetization, measured in similar conditions as the Fe2TiSb

sample.

60
 As in that case, the temperature dependence appears to be a superposition of weak

paramagnetism of the dominant Fe1.5TiSb phase and a small ferromagnetic component due to the

Fe-rich segregations at the grain boundaries. The magnetization does not change significantly

over the whole temperature range, aside from the upturn in temperatures below 10 K.

Figure 3.10 (a) The magnetization of Fe1.5TiSb shows no indication of saturation for the fields
up to 7200 kA/m at T = 5 K, showing the paramagnetic nature of the dominant phase. The inset
shows small coercivity related to Fe-rich phase precipitates. (b) Magnetization of the sample as a
function of temperature measured in an applied field of H = 80 kA/m during heating. The inset
shows the magnetization curves measured at temperatures 5, 300, and 400 K.

Due to the smaller concentration of Fe-rich precipitates in the Fe1.5TiSb sample

(estimated as ∼1.5%), the maximum magnetization measured is much smaller than for the

Fe2TiSb sample [Figure 3.10(b) inset].

3.4.3 FeTiSb

Finally, the half-Heusler FeTiSb composition was expected from our theoretical survey to

also have a negative energy of formation and to be a near half-metallic compound. As-cast

samples of nominal composition FeTiSb were prepared and heat treated in a similar fashion to

the previous samples.

61
 Microstructural analysis on samples annealed to 900 °C for 7 days shows that the

material in fact decomposes into two separate phases. One is again the iron rich Fe1.5TiSb phase,

but with a significant presence of a Ti-Sb rich phase.

Figure 3.11 displays optical and SEM micrographs. The composition of the Ti-Sb rich

phase was found by EDS to be approximately Sb1.8Ti1.5Fe. While once again the dominant phase

is Fe1.5TiSb, in previous cases it was excess Fe, which segregated to the grain boundaries. In

contrast, in this case, excess Ti and Sb have formed a new granular phase, indicating that the

level of solubility of the dominant phase for Ti and Sb might be less than its solubility level for

Fe. Figure 3.12 shows XRD pattern of the sample. It clearly indicates the presence of the L21

phase again, which originates from the Fe1.5TiSb grains, as well as small yet-unidentified peaks

at low angles, which are presumably attributable to the Ti-Sb rich grains. The relative peak

intensities in this case are probably a reflection of an induced texture in the sample.

Figure 3.11 (a) An optical metallography image of FeTiSb exhibiting two distinct granular
phases present. (b) Shows the SEM micrograph of an iron poor grain, whose composition was
found to be Sb1.8Ti1.5Fe. (c) Shows an iron-rich grain, whose composition was found to be
Fe1.5TiSb.

Magnetization versus field of the FeTiSb sample, illustrated in figure 3.13, has a similar

behavior to Fe1.5TiSb. We again attribute the paramagnetic response to the Fe1.5TiSb phase, but

in this case the contribution we attribute to the precipitates is different.

62
Figure 3.12 The XRD pattern measured using Co-Kα radiation of the as-cast FeTiSb compound.
Most peak positions can be attributed to an L21 structure while there are some yet-unidentified
peaks possibly arising from the Ti-Sb rich phase.

Figure 3.13 (a) Magnetization curves for FexTiSb (x = 1, 1.5, 2)

Specimens. (b) Temperature-dependent magnetization of these samples
Measured in the applied field of H = 80 kA/m during the heating.

63
 Comparing the magnetization results from FexTiSb (x = 1, 1.5, 2) samples, the common

presence of the stable and weakly paramagnetic Fe1.5TiSb phase is apparent, with a varying

ferromagnetic contribution of Fe-rich precipitates for Fe2TiSb and Fe1.5TiSb samples and TiSb-

rich phase for the case of FeTiSb. Magnetization versus temperature (Figure 3.13) gives a similar

conclusion: ferromagnetic precipitates give a slowly-varying contribution to the magnetization.

In order to verify the nature of magnetization versus temperature upturn for T <10 K, field-

cooling (FC) and zero-field-cooling (ZFC) measurements were performed to check for the

presence of superparamagnetism in the samples, which would help determine if observed

paramagnetism is indeed due to the dominant Fe1.5TiSb grains or small precipitates. The

outcome, shown in figure 3.14 does not show the expected bifurcation for superparamagnetic

effects for either the Fe2TiSb or Fe1.5TiSb samples. Therefore the main source of the increased

magnetization for T <10 K, as well as the nonsaturated magnetization versus field at low

temperatures, is most likely due to weak paramagnetism of the dominant phase Fe1.5TiSb.

64
 Figure 3.14 Field-cooling and zero field cooling of Fe1.5TiSb and Fe2TiSb samples confirm the
absence of superparamagnetism in these samples.

Overall, one concludes at this point that regardless of the starting composition, it is the

Fe1.5TiSb phase that is most stable, and it appears to dominate after heat treatment. This phase

crystallizes in the L21 structure, albeit one with iron sites which are approximately 2/3 to 3/4

occupied, and seems to exhibit weak paramagnetism. Starting from a Fe-rich composition leads

to segregation of excess Fe to grain boundaries of the Fe1.5TiSb grains, while starting from an

Fe-poor composition leads to a secondary Ti-Sb-rich phase appearing. The main question to be

addressed from a theoretical point of view is whether the greater stability of the Fe1.5TiSb phase

compared to FeTiSb or Fe2TiSb can be explained, whether its existence in a (defective) L21 or

related structure can be explained, and whether one expects it to be nonmagnetic.

65
3.5 Theoretical Cluster Expansion Treatment of the Fe-Ti-Sb Alloy System

The experimental work described in previous section makes clear that the lowest energy

structure in the Fe-Ti-Sb system is based on the L21 structure. It is also clear that it is not simply

a Fe2TiSb full-Heusler compound, and specifically, it is Fe deficient. Given this experimental

input, we performed a binary cluster expansion (CE) of Fe and vacancies (hereafter, a vacancy is

indicated as □) on the Fe sublattice of the Fe2TiSb L21 structure in an attempt to explain the

predisposition of the Fe-Ti-Sb to tolerate Fe deficiency.

We calculated the formation energy ΔHf (per mixing site) of a compound with

composition Fex□2-xTiSb with respect to the end members Fe2TiSb and□2TiSb using:

ΔHf = 1/2[E(Fex□2-xTiSb)−x E(Fe2TiSb) − (2 − x)E(□2TiSb)] (3.4)

Where all the required energies are calculated using density functional theory (DFT). We

fit a binary CE to the formation energies calculated using equation 3.4 by minimizing the sum of

squares of the residuals. We generated new structures, used their formation energies in the fit,

and added terms to the CE until the cross-validation (CV) score was minimized, and adding

further structures resulted in overfitting. We found a well-converged CE, with a CV score of ∼

20 meV/mixing atom by using 66 Fex□2-xTiSb structures and their formation energies in our fit

(Figure 3.15). We searched for ground states by computationally scanning through >103

structures, which corresponds to a complete enumeration of all possible Fe/□ orderings with <20

atoms per cell.21 Out of this extremely large configuration space, we predict a novel, stable

ground state phase with a composition Fe1.5TiSb, i.e., between the full-Heusler and half-Heusler

compositions, belonging to the trigonal R3m space group, and with a formation energy of −0.223

eV/mixing atom.

66
Figure 3.15 Energies of 66 different arrangements of Fe and □ on the Fe sublattice of Fe2TiSb in
the L21 structure, calculated using DFT, and those predicted by a binary cluster expansion (CE)
fit to the DFT energies. The DFT convex hull is captured well by the CE, with a cross-validation
error of ∼20 meV/mixing atom.

This structure corresponds to a layering of the L21 and C1b structures along the [111]

direction of the underlying cubic lattices (see Fig. 3.16). We provide the lattice parameters and

atomic coordinates of the Fe1.5TiSb phase in Ref. [22].

We calculated the density of electronic states in the ground state R3m structure and find

that it is a nonmagnetic semiconductor with a Kohn-Sham gap of ∼0.64 eV (see figure 3.17).

67
Figure 3.16. Structure of the low energy R3m Fe1.5TiSb phase obtained from a cluster expansion
illustrating the layering of the L21 Fe2TiSb and C1b FeTiSb along the [111] direction of the
underlying cubic lattices. A (111) “defect planes” formed by vacancies is highlighted. Only a
part of the periodic cell is shown here for clarity. (Sb = green, Ti = blue, Fe = red, and
vacancies= unfilled spheres.)

Figure 3.17 The density of electronic states of the low-energy

Fe1.5TiSb R3m structure predicted by the cluster expansion. It is
a nonmagnetic semiconductor with gaps in both the spin channels.

68
 We observe that gaps open in both the spin channels and the Fermi energy is in the gap.

As discussed in section 2.2, the full-Heusler Fe2TiSb compound is a Slater-Pauling near half-

metal with 25 electrons, a moment of ∼1 Bohr magneton per formula unit (µB/f.u.), and a gap in

the minority spin channel (see figure 3.1). The half-Heusler FeTiSb compound is also a Slater-

Pauling near half-metal with 17 electrons, a moment of ∼1 µB/f.u. but with a gap in the majority

spin channel (see figure 3.2). When the full-Heusler and the half-Heusler are layered to form the

Fe1.5TiSb compound, there are three electrons per atom in each spin channel, and therefore a

Slater-Pauling gap can exist in both spin channels. This gap in both spin channels is precisely

what we observe in the Fe1.5TiSb compound (see figure 3.17), consistent with Slater-Pauling

behavior of these compounds. And this opening of gaps in both spin channels seems to

contribute to the stability of the R3m structure. Thus, in full agreement with experimental data,

our calculations demonstrate the existence of a Fe1.5TiSb phase, that is, (a) stable relative to

Fe2TiSb and FeTiSb, (b) an ordered Fe/□arrangement based on the Heusler structure, and (c) is a

nonmagnetic semiconductor.

3.6 Conclusion

We investigated the structural and magnetic properties of the FexTiSb material system,

including the half-Heusler FeTiSb (x = 1) and the full-Heusler Fe2TiSb (x = 2) compositions as

well as the intermediate Fe1.5TiSb composition. In all cases, after heat treatment a dominant

Fe1.5TiSb phase was revealed. Heat treatment of as-cast Fe2TiSb results in Fe1.5TiSb grains with

approximately 10% by volume of an Fe-rich precipitate.

69
 The Fe-rich precipitate is the main contributor to the magnetic behavior, with the

Fe1.5TiSb phase appearing to be weakly paramagnetic. Heat treatment of as-cast FeTiSb also

results in the dominant Fe1.5TiSb phase, but with a Ti-Sb-rich precipitate of nominal composition

Sb1.8Ti1.5Fe. The cluster expansion results show that the FeTiSb half-Heusler is unstable with

respect to two-phase decomposition into Fe1.5TiSb and a Fe-deficient phase, Fe0.5TiSb. We note

that this Fe-deficient composition is similar to the Sb1.8Ti1.5Fe phase found experimentally.

Future experiments are needed to clarify the structure and properties of this phase. Direct

synthesis of Fe1.5TiSb resulted in a single phase material with only ∼1.5% precipitates, in

accordance with our initial findings and suggesting that it is the dominant phase within the entire

metallurgical system. Directly synthesized Fe1.5TiSb was also observed to be weakly

paramagnetic. The crystal structure of the Fe1.5TiSb material is related to the cubic L21 structure,

with Fe in the 8c Wyckoff position and Ti and Sb in 4a and 4b positions, respectively. The main

difference in contrast to the postulated full Heusler compound is the overall composition of

Fe1.5TiSb, from which we conclude that in the 16 atom L21 fcc cell there should be two Fe

vacancies.

Our cluster expansion calculations revealed that Fe1.5TiSb is a stable ground-state

composition in the FexTiSb system, and is expected to be a nonmagnetic Slater-Pauling

semiconductor.

We predict the stable, zero-temperature low-energy phase to have a new, interesting

structure type, with Fe□ ordering on the 8c Wyckoff sites, occurring in the R3m space group.

The special stability of this structure appears to come from the formation of a gap at the Fermi

energy for both spin channels, a mechanism which should apply to other Heusler systems.

70
 While experimental studies are as yet unable to draw unambiguous conclusions about the

positions of the Fe vacancies to confirm the predicted R3mstructure, the observed Fe1.5TiSb

composition coupled with the nonmagnetic behavior is consistent with the hypothesized layered

structure.

From the XRD data, peaks which would distinguish between a defective L21 structure

and an R3m structure are predicted to be of low intensity, making unambiguous identification of

the R3m structure difficult. While the R3m structure is predicted to be the lowest energy phase, it

is also possible that kinetics on the experimental time scale are insufficient to obtain a fully

ordered R3m structure, and the structure experimentally observed may be a disordered state

somewhere between an R3m and a defective L21 structure. Further theoretical studies evaluating

the free energy and configurational contribution to the entropy would help clarify the situation,

as would more detailed structural characterization combined with extended heat treatment.

Concerning the nonmagnetic ground state, recall that Fe2TiSb has a predicted Slater-Pauling

state with a moment of +1µB per formula unit, while FeTiSb has a predicted Slater-Pauling state

with a moment of −1 µB per formula unit. The Fe1.5TiSb material has zero moment—in fact there

is no moment on any atom in the calculated phases—a bit peculiar for Fe. Locally, however,

there will be fluctuations in the concentrations of the Fe vacancies, and with these will be

fluctuations in the local magnetic moments. We speculate that at finite temperature, these

moments are likely to average to zero, but may also be partly responsible for the upturn in

magnetization at low temperature. Further theoretical work on disordered structures would help

clarify the magnetic state.

71
3.7 References

1. T. Graf and C. Felser, in Spintronics: From Materials to Devices, edited by C. Felser and G.
H. Fecher ,Springer, Dordrecht, 2013.
2. C. Felser, G. H. Fecher, and B. Balke, Angew. Chem. Int. Ed. 46, 668 (2007).

3. B. Balke, S. Wurmehl, G. H. Fecher, C. Felser, and J. K𝑢̈ bler, Sci. Technol. Adv. Mater. 9,
014102 (2008).
4. D. Serrate, J. De Teresa, and M. Ibarra, J. Phys. Condens.Matter 19, 023201 (2007).

5. H. Kandpal, V. Ksenofontov, M. Wojcik, R. Seshadri, and C. Felser, J. Phys. D: Appl. Phys.

40, 1587 (2007).
6. J. E. Saal, S. Kirklin, M. Aykol, B. Meredig, and C.Wolverton, J. Miner. Metals. Mater. Soc.

65, 1501 (2013).

7. S. Kirklin, J. E. Saal, B. Meredig, A. Thompson, J.W. Doak, M. Aykol, S. Ruhl, and C.

Wolverton (unpublished).
8. W. H. Butler, A. Ghosh, K. Munira, Tim Lovorn, J. Romero, C. Lample, L. Wilson, J. Ma, Y.
Xie, S. Keshavarz, and D. Mildebrath, Heusler Database, http://heusleralloys.mint.ua.edu.
9. I. Galanaki, Ph. Mavropoulos, and P. H. Dedericks, J. Phys. D: Appl. Phys. 39, 765 (2006).

10. I. Galanakis, E. S¸ as¸ıog˘lu, and K. O¨ zdog˘an, Phys. Rev. B 77,214417 (2008).

11. S. Skaftouros, K. O¨ zdog˘an, E. S¸ as¸ıog˘lu, and I. Galanakis, Phys. Rev. B 87, 024420
(2013).
12. J. C. Slater, J. Appl. Phys. 8, 385 (1937).

13. L. Pauling, Phys. Rev. 54, 899 (1938).

14. R. M. Bozorth, Phys. Rev. 79, 887 (1950).

15. P. Brown, A. Fox, E. Maslen, M. O’Keefe, and B. Willis, in International Tables for
Crystallography Volume C: Mathematical, Physical and Chemical tables, edited by E. Prince
(Kluwer, Dordrecht, 2006), Chap. 6.1, pp. 554–595.

72
16. U. Arndt, D. Creagh, R. Deslattes, J. Hubbell, P. Indelicato, E. Kessler, Jr., and E. Lindroth,
in International Tables for Crystallography Volume C: Mathematical, Physical and Chemical
tables, edited by E. Prince (Kluwer, Dordrecht, 2006), Chap.4.2, pp. 191–258.
17. J. Rodriguez-Carvajal, Physica B 192, 55 (1993).

18. P. J. Webster, J. Phys. Chem. Solids 32, 1221 (1971).

19. K. R. A. Ziebeck and P. J. Webster, J. Phys. Chem. Solids 35, 1 (1974).

20. Y. Takamura, R. Nakane, and S. Sugahara, J. Appl. Phys. 105, 07B109 (2009).

21. G. L. W. Hart and R. W. Forcade, Phys. Rev. B 77, 224115 (2008).

22. See Supplemental Material at http://link.aps.org/supplemental/

10.1103/PhysRevB.93.104424 for additional crystal structural information.
23. W. Kohn, A. Becke, and R. Parr, J. Phys. Chem. 100, 12974 (1996).

24. G. Kresse and J. Hafner, Phys. Rev. B 47, 558 (1993).

25. G. Kresse and J. Furthm¨uller, Phys. Rev. B 54, 11169 (1996).

26. P. E. Bl¨ochl, Phys. Rev. B 50, 17953 (1994).

27. G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999).

28. J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996).

29. Y. Wang and J. P. Perdew, Phys. Rev. B 44, 13298 (1991)

30. J. M. Sanchez, F. Ducastelle, and D. Gratias, Physica A 128, 334 (1984).

31. M. Asta, C. Wolverton, D. de Fontaine, and H. Dreysse𝑒́ , Phys. Rev. B 44, 4907 (1991).

32. C. Wolverton, M. Asta, H. Dreyss´e, and D. de Fontaine, Phys. Rev. B 44, 4914 (1991).

73
33. J. W. D. Connolly and A. R. Williams, Phys. Rev. B 27, 5169 (1983).

34. A. Zunger, in Statics and Dynamics of Alloy Phase Transformations edited by P. E. A.

Turchi and A. Gonis, NATO ASI Series (Springer US, New York, 1994).
35. A. van de Walle, M. Asta, and G. Ceder, Calphad 26, 539 (2002).

36. A. van de Walle and G. Ceder, J. Phase Equilib. 23, 348 (2002).

74
 CHAPTER 4

STRUCURAL AND MANGETIC PROPERTIES ANALYSES OF THE BULK FexCo1-xTiSb

ALLOY SYSTEM: THE Fe0.5Co0.5TiSb COMPOUND AS PROTOTYPICAL HALF
HEUSLER

4.1 Introduction

The Heusler alloys and compounds have become a household name in the field of

materials science since the discovery of the Cu2MnSn alloy by Friedrich Heusler as a ternary

compound with peculiar magnetic behavior. During the course of the past hundred years they

have shown to exhibit many interesting electronic properties such as ferromagnetism,

ferrimagnetism, half metallicity, superconductivity, semiconductivity, thermoelectricity and

topological insulation. The technological applications of this class of intermetallics are found in

diverse fields.1-5

The so-called “half Heusler” subclass of these compounds has been considerably

exploited to harvest their potential spintronic and thermoelectric capabilities. This includes a

very recent report of the antiferromagnetic compound CuMnAs used as the building block of an

all-antiferromagnetic memory with potentially tangible increase in speed of operation,

robustness, energy efficiency and storage density which could be an instance of the importance

of the antiferromagnetic spintronics.9 This subclass crystallizes in a C1b cubic structure (space

group 216) and has an equiatomic ternary stoichiometry of XYZ in which X and Y are usually

transition metals and Z is a metalloid element.

75
 Following our own findings on the Co-Ti-Sb and Fe-Ti-Sb systems, we decided to exploit

the quaternary amalgamation of the mentioned alloys. 6 There were some reports in the literature

indicating that the Cobalt-rich side of the FexCo1-xTiSb was studied for their thermoelectric

capacities and results indicated that there were impurities present within the alloy microstructure

especially in the Fe0.5Co0.5TiSb composition, the hypothetical half Heulser of the alloy system. 7

This all prompted us to consider a metallurgical evaluation of the entire alloy including the iron-

rich side of the system specifically in order to consider the existence of a half Heusler quaternary

compound.

4.2 Experimental

Stoichiometric ratios of pure elements Fe(Alfa Aeaser, 99.98%) , Co (Alfa Aesar,

99.95%), Ti (Alfa Aeasar, 99.9%) and Sb (Alfa Aeasr, 99.999%) were mixed together and

molten using an Edmund Buehler Mini MAM-1 compact arc melting system under an ultra-high

purity Argon pressure of 0.02 Pa. A 10% excess of Antimony was added to the initial melt to

compensate for the mass of the matter lost due to the volatility. A titanium oxygen getter was

melted before all the runs to avoid any potential oxidization of the sample. The ingots were

flipped and remelted at least 5 times to ensure the proper mixing of the all the three elements and

eventual homogeneity of the material. The bulk samples were then cut and mounted for

metallographic studies. A JEOL 7000 Field emission scanning electron microscope (SEM) was

used for the energy-dispersive X-ray spectroscopy (EDS) to ensure the chemical composition of

the as-cast ingots had not deviated from the intended target stoichiometry. The heat treatment

samples were wrapped with Tantalum and were sealed under vacuum in quartz tubes in a

vacuum atmosphere to avoid oxidation.

76
 The pieces were heat treated at various temperature and dwell times out of which a cycle

of treatment at 900 °C for 3 days and cooled in the furnace with an approximate rate of 90 °C per

hour appeared to result in the optimum C1b structure and a homogenous microstructural

stoichiometry. The crystal structure was then studied using a Bruker D8 Discover X-Ray

Diffraction (XRD) system with a monochromatic cobalt Kα radiation. In order to minimize the

surface and crystal orientation effects the samples were mounted, polished and rotated in the φ

axis during the XRD analysis. Also electron backscatter diffraction (EBSD) phase map was

performed to affirm the structure observed on the XRD pattern using the JEOL 7000 FE-SEM

system. Care was taken so that the EBSD step size does not exceed the tenth of smallest

microstructural feature so that no phase is left out during the scanning process. The simulated

XRD patterns were obtained by using a CaRine crystallography 4.0 software and the Rietveld

refinement was carried out through a Match! Software based upon FullProf algorithm. The

magnetic and electric properties were measured using a Dynacool VSM and a PPMS system

respectively. The Transmission Electron Microscopy (TEM) selected area electron diffraction

(SAED) pattern was taken on a FEI Tecnai F-20TM system using a camera length 670 millimeters

for generating the ring pattern. The TEM sample was prepared using a TESCAL Lyra3 TM

Focused Ion Beam (FIB) SEM system using a Gallium ion source.

4.3 Results and discussion

4.3.1 CoTiSb

All the samples heat treated at 900 °C showed a mostly granular microstructure. The EDS

test showed that the composition of the grains to be CoTiSb with an uncertainty of

approximately 5 % with respect to each element and therefore quite near the intended

stoichiometry. Figure 4.1 shows an example of such microstructure and its relevant EDS result.

77
Figure 4.1 (a) The Metallography image of a CoTiSb Sample heat treated at 900 °C for 7 days.
The sample was etched for 30 seconds using the marble etchant. The microstructure is evidently
single grain. (b) SEM micrograph of the same surface showing the grain structure and its
relevant EDS spectrum .The composition of the grains was observed to be CoTiSb.

Figure 4.2 exhibits XRD pattern of the polished and etched sample, indicative of an FCC

structure. The presumed Wyckoff positions of the constituting elements is the same as the ones

reported by Barth and collaborators in which Titanium was positioned at 4a (0,0,0) , Antimony at

4b ( 1/2 , ½ , ½ ) and Cobalt at 4c ( ¼, ¼, ¼ ).8 There is further concordance with their findings

as the calculated lattice constant from the experimental data here is 5.898 Å, which is quite

similar to the reported 5.884 Å. The intensities of (111) and (200) peaks are higher relative to

that of (220), which is attributed to the relatively large grain size and the consequently

considerable texture effect.

Figure 4.3 shows the hysteresis loop of this compound obtained at 5 Kelvin. The material

shows paramagnetic behavior with almost zero moment which is most probably originating

from some negligible magnetic elements such as unreacted cobalt.

78
 Figure 4.2 The XRD diffraction data of the CoTiSb sample heat treated at the 900 °C for 7
days. The spectrum is correlated to an C1b structure with a lattice parameter of 5.898 Å using
cobalt Kα X-Ray source.

Figure 4.3 The near zero Kelvin hysteresis loop of the CoTiSb compound heat treated at 900 °C
for 7days.
The paramagnetic behavior is in line with the findings in literature that CoTiSb is

semiconductor and consequently not magnetic.

79
4.3.2 Fe0.25Co0.75TiSb

The sample alloy was cast and heat treated for 7 days at 900 °C. The optical microscopy

and EDS revealed that the material is comprised of large grains with a CoTiSb chemical

composition and the remainder iron almost completely segregating between the mentioned

grains. Figure 4.4(a) is a metallographic image showing the grain structure and the figure 4.4(b)

is the EPMA relevant of the same region showing a possible segregation of iron within the grain

boundaries.

The XRD pattern of the sample is shown in Figure 4.5. The signature of the CoTiSb has

an overwhelming presence with some peaks attributable to the iron rich composition within the

grain boundaries. This is indicative of the microstructural data with the material being composed

of CoTiSb grains and iron rich intergranular segregations.

Figure 4.4 (a) The micrograph of the Fe0.25Co0.75TiSb sample heat treated at 900 °C for 7 days
and (b) its EPMA elemental map based on the iron element showing the grain segregation. The
chemical composition of the grains matched a CoTiSb stoichiometry.

80
 Figure 4.5 The XRD Pattern of the Fe0.25Co0.75TiSb heat treated for 7 days at 900 °C. Most of
the peaks are related to the CoTiSb grains and the ones marked with asterisk can be attributed the
iron rich phase segregated within the grain boundaries.

The magnetometry results showed that the alloy behaves still exhibits very weak

magnetic moment as it was the case with the CoTiSb compound. This is in concordance with the

presence of the CoTiSb as the dominant phase in the Fe0.25Co0.75TiSb forming a grain structure.

The weakly observed magnetization is stemming from the low amounts of possible iron

segregates based in the grain boundaries. The evaluation of both room temperature and 5 K

magnetic behavior confirms the prevalence of a paramagnetic phase within the system. The

relevant results are shown at figure 4.6.

81
 Figure 4.6 The magnetic behavior of the Fe0.75Co0.25TiSb alloy system at 5 Kelvin and at 300
Kelvin.

4.3.3 Fe0.5Co0.5TiSb

The prototypical half Heusler within the FexCo1-xTiSb system is a Fe0.5Co0.5TiSb alloy

stoichiometry which was considered worthy of synthesis itself. After the sample was heat treated

at 900 °C for 3 days a single grain microstructure was observed. There were no sign of any

precipitates as it was the case with previously mentioned stoichiometries. Figure 4.7 presents the

related results. The EDS analysis exhibited a consistent chemical composition on different

dimensional i.e. magnification levels conforming to the overall targeted stoichiometry of

approximately Fe17Co17Ti32Sb32.

82
Figure 4.7 (a) The optical micrograph of a Fe0.5Co0.5TiSb heat treated at 900 °C for 7 days. The
sample was etched for 10 seconds using the Adler etchant. The microstructure reveals single
grains without the presence of any secondary phase in the grain boundaries (b) SEM micrograph
of the same surface showing the grain structure and its relevant EDS composition. The grains
were observed to be approximately Fe17Co17Ti32Sb32 conforming to the intended homogenous
stoichiometry.

The atomic structure obtained from the XRD data indicates the presence of a C1b crystal

structure signature. The calculated lattice constant is 0.5918 nm. Crystal lattice simulations were

used to propose a model based on the presumed C1b lattice. In such model two atoms and two

atoms of cobalt are positioned within the four possible Wyckoff locations in 4c (1/4,1/4,1/4) sites

while four titanium atoms and four antimony ones are located on 4a (0,0,0) and 4b (1/2,1/2,1/2)

sites respectively. This model conforms quite well to the experimental spectrum of the material

as no additional peaks are observed with respect to the model and no absent peaks from the

spectrum are encountered either. To quantitatively evaluate this conformity a Rietveld

refinement fitting to the proposed model was also performed. The resultant fitting parameters

pointed toward a good agreement between the proposed model and the real-time experimental

data with an RBRAGG factor of only 0.1 and a reduced χ2 factor of 8.1.

83
 The relatively higher latter fitting parameter can be attributed to some peaks having

intensities somewhat different from the hypothetical XRD pattern which can be due to the fact

that the grain size is substantially large in this material standing at an average of 300 µm leading

to a considerable crystal orientation and texture effects resulting in the intensities not completely

concordant with the theoretical values. The XRD experimental pattern and its relevant Rietveld

refinement fit are shown in figure 4.8

Figure 4.8 The XRD spectrum of the Fe0.5Co0.5TiSb compound. All peak positions can be
attributed to a C1b structure with a lattice constant of 0.5918 nm. No ambiguous peaks were
observed. The Rietveld different plot has a signature of good fit with the proposed crystal model
of the hypothetical compound.

In order to verify both the presence of texture and also phase purity of material, an EBSD

phase map was performed on the microstructure. Figure 4.9 shows the band contrast, phase map

and the texture map of the grain structure.

84
 As it is shown on the phase map about 90 percent of the entire microstructure on a

relatively large cross section of the material (500 µm by 500 µm) has been colored red which is

significant of the C1b Fe0.5Co0.5TiS with the rest of microstructure being porous. This further

confirms the high purity of the material and its structural nature being a prototypical half

Heusler. The inverse pole figure map also shows large grain areas having different crystal

orientations which is also the reason that some of the peak intensities on the XRD pattern of the

material not completely overlapping the generated pattern by the crystal simulation and the

consequent higher degree of reduced χ2 factor compared to its RBRAGG counterpart.

Figure 4.9 a) The electron backscatter band contrast image of a relatively large area on the
surface of the Fe0.5Co0.5TiSb sample. b) The relevant EBSD phase map scan image of the same
area showing that almost the entirety of the surface corresponds to the hypothetical half-Heusler
C1b phase c) The orientation map of the same location and its inverse pole figure showing the
texture formation of the sample.

The case for single-phase nature of this alloy was further studied using the TEM method

to gather evidence for homogeneity at very dimensions and also to verify the consistency of the

assumed compound structure. Consequently, a FIB sample was taken out of a junction of three

grains so that sufficient statistical data was gathered and the sampling was not limited to only

one grain spot. The elemental map by EDS showed no sign of segregation among the main

elements in within the sample.

85
 Also different EDS taken from different spots on the sample indicated that the

stoichiometry averaged to a Fe16Co17Ti32Sb32 which is in good agreement with the target

compound composition of Fe0.5Co0.5TiSb. A SAED ring pattern taken from the sample is shown

in figure 4.10. The ring diameters overlap quite well with the expected values for a C1b structure

peaks with a lattice constant of 0.5918 nm. No other obscure spots were detected related to a

possible inhomogeneity. Overall the TEM findings also conform to our previous

characterizations of the Fe0.5Co0.5TiSb being indeed a C1b compound.

Figure 4.10 A SAED pattern of Fe0.5Co0.5TiSb heat treated at 900 °C for

3 days. There are spots visible belonging to the assumed C1b structure with a lattice constant of
0.5918 nm. No obscure spots were detected.

The magnetometry results of the compound showed that at many different temperatures

the material had no considerable magnetic moment as shown in the isotherms of figure 4.11 (a).

We also evaluated the resistance-temperature response of the material to configure the electrical

properties of the compound.

86
 The corresponding results are exhibited in figure 4.12. There were no hysteretic effects

witnessed in either presence or absence of magnetic field during the measurement. (Figure

4.12(a) and 4.12(b)). As the resistance rose from 10 Kelvin below, it was suggested that a

semiconducting behavior was dominant. Therefore an exponential fit for the lower temperature

regime was considered and eventually applied with good mathematical juxtaposition. The

finalized fit hinted to a semiconductor with a bandgap range of 0.12 meV to 0.16 meV. (Figure

4.12(c)).

At this stage the compound was considered to be a semiconductor and a paramagnet but

the theory calculations by our colleagues in the Northwestern university led to a hypothetical

antiferromagnetic coupling between iron and cobalt atoms resulting in a specific energy and a

predicted Neel temperature at 10 Kelvin. (Section 4.4) In order to detect any Neel transition a

susceptibility versus temperature measurement was done with the result manifested in figure

4.11 (b). As it seen, there is a maximum peak observed at approximately 8 Kelvin, showing an

entry into an antiferromagnetic phase from this temperature upwards. This finding is in good

agreement with the prediction of antiferromagnetic Neel temperature of 10 K.

87
 Figure 4.11 (a) Moment versus temperature isotherms of the Fe0.5Co0.5TiSb at a range of 5
Kelvin up to the room temperature. No considerable moment is detected in the material. (b) The
susceptibility versus temperature diagram of the compound. As it is seen from the inset graph,
there is maximum seen at 8 Kelvin in line with a Neel antiferromagnetic behavior.

Figure 4.12 (a) The Resistance-Temperature behavior of the Fe0.5Co0.5TiSb compound at zero
field (b) The Resistance-Temperature behavior of the Fe0.5Co0.5TiSb compound at 1kOe. (c) The
exponential fit at the less than 10 Kelvin.

4.3.4 Fe0.75Co0.25TiSb

The alloy was cast by arc melting the constituting elements following the same

methodology as the other compositions and a consequent heat treatment at 900 °C as it was the

case with the Fe0.5Co0.5TiSb compound mentioned in the previous section.

88
 In order to obtain the structural signature of alloy an XRD pattern was taken from a

sample with the aim to examine the possibility of a single-phase compound emergence as it was

the case with the Fe0.5Co0.5TiSb compound. The corresponding spectrum is illustrated in figure

4.13 which is still showing a C1b signature present with extra peaks which are not attributable to

the know phases. The optical and electron microscopy results showed the evolution of a dual

phase microstructure with a granular formation and a secondary intergranular phase emerging in

some locations.

These observed morphologies were encountered under different heat treatment cycles

and no single phase microstructure could be achieved even at higher temperatures or lengthier

dwell times exhibiting the stability of these phases. The EDS analysis showed that the grain

composition was somewhat deviating from the initial alloy stoichiometry related to the

secondary being almost completely void of cobalt, meaning that there were diffusion of the

other three constituting elements to the intergranular loci .

89
Figure 4.13 The XRD Pattern measured using a Co Kα radiation of the Fe0.75Co0.25TiSb sample
heat treated at 900 °C for 14 days. The signature is still consistent with a C1b structure while the
red stars indicate the presence of unidentified phase(s).

Interestingly the composition of the intergranular phase appeared to be approximately

Sb1.8Ti1.5Fe which is identical to the phase we observed during the heat treatment of FeTiSb

alloy.6 The EDS data taken from at least 13 different grains within the microstructure culminated

that the grain composition could be averaged to a stoichiometry of Fe0.64Co0.25Ti0.77Sb0.71 with

statistical confidence level of 95 percent, which again indicates that the grains have deviated

from the starting composition, Fe0.75Co0.25TiSb and with iron, titanium and antimony diffusing

into the grain boundaries and forming the ternary Sb1.8Ti1.5Fe composition with cobalt basically

present within the grains only. Figures 4.14 (a) and 4.14 (b) show the optical micrograph and a

corresponding SEM image of the microstructure with grains and the secondary phase

respectively. It is also noteworthy the intergranular phase has the same morphology witnessed in

the FeTiSb system which is somewhat lamellar.6

90
 Figure 4.14 (a) An optical micrograph of the Fe0.75Co0.25TiSb alloy showing grain formations
and evolution of a secondary phase between them. (b) SEM image exhibiting the grains and the
secondary phase found to have a stoichiometry of Sb1.8Ti1.5Fe through EDS.

As the XRD pattern of this alloy had a C1b signature while the stoichiometry of the grains

was somewhat deviated from the C1b composition itself, we decided to perform an EBSD phase

mapping on the microstructure of this alloy with the target phase being the C1b Fe0.5Co0.5TiSb so

that we could see if the mapping process can attribute the C1b phase to the off-stoichiometric

grains so that the source of the C1b signature in the XRD pattern is thus elucidated. Figure 4.15

shows the results of such an EBSD scan. Figure 4.15 (b) clearly shows a full coloring of the

grains in red, signifying the presence of the target C1b there. The assigned volume fraction of the

C1b stands at approximately 75 percent with the rest attributable to the ternary secondary

intergranular phase, voids and porosities.

91
 Figure 4.15 (a) Electron Backscatter band contrast image of a large section of the
Fe0.75Co0.25TiSb alloy. (b) The phase map taken from the cross section attributes the C1b
Fe0.5Co0.TiSb structure to the grains (colored red). The rest of the microstructure is colored
black. (c) The relevant orientation map and inverse pole figure of the cross section.

Magnetometry behavior of the material obtained from VSM measurements show a

mostly paramagnetic behavior dominating with some weak contribution from possible magnetic

impurities within the material. This negligible magnetic effect might arise from small magnetic

phases of iron or cobalt emerging at this alloy concentration. Figure 4.16 shows the room

temperature and near-absolute zero hysteresis loops.

Figure 4.16 The 5 Kelvin and 300 Kelvin hysteresis loops of the Fe0.75Co0.25TiSb alloy.

92
4.3.5 FeTiSb

Here we refer to our previous findings in the Fe-Ti-Sb system. In our initial theoretical

studies, the half Heusler FeTiSb compound was expected to be a stable half metallic compound.6

Microstructural analysis showed that the material in fact decomposed into two separate phases.

One is the iron rich Fe1.5TiSb phase with a strong presence of a Titanium-Antimony rich phase.

Figure 4.17 displays the optical micrograph and also EDS analysis of these two separate phases.

The composition of the Titanium and Antimony rich phase was found out to be approximately

Sb1.8Ti1.5Fe which was encountered in the previous section’s quaternary alloy as well. Presence

of this phase can be attributed to the iron phase Fe1.5TiSb to be the dominating phase and as the

overall composition of FeTiSb is iron deficient compared to the dominant phase, unlike the case

of Fe2TiSb in which the excess iron precipitated on the grain boundaries in this case the excess

Titanium and Antimony have formed a new granular phase indicating that the level of the

solubility of the dominant phase for Titanium and Antimony might be less than its solubility

level for iron.

Figure 4.18 shows the result of the XRD analysis of the sample. It shows the presence of

the L21 which originate from the Fe1.5TiSb while there are obscure peaks in the low angle which

could be attributed to the titanium and antimony rich grains.

93
 Figure 4.17 (a) An optical metallography image of FeTiSb exhibiting two distinct granular
phases present. (b) The SEM image and the relevant EDS of the intergranular secondary phase
being Sb1.8Ti1.5Fe. (c) The composition of the grains was found to be Fe1.5TiSb.

Figure 4.18 The XRD spectrum of the FeTiSb compound. Most peak positions can be attributed
to an L21 structure while there are some obscure peaks possibly arising from Titanium and
Antimony rich phase.

4.4 First-Principle Calculations of Fe0.5Co0.5TiSb Electronic Bandstructure

Using first-principles calculations based on density functional theory (DFT), we

investigate the structural order, energetics, and electronic structure of the single-phase

Fe0.5Co0.5TiSb.

94
 Since our experimental measurements indicate that the material assumes a C1b-type

lattice, we begin by systematically investigating the energetics of various atomic configurations

at the composition. We construct a cluster expansion for the energy of mixing of Fe, Co atoms

on the X-site (¼, ¼, ¼) of a XYZ C1b structure and identify the lowest-energy ordered structure.

We find an overall low energy of mixing, of the order of 0.020 eV/site, indicating possible Fe-Co

disorder in the material at RT or higher temperatures. Our calculations predict that the energy

gain (i.e., increase in stability) due to ferromagnetic order (FM) in the material, when compared

to the non-spin polarized, non-magnetic (NM) structure is small: ~0.007 eV/atom. Nonetheless,

all the atomic configurations sampled at the Fe0.5Co0.5TiSb composition, exhibited a near half-

metallic electronic structure with the Fermi level falling at the edge of the gap in one spin

channel (minor) and a finite density of states in the other spin channel (major).

Based on (a) the insignificant increase in stability due to FM order, and (b)

semiconducting behavior of Fe0.5Co0.5TiSb in experiments, we explore other magnetic

configurations, namely, different kinds of antiferromagnetic (AFM) order in the material. Of the

AFM configurations sampled, we find a few (see Figure 4.19) that are almost degenerate (i.e.,

within 0.001-0.002 eV/atom) with the lowest energy FM configuration. This is an indication of

competition between different exchange interactions, and that a more-complex AFM ordering

and/or spin frustration in the lattice may be energetically more favorable than the FM ordering.

Further, such energetics predict an extremely low Neel temperature if the 0 K ground state is

indeed AFM. As expected, all the AFM configurations sampled show a (near) metallic electronic

structure with the Fermi energy falling within the valence band but close to its maximum. This

delicate electronic structure is likely to be destroyed by local off-stoichiometry, impurities, etc.

due to the resulting change in the electron count.

95
 The antiferromagnetic atomic configuration is also shown at figure 4.20. Once again the

energy competition between such an antiferromagnetic phase and the potential ferromagnetic is

near enough for the former to become possibly stable.

Figure 4.19 Calculated electronic density of states (DOS) of Fe0.5Co0.5TiSb with (left)
ferromagnetic (FM) and (right) antiferromagnetic (AFM) order. The FM configuration is a near
half-metal with a finite DOS at Fermi energy EF in only one spin channel, whereas the AFM
configuration is a metal with EF at the edge of the valence band and close to its maximum.

Figure 4.20 Schematics showing a few possible antiferromagnetic configurations in a C1b

Fe0.5Co0.5TiSb crystal structure. One such configuration (left) is within 0.9 meV/atom of the
ferromagnetic configuration, indicating competition between different exchange interactions and
possibly complex magnetic ordering in the system.

96
4.5 Conclusion

Our investigation into the FexCo1-xTiSb alloy system showed that in the cobalt-rich part

of the stoichiometry the CoTiSb compound was dominating the system with iron segregating

into the grain boundaries. No sign of a single-phase alloy was found in this region.

Interestingly with iron content reaching the equiatomic ratio with cobalt, i.e. Fe0.5Co0.5TiSb,

enough evidence was garnered to confirm the existence of a postulated C1b half-Heusler phase

when the alloy was heat treated at 900 °C for 3 days.

The further addition of iron though resulted in loss of the single-phase state, with a

ternary Sb1.8Ti1.5Fe phase emerging and the grains becoming somewhat off-stoichiometric

though still retaining a C1b structure.

It is recommended that advanced structural analysis such as Mossbauer spectroscopy and

neutron diffraction be performed to further analyze the disorder and atomic defect in the crystal

structure of the Fe0.5Co0.5TiSb phase.

4.6 References

1. C. Felser, L. Wollmann, S. Chadov, G.H. Fecher and S.S.P. Parkin in Heusler Alloys:
Properties, Growth and Applications, Edited by C. Felser and A. Hirohata ( Springer
International, Switzerland, 2016).

2. T. Klimczuk, C. H. Wang, K. Gofryk, F. Ronning, J. Winterlik, G. H. Fecher, J.-C. Griveau,

E. Colineau, C. Felser, J. D. Thompson, D. J. Safarik, and R. J. Cava, Phys. Rev. B. , 85, 174505
(2012).

3. I. Galanakis, K. Özdoğan, and E. Şaşıoğlu, AIP. Adv., 6, 055606 (2016).

4. Serrate D, De Teresa JM, Ibarra MR J. Phys. Condens. Matter 19:023201, (2007).

97
5. Kandpal HC, Ksenofontov V, Wojcik M, Seshadri R, Felser C J Phys D, Appl Phys. , 40:1587,
(2007).

6. N. Naghibolashrafi, S. Keshavarz, Vinay I. Hegde, A. Gupta, W. H. Butler, J. Romero, K.

Munira, P. LeClair, D. Mazumdar, J. Ma, A. W. Ghosh, and C. Wolverton, Phys. Rev. B. , 93 ,
104424, (2016).

7. T. Wu, W. Jiang, X. Li, Y. Zhou and L. Chen, J. Appl. Phys. , 102, 103705 , (2007).

8. J. Barth, B. Balke, G.H. Fecher, H. Stryhanyuk, A. Gloskovskii, S. Naghavi and C. Felser , J.

Phys. D. : Appl. Phys. 42 , 185401, (2007).

9. P. Wadley B. Howells, J. Železný, C. Andrews V. Hills, R. P. Campion, V. Novák, K.

Olejník, F. Maccherozzi, S. S. Dhesi, S. Y. Martin, T. Wagner, J. Wunderlich, F. Freimuth, Y.
Mokrousov, J. Kuneš, J. S. Chauhan, M. J. Grzybowski, A. W. Rushforth, K. W. Edmonds, B.
L. Gallagher, T. Jungwirth , Science, Vol. 351, 6273, (2016).

98
 CHAPTER 5

SYNTHESIS AND CHARACTERIZATION OF Co-Ti-Sn ALLOYS:

Co1.5TiSn AS A HALF METAL BASED ON THE Fe1.5TiSb PROTOTYPE OF LAYERED

HEUSLER

5.1 Introduction

The half-metallic materials are optimal candidates for spintronic applications such as spin

torque transfer random access memories.1,2,3 Their inherently high spin polarization and Curie

temperatures are highly sought for novel magnetic recording devices. Half metallic materials

have been found among the family of intermetallic compounds called Heuslers. Theoretical and

experimental studies both have hint towards presence of many possible half metals within this

class of intermetallics.4, 5

Our recent findings on the Fe-Ti-Sb alloy system resulted in the possibility of the

existence of an intermediate Heusler stoichiometry which neither follows the 2:1:1 full Heusler

atomic ratio nor the 1:1:1 half Heusler one, but instead has the 1.5:1:1 ratio between its

constituting elements and therefore can be called a ‘layered’ Heusler. 6 The prototype which was

Fe1.5TiSb interested us in searching for the possibility of a hidden class Heuslers with this

specific – and yet undiscovered – stoichiometry. Our theoretical calculations led by William H.

Butler and Christopher Wolverton and their respective research groups arrived at the possibility

of the Co1.5TiSn being a material of interest.

Initial theoretical calculations showed that similar to Fe1.5TiSb, the stacking of Cobalt

vacancies in the [111] direction will result to an energetically stable material with an electronic

bandstructure of a half metal as shown in figure 5.1.

99
 Figure 5.1 Electronic Bandstructure of the Co1.5TiSn in an R3m vacancy stacking. The Fermi
level in a minority channel gap while there are majority channel states available leading to a half
metallic behavior.

The cluster expansion analysis of the Co-Ti-Sn resulted not only in the Co1.5TiSn

compound being a stable ground state , but other alloys were seen to be ground states and so

stable compounds such as Co1.33TiSn, Co1.67TiSn and CoTiSn.

Our initial literature survey showed that although that Co2TiSn has been extensively

studies in the past decades as a Heusler compound 7,8,9 and more recently established

theoretically and experimentally as half metal by the means of ab-initio calculations and Andrew

Reflectivity respectively. 11,12,13 There was only one instance related to the Co1.5TiSn being

mentioned in the scientific literature in which the named compound was confirmed to be the

main phase in the CoTiSn alloy alongside some titanium and tin phases. The study was solely

concentrated on neutron diffraction of the CoTiSn alloy.14

100
 Encouraged by the previous report on Co1.5TiSn being stable, and also capitalizing on the

fact that microstructural analyses on the bulk Co-Ti-Sn were scarce in the scientific literature, we

found impetus to embark on studying this alloy system using the very same methodology we

adopted in successfully characterizing the Fe-Ti-Sb alloys which led to finding of Fe1.5TiSb as an

unknown phase.

5.2 Experimental

Stoichiometric ratios of pure elements , Co (Alfa Aesar, 99.5%), Ti (Alfa Aeasar, 99.9%)

and Sn (Alfa Aeasr, 99.99%+) were mixed together and molten using an Edmund Bühler Mini

MAM-1 compact arc melting system under an ultra-high purity Argon pressure of 0.02 Pa. A

titanium oxygen getter was melted before all the runs to avoid any potential oxidization of the

sample. The ingots were flipped and remelted at least 5 times to ensure the proper mixing of the

all the three elements and eventual homogeneity of the material. The bulk samples were then cut

and mounted for metallographic studies. A JEOL 7000 Field emission scanning electron

microscope (SEM) was used for the energy-dispersive x-ray spectroscopy (EDS) to ensure the

chemical composition of the as-cast ingots had not deviated from the intended target

stoichiometry. The heat treatment samples were wrapped with Tantalum and were sealed under

vacuum in quartz tubes in a vacuum atmosphere to avoid oxidation. The pieces were heat treated

in various temperature and dwell times out of which a cycle of treatment at 900 °C for 3 days

and cooled in the furnace with an approximate rate of 90 °C per hour appeared to result in the

optimum structure and a homogenous microstructural stoichiometry for the mainly-sought

Co1.5TiSn compound. The heat treated sampled were then mounted in a conductive

metallographic mount and grinded and polished with polycrystalline diamond and colloidal silica

solutions. The crystal structure was then studied using a Bruker D8 Discover X-Ray Diffraction

(XRD) system with a monochromatic cobalt Kα radiation.

101
 In order to minimize the surface and crystal orientation effects the samples were rotated

in the φ axis during the XRD analysis. Also, electron backscatter diffraction (EBSD) phase map

was performed to affirm the structure observed on the XRD pattern using the JEOL 7000 FE-

SEM system. Care was taken so that the EBSD step size does not exceed the tenth of smallest

microstructural feature so that no phase is left out during the scanning process. The simulated

XRD patterns were obtained by using a CaRine crystallography 4.0 software and the Rietveld

refinement was carried out through a Match! Software based upon FullProf algorithm. The

magnetic and electric properties were measured using a Quantum Design DynacoolTM system

ranging from 5 Kelvin to 400 Kelvin.

5.3 Results and Discussion

5.3.1 Co1.5TiSn

The alloy was arc melted using stoichiometric ratios of cobalt, titanium and tin. The

workpiece was remelted several types to ensure complete mixing of the constituting elements.

EDS analysis of the different sections and magnifications of the as-cast sample showed a 1.5:1:1

ratio existing between the main elements as intended.

Optimal homogenization cycle was found to be a 3-day heat treatment at 900 °C

resulting in a single grain microstructure. It should be noted that dwell times longer than 3 days

and temperatures higher than 900 °C proved to be detrimental in the sense that the impurity

phases started to emerge in the microstructure under such conditions.

Optical metallography revealed a microstructure in forms of grains and no trace of a

secondary phase. Only natural porosities were observed. The electron microscopy under SEM

also showed the absence of any impurity other than the grains.

102
 The EDS also confirmed that the stoichiometry of the grains has an average of

Co42at%.Ti28at.%Sn28at.% which is in very good agreement of the starting target composition. The

microstructure micrographs are shown in figure 5.2.

Figure 5.2 (a) The optical micrograph of a Co1.5TiSn sample homogenized at 900 °C for 3 days.
The presence of grains and some gas porosities are seen. (b) The SEM image of the same
sample. The selected EDS region showed a good agreement with the final Co1.5TiSn hinting the
grains having the intended composition.

In order to study the crystal structure of this single-phase alloy, a lattice cell was

constructed. The model for this cell was based on our assumption that similar to the Fe1.5TiSb ,

the Co1.5TiSn is also basically an L21 structure with two atoms missing in the 8c (1/4, ¼, ¼ )

position and titanium in 4b (1/2,1/2,1/2) and tin at the 4a (0,0,0) positions respectively. This

meant that in the crystal simulations a 75 percent occupancy level at the 8c position should be

considered, as for a 1.5:1:1 stoichiometry it is needed to reduce the multiplicity number of the 8c

position from 8 to 6 in order to meet the required 6:4:4 ratio between the atomic positions

multiplicity numbers.

The assessment of the presumed crystal model was done through XRD diffraction and its

consequent Rietveld refinement. The obtained spectrum showed a signature of L21 lattice and no

unrelated peak was observed further indicating the presence of a single-phase material as it was

previously seen under the microstructure.

103
 The Rietveld fit to the presumed crystal model resulted in good agreement with highly

satisfactory fitting parameters of reduced χ2 of only 1.4 and weighted average RBragg factor of

0.4.

The initially adopted Wyckoff positions were not changed after final convergence

showing the validity of the model. The fitted lattice constant of the material was 6.001

Angstroms. The experimental XRD pattern and its Rietveld fit can be seen in figure 5.3.

Figure 5.3 The experimentally observed pattern of Co1.5TiSn, the calculated fit and the
difference pattern.

The microstructural and structural data and their good agreement to the assumed model

inspired us to further analyze the material through a phase map using EBSD method. The goal

was to search for the target phase in the microstructure and any possible impurity present.

Therefore the same atomic coordinates and lattice constant explained previously were adopted as

the target phase. The material was scanned at a relatively large area size to include as many

morphological and secondary phases as possible. The adopted step size of 0.3 micron was much

less than the average grain size, which was 60 microns, to ensure a thorough scanning of the

entire microstructure.

104
 The obtained EBSD phase map showed an impressive 98 percent match with the

microstructure. The remainder unresolved morphologies could be attributed to grain boundaries

and surface effects. The respective maps are shown in figure 5.4.

Figure 5.4 (a) The band contrast image of the Co1.5TiSn alloy microstructure. An fully equiaxed
cross section is observed. (b) The phase map of the same cross section. All the morphological
features are colored red, which indicates a 98 percent match with the target Co1.5TiSn phase
structure.

Figure 5.4 (a) shows the band contrast image of the chosen microstructure cross section

which shows a granular microstructure with no sign of any secondary morphology. In figure 5.4

(b) the grains are all colored red which stands for the assumed Co1.5TiSn structure and thus is

indicative of a homogenized single phase alloy having the L21 structure with Cobalt vacancies in

the 8c position as the initial model was adopted.

The characterization of the structure and microstructure revealed a single-phase alloy and

so we decided to obtain the magnetic properties of this compound, which showed a very good

agreement with our crystallographic target model. The room temperature hysteresis loop

measurement showed a paramagnetic behavior as seen in figure 5.5 (a). Interestingly the near

absolute zero loop obtained at 5 Kelvin not only a magnetic behavior resembling a Heusler

compound with a soft magnetic hysteresis , but also the saturation moment was measured to be

0.63 µB/f.u. which is interestingly quite near the predicted 0.5 µB/f.u. value for the half metallic

compound. (Figure 5.5 (b)).

105
 Figure 5.5 (a) Paramagnetic behavior of Co1.5TiSn compound at 300 Kelvin. (b) The hysteresis
loop of the material at 5 Kelvin. The saturation moment stands at 0.63 µB/f.u. A comparison
between both the room temperature and the near-absolute zero temperature behavior is also
shown.

The results so far hinted towards a high probability that the Co1.5TiSn is indeed a single-

phase material with the predicted structure and half metallic moment. As the material showed

paramagnetic behavior at room temperature, we considered to test the Rhodes-Wohlfarth model

on it to see if the effective paramagnetic moment is much smaller compared to its saturation

moment at near absolute zero temperatures, which has been suggested as a hallmark of half

metallicity.

Rhodes-Wohlfarth was initially proposed to evaluate the nature of electronic contribution

in the ferromagnetic material, if the ratio was around 1, the material behaved as a Heisenberg

magnet with localized electron contribution while ratios higher than one was attributed to

itinerant electrons dominance. 15 The pioneers of the half metallicity encountered an unusually

low Rhodes-Wohlfarth ratio for the established half metals which was much lower than 1. 16, 17, 18

106
 This unprecedented behavior was attributed to drastic reduction of magnetic moment

upon entry into paramagnetic moment as the density of states available was considerably reduced

due to the presence of a bandgap in the half metallic material. 19, 20 the ratio is defined as pc/ps

with ps being the saturation moment per formula unit at zero Kelvin and pc defined as μeff2 = pc

(pc+2).

µeff. being the effective paramagnetic moment, calculated from Curie-Weiss law, C= (Nμeff2) /

3A.k, and C being Curie-Weiss constant derived from the inverse mass susceptibility formula, χ-1

= (T-θ)/C. The Inverse mass susceptibility of Co1.5TiSn compound was obtained with respect to

temperature to derive the C constant, which was found to be 2.437 × 10-5 emu.Kelvin/gr.Oe .

(Figure 4.6).

Figure 5.6 (a) The inverse mass susceptibility behavior of Co1.5TiSn with respect to temperature
at 2 Tesla (b) Linear portion of the paramagnetic region of the same graph. The slope, equating
to 1/C, was found to be 41027 gr.Oe/emu.Kelvin.

Using the C value, the µeff. was calculated to be 0.223 µB and consequently pc was

obtained as only 0.02 µB . Dividing the pc by ps, which is the saturation moment at 5 Kelvin, or

0.64 µB, we found the Rhodes-Wohlfarth ratio to be 0.03 which is much lower than 1, bearing a

signature of a half metallic behavior.

107
5.3.2 Co2TiSn

Encouraged by our findings on Co1.5TiSn compound, we decided to test our methodology

further on the established half metallic compound of the Co-Ti-Sn, or Co2TiSn. Another impetus

for our consideration was the fact that the microstructural analysis of this compound was scantly

reported giving us an ample ground for analysis.

The starting sample was prepared through arc melting of appropriate weight percentages

of constituting elements. The sample was remelted at least 3 times to ensure full mixing and lack

of any macrosegregation. The optimal homogenization cycle was found to be a heat treatment at

1000 °C for 7 days. It should be mentioned that the heat treatment at 900 °C rendered a dual-

phase microstructure with cobalt impurities and treatment done at temperatures lower than 800

°C resulted in the inhomogeneous dendritic structure.

The microstructure of the homogenized sample is shown at figure 5.7

Figure 5.7 (a) The optical metallography of the Co2TiSn alloy heat treated at 1000 Celsius for 7
days. The microstructure is single-grain with some naturally occurring porosity. (b) The SEM
image of the same cross section and EDS of a grain which exhibited a target stoichiometry of
Co49at.%Ti24at.%Sn24at.%

108
 The figure 5.7 (a) reveals a homogenized microstructure with no sign of dendrites or

intergranular phases but a fully equiaxed structure with relatively large grains beyond a 100

micron margin. The SEM also confirmed no presence of any secondary phase with EDS

indicating a target composition of Co2TiSn for the grains in figure 5.7 (b).

As the initial microstructural results hinted towards the alloy being pure phase and having

the intended stoichiometry, an XRD spectrum was also obtained to evaluate the crystal structure.

Once again as reported in the literature, we expected to observe a full Heusler L21 signature and

this was exactly seen in the data. Figure 5.8 shows the obtained pattern. All the peaks of the L21

signature are seen and marked with no excess peak attributable to any possible secondary phase.

The lattice constant was derived to be 6.078 Angstroms which is not far from the value reported

in the literature for this material. 9

Figure 5.8 The XRD Pattern of the Co2TiSn using a Kα Cobalt source. All the peaks are related
to the L21 expected pattern.

109
 In order to further strengthen the case for the presence of a single-phase Co2TiSn alloy

with no impurities and an L21 full Heusler crystal structure, a phase mapping of microstructure

was also performed using the Wyckoff positions of the L21 prototype and a lattice constant of

6.078 Angstroms. The respective maps are shown in figure 5.9. The area was chosen large

enough to cover as much as microstructural features as possible.

Figure 5.9 (a) Band contrast image of a Co2TiSn cross sectional area with relatively large grains
and some intergranular porosity. (b) The relative phase map showing almost the entirety of the
microstructure colored red signifying a high match with the intended L21 phase.

Figure 5.9 (a) reveals a relatively large granular structure with grain sizes at least 200

microns and beyond. The figure 5.9 (b) is the phase map taken from the same area with a 92

percent matching which indicated there is a good agreement between the microstructure and the

sought phase. The remaining unresolved 8 percent can be attributed to the porosities, grain

boundaries and surface noise effects.

The evidence for a pure phase material was sufficient for us to proceed to the magnetic

evaluation of the Co2TiSn. The material was found to be behaving magnetically at room

temperature as a full hysteresis loop was obtained. The near-absolute zero hysteresis loop was

obtained at 5 Kelvin.

110
 The material was still exhibiting a hysteresis loop typical of Heuslers with no low

coercivity and a soft magnetic behavior due to lack of magnetocrystalline anisotropy of the cubic

Heusler structure. The saturation moment at 5 Kelvin was found to be 2.11 µB/f.u., which is in

very good agreement with the expected Slater-Pauling moment of the half metallic Co2TiSn. The

value is quite similar as those reported previously on this compound. The relevant loops are

shown in figure 5.10.

Figure 5.10 (a) The hysteresis loop of the homogenized Co2TiSn at 300 Kelvin. (b) The
hysteresis loop at 5 Kelvin. The saturation moment is calculated to be 2.11 µB/f.u.

As the saturation moment here was in concordance with its expected half metallic Slater-

Pauling moment, it was decided to apply the Rhodes-Wohlfarth model to see if the ratio will

again drop to less than 1. For this goal the inverse mass susceptibility versus temperature of the

material was obtained in both cooling and heating modes from low temperatures to the high-

temperature paramagnetic region of the material. Also a linear fit for the paramagnetic region

was drawn whose slope equals 1/C and the related value was found 7510 gr.Oe/emu.Kelvin. The

inverse mass susceptibility plot and the linear fit of its paramagnetic region is shown at figure

5.11.

111
 Figure 5.11 (a) The inverse mass susceptibility behavior of Co2TiSn versus temperature at 2
Tesla (b) The linear fit to the paramagnetic portion of the previous plot. The slope value was
found to be 7510.

Introducing the C value into the formula C=(Nμeff2 ) / 3A.k , the µeff. will be equal to

0.55 μB . And considering μeff2 = pc( pc+2) , the pc will be 0.14 µB . As the saturation moment

was 2.11 µB, the Rhodes-Wohlfarth ratio will be 0.07, which is much less than unity and

indicates a high presence of spin polarized electrons or a half metallic signature like the previous

case of Co1.5TiSn.

5.3.3 Co1.75TiSn

The findings of the two L21-based compounds, Co2TiSn and Co1.5TiSn, with half metallic

properties and also theoretical predictions by our colleagues in professor Wolverton group

induced us to consider an intermediate alloy between the named compounds which obviously

was the Co1.75TiSn. The sample was arc melted based on its formula and the stoichiometric

relation between constituting elements. The homogenization cycle was empirically found to be

15 days at 1000 °C. Lower dwell times and temperatures all resulted in inhomogeneous

microstructures. The related optical and electron micrographs are shown in figure 5.12.

112
 Figure 5.12 (a) Optical metallography image of the homogenized Co1.75TiSn sample. Large
grains with some porosity are seen. (b) The SEM image of the same area and an EDS taken from
a grain indicate that the composition is consistent with the intended stoichiometry of Co1.75TiSn.

The 5.12 (a) image shows a homogenized microstructure with large grains beyond a

margin of 200 microns which is expected considering the high temperature and lengthy treatment

cycle of heat treatment. 5.12 (b) shows a SE image and an EDS taken from a grain. The average

composition was standing at Co47at.%Ti27at.%Sn27at.% which translates into the target stoichiometry

and hints a full homogenization.

The XRD analysis of the pattern obtained from the sample still showed the same L21

signature observed for the previous cases of the Co1.5TiSn and Co2TiSn. The corresponding

spectrum is shown at figure 5.13. This might be an interesting find in the sense that it infers the

possibility of a solid-solution region present from the chemical range of Co1.5TiSn to Co2TiSn, as

not only the microstructure is single-phase in this range but also the crystal structure remains the

same within this area.

The magnetic measurements of the material showed also a mitigated saturation moment

between the values of Co1.5TiSn and Co2TiSn at 5 Kelvin with a value of 1.68 µB/f.u. Also

similar to the Co2TiSn the material was still behaving magnetic at room temperature. (Figure

5.14)

113
 Figure 5.13 The XRD Spectrum of Co1.75TiSn compound. All peaks are attributed to an L21
signature.

Figure 5.14 The hysteresis loop of the Co1.75TiSn compound at 5 Kelvin and 300 Kelvin.

The application of the Rhodes-Wohlfarth model also bore interesting results. The slope of

the paramagnetic part of the inverse mass susceptibility was found to be 8721 using which we

arrived at a µeff. equal to 0.424 μB and a consequent pc of 0.193 μB , which if divided by

saturation moment at 5 Kelvin, 1.68 µB, will bear a Rhodes-Wohlfarth ratio of only 0.11, very

similar to values witnessed for Co2TiSn and Co1.75TiSn . (Figure 5.15).

114
Figure 5.15 (a) Inverse mass susceptibility of Co1.75TiSn with respect to temperature at 3 Tesla
(b) The linear fit to the paramagnetic region.

As this compound is also showing a hallmark of half metallicity as far as the Rhodes-

Wohlfarth model is concerned while also exhibiting a L21 signature, the interesting question here

might be the possibility of a half metallic ‘region’ instead of only a compound being present in

the range of Co1.5TiSn all the way to Co2TiSn .

5.3.4 CoTiSn

The CoTiSn composition has been reported to be a compound in some literature although

no microstructural or structural evidence to verify such a claim has been offered. 9 A more recent

report by Nobata, Kurisu and their coworkers based mainly on neutron diffraction analyses

showed that the CoTiSn alloy basically decomposes into mostly a Heusler-type Co1.5TiSn (70

volume percent ) plus some titanium and tin compounds. 14 Here also a microstructural

evaluation was lacking so it inspired us to do the same course of research we did before for this

CoTiSn alloy as well especially now that no microstructural evidence was found on this system

and considering that we already had synthesized a single-phase Heusler-type Co1.5TiSn similar to

the Nobata report.

115
 So once again an alloy was formed through arc melting of main elements and the EDS of

the as-cast sample proved the exact targeted ratio. Different heat treatment cycles showed that up

to 900 °C and up to 15 days the material remained inhomogeneous as the microstructure were

overwhelmingly dendritic.

The dendritic morphology disappeared at 1000 °C and a dwell time of 7 days giving way

to a multiple phase microstructure. There were three distinct phases visible under the optical

microscope: A matrix upon which other phases are embedded, a larger and brighter phase and a

smaller and darker one. A metallographic optical micrograph of the microstructure is shown at

figure 5.16.

Figure 5.16 The microstructure of the CoTiSn heat treated for 7 days at 1000 °C. Three distinct
morphological features can be differentiated.

The EDS analysis of the microstructure during the SEM showed that the matrix mostly

has a stoichiometry of Co44at.%Ti29at.%Sn29at.% , which can be rewritten as Co1.5TiSn . The larger

sized embedded phase which looked brighter under the optical microscope was found to be pure

tin. The darker and smaller sized feature in the meantime showed a binary stoichiometry of

Ti53at.%Sn47at%.The related SEM micrographs are shown below in figure 5.17.

116
 Figure 5.17 (a) EDS data taken from a rectangular area of the matrix. The composition
conformed to Co1.5TiSn. (b) The embedded phase in figure (a) is measured to be Ti53at.%Sn47at%.
(c) The smaller-sized features were found to be pure Sn.

The results are in good agreements with Nobata and Kurisu’s findings. According to their

neutron diffraction and subsequent Rietveld analyses the CoTiSn decomposed into mostly

Co1.5TiSn, which is exactly what we observed with this compound being the matrix of the

microstructure. Also according to their report the remaining phases were Sn and Ti6Sn5 both of

which are seen in our study with the latter phase being Ti53Sn47 in our case.

To further analyze the structure a XRD test was done on the sample. The data resulted in

a spectrum showing a Co1.5TiSn signature plus some peaks which could be attributed to tin and

also some to a Ti6Sn5 phase. The spectrum is shown at figure 5.18.

117
 Figure 5.18 The XRD spectrum of a CoTiSn sample heat treated at 1000 °C for 7 days. The
Co1.5TiSn, Ti6Sn5 and Sn peaks are assigned with red, green and blue asterisks respectively.

This is in concordance with our microstructural findings as the all the phases seen by

microscopy are also present in the XRD pattern and once again phases marked by Nobata group

using neutron diffraction is also witnessed through the X-Ray pattern .

5.3.5 Co1.33TiSn

As initial cluster expansion prediction by our colleagues in the Wolverton group

considered a possible Co1.33TiSn ground state and a compound, measures for preparing this alloy

were taken. The alloy was processed by arc melting the constituting elements and the as-cast was

remelted to ensure homogeneity which was verified by the EDS of the as-cast pieces showing the

target stoichiometry. The optimal homogenization cycle was found to be 7 days at 900 °C

following different empirical heat treatment trials.

118
 The metallography revealed that there were two distinct morphologies present in the

system: A granular phase which comprised the most of the microstructure and a secondary phase

located in-between the grains. The related images are shown in figure 5.19 (a). In order to

analyze the chemical composition of these morphologies an EDS under the SEM was performed.

The results pointed out that the grains which were the dominant phase as far the volume fraction

was concerned were in fact having the Co1.5TiSn, similar to the case of the multiple-phase

CoTiSn alloy which also had the Co1.5TiSn phase dominating the microstructure in the form of a

matrix.

The SEM micrograph taken from the several features of the microstructure is also shown

in figure 5.19 (b). The intergranular secondary phase showed a stoichiometry of Ti6Sn5 similar to

the case of CoTiSn. In contrast to the CoTiSn system here no tangible trace of tin was seen.

Figure 5.19 (a) The optical micrograph of a Co1.33TiSn alloy heat treated at 900 °C for 7 days.
Two different phases are observable. (b) The SEM micrograph of the same sample with
respective EDS of the features showing the dominant phase to be Co1.5TiSn and the intergranular
one to be Ti6Sn5.

The structural analysis of this alloy by XRD was also in agreement with the

microstructural observations.

119
 The obtained spectrum had a Co1.5TiSn signature alongside additional peaks related to

the Ti6Sn5 phase. Here also no sign of the tin phase seen in the CoTiSn alloy were witnessed.

The result is shown in figure 5.20

Figure 5.20 The XRD Spectrum obtained from a Co1.33TiSn Sample. The pattern shows peaks
related to the Co1.5TiSn and Ti6Sn5 phases.

The presence of a phase besides the dominant Co1.5TiSn indicates that Co1.33TiSn is not a

compound but a mixed phase like the case of the CoTiSn system. The magnetic behavior of these

two alloys was studied by obtaining hysteresis loops at 5 degree Kelvin. The saturation moment

of the Co1.33TiSn alloy was found to be 0.55 µB/f.u. which is lower than the value seen for the

Co1.5TiSn system.

120
 The reason for this reduction is that the Co1.33TiSn system has a magnetic component

which is Co1.5TiSn, that is also the dominant microstructural phase, and some non-magnetic

impurities which contribute to this relative decrease in the saturation moment. The respective

hysteresis loops of these two alloys are shown in figure 5.21.

Figure 5.21 (a) Hysteresis loop of Co1.33TiSn ally at 5 Kelvin. The saturation moment was found
to be 0.55 µB/f.u. (b) Hysteresis loop of CoTiSn alloy at 5 Kelvin. The saturation moment was
found to be 0.47 µB/f.u., which is less than the amount for Co1.33TiSn.

The CoTiSn alloy shows a lower value of the saturation moment with respect to the

Co1.33TiSn system, as in the former the volume fraction of the dominant magnetic phase which is

Co1.5TiSn is now lower as alongside the Ti6Sn5 secondary phase there are also tin impurities

emerging in the microstructure contributing to a reduction in moment as tin is also non-magnetic.

One can summarize our findings on the magnetic behavior and the phase distribution of the Co-

Ti-Sn Heusler system in the form of a graph such as the one presented in figure 5.22. It is

observable that in the single-phase region, the reduction on cobalt content will lead to a tangible

reduction in the compound moment as well. This can be explained in the way that the L21 lattice

is quite sensitive to the presence of the magnetic cobalt atom in the tetrahedral positions and any

reduction of it in the structure will result in a considerable moment decrease.

121
 Here should be mentioned that we also synthesized a Co1.62TiSn alloy which like its

adjacent counterparts appeared to be a L21 compound with a single-phase microstructure and a

Rhodes-Wohlfarth model of half metallicity, but for the sake of brevity the details of process and

characterization are not mentioned.

As soon as the cobalt content goes below the Co1.5TiSn borderline, the alloys deviate

from their previous single-phase behavior and multiple phase start to emerge in the structure.

Still the dominant phase was seen to be the Co1.5TiSn, while the rest of the phases were non-

magnetic therefore here the magnetic behavior almost appeared to be reaching a steady-state,

with magnetic moment steadily decreasing with emergence of the non-magnetic contributors

within the alloy the system.

Figure 5.22 The magnetization and phase behavior in the CoxTiSn.

122
 It is interesting to add that the same trend witnessed in the figure 5.22 was also reported

by the Pierre and coworkers. 9 Nevertheless in the mentioned report , all the alloys of the same

range in our figure 5.22 was reported to be pure-phase compounds without any microstructural

or structural evidence , while our studies shows that the single-phase behavior only exists in the

Co1.5-2TiSn stoichiometric range and the CoTiSn decomposes into three phases almost exactly as

the report by Nobata. 14

5.4 Conclusion

The Co-Ti-Sn alloys from their half to full Heusler equivalents were synthesized. It was

seen that the Co1.5TiSn was indeed a compound as predicted by our theoretical calculations with

a hallmark of L21 structure with cobalt vacancies in the tetrahedral positions similar to our

previous findings on Fe1.5TiSb compound. The magnetometry measurement also was in good

agreement with theory prediction of this compound being a half metal. Further magnetic analysis

by Rhodes-Wohlfarth also hinted into an unusually low ratio, which also is a fingerprint of half

metallicity. It was also seen that the Co1.5TiSn to Co2TiSn has an L21 signature and also

microstructurally single-phase therefore raising the possibility of a solid-solution region in this

chemical range. The compounds of this range also showed the unusually lower than unit Rhodes-

Wohlfarth ratio which again is a hallmark of half metallicity.

The alloys with cobalt content less than the one for Co1.5TiSn decomposed into a

dominant magnetic phase of Co1.5TiSn and intergranular Ti6Sn5, which also included another

phase which was found to be tin in the case of CoTiSn.

123
 This finding might open a door to the possibility of a 1.5:1:1 stoichiometry of Heusler

compounds present which has been overlooked until now. The exploitation of this ‘layered’

Heuslers might be assistive in various functional electronic materials applications.

5.5 References

1. I. Galanakis and P.H. Diedreich in Half-Metallic Alloys: Fundamentals and Applications,

Springer-Verlag Berlin Heidelberg, 2006.

2. D.D.Tang and Y.J. Lee in Magnetic Memory: Fundamentals and Technology, Cambridge
University Press, 2010.

3. C. Felser and G.H. Fecher in Spintronics : From Materials to Devices , Springer

Science+Business Media, Dordrech, 2013.

4. T. Graf, F. Felser and S.S.P. Parkin , Prog. Solid State Chem. 39, 1 (2011).

5. I. Galanakis in Heusler Alloys, edited by C. Felser and A. Hirohata, Springer Publishing

Switzerland, 2016.

6. N. Naghibolashrafi, S. Keshavarz, Vinay I. Hegde, A. Gupta, W. H. Butler, J. Romero,

K. Munira, P. LeClair, D. Mazumdar, J. Ma, A. W. Ghosh, and C. Wolverton, Phys. Rev.
B. , 93 , 104424, (2016).

7. W. Leiper, G.R. Mackay and C. Blauw, J. Appl. Phys. 49, 1552 (1978).

8. S. Ishida, S. Akazawa, Y. Kubo and J. Ishida, J. Phys. F: Met. Phys., 12, 6, (1982).

9. J. Pierre , R.V. Skolozdra and Y.V. Stadnyk, J. Magn. Magn. Mater. 128 , 1, (1993).

10. S. C. Lee, T. D. Lee, P. Blaha, and K. Schwarz, J. Appl. Phys. 97, 10C307, (2005).

11. . Barth, G. H. Fecher, B. Balke, S. Ouardi, T. Graf, C. Felser, A. Shkabko, A.

Weidenkaff, P. Klaer, H. J. Elmers et al., Phys. Rev. B 81, 064404 (2010).

12. J. Barth, G. H. Fecher, B. Balke, T. Graf, A. Shkabko, A. Weidenkaff, P. Klaer, M.

Kallmayer, H.-J. Elmers, H. Yoshikawa, S. Ueda, K. Kobayashi, and C. Felser, Philos.
Trans. R. Soc. A 369, 3588 (2011).

124
13. L.Bainsla and K.G. Suresh, Current Appl. Phys., 16, 1, (2016).

14. T. Nobata, G. Nakamoto, M. Kurisu , Y. Makihara, K. Ohoyama and M. Ohashi, J. Alloy.

Compd. , 347 , 1, (2002).

15. P. Rhodes and E.P. Wohlfarth, Proc. Roy. Soc. Lond.A., 273, 247 (1963).

16. P.G. Van Engen, K.H.J. Buschow, R. Jongebreur and M. Emran, Appl. Phys. Lett. , 42 , 2
(1983).

17. K.H.J. Buschow, P.G. Van Engen and R. Jongebreur, J. Magn. Mag. Mat. , 38 , 1 (1983).

18. R.A. de Groot, F.M. Mueller, P.G. Van Engen and K.H.J. Buschow, Phys. Rev. Lett. 50,
25 (1983).

19. M.J. Otto, R.A.M. van Woerden, P.J. van der Valk, J. Wijngaard., C.F. van Bruggen , C.
Haas and K.H. J. Buschow, J. Phys. Condens. Matter, 1 , 2341 ( 1989).

20. M.I. Katsnelson, V.Y.Irkhin, L.Chioncel, A.I. Lichtenstein and R.A. de Groot , Rev. Mod.
Phys. , 80 , 2 (2008).

125
 CHAPTER 6

CONCLUSION

Following our theorist colleagues recommendations, we obtained an extensive list of

possible candidates of Heuslers with potential half metallic bandstructure. Among all of this

large number of possible compounds, the Fe2TiSb attracted our interest as the initial literature

survey did not result in any scientific report on the experimental process of this metallic

compound and also according to our density functional theory predictions this compound had a

half metallic bandstructure with a relatively low moment of 1 Bohr magneton per formula unit,

making it a viable candidate for spintronic explorations.

Our first processing attempts through arc melting and eventual homogenization

treatments showed that at 900 °C the alloy was fully homogenized with equiaxed grains forming.

The scanning electron microscopy and subsequent energy-dispersive X-Ray spectroscopy studies

reveled that in fact these newly formed grains were deviating from the initial stoichiometry of

the Fe2TiSb, and were actually showing a Fe1.5TiSb composition. Further analyses confirmed

that this stoichiometry stood the tests and also indicated that indeed iron-rich precipitates were

forming between the Fe1.5TiSb grains, accounting for the absent iron with respect to the starting

stoichiometry of Fe2TiSb.

The processing of the Fe1.5TiSb confirmed the single-phase nature of the alloy,

establishing it as a possible compound. The X-Ray diffraction studies revealed a L21 structure, in

line with the Heusler character of the compound.

126
 This was also corroborated through more elaborate diffractive methods such as electron

backscatter diffraction maps and transmission electron microscopy selected area diffraction.

The magnetic measurement revealed a material void of any significant response with a

paramagnetic behavior. Such a result inspired the possible assumption of the Fe1.5TiSb

compound being a semiconductor. This called for a theoretical treatment of such compound

through DFT methodology and subsequent bandstructure derivations, which eventual revealed

that indeed the consequent bandstructure is of a semiconductor.

An even more important aspect to the theory of this compound was that according to the

Slater-Pauling rule, the Fe2TiSb compound should have an expected moment of 1 µB/f.u., and the

hypothetical FeTiSb a value of -1 µB/f.u. Therefore if we consider the Fe1.5TiSb, to be “lying”

intermediately as far as the iron content is concerned between the Fe2TiSb and FeTiSb

hypothetical compounds, the corresponding moment which is zero, is also an average of the

hypothetical 1 and -1 µB/f.u. values as well. This is a novel finding as it indicates that the

Fe1.5TiSb is in fact ‘layered’ between its two full and half Heusler counterparts and still follows

the Slater-Pauling rule. Our theorists, led by Professor Butler, consequently named this

compound a newly discovered layered Slater-Pauling phase.

Further on, we tried to process the half Heusler FeTiSb compound, which during our

various efforts happened to be an alloy system composed of two different phases one being the

previously observed Fe1.5TiSb alongside some Ti/Sb rich phase. The theoretical energetics

studies through cluster expansion also interestingly hinted that both Fe1.5TiSb and a Ti/Sb rich

phase are also energetically stable phases, in line with our experimental observations.

127
 The scientific significance of such conclusions lies in the possibility of a yet unexplored

stoichiometry available in the Heusler family, which is intermediate between full and half

stoichiometries and therefore has a 1.5:1:1 atomic ratio between the ternary X, Y and Z elements.

The tetrahedral positions in such compound are in fact vacant in the sense that there are 6 atoms

of the X element present. This possible subclass of Heuslers could be an interesting subject of

study for both theoretical condensed matter physicists and experimental materials scientists, in

order to exploit potentially novel functional materials in the family of Heusler compounds.

The next level of our research concentrated on a quaternary system: As we already had explored

the Fe-Ti-Sb Heusler alloys, we decided to take a step further on and add cobalt to this system

and perform research on the Fe-Co-Ti-Sb possible Heuslers. The impetus this phase of our

research came from our previously mentioned findings and also lack of extensive research on

this specific quaternary system with some instances hinting towards of a multiphase alloy system

present alongside some reports avoiding the heat treatment of antimony-rich system altogether to

bypass the possible loss of the element due to its volatility.

Our course of research was concentrated on the half Heusler section of the alloy system

with adding iron as the main doping element to the ternary CoTiSb compound and gradually

reaches the iron-rich FeTiSb alloy.

The study revealed that there indeed was a half Heusler compound, Fe0.5Co0.5TiSb,

observable with an established chemical composition and a C1b signature. Various

characterization techniques confirmed the homogeneity of the alloy, and the half Heusler nature

of it. There were no reports of any single-phase compound in this quaternary system previously.

128
 Magnetic analysis eventually showed an antiferromagnetic behavior which was also in

line with the theoretical energetic studies showing a favorable competition for the stability of a

C1b Fe0.5Co0.5TiSb compound becoming antiferromagnetic rather than ferromagnetic.

No other compound were observed in this system and there were a prevalence of the

CoTiSb phase in the cobalt-rich side of the alloys and in the iron-rich side there were

competition between the possible C1b and titanium/antimony phases appearing in the

microstructure, the latter both observed and theoretically found to be stable in our previous

research on the Fe-Ti-Sb system.

The last phase of our research was based on the Fe1.5TiSb layered Heusler we discovered

before: Following this interesting finding, Professor Butler took interest in a hypothetically

similar compound, Co 1.5TiSn. Initial first-principle studies by Professor Butler and our

colleagues in the Northwestern University resulted that a possible layered Heusler of the named

composition would attain a half metallic bandstructure with an estimated saturation moment of

0.5 µB/f.u.

Our literature survey showed the scarcity of microstructural analysis on the Co-Ti-Sn

system with the Co2TiSn being more exploited as a half metallic compound. No instance of the

Co1.5TiSn was found with an exception of a neutron diffraction study of the CoTiSn alloy system

leading to a possible evolution of a structure with a Co1.5TiSn and tin and titanium impurities.

Our experimental efforts showed that not only the Co1.5TiSn was achievable with all the

structural hallmarks of a Heusler compound through extensive characterization methods like

SEM, EDS, EBSD and XRD, but also low temperature moment of the compound was 0.6 µB/f.u.,

being in very good agreement with its half metallic value.

129
 The subsequent metallurgical tests showed that in fact a FCC solid solution region exists

between Co1.5TiSn up to Co2TiSn, in which the microstructure is single-phase and a L21

signature in the structure is visible.

Below the Co1.5TiSn stoichiometry, the alloy system starts decomposing into regions of

the dominant Co1.5TiSn with impurities of tin and titanium detectable under optical and electron

microscopes, further corroborating the reports on the neutron diffraction of the CoTiSn alloy for

the very first time with visual microscopic evidence.

Even further interesting was the test of a relatively less known criterion for half

metallicity, called the Rhodes-Wolfarth model, on the Co1.5TiSn : The compound successfully

stood the test of the model in exhibiting a half metallic signature through the comparison of its

effective paramagnetic moment to its zero Kelvin saturation one. The ratio was less than one, a

signature of half metals. Also this unusually low ratio was observed for all the alloys within the

solid-solution region between Co1.5TiSn to Co2TiSn, hinting towards the possibility of a half

metallic ‘region’ instead of a singular ‘compound’ being present.

The future potential plan suggestions built on all the findings mentioned in this thesis

could be more elaborated analysis of the found phases, e.g. Fe1.5TiSb, Fe0.5Co0.5TiSb and

Co1.5TiSn.

Methods such as high-resolution transmission electron microscopy, neutron diffraction,

Mossbauer spectroscopy and nuclear magnetic resonance could be useful in further exploring

these newly found Heuslers, especially with regard to the possible detection of the location of the

vacancies in the layered Heuslers and also their distribution within the crystal lattice.

130
 Besides characterization, new processing methods could also be interesting pathways to

study these materials, especially thin film deposition methods which are naturally more

important in realizing viable electronic devices than the bulk synthesis. For this goal, physical

deposition methods such as pulsed laser deposition (PLD) or sputtering can be useful in

production of films of good quality and utilization of such films in more elaborate stacks leading

to potential spintronic memory devices.

Also at the moment further theoretical calculations, especially on the energetics and

cluster expansion of both Fe-Co-Ti-Sb half Heusler system and the ternary Co-Ti-Sn are under

way for the eventual article publication.

Lastly, I would recommend that other quaternary systems similar to the ones we covered

in this scope of research be considered for processing and characterization. Two which I

personally propose are FeCoTiGe and FeCoMnSb systems, both of which have never been

experimentally realized and therefore have good potential for an interesting metallurgical

evaluation.

131
 REFERENCES

1.F. Heusler, Verh. dt. phys. Ges. 5, 219 (1903).

2.T. Graf, F. Felser and S.S.P. Parkin , Prog. Solid State Chem. 39, 1 (2011).

3.M. Meinert, Ferrimagnetic Heusler Compounds, Bielefeld University, Department of Physics

(2011).

4.F. Heusler and W. Starck, Haupt E. Verh DPG 5, 220 (1903).

5.T. Nakamichi and C.V. Stager, J. Magn. Magn. Mater, 31, 85 (1983).

6.P.C. Canfield, J.D. Thompson, W.P. Beyermann, A. Lacerda, M.F. Hundley, E. Peterson,
Z.Fisk and H.R. Ott, J. Appl. Phys. 70, 5800 (1991).

7.J. Tobola, J. Pierre, S. Kaprzyk, R.V. Skolozdra and M.A. Kouacou, J. Phys. Condens.Matter
10, 1013 (1998).

8.J. H. Wernick, G. W. Hull, J. E. Bernardini, and J. V. Waszczak, Mater. Lett. 2, 90 (1983).

9.M.M. Kirillova, A.A. Makhnev, E.I. Shreder, V.P. Dyakina and N.B. Gorina, Phys. Stat. Sol.
(b)187, 231 (1995).

10. H. Nakamura, Y. Kitaoka, K. Asayama, Y. Nuki and T. Komatsubara, J. Phys. Soc. Jpn.
57, 2276, (1988).

11. R. Carey, D. M. Neumann, and M. L. Wears, Phys. Rev. B 62, 1520 (2000).

12. S. Sakurada and N. Shutoh, Appl. Phys. Lett. 86, 2105 (2005).

13. S.J. Patel, J.K. Kawasaki, J. Logan, B.D. Schultz, J. Adell, B. Thiagarajan, A. Mikkelsen
and C.J.Palmstrøm, Appl. Phys. Lett. 104, 201603 (2014).

14. A. Hirohata, J. Sagar, L.R. Fleet and S.S.P. Parkin in Heusler Alloys, edited by C. Felser
and A. Hirohata, Springer Publishing Switzerland, 2016.

132
15. 29. Y. Sakuraba, J. Nakata,M. Oogane, H. Kubota, Y. Ando, A. Sakuma and T. Miyazaki,
Jpn. J. Appl.Phys. 44, L1100 (2005).

16. Fitriani , R. Ovik , B.D. Long , M.C. Barma , M. Riaz , M.F.M. Sabri , S.M. Said and
R. Saidur, Renew. Sustain. Energy Rev. , 64, 635 (2016).

17. https://flowcharts.llnl.gov/. (2014).

18. J. Yang and F.R. , J. Electron. Mater. , 38, 7 (2009).

19. J.W.G. Bos and R.A. Downie, J. Phys.: Condens. Matter, 26, 433201 (2014) .

20. S. Chen, Z. Ren, Mater. Today, 16 387 (2013).

21. J.F.T. Young, Phil. Mag, 46, 291 (1923).

22. O. Heusler, Ann. Phys., 19 , (1934) .

23. A.J. Bradley, J.W. Rodgers, Proc. R. Soc. A. , 144 (1934).

24. J. Nuss and M. Jansesn , Z. Anorg. Chem. , 628 , 1152 (2002).

25. U. Eberz, W. Seelentag and H.U. Schuster,Z. Naturforsch., 35b , 1341 ( 1980 ).

26. V. Alijani, J.Winterlik, G.H. Fecher, S.S. Naghavi and C. Felser, Phys. Rev. B 83, 184428
(2011).

27. D. Jung, H.J.Koo and M.H. Whangbo, J. Mol. Struct. Theochem. , 527, 113 (2000).

28. P.G. Van Engen, K.H.J. Buschow, R. Jongebreur and M. Emran, Appl. Phys. Lett. , 42 , 2
(1983).

29. K.H.J. Buschow, P.G. Van Engen and R. Jongebreur, J. Magn. Mag. Mat. , 38 , 1 (1983).

30. R.A. de Groot, F.M. Mueller, P.G. Van Engen and K.H.J. Buschow, Phys. Rev. Lett. 50,
25 (1983).

31. I. Galanakis in Heusler Alloys, edited by C. Felser and A. Hirohata, Springer Publishing
Switzerland, 2016.

32. C. Felser, L. Wollman, S. Chadov , G.H. Fecher and S.S.P. Parkin in Heusler Alloys,
edited by C. Felser and A. Hirohata, Springer Publishing Switzerland, 2016.

33. J.M.D.Coey in Magnetism and Magnetic Materials, Cambridge University Press,2009.

133
34. P. Rhodes and E.P. Wohlfarth, Proc. Roy. Soc. Lond.A., 273, 247 (1963).

35. M.I. Katsnelson, V.Y.Irkhin, L.Chioncel, A.I. Lichtenstein and R.A. de Groot , Rev. Mod.
Phys. , 80 , 2 (2008).

36. M.J. Otto, R.A.M. van Woerden, P.J. van der Valk, J. Wijngaard., C.F. van Bruggen , C.
Haas and K.H. J. Buschow, J. Phys. Condens. Matter, 1 , 2341 ( 1989).

134