Synthesis and Characterization of Nanostructures

Synthesis and Characterization of

Nanostructures

By

Rafaqat Hussain

CIIT/SP04-PPH-006/ISB

Ph.D Thesis

In

Physics

COMSATS Institute of Information Technology

Islamabad- Pakistan

Spring, 2011

ii

COMSATS Institute of Information Technology

Synthesis and Characterization of Nanostructures

A Thesis Presented to

COMSATS Institute of Information Technology, Islamabad

In partial fulfillment

of the requirement for the degree of

Ph.D (Physics)

By

Rafaqat Hussain


Spring, 2011

iii


A Post Graduate Thesis submitted to the Department of Physics as partial

fulfillment of the requirement for the award of Degree of M.S/Ph.D (Physics).

Name Registration Number

Rafaqat Hussain CIIT/SP04-PPH-006/ISB

Supervisor

Dr. Ehsan Ullah Khan, (T.I.)

Professor in Physics

Co-Supervisor

Dr. Syed Tajammul Hussain

Director, Nanoscience and Catalysis Division

National Centre for Physics, Islamabad, Pakistan.

June, 2011

iv

Final Approval

This thesis titled

Synthesis and Characterization of Nanostructures By

Rafaqat Hussain

CIIT/SP04-PPH-006/ISB Has been approved

For the COMSATS Institute of Information Technology, Islamabad

External Examiner 1: ___________________________________________________

Prof. Dr. Asghari Maqsood, S.I

Meritorious Professor, Department of Physics, CESET, Islamabad

External Examiner 2: ___________________________________________________

Prof. Dr. Younis Nadeem

Department of Physics, Bahauddin Zakariya University, Multan

Supervisor: ___________________________________________________________

Prof. Dr. Ehsan Ullah Khan, T.I

Department of Physics, CESET, Islamabad

Co-Supervisor: ________________________________________________________

Prof. Dr. Syed Tajammul Hussain, PoP

Director NS & CD, National Centre for Physics, Islamabad

HoD: ________________________________________________________________

Prof. Dr. Mahnaz Haseeb

Department of Physics, Islamabad Campus, CIIT

Chairman:_____________________________________________________________

Prof. Dr. Sajid Qamar

Department of Physics, CIIT, Islamabad

Dean, Faculty of Sciences: ___________________________________________

Prof. Dr.Arshad Saleem Bhatti, T.I

v

Declaration

I Rafaqat Hussain, registration number CIIT/SP04-PPH-006/ISB hereby declare that I

have produced the work presented in this thesis, during the scheduled period of study. I

also declare that I have not taken any material from any source except referred to

wherever due that amount of plagiarism is within acceptable range. If a violation of HEC

rules on research has occurred in this thesis, I shall be liable to punishable action under

the plagiarism rules of the HEC.

Date: ________________ Signature of the student:

_____________________________

Rafaqat Hussain


vi

Certificate

It is certified that Mr. Rafaqat Hussain, registration number CIIT/SP04-PPH-006/ISB has

carried out all the work related to this thesis under our supervision at the Department of

Physics, COMSATS Institute of Information Technology, Islamabad and the work fulfills

the requirement for an award of PhD degree.

Date: ____________________

Supervisor:

___________________________

Prof. Dr. Ehsan Ullah Khan, TI

Professor in Physics

Co-Supervisor:

___________________________

Prof. Dr. Syed Tajammul Hussain

Director NS & CD, NCP, Islamabad

Head of Department:

_______________________

Department of Physics

vii

Dedicated

To

Holy Prophet Muhammad (Peace be upon him) and his

companions who laid the foundations of modern

civilization and paved the way for social, moral,

political, economical, cultural and physical revolution

xi

Table of Contents

1.1 Motivations................................................................................................................1

1.2 Current Status of the Field .........................................................................................5

1.3 Objectives ..................................................................................................................7

1.4 Summary of the thesis ................................................................................................7

2.1. The Magnetic Moment and Magnetisation ................................................................8

2.1.1 Origins of Magnetism in Bulk Materials ..............................................................8

2.1.2. Thin Film Magnetism ....................................................................................... 10

2.2 Demagnetizing Field ................................................................................................ 11

2.3 Magnetic Anisotropies ............................................................................................. 11

2.3.1 Magnetocrystalline Anisotropy .......................................................................... 12

2.3.2 Exchange Anisotropy ........................................................................................ 12

2.3.3 Shape Anisotropy .............................................................................................. 13

2.3.4 Other Contributions to Anisotropy ..................................................................... 15

2.4 Magnetic Domain .................................................................................................... 15

2.5 Reversal Mechanisms .............................................................................................. 16

2.5.1 (Coherent reversal) Stoner Wohlfarth model ...................................................... 17

2.5.2 Incoherent Reversal ........................................................................................... 18

2.5.3 Domain wall pinning mechanism or domain nucleation ..................................... 21

2.5.4 Reversal process in EB thin films ...................................................................... 23

2.6 Exchange Bias ......................................................................................................... 26

2.6.1 Earlier Theories ................................................................................................. 26

2.6.2 Technological Importance ................................................................................. 27

2.6.3 Recent development in the field of EB............................................................... 28

2.7. York Protocol ......................................................................................................... 29

2.7.1. Theory .............................................................................................................. 29

xii

2.7.1.1. Grain Size Distribution .............................................................................. 29

2.7.1.2 Blocking Temperature ................................................................................. 31

2.7.1.3 Measurement of EB from York Protocol ..................................................... 33

2.8 Training Effect in Exchange Bias ............................................................................. 34

2.9 Diluted Magnetic Semiconductors ........................................................................... 36

2.9.1 Types of interactions in DMS ............................................................................ 37

3.1. Synthesis Technologies ........................................................................................... 42

3.1.1. High Target Utilization Sputtering (HITUS) ..................................................... 42

3.1.2. Aerosol Assisted Chemical Vapor Deposition (AACVD) ................................. 44

3.1.2.1. Advantages of AACVD over conventional CVD ........................................ 45

3.2. Characterization Techniques ................................................................................... 46

3.2.1. X-Ray Diffractometer ....................................................................................... 46

3.2.2 Field Emission Electron Microscope (FESEM).................................................. 47

3.2.3. Zeiss Particle Size Analyzer ............................................................................. 48

3.2.4. Transmission electron microscopy (TEM) ........................................................ 49

3.2.5 Magnetometers .................................................................................................. 51

3.2.5.1. Vibrating Sample Measurement (VSM) ..................................................... 51

3.2.5.2 Alternating Field Gradient Magnetometer (AGFM)..................................... 53

3.2.5.3. Superconducting Quantum Interference Device (SQUID) .......................... 54

3.2.6. Rutherford Back scattering (RBS) .................................................................... 55

4.1 Introduction ............................................................................................................. 58

4.2 Experimental ........................................................................................................... 59

4.3 Results and Discussion ............................................................................................ 61

4.3.1 Effect of substrate cutting .................................................................................. 61

4.3.2 Effects of sample shape ..................................................................................... 63

Summary ....................................................................................................................... 64

5.1 Experimental ........................................................................................................... 66

5.1.1 Fabrication process and conditions .................................................................... 66

5.1.2 Setting process .................................................................................................. 67

xiii

5.1.2.1 York Protocol ............................................................................................. 67

5.2 Results and Discussion ............................................................................................ 69

5.2.1 Grain Size Analysis ........................................................................................... 69

5.2.2 Exchange Bias ................................................................................................... 72

5.2.3 Blocking Temperature ....................................................................................... 74

5.3 Training Effect......................................................................................................... 77

5.3.1 Sample Preparation............................................................................................ 78

5.4. Results and Discussion............................................................................................ 79

Summary ....................................................................................................................... 82

6.1 Introduction ............................................................................................................. 84

6.2 Nickel Doped TiO2 Thin Films ................................................................................ 85

6.2.1 Experimental ..................................................................................................... 86

6.2.2 Results and Discussion ...................................................................................... 87

6.2.2.1 XRD analysis .............................................................................................. 87

6.2.2.2 Rutherford Back Scattering ...................................................................... 92

6.2.2.3 Scanning electron microscopy ..................................................................... 96

6.2.2.4 Magnetic properties .................................................................................... 98

6.3 Cobalt Doped TiO2 Thin Films .............................................................................. 101

6.3.1 Experimental ................................................................................................... 101

6.3.2 Results and Discussion .................................................................................... 102

6.3.2.1 XRD Analysis ........................................................................................... 102

6.3.2.2 Rutherford Back Scattering ....................................................................... 105

6.3.2.3 Scanning electron microscopy ................................................................... 106

Summary ..................................................................................................................... 112

7.1 Exchange Bias ....................................................................................................... 113

7.2 Diluted magnetic semiconductors .......................................................................... 115

7.3 Future Work .......................................................................................................... 117

References ................................................................................................................... 131

xiv

List of Publications ...................................................................................................... 131

LIST OF FIGURES ....................................................................................................... xv

LIST OF TABLES .................................................................................................... xviii

xv

LIST OF FIGURES

_____________________________________________________________

Fig 1. 1 a) magnetic semiconductor b) a nonmagnetic semiconductor c) a dilute magnetic

semiconductor .................................................................................................................4

Fig 2.1: Schematic representation of a magnetisation curve for a typical ferromagnet,

showing the hysteretic behavior and the main characterizing parameters….……………10

Fig 2.2: Hysteresis loop of CoO/Co particles after sample was cooled through TN of CoO

in a saturating magnetic field ......................................................................................... 13

Fig 2. 3: Domain formation and minimization of energy ................................................ 15

Fig 2.4: Coherent rotation of magnetisation by rotation – the Stoner-Wohlfarth model. . 17

Fig 2.5: Variation of coercivity with axial ratio using the Stoner-Wohlfarth model of

reversal. ......................................................................................................................... 19

Fig 2.6: Fanning (a) and coherent (b) reversal modes for an N = 2 chain of spheres. ...... 20

Fig 2.7: Magnetization Reversal in Exchange Bias Thin Films....................................... 25

Fig 2. 8: Magnetization Reversal in soft magnetic film .................................................. 25

Fig 2.9: Schematic of the energy barrier reversal, showing the proportion of AFM grains

set parallel or anti-parallel to the original set direction. .................................................. 32

Fig 2.10: Comparison of Blocking Temperature (TB) measured from a) Conventional

method b) York Protocol ............................................................................................... 32

Fig 2.11: Schematic of the grain size distribution after the setting of the AFM and cooling

to a temperature at which the AFM is thermally unstable. .............................................. 33

Fig 2. 12: Consecutive hysteresis loops of a Co − CoO system measured with torque

balance. The observed overshoot is an instrumental effect. ............................................ 34

Fig 2. 13: Schematic showing the FM and AFM sublattice magnetizations in an exchange

bias system where the AFM anisotropy has biaxial symmetry during the 1st and 2nd

hysteresis loop measurements. ...................................................................................... 36

xvi

Fig 2. 14: Interaction of two bound magnetic polarons. The polarons are shown with gray

circles. Small and large arrows show impurity and hole spins, respectively. Shaded region

shows the effect of two BPMs on impurity spins. ........................................................... 40

Fig 3. 1: Schematic representation of HiTUS sputtering technology……………………42

Fig 3. 2: Schematic representation of AACVD .............................................................. 46

Fig 3. 3: Schematic Diagram of a Transmission Electron Microscope ............................ 50

Fig 3. 4: Schematic of a standard VSM. ......................................................................... 52

Fig 3. 5: Schematic of a standard AGFM. ...................................................................... 53

Fig 3. 6: Schematic diagram Josephson junction ............................................................ 55

Fig 3. 7: Schematic diagram of RBS basic function ....................................................... 56

Fig 4. 1: Sample structure………………………………………………………………..59

Fig 4. 2: SEM images of the edges of the three sample types ......................................... 60

Fig 4. 3: Hysteresis loops for three samples produced by (a) cutting with a diamond

scribe, (b) depositing through a mask and (c) cutting with an ultrasonic cutter ............... 62

Fig 4. 4: Hysteresis curve a) and b) shows the effect of sample shape ............................ 63

Fig 5. 1: Sample structure with Mn doping……………………………………………...67

Fig 5. 2: a) Schematic diagram and b) measurement steps of the York protocol. 14

......... 68

Fig 5. 3: TEM Image ..................................................................................................... 70

Fig 5. 4: Graph showing the log-normal distribution of grain sizes................................. 71

Fig 5. 5: Typical hysteresis loop obtained using the York protocol. ................................ 73

Fig 5. 6: Typical loop shift for different temperatures under a constant reverse field. ..... 75

Fig 5. 7: Measurement of Blocking Temperature (TB) by York Protocol ........................ 76

Fig 5. 8: Schematic diagram of sample structure ............................................................ 78

Fig 5. 9: Training effect with a) NiCr seed layer b) Cu seed layer .................................. 79

Fig 5. 10: a) comparison of bias voltage vs. training effect for NiCr and Cu under layer b)

grain size vs. training effect for NiCr under layer ........................................................... 80

Fig 6. 1: a)Topography image and corresponding b) MFM images of Ni doped TiO2 thin

films ……………………………………………………………………………………85

Fig 6. 2. XRD pattern of Ni (2%) doped TiO2 ................................................................ 88



xvii


Fig 6. 7: XRD pattern of Ni (15%) doped TiO2 thin films .............................................. 90

Fig 6. 8: Crystal structure of [Ni2Ti2(OEt)2(l-OEt)6(acac)4] ............................................ 91

Fig 6. 9: Comparison of experimental and simulated RBS spectra on Ni doped TiO2 thin

film................................................................................................................................ 94

Fig 6. 10: RBS spectra of Ni doped TiO2 thin films with various concentrations ............ 95

Fig 6. 11: SEM images of Ni doped TiO2 thin films with a) 2wt.% Ni b) 4wt.% Ni c)

6wt.% Ni d) 8wt.% Ni and e) 15wt.% Ni doping ............................................................ 97

Fig 6. 12: Magnetic moment of Ni doped TiO2 thin films at 100K ............................... 100

Fig 6. 13: Magnetic moment of Ni doped TiO2 thin films AT 300K ............................. 100

Fig 6. 14: XRD pattern of Ni doped TiO2 thin films with a) 2wt.% Ni b) 4wt.% Ni c)

6wt.% Ni d) 8wt.% Ni and e) 15wt.% Ni doping .......................................................... 104

Fig 6. 15: RBS spectra of Co doped TiO2 thin films with various concentrations ......... 105

Fig 6. 16: SEM images of Co doped TiO2 thin films with a) 2wt.% Co b) 4wt.% Co c)

6wt.% Co d) 8wt.% Co and e) 15wt.% Co doping ........................................................ 108

Fig 6. 17: Hysteresis loop of Co doped TiO2 thin films with various Co doping

concentrations ............................................................................................................. 109

xviii

LIST OF TABLES

_____________________________________________________________

Table 2. 1: calculated values of c for elongated iron particles26

................................... 19

Table 5. 1: The results obtained for the average grain size at different bias voltages…...72

Table 5. 2: The results obtained for and c at different bias voltages....................... 73

Table 5. 3: The results obtained for TB at different reverse fields. ................................... 77

Table 5. 4: Amount of training for Cu and NiCr underlayer ........................................... 81

Table 6. 1. Lattice parameters, Cell volume and crystallite size calculated from XRD

data……………………………………………………………… ………….92

Table 6. 2: Ni concentration and film thickness as calculated from RBS spectra ............ 95

Table 6. 3: Magnetic moment of Ni doped TiO2 thin films at 100K and 300K ............. 101

Table 6. 4: Co concentration and film thickness as calculated from RBS spectra .......... 106

Table 6. 5: Magnetic moment of Co doped TiO2 at 300K ............................................. 111

viii

ACKNOWLEDGEMENTS

First of all I thank Almighty ALLAH, the most Merciful, and the most

Beneficent, Who blessed me with sound health and opportunity to complete this research

work successfully. I pay gratitude to my supervisors Prof. Dr. Syed Tajammul Hussain

and Prof. Dr. Ehsan Ullah Khan (T.I.) for their kind supervision, guidance and

cooperation during this research work. I also appreciate continuous support from NCP

and Department of Chemistry, QAU, Islamabad during the course of my PhD work. I pay

my gratitude to all the lab fellows and staff members in NCP and QAU for their good

wishes and support to my research. My whole heartedly appreciation to my foreign

supervisor Prof. Kevin O’ Grady, Department of Physics, the University of York, UK for

his kind guidance, encouragement and cooperation during my six months stay at the

Department of Physics, the University of York, UK to perform experimental work.

Without his careful consideration and encouragement this research work could have

never been completed. I am obliged to Dr. Barbara Kaeswurm for her guidance and keen

interest to help me to work in the project. I am also thankful to Mr. Nick Cramp and all

the other lab fellows in the University of York for their help in the project and enjoyable

company.

I am thankful to Dr. Tadachika Nakayama, the University of Nagaoka, Japan and

Ms. Naila Jabeen Berkeley labs, USA for their help to characterize my samples.

I commend the cooperation and the supports extended to me by Prof. Arshad

Saleem Bhatti, Dean Faculty of Sciences and appreciate his arduous efforts to prop up

post graduate research program. I am gratified to Dr. Ishaq Ahmad, Head of the

Department of Physics, for creating a lively scientific environment.

Extraordinary credit to the authorities of HEC, Government of Pakistan, for

bestowing upon me the scholarship under Indigenous 5000 Scheme and reverend

authorities in the University of York for giving me opportunity to work in Magnetism

Lab in the University of York.

I, from the core of my heart, thank all of my fellows at CIIT, NCP and QAU for

their moral and manual help throughout this research work. I am enormously grateful to

my loving parents, brothers and sisters who always pray for my success in every walk of

life.

ix

ABSTRACT


This thesis is mainly focused on synthesis and characterization of (magnetic)

nanostructures in the form of multilayers and magnetic oxides thin films for spintronics

applications. Exchange bias phenomenon which has a critical role in

ferromagnetic/antiferromagnetic multilayer system was studied experimentally with a

theoretical understanding of very recent model of exchange bias namely York Model.

Standard IrMn and CoFe multilayer system (Si/Cu/IrMn/CoFe/Ta) was fabricated using

High Target Utilization Sputtering (HiTUS) to study various aspects of exchange bias.

Effect of Mn doping showed a decrease in the blocking temperature. Chemical reaction

of Mn at the interface and diffusion of Cu from the under layer in IrMn layer were

considered to be cause of this decrease. Training effect in exchange coupled IrMn and

CoFe multilayer thin films was investigated for varying grain size that was controlled

during the fabrication process through bias voltage. It was observed that smallest grains

gave rise to a larger training effect as larger anti-ferromagnetic grain volumes give rise to

thermally stable bias fields and consequently smaller training effects. The result is found

reproducible and in agreement with the literature. The effects of nucleation were also

studied. It was determined that nucleation arises from both sample shape effects and the

process used to cut the sample. The obtained results showed that sample edge roughness

leads to a distribution of nucleation fields and hence changes the shape of the hysteresis

loop. It was concluded that the best way to cut samples of nucleation controlled materials

is by cracking for the application in spintronics devices.

Second part of the study was about Ni and Co doped TiO2 diluted magnetic

semiconductors thin films grown by Aerosol Assisted Chemical Vapor Deposition

(AACVD). AACVD method was adopted for synthesizing these films due to certain

advantages over other chemical routes. Further, synthesis routes may vary various

properties and there are only a handful reports in the literature in which AACVD method

x

was employed to synthesize diluted magnetic oxides. Ni and Co doped TiO2 films were

prepared at 450 C and 650 C respectively with Argon as a carries gas. XRD, FESEM

and RBS were carried out to see phase, morphology, and stoichiometry and film

thickness. Magnetic properties of the films were investigated using SQUID. Ni and Co

doping resulted in ferromagnetism in TiO2 at room temperature attributed to the

formation of Bound Magnetic Polaron (BMP).

Chapter 1

1

Chapter 1

Introduction

1.1 Motivations

The study of the materials at nano-scale has been a focus of recent investigations.

But to initiate, what is nano? Here we do not go into definitive details. Simply,

nanoscience is an emerging area of science and technology. It has gotten attention from

researchers all over the world. The word nano describes physical length scales that are

equivalent of billionth of a meter in length.

The increasing interest in the nanomaterials is because of the numerous

possibilities that they can induce when they are on atomic scales. Nanomaterial can

induce significant change in the mechanical, optical, chemical and magnetic properties,

which are very dissimilar in comparison with bulk materials of the same composition.1 In

this respect different fields like physics, chemistry, biology and engineering, strived to

explain various phenomena that these materials exhibit.2

In the past, materials science focused basically upon utilization of the natural

elements (iron, silicon, etc.) for developing new compounds. As a result scientists have

attained knowledge of making devices using artificial structures in which the atoms are

deposited layer by layer and later on they managed to redesign the device structures to

control the properties at molecular level. This allowed fabricating the system with novel

properties. With these technological advances it has become possible now to make

artificial nanometer systems in which the effects of the quantum confinement is

pronounced.

Chapter 1

2

How can nanoscale materials be used to improve our lives whether in health,

environment, energy needs of the day and improved performance of existing electronic

devices etc.? The quest for answer compels us to make research in this field.

Investigation on nanoparticles has also made its contribution in many other fields as well.

The size of the devices is continuously reducing, while the speed and efficiency is

improving. The advancements are being made and new concepts for reducing size,

power utilization and exploring multi functions of material are being investigated

regularly. The use of the spin electrons, hole, nuclei, or ions to explore and enhance new

functionalities analog and digital electronics is one of the favored topics today.3-4

The

charge, mass and spin of the electron lays the foundation for the information technology,

data storage and many other areas being in use at present time. Semiconducting materials

are in used for construction of Integrate circuit and high-frequency devices.4-5

On the

other hand, mass storage of information is carried out by magnetic recording such as

magnetic tapes, hard disks and optical discs. All this can be achieved by using spin of the

electron in a ferromagnetic metal. 6-7

So far the charge and spin is used separately. However the combination of this

two degree of freedom may bring an enhancement in the performance in current

technology. It may also bring new functionalities which are not possible by using one out

of them. If we can combine these two properties we can actually store and process the

information together. On the other hand, it is possible to inject spin-polarized current to

control the spin state of carriers into semiconductors, which may allow us to work on

qubit (quantum bit) operation which is essential for quantum computing. This branch of

the electronic in which we play with both charge and spin of electron at the same time is

called ―spintronics‖.8

Due to the difficulties to integrate such ferromagnetic materials with conventional

semiconductors being used, such as silicon (Si) and gallium arsenide (GaAs), these

devices could not be made yet. It is because of huge difference in physical and chemical

properties, crystal structure and lattice parameters between these two classes of materials.

Chapter 1

3

Further, Si and GaAs do not have magnetic ions and have nonmagnetic properties. To

achieve a meaningful variance in the energy between the two possibilities of orientation

of spin (up and down) the required applied magnetic field would be too high for common

use.9

In some ferromagnetic semiconducting materials (Fig 1.1a) magnetism and

semiconducting properties are known to exist simultaneously, such as ferromagnetic and

ferrimagnetic semiconducting spinels and europium chalcogenides 10-11

. These materials

have a periodic array of magnetic element. In practice, these classes of semiconductors

are difficult to fabricate and are incompatible with the current industrial semiconductors

like Silicon and Gallium arsenide. The reason is mismatch in the crystalline structure of

these materials and have low Curie Temperatures (Tc) which is about 100 K.9

Diluted magnetic semiconductors (Fig 1.1c) can be one possible solution of this

problem. When small concentrations of magnetic ions are doped into the non-magnetic

host semiconductors it can give sufficiently high Curie temperature as far as the

theoretical predictions are concerned (Fig 1.1b). This category of semiconductor is then

known as dilute magnetic semiconductor or (DMSs). In recent time many non-magnetic

impurities have also shown ferromagnetism at room temperature.

It provides the possibility of using and studying a variety of magnetic and

magneto-optical phenomena which are seldom present in conventional non-magnetic

semiconductor. Recently a large number of investigations have been focused in the

elaboration of new DMS material in different semiconductor host.12

These efforts have

been made to develop ferromagnetism on room temperature to find a new class of

spintronics devices such as transistors, spin valves, magnetic sensors, light emitting

diodes, non-volatile memory, optical isolators, logic devices and ultra-fast optical

switches. The predicted advantages of these spintronics devices will be of greater

efficiency, higher speed, and better stability, in addition to the low energy requirement

for flipping a spin.4

Chapter 1

4

Fig 1. 1 a) magnetic semiconductor b) a nonmagnetic semiconductor c) a dilute magnetic

semiconductor9

Diluted magnetic semiconductors are the class of materials which have the spin

polarized electrons retaining the semiconducting properties, along with the magnetism.

However, it is essential to enhance Curie temperature well above the room temperature to

make them useful for their application in spintronics devices.9

Besides DMS, multilayered structures of ferromagnetic antiferromagnetic

materials coupled together have a range of application in the spintronics devices such as

magnetic random access memory (MRAM), spin valve etc. The working principle of

these structures is mainly based on Exchange Bias (EB). Exchange bias is the

phenomenon in which the characteristic hysteresis loop which determines the properties

of the magnetic materials is shifted on one side of the origin.

Exchange bias played a vital role in the development of the spin valve.13

Without

this the giant magneto resistant (GMR) read head would not have been possible, and

consequently the storage densities of the modern hard disk drive (HDD) would not have

been technologically feasible. The development of MRAM has also seen considerable

focus. This technology would compete with current SRAM and DRAM technologies and

the development of which would bring the GMR sensor into another large market.

Element 1: Nonmagnetic Element 2: Nonmagnetic Element 3: Magnetic

Chapter 1

5

To understand the operation of the spin-valve a basic understanding of GMR is

required. Thompson13

has recently written an in depth review on this topic. In

Ferromagnetic materials (FMs) spin up and spin down electrons experience different

probabilities of scattering. This is due to a difference in the number of each electron in

the d band, known as a spin-split structure. As such it can be considered that the current

of spin up electrons is separate to that of the spin down electrons, as one current is

favored and the other is not, they are called the majority and minority electrons. In a

simple FMs bi-layer, two spin channels can be considered. In the first case, when the

FMs magnetizations are parallel, one spin channel carries only the majority electrons

whilst the other carries only minority electrons. In such a case the majority electron spin

channel is greatly favored for carrying charge and as such the overall resistance is low. In

the second case, when the FMs magnetizations are anti-parallel, both spin channels must

carry both majority and minority electrons. As such neither channel is favored, as they

have equal resistance, and so the overall resistance is higher.

For a spin-valve device to be possible a pinned and free ferromagnetic (FM) layer

is required. Due to exchange bias this is possible. An antiferromagnetic (AFM) layer is

used to pin one of the FM layers. The exchanged biased bi-layer is then separated with a

spacer that destroys any exchange bias that could arise between the two FM films. This

allows for one of the FM layers to maintain a constant magnetisation, whilst the other is

free to rotate within an applied field.

1.2 Current Status of the Field

The current state of exchange bias was reviewed and challenged by O‘Grady et.

al14

with the proposal of new definitions and explanations for a number of phenomena

associated with the magnetic measurement and characterization of sputtered

polycrystalline thin films of exchange coupled ferromagnetic/antiferromagnetic bilayer

systems in general.

Chapter 1

6

The largest of contributions by O′grady et al was that of the York Protocol14

, a

series of steps in which the magnetic history of samples could be controlled and therefore

reproducible measurements made. This allowed for comparable measurements of effects

that could not previously be compared. This also gave rise to a new definition of the

blocking temperature (TB) as well as a new explanation for the main contributing factors

to the value of Exchange field ( ), both of which will be elaborated on in later sections.

This knowledge has allowed the design of AFM/FM materials for specific applications

and setting conditions.

It is also very important to investigate the reproducibility of the measurements.

Also, the errors and the losses in the magnetic energy to make the devices more efficient

are essential to study. Very small parameters may bring a big impact on the functionality

of the devices e.g. the sample fabrication techniques currently used in the industry is

ultrasonic cutting. This thesis will provide a discussion on the best fabrication techniques

to be applied.

The present thesis is a discussion on the determination of the structural and

magnetic properties, of the synthesis of transition metal doped TiO2 thin films by Aerosol

Assisted Chemical Vapour Deposition (AACVD). There are many advantages of using

this technique such as fast evaporation of precursor, relatively shorter delivery time and

higher deposition rate, low cost and precise stoichiometry.15-17

Due to its versatility to use

a variety of precursors and the reaction environment, this method allows us to control the

size, shape, and size distribution of grains, which are hard to achieve through other

fabrication techniques. Until recently a great attention is given to synthesize diluted

magnetic semiconductor nanostructures from various chemical methods and physical

techniques.18-22

However direct chemical approaches are generally compatible with large-

scale production. Despite the fact that there is huge uncertainty in understanding the

origin of room temperature ferromagnetism in doped semiconductors with different

dopant as for as the small curie temperature is concerned than the theoretically predicted

value.9 This has opened a need to investigate the new synthesis routes to achieve the

Chapter 1

7

theoretical values of the magnetization to make a better understanding of factors affecting

on it.23

1.3 Objectives

The objective of the present research can be summarized as follows.

Investigate the effect of Mn doping on the Blocking temperature (TB).

To study the Sample Shape and fabrication process effect on Exchange Bias (EB)

Grain size effect on Training Effect

Synthesis and characterization of Ni and Co doping in TiO2 by AACVD

1.4 Summary of the thesis

A brief outline of the content in the thesis is listed below.

Chapter 2 is devoted to introduction and literature review on history of

magnetism, the related articles on exchange bias and the description of basic ideas to

establish a background necessary to understand the discussion chapters.

Chapter 3 describes the experimental setup/characterization facilities, working

principals and how their use in current research work was carried out.

Chapter 4 is devoted to results and discussion on the effect of shape and the

fabrication process on exchange bias (EB)

Chapter 5 describes effect of Mn doping in the antiferromagnetic (AFM) layer on

the Blocking Temperature (TB). Effect of grain size on the training effect is also included

in this chapter.

Chapter 6 is devoted to the synthesis of Ni and Co doping in TiO2 by AACVD.

The structure and magnetic properties are also discussed.

Chapter 7 is devoted to conclude the output of the research and the suggestions

for the future work.

Chapter 2

8

Chapter 2

Theory of Magnetism in Exchange Bias and Diluted

Magnetic Semiconductors

2.1. The Magnetic Moment and Magnetisation

Magnetic moment μ=|μ| is the most fundamental property in magnetism. It can be

expressed in terms of an analogy to a simple current loop in classical picture 24

. However,

in reality, the magnetic response of materials is quantum mechanical phenomenon which

arises from spin interaction of unpaired electron. In ferromagnetic materials, the unpaired

spins interact via indirect exchange mechanism and hence give rise to the magnetic

moment.

In bulk materials another most important characterizing parameters is the

magnetisation, M= |M|. Magnetisation is defined as the magnetic moment per unit

volume, μ/V. However, this statement is valid for homogeneous systems only. Actual

systems, such as magnetic thin films used in data storage are rarely homogeneous. Single

domain Co-alloy thin films are used for storage material. The system of Co-alloy thin

films are highly non-uniform as the grains of the thin film are decoupled from one

another which is a need to increase the storage capability.

2.1.1 Origins of Magnetism in Bulk Materials

The magnetic materials can be divided into three distinct classes owing to their

response in an external field. The weakest form of response is known as diamagnetism

and arises due to change in orbital velocity of electron in external field HA; hence all

materials with paired core electrons possess a diamagnetic component. Hence all

materials are diamagnetic. The change in velocity opposes the field that causes it and

Chapter 2

9

therefore susceptibility χ, the ratio of magnetisation to field, is negative. The diamagnetic

response is linear, independent of temperature and decreases with the magnitude of the

applied field. In general, the effect is very weak but in small particles and thin metallic

films it can be significant.

Another class of magnetic materials is known as paramagnetic materials. Like

diamagnetic behavior, the response to an applied field is positive although it is relatively

weak. Paramagnetism is temperature dependent in that the susceptibility χ, the ratio of

magnetisation to field, varies with the inverse of temperature. The origin of

paramagnetism is the fixed dipole moments atom or molecule in a material from there

being no complete cancellation of the spin and orbital components of angular

momentum25

. In the absence of any external field, all the moments are randomly oriented

and so the net magnetisation is zero. Eventually the moments will all align, typically at

high field ~ 10T and low temperatures T=4.2K. However, the degree of alignment is

limited owing to thermal effects that will attempt to randomize the orientation of the

moments and hence the paramagnetic effect is weak but increases with the magnitude of

the applied field.

Ferromagnetic materials (Fe, Co and Ni) are the most striking manifestation of the

magnetism due to their spontaneous magnetization that does not require an applied field

in order to be induced. The magnetic moment μ arises from the ordering of the atomic

spins via quantum mechanical exchange coupling between them. A ferromagnet is

characterized by their most prominent phenomenon known as hysteresis (Fig 2.1).

Hysteresis, from the Greek hysteresis, describes the situation where an effect lags behind

its cause. In the field of magnetism, this is manifested in the lagging of the magnetisation

M behind a swept applied field HA and it is this behavior that is central to the use of

ferromagnetic materials as data storage media.

Chapter 2

10

Fig 2.1: Schematic representation of a magnetisation curve for a typical ferromagnet,

showing the hysteretic behavior and the main characterizing parameters26

For a typical hysteresis loop plot, the applied magnetic field A and normalized

magnetisation M/Ms are used as the abscissa and the ordinate respectively. The

normalization is important to eliminate the large error associated with measuring the

magnetically-active volume of a thin-film sample. A complete loop contains two

important parameters as indicated in Fig 2.1: the remanent magnetisation (remanence) Mr

and the coercive field (intrinsic coercivity) c. The properties of both can be used to

assess the suitability of a material for use as a potential storage medium. The remanence

constitutes an effective ‗memory‘ for the material and the coercivity is the field required

to reduce the magnetisation to zero.

2.1.2. Thin Film Magnetism

In the initial applications of thin-film recording media, the reversal mechanism

was primarily dictated by an irreversible process induced by domain wall motion and the

associated pinning of the domain walls by strains27

or inclusions28

. With the rapid

advancement of recording media technology, the reduction in grain size has seen a move

towards single domain-type grains which has also resulted in an enhanced coercivity. In a

medium comprising single domains with uniaxial anisotropy, Stoner-Wohlfarth type

Chapter 2

11

behavior results and therefore there is a clear switch between two magnetisation states.

In the ideal Stoner-Wohlfarth case, a square hysteresis loop will come as a consequence

of having a system of moments with near-perfect orientation. Such a requirement is hard

to achieve in longitudinal media where all the moments lay in-plane and a circumferential

texture is necessary. It is not possible to obtain such a texture that will give the optimum

alignment.

2.2 Demagnetizing Field

According to Maxwell‘s equations for magnetism the magnetic field H and flux

density (|B|) must be continuous in both materials and free space. This is also known as

law of continuity of flux. When a field is applied on a piece of magnetic material it is

magnetised along a certain direction, north and south ‗poles‘ will form at each end. Thus

a magnetic field opposite to the applied field will arise. Due to its direction, this field is

known as the demagnetizing field HD. The magnetizing field depends both on the

magnetisation of the material and its shape. The demagnetising field can be calculated

from

(2.1)

Where M is the magnetisation of the sample and ND is demagnetising factor

which depends upon the geometry.

2.3 Magnetic Anisotropies

Due to the bonding of the ions in the crystals the spins on the atoms cannot take

up all orientations. Electron orbits in the solids are fixed by the crystal field and spins are

coupled to the orbits via L-S coupling. Hence to orient the spins requires that either the

L-S coupling is overcome or the hard movement. This means that the properties vary

along crystal axes due to varying atomic separation.

Chapter 2

12

2.3.1 Magnetocrystalline Anisotropy

Magnetocrystalline anisotropy is where a magnetic material is more easily

magnetised along a particular crystallographic direction. The origin of this effect is in the

spin-orbit coupling and the coupling of orbital magnetic moments to the crystal lattice.

The orbital magnetic moments are quenched so there is no orbital contribution to the

atomic magnetic moment. Consequently large magnetic fields will have no effect on the

orientation of the electron orbit. The orientation is therefore very strongly fixed to the

crystal lattice. When a field is applied, the electron spins will try to align. But as the spins

are coupled to the orbital angular momentum which is in turn fixed to the lattice, the

material will resist the reorientation of the spins. Therefore an anisotropic material will be

easier to magnetise along a certain crystallographic direction known as the easy axis, and

the axis where magnetisation is hardest defined as the hard axis.

2.3.2 Exchange Anisotropy

Exchange anisotropy was first discovered by Meiklejohn and Bean (1956).29

Ferromagnetic Co particles with diameter of ~20nm were oxidized forming a shell of

antiferromagnetic CoO around a ferromagnetic Co core. When the Co/CoO particles were

heated above the Néel temperature, TN, of the AFM CoO their magnetic properties were

the same as Co particles. However when the Co/CoO particles were cooled from 300K to

77K though the TN of the CoO, TN = 293K, with a saturating magnetic field applied the

hysteresis loop of the particles was shifted as shown in Fig 2.2.

Chapter 2

13

Fig 2.2: Hysteresis loop of CoO/Co particles after sample was cooled through TN

of CoO in a saturating magnetic field29

The shift in the hysteresis loop is called exchange bias. This type of anisotropy

exists due to the exchange coupling of ferromagnetic and antiferromagnetic materials

whether in the form of particles or thin films. The moment existing on the interface

opposes the reversal direction. Hence the coercivity is not the same in each direction and

the loop is no longer symmetrical. In extreme cases loop can shift entirely towards the

field cooling side.

2.3.3 Shape Anisotropy

In some materials, especially in polycrystalline thin films, lack of preferred grain

orientation gives rise to absence of crystal anisotropy. In such a case, shape of the

particles or grains gives rise to demagnetising field. A finite sample exhibiting poles at its

surfaces leads to a stray field outside the sample which results in a demagnetizing field

inside the sample. In the presence of its own stray field, energy of a sample is given be

the eq,26

Chapter 2

14

(2.2)

Where, is the demagnetizing field inside the sample. The solution of this eq is

much complicated for general shape. For symmetric shapes, it follows this way,

An ellipsoid contain demagnetizing field given in equation (2.1)

The stray field energy is thus given as

(2.3)

(2.4)

Where, is the volume of the sample.

Using these equations and taking the dimensions into account, the stray field energy

density for long cylinder can be calculated by using eq,

(2.5)

For infinitely extended thin sheet with a = b = , the stray field energy density amounts

to,

(2.6)

For thin magnetic films and layers, the above eq. can be rewritten as

(2.7)

With

At θ=90 , the stray field energy gets the minimum value of energy. This means that the

shape anisotropy dominates the magneto crystalline anisotropy in thin films which result

in, in-plane magnetization of the thin films.30

Chapter 2

15

2.3.4 Other Contributions to Anisotropy

Anisotropy can also be induced by stress in some specific direction, annealing in a

magnetic field or due to plastic deformation.

2.4 Magnetic Domain

Different regions of a macroscopic system break symmetry in different ways.

Weiss26

proposed that a ferromagnet contains a number of small regions called domains,

with each of which the local magnetization reaches the saturation value. Domains are

separated by domain walls. Domain structure is a natural consequence of various

contributions to the energy, exchange energy ( ), anisotropy energy ( ),

demagnetizing energy ( ) and Zeeman energy ( ). So the total energy without an

external field will be

(2.8)

If there is no external field the = 0 and = constant

(2.9)

If no dipole setup at the surface of a ferromagnetic material, there would be no

domains. Domain form solely to minimize the magnetostatic energy, that results when

m.n≠0 at an interface. Fig 2.3. shows how the domains are formed and how the energy of

the system minimizes.

Fig 2. 3: Domain formation and minimization of energy26

In Fig 2.3. a) Single domain is formed as a consequence of magnetic dipole

formed on the surface of the crystal. This configuration will have the energy as

a) b) c) d)

Chapter 2

16

(2.10)

i.e. if as is opposite to

So is positive

In case of Fig 2.3 b) the magnetic energy is reduced to ½ of a) and in case of c) if

there are N number of grain, the energy will reduce 1/N of a). In fig d) and e) the domain

arrangement is in such a way that the magnetic energy is zero. So domain always has its

origin in the possibility of lowering the energy of the system by going from a saturated

configuration with high energy to a domain configuration with lower energy.

Because of the domain walls formation, the demagnetizing energy decreases

but domain wall energy decrease which is equal to the sum of and .

The domain walls can be classified according to the angle between them. The

magnetization in the two domains lies as under. A domain 180 domain wall separates

domains of opposite magnetization. A 90 domain wall separates domain of

perpendicular magnetization. The most common of the domain walls is Bloch wall in

which the magnetization rotates in a plane parallel to the plane of the wall. Another

possible configuration is the Neel wall in which the magnetization rotates in a plane

perpendicular to the plane of the wall.

2.5 Reversal Mechanisms

The process of magnetisation reversal in ferromagnet generally follows one of

three possible mechanisms. The first of these is the mechanism of coherent rotation that

occurs in single domain particles either in powder form or in the form of a thin film

where the individual particles are not exchange coupled. Here the mechanism of

reversal is by coherent or incoherent rotation of the atomic moments over an energy

barrier determined by the anisotropy energy density and the grain volume. Generally

the energy barrier in zero field is of the form

Chapter 2

17

ΔE = KV (2.11)

However in small elements or at the edges of even large samples the energy barrier is

modified by shape demagnetising effects so that even in zero field

ΔE = KV′ (1-Hd /HK) 2

(2.12)

Where, V′ is now the element volume or the volume of an asperity at the sample

surface.

2.5.1 (Coherent reversal) Stoner Wohlfarth model

In 1948 Stoner and Wohlfarth predicted that reversal mechanism of magnetisation

in uniaxial single domain particles is by rotation. The model was based on the assumption

that the spins of the atoms in these single domain particles remain aligned during the

reversal process. Stoner-Wohlfarth model of magnetization reversal gives the idea of

coherent rotation during the reversal process in single domain particles.

Fig 2.4: Coherent rotation of magnetisation by rotation – the Stoner-Wohlfarth model.31

Chapter 2

18

For the case of an ellipsoid, with minor and major axes a and c respectively, an

applied field A will attempt to rotate the magnetisation vector Ms away from the

preferred easy axis direction (along c). This is resisted by some kinds of anisotropy of the

particle usually the shape, stress, or crystal anisotropy, or some combination of these. The

form of rotation of Ms is known as coherent reversal and is ultimately dependent upon

the precise alignment between the applied field and easy axis direction. In addition,

thermal activation will affect the nature of reversal for any system that is not at absolute

zero. The non-stationary nature of the grains at a finite temperature will entail that perfect

alignment is never achieved and so the grains in the system will possess a distribution of

energy barriers.

Stoner-Wohlfarth particles can provide insight into the reversal behavior of both

continuous and patterned perpendicular media in an applied field. This is because the

grains in such a medium are single-domain in nature and are exchange decoupled from

each other.

2.5.2 Incoherent Reversal

The Stoner-Wohlfarth model of coherent reversal is remarkable; however, it is not

successful in describing the reversal properties of some real materials, especially in the

case of multilayer thin films, where the domains have a very little or almost no

elongations. Fig 2.5 shows the coercivity data of system of non-interaction single domain

particles given by Stoner-Wohlfarth model.

Chapter 2

19

Fig 2.5: Variation of coercivity with axial ratio using the Stoner-Wohlfarth model of

reversal.31

Coercivity data due to shape anisotropy for aligned elongated particle for Stoner

Wohlfarth type of coherent reversal is given as

c = (Na-Nc)Ms (2.13)

Where Na and Nc are the demagnetising factors for short and long axes.

The calculated values of c for elongated iron particles by above equation is given by

B.D. Cullity26

in reference book reproduced in table 2.1

Table 2. 1: calculated values of c for elongated iron particles26

Shape anisotropy

c/a

Na-Nc

(SI)

Coercive Field

c (Oe)

1.0 9 0

1.1 0.075 810

1.5 0.301 3,240

5 0.833 8,950

10 0.939 10,100

20 0.980 10,500

1 10,800

Chapter 2

20

As can be seen in above in table 2.1, the observed value of c is 10,100Oe for c/a ratio of

10 and 10,800Oe for infinite elongation of iron particles. However, later on, Luborsky32

showed that c cannot exceed more than 1800Oe for c/a above 10, which is less than

20% than the theoretical value. This contradiction imposed a need to study other

incoherent rotation modes that would give a better match of experimental coercivities.

The reassessment of the coherent reversal by rotation process was initiated by

experimental coercivity data on systems of fine particles being considerably lower than

the predicted values33

. In an attempt to refine the description of magnetisation reversal in

a fine particle system, Jacobs and Bean proposed the 'chain of spheres' model. In this

model, the reversal occurs via a so-called incoherent mechanism known as fanning.

Fig 2.6: Fanning (a) and coherent (b) reversal modes for an N = 2 chain of spheres.33

Jacobs and Bean proposed a ―chain of spheres‖ model for ―peanut shaped‖ particles

which were seen on electron microscopy in their experiment on iron particles as shown in

the Fig 2.6.33

They modified expression for the intrinsic coercivity brings the predicted

values more in line with that observed in experiments.

In their model they suggested two possible mechanisms of reversal

Chapter 2

21

1. Symmetric fanning in which Ms vectors in the adjacent spheres have the

alternate directions. (Shown in Fig 2.6 a)

2. Coherent rotation in which Ms Vector lies parallel.

The coercivities calculated are compared to as predicted by Stoner-Wohlfarth model in

prolate spheroid. In this way they modified expression for the intrinsic coercivity brings

the predicted values more in line with that observed in experiments.

In 1957 Frei et al.,34

proposed another model of incoherent rotation based on the

micromagnetics calculation of shape anisotropy. The model is based on the assumption

that the particle is a single domain with all spins initially parallel to the +z direction in

zero fields. Crystal anisotropy and thermal effects are ignored in this type of reversal

mode. The main observation of this mode of incoherent reversal mechanism is that the

coercivity strongly depends upon the shape and size of the particle. They found that there

is a critical size of the particles below which, coherent rotation is favored. The particles

above that critical size will reverse by curling.

A number of models were put forward, such as the row-reversal model35

(Andrä

et al., 1984), various mean-field theories (Wuori and Judy, 198436

) and theories involving

domain wall motion (Wielinga et al., 198237

; Andrä and Danan, 198738

). However, many

of these interpretations failed to capture all of the experimentally observed phenomena,

be it incorrect values for the coercivity or the unlikely existence of a comparatively large

domain wall contained within very small grains.

2.5.3 Domain wall pinning mechanism or domain nucleation

The alternative mechanisms that initiate reversal are either a domain wall

pinning mechanism or domain nucleation. In the case of domain wall pinning a reverse

domain may already exist within the sample after it has been saturated and the field

reduced to zero. The reverse domain will have been injected due to the magnetostatic

energy of the sample or simply by the effects of thermal energy acting on the sample. A

Chapter 2

22

dispersion of easy axis orientations will aid this process as the moment returns to the

easy axis after having been saturated in the direction of the applied field. As larger

reverse fields are applied generally reversal proceeds by domain wall motion

throughout the sample but where strong pinning sites exist within the material, for

example at grain boundaries or at defects or inclusions, further reverse domain

nucleation may occur where it is energetically favorable. For an overview of reversal

processes see 26

.

Certain materials and particularly those having high anisotropy such as NdFeB

or SmCo5 exhibit a different form of hysteresis due to the very high anisotropy and the

consequent difficulty of nucleating the reverse domain when a strong easy axis

alignment is present. Under these circumstances a saturated material remains at

saturation even when the field is reversed to zero and a reverse domain cannot form

until a nucleation field Hn is reached. The value of Hn is then greater than any domain

wall pinning fields within the material and hence the reverse domain, once nucleated,

sweeps out across the sample resulting in an almost perfectly square hysteresis loop.

Such materials are known as nucleation controlled magnets 39

.

Nucleation controlled hysteresis is not confined to permanent magnet materials

or other hard materials. Similar effects are observed in very soft materials such as

Permalloy in thin film form. Here the fact that the magneto static energy in the plane of

the film is very low means that a significant reverse applied field is required to nucleate

a reverse domain. Due to the low anisotropy of the alloy and the strong exchange

coupling between the grains, the domain wall is again able to sweep through the sample

resulting in a square hysteresis loop 40

.

The energy barrier to domain wall nucleation depends upon a number of factors.

However in a granular material the nucleation process is similar to that involving a

reversal of a region of volume V over an energy barrier in a similar manner to that

which occurs in a single domain particle and to a first approximation can also be

Chapter 2

23

described by equations 1 or 2. Hence for materials where the reversal process is

dominated by nucleation or even in the case where ongoing reverse domain nucleation

occurs as domain walls sweep through the sample, the effects of local demagnetising

fields can be of critical importance.

Other than for perfect ellipsoids of revolution, the demagnetising field in a

sample is generally non uniform. This non uniformity can be manifest, for example at

the corners of square samples, or where any sample edge roughness occurs due to the

process by which the sample was fabricated. Such effects are well known when samples

of NdFeB are produced for measurement. Here dramatic reductions in the coercivity

can be observed unless samples are produced with very smooth surfaces to prevent

nucleation on surface asperities and other defects41

.

2.5.4 Reversal process in EB thin films

One of the most prevalent is so-called domain wall-assisted reversal. The reversal

process is as follows: an applied field induces the nucleation of a domain in the soft layer,

followed by its propagation towards the interface with the hard layer where it is pinned

and compressed. This compression continues until the field is sufficient to de-pin the

domain wall, allowing its propagation through the hard layer. For example, this type of

behavior has been modeled for a single FePt (hard)/Fe (soft) grain 42

. Of course, such a

mechanism is only possible if a domain wall can fit inside the layers. It has been

suggested that such structures can offer advances over conventional perpendicular media

in terms of data density 43

. Experimental measurements of the coercivity and remanent

coercivity have been undertaken to support this reversal hypothesis 44

.

Goodman et al.,45

proposed a comprehensive detail of reversal process in

exchange bias thin films (Fig 2.7)

At point 1 in the hysteresis loop the ferromagnet (FM) is fully saturated and

contains a single domain.

Chapter 2

24

At point 2 the reversal of the pinned layer is initiated. At this stage in the

hysteresis curve the nucleation of reverse domains in the ferromagnet is indicated

by degree of saturation at point 1.

At point 3 the domains nucleated at point 2 are growing. This growth is promoted

by applied field, the coupling field within layer and characteristic thermal

activation process. The coercivity at point3 exhibits normal magnetic viscosity.

At point 4 the ferromagnet is now saturated in negative bias and thermal

activation process in antiferromagnetic (AFM) are driven by exchange field from

FM layer. Goodman et al.45

proposed that at this point there is a growth of

domains within the AFM.

At point 5 it would be expected that FM layer would reverse. However, this is not

the case, to explain this, additional energy barriers have to be applied in the

system. Goodman et al.(2001)45

proposed that those energy barriers arose from

the pinning of domain walls in AFM. This results in the introduction of an

additional effective anisotropy in the FM via the exchange between the two

layers.

At point 6 the reversal of the FM layer starts. There is a difference between the

nature of the reversal at point 2 and 6. Goodman et al.45

proposed some initial

domain nucleation between 5 and 6. These domains could develop either via

domain wall movement over pinning sites or domain rotation. The viscosity at

point 4 suggests that pinning had to be significant.

At point 7 anomalous behavior of the second coercivity was observed. Goodman

et al.(2001)45

proposed that c 2 was intermediate and critically depends on the

amount of the time spent at.

Chapter 2

25

Fig 2.7: Magnetization Reversal in Exchange Bias Thin Films45

Craig et al.,46

studied the domain wall nucleation and its propagation with the help of

Lorentz microscopy (Fig 2.8)

Fig 2. 8: Magnetization Reversal in soft magnetic film46

Chapter 2

26

2.6 Exchange Bias

In 1956 Meiklejohn and Bean29

observed a new type of magnetic anisotropy in

compacts of oxidized Co particles. It was described by them as exchange anisotropy, or

exchange bias, as the effect arose due to the interaction of the ferromagnetic (FM) Co

particles with their anti-ferromagnetic (AFM) CoO shells when field cooled through their

Néel temperature ( ). In over 50 years since the discovery, numerous materials have

been made, and measured, that displayed this effect to various degrees. However there is

still lack of a comprehensive theory that can explain all the pragmatic affects such as the

hysteresis loop shift ( ) and the enhancement of coercivity ( c), which is described as

half the loop width.

2.6.1 Earlier Theories

A number of models attempted to explain the exchange bias phenomena. The first

attempt was made by Meiklejohn and Bean 29

. Their model was based on their studies of

single domain oxidized Co particles having uniaxial anisotropy and an easy axis in the

direction of the applied field. This model was based on the assumption that AFM spin

structure is perfectly uncompensated at the interface. Other assumption is that these

uncompensated spins remains aligned along the easy axis due to a large anisotropy in the

AFM. Though this model is successful in other exchange bias systems47-49

, however, it

predicts two orders of magnitude larger than observed values of in smaller grain size

polycrystalline films. The second model, chronologically, was that of Néel 50

in which he

proposed an uncompensated AFM spin structure at the interface. However he pointed out

that during the reversal of the FM layer the AFM spin structure at the interface would be

subject to irreversible changes, therefore the values of and c would change by the

changes in the AFM spin structure during reversal of the FM layer. This model too

predicts unreasonable values for . A more successful, model was that of Fulcomer and

Charap 51

in which it was predicted that changes in the AFM due to thermal activation

would occur due to the exchange field from the FM layer. This model predicts a

Chapter 2

27

distribution of particle sizes and shapes in AFM. The model gave good agreement of

temperature dependence of and c with experimental observations.52

Some other

granular models were also proposed on the basis of assumptions proposed by Fulcomer

and Charap.53-54

More recently Xi55

extended Fulcomer and Charap model to describe

thermal effects. However this model is restricted to single grain volume and do not fits

well for real systems.

However, models discussed above and many other attempted models failed to

exchange correct values of EB in real systems and to provide a road map develop new

materials for practical applications. Major shortcomings in all previous models are that

none of them could explain effect of film thickness and grain size, role interface and

grain size on EB.14

The reasons for failure of models are that none of them considered possible

thermal instability and initial degree of order in AFM layer. Study of wide variety of

systems is another reason for the failure of models.

In 1999 Berkowitz and Takano 56

gave a comprehensive review of the field. They

also proposed that a successful model of the exchange bias phenomena would have to

answer a number of problems. In answer to this O‘Grady et al. 14

proposed new model for

the exchange bias in polycrystalline thin films which can answer the majority of the

problems. As such it is this theoretical model that this study will utilize.

2.6.2 Technological Importance

Exchange bias played a vital role in the development of the spin valve 13

. Without

this the giant magneto resistant read head would not have been possible, and

consequently the storage densities of the modern hard disk drive (HDD) would not have

been technologically feasible. The development of magnetic random access memory

(MRAM) has also seen considerable focus. This technology would compete with current

Chapter 2

28

SRAM and DRAM technologies and the development of which would bring the GMR

sensor into another large market.

To understand the operation of the spin-valve a basic understanding of GMR is

required. In FMs spin up and spin down electrons experience different probabilities of

scattering. This is due to a difference in the number of each electron in the d band, known

as a spin-split structure. As such it can be considered that the current of spin up electrons

is separate to that of the spin down electrons, as one current is favored and the other is

not they are called the majority and minority electrons. In a simple FM bilayer two spin

channels can be considered. In the first case, when the FM‘s magnetizations are parallel,

one spin channel carries only the majority electrons whilst the other carries only minority

electrons. In such a case the majority electron spin channel is greatly favored for carrying

charge and as such the overall resistance is low. In the second case, when the FM‘s

magnetizations are anti-parallel, both spin channels must carry both majority and

minority electrons. As such neither channel is favored, as they have equal resistance, and

so the overall resistance is higher. Thompson 13

has recently written an in depth review of

this topic.

For a spin valve device to be possible a pinned and free FM layer is required. Due

to exchange bias this is possible. AFM is used to pin one of the FM layers. The

exchanged biased bi-layer is then separated with a spacer that destroys any exchange bias

that could arise between the two FM films. This allows for one of the FM layers to

maintain a constant magnetization, whilst the other is free to rotate within an applied

field.

2.6.3 Recent development in the field of EB

The current state of exchange bias was reviewed and challenged by O‘Grady et

al.14

with the proposal of new definitions and explanations for a number of phenomena

Chapter 2

29

associated with the magnetic measurement and characterization of sputtered

polycrystalline thin films.

The largest of contributions by O'Grady et al 14

was that of the York Protocol, a

series of steps in which the magnetic history of samples could be controlled and therefore

reproducible measurements made. This allowed for comparable measurements of effects

that could not previously be compared. This also gave rise to a new definition of the

blocking temperature ( B) as well as a new explanation for the main contributing factors

to the value of , both of which will be elaborated on in later sections. This knowledge

has allowed the design of AFM/FM materials for specific applications and setting

conditions.

2.7. York Protocol

2.7.1. Theory

2.7.1.1. Grain Size Distribution

There are a number of probability densities used to describe experimental grain

growth and grain size distributions in thin films. Of these it has been show that the grain

size distribution in polycrystalline thin films follows a log-normal distribution 57

.

Furthermore, the log-normal distribution only returns positive values, which is necessary

for the description of grain sizes. Another advantage is that if a variable D follows the

log-normal distribution then so must both 2 and

3 which is especially significant for

magnetic materials as their properties are volume dependant. For a linear interval, dD, the

log-normal distribution is written as:

(2.14)

Where D is the grain diameter, σ is the standard deviation of ln and μ the mean of ln .

The standard deviation of the log-normal distribution is, however, given by:

(2.15)

Chapter 2

30

While the mean grain diameter is:

(2.16)

Both of which will be required to give a full physical description for comparisons.

In order to make a quantitative analysis of the experimentally found grain sizes,

the experimental histograms must be fitted using the log-normal distribution function.

The most direct method of achieving this would be to use standard data analysis software,

calculate σ and μ then recompose equation (2.14) and directly fit the experimental data.

However this requires a large number of grains, ≥1000 grains, to make a reproducible

distribution 57

. Vopsaroiu et al. 57

proposed the usage of the cumulative percentage

method (CPM). The advantage to this is that a much smaller number of grains are

required to form a reproducible distribution. Through this method σ and μ can be

calculated and then used to recompose equation (2.14).

The cumulative percentage data for a specific grain size is calculated using the

percentage undersize sum , where is the percentage of the total number of

particles with a given grain size diameter . This is then plotted against ln . As ln is

normally distributed it is possible to graphically calculate μ as it corresponds to 50% of

the cumulative percentage. It is similarly simple to calculate σ using:

(2.17)

Where, 84 and 16 correspond to the 84% and 16% points on the cumulative

percentage curve. Armed with both σ and μ the log-normal function can be generated and

fitted to the experimental data. 57

Chapter 2

31

2.7.1.2 Blocking Temperature

Conventionally the blocking temperature con is described as the temperature at

which the exchange bias reduces to zero. To determine con the procedure was based

on the idea of taking hysteresis loops whilst increasing the temperature until such a point

that loop shift became zero. At temperatures above con the exchange bias remains

equal to zero. Fulcomer and Charap 51

stated that the value of con would correspond to

the individual blocking temperature of the AFM grain with the largest anisotropy energy.

Following from this it can be said that each AFM grain in a polycrystalline system has its

own blocking temperature and so the bulk AFM is characterized by a distribution of

blocking temperatures. While using the conventional method of measurement, the AFM

would be prone to thermal activation during measurement at a logarithmic rate 58

. This

lead to a lack of reproducibility due to changes in the state of order of the AFM from

measurement to measurement.

O‘Grady et al.,14

demonstrated that due to the thermal activation of the AFM layer

it was possible to shift the hysteresis loop of an exchange biased system to the opposite of

that from when it was set. They further showed that using a specific set of procedures,

known as the York Protocol, this could be specifically controlled to measure the mean

blocking temperature . Thus by heating the FM layer whilst reversed the AFM

would experience changes in order from the original state to the opposite orientation, as

shown in Fig 2.9. As such they concluded that the amount of the AFM that undergoes

reversal will rely solely on the temperature and the exchange field from the FM layer.

Chapter 2

32

Fig 2.9: Schematic of the energy barrier reversal, showing the proportion of AFM grains

set parallel or anti-parallel to the original set direction.14

The York Protocol thus gives a different definition of < > and O‘Grady et al.

define it as ―The point where equal fractions of the volume of the AFM grains are

orientated in opposite senses‖. As such the value of can be described as proportional

to the difference between the fractions of AFM grains that are orientated in opposite

directions:

(2.18)

Fig 2.10: Comparison of Blocking Temperature (TB) measured from a) Conventional

method b) York Protocol14

Chapter 2

33

2.7.1.3 Measurement of EB from York Protocol

O‘Grady et al.,14

demonstrated that in an exchanged biased system the value of

is dependent on the amount of AFM grains that are thermally set. However as for

IrMn, the AFM used in most technological applications, is far higher than room

temperature it is not possible to fully set the AFM without damaging the structure of the

sample. As such the sample has to be set below , in which the setting process is via

thermal activation. Fig 2.11 demonstrates a situation where a sample has been set at a

temperature set for a time in which set was not sufficient to set the entire AFM

distribution. As such a fraction of the AFM grains with V>Vset are not aligned with the

FM layer, whilst a fraction of AFM grains with V<V were thermally unstable at the

temperature of measurement Tms. Consequently only the grains of volumes between VC

and Vset will contribute to . O‘Grady et al. assumed a constant value of KAF, the AFM

anisotropy constant, and wrote the proportion of:

Fig 2.11: Schematic of the grain size distribution after the setting of the AFM and cooling

to a temperature at which the AFM is thermally unstable.14

(2.19)

Chapter 2

34

It has been shown in practice that the application of a higher field during the

setting process increases the value of by orders of up to 20% and more. This implies

that the applied field somehow affects the interfaces while not affecting the bulk material.

O‘Grady et al. 14

showed there to be a correlation between the texture of the AFM and the

composition of the FM layers at the interface. Although they had insufficient data to

define the origin of this effect they were able to write an equation linking with the

degree of order and stability of the AFM and the behavior of interfacial coupling:

(2.20)

2.8 Training Effect in Exchange Bias

Training effect is one of the direct consequences of the exchange bias. It is found

that that with repeated cycling of the applied field there is a reduction in the exchange

bias in FM and AFM coupled system. The decrease of the shift of hysteresis loop with

consecutive cycles of the applied field is called training effect. Training effect may also

come from the change in the shape of the loop and change in the coercive field.

The training effect was discovered by Paccard et al59

in 1966 in the consecutive

loop of Co-CoO system. He found that the shift in the first and second loop is more

significant than the others.

Fig 2. 12: Consecutive hysteresis loops of a Co − CoO system measured with torque

balance. The observed overshoot is an instrumental effect.59

Chapter 2

35

He also found change in the shape of the loop and decrease in coercivity as shown

in the Fig 2.12. Thus training effect can be categorized in two classes. One is due to the

shift of 1st two measurements, which is more pronounced. The reason is the non-

equilibrium state of AFM caused during the cooling process. Second type of the training

is due to the further shift of the loop other than first two loops, which decreases with

increasing number of cycles. Paccard et al 59

shown that 2nd

type of training obeys inverse

power law. Fluctuations of the FM-AFM coupling, due to spins reconfiguration or the

domain state of the AFM, during consecutive cycles, is the cause for the 2nd

type of

training.

This occurs in both single crystal and polycrystalline thin film samples although

the effect is more important in polycrystalline systems60

. Training has been observed in

polycrystalline IrMn thin films such as those studied here 61.Training effect is important

due to the fact that it is the measure of the stability of exchange bias bilayer which is

inside many devices.

Hoffman (2004)62

, for the first time, put forward a theoretical model describing

the training effect successfully. He showed that training on the first loop may be a result

of multiple easy axes for the AFM layer62

. This model described that for AFM with

biaxial anisotropy there will be training effects between the first and second measured

hysteresis loops, however he ruled out any possibility of training effect in AFM

anisotropy that has uniaxial symmetry. This is in accordance with experimental data

where systems with an AFM having multiple easy axes (eg. cubic anisotropy), such as

IrMn, training is observed. However no training observed in FeF2 having AFs with

uniaxial anisotropy 63 .

Chapter 2

36

Fig 2. 13: Schematic showing the FM and AFM sublattice magnetizations in an exchange

bias system where the AFM anisotropy has biaxial symmetry during the 1st and 2nd

hysteresis loop measurements. 62

In Hoffman‘s model the training effect is caused by the AFM spins which are

non-collinear initially, relaxes into a collinear arrangement after the reversal of the field.

This is shown schematically in Fig 2.13

Another study of training in FeMn/CoFe systems by Fernandez-Outon et al64

showed that there were two contributions to the training effect. By measuring the training

at a temperature where the AFM grains that make up the AFM layer were thermally

stable, TNA, an athermal training effect where only the first branch of the hysteresis loop

shifts between the first and second hysteresis loops is observed. When the sample was

measured above TNA, both branches of the hysteresis loops shifted due to thermal

activation effect.

2.9 Diluted Magnetic Semiconductors

Diluted magnetic semiconductors are the class of materials which have the spin

polarized electrons retaining the semiconducting properties, along with the magnetism.

However it is essential to enhance Curie temperature well above the room temperature to

make them useful for their application in spintronics devices. Also the origin of

magnetism in such materials is still controversial.

Chapter 2

37

The DMS got much attention after the earliest observation of RT ferromagnetism

in the Co doped TiO2 system by Matsumoto et al.65

. Matsumoto and his coworkers

synthesized Ti1-xCoxO2 anatase films (0 ≤ x ≤ 0.08) on LaAlO3 (0001) and SrTiO3 (001)

substrates by laser molecular beam epitaxy. The same research group reported RT

ferromagnetism in rutile phase Ti1-xCoxO2 (0 ≤ x ≤ 0.05) thin films by same experimental

technique onto α-Al2O3 substrates. Since then, several physical and chemical deposition

techniques were employed to fabricate Co and many other transition metal doped TiO2

systems. Many techniques such as pulsed laser deposition (PLD)19-22, 66

, plasma-assisted

molecular beam epitaxy67-69

reactive co-sputtering70

, metal-organic chemical vapor

deposition (MOCVD)71

, molecular beam epitaxy (MBE)72-74

, and solgel method75

have

been used.

Below are the few theoretical models which are currently available to describe

this phenomenon.

2.9.1 Types of interactions in DMS

2.9.1.1 sp-d exchange interactions

The model of band structure of DMS can be used to start to interpret the source of

their properties. In this band structure two electronic subsystems are compared: one

consisting of the magnetic impurity electrons with magnetic moments localized in the

ionic open 3d (or 4f) shell, and the other containing delocalized, band electrons built

primarily of outer s and p orbital of constituting atoms. The magnetic properties produce

in the DMS system due to the strong spin dependent sp-d(f) exchange interactions and the

localized magnetic moments (d-d) interaction between these two systems.

The localized impurity magnetic moments in DMS and the spin dependent

interactions between band carriers are due to the two autonomous mechanisms, called the

hybridization mediated kinetic exchange and the direct Coulomb exchange 76

.

Chapter 2

38

2.9.1.2 d-d interactions between magnetic ions

DMS with magnetic properties are due to the interactions that couple the spins of

magnetic ions. There are a number of microscopic mechanisms that bring about the spin-

spin (d-d) interactions between two magnetic ions. The interaction can be considered as a

virtual transition between the ions and neighboring anions, in two main mechanisms,

namely the double exchange and superexchange. The spins of two ions are interrelated

due to the spin-dependent kinetic exchange interaction between each of the two ions and

the s, p band in the super-exchange mechanism. The magnetic behavior comes from the

superexchange is due to the dominant spin-spin interaction for group II-VI DMS and

have shown by Larson et al.77

. When the magnetic ions in DMS have the same chemical

but different charge state then the double exchange interaction occurs78

. The virtual

hopping of an ‗additional‘ electron from one ion to the other through interactions with the

p-orbitals shows the coupling of magnetic ions in different charge state related by double

exchange. Zener proposed this mechanism 79

in the 1950s and was applied to understand

ferromagnetism caused by the coexistence of the Mn2+

and Mn3+

in ZnO80

The

suprexchange interactions for many magnetic dopant should actually be

antiferromagnetic, and thus no contribution to the ferromagnetism.

2.9.1.3 RKKY interaction

The ion-ion interaction in DMS only when a high concentration of free carriers is

present could be described by using RKKY interaction. It can be understood as the

interaction between the surrounding electron gas and a local magnetic impurity. The

polarizable medium is the carriers that transmit spin polarization from one atomic site to

another. We can prove the different spin electrons scattered differently, as spin-up and

spin-down electrons feel a different potential in the neighborhood of a spin-polarized

impurity ion. The oscillation of the spin-up electron density is shifted comparative to the

oscillation of the spin down density due to the different phase shift. The superposition of

these two charge densities yields an oscillatory magnetization which decays according to

Chapter 2

39

the dimensionality of impurity considered. Atoms at a given distance from the impurity

experience either a negative or a positive polarization and consequently their orientation

describes their magnetic moment. The carrier induced RKKY interaction, adequately

long range to account for the magnetic interaction in dilute systems, and has been

propose to explain the carrier (hole) induced ferromagnetism observed in IV-VI thin

films and Mn-based III-V81-83

.

2.9.1.4 Zener model

Dietl et al 84-85

shown that , the Zener model has to be invoked to explain the

observed properties of Mn-based II–VI and III–V thin films and heterostructures when

the mean ion-ion distance is small with respect to 1/kF. This is due to the significance of

the kp, carrier-carrier interactions and spin-orbit coupling which are hard to take into

account within the RKKY model. The mean field values of the ordering temperature

deduced from the Zener equals to the RKKY model, when these interactions are

neglected. In Zener model79

, the spin polarization of the localized spins results in a spin

splitting of the bands and in this situation the exchange coupling between the carriers and

the localized spins leads to ferromagnetism. The redeployment of the carriers between the

spin sub-bands lowers the energy of the holes carriers, which at suitably low temperatures

overcompensates an increase of the free energy related with a decrease in Mn entropy.

Dietl et a84

suggested that the holes in the extended or weakly localized states mediate

the long range interactions between the localized spins on both sides of the Anderson-

Mott metal-insulator transition (MIT) in the Mn doped II-VI and III-V DMS. They also

showed that the holes transmit magnetic information efficiently between the Mn spins

due to the large density of states in the valence band and strong spin-dependent p-d

hybridization.

The p-d Zener model has been successful in explaining a number of properties

observed in ferromagnetic DMS, particular (Ga,Mn)As and (In, Mn)As, including the

magnetocrystalline anisotropy 86

, ferromagnetic transition temperature87

etc.

Chapter 2

40

2.9.1.5 Polaron Percolation model

The polaron percolation theory has been developed to understand the

ferromagnetic ordering of the DMS with strongly localized carriers. The formation of

bound magnetic polaron (BMP) results from the exchange interaction of those strongly

localized carriers with magnetic impurities. Since the carrier concentration is much less

than the magnetic impurities density, a localized hole is surrounded by the impurity spins

in a BMP as shown in Fig.2.14 88

. Even though direct exchange interaction of the

localized carriers may be antiferromagnetic, the interaction between bound magnetic

polaron can be ferromagnetic89

if the concentration of the magnetic impurities is large

enough. The localized holes (large arrows) produce an effective field for the impurity

spins (small arrows). Shaded area shows overlap effect of two BMPs on impurity spins.

The maximum of this effective magnetic field is achieved when the spins of the localized

holes are parallel. When the direction of impurity spins is parallel to the effective field,

minimum energy and maximum field are also achieved. Therefore at low temperatures

the system should eventually reach the state where the spins of all holes point in the same

direction, and all impurity spins point in the same or in the opposite direction, depending

on the sign of the impurity-hole exchange interaction.

Fig 2. 14: Interaction of two bound magnetic polarons. The polarons are shown with gray

circles. Small and large arrows show impurity and hole spins, respectively.88

Shaded

region shows the effect of two BPMs on impurity spins.

Chapter 2

41

Taking into account the high defect concentration in a typical magnetic

semiconductor material, the localized charge carrier density in the systems is highly

inhomogeneous too. Since the exchange interaction between magnetic impurities is

transmitted through the charge carriers, this interaction must also be highly

inhomogeneous 90

. When the temperature is lowered, in the regions with higher charge-

carrier density, the ferromagnetic transition will first occur locally (align a spin in parallel

or antiparallel with all the impurity spins in the vicinity), leading to the formation of a

BMP. As temperature falls further, the polaron grows in size until its radius overlaps that

of neighboring polaron, enabling long-range interactions between TM ions and

ferromagnetic ordering in low carrier density systems. BMP begins to form at a certain

temperature and their diameter will increase with decreasing temperature and eventually

spreads over the whole system at the Curie temperature to produce ferromagnetism.

Chapter 3

42

Chapter 3

Experimental

3.1. Synthesis Technologies

3.1.1. High Target Utilization Sputtering (HITUS)

HITUS was used for study of exchange bias effect in multilayer thin film at

Department of Physics, University of York, UK. Details of its working and conditions for

current studies are discussed below:

Exchange bias effects are seen in multiple types of samples each with distinct

morphologies controlled by their formation processes. The most common sample types

are: nanoparticles, which have non-flat interfaces51

, epitaxial thin films, which have near-

flat interfaces91

and sputtered polycrystalline thin films, which have significantly rough

interfaces.92

It is, interestingly, the later that produces the greatest exchange bias at room

temperature.14

Fig 3. 1: Schematic representation of HiTUS sputtering technology.57

Chapter 3

43

Vopsaroiu et al,57, 93

describes a novel sputtering technology that allows direct control

over the grain size of the sputtered sample through close and accurate control of the

growth rate. Magnetron sputtering, an older sputtering method, does not allow sputtering

from thick ferromagnetic targets and has non-uniform wear of the target making it

inefficient for magnetic thin films. Whilst triode and diode sputtering systems may have

the advantage of uniform target wear, they are not suitable for reactive sputtering and

have low deposition rates. As such the sputtering technology as described by Vopsaroiu

et al.57

is given the acronym HiTUS (high target utilization sputtering >90% of the

target). A further benefit to the technology is its ability to both sputter from large

ferromagnetic targets as well as carry out reactive sputtering.57, 93

Samples were prepared using the sputtering system introduced above and shown

schematically in Fig 3.1. The system is based on high intensity plasma which is produced

in a separate arm attached to the main deposition chamber. The plasma is created using a

radio frequency (RF) electric field (max. 2.5kW) and is launched into the main chamber

through the interaction between the RF field and the launch electromagnet. A second

electromagnet is then used to steer the plasma onto the target. Sputtering cannot occur,

however, without applying a negative dc bias (-1V to -1000V) to the target. At voltages

over -100V the target current becomes saturated and independent of the voltage. This

allows for control over the energy of the Ar ions incident upon the target. As the number

of Ar ions is controlled by the energy of the RF field the sputter rate can be controlled by

keeping the RF field constant and varying the bias voltage. This in turn controls the grain

size.57

All samples were sputtered under the same conditions after pumping down to a

base pressure of 2.75x10-3

mbar. The average process pressure was 2.75x10-3

mbar and the

RF power was held at 1.5kW. A rotating substrate table allowed up to six samples to be

grown separately without breaking the vacuum. The design of the substrate table meant a

Si substrate and Transmission Electron Microscope (TEM) grid could be sputtered on

simultaneously. The distance between substrate and target was ~25cm which is quite

Chapter 3

44

large than normal, thus minimizing interaction between them and the film on both

materials was identical.93

Metallic targets (Cu, NiCr, IrMn, CoFe) used for this study

were pure up to 99.9999%. Prior to the sputtering of each layer a 60sec substrate/target

plasma cleaning process was employed to eliminate contaminants and any oxide layers.

Substrate heating was not employed and a temperature below 100°C was maintained.

Additionally, a permanent magnet was placed above the substrate table to attempt to

induce order in the AFM layer during growth.

Sample growth was predominately automated; however substrate exposure had to

be controlled manually. Details of growth conditions were recorded during growth for the

sputtering of each layer. In each series five samples were grown with variable bias

voltage (200,400,600,800 and 1000V) whilst keeping the composition constant.

3.1.2. Aerosol Assisted Chemical Vapor Deposition (AACVD)

Several recent studies suggest that any magnetic ordering in diluted magnetic

semiconductors depends on the synthesis routes and is sensitive to chemical ordering of

the TM ions and defects which may be vacancies or interstitials 94-96

.Many reported

properties of DMS have shown lot of variation in the experimental data depending

strongly on preparation technique and growth conditions. This has created a lot of

complications in interpretation of experimental data. Thus on one hand more studies are

required for the optimized results, on the other hand new and more sophisticated

synthesis techniques are required which are economical.

AACVD is a novel technique in which precursor solution is atomized into fine,

submicron size aerosol droplets that are transferred to heating zone through evaporation.

In heating zone it go through decomposition and homogeneous and/or heterogeneous

chemical reactions to form the desired product. AACVD has an advantage over the

conventional CVD for it overcomes the availability and delivery problems of the

precursors found in the conventional CVD.

Chapter 3

45

Samples were prepared on Si (100) and commercially available soda glass

substrates using the house-build AACVD assembly introduced above and shown

schematically in Fig 3.2. The system is based on a PIFCO ultrasonic air humidifier with a

piezoelectric modulator. Other parts of the assembly are the cylinder of carrier gas which

may be air or any other gas according to the requirements of the experiments. A micro

controller is used to control the gas flow and hence the deposition rate which alternatively

controls the films morphology, porosity and thickness over a certain period of time. The

aerosols are created using PIFCO ultrasonic air humidifier and launched into the heating

zone through with the help of carrier gas where precursor decomposes to form thin films

of the required metal oxide.

3.1.2.1. Advantages of AACVD over conventional CVD

AACVD has following distinctive advantages over the conventional CVD process15-17

i. AACVD enables a fast evaporation of the precursor. This give an advantage

relatively short delivery time to the heating zone and high deposition rate.

ii. Fabrication of multi component films with highly precise stoichiometry

iii. A number of different precursors may be used for AACVD which need not to be

volatile, but just to be merely soluble. Thus eliminating the need of volatile

precursor, where essential for conventional CVD.

iv. Condition of reaction environment during AACVD is flexible. It can be

performed both at low pressure and in open atmosphere.

v. AACVD is a low cost process, it do not need complicated and costly

instrumentation.

vi. Variety of products may be achieved through AACVD such as thin films, nano-

powder and nanotubes etc.

However, AACVD also have certain limitations in comparison with other techniques

e.g.

i. AACVD is not suitable technique to grow thin films/coatings on thermally

unstable substrates 97

Chapter 3

46

ii. Gas phase reaction produces defects like pin holes in the films.

iii. Adhesion of the film may become weak with time.

iv. Difficult to control film thickness.

Fig 3. 2: Schematic representation of AACVD

3.2. Characterization Techniques

3.2.1. X-Ray Diffractometer

The diffractometer determines the identity of crystalline solid based on the atomic

structure of the material. The XRD pattern gives the direct information of two things

i. Relative positions of the XRD peaks give size and shape of the unite cell

ii. Relative intensities of the peaks determine the atomic positions in the unit cell.

In diffractometer when a sample is exposed to the monochromatic x-rays

diffraction occurs when atoms in a periodic array scatters radiation coherently, producing

intensive constructive interference at specific angles. The electrons in an atom interact

with the oscillating electric field of the light wave and produce coherent scattering. In this

way Atoms in a crystal form a periodic array of coherent scatterers. Diffraction from

different planes of atoms produces a diffraction pattern, which contains information about

the atomic arrangement within the crystal. The simplistic model to understand the

conditions required for the interference is Bragg‘s law98

Flow Controller

Car

rier

Gas

Piezoelectric Modulator

Precursor

Solution

Aerosol

s

Glass Jar

Substrate

Hot

Plate

Chapter 3

47

(3.1)

The space between diffracting planes of atoms determines peak positions. The

peak intensity is determined by what atoms are in the diffracting plane.

The XRD used in this study was the PAN analytical, X‘pert PRO operated on

40kV and 22.5mA. The high intensity monochromatic Cu-K radiations

(λ=0.154184nm) source and PSD detector are fitted. The majority of the control of the

system was through an attached computer where scanning parameters could be controlled

using the proprietary software. To increase efficiency and accuracy of the measurements

the intervals between recordings were carefully chosen. A very slow scanning rate of 1

step/sec was applied where each step was 0.02 .

3.2.2 Field Emission Electron Microscope (FESEM)

The SEM takes advantage of the concepts put forward by Ernest Ruska (1906-

1987) and applies them to the potential applications of electrons in microscopy. This

allows high resolution images to be taken at high magnifications with relative ease.

FESEM is a high resolution SEM and lithography system. A cold cathode field

emitter or simply field emission source at the top of the apparatus accelerates the

electrons through a potential difference 20kV. The tube is kept at a constant high pressure

<1x10-7

to decrease the discharge through electron-particle interactions. Upon reaching

the condenser the electrons are condensed into the beam, which is the focused on the

sample by using the objective aperture and the magnetic lenses. This is an important step

as the magnification of the sample is proportional to the distances between the sample

and the plane of the lenses. As the beam is incident upon the sample the electrons are

diffracted by the atoms in the sample. These diverged beams then pass through the next

set of lenses and are focused on the phosphorus screen or charge-coupled device (CCD)

to produce an image.

Chapter 3

48

The FESEM used in this study was the FEI Sirion S-FEG FESEM. Control of the

system was done mostly through the integrated electronics and computers where the

electron gun voltage, position of the image and capturing process could be controlled

using the in-built software and electronics. To help improve the image quality an

additional aperture was added to reduce interference on the image. In addition to this

spectrometry data could be collected from emitted x-rays so as to confirm the

composition of the samples.

The Sirion S-FEG allows a very short working distance of 5mm to help to

increase the image resolution. Higher image resolution can be obtained at a very high

voltage of 20kV using conventional SEI detector or through a combination of the lenses

system. It has a facility to work on low voltage (1-5kV) images for bulk and non-

conducting samples. However, the standard working conditions are 5kV. Energy

Dispersive X-ray Spectroscopy (EDX) is fitted with the FESEM for chemical analysis of

the samples having Z > 4. EDX is operated using Oxford INCA analysis system which is

further upgraded by applying 30mm 2 light element capable ATW detector

3.2.3. Zeiss Particle Size Analyzer

The Zeiss Particle Size Analyzer works by using an equivalent circle method. It

has 48 units of measurements, or bins, which have been calibrated so that each bin

corresponds to a specific diameter. For a measurement to be made a TEM image of the

grains is placed on a light box in which an iris focuses a spot of light. It is then possible

to match the spot of light with an equivalently sized grain. Registration is then made in a

bin for this grain size using an especially written lab view program. The program then

keeps track of the grain sizes within a specified range and creates a frequency distribution

accordingly.

Chapter 3

49

3.2.4. Transmission electron microscopy (TEM)

The TEM takes advantage of the concepts put forward by M. Planck and L. De

Broglie and applies them to the classical light microscope by replacing photons with

electrons. This allows high resolution images to be taken at high magnifications with

relative ease.

An electron gun at the top of the apparatus accelerates the electrons through a

potential difference, 200kV in this instance, down a thin tube in the centre of the device.

This tube is kept at a constant high vacuum so as to decrease the chance of an electron-

particle interaction. Upon reaching the condenser aperture the electrons are condensed

into a beam, which is then focused onto the sample by the objective aperture and lens.

This is an important step as the magnification of the sample is proportional to the

ratio of distances between the sample and plane of the lens. As the beam is incident upon

the sample the electrons are diffracted by the atoms in the sample. These diverged beams

then pass through the next set of lenses and are focused on the phosphorus screen99

or

charge-coupled device (CCD) to produce an image. This is demonstrated schematically in

Fig 3.3.

Chapter 3

50

Fig 3. 3: Schematic Diagram of a Transmission Electron Microscope100

The TEM used in this study was the JEOL JEM-2100 TEM. Control of the system was

done mostly through the integrated electronics and computers where the electron gun

voltage, position of the image and capturing process could be controlled using the in-built

software and electronics. To help improve the image quality an additional aperture was

added to reduce interference on the image. In addition to this spectrometry data could be

collected from emitted x-rays so as to confirm the composition of the samples.

To reduce error ten images at 80,000x multiplication were taken per sample. The

criteria for an image were for there to be 50 or more grains clearly visible. Equivalent

circle method was then implied to study grain size. Two images would be taken per area

on the film so as to obtain an even distribution of grain sizes.

Chapter 3

51

3.2.5 Magnetometers

There are multiple methods for making measurements of the magnetic properties

of materials, each with its own advantages and disadvantages. The first method is the

Susceptometer Magnetometer which is based on Faraday‘s law of induction. By moving

a magnetised sample between two ‗sensing‘ coils a signal is produced which is

proportional to the magnetic moment of the sample. A major benefit to this is that the

susceptometer measures in absolute units of both magnetic susceptibility and moment

and therefore no calibration is required101

. The second method is the Superconducting

Quantum Interference Device (SQUID) Magnetometer. The SQUID magnetometer

operates by measuring the flux change, due to the magnetised sample, on a

superconducting detection coil attached to a SQUID device. The signal that results is

proportional to the magnetic moment of the sample. This method results in exceedingly

high sensitivity; however the system must be operated at superconducting temperatures

101. The third method is that of the Vibrating Sample Magnetometer (VSM). This method

is the most commonly used and versatile method of magnetic characterization and is the

method used solely in this study and so will be covered in more detail in the following

sections.

3.2.5.1. Vibrating Sample Measurement (VSM)

The basic principle of the VSM is if any sample is placed inside a uniform

magnetic field, created between the poles of two electromagnets, a magnetic dipole will

be induced. If the sample is then vibrated with sinusoidal motion then a sinusoidal

electric signal will be induced in the pickup coils. This induced signal will then have the

same frequency of vibration; however the amplitude will be proportional to the magnetic

moment, the amplitude of vibration and the relative position of the sample to the coils 101

.

The setup of this system is shown schematically in Fig 3.4.

Chapter 3

52

Fig 3. 4: Schematic of a standard VSM.

The VSM used in this study was the ADE Vector Model 10 VSM. The majority of the

control of the system was through an attached computer where temperature, field angle

and applied field could be controlled using the proprietary software. To increase

efficiency and accuracy of the measurements the intervals between recordings were

carefully chosen. During magnetic saturation of the FM large step sizes were used, 200

Oe, while during the loop shift small step sizes, 25 Oe, were used. This allowed for an

increase in resolution around the points of interest.

Due to the automation of the system the main sources of error were in how the

sample was placed in the machine. As the sample would be vibrating it was important to

ensure that it was rigidly attached to the sample holder to reduce excess modes of

vibration. This was achieved using both vacuum grease and a non-magnetic tape.

Secondly the ‗saddle point‘ of the field had to be found. Using knobs that varied the

Pickup

Coil

Vibration

direction

Probe body

Hall probe

Sample

Electromagne

t

Electromagne

t

Chapter 3

53

position of the sample it was possible to find the point at which a minimum signal was

detected in both the x and y planes92

.

3.2.5.2 Alternating Field Gradient Magnetometer (AGFM)

The Alternating Gradient Field Magnetometer (AGFM) operates in almost an

opposite fashion to that of the VSM. The sample is attached to a sample holder which is

placed in a magnetic field gradient and, due to Faraday‘s law of induction, a force felt. If

the field gradient is then varied with a frequency equal to that of the natural frequency of

the sample then, the sample will vibrate sinosodially with the amplitude at its maximum.

This vibration will then be detected by the piezoelectric bimorph which is attached, via

quartz legs, to the sample holder. The sinusoidal motion of the sample then induces a

proportional current in the bimorph. Like the VSM the frequency of this signal will be

equal to that of the alternating field; however the amplitude will be proportional to the

magnetic moment, the amplitude of the alternating field and the mass of the sample101

.

The setup of the system is shown schematically in Fig 3.5.

Fig 3. 5: Schematic of a standard AGFM.

Gradient

Coil

Piezo-electric

biomorph

Probe body

Hall probe

Sample

Quartz legs

Electromagn

et

Chapter 3

54

The AGFM used in this study was the MicroMag 2900 AGFM. The majority of

the control of the system was through an attached computer where the frequency and

magnitude of the magnetic field and sensitivity could be controlled using the proprietary

software. The natural frequency of the sample was found when the difference between

two auto-tunes was in the order of 0.25Hz in difference. Due to the speed of

measurement in the AGFM step size was not a concern and so step sizes of 30 Oe were

used.

Due to the fragility of the probe and operation of the system the main sources of

error were dependant on the mass of the sample, the condition of the probe and the

position of the sample between the magnets. The probe had to be handled with extreme

care so as not to damage or break the legs as this would drastically affect the signal.

However the mass of the sample was not an issue due to its small size whereas the

position of the sample was vital. This was controlled in identical fashion to that of the

VSM.

3.2.5.3. Superconducting Quantum Interference Device (SQUID)

Superconducting Quantum Interference Device (SQUID) magnetometer is an

extremely sensitive magnetometer (Typically 10-7

emu) based on the Josephson Effect. It

can give measurements on a range of temperature ranging from 2K (liquid He

temperature) to 350K. The Helium-cooled magnetometer could operate in the range of -

70 to +70 kOe.

RF SQUID design is very much similar to that Josephson junction shown in the

schematic diagram. Josephson junction is based on a superconducting coil sandwiching

an insulator. There is phase coherence in the electron pairs which are responsible for the

zero resistance in the superconductor. Therefore these electrons carrying current travel in

phase throughout the coil. Any minute change in the flux may change the phase of the

electrons. Due to quantum mechanical boundary conditions this phase change must be

Chapter 3

55

integral multiple of 2 i. The change in phase is an estimate in the allowed quantized

value of change in flux throughout the coil. Any small change in the flux can be

compensated by an additional small current. Voltage across the Josephson junction

arising due to the compensating current enables to measure any small change in the

magnetic flux.

Fig 3. 6: Schematic diagram Josephson junction

RF SQUID is based on the Josephson junction coupled with an inductor in an LC-tank

circuit also known as transformer. LC- tank is excited at its resonant frequency by

applying RF current. The sample is mounted at the centre of this LC tank. Mutual

induction between vibrating sample and the flux transformer induced a change in flux.

The corresponding voltage measured across the Josephson junction is used to measure the

magnetic induction of the sample.

3.2.6. Rutherford Back scattering (RBS)

RBS is a multi-elemental analytical technique based on the electrostatic repulsion

between high energy incident ions and target nuclei. Besides elemental analysis of the

sample it can measure properties such as the thickness, chemical composition at the

Chapter 3

56

surface, crystalline quality and contamination. Its depth resolution ability is from 0-1mm.

It is a fast (typically 10 min or less) and non-destructive technique with high precession

of ± 3% with a great sensitivity.

RBS is based on the elastic scattering of light nuclei H, He, Li having energy in

the range of few MeV. These accelerated ions penetrate into the samples where they lose

energy by two different scattering processes which are explained in next paragraph.

Backscattered ions again lose energy while coming out and finally enter into a detector

which can measure number of backscattered ions as well as their energies. Experimental

setup is shown in the schematic diagram

Fig 3. 7: Schematic diagram of RBS basic function

In RBS back scattered ions lose energy by two different scattering processes

i. Scattering with sample nuclei

ii. Scattering through sample electrons

The first process depends upon the scattering cross-section of sample nuclei and

therefore on atomic number and mass of the sample nuclei. The incident ions thus scatter

on different angles due to the discreet lose in energy, producing a separate peak for

Chapter 3

57

different nuclei. Position of the peaks is finger prints for each nuclei and the height of the

peaks gives the relative intensities of the nuclei present in the sample.

Backscattering of incident ions through results in continuous lose in energy as the

ions pass through certain depth occupied by some specific element as. The energy lose

depends upon the electron density and the distance travelled within the sample.

Continuous lose in energy results in shift of the peak on energy spectrum of the specific

element. The peak fades toward the lower energies. The lose in energy ∆E is directly

proportional to the thickness of the film ∆t as follows

(3.3)

Where = lab scattering angle, k is kinematic factor. Subscripts ‗in‘ and ‗out‘ are

for the rate of energy lose of incident and the backscattered ions.

In short compositional depth profile can be determined from measurement of the

energy spectra. The positions of peaks in the energy spectrum give information of the

elements contained. The width and shifted position of the peaks give the depth profile;

peak heights contain the information of the relative concentration.

By examining the crystal structure through RBS can give an idea about the

chemical structure, however, energy spectra cannot give any information of chemical

structure directly.

Chapter 4

58

Chapter 4

Sample Shape and Fabrication Effects in Exchange Bias

Systems

The magnetic properties of materials are determined by the reversal process. In

single domain granular systems the reversal process is by coherent or incoherent rotation.

In thin films this mechanism can apply but for exchange coupled granular films the

reversal process is usually that of reverse domain nucleation and domain wall

propagation. Hence local effects such as demagnetising fields have a significant influence

since these can affect the nucleation process significantly. In this work the effects of

nucleation is described arising from both sample shape effects and the process used to cut

the sample. It is found that cutting techniques such as the use of ultrasonic cutters leads to

a large increase in nucleation which distorts the hysteresis loop. Deposition through

masks causes shadowing effects at the edges that also distort the loops. Cutting with a

diamond scribe appears to give the best outcome. Implications for devices based on nano-

elements are discussed.

4.1 Introduction

In this work an exchange bias system consisting of a CoFe ferromagnetic layer

deposited on top of an IrMn antiferromagnetic layer is studied which generally exhibits

nucleation controlled behavior. Effects of sample shape and also the effect fabrication

method for the production of the sample on exchange bias reversal mechanism is

studied. Additionally details of the edges of samples produced by different techniques

in an attempt to establish the best method for sample fabrication for nucleation

controlled materials are also studied. Multilayered IrMn/CoFe system is studied

because the exchange bias effect will be critical in the development of GMR, TMR or

spin-torque switched MRAM devices which will consist of lithographically defined

elements39, 102

.

Chapter 4

59

4.2 Experimental

`

All samples studied in this work were produced by sputtering using a HiTUS

sputtering system described by O′grady et al. 93

.

Fig 4. 1: Sample structure

Details of growth conditions are discussed in section 3.1.1. Five samples were

grown with variable bias voltage (200,400,600,800 and 1000V) whilst keeping the

composition. Amorphous Ta was used as seed layer in order to remove substrates

effects due to the availability of similar substrates. Only sample grown at 800V is used

to carry out further studies.

Exchange bias system with the structure shown schematically in Fig 4.1 which

is used in the current studies. This is a standard exchange bias system which is already

reported103

and which has a square hysteresis loop with the reversal controlled on the

first part of the loop by a nucleation process. In exchange bias systems with this type of

structure it is generally the case that the forward going part of the loop is nucleation

controlled.

Samples were prepared by three methods:

1. Samples were cut from a continuous film using a diamond scribe which was

used to score the uncoated side of the substrate and subsequently snapping the

Chapter 4

60

sample.

2. The sample was sputtered through a thin stainless steel mask with a circular hole

cut by either a simple mechanical drill or by laser cutting. In practice It is found

that, for the samples produced by sputtering through a mask, the mask cutting

technique resulted in no discernible change in properties.

3. An ultrasonic cutting device was used that produces circular samples but where

the sample is cut through from the coated side.

In addition to examining the effect of sample cutting; effect of sample shape has

also studied. This was achieved via the cutting of square and round samples as

described above but also by direct growth of identical stacks onto preformed round and

square substrates. In this way any intrinsic effect of sample shape can be distinguished

from sample edge roughness caused by a cutting technique.

Fig 4. 2: SEM images of the edges of the three sample types

Fig 4.2.a) shows the image for the sample cut with a diamond scribe as can be seen in

this image a very crisp and smooth edge results. Fig 4.2.b) shows the edge of a sample

Chapter 4

61

deposited through a circular mask. Here a marked gradation in the film thickness is

clear due to the shadowing effect at the edge of the mask. Fig 2.4 c) shows the sample

cut with an ultrasonic cutter. Here an extremely jagged and fragmented edge results due

to the nature of the cutting process with these devices.

Samples grown onto preformed substrates were a 5mm square and a 5mm

diameter circle. Unfortunately substrates of exactly the same type were not available in

different shapes. However this effect was negated by growing on a Ta underlayer with

an amorphous structure. Hence there are no substrate related texture effects in the films.

4.3 Results and Discussion

4.3.1 Effect of substrate cutting

Fig 4.3. shows hysteresis loops for three samples produced by (a) cutting with a

diamond scribe, (b) depositing through a mask and (c) cutting with an ultrasonic cutter.

Only data for the circular mask produced with a laser cutter is shown because the data

for a similar mask produced with a simple mechanical drill was identical.

Clearly the curves from the three samples show dramatically different effects. The

hysteresis loop for the sample produced by cutting with a diamond scribe (a) shows

clear nucleation controlled behavior with a single reversal occurring just above 500 Oe

and then a rapid reversal to negative saturation. The sample produced by sputtering

through a mask (b) appears to show multiple nucleation events occurring even above

zero field but interestingly shows a similar measured coercivity to that of the sample

cut with a diamond scribe. The sample produced using the ultrasonic cutter (c) shows

an even greater degree of nucleation at low fields and shows a slightly reduced

coercivity compared to the other two samples.

Chapter 4

62

Fig 4. 3: Hysteresis loops for three samples produced by (a) cutting with a diamond

scribe, (b) depositing through a mask and (c) cutting with an ultrasonic cutter

The reason for these effects can be seen in the images in Fig. 4.2. The smooth

edge from the diamond scribe cut sample means that there appears to be very little, if

any, nucleation prior to the major nucleation event at just over 500 Oe. The sample

deposited through a mask shows a clear shadowing effect at its edge which will of

course change the nucleation field as the thickness of the deposited film changes the

demagnetising field at the edges. The effect of using an ultrasonic cutter is catastrophic

with many chips along the film edge leading to areas with a strong demagnetising field

that will readily nucleate multiple reversal events.

However, the interesting feature is that all three samples have a similar

coercivities even if the loop shape as the sample is demagnetised varies dramatically.

This is not a true nucleation field in the conventional sense but rather a field at which

-1.0 -0.5 0.0 0.5

-1.0

-0.5

0.0

0.5

1.0

H [kOe]

a) diamont cut

b) deposted through mask

c) ultrasonic cutter

M/Ms

Chapter 4

63

the domain wall or walls nucleated in the ferromagnet can overcome the exchange

coupling between the AFM layer and FM layer allowing the domain wall to sweep

through the sample.

A further unexpected result occurs on the recoil loop where the sample

deposited through a mask has a significantly different coercivity than the other two

materials. Understanding of the origin of the coercivity in exchange bias systems is not

clear at this stage of research in the related field but this variation in the return

coercivity was observed for both samples sputtered through masks but to differing

degrees. Although this variation is not explainable at this stage yet it is worth to note

that the results are reproducible.

4.3.2 Effects of sample shape

Fig 4.4. a) and b) shows the effect of sample shape. A comparison of samples

deposited directly onto pre-prepared substrates without the use of masks or cutting is

presented. For both square and circular substrates the curves reproduce and both give

the same shape of hysteresis loop and the same value of coercivity.

Fig 4. 4: Hysteresis curve a) and b) shows the effect of sample shape

-1.0 -0.5 0.0

-1.0

-0.5

0.0

0.5

1.0

a) square

b) round

M/Ms

H [kOe]

Chapter 4

64

This result implies that sample shape effects in nucleation materials are

relatively small and that the strong demagnetising fields that would occur for example

in the corner of the square sample, have a very limited effect. This indicates that the

coupling to the AFM layer is dominating over demagnetising effects from samples with

relatively smooth edges. Hence the exact shape of the sample appears to be much less

significant than the sample edge roughness. However it should be noted that when

samples such as these are measured using the vibrating sample magnetometer the

geometric response of the coils or other sensor in the magnetometer means that great

care should be exercised to ensure that the magnetometer is calibrated with a sample of

similar size, moment and particularly shape to that of the sample which is to be

measured.

Nucleation and domain reversal processes in small elements have been

examined in detail using Lorentz TEM by Craig et al46

. These workers found that sharp

corners in the elements played a major role in determining the onset and progress of

reversal. However the samples studied were Permalloy which is magnetically soft ( c

<100 Oe) and hence demagnetising fields due to a sample shape may easily nucleate

reverse domains. In a real MRAM device element a significantly higher reversal field

will be required. From our study it would appear that in these circumstances overall

element shape will be less critical than edge roughness. This will pose a significant

challenge for the lithography since edge roughness will cause an overall change in the

reversal process of an element. This will also contribute significantly to a switching

field distribution across the sample due to non-uniform roughness in different elements.

Summary

From this work it is concluded that great care must be exercised when making

magnetic measurements on magnetic thin films where the reversal process is controlled

by domain nucleation. Surprisingly the effect of sample shape is relatively minor in

those materials is studied here but this may not always be the case for samples having a

Chapter 4

65

lower anisotropy than the exchange bias materials studied here. The critical parameter

appears to be the edge roughness of the samples where it is essential to obtain sharp

edges that are smooth using tools such as a diamond scribe but presumably not a

diamond saw where chipping at the edge of the samples would also occur. The use of

ultrasonic cutters, which are popular for taking samples from films deposited on wafers

or thin film discs, are completely inappropriate as they will give rise to a completely

false shape to the hysteresis loop. The use of masks is only appropriate with great care

due to the shadowing effect which occurs at their edge. It is found that the samples for

VSM measurements and other open circuit techniques should always be prepared on

pre-cut substrates or by using a diamond scribe or similar cracking tool to cut the

sample. Even when using diamond scribes the samples should always be cut from the

reverse side to avoid chipping the magnetic thin film thereby leading to an excess of

nucleation sites. The significance of edge roughness will make demands on lithography

for MRAM technology.

Chapter 5

66

Chapter 5

Interface Modification and Training Effect in Exchange

Bias Mutilayer System

The effects of inserting mono-atomic layers of Mn at the interface of IrMn/CoFe

bi-layers were investigated. Samples were grown using HiTUS sputtering technology at

200, 400, 600, 800 and 1000V bias voltages. Magnetic characterization was carried out

using a MicroMag 2900 AGFM and ADE Vector Model 10 VSM whilst grain size

analysis was carried out using a JEOL JEM-2100 TEM and a Zeiss particle analyser.

Exchange bias of the bi-layer grown at 800V was increased from (-411±10) Oe to (-

479±10) Oe with the insertion of 2Å thick Mn layer whilst coercivity was increased from

(212±10) Oe to (252±10) Oe and blocking temperature was decreased from 408.3 K to

380.0 K. In addition to this it was found that blocking temperature decreased by 7.5 K

when set at 20 kOe as opposed to 5 kOe. The cause of the changes to the coercivity,

exchange bias and blocking temperature could be due to modification of the anti-

ferromagnetic spin structure at the interface.

5.1 Experimental

5.1.1 Fabrication process and conditions

All samples were sputtered under the same conditions as described in section

3.1.1. A few samples were grown without the 0.2nm Mn layer as a control at 400, 600

and 800V in order to measure the effect of the Mn . The sample structure is shown in Fig.

5.1

Chapter 5

67

Fig 5. 1: Sample structure with Mn doping

5.1.2 Setting process

5.1.2.1 York Protocol

The measurement of the exchange bias phenomenon presents a number of

challenges. Firstly, for materials which have technological applications it is difficult to

achieve TN of the AFM layer without causing damage to the sample as a whole. For all

current applications, as well as within this study, the alloy used for AFM layers is IrMn3.

This material has a value of TN 690K, at which point diffusion of the multilayer would

occur. It is important to note, however, that field cooling from temperatures as low as

475K has been reported to result in the setting of IrMn layers. This is due to the thermal

activation of the orientation of the AFM lattice within each grain58

. Secondly, due to the

AFM grains being thermally unstable, the state of order of the AFM can change during

measurement, as shown by Fulcomer and Charap 51

. This means that for measurements of

the parameters and c results are highly non-reproducible. Thirdly, when a

conventional magnetic measurement is made the AFM layer itself gives no signal.

Chapter 5

68

Fig 5. 2: a) Schematic diagram and b) measurement steps of the York protocol. 14

O‘Grady et al. 14

proposed that through the careful management of the thermal

and magnetic history of the AFM a uniform initial state could be produced. This would

allow for reproducible results and therefore comparability between multiple

measurements. O‘Grady et al. 14

also suggested an interesting solution to the blindness of

the AFM layer to an applied field. Through changes in the state of the AFM grains

changes in the properties of the adjacent FM layer occur. Thus through the careful control

of the magnetic and thermal history one can infer changes in the AFM grains through

observations in the changes of the FM layer.

To control the magnetic and temperature history of the samples a set of protocols

were used 14

. Firstly, the AFM layer was set in a reproducible manner. This was done by

Chapter 5

69

applying a magnetic field in the direction of the known FM layer easy axis that was

sufficient to saturate the FM layer. Then the sample was heated to a maximum

temperature ( ) at which interfacial diffusion did not occur. For this study the setting

of the samples was undertaken using both a hand built ‗annealing‘ furnace and an ADE

Vector Model 10 VSM with a value of of 90mins and a maximum temperature of

498K. Secondly, the temperature at which no thermal activation occurs ( ) was

established. This was done by cooling down the sample, with the setting field still

applied, to which was determined by first cooling down the sample to a trial .

The sample was then held in the state of magnetisation for a short period (1min) and a

hysteresis loop was measured. The process was then repeated however with the FM layer

reversed for a period of 30mins before a measurement was taken. If the hysteresis loop

was not reproduced then thermal activation had occurred and a lower value of would

need to be used. It is important to take note that two hysteresis loops should be taken at

each stage to eliminate the training effect. The training effect occurs only for the first

hysteresis loops and is thought to be due to spin-flop coupling which is removed by the

first hysteresis loop 104

. In the case of this study all samples used were thermally inactive

at room temperature so > 298K. This process is shown schematically in Fig 5.2 a) and

the precise measurement sequence in Fig 5.2 b) 14

.

5.2 Results and Discussion

5.2.1 Grain Size Analysis

The image shown in Fig 5.3 gives a good example of a high resolution TEM

image taken at a magnification of 80,000x. The grains can be clearly seen as black spots

of varying sizes.

Chapter 5

70

Fig 5. 3: TEM Image

These grains were counted using the Zeiss grain size analyzer, as described above,

and then plotted to a lognormal distribution as shown in Fig 5.4. As can be quite clearly

seen there is an equivalently equal increase in average grain size with the increase in bias

voltage as well as an increase in the standard deviation of the grain sizes for larger bias

voltages. However it is important to note the mediocre fit of the log-normal distribution

curve to the data as well as the difference in height between the 400V and 600/800V

samples. Although the CPM was utilized it is highly likely that too few grains were

measured. To obtain a good fit between the theory and data 150-200 more grains should

have been measured.

Chapter 5

71

Fig 5. 4: Graph showing the log-normal distribution of grain sizes.

400V

600V 800V

Chapter 5

72

Curve fitting was obtained by the equation 2.5 and rewritten below as

This data, in comparison with the recorded sputter rates along with the calculated

average grain sizes, is shown in Table 5.1. As expected there is a near constant increase

between each bias voltage. This is in line with the theory as described by Vopsaroiu et al.

57 in which it was found that the mean grain size depends on both the sputter rate and

nucleation rate. As mentioned an increase in standard deviation was observed, which

again confirms the theory in 57

, where it is shown that as sputter rate increases the range

of possible grain sizes increases. This increase in standard deviation could also be used to

explain the flatness of the 600/800V curves. However this is unlikely as although

flattening would be expected it would not occur at such a magnitude.

Table 5. 1: The results obtained for the average grain size at different bias voltages.

Bias Voltage

Sputtering Rate

of AFM (nm/s)

Average Grain

Size (nm)

Standard

Deviation

400V 0.10 4.9 0.54

600V 0.12 5.8 0.42

800V 0.14 7 0.35

5.2.2 Exchange Bias

The hysteresis loop, as shown in Fig 5.5, demonstrates a clear example of an

exchange biased system measured using the York protocol. The values for and c are

straightforwardly obtained using the standard methods: is the offset of the loop centre

from zero fields; c is half its width. Errors in these calculations were calculated by

observing the nearest data point to zero magnetisation and making a reasonable estimate

as to the possible variance in value.

Chapter 5

73

Fig 5. 5: Typical hysteresis loop obtained using the York protocol.

The effects on both and c of a 2Å Mn dusting layer at the AFM/FM interface for

different bias voltages are shown in Table 5.2. The most striking feature is that in the

600/800V cases an increase in of ~10% is observed, accompanied by an enhancement

in by ~13%. This in agreement with the work of Tsunoda et al.105

in which the

interfacial layers of 0-1nm was studied. They believed that the cause of the enhancement

of and c was due to the modification of the AFM spin structure at the interface. This

is a fair assessment; however results in the following section raise more questions as to

the effect.

.

Table 5. 2: The results obtained for and c at different bias voltages.

Bias

Voltage

400V

Control

400 Mn

doped

600V

Control

600 Mn

doped

800V

Control

800 Mn

doped

Hex

(±10.0)

(Oe)

-532

-569 -429 -496 -411 -479

Hc

(±10.0)

(Oe)

-271

-241 222 256 212.3 252

Chapter 5

74

In the 400V case there appears to have been no real change in , however an

enhancement in C by ~11% is observed. The lack in enhancement is odd judging a

similar increase in with that of the 600 and 800V cases is seen. Due to the 200V

samples being thermally unstable at room temperature it is reasonable to assume that the

400V sample suffered from a slight instability.

5.2.3 Blocking Temperature

The effect of heating a sample under a field opposite to that of the set

magnetisation is shown in Fig 5.6. The loop shift is clearly identifiable, from which the

value for can be easily obtained by York protocol 14

; is the temperature at which

goes to zero under a reverse field. Errors in these calculations were estimated by

taking into account the errors as shown in Table 5.1 as these were used, in association

with other results, to calculate the blocking temperatures.

Chapter 5

75

Fig 5. 6: Typical loop shift for different temperatures under a constant reverse field.

Chapter 5

76

The effects on of a 2Å Mn dusting layer at the AFM/FM interface for different

bias voltages and reverse fields are shown in Table 5.2. The most obvious feature is that

in all situations there was ~5% decrease in . This result seems to go against the first

impression of what should occur. With the value of being determined by the number

of AFM grains that are both set and thermally stable, an increase in implies that there

are a larger number of AFM grains contributing to the total value. This could be

interpreted as either more grains have become set or thermally stable. In the former

situation should increase; however in the latter situation should decrease. As such,

the decrease in could be due to the high moment of the Mn holding the grains

previously too small to be set in place.

Fig 5. 7: Measurement of Blocking Temperature (TB) by York Protocol

Chapter 5

77

Table 5. 3: The results obtained for TB at different reverse fields.

600V

Control

0.5T

600V

Control

2T

600V

Mn

doped

0.5T

600V

Mn

doped

2T

800V

Control

0.5T

800V

Control

2T

800V

Mn

doped

0.5T

800V

Mn

doped

2T

TB

(K) 421.9

417.21

409.7 396.9 412.6 408.3 388.4 380.0

A less obvious, but even more interesting feature, is that in both the 600/800V Mn

doped samples there was ~1-2% decrease in blocking temperature between the 0.5T and

2T reverse fields. This result is intriguing as at such fields the FM layer is completely

saturated and in theory the AFM shouldn‘t be able to see the field. A possible source for

this feature could be dirt or other contaminants at the interfacial layer, however further

studies are needed to confirm this.

5.3 Training Effect

The training effect of IrMn/CoFe bilayer system with different seed layers Cu and

NiCr is studies. Training effect have dependence on many factors e.g. temperature,

thickness of the FM and AFM layers, grain structure, crystallinity as well as the

orientation on FM/AFM on the interface. However complete mechanism is still under

debate. Only theoretical model at present is described by Hoffman62

showing that the

training effects can be seen for AFM with biaxial anisotropy. Few researchers have

shown the particle or grain size dependence of the training effect106

. Recently Kevin

O′grady successfully proposed a theoretical model describing various features of

exchange bias. In this work the training effect of sputtered IrMn/CoFe bilayer is

described by relating to the average grain size. It is found that training effect if maximum

for smaller grain size and vice versa for Cu and NiCr seed layers used in this study. Also

the dependence of training effect on seed layer is also discussed.

Chapter 5

78

5.3.1 Sample Preparation

All samples studied in this work were produced by sputtering using a HiTUS

sputtering system described by O′grady et al93

and also described in section 3.1.1.

A dc field of 500 Oe was applied during the deposition to pin the direction of

IrMn/CoFe system. The sputtering conditions were kept same for all the spattered films.

Each sample had 5nm NiCr as a seed layer and caped with 10 nm Ta layer to avoid any

oxidation. For comparison similar samples with 5nm Cu seed layers was also prepared

under the same condition. Each series consists of 6 samples grown with variable bias

voltage (200,400,600,800, 980 and 1000V) whilst keeping the composition other than

the difference in type of the seed layer. The magnetic properties of all the samples were

determined with VSM. The training effect of the hysteresis loops was measured by

cycling the applied field continuously starting from pinned direction anti-parallel to

pinned direction. Only first two cycles were recorded. All the measurements are taken

from 2000 Oe to -2500 Oe so that the ascending and descending arm of the loop feel

the same amount of applied field during the measurement.

Fig 5. 8: Schematic diagram of sample structure

Exchange bias system with the structure shown schematically in Fig 5.8 which is used

in the current studies. This is a standard exchange bias system which is already reported

and discussed in the previous sections of this thesis

Chapter 5

79

5.4. Results and Discussion

Fig 5.9 shows the normalized hysteresis loops of samples produced with different

seed layer namely NiCr and Cu fabricated with HiTUS and characterized according to

York protocol. Only first two loops are taken to observe the training effect in our samples

to compare with size and seed layer structure. A clear shift in the descending curve is

pronounced, while almost no shift in the ascending curves. The decrease in coercivity and

exchange field can be observed and also presented in the table 5.4.

Fig 5. 9: Training effect with a) NiCr seed layer b) Cu seed layer

Chapter 5

80

Table 5.4 and Fig 5.10 shows a comparison between the Bias voltage and the

observed training effect. The bias voltage is in turn related to the grain size due to the

characteristic fabrication process by HiTUS. This is also reported in the literature and

further confirmed in Fig 5.10 b) by plotting training amount vs. the grain size for NiCr

seed layer. The grain size follows lognormal distribution which is discussed in section

5.2.3 with median grain sizes shown in Table 5.4. It can be seen that the grain size is

maximum at an operating bias voltage of 200V and decreases with the increase bias

voltage. However the grain size for 1000V bias voltage again decreases unexpectedly.

This might be due to the fact that Ar plasma used for sputtering was not stable at 1000V

and the sputtering rate was too fast. For confirmation another sample was fabricated with

bias voltage 980V with the same and before sputtering plasma was allowed to stabilize

for 5-10 minutes. The grain size for the samples fabricated at 980V is in good agreement

and in sequence with the grain sizes of the samples grown on bias voltage 200,400,600

and 800V.

Fig 5. 10: a) comparison of bias voltage vs. training effect for NiCr and Cu under layer b)

grain size vs. training effect for NiCr under layer

Chapter 5

81

Table 5. 4: Amount of training for Cu and NiCr underlayer

Bias Voltage

(V) Training Field ( 10 Oe) Grain size (nm)

NiCr under layer Cu under layer NiCr underlayer

200 28.5 73.2 3.9

400 21.1 53.2 4.2

500 20 52.5 4.4

600 17.5 50.9 5.2

800 5.3 42.3 5.9

980 3.8 40.3 6.2

1000 11.3 52 5.2

The training effect found in the sputtered films in the 1st two loops might be due

to presence of metastable biased spins configuration in the AFM layer. During each cycle

these AFM spins turn into the energetically favorable spins. This change in initial spin

distribution brings a decrease in the exchange anisotropy of AFM layer. In

nanostructured crystalline material large numbers of spins are located at the grain

boundaries. The disordered structure of the spins results in fluctuations in the magnetic

anisotropies which alternately affects the exchange strength. Thus strength of the

training effect is more pronounced in small grain size107

.

Fig 5.10 b) shows a plot between bias voltage which alternately predict the grain

sizes and the training effect. It is found that the training effect is found more pronounced

for the films grown on the Cu seed layer and weak for the films grown on the NiCr seed

layer. However both plots follow the same trend, again confirming the dependence of

training effect on the grain size.

Marian Fecioru-Morariu108

has reported that a better quality growth of the (110)

FeMn has less structural defects as compared to the (001) oriented film. He concluded

that rearrangement takes place during field cycling of the AFM domain structure. This

rearrangement arises from the presence of defects in the AFM lattice, which is more

Chapter 5

82

pronounced in (001) grown FeMn as compared to (110) FeMn. This explains why the

training effect is stronger for the (001) sample as compared to the (110) sample.

Nick et al.109

XRD scans found that there is strong peak from the FCC IrMn (111)

planes when growing a similar sample structure. There was also a possible peak from the

BCC CoFe (110) planes. For the samples with Cu seed layers no peaks were observed

indicating that there is no strong texture and poor crystallinity in general. This indicates

that the IrMn when grown on NiCr has a strong in-plane (111) texture where the (111)

planes lie parallel to the substrate. It is therefore concluded that in the set of samples with

Cu seed layer provides poor crystallinity which alternately provides more structural

defects and grain boundaries of the polycrystalline IrMn is responsible for the more

pronounced increase in the training effect value.

Summary

The effects of placing a 2Å Mn dusting layer at the AFM/FM interface on the

exchange bias, coercivity and blocking temperature of IrMn/CoFe bilayers were

investigated. The purpose of placing Mn dusting layer was to modify AFM/FM interface

on which all the above properties depend. The study was taken in search of enhancement

in TB. of the bilayer was mildly enhanced from (-411±10) Oe to (-479.7±10) Oe,

while was increased from (222.3±10) Oe to (256.5±10) Oe. It is believed that the

enhancement of and is due to the modification of the AFM spin structure at the

interface.

However as the blocking temperature was reduced from 408 K to 380 K with the

addition of Mn, it is believed that grains that were previously too small to be thermally

stable were somehow set. This could be due to either a chemical reaction with the Mn at

the interface or due to exchange effects.

Chapter 5

83

Training effect is found more pronounced in smaller grains, which is due to the

more disordered structure of the spins at the grain boundaries. Also it is found that

training effect have a significant influence on the crystallinity of AFM at the interface.

Chapter 6

84

Chapter 6

Transition Metal Doped TiO2 Thin Films by AACVD

This chapter is devoted to synthesis of Ni, and Co doped TiO2 thin films, optimization of

the parameters of the AACVD grown thin films, the study of structural, morphological

and magnetic properties. The optimization of the process parameters (temperature, carrier

gas flow rate, and deposition time) for the thin film deposition by AACVD technique was

defined. The growth temperature for thin films on silicon substrates were 450°C for Ni

doped TiO2 and 650°C for Co doped TiO2,The carrier gas flow rate was kept 120mL/ min

and a deposition time of 20 minutes for all the films grown in current studies. Anatase

phase formation was obtained for Ni doped TiO2 thin films as shown in XRD patterns.

For Co doped TiO2 thin films; at a synthesis temperature around 650°C oxygen deficient

rutile phase namely magneli phases (Ti3O5, Ti4O7, TiO etc) were found. SEM images

show that as grown Ni and Co doped TiO2 thin films in this study were polycrystalline.

SEM images also confirm that the films were highly crystalline. Rutherford Back

Scattering (RBS) was carried out to confirm the TM doping concentration and to find the

thickness of the films. The thickness of the films found to be in the range 200-260nm

except sample doped with 2wt. % Ni has film thickness of ~800nm. The SQUID

measurements of TM doped TiO2 thin films used to show room temperature

ferromagnetism in the films.

6.1 Introduction

A range of metal oxide semiconductors are reported in the literature including

ZnO, TiO2 in the form of thin films and nanoparticles for various applications12, 110-128

.

Cynthia Edusi et al129

optimized the conditions in controlling the TiO2 phase. They have

shown that at 400 C anatase phase of the TiO2 can be achieved on the glass substrate.

However there are only a handful of reports on transition metal doped TiO2 thin films

prepared by AACVD technique130-131

.

Chapter 6

85

In this section Ni doped TiO2 thin films are reported synthesized by AACVD. The

optimized conditions and choice of precursor are made to get the most suitable anatase

phase for DMS applications. Suitable precursor to get the proper stoichiometry is

designed by modifying the synthesis procedure adopted by Tahir et. al.132

.

6.2 Nickel Doped TiO2 Thin Films

There are only a handful of reports in literature on Ni doped TiO2 thin films

which may be due to difficulties to prevent secondary phases and precise understanding

in RTFM in this system. However, there are few reports on Ni doped TiO2 films and

particles with various synthesis techniques and results. Earlier Hong et al. 133

reported

RTFM in Ni doped TiO2 anatase and rutile phases grown on SrTiO3 and LaAlO3 by laser

ablation. Cho et al.134

reported highly resistive NixTi1-xO2 films by solgel method. They

concluded with two different mechanisms for justification of RTFM in their report i.e.

due to formation of Ni and NiO clusters. However possibility of carrier mediated

ferromagnetism was ruled out due to high resistivity in their samples134

.

D.L. Hou et al.135

reported some exiting results in NixTi1-xO2 anatase films grown

by reactive magnetron sputtering on amorphous SiO2 substrates. MFM results shown

regular and clear magnetic domains for the first time thus predicting intrinsic

ferromagnetism in Ni doped TiO2 anatase films135

.

Fig 6. 1: a)Topography image and corresponding b) MFM images of Ni doped TiO2 thin

films135

Chapter 6

86

Few other reports also show variation in RTFM in Ni doped TiO2 structures. Uhm and

coworkers136

prepared Ni:TiO2 powder by mechanical alloying by varying Ni

concentration 0-12wt.%. All the three phases were found in XRD data and an increase in

ferromagnetic behavior upto 8 wt.% of Ni concentration. Cabrera et al.137

reported TM

(Fe, Ni, Mn and Co) doped TiO2 powder obtained by ball milling. However Ni clusters

were found magnetically ordered in their report. More recently Hoa et al.138

reported

intrinsic RTFM in Ni doped TiO2 nanowires synthesized via solvothermal technique.

Coesivity value of ~125Oe confirmed the intrinsic nature of RTFM in Ni doped TiO2

nanowires. Owing to higher Ms value in undoped samples also suggested surface defects

playing important role in ferromagnetism. Badaur et al.139

reported RTFM solgel derived

Ni doped TiO2 powder. Any contribution in RTFM from any secondary phase was ruled

out by using HRTEM and XPS analysis. So far there is no report with any TM doped

TiO2 thin film by AACVD for DMS applications. It is important because various

properties are strongly influenced by the synthesis routes adopted as can be seen in above

paragraph and discussed in section 6.1.

6.2.1 Experimental

Thin films of Ti1-xNixO2 (x=0.02-0.15) were prepared by Aerosol Assisted

Chemical Vapor Deposition (AACVD). Ni doped TiO2 films were deposited by AACVD

on silicon substrates. The precursor [Ni2Ti2(OEt)2(l-OEt)6(acac)4] synthesized for current

studies has already been reported by Asif et al.132

. The stated precursor was synthesized

for 1:1 molar ratio of Ni to Ti. For current studies precursor was prepared to achieve

2,4,6,8 and 15 wt.% Ni concentration. In detailed synthesis process 1.00 g (4.36mmol) of

Ti(OEt)4 to 15 ml toluene and 1.11g (4.36mmol) Ni (acac)2 was dissolved in 15 ml of

toluene separately. Then stoichiometric amount of Ni (acac)2 solution added to Ti(OEt)4

solution to obtain 2,4,6,8 and 15 wt.% Ni. The solution thus formed homogeneous

mixture of [Ni2Ti2(OEt)2(l-OEt)6(acac)4] and Ti(OEt)4. AACVD was performed using the

aforementioned precursor.

Chapter 6

87

Substrates were washed with acetone, isoproponol and ethanol ultrasonically. The

substrate kept in ethanol before use. Modified AACVD apparatus was used as reported

by Asif et al.132

. Instead of using reactor chamber, hot plate was used to decompose the

precursor on Si substrates. PIFCO air humidifier was used to generate the aerosols and

Argon gas was used as a carrier gas in order to prevent any possible reaction of precursor

with air before deposition. All the films were grown on 450 C and the Argon flow rate

was kept on 120 mL/min with the help of a micro controller. Films were deposited for 20

min each. After deposition hot plate and Argon flow turned off until the room

temperature was achieved. Scotch tap test was conducted to check the adhesion of the

films.

6.2.2 Results and Discussion

6.2.2.1 XRD analysis

The crystallinity, phase and preferred orientation of the films was investigated

using PANalytical X-ray diffractometer model XPert PRO with primary monochromatic

high intensity Cu-K (λ = 1.54184 Å) radiation. Data was acquired over the range of 2θ

from 10 to 80 with a slow scanning speed of 1 step/s and each step was 0.02 . The

results show anatase up to 8 wt.% Ni concentration. In general, Oxygen deficient

environment is favorable for the growth of rutile phase due to 3 times smaller c/a ratio of

rutile (2.52) to that of anatase (1.54). Formation of predominantly anatase phase here is

due to the fact that acac ligand and chelating ethoxide used for the synthesis of films

contain oxygen atoms which are coordinatively saturating both Ni and Ti as shown in

Fig. 2.6. 132

This oxygen thus eliminates need of any additional oxygen for the formation

of anatase TiO2. Post deposition cooling in air also pronounced anatase phase.

XRD shows an overall increase in crystallinity with increasing Ni contents, plus a

greater peak intensity in the (101) reflection. This indicates an increase in the c-axis

crystal orientation perpendicular to the substrate surface with increased doping

concentration. Further there is a slight shift in the 2θ values towards lower values,

Chapter 6

88

specially, considerable in (101) plane suggesting that Ni contents are doped within the

lattice of TiO2.

The small shifting in (101) peak towards lower angle suggest that Ni+2

has

replaced Ti+4

rather Ni+3

. It can be explained by taking comparison of the ionic radii of

Ti4+

(0.745 Å) ion with Ni2+

(0.830 Å) and Ni3+

(0.740 Å) ions.140

Further there is no

additional peak of Ni or NiO from 2-8wt.% Ni doping.

The crystallite size using Sherrer formula for 2,4,6,8 and 15 wt% Ni doped

samples are 25.1nm, 29.6nm, 25.8nm, 23.4nm and 13.4nm respectively as calculated

from maximum intense peak (101).

Fig 6. 2. XRD pattern of Ni (2%) doped TiO2

Chapter 6

89



Chapter 6

90


Fig 6. 6: XRD pattern of Ni (15%) doped TiO2 thin films

Chapter 6

91

Fig 6. 7: Crystal structure of [Ni2Ti2(OEt)2(l-OEt)6(acac)4]132

XRD shows an overall increase in crystallinity with increasing Ni contents, plus a

greater peak intensity in the (101) reflection. This indicates an increase in the c-axis

crystal orientation perpendicular to the substrate surface with increased doping

concentration. Further there is a slight shift in the 2θ values towards lower values,

specially, considerable in (101) plane suggesting that Ni contents are doped within the

lattice of TiO2.

The small shifting in (101) peak towards lower angle suggest that Ni+2

has

replaced Ti+4

rather Ni+3

. It can be explained by taking comparison of the ionic radii of

Ti4+

(0.745 Å) ion with Ni2+

(0.830 Å) and Ni3+

(0.740 Å) ions.140

Further there is no

additional peak of Ni or NiO from 2-8wt.% Ni doping.

Chapter 6

92

Table 6. 1. Lattice parameters, Cell volume and crystallite size calculated from XRD data

composition a(Å) c(Å) Lattice Distortion (c/a)

Cell volume V (Å)3

Crystallite Size (nm)

Reference (83-2243)

3.7842 9.5146 2.5142 136.25 --------

0.02 3.765 9.4396 2.507198

133.8085 25.2

0.04 3.7842 9.4390 2.494318 135.1681 30.2

0.06 3.766 9.4368 2.505789 133.8398 26

0.08 3.769 9.456 2.508888 134.3259 26

Reference (83-0198, 33-960)

5.0311 13.7960 ----------- 302.42 ---------

0.15 4.9989 13.785 2.757607 344.4734 14.4

The crystallite size using Sherrer formula for 2,4,6,8 and 15 wt% Ni doped

samples are 25.1nm, 29.6nm, 25.8nm, 23.4nm and 13.4nm respectively as calculated

from maximum intense peak (101) and given in Table 6. 1.

At a Ni concentration of 15wt.% NiTiO3 phase appeared. However no Ni or NiO

peaks are observed within sensitivity range of XRD. These phases might exist in

amorphous phases, or NiO has reacted with TiO2 to form NiTiO3 at 15wt.% Ni

concentration.

A small peak of TiO2 rutile phase can be seen in all samples. Ni would have

increased the number of oxygen vacancies which might be responsible for the

transformation of anatase to rutile and NiTiO3. 141

6.2.2.2 Rutherford Back Scattering

Rutherford backscattering spectrometry (RBS) measurement of Ni doped TiO2

samples were performed using Accelerator Facility at Experiential Physics Labs, NCP.

RBS is a non-destructive analysis technique for thickness and composition of elements of

materials.

A collimated 2.0 MeV He+ beam produced by 5UDH-2 Pelletron was used for

Chapter 6

93

RBS channeling measurements. The sample was mounted on a high precision (0.01 )

five-axis goniometry in a vacuum chamber, so that the orientation of this sample relative

to the He+ beam could be precisely controlled. The backscattered particles were collected

by Silicon Surface Barrier (SSB) detector (FWHM 11 keV and area 50 mm2) using

energy resolution of 25 keV placed at angle of 170 .

Simulation was performed by SIMNRA software for RBS data from Accelerator.

RBS spectrum of Ni doped TiO2 thin films (black line) along with simulation spectrum

(red line) is shown in Fig 6.2

The overlaid spectra of Ni doped TiO2 films with all five compositions are also

shown in the Fig 6.3. The peaks at high energy (channel number) correspond to the

scattering from the Ti and Ni. However no Ti or Ni peaks corresponding to the scattering

from Si are observed. By measuring the intensities (yield) of the Ti and Ni signals and

correcting from the scattering crossection of the Ti and Ni exact ratio of the Ni/Ti is

determined and shown in the Table 6.1. From width of the RBS peaks thickness and

composition of the films were determined and shown in the table 6.1. The composition of

the oxide films and the Ni concentration were determined by comparing the experimental

RBS spectra with those obtained by simulation.

Chapter 6

94

Fig 6. 8: Comparison of experimental and simulated RBS spectra on Ni doped TiO2 thin

film

Both spectra indicate that the Ni doped TiO2 films are stoichiometric. The small

projection at the edge of the Ti portion of the spectra may be due to the close atomic

weight of the Ni and Ti the peaks could not be resolved due to the limitations of the RBS

studies.

0 500 1000 1500 2000

Channel

0

5

10

15

20

25

30

35

Nor

mal

ized

Yie

ld

0.5 1.0 1.5 2.0

Energy (MeV)

Ni1(2.085MeV)

Ni

Ti

Ca & Si from substrateO

Chapter 6

95

Fig 6. 9: RBS spectra of Ni doped TiO2 thin films with various concentrations

Table 6. 2: Ni concentration and film thickness as calculated from RBS spectra

Sample Doping concentration (wt.%)

Calculated Measured

Films thickness

(nm)

Ni-2 2 1.9 0.1 789 5

Ni-4 4 3.8 0.1 228 5

Ni-6 6 5.9 0.1 217 5

Ni-8 8 7.7 0.1 255 5

Ni-15 15 14.3 0.1 208 5

Chapter 6

96

6.2.2.3 Scanning electron microscopy

SEM images of Ni doped TiO2 thin films deposited by AACVD (Fig 6.4) show

compact and smooth film morphologies with homogenously dispersed grains. Individual

grains are well defined and clear grain boundaries can be seen. The packing density of the

microstructure and the grain sizes apparently seem to be affected by variation of the Ni

concentration SEM images. Further it can be seen that all the films are nanoporous with a

very uniform grain size distribution.

The average grain size lies in the range of (40-60nm) which is 2-3 times greater

than crytallite size determined by XRD data which is righly justified. Shape of the grains

changed with increase in Ni concentration as shown in Fig. 6.4. The change is the

structure of the grains is attributed to the growth stress of the films, radius and the

concentration of the Ni ions.

Apparantly the films prepared by different concentrations of Ni are of different

colors, with a thickness of as calculated by the RBS. Apparently the films turns into

yellow green, with the increase in the Ni concentration. As widely known, the color of

Ni+2

ion is generally green, which corresponds with the colors of film, which illustrates

that the Ni+2

may have come into the films142

. In case of sample with 2wt.% Ni, a

compact dense morphology was obtained. The grains are of flower like shape. At 4 wt.%

Ni concentration grains have retained their flower like shape and the roughness of film is

pronounced with increased porosity. A deeper examination shows that even small grains

are aggloromated to form a bigger grain. The SEM image of the films with 6wt.% Ni

concentration shows that the growth of particles is turned into elongated shape. There is a

significant enhancemend in the porosity of the films. Another change in the shape of the

grains is observed in the films with 8wt.% Ni concentration. The film seems to be very

compact. With the increase in the number of the grains the porosity is decreased as can be

seen in Fig 6.4 e). In 15wt.% Ni concentration the shape of the grains is changed into leaf

like structures.

Chapter 6

97

Fig 6. 10: SEM images of Ni doped TiO2 thin films with a) 2wt.% Ni b) 4wt.% Ni c)

6wt.% Ni d) 8wt.% Ni and e) 15wt.% Ni doping

a) b)

c) d)

e)

Chapter 6

98

6.2.2.4 Magnetic properties

Room temperature magnetic hysteresis loops of Ti1-xNixO2 for x=0.02, 0.04, 0.06,

0.08 and x=0.15 are shown in Fig 6.5 respectively. All the samples exhibit

ferromagnetism both at 100K and 300K. The ferromagnetic behavior is found to have a

strong dependence on Ni concentration and the synthesis route. Fig. 6.5 shows saturation

magnetization, Ms, variation with Ni concentration at two different temperatures i.e.

100K and 300K. It can be seen that the magnetization increases with increasing Ni

concentration from 2wt.% Ni concentration to 8wt.% Ni concentration. However, for the

sample with 15wt.% Ni concentration the sample is not saturated and paramagnetic effect

is dominant which is attributed to the secondary phases present in the sample like

NiTiO3.

There are many possible mechanisms which can establish magnetic coupling in

semiconducting TiO2 i) super exchange, ii) double exchange, iii) RKKY interaction, and

iv) impurity band exchange. Carrier mediated interactions occurs in situation where high

carrier concentration (electrons or holes) exists. These interactions include RKKY and

double exchange. Transition metals are normally doped in low concentration due to the

unfavorable precipitation problems in the form of secondary phases in diluted magnetic

oxides. Now, if TM doping concentration is low than percolation threshold (xp) which is

essential for nearest neighbor coupling; super exchange and double exchange interaction

becomes irrelevant for the explanation of ferromagnetic ordering. In oxides xp is

typically 25-30% which can be calculated from xp~2/Z, where Z is the cation

coordination number.79

However, in relevance to our synthesis technique and doping concentration

formation of bound magnetic polaron143

is most likely the reason in establishing the long

range ordering. Since the Argon gas is used as the carrier gas which may have resulted in

the oxygen vacancies. It is believed that local carrier are involved in RTFM in our

Chapter 6

99

synthesized Ni doped TiO2 thin films. Further formation of oxygen deficient magneli

phases in small amount cannot be ruled out since films are synthesized in inert

atmosphere, although, there is no such clear evidence in XRD. However the oxygen

vacancies are the most likely source of RTFM. Bound Magnetic Polaron (BMP) model143

more appropriate for elucidation of magnetic behavior for current study. From analysis of

XRD it is concluded that Ni+2

may have replaced Ti+4

on octahedral site. In order to

maintain the charge neutrality, substitution of Ni+2

results in generation of positively

charged oxygen vacancies. Theses oxygen vacancies capture donor electrons and

constitute an F-centre. The strong electron-phonon interactions establish a strong

polaronic effect in TiO2 which enhances the carrier effective mass.144-145

Polaronic

electron captured in such F-centers tends to spend their time in hydrogen like orbital

which effectively overlap the d shells of the neighboring magnetic atoms. Thus, F-center

bound magnetic polaron formed by an electron trapped by an oxygen vacancy,

surrounded by magnetic impurity ions (Ni+2

in this study) is a possible mechanism for the

ferromagnetism 143

.

These F-centre BMPs grow at low temperature resulting in long range ordering.143

This may be the reason that at low temperature 100K the moment in all the samples has

increased. In sample with 15wt.% Ni concentration is not saturated. XRD results show

NiTiO3 and TiO2 rutile phases at 15wt.% Ni concentration with a low crystallinity.

NiTiO3 shows antiferromagnetic behavior below 23K as evident from neutron

diffraction and magnetic susceptibility.146

Also NiTiO3 is paramagnetic at room

temperature. However rutile phase may have contained some amount of Ni doped in

amorphous form, which shows hint of ferromagnetism.

Chapter 6

100

Fig 6. 11: Magnetic moment of Ni doped TiO2 thin films at 100K

Fig 6. 12: Magnetic moment of Ni doped TiO2 thin films AT 300K

Chapter 6

101

Thus we can summarize the BMP model as the interaction of the trapped electron

with the host lattice that lies within its orbit ferromagnetically. This leads to a bound

polaron with a large magnetic moment. If the density of BMP is less, then they do not

strongly interact resulting in an insulating paramagnetic phase. However, for a certain

polaron density they couple ferromagnetically.

Table 6. 3: Magnetic moment of Ni doped TiO2 thin films at 100K and 300K

Sample Moment (emu/cc)

100K 300K

Ni-2 3.9776 3.348

Ni-4 10.277 7.7278

Ni-6 26.721 17.6859

Ni-8 32.50 30.5075

Ni-15 No saturation

6.3 Cobalt Doped TiO2 Thin Films

6.3.1 Experimental

Thin films of Ti1-xCoxO2 (x=0.02-0.15) were prepared by Aerosol Assisted

Chemical Vapor Deposition (AACVD). Co doped TiO2 films were deposited by AACVD

on silicon substrates. The precursor for Co doped TiO2 films was synthesized by using

Titanium Isopropoxide Ti(OCH(CH)3)2)4 and Co(acac)2 as the starting materials. We

have used this precursor by varying the Co concentration 2,4,6,8 and 15 molar ratio. In

detailed synthesis process 1.00 g (4.36mmol) of Ti Isopopoxide and 1.02g (4.36mmol)

Co(acac)2 was dissolved in 15 ml of toluene separately. Then stoichiometric amount of

Co(acac)2 solution added to Ti Isopropoxide solution to obtain 2,4,6,8 and 15 wt% Co.

The precursor thus form is clear solution suggesting that both the starting materials are

mixed uniformly. AACVD was carried on by using this precursor.

Substrates were washed with acetone, isoproponol and ethanol ultrasonically. The

Chapter 6

102

substrate kept in ethanol before use. AACVD apparatus used as shown in the chapter 2 as

discussed above. All the films are grown on 650°C and the Argon flow rate was kept on

120 mL/min with the help of a micro controller. Films were deposited for 20 min each.

After deposition hot plate turned off until the room temperature achieved while Argon

flow was kept on. Scotch tap test was conducted to check the adhesion of the films.

6.3.2 Results and Discussion

6.3.2.1 XRD Analysis

The XRD results show the existence of different phases including the non

stoichiometric Titanium oxide magneli phases, which follow the integer formula TinO2n-1

(n<4<10) (putt formula here). At 2wt.% Co concentration, peaks for Ti6O, TiO, Co3Ti3O,

(TiO1.20)3.12, and TiO2 rutile phases are shown at their respected angles (however it is

very difficult to identify them exactly from XRD). For the films with 4wt.% Co

concentration Ti3O5, TiO2 rutile, Ti5O9 and TiO1.95 phases are pronounced with a sharp

peak for rutile TiO2 phase. Further increase in Co concentration up to 6wt.% have the

same peaks but over all crystallinity is lost. For 8wt.% Co concentration the magneli

phases disappeared. Only the Rutile TiO2 phase was obtained with further decrease in the

crystallinity. Sample with 15wt.% Co concentration a highly crystalline rutile phase is

obtained.

Many reaction mechanism studies147-150

carried out on Ti(OCH(CH)3)2)4. Authors

proposed reaction mechanisms at various temperatures. Fictorie et al.149

employed

temperature programmed reaction spectroscopy and molecular beam spectroscopy to

investigate reaction kinetics of Ti(OCH(CH)3)2)4 decomposition in inert carrier gas. They

observed TiO2, acetone CH3COCH3, propane C3H6 and hydrogen (H2) decomposition

products of Ti(OCH(CH)3)2)4 at temperature higher that 550 C as shown in equation

Ti(OCH(CH)3)2)4 (g) TiO2 (s) + CH3COCH3 (g) + H2(g) + C3H6(g) (6.1)

Chapter 6

103

Since Co doped titanium oxide film are deposited at 650°C (well above 550°C) in

inert environment, the above equation fits well in describing the reaction mechanism.

Further, reaction products H2 and carbon from the CH3COCH3 and C3H6 can reduce T+4

to Ti+3

and even into Ti+2

and Ti+1

states. Liu et. al.151

proposed three types of H2 and

TiO2 interactions with increasing temperature in their report. They proposed electrons are

transferred from H2 to oxygen present in TiO2 lattice; reacts with H2 to form H2O, leaving

behind an oxygen vacancy. When temperature is above 560 C, electrons present in the

oxygen vacancies are transferred to Ti+4

to form Ti+3

. Above arguments successfully

discuss the reason behind the formation of unstable magneli phases from 2-6 wt.% Co

doped TiO2 films in current studies.

Besides above stated chemical reaction, another reaction may be proposed in

which H2 produced during the decomposition of Titanium isopropoxide may react with

oxygen present in the Co(acac)2. With the increase in Co(acac)2 amount to get 8wt.% and

15wt.% Co doped samples, this reaction becomes more prominent. This might be the

reason for stable rutile phases at higher doping concentration.

From XRD data it is concluded that 2-8wt.% Co concentration has stabilized the

unstable oxygen deficient magneli phases. However there was a decrease in the

crystallinity from 2-8wt.% Co concentration which is attributed towards the formation of

new phases. With further increase in Co concentration up to 15wt.% pure rutile phase is

confirmed with very high crystallinity. In all the samples no Co or Co oxide peaks were

found. A slight shift of rutile peak is observed at 2θ 20 . It is known that Co has the

larger atomic radius (0.72Å) than Ti atomic radius (0.68 Å). When an atom with larger

atomic radius replaces a small atom in the lattice, it is observed that peak is shifter

towards the lower 2 value. Therefore a slight shift on rutile peak at 20 suggests that Co

is incorporated within the TiO2 matrix.

Chapter 6

104

Fig 6. 13: XRD pattern of Ni doped TiO2 thin films with a) 2wt.% Ni b) 4wt.% Ni c)

6wt.% Ni d) 8wt.% Ni and e) 15wt.% Ni doping

Chapter 6

105

6.3.2.2 Rutherford Back Scattering

Rutherford backscatering spectrometry (RBS) mearurements of epilayer Co doped

TiO2 samples are presented in Fig 6.8 and table 6.3. All the measurements were taken

with the same method as discussed above for the Ni doped TiO2.

Fig 6.8 shows RBS for Co doped TiO2 thin films.

Fig 6. 14: RBS spectra of Co doped TiO2 thin films with various concentrations

The peaks at high energy correspond to the scattering from the Ti and Co. There is no

evedience for the formation of silicate with Ti or Co . The Co concentration calculated

by RBS analysis is slightly low than the experimental value. This may be due to low

solubility of Co(acac)2 in taluene. Films thicknesses were found in the range of 150-

250nm as measured from RBS spectra and shown in the table 6.3

Chapter 6

106

Table 6. 4: Co concentration and film thickness as calculated from RBS spectra

Sample Doping concentration(wt.%)

Stiociometric Calcutated

Films thickness

(nm)

Co-2 2 1.85 166

Co-4 4 3.77 219

Co-6 6 5.4 204

Co-8 8 7.1 246

Co-15 15 13.22 160

. Both spectra indicate that the Co doped TiO2 films are non-stoichiometric. This

result is in agreement to the result found from XRD analysis. Arguments for non-

stoichiomery in the Co doped TiO2 films are established in section 6.3.2.1.

6.3.2.3 Scanning electron microscopy

SEM images of Co doped TiO2 thin films deposited by AACVD (Fig 6.9) shown

compact and smooth film morphologies with homogenously dispersed particles.

Individual grains are well defined and clear grain boundaries can be seen. The packing

density of the microstructure and the grain sizes apparently seem to be affected by

variation of the Co concentration. SEM images further reveal that all the films are

nanoporous structure with a very uniform grain size distribution. The average grain size

lies in the range of (40-60nm). Shape of the grains changed with increase in Co

concentration as shown in Fig. 6.9. The change is the structure of the grains is attributed

to the growth stress of the films and radius and the concentration of the Co ions.

In the case of sample with 2wt.% Co a compact dense morphology was obtained.

The grains are of spherical in shape. Average grain size is 25-50nm and the size

Chapter 6

107

distribution throughout the film is very uniform. Pore size in the film is found only a few

nanometers. At 4 wt.% Co concentration shape of the grains is changes into elongated

pallet like with a length between 50-100nm and width is only a few nanometer. Gains are

uniformly distributed with uniform grain size distribution. However there are some

flakes or the larger aglaromated grains can be seen with a small concentration. The SEM

image of the films with 6wt.% Co concentration shows that the growth of particles has

retained their shape but the grain size terned into smaller range . A decrease in the

porosity can also be seen clearly. Another change in the shape of the grains is observed in

the films with 8wt.% Co concentration. However some large size grains can also be seen

with average grain size in the range of 100-200nm. In 15wt.% Co concentration the shape

of the grains is changed into beed like structures.

The change in the grain shape in can be explained by analyzing XRD data. From

XRD data it can be seen that there is considerable variation in the maximum intense peak,

indicating preffered growth direction are different for each phase formed with increase in

doping concentration. Also decomposition reaction kienetics at the substrate surface

define the grain shape and size.

Chapter 6

108

Fig 6. 15: SEM images of Co doped TiO2 thin films with a) 2wt.% Co b) 4wt.% Co c)

6wt.% Co d) 8wt.% Co and e) 15wt.% Co doping

a) b)

c) d)

e)

Chapter 6

109

6.3.2.4 Magnetic properties

Room temperature magnetic hysteresis loops of Ti1-xCoxO2 for x=0.02, 0.04, 0.06,

0.08 and x=0.15 are shown in Fig 6.10

Fig 6. 16: Hysteresis loop of Co doped TiO2 thin films with various Co doping

concentrations

Chapter 6

110

All the samples exhibit ferromagnetism (300K) as shown in the Fig. 6.10

Magnetisation has increased with increasing Co concentration. But at 8wt.% Co

concentration anomalous behavior is observed. There can be number of possibilities

which can cause RTFM in Co doped Oxygen deficient TiO2 structures as follows:

1. Co clusters in metallic form and secondary phases

2. Ti+3

and/or Ti+2

interstitial defects

3. Ti+3

doped TiO2 (self doping)

4. Ti+3

and/or Ti+4

replaced by Co+2

5. Oxygen vacancies

No clustering from metallic Co and other ferromagnetic phases observed in XRD

analysis. Further magnetic saturation in samples with 2-8wt.% Co doping concentration

occurs well below than that for cobalt metal films (H~1.5-2 Tesla).18

Therefore

formation of BPMs and their overlap is one of the possible source of RTFM in the films.

Further magneli phases with lower stoichiometry (TiO1.75, TiO1.80 and TiO1.83 etc.)

are antiferromagnetic with Neel temperature ~130K and paramagnetic at higher

stoichiometry(Ti3O7 and Ti4O9 etc.)152

. Therefore contribution in RTFM from these

magneli phases can be ruled out at this level.

As discussed in XRD section, presence of T+3

and Ti+2

cannot be ruled out. Ti+3

and Ti+2

may itself source of ferromagnetism due to their unfilled d-shell with electronic

configuration as 3d1 and 3d

2, however it these ions are isolated then there may only be a

paramagnetic effect.153

Recently Hua et al.138

reported RTFM in Ti+3

doped TiO2 nanowires prepared by

solvothermal method, The Ms value was found to be ~23.6 memu/g. However in the

presence of dopant, it is very difficult to distinguish whether RTFM is from Ti or Co

doping.

Chapter 6

111

RTFM in current study can be explained with the help of BPM model for both

cases i.e. Co+2 and Ti+3 doping. In this case Co+2

can replace Ti+4

as well as Ti+3

. In both

cases it can introduce positively charged vacancy. However, Co+2

most likely replace Ti+3

due to comparable oxidation state as compared to Ti+4

. These vacancies can capture an

electron in quasi hydrogenic orbit surrounded by Co+2

ions and construct a BMP. Co+2

ions may couple via donor electron ferromagnetically. As doping concentration increases,

BMPs concentration increases throughout the samples. The overlap of BMPs establishes

long range ferromagnetic ordering. Co-Co, Ti-Ti and Ti-Co may interact via BPMs.

Thus net magnetization may be the combined result of ferromagnetic interaction of

different types of BMPs.

At 8wt.% Co doping concentration ferromagnetic signal coupled with a large

paramagnetic signal is observed. Since unstable oxygen deficient magneli phases152

are in

transition to stable rutile phase from 2-8wt.% Co doped samples and overall crystallinity

of sample is reduced in the process as can be seen in XRD patterns. It is therefore quite

possible that paramagnetic magneli phases are still present in amorphous form. An

increase in the moment observed in 15% Co doped film. XRD results shown that the

there is a pure rutile phase at this doping concentration. Magnetic moment seems to

increase with crystallinity and also in the stable rutile phase.

Table 6. 5: Magnetic moment of Co doped TiO2 at 300K

Sample Moment (emu/cc)

300K

Co-2 4.90

Co-4 6.29

Co-6 7.11

Co-8 No Saturation

Co-15 10.54

Chapter 6

112

Summary

In summary, Ni and Co doped TiO2 films were synthesized by AACVD under

oxygen deficient Argon environment. Working temperature was 450 C and 650 C for Ni

and Co doped films respectively. Analyses of the XRD suggests that anatase phase in the

dominant phase up to 8% Ni concentration, however at 15% Ni doping a change in the

phase observed. Due to the fact that Ni reacted with the TiO2 matrix NiTiO3 phase is

formed. With Co doping magneli phases are observed along with the rutile phase. At 15%

pure rutile phase with high crystallinity was observed. SEM has shown uniform

distribution on the grain sizes in both the cases. All samples are ferromagnetic at room

temperature i.e. 300K..

Chapter 7

113

Chapter 7

Conclusions and Future Work

This chapter is based on some important conclusions drawn from the thesis. The

thesis was comprised of two studies, both having the great impact in the field of

spintronics. First part was about the studies of Exchange Bias and its related

phenomenon. The second half was about the synthesis of Nickel and Cobalt Doped TiO2

by Aerosol Assisted Chemical Vapour Deposition (AACVD) keeping in view the

importance of finding new synthesis routes, as properties may vary significantly in such a

systems.

In section 7.1 conclusions drawn on the basis of experiments on Exchange Bias

phenomenon are summarized. Section 7.2 is devoted for the come out of research on

synthesis and structural and magnetic properties of Ni and Co doped TiO2 thin films.

7.1 Exchange Bias

In chapter 4 the effects of sample shape and fabrication process on the reversal

mechanism in Exchange Bias multilayer thin films were studied.

To study the effects, all samples studied in this work were produced by sputtering

using a HiTUS sputtering system. The average process pressure was 2.75×10−3mbar and

the RF power was held at 1.5kW. During fabrication of the samples a field of 300Oe was

applied to give an easy axis to the exchange coupled films. Five samples were grown

with variable bias voltage (200,400,600,800 and 1000V) whilst keeping the composition.

However sample grown on 800V was used to carry out further studies being the best

sample in the series. To study the effect of fabrication technique Samples were prepared

by three methods:

1. Samples were cut from a continuous film using a diamond scribe and then

cracked.

Chapter 7

114

2. The sample was sputtered through a thin stainless steel mask

3. And a sample was cut with ultrasonic cutter.

The effect of nucleation was described arising from both sample shape effects and the

process used to cut the sample. It was found that cutting techniques such as the use of

ultrasonic cutters leads to a large increase in nucleation which distorts the hysteresis loop.

Deposition through masks causes shadowing effects at the edges that also distort the

loops. Cutting with a diamond scribe appears to give the best outcome. Finally it was

concluded that

The sample edge roughness leads to a distribution of nucleation fields and hence

changes the shape of the hysteresis loop which has consequences for MRAM or

spintronics devices.

The best way to cut samples of nucleation controlled materials is by cracking.

The effect of using an ultrasonic cutter is catastropic with many chips along the

film leading to areas with a strong demagnetising field that will readily nucleate

multiple domain nucleation.

Shadow masks cause thinning at the edges which also affects nucleation.

The overall shape of a sample is less critical than edge roughness.

The geometric response in coils or other sensors in magnetometers has to be taken

into account and a calibration sample of similar size, moment and shape has to be

used.

A different coercivity is observed in the recoil loop for the sample deposited

through a mask.

Our understanding of the coercivity of exchange bias systems is not clear enough.

In chapter 5 The effects of inserting mono-atomic layers of Mn at the interface of

IrMn/CoFe bi-layers were investigated. The insertion of Mn alters spin structure at

AFM/FM interface which may bring an enhancement in TB. Samples were grown using

HiTUS sputtering technology at 200, 400, 600, 800 and 1000V bias voltages.

Chapter 7

115

The effects of placing a 2Å Mn dusting layer at the AFM/FM interface on the

exchange bias, coercivity and blocking temperature of IrMn/CoFe bilayers were

investigated. of the bilayer was mildly enhanced from (-411±42.4) Oe to (-

479.7±44.3) Oe, while was increased from (222.3±23.9) Oe to (256.5±37.4) Oe. It is

believed that the enhancement of and is due to the modification of the AFM spin

structure at the interface.The blocking temperature was reduced from 408.3 K to 380.0 K

with the addition of Mn; it is believed that grains that were previously too small to be

thermally stable were somehow set. This could be due to either a chemical reaction with

the Mn at the interface or due to exchange effects.

In chapter 5 effect of grain size on the training effect was also investigated. It is

an important parameter due to the fact that it is the measure of the stability of exchange

bias bilayer which is inside many devices.

The training effect is strongly reinforced by the reduced grain size of the AFM

layer.

It was confirmed that nanocrystalline structure have large ratio of spins located

at grain boundaries. These spins are disordered which bring out variation in the

magnetic anisotropy. This alternately changes the strength of the exchange bias

thus promoting the training effect.

Larger AFM grain-volumes give rise to thermally stable bias fields and

consequently smaller training effects.

7.2 Diluted magnetic semiconductors

Thin films of Til-XNiXO2 (x=0.02-0.15) were prepared by Aerosol Assisted

Chemical Vapor Deposition (AACVD) on Si substrates in Argon atmosphere on 450 C.

The precursor [Ni2Ti2(OEt)2(l-OEt)6(acac)4] was used for Ni doped TiO2 films. A

modified design of AACVD was used for deposition. XRD pattern shows that up to 8%

Ni doping and an overall increase in crystalline with increasing Ni contents, plus a

greater peak intensity in the (101) reflection. However No secondary phase from Ni or

Chapter 7

116

NiO obtained suggesting that the Ni contents are doped within the TiO2 matrix. At 15%

Ni concentration NiTiO3 phase appeared. This change in phase was due to the fact that Ni

would have increased the number of oxygen vacancies which might be responsible for

the transformation of anatase to rutile. RBS data showed the Ni and Ti composition was

close to the stoichiometry. The Depth profile showed that composition is very uniform

throughout the film thickness. Thickness for 2% to 8% Ni concentration is between 200-

260nm however for 15% Ni concentration it was more than 800nm. SEM images show

uniform distribution on the grains with clear grain boundaries. Average grain size was

measured between 40-60nm, however, for 15% Ni doped samples grains or the grain

boundaries are not clearly visible. Magnetic measurements taken from SQUID show that

all the samples exhibit ferromagnetism at room temperature (RTFM). It can be concluded

that this ferromagnetism is intrinsic as no Ni peaks can be seen within the XRD limit.

Further no oxide of nickel or bimetallic oxides of Ni and Ti are ferromagnetic.

RTFM is attributed to the oxygen vacancies. Oxygen vacancies were induced by the

oxygen deficient environment and due to the incorporation of Ni doping. Bound

Magnetic Polaron model is successful model to describe RTFM in Ni doped TiO2 films.

This can be further confirmed by taking SQUID measurements at low temperature

(100K). At low temperature BMPs formed due to the presence of oxygen vacancies grow

resulting in enhancement of RTFM.

Co doped TiO2 films were deposited with the same scheme as for Ni doped TiO2

films, however, on a higher temperature of 700 C. At higher temperature in the oxygen

deficient environment so called magneli phases of TiO2 appeared. However From 2-8%

Co concentration has stabilized the unstable oxygen deficient magneli phases. This may

be due to the inert atmosphere and high temperature during the synthesis of the films. A

slight shift of rutile peak is observed at 2θ 20 due to the fact that Co has the larger

atomic radius (0.72Å) than Ti atomic radius (0.68 Å).This shift is evident that Co ions are

incorporated within the TiO2 lattice. Grain size was uniform throughout the series as seen

by SEM. There was a slight decrease in concentration of Co than the stoichiometric

amount, which is due to less solubility of Co(acac)2 in toluene. All the samples exhibit

Chapter 7

117

RTFM. This RTFM is a combination of Co doping and the oxygen deficient magneli

phases.

7.3 Future Work

For further investigation few suggestions drawn from the research out of the thesis are:

1. To further investigate the effects of the interfacial layers it would be highly

beneficial to investigate the effect of varying the dopant layer thickness.

2. In the case of Mn it is vital to attempt to characterize the phenomena of the

decrease in with higher setting fields.

3. In case of Mn doping, the elimination of experimental error as a source of this

effect would be a result of further measurements.

4. Theoretic understanding of coercivity mechanism in exchange bias films, it is

therefore an area to investigate.

5. York protocol should be implied to investigate the theoretical understanding of

Training Effect. It would produce high impact on spintronics industry if the entire

exchange bias related phenomenon be related with the grain size.

6. Synthesis through AACVD provides a good opportunity to control number of

parameter like reaction environment, temperature, selection of precursor.

Presence of magneli phases alone can induce RTFM. It is worth to study TiO2

films at various temperatures without doping.

7. Number of transition metals and non-metallic doping cases are not studied with

AACVD. It is in the scientific to compare various transition metal doped DMS

properties synthesized with various routes. AACVD can be one of them

Chapter 8

118

Chapter 8

References 1. Xia Y (2004) The Chemistry of Nanomaterials. Edited by C. N. R. Rao, A.

Müller, A. K. Cheetham. Chem Phys Chem 5, 1913-1914.

2. Žutić I, Fabian J, Das Sarma S (2004) Spintronics: Fundamentals and

applications. Rev Mod Phys 76, 323-410.

3. Chambers SA (2002) A potential role in spintronics. Mat Today 5, 34-39.

4. Wolf SA, Awschalom DD, Buhrman RA, Daughton JM, von Molnár S, Roukes

ML, et al. (2001) Spintronics: A Spin-Based Electronics Vision for the Future.

Science 294, 1488-1495.

5. Gardelis S, Smith CG, Barnes CHW, Linfield EH, Ritchie DA (1999) Spin-valve

effects in a semiconductor field-effect transistor: A spintronic device. Phys Rev B

60, 7764-7767.

6. Mansuripur M, Sincerbox G (1997) Principles and techniques of optical data

storage. Proceedings of the IEEE 85, 1780-1796.

7. Ohno H, Matsukura F, Ohno Y. Semiconductor spin electronics. In: J

International (ed). Cutting Edge 1, 2002.

8. Akinaga H, Ohno H (2002) Semiconductor spintronics. Nanotechnology, IEEE

Trans 1, 19-31.

9. Ohno H (1998) Making Nonmagnetic Semiconductors Ferromagnetic. Science

281, 951-956.

10. Mauger A, Godart C (1986) The magnetic, optical, and transport properties of

representatives of a class of magnetic semiconductors: The europium

chalcogenides. Phys Rep 141, 51-176.

11. Rebecca J, et al. (2005) Transition metal-doped TiO2 and ZnO—present status of

the field. J Phys: Cond Matt 17, R657.

12. Fukumura T, Toyosaki H, Yamada Y (2005) Magnetic oxide semiconductors.

Semicond Sci and Tech 20, S103.

13. Fukumura T, Toyosaki H, Yamada Y (2006) Magnetic Oxide Semiconductors.

ChemInform 37, no-no.

Chapter 8

119

14. Thompson SM (2008) The discovery, development and future of GMR: The

Nobel Prize 2007. J Phys D: Appl Phys 41, 093001.

15. O‘Grady K, Fernandez-Outon LE, Vallejo-Fernandez G (2010) A new paradigm

for exchange bias in polycrystalline thin films. JMag Mag Mat 322, 883-899.

16. Hou X, Choy KL (2006) Processing and Applications of Aerosol-

Assisted Chemical Vapor Deposition. Chem Vap Dep 12, 583-596.

17. Palgrave RG, Parkin IP (2006) Aerosol Assisted Chemical Vapor Deposition

Using Nanoparticle Precursors: A Route to Nanocomposite Thin Films. J Amer

Chem Soc 128, 1587-1597.

18. Wang HB, Meng GY, Peng DK (2000) Aerosol and plasma assisted chemical

vapor deposition process for multi-component oxide La0.8Sr0.2MnO3 thin film.

Thin Solid Films 368, 275-278.

19. Higgins JS, Shinde SR, Ogale SB, Venkatesan T, Greene RL (2004) Hall effect in

cobalt-doped TiO_{2-δ}. Phys Rev B 69, 073201.

20. Kim DH, Yang JS, Lee KW, Bu SD, Noh TW, Oh SJ, et al. (2002) Formation of

Co nanoclusters in epitaxial Ti0.96Co0.04O2 thin films and their ferromagnetism.

Appl Phys Lett 81, 2421-2423.

21. Stampe PA, Kennedy RJ, Xin Y, Parker JS (2002) Investigation of the cobalt

distribution in TiO2:Co thin films. J Appl Phys 92, 7114-7121.

22. Yamada Y, Toyosaki H, Tsukazaki A, Fukumura T, Tamura K, Segawa Y, et al.

(2004) Epitaxial growth and physical properties of a room temperature

ferromagnetic semiconductor: Anatase phase Ti1 - xCo xO2. J Appl Phys 96, 5097-

5102.

23. Yang HS, Choi J, Craciun V, Singh RK (2003) Ferromagnetism of anatase Ti1 -

xCo xO2- films grown by ultraviolet-assisted pulsed laser deposition. J Appl Phys

93, 7873-7875.

24. Sato K, Dederics PH, Katayama-Yoshida H (2003) Curie temperatures of III–V

diluted magnetic semiconductors calculated from first principles. EPL Eur phys

Lett 61, 403.

25. Blundell S (2001) Magnetism in Condensed Matter. Oxford University Press,

Oxford, UK.

Chapter 8

120

26. Langevin P (1905) Magnétisme et theorie des électrons. Annales de Chem et Phys

8, 70-127.

27. Cullity BD, Graham CD. Introduction to magnetic materials. Hoboken, NJ: John

Wiley & Sons : IEEE Press, 2009.

28. Becker R (1932) Elastic tensions and magnetic characteristics. ZEITSCHRIFT

FÜR PHYSIK 33, 905-913.

29. Kersten M (1943) Grundlagen einer Theorie der Ferromagnetischen Hysterese

und derKoerzitivkraft. Hirzel, Leipzig.

30. Meiklejohn W, Bean C (1957) New Magnetic Anisotropy. Phys Rev 105, 904-

913.

31. Getzlaff M {Fundamentals of Magnetism - Mathias Getzlaff - Google Books}.

32. Stoner EC, Wohlfarth EP (1948) A Mechanism of Magnetic Hysteresis in

Heterogeneous Alloys. Phil Trans Roy Soc Lond A 240, 599-642.

33. Luborsky FE (1961) Development of Elongated Particle Magnets. J Appl Phys

32, S171-S183.

34. Jacobs IS, Bean CP (1955) An Approach to Elongated Fine-Particle Magnets.

Phys Rev 100, 1060-1067.

35. Frei E, Shtrikman S, Treves D (1957) Critical Size and Nucleation Field of Ideal

Ferromagnetic Particles. Phys Rev 106, 446-455.

36. Andrä W, Danan H, Röpke U (1984) Models for Perpendicular Hysteresis of Thin

Magnetic Films. IEEE Trans Magn MAG-20, 102-104.

37. Wuori ER, J. H. Judy (1984) DIoaS-W, Particle Array ITM, vol. MAG-20, pp.

1867-1869. (1984) Demagnetizing Influence on a Stoner-Wohlfarth Particle

Array. IEEE Trans Magn MAG-20, 1867-1869.

38. Wielinga T, Lodder JC, Worst J (1982) Characteristics of RF-Sputtered Co-Cr

Films. IEEE Trans Magn MAG-18, 1107-1109.

39. Andrä W, Danan H (1987) Inhomogeneous Cr Concentration as Origin of the

Perpendicular Coercivity in Co-Cr Films. IEEE Trans Magn MAG-23(1), 62-64.

40. Gaunt P, Mylvaganam CK (1977) Nucleation and pinning at 360 domain walls in

SmCo5 and related alloys. JAppl Phys 48, 2587-2590.

Chapter 8

121

41. Donnet DM, Lewis VG, Chapman JN, Ogrady K, Vankesteren HW (1993)

Microstructure and Hysteresis in Co/Pt Multilayers. J Phys D-Appl Phys 26,

1741-1745.

42. Hono K. Tsukuba, Japan.: NIMS, March, 2010.

43. Goll D, Macke S, Kronmuller H (2008) Exchange coupled composite layers for

magnetic recording. Physica B: Cond Matt 403, 338-341.

44. Dobin AY, Richter HJ (2007) Domain wall assisted magnetic recording (invited).

J Appl Phys 101, 09K108.

45. Girt E, Dobin AY, Valcu B, Richter HJ, Wu X, Nolan TP (2007) Experimental

Evidence of Domain Wall Assisted Switching in Composite Media. IEEE Trans

Magn 43, 2166-2168

46. Goodman AM, O'Grady K, Laidler H, Owen NW, Portier X, Petford-Long AK, et

al. (2001) Magnetization reversal processes in exchange-biased spin-valve

structures. IEEE Trans Magn 37, 565-570.

47. Craig BR, McVitie S, Chapman JN, O‘Donnell DO, Johnston AB (2007)

Micromagnetic reversal behavior of multiscale permalloy elements. J Appl Phys

102, 013911-013916.

48. van der Zaag PJ, Wolf RM, Ball AR, Bordel C, Feiner LF, Jungblut R (1995) A

study of the magnitude of exchange biasing in [111] Fe3O4/CoO bilayers. JMag

Mag Mat 148, 346-348.

49. Nogues J, Lederman D, Moran TJ, Schuller IK, Rao KV (1996) Large exchange

bias and its connection to interface structure in FeF2-Fe bilayers. Appl Phys Lett

68, 3186-3188.

50. Jiang JS, Felcher GP, Inomata A, Goyette R, Nelson C, Bader SD (2000)

Exchange-bias effect in Fe/Cr(211) double superlattice structures. Phys Rev B 61,

9653-9656.

51. Neél L (1967) Ann Phys (Paris) 2, 61.

52. Fulcomer E, Charap SH (1972) Temperature and frequency dependence of

exchange anisotropy effects in oxidized NiFe films. J App Phys 43, 4184-4191.

53. Grimsditch M, Hoffmann A, Vavassori P, Shi H, Lederman D (2003) Exchange-

Induced Anisotropies at Ferromagnetic-Antiferromagnetic Interfaces above and

below the Néel Temperature. Phys Rev Lett 90, 257201.

Chapter 8

122

54. Nishioka K, Hou C, Fujiwara H, Metzger RD (1996) Grain size effect on ferro-

antiferromagnetic coupling of NiFe/FeMn systems. J Appl Phy 80, 4528-4533.

55. Nishioka K, Shigematsu S, Imagawa T, Narishige S (1998) Thickness effect on

ferro/antiferromagnetic coupling of Co/CrMnPt systems. J Appl Phy 83, 3233-

3238.

56. Xi H (2005) Theoretical study of the blocking temperature in polycrystalline

exchange biased bilayers. JMag Mag Mat 288, 66-73.

57. Berkowitz AE, Takano K (1999) Exchange anisotropy-A Review. JMag Mag Mat

200, 552-570.

58. Vopsaroiu M, Vallejo-Fernandez G, Thwaites MJ, Anguita J, Grundy PJ,

O‘Grady K (2005) Deposition of polycrystalline thin films with controlled grain

size. J Appl Phys D 38, 490-496.

59. O‘Grady K, Holloway L, Antel WJ (2002) Measurement of the Energy Barrier

Distribution in the Antiferromagnetic Layer of Exchange-Biased Materials. IEEE

Trans Magn 38, 2741-2743.

60. Paccard D, Schlenker C, Massenet O, Montmory R, Yelon A (1966) A New

Property of Ferromagnetic-Antiferromagnetic Coupling. phys status solidi (b) 16,

301-311.

61. Nogués J, Schuller IK (1999) Exchange bias. JMag Mag Mat 192, 203-232.

62. Xi H, White RM, Mao S, Gao Z, Yang Z, Murdock E (2001) Low-frequency

dynamic hysteresis in exchange-coupled Ni81Fe19/Ir22Mn78 bilayers. Phys Rev B

64, 184416.

63. Hoffmann A (2004) Symmetry Driven Irreversibilities at Ferromagnetic-

Antiferromagnetic Interfaces. Phys Rev Lett 93, 097203.

64. Fitzsimmons MR, Yashar P, Leighton C, Schuller IK, Nogués J, Majkrzak CF, et

al. (2000) Asymmetric Magnetization Reversal in Exchange-Biased Hysteresis

Loops. Phys Rev Lett 84, 3986-3989.

65. Fernandez-Outon LE, Vallejo-Fernandez G, Manzoor S, O‘Grady K (2006)

Thermal instabilities in exchange biased materials. JMag Mag Mat 303, 296-301.

66. Matsumoto Y, Murakami M, Shono T, Hasegawa T, Fukumura T, Kawasaki M, et

al. (2001) Room-Temperature Ferromagnetism in Transparent Transition Metal-

Doped Titanium Dioxide. Science 291, 854-856.

Chapter 8

123

67. Higgins JS, Shinde SR, Ogale SB, Venkatesan T, Greene RL (2004) Hall effect in

cobalt-doped TiO2- delta. Phys Rev B 69, 073201.

68. Chambers SA, Droubay T, Wang CM, Lea AS, Farrow RFC, Folks L, et al.

(2003) Clusters and magnetism in epitaxial Co-doped TiO2 anatase. Appl Phys

Lett 82, 1257-1259.

69. Chambers SA, Thevuthasan S, Farrow RFC, Marks RF, Thiele JU, Folks L, et al.

(2001) Epitaxial growth and properties of ferromagnetic co-doped TiO2anatase.

Appl Phys Lett 79, 3467-3469.

70. Chambers SA, Wang CM, Thevuthasan S, Droubay T, McCready DE, Lea AS, et

al. (2002) Epitaxial growth and properties of MBE-grown ferromagnetic Co-

doped TiO2 anatase films on SrTiO3(001) and LaAlO3(001). Thin Solid Films

418, 197-210.

71. Park WK, Ortega-Hertogs RJ, Moodera JS, Punnoose A, Seehra MS (2002)

Semiconducting and ferromagnetic behavior of sputtered Co-doped TiO2 thin

films above room temperature. J Appl Phys 91, 8093-8095.

72. Seong N-J, Yoon S-G, Cho C-R (2002) Effects of Co-doping level on the

microstructural and ferromagnetic properties of liquid-delivery metalorganic-

chemical-vapor-deposited Ti1 - xCo xO2 thin films. Appl Phys Lett 81, 4209-4211.

73. Kim JY, Park JH, Park BG, Noh HJ, Oh SJ, Yang JS, et al. (2003)

Ferromagnetism Induced by Clustered Co in Co-Doped Anatase TiO2 Thin Films.

Phys Rev Lett 90, 017401.

74. Murakami M, Matsumoto Y, Hasegawa T, Ahmet P, Nakajima K, Chikyow T, et

al. (2004) Cobalt valence states and origins of ferromagnetism in Co doped TiO2

rutile thin films. J Appl Phys 95, 5330-5333.

75. Toyosaki H, Fukumura T, Yamada Y, Nakajima K, Chikyow T, Hasegawa T, et

al. (2004) Anomalous Hall effect governed by electron doping in a room-

temperature transparent ferromagnetic semiconductor. Nat Mater 3, 221-224.

76. Soo YL, Kioseoglou G, Kim S, Kao YH, Devi PS, Parise J, et al. (2002) Local

environment surrounding magnetic impurity atoms in a structural phase transition

of Co-doped TiO2 nanocrystal ferromagnetic semiconductors. Appl Phys Lett 81,

655-657.

77. Kacman P (2001) Spin interactions in diluted magnetic semiconductors and

magnetic semiconductor structures. Semicond Sci Tech 16, R25.

Chapter 8

124

78. Larson BE, Hass KC, Ehrenreich H, Carlsson AE (1988) Theory of exchange

interactions and chemical trends in diluted magnetic semiconductors. Phys Rev B

37, 4137.

79. Blinowski J, Kacman P (1996) Double exchange in mixed-valency diluted

magnetic semiconductors. Acta Phys Polon A 90.

80. Zener C (1951) Interaction between the d-Shells in the Transition Metals. II.

Ferromagnetic Compounds of Manganese with Perovskite Structure. Phys Rev

82, 403.

81. Medvedkin GA, Ishibashi T, Nishi T, Hayata K, Hasegawa Y, Sato K (2000)

Room temperature ferromagnetism in novel diluted magnetic semiconductor Cd1-

xMnxGeP2. Japan J Appl Phys 39, L949-L951.

82. Dietl T, Cibert J, Ferrand D, Merle d'Aubigné Y (1999) Carrier-mediated

ferromagnetic interactions in structures of magnetic semiconductors. Mat Sci

Engg B 63, 103-110.

83. Fukuma Y, Asada H, Arifuku M, Koyanagi T (2002) Carrier-enhanced

ferromagnetism in Ge1 - xMnxTe. Appl Phys Lett 80, 1013-1015.

84. Lazarczyk P, Story T, Arciszewska M, GaLazka RR (1997) Magnetic phase

diagram of Pb1-x-ySnyMnxTe semimagnetic semiconductors. J JMag Mag Mat

169, 151-158.

85. Dietl T, Ohno H, Matsukura F (2001) Hole-mediated ferromagnetism in

tetrahedrally coordinated semiconductors. Phys Rev B 63, 195205.

86. Dietl T, Ohno H, Matsukura F, Cibert J, Ferrand D (2000) Zener Model

Description of Ferromagnetism in Zinc-Blende Magnetic Semiconductors.

Science 287, 1019-1022.

87. Takamura K, Matsukura F, Chiba D, Ohno H (2002) Magnetic properties of

(Al,Ga,Mn)As. Appl Phys Lett 81, 2590-2592.

88. Jungwirth T, Mascaronek J, Wang KY, Edmonds KW, Sawicki M, Polini M, et al.

(2006) Low-temperature magnetization of (Ga,Mn)As semiconductors. Phys Rev

B 73, 165205.

89. Kaminski A, Das Sarma S (2002) Polaron Percolation in Diluted Magnetic

Semiconductors. Phys Rev Lett 88, 247202.

90. Wolff PA, Bhatt RN, Durst AC (1996) Polaron2010;polaron interactions in

diluted magnetic semiconductors. J Appl Phys 79, 5196-5198.

Chapter 8

125

91. Kaminski A, Das Sarma S (2003) Magnetic and transport percolation in diluted

magnetic semiconductors. Phys Rev B 68, 235210.

92. Sort J, Langlais V, Doppiu S, Dieny B, Suriñach S, Muñoz JS, et al. (2004)

Exchange bias effects in Fe nanoparticles embedded in an antiferromagnetic

Cr2O3 matrix. Nanotechnology 15, S211-S214.

93. Wee L, Stamps R, Malkinski L, Celinski Z, Skrzypek D (2004) Thermal training

of exchange bias in epitaxial Fe/KNiF3. Phys Rev B 69.

94. Vopsaroiu M, Thwaites MJ, Rand S, Grundy PJ, O'Grady K (2004) Novel

sputtering technology for grain-size control. Magnetics, IEEE Trans 40, 2443-

2445.

95. Gupta S, Fenwick WE, Melton A, Zaidi T, Yu H, Rengarajan V, et al. (2008)

MOVPE growth of transition-metal-doped GaN and ZnO for spintronic

applications. J Crystal Growth 310, 5032-5038.

96. Park JH, Kim MG, Jang HM, Ryu S, Kim YM (2004) Co-metal clustering as the

origin of ferromagnetism in Co-doped ZnO thin films. Appl Phys Lett 84, 1338-

1340.

97. Ueda K, Tabata H, Kawai T (2001) Magnetic and electric properties of transition-

metal-doped ZnO films. Appl Phys Lett 79, 988-990.

98. Sakka S. Handbook of Sol-Gel Science and Technology: Processing,

Characterization, and Applications. In: S Sakka (ed): Kluwer Academic

Publishers, 2005.

99. Bragg WL (1914) The Diffraction of Short Electromagnetic Waves by a Crystal.

Proceedings of the Cambridge Philosophical Society 17, 43-57.

100. Heidenreich RD. Fundamentals of Transmission Electron Microscopy.

Interscience Publishers, 1964.

101. Gringer. Diagram outlining the internal components of a basic TEM system. Free

Software Foundation, 2009:English: Diagram outlining the internal components

of a basic TEM system. SVG version of File:Scheme TEM en.png.

102. Jiles D (1998) Introduction to magnetism and magnetic materials. Chapman and

Hall.

Chapter 8

126

103. Gallagher WJ, Parkin SSP (2006) Development of the magnetic tunnel junction

MRAM at IBM: From first junctions to a 16-Mb MRAM demonstrator chip. IBM

J Res Dev 50, 5-23.

104. Kaeswurm B, O'Grady K (2010) Interfacial and bulk order in polycrystalline

exchange bias systems. J Appl Phys 107, 09D727-723.

105. Brown H, Dahlberg ED, Hou C (2001) Exchange bias measurements of

CoFe/IrMn. J Appl Phys 89, 7543.

106. Tsunoda M, Yoshitaki S, Ashizawa Y, Kim DY, Mitsumata C, Takahashi M

(2007) Enhancement of exchange bias by ultra-thin Mn layer insertion at the

interface of Mn-Ir/Co-Fe bilayers. phys status solidi (b) 244, 4470-4473.

107. Manzoor S, Vopsaroiu M, Vallejo-Fernandez G, O'Grady K (2005) Grain-size

effects in exchange-biased FeMn/NiFe bilayers. J Appl Phys 97, 10K118-113.

108. Prados C, Pina E, Hernando A, Montone A (2002) Reversal of exchange bias in

nanocrystalline antiferromagnetic–ferromagnetic bilayers. J Phys: Cond Matt 14,

10063.

109. Fecioru-Morariu M. Exchange bias of metallic ferro-/antiferromagnetic bilayers.

Effects of structure, dilution, anisotropy and temperature. Master of Science: aus

Iasi, Rumänien, 2008.

110. Aley NP, Vallejo-Fernandez G, Kroeger R, Lafferty B, Agnew J, Lu Y, et al.

(2008) Texture Effects in IrMn/CoFe Exchange Bias Systems. Magnetics, IEEE

Trans 44, 2820-2823.

111. Hong NH, Sakai J, Huong NT, Brize V (2005) Room temperature ferromagnetism

in laser ablated Ni-doped In 2O3 thin films. Appl Phys Lett 87, 102505-102503.

112. Seshadri R (2005) Zinc oxide-based diluted magnetic semiconductors. Current

Opinion in Solid State and Materials Science 9, 1-7.

113. Noice L, Seipel B, Grathoff G, Gupta A, Moeck P, Rao VK (2005) Structural

Studies of ZnO Calcined with Transition Metal Oxides. MRS Online Proceedings

Library 891, M3 - 10.1557/PROC-0891-EE1510-1510.

114. Weng H, Dong J, Fukumura T, Kawasaki M, Kawazoe Y (2006) First principles

investigation of the magnetic circular dichroism spectra of Co-doped anatase and

rutile TiO2. Phys Rev B 73, 121201.

115. Stamenov P, Venkatesan M, Dorneles LS, Maude D, Coey JMD (2006)

Magnetoresistance of Co-doped ZnO thin films. J Appl Phys 99, 08M124-123.

Chapter 8

127

116. Katayama M, Ikesaka S, Kuwano J, Yamamoto Y, Koinuma H, Matsumoto Y

(2006) Field-effect transistor based on atomically flat rutile TiO[sub 2]. Appl

Phys Lett 89, 242103-242103.

117. Misra P, Sharma TK, Porwal S, Kukreja LM (2006) Room temperature

photoluminescence from ZnO quantum wells grown on (0001) sapphire using

buffer assisted pulsed laser deposition. Appl Phys Lett 89, 161912-161913.

118. Jayakumar OD, Gopalakrishnan IK, Kulshreshtha SK (2006) Surfactant-Assisted

Synthesis of Co- and Li-Doped ZnO Nanocrystalline Samples Showing Room-

Temperature Ferromagnetism. Adv Mat 18, 1857-1860.

119. Ramamoorthy K, Kumar K, Chandramohan R, Sankaranarayanan K (2006)

Review on material properties of IZO thin films useful as epi-n-TCOs in opto-

electronic (SIS solar cells, polymeric LEDs) devices. Mat Sci Engg B 126, 1-15.

120. Ramaneti R, Lodder JC, Jansen R (2007) Anomalous Hall effect in anatase

Co:TiO[sub 2] ferromagnetic semiconductor. Appl Phys Lett 91, 012502-012503.

121. Mallia G, Harrison NM (2007) Magnetic moment and coupling mechanism of

iron-doped rutile TiO_{2} from first principles. Phys Rev B 75, 165201.

122. Xu Q, Hartmann L, Schmidt H, Hochmuth H, Lorenz M, Schmidt-Grund R, et al.

(2007) Magnetoresistance and anomalous Hall effect in magnetic ZnO films. J

Appl Phys 101, 063918-063915.

123. Chen G, Zeng F, Pan F (2009) Enhanced spin injection and voltage bias in

(Zn,Co)O/MgO/(Zn,Co)O magnetic tunnel junctions. Appl Phys Lett 95, 232508-

232503.

124. Erokhin SG, Deych LI, Lisyansky AA, Granovsky AB (2009) Light modulator on

excitons in a quantum well of an optical microcavity. Tech Phys Lett 35, 785-788.

125. Sedlmayr N, Dugaev VK, Berakdar J (2009) Current-induced interactions of

multiple domain walls in magnetic quantum wires. Phys Rev B 79, 174422.

126. Fukumura T, Yamasaki T, Yamada Y, Ueno K, Kawasaki M (2011) Oxide

Semiconductor Spintronics. Hyomen Kagaku 32, 134-138.

127. Kobayashi M, Muneta I, Schmitt T, Patthey L, Ohya S, Tanaka M, et al. (2012)

Digging up bulk band dispersion buried under a passivation layer. Appl Phys Lett

101, 242103-242104.

Chapter 8

128

128. Yao Z, Jia F, Tian S, Li C, Jiang Z, Bai X (2010) Microporous Ni-Doped TiO2

film Photocatalyst by Plasma Electrolytic Oxidation. ACS Appl Mat & Interfaces

2, 2617-2622.

129. Yeo Y-C, King T-J, Hu C (2002) Metal-dielectric band alignment and its

implications for metal gate complementary metal-oxide-semiconductor

technology. J Appl Phys 92, 7266-7271.

130. Edusi C, Hyett G, Sankar G, Parkin IP (2011) Aerosol-Assisted CVD of Titanium

Dioxide Thin Films from Methanolic Solutions of Titanium Tetraisopropoxide;

Substrate and Aerosol-Selective Deposition of Rutile or Anatase. Chem Vap Dep

17, 30-36.

131. Huang Y, Ho W, Ai Z, Song X, Zhang L, Lee S (2009) Aerosol-assisted flow

synthesis of B-doped, Ni-doped and B–Ni-codoped TiO2 solid and hollow

microspheres for photocatalytic removal of NO. Appl Catalysis B: Environmental

89, 398-405.

132. Huang Y, Zheng Z, Ai Z, Zhang L, Fan X, Zou Z (2006) Core−Shell

Microspherical Ti1-xZrxO2 Solid Solution Photocatalysts Directly from Ultrasonic

Spray Pyrolysis. J Phys Chem B 110, 19323-19328.

133. Tahir AA, Mazhar M, Hamid M, Wijayantha KGU, Molloy KC (2009)

Photooxidation of water by NiTiO3 deposited from single source precursor

[Ni2Ti2(OEt)2( -OEt)6(acac)4] by AACVD. Dalton Trans, 3674-3680.

134. Hong NH, Prellier W, Sakai J, Hassini A (2004) Fe- and Ni-doped TiO[sub 2]

thin films grown on LaAlO3 and SrTiO3 substrates by laser ablation. Appl Phys

Lett 84, 2850-2852.

135. Cho JH, Hwang TJ, Joh YG, Kim EC, Kim DH, Lee KJ, et al. (2006) Room-

temperature ferromagnetism in highly-resistive Ni-doped TiO2.Appl Phys Lett

88, 092505-092503.

136. Hou DL, Meng HJ, Jia LY, Ye XJ, Zhou HJ, Li XL (2007) Oxygen vacancy

enhanced the room temperature ferromagnetism in Ni-doped TiO2 thin films.

Phys Lett A 364, 318-322.

137. Uhm YR, Woo SH, Kim WW, Kim SJ, Rhee CK (2006) The characterization of

magnetic and photo-catalytic properties of nanocrystalline Ni-doped TiO2 powder

synthesized by mechanical alloying. JMag Mag Mat 304, e781-e783.

138. Cabrera AF, Errico L, Rodríguez Torres CE, Sánchez FH (2007) Influence of

thermal treatments on phase formation and magnetic behaviour in metal transition

doped TiO2. Physica B: Cond Matt 389, 103-106.

Chapter 8

129

139. Hoa N, Huyen D (2012) Comparative study of room temperature ferromagnetism

in undoped and Ni-doped TiO2 nanowires synthesized by solvothermal method. J

Mat Sci: Mater Electron, 1-6.

140. Bahadur N, Pasricha R, Govind, Chand S, Kotnala RK (2012) Effect of Ni doping

on the microstructure and high Curie temperature ferromagnetism in sol–gel

derived titania powders. Mat Chem Phys 133, 471-479.

141. Shannon R (1976) Revised effective ionic radii and systematic studies of

interatomic distances in halides and chalcogenides. Acta Crystallographica

Section A 32, 751-767.

142. Riyas S, Yasir V, Das P (2002) Crystal structure transformation of TiO2 in

presence of Fe2O3 and NiO in air atmosphere. Bulletin of Mat Sci 25, 267-273-

273.

143. Coey JMD, Venkatesan M, Fitzgerald CB (2005) Donor impurity band exchange

in dilute ferromagnetic oxides. Nat Mat 4, 173-179.

144. Yagi E, Hasiguti RR, Aono M (1996) Electronic conduction above 4 K of slightly

reduced oxygen-deficient rutile TiO2-x. Phys Rev B 54, 7945.

145. Tang H, Berger H, Schmid PE, Lévy F (1994) Optical properties of anatase

(TiO2). Solid State Commun 92, 267-271.

146. Jacob KT, Saji VS, Reddy SNS (2007) Thermodynamic evidence for order–

disorder transition in NiTiO3. The J Chem Therm 39, 230-235.

147. Komiyama H, Kanai T, Inoue H (1984) Preparation of porous, amorphous, and

ultrafine TiO2 particles by chemical vapor deposition. Chem Lett 13, 1283-1286.

148. Chen S, Mason MG, Gysling HJ, Paz-Pujalt GR, Blanton TN, Castro T, et al.

(1993) Ultrahigh vacuum metalorganic chemical vapor deposition growth and in

situ characterization of epitaxial TiO2 films. J Vac Sci & Tech A: Vac, Surf and

Films 11, 2419-2429.

149. Fictorie CP, Evans JF, Gladfelter WL. Kinetic and mechanistic study of the

chemical vapor deposition of titanium dioxide thin films using tetrakis-

(isopropoxo)-titanium(IV). Orlando, Florida (USA): AVS, 1994:1108-1113.

150. Rahtu A, Ritala M (2002) Reaction Mechanism Studies on Titanium

Isopropoxide–Water Atomic Layer Deposition Process. Chem Vap Dep 8, 21-28.

Chapter 8

130

151. Liu H, Ma HT, Li XZ, Li WZ, Wu M, Bao XH (2003) The enhancement of TiO2

photocatalytic activity by hydrogen thermal treatment. Chemosphere 50, 39-46.

152. Keys LK, Mulay LN (1967) Magnetic Susceptibility Measurements of Rutile and

the Magnéli Phases of the Ti-O System. Phys Rev 154, 453-456.

153. Soack Dae Y, Yajie C, Aria Y, Trevor LG, Xu Z, Dario AA, et al. (2006)

Oxygen-defect-induced magnetism to 880 K in semiconducting anatase TiO2−δ

films. J Phys: Cond Matt 18, L355.

131

List of Publications

1. R. Hussain, B. Kaeswurm and K. O'Grady

Sample fabrication effects in exchange bias systems

J. Appl. Phy. 109 (7), 07E533-533 (2011).

2. S. T. Hussain, K. Khan and R. Hussain

Size control synthesis of sulfur doped titanium dioxide (anatase) nanoparticles, its

optical property and its photo catalytic reactivity for CO2 + H2O conversion and

phenol degradation

J.Nat. Gas Chem. 18, 383-391 (2009).

3. S. T. Hussain, N. Niaz, A., I. Amed and R. Hussain

New Synthesis Procedure for the Production of Trimetallic Composite for

Hydrogen Storage

Intl.l Rev. Chem. Engg. 1, 238-242 (2009).

4. N. A. Niaz, I.Ahmad, S. Nasir, Z.Wazir, R. Hussain, N. R. Khalid and S. T.

Hussain Synthesis and Characterization Of Mg-al Alloys For Hydrogen Storage

Applications Digest Journal of Nanomaterials and Biostructures. 8(1), 423-431

(2013).

5. A. Nisar, S.T. Hussain, M.A.Iqbal, H. Ayesha, M. Arshad, S. Akram, Z. Ali, N.

Ahmad, S. M. Abbas and R. Hussain

Optoelectronic properties of evaporated antimony tin sulfide thin films for solar

cell applications

Elixir Renewable Energy Engg. 55, 13129-13132 (2013).

6. N. Ahmad, S.T. Hussain, B. Muhammad, T. Mahmood, Z. Ali, N. Ali, S.M.

Abbas, R. Hussain and S.M. Aslam

Effect of the Reaction Conditions on Al-Pillared Montmorillonite Supported

Cobalt-Based Catalysts for Fischer Tropsch Synthesis

Digest Journal of Nanomaterials and Biostructures. 8(1), 347-358 (2013).

Documents

Synthesis and Characterization of Nanostructures