Investigation on the structure and magnetic properties of co2mnsi heusler alloy for spintronic application

33 Chapter 3: MgO buffer layer effect on the structure and magnetic properties of Co 2 MnSi CMS films on MgO substrates 34 3.1 Study of CMS thin films on MgO substrates no buffer layer

Acknowledgements

This author would like to express her heartfelt gratitude to her supervisors,

Dr Chen Jingsheng and Dr Han Guchang for their guidance, advice, concern and

encouragement throughout the course of the project

The author would like to thank Dr Qiu Jinjun for the introduction and

maintenance of ULVAC sputter system, and Miss Luo Ping for the help in XPS

analysis

In addition, the author is grateful for the friendship and support of Huang

Lisen, Yang Yang, Ho Pin, Xu Dongbin, Si Huayan, Chen Chin, Li Huihui, Ko

Viloane, Lim Boon Chow, and other research staff and students in the Department

of Materials Science and Engineering and Data Storage Institute

Last but not least, the author would like to thank her family for their love,

support and understanding during the period of research project

Table of Contents

1.1 Gaint magnetoresistance (GMR) read head technology 3

1.1.1 Physics of GMR effect 3

1.1.2 Current-in-plane (CIP) spin valve 6

1.1.3 Current-perpendicular-to-plane (CPP) spin valve 8

1.2 Tunnel magnetoresistance (TMR) read head technology 10

1.3 Challenges for next generation read head over 1 Tbits/in2 11

1.4 Literature review of Co2MnSi half-metallic Heusler alloys 12

1.4.1 Basic properties of Heusler alloys 13

1.4.1.1 Origin of the bandgap 14

1.4.1.2 Slater-Pauling behavior 15

1.4.2 Parameters affecting MR ratio of CPP-GMR head using Co2MnSi Heusler alloy 17

1.4.2.1 Effect of lattice parameter 17

1.4.2.2 Spin-orbit coupling 19

1.4.2.3 Effect of temperature on spin polarization 19

1.4.2.4 Effect of disorder, defects, and doping 20

1.4.2.5 Other factors from device perspective 21

1.5 Objective of thesis 22

1.6 Outline of thesis 24

Chapter 2: Experimental methodologies 25

2.1 Sample fabrication by magnetron sputtering 25

2.2 Alternating gradient force magnetometer 26

2.3 X-ray diffraction 27

2.3.1 θ - 2θ x-ray scans 29

2.3.2 Rocking curve 29

2.3.3 Phi-scans 30

2.4 Transmission electron microscopy 31

2.5 X-ray photoelectron spectroscopy 32

2.6 Atomic force microscopy 33

Chapter 3: MgO buffer layer effect on the structure and magnetic properties of Co 2 MnSi (CMS) films on MgO substrates 34 3.1 Study of CMS thin films on MgO substrates (no buffer layer) 34

3.1.1 Experimental methods 35

3.1.2 Results and discussion 36

3.1.2.1 Structure properties 36

3.1.2.2 Magnetic properties 39

3.2 Study of CMS thin film on MgO-buffered MgO substrates 44

3.2.1 Experimental details 44

3.2.2 Results and discussion 45

3.2.2.1 Crystallographic structure 45

3.2.2.2 In-plane epitaxial growth relationship 46

3.2.2.3 Heusler L21 texture 48

3.2.2.4 Microstructure 49

3.2.2.5 Magnetic properties 50

3.3 Summary 54

Chapter 4: Cr buffer layer effect on the texture and magnetic properties of

4.2 Results and discussion 58

4.2.1 Effect of CMS in-situ annealing temperature 58

4.2.1.1 Crystallographic structure 58

4.2.1.2 Cr/CMS interface and multi-layer roughness 63

4.2.1.3 Magnetic properties 64

4.2.1.4 Microstructure 68

4.2.2 Effects of Cr in-situ annealing temperature 69

4.3 Summary 72

Chapter 5: Conclusions and future work 74 5.1 Conclusions 74

5.2 Future work 75

Summary

The effects of MgO and Cr buffer layers on the structures and magnetic

properties of Co2MnSi (CMS) thin films have been studied The CMS thin films

were deposited on MgO (001) single crystal substrates with and without MgO or Cr

buffer layer at room temperature by magnetron sputtering deposition and annealed

at various temperatures From 2-theta and Phi-XRD analysis, it was found that both

MgO and Cr buffer layers could help induce the (001) epitaxial growth of the CMS

thin films with B2 or L21 structure, while A2 structure was formed in the films

without buffer layer A four-fold 45º shift between MgO (022) peaks and Cr (022)

or CMS (022) peaks was obtained in the Phi-scan analysis of the post-annealed

MgO- and Cr-buffered CMS thin films, which confirmed the epitaxial relationship

of CMS [110](001) || MgO [100](001) and CMS [110](001) || Cr [110](001) || MgO

[100](001) A smoother MgO buffer layer surface was obtained in an Ar+O2

atmosphere compared to that obtained in Ar atmosphere, which resulted in a

stronger CMS (001) texture Moreover, it was found that the in-situ annealing of

the Cr buffer layer at 500°C destroyed the homogenous surface, while the

RT-deposited Cr layer showed a very smooth surface and subsequently gave a better

CMS crystallinity The saturated magnetization of the CMS thin films increased

with annealing temperature below 600ºC and then decreased when annealed further

at 600ºC A chemical analysis of inter-diffusion was examined and the results

indicated significant diffusion of Pt and Co in the MgO-buffered CMS thin films

annealed at 600°C for 1h and Cr in the Cr-buffered CMS thin films annealed at

600°C for 15min Large initial susceptibilities were found in the Cr buffered CMS

thin films For the MgO/CMS/Pt samples post-annealed at 600ºC, three-stage initial

curves were obtained, suggesting a pinning behavior in the magnetic reversal

mechanism

List of Figures

Figure 1.1: Data storage roadmap of HDDs Adapted from [1] 1Figure 1.2: Magnetoresistance magnitude of GMR versus TMR Adapted from [5] 3Figure 1.3: Normalized resistance versus applied magnetic field at 4.2K with CIP

Figure 1.4: Two current model of GMR effect with F/C multi-layers 5Figure 1.5: Schematic representation of an exchange biased spin valve structure, in

which the magnetization of one ferromagnetic layer is fixed by exchange coupling to an antiferromagnetic layer The other ferromagnetic layer has it magnetization direction changed by external

Figure 1.6: Different spin valve structures: (a) bottom type; (b) top type; (c) dual

type, Cap: capping layer; FL: free layer; NM: non-magnetic layer; RL: reference layer; PL: pinned layer, AFM: anti-ferromagnetic layer; SL:

Figure 1.10: Structure of (a) C1b half-Heusler alloys; (b) L21 full-Heusler alloys; (c)

B2 full-Heusler alloy; (d) A2 full-Heusler alloy Adapted from [32] 13Figure 1.11: Major combination of Heusler alloy formation Adapted from [32] 14Figure 1.12: Origin of the minority band gap in (a) NiMnZ and (b) Co2MnZ

Figure 1.13: Calculated total spin moments, in which dashed line represents the

Figure 1.14: DOS for NiMnSb and CoMnSb Adapted from [40] 18Figure 2.1: M-H loop of ferromagnetic materials Adapted from [69] 27Figure 2.2: Illustration of x-ray diffraction Adapted from [74] 28Figure 2.3: X-ray powder scans geometry Adapted from [75] 29

Figure 3.1: (a) L21 structure of CMS; (b) top view of MgO/Co2MnSi crystal

Figure 3.2: Schematic diagram of MgO(sub)/CMS/Pt layered structure 36Figure 3.3: AFM image of MgO substrate surface topography 36Figure 3.4: 2-theta XRD spectra of CMS films deposited at RT and post annealed at

400°C and 600°C All peaks labeled in * are from MgO substrate 37Figure 3.5: In-plane hysteresis loops of CMS films: (a) deposited at room

temperature; (b) post-annealed at 400°C; (c) post-annealed at 600°C; (d) extracted Ms and Hc curves as a function of temperature 40Figure 3.6: Line-scan EDX spectra of CMS thin films deposited on MgO substrates

at room temperature and annealed at (a) 400°C and (b) 600°C 41 Figure 3.7: The initial curves of both as-deposited CMS films and CMS films post-

annealed at 400°C and 600°C (1), (2), (3) represents for initial stage,

Figure 3.8: Schematic of MgO(sub)/MgO/CMS/Pt multi-layer structure 44Figure 3.9: The 2-theta XRD spectra of CMS deposited at RT and post-annealed at

various temperatures with 7nm MgO buffer layer deposited in (a)

Figure 3.10: The three dimension model of crystal structure of epitaxial grown

Figure 3.11: Phi-scan images of CMS {022} and MgO {022} planes in

MgO-buffered CMS thin films annealed at (a) 400°C and (b) 600°C 47 Figure 3.12: Phi-scan spectra for CMS (111) peaks of MgO-buffered CMS thin

Figure 3.13: Cross-sectional TEM images and IFFT image of the atomic lattice of

MgO-buffered CMS with MgO deposited in (a) Ar; (b) Ar+O2 gas

Figure 3.14: M-H loops of MgO-buffered CMS with MgO deposited in (a) Ar+O2;

(b) Ar gas atmosphere; extracted (c) Ms and (d) Hc curves as a function

Figure 3.15: The XPS depth profile of MgO(7nm)-buffered CMS films: (a)

deposited at room temperature; (b) annealed at 400°C; (c) annealed at

Figure 4.1: (a) L21 structure of CMS; (b) Top view of MgO/Cr/CMS crystal

Figure 4.2: Schematic of MgO(sub)/Cr/CMS/Pt layer structure 57Figure 4.3: The spectra of 2-theta XRD of Cr/CMS/Pt multi-layers 58Figure 4.4: (a) CMS (002) and (004) Peak-integration and (b) Rocking curve of

Figure 4.5: The 3-D crystal structure model of epitaxial grown MgO(sub)/Cr/CMS

Figure 4.6: Phi-scans of MgO, Cr and CMS {022} planes of Cr-buffered CMS thin

films annealed at (a) 300°C; (b) 400°C; (c) 500°C; (d) 600°C 62Figure 4.7: Phi-scan of CMS (111) peaks of Cr-buffered CMS films annealed at

Figure 4.8: AFM images of as-deposited Cr buffer layer 63Figure 4.9: AFM images of Cr/CMS/Pt multi-layers: (a) deposited at room

temperature; (b) annealed at 300°C; (c) annealed at 400°C; (d) annealed

Figure 4.10: (a) In-plane hysteresis loop of Cr-buffered CMS films deposited at

room temperature and post-annealed at various temperatures; (b) extracted Ms and Hc curves as a function of temperature 65

Figure 4.12: XPS depth profile of Cr/CMS/Pt multi-layers in which CMS thin films

were: (a) deposited at room temperature; (b) annealed at 300°C; (c)

Figure 4.13: Cross-sectional TEM images of Cr/CMS/Pt layers in which CMS films

were annealed at (a) 300ºC; (b) 400ºC; (c) 500ºC, (d) high resolution TEM image of Cr/CMS/Pt multi-layers in which CMS films were

Figure 4.14: 2-theta spectra of Cr/CMS/Pt multi-layers with various Cr in-situ

Figure 4.15: AFM images of Cr(10nm)-buffered MgO substrates with Cr in-situ

annealing for 30min at (a) 300°C; (b) 400°C; (c) 500°C; (d) 600°C 71

AFM Atomic force microscopy

AMR Anisotropic magnetoresistance

AGFM Alternating gradient force magnetometry

dhkl Interplanar spacing

DF Dark-field

DSI Data Storage Institute

DOS Density of States

EDX Energy-dispersive X-ray spectroscopy

△ RA Resistance Change per Unit Area

RAMAC Random Access Method of Accounting and Control

RF Radio frequency

Ra Surface roughness average

Rrms Mathematical Root Mean Square Roughness

SV Spin valve

SOC Spin-orbit coupling

TMR Tunneling Magnetoresistance

T Temperature

Tc Curie temperature

TEM Transmission electron microscopy

XRD X-ray diffraction

XPS X-ray photoelectron spectroscopy

Zt Total valence electrons number

Zavg Average value of valence electrons number

Zi Current value of valence electrons number

Chapter 1: Introduction

Hard disk drives (HDD) have a leading position in the digitalized information storage

area, accompanied by a dramatic increase in storage density at a rate larger than 40% per year

Fig 1.1 shows the evolution and roadmap of HDD data storage [1] The first magnetic HDD

called the Random Access Method of Accounting and Control was introduced by IBM with an

areal density (AD) of 2000 bits/in2 To date, HDDs with AD of around 500 Gbits/in2 are widely

used Recently, HDD with AD of around 803 Gbits/in2 has been demonstrated in the lab [2]

Figure 1.1: Data storage roadmap of HDDs Adapted from [1]

The fast developing speed in HDD industry is closely related to the development in the

field of spintronics in read head technology The first breakthrough came in 1991 when the

inductive read head was replaced by the anisotropic magnetoresistive (AMR) read head

Magnetoresistance (MR) refers to the resistance variation, between maximum and minimum

resistances, normalized against the minimum resistance value when the magnetic field changed

The readout signal is proportional to the MR ratio Although the MR ratio of AMR head was

only 1%, it was more than double the value demonstrated by inductive read head

The second major development of the read head technology came with the discovery of

the giant magnetoresistance (GMR) effect by Fert’s group [3] in France as well as Grunberg’s

group [4] in Germany in 1988 In 1997, the first GMR read head was introduced in commercial

HDDs with AD of around 2 Gbits/in2 However, this GMR read head with current-in-plane (CIP)

geometry possessed some key limitations As such, both current-perpendicular-to-plane

(CPP)-GMR and tunneling magnetoresistance (TMR) read heads were investigated as promising

candidates to replace this CIP-GMR read head Fig 1.2 shows the magnetoresistance of GMR

versus TMR [5] The output signal of MgO and Al2O3 based TMRs indicated a larger

magnetoresistance than the CPP-GMR read head In 2005, Seagate introduced the first TMR

read head with both MgO and Al2O3 insulators

Figure 1.2: Magnetoresistance magnitude of GMR versus TMR Adapted from [5]

1.1 Gaint magnetoresistance (GMR) read head technology

1.1.1 Physics of GMR effect

The giant magnetoresistance (GMR) effect was discovered in iron (magnetic) and

chromium (non-magnetic) alternating layers In these structures, when the magnetization

directions in neighbouring iron layers changed from antiparallel to parallel, a significant change

in resistance was observed This phenomenon was known as GMR For samples with 9 Å Cr

layers, the MR (MR = [ρAP –ρP]/ ρP) ratio increased four-fold from 20% at room temperature to

80% at 4.2 K, as shown in Fig 1.3

Figure 1.3: Normalized resistance versus applied magnetic field at 4.2K with CIP current Adapted from [19]

In these multilayer structure, the anti-parallel arrangement of magnetization between

adjacent ferromagnetic (FM) layers resulted in high resistivity while a parallel arrangement of

the magnetization led to an obvious reduction of resistivity magnitude However, the saturation

field Hs required to overcome the antiferromagnetic interlayer coupling effect and align the

magnetization of consecutive layers was too large for GMR to be applicable in real devices On

the other hand, further investigation revealed that the anti-ferromagnetic coupling arrangement

was not a prerequisite for the GMR effect [6]

The physical origin of GMR effect can be explained by the effect of the electron spin on

the electronic transport in ferromagnetic conductors, i.e spin-dependent scattering A qualitative

understanding of GMR effect was given by Mott’s two-current model [7], as shown in Fig 1.4

The key point of this model is that two independent conduction carriers – spin up and spin down

electrons existed in FM conductors

Figure 1.4: Two current model of GMR effect with F/C multi-layers

As the resistance of the multi-layer structure arose from the scattering processes of spin

electrons, strong scattering led to short mean free path, while weak scattering led to long mean

free path When the magnetization directions of two ferromagnetic layers were parallel, the

spin-up electrons, assumed to be parallel to magnetization, passed through the multi-layers with

almost zero scattering On the other hand, the spin-down electrons were scattered strongly as

their spin aligned anti-parallel to the magnetization direction Thus, parallel configuration

resulted in low resistivity If the magnetizations in the two ferromagnetic layers were

anti-parallel, high resistivity would be induced as both spin-down and spin-up electrons would be

scattered strongly

1.1.2 Current-in-plane (CIP) spin valve

Figure 1.5: Schematic representation of an exchange biased spin valve structure, in which the magnetization of one ferromagnetic layer is fixed by exchange coupling to an antiferromagnetic layer The other ferromagnetic layer has it magnetization direction changed by external magnetic field

With the discovery of the GMR effect in the CIP case, S.S Parkin’s group proposed a

CIP spin valve (SV) structure, as shown in Fig 1.5 This SV structure, composed of four layers,

had a much smaller switching magnetic field (Hs) which make the GMR effect practical for

spintronic devices [8] The magnetization of the pinned layer (PL) would be pinned along the

anti-ferromagnetic (AFM) layer field cooling direction, while the magnetization of the free layer

(FL) would change with the external field The pinning effect was not only the result of

unidirectional anisotropy generated by domains but also that of interface exchange coupling

Figure 1.6: Different spin valve structures: (a) bottom type; (b) top type; (c) dual type, Cap: capping layer; FL: free layer; NM: non-magnetic layer; RL: reference layer; PL: pinned layer, AFM: anti-ferromagnetic layer; SL: the seedlayer

In a read head sensor, it would be crucial to reduce the interaction influence of the pinned

layer on the free layer, due to interlayer and magnetostatic coupling Hin On one hand, Hin can be

reduced by increasing the NM layer thickness However, this would reduce the MR signal

through scatterings which decreased the flow of polarized conduction electrons [9] On the other

hand, a synthetic anti-ferromagnetically coupled pinned layer [10] can be used, as shown above

in Fig 1.6 (a) This spin valve structure with pinned layer deposited first (bottom type) would

reduce or even cancel off the stray field generated from the pinned layer Fig 1.6 (b) gives a

reversed strucuture (top type) where the free layer was deposited first The dual spin valve in Fig

1.6 (c) shows a combination of both top and bottom types Although this structure gave higher

MR ratio, it required more spacing between shields in devices Therefore, this dual spin valve

was limited for ultrahigh density recording

1.1.3 Current-perpendicular-to-plane (CPP) spin valve

Since the discovery of the GMR effect, much works had been done with the CIP spin valve

structure It was relatively easy to measure the resistance of thin films with CIP geometry, while

the resistance measurement of CPP spin valves required complicated nano-scale device

fabrication process as the sensor length was quite small However, CIP-GMR had two main

limitations, as shown in Fig 1.7 (a)

(a) CIP (b) CPP Figure 1.7: The (a) CIP and (b) CPP GMR read head geometry Adapted from [11]

(1) Two insulator gaps existed between the sensor and the two shields A smaller gap would

promote higher linear density As such, the ideal condition was to eliminate the gap

However, this would result in a short circuit between the shield and sensor

(2) The reduction of track width could result in a conflict between higher track density and

linearly decreased output signal

Unlike the CIP geometry, the two gaps were not necessary in a CPP spin valve head as

the current flowed from the top shield to the bottom shield through sensor stack In addition, both

theoretical [12] and experimental results [13] revealed that the CPP-GMR showed higher

intrinsic MR ratio compared to CIP-GMR The physics involved in the MR measurement of CIP

and CPP geometry was significantly different In a theoretical paper by T Valet and A Fert in

1993 [14], they showed that the most important difference between the CPP and CIP was

induced by the spin transport process For CPP, the spin transportation was perpendicular to film

interface, which included a spin accumulation effect This effect allowed the spin transportation

in CPP-GMR to be dependent on the long spin diffusion length rather than the short mean free

path in CIP geometry Moreover, a spin dependent interface resistance introduced by specular

reflection can also be found in CPP spin valve The famous Fert-Valet model, which mainly

focused on the effect of bulk and interface spin-asymmetry coefficients, was then widely used in

the investigation of CPP-GMR devices

Unfortunately, the MR ratio of CPP-GMR still fell below the value required to achieve

sufficient bit error rate for areal density exceeding 300 Gbits/in2 [15-18] after several years of

investigation The current used read head is based on CPP-TMR

1.2 Tunnel magnetoresistance (TMR) read head technology

When the metallic NM layer in a tri-layer GMR spin valve is replaced by a thin insulator,

the mechanism for MR becomes spin dependent tunneling This phenomenon is called tunneling

magneto-resistance (TMR) and the junction is called magnetic tunnel junction (MTJ)

Figure 1.8: Schematic density of states for both magnetic electrodes with the parallel and paralle arrangement of magnetizations The conductivity of each spin channel is proportional to the spin DOS in the emitter and collector electrode Adapted from [19]

anti-The tunneling process was not only dependent on the available electronic channels in FM

electrodes like in GMR, but also dependent on thickness and height of barrier Julliere reported

the first work of TMR measurement in a Co/Oxided-Ge/Fe MTJ in 1975 [20] In his article,

Julliere proposed an explanation of spin polarized tunneling effect Assuming spin conservation

in the tunneling process, the conductance can be understood by the sum of two independent

channels, as shown in Fig 1.8 [19] The famous Julliere formula relates the relative change of

conductance with the density of states of each spin channel:

TMR=

2 1

2 11

2

P P

P P

Di Di

D is the DOS of the FM electrodes for spin-up and spin-down direction at Fermi level

Researches on TMR have been very active since 1995 and major breakthroughs were

made in 2004 at both Tsukuba (Yuasa et al) and IBM (Parkin et al) It was found that very large

TMR ratio (200% at room temperature) could be obtained from MTJ with high quality MgO

barrier TMR ratio of about 600% was reported later [21] However, the major challenge of TMR

head was its high resistance, which limited the working frequency and thus reading speed

Reducing the insulating layer thickness also introduced pin-holes into the structure and

deteriorated its performance

1.3 Challenges for next generation read head over 1 Tbits/in2

As discussed above, TMR-based sensors with higher output signal have replaced

CIP-GMR sensors in HDD read heads However, there is increasing interest to replace TMR-heads

with CPP-GMR heads for 1 Tbit/in2 read heads This is mainly due to the low resistance area

product in the all-metallic layer structure and the lower capacitance of CPP spin valves, enabling

higher data transfer rates However, the major drawback of the current CPP-GMR device is the

low MR signal at room temperature

1.4 Literature review of Co2MnSi half-metallic Heusler alloys

In recent years, half-metallic ferromagnetic materials have attracted much attention due

to possible applications in the field of spin-electronics The existence of these materials was

predicted using ab-initio calculations by de Groot et al 1983 [22] As shown in Fig 1.9, spin up

and down electrons in the band structure of these materials showed completely different

behaviors Half-metals can essentially be treated as hybrids of metals and semiconductors As the

minority spin band showed semiconductor-like behaviour with a gap at Fermi level, these

materials exhibited a 100% polarization at Fermi level Half-metallic ferromagnets can thus be

expected to maximize the efficiency of spin-electronic devices, giving high MR ratios in

CPP-GMR read heads

Many materials have been predicted to be half-metallic by ab-initio calculations, such as

transition metal chalcogenides (e.g CrSe) and pnictides (e.g CrAs) [23-26], oxides CrO2 and

Fe3O4 [27], europium chalcogenides (e.g EuS) [28], double perovskites (e.g Sr2FeReO6) [29],

and other kinds of materials [30,31] However, amongst all these materials, the so-called half and

full-Heusler alloys have attracted much more interest due to their high Curie temperatures [32]

which is a requirement for practical devices The basic properties and research progress on

Heusler alloys will be discussed below

Figure1.9: Schematic representation of the DOS for a half-metal with respect to normal metals and semiconductors [33]

1.4.1 Basic properties of Heusler alloys

Figure 1.10: Structure of (a) C1 b half-Heusler alloys; (b) L2 1 full-Heusler alloys; (c) B2 full-Heusler alloy; (d) A2 full-Heusler alloy Adapted from [32]

The Heusler alloys can be characterized into half-Heusler alloy (XYZ) in C1b structure

and full-Heusler alloy (X2YZ) in L21 structure, as shown in Fig 1.10 (a) and (b), respectively X

and Y atoms represent transition metals, while Z is either a non-magnetic metal or a

semiconductor, as shown in Fig 1.11 [22, 32] The unit cell consists of four interpenetrating face

centered cubic (fcc) sublattices, in which the C1b structure is shaped by removing one of the X

sites in L21 structure In addition, Y-Z atomic disorder in L21 structure of full-Heusler alloy will

result in the formation of B2 structure, while A2 structure will form when X-Y and X-Z disorder

occur

Figure1.11: Major combination of Heusler alloy formation Adapted from [32]

1.4.1.1 Origin of the bandgap

According to the calculations by Galanakis et al [34], the origin of the bandgap in

Heusler alloys is caused by the d-d states hybridization of X and Y transition metals, as the DOS

in the vicinity of EF is dominated by the d-states The formation of this gap in half-Heusler alloys

(Fig 1.12(a)) and full-Heusler alloys (Fig 1.12(b)) is not exactly the same In the case of

half-Heusler alloys, the gap is formed by the hybridization states between elements X and Y directly,

while in full-Heusler alloys, the hybridization between the elements X happened before the X-Y

elements hybridization, in which only two bonding states among these four X–X orbitals

eventually hybridized with the Y element

Figure1.12: Origin of the minority band gap in (a) NiMnZ and (b) Co 2 MnZ Adapted from [32]

1.4.1.2 Slater-Pauling behavior

Galanakis et al [34,35] reported analogous Slater-Pauling behaviour in the Heusler alloys

with binary transition metal alloys, which is decribed as Mt = Zt - 18 (half Heusler) and Zt – 24

(full Heusler), where Mt represents total moments per formula unit, and Zt represents total

valence electrons number This behavior shown in Fig 1.13 is a theoretical guide to achieve

desired magnetic properties by substituting Y atoms with other transition metals in Heusler

alloys According to Fig 1.11, there are more than 2000 possibilities to form Heusler alloys

However, there are only tens of alloys which have been reported based on this behavior For

example, a great improvement of Tc to about 750 K was successfully made by the substitution of

Cr with Fe atoms in Co2CrAl HMFs (Tc around RT) with a new composition of Co2Cr0.6Fe0.4Al

[36-39]

Figure 1.13: Calculated total spin moments, in which dashed line represents the Slater-Pauling behavior Adapted from [40]

1.4.2 Parameters affecting MR ratio of CPP-GMR head using Co2MnSi Heusler alloy

Amongst all the Heusler alloys, Co2MnSi, with the full-Heusler L21 structure (space

group Fm3m) has attracted much attention as it was predicted to be a stable half-metal due to its

large band-gap of 0.4 eV [41] in the minority spin band and high Curie temperature of 712°C

[42] Polarization of around 60% of both bulk and thin film full-Heusler alloy Co2MnSi has been

obtained by point contact andreev reflection spectroscopy measurements [43-47] Unfortunately,

experimental attempts on CPP-GMR devices using Heusler alloy by Yakushiji et al achieved a

MR ratio of only 2.4% at RT in a pseudo spin valve structure of Co2MnSi/Cr/Co2MnSi [17] This

section serves to give a brief review on the parameters which affect spin polarization in full

-Heusler alloys

1.4.2.1 Effect of lattice parameter

Magnetic and electronic properties of both half-Heusler in C1b structure and full-Heusler

in L21 structure are dependent on the magnitude of the lattice parameter Density of States (DOS)

in both NiMnSb and CoMnSb with a lattice parameter change of ±2% was calculated [40], as

shown in Fig 1.14 Although the half-metallic property was conserved in both expansion and

compression scenarios, shifts of Fermi level occurred Based on their calculations, the shift of EF

was attributed to the larger extension of p states compared to the d states of Sb Movement of

Fermi level towards conduction band took place in compression, while Fermi level moved

towards the valence band during expansion In addition, the dominant strong hybridization

between Mn d and Ni or Co d states led to a slight increase in the size of the gap during

compression Similar behaviour is expected of other Co2MnZ alloy compounds where the

half-metallicity is preserved with lattice constant change of ±2% [34] In addition, a lattice

compression of 4% was reported to lead to a strong increase in band gap energy of 23% [48]

Figure 1.14: DOS for NiMnSb and CoMnSb Adapted from [40]

1.4.2.2 Spin-orbit coupling

Spin-orbit coupling (SOC) was neglected during the calculations of half-metallicity

mentioned above Taking into account the SOC, the electron spin would no longer be a good

quantum number As a result, the electron eigenfunction would not conserve the spin degree,

even at 0 K However, DOS within the gap is expected to be less in materials which have weak

SOC effect and its polarization is close to 100% [49, 50] The Heusler alloys like Co2MnSi,

Co2MnGe and Co2MnSn showed small orbit moments based on the calculations of Galanakis et

al [48, 51]

1.4.2.3 Effect of temperature on spin polarization

Several groups have investigated the temperature effect on the polarization of Heusler

alloys qualitatively and quantitatively based on different assumptions and theories, such as

tight-binding model, constrained density-function approach, dynamical mean-field way, and

double-exchange theory [52-57] For example, based on the constrained density-functional theory,

studies on NiMnSb showed that the minority spin-bands shifted cross the Fermi level gradually

as temperature increased and finally a collapse in polarization occurred at around 0.4 Tc (RT)

These calculations were consistent with experimental results of MR ratio loss as temperature

changes from 4.2 K to RT

1.4.2.4 Effect of disorder, defects, and doping

As we introduced in Fig 1.10, Heusler alloys can form B2 and A2 structures when X-Y

and X-Z/Y-Z disordering occur in L21 structure at temperatures below the melting point This

would happen during the deposition of Heusler alloy thin films in the fabrication process of

CPP-GMR devices Investigations showed that some Heusler alloys retained their half-metallicity in

B2 structure, while A2 disorder degraded the spin polarization significantly [58, 59] Picozzi et

al [58] investigated the formation of defects in full-Heusler alloy, in particular Co2MnGe and

Co2MnSi They found that the Mn antisites had the lowest formation energy and did not destroy

the half-metallicity in contrast to Co antisites In addition, large formation energies of the Mn-Si

and Mn-Co atomic swaps were found However, these results cannot be generalized to all

Heusler alloys Recently, the investigation of Nd doping effect on the transport and magnetic

properties of CMS had been reported by K Hono et al [60] From the resistivity measurements

at low temperatures, it was concluded that electron-magnon scattering was suppressed in Nd-rich

CMS phase This was based on the understanding that small density of states near the Fermi

level in the spin down mode was related to the mixing of spin up and down DOS caused by

inelastic electron-magnon scattering [61] There were many other factors contributing to the loss

of spin polarization of CMS For example, Suk J Kim reported that fcc Co precipitated together

with Co MnSi at annealing temperature of 600°C, indicating a meta-stable phase of Co MnSi

1.4.2.5 Other factors from device perspective

According to Valet-Fert model, the use of highly spin-polarized metallic Heusler alloys

as FM electrodes of CPP-GMR devices was a feasible way to enhance the MR ratio by

increasing the bulk spin asymmetry However, in a multi-layered CPP-GMR device structure,

many other factors may take effect on the output of CPP-GMR devices Firstly, low resistivity

and large spin-diffusion length were required for the space layer to obtain large CPP-GMR

values Low resistivity was also desirable for the buffer layer, which served as the bottom

electrode for the measurement of CPP-GMR, to further decrease the total resistance of the CPP

structure A relatively high MR ratio of 14% at 6 K was obtained using Ag buffer layer with

Co2FeAl0.5Si0.5 full-Heusler alloys electrodes [62] Secondly, the interlayer diffusion between the

FM and NM layer caused by high deposition or annealing temperature of full-Heusler alloys had

to be reduced in the CPP-GMR devices, as the inter-diffusion can result in the formation of

magnetically dead layers Hence, a compromise must be made between the crystal quality and

possible interfacial conditions [63] Thirdly, the interfacial spin-dependent scattering played an

important role in the MR ratio of CPP-GMR devices Ambrose and Mryasov [33] proposed a

selection criterion for maximizing the interface spin asymmetry by changing the ferromagnetic

metal and non-magnetic space layers They pointed out that both band matching for majority

spin channel and mismatching for minority spin channel at EF played important roles The good

band matching allowed the spin-up electrons to propagate across the interface On the other hand,

poor matching increased the scattering of spin-down electrons Seagate introduced an all-Heusler

alloy CPP-GMR spin valve using ferromagnetic Co2MnGe and non-magnetic Rh2CuSn Based

on band structure calculations, the interface spin asymmetry of this structure would be

maximized [64] However, some degree of disorder caused a loss of polarization at EF and hence

limited the MR ratio (6.7%) of this system

1.5 Objective of thesis

As discussed in the previous sections, half metallic alloys, Co2MnSi (CMS), with the

full-Heusler L21 structure (space group Fm3m), have attracted much attention in the field of

spin-electronics due to its large band gap of 0.4eV in the minority spin band and high Curie

temperature of 712°C According to Valet-Fert model, the use of highly spin-polarized metallic

Heusler alloys as FM electrodes of GMR devices enhanced the MR ratio mainly by increasing

bulk spin asymmetry However, there are two main issues which could degrade the performance

of GMR devices using CMS as FM electrodes One of the issues involved the loss of spin

polarization caused by Co-Mn or Co-Si disorder, which is represented by A2-type structure The

other is surface roughness, which plays an important role in spin-electronics applications

Different buffer layers had been used to induce the (001) texture of CMS for further achieving

B2 or L21 structure For instance, (001) texture of CMS on MgO buffered MgO substrate ha d

been reported, with MgO buffer layer deposited by e-beam evaporation [65]

The main objective of this research work is to investigate the effects of different buffer

layers on the structures and magnetic properties of CMS thin films

1) Co2MnSi full-Heusler alloy thin films were sputter-deposited onto MgO (001) single

crystal substrates without buffer layer to study the structures and magnetic properties of

CMS thin films

2) MgO buffer layers were deposited in different gas atmospheres (Ar and a mixture of Ar

and oxygen) on MgO substrate by magnetron sputtering After which, the structures,

magnetic properties and interfacial morphologies of these MgO buffered CMS thin films

were studied

3) Cr was used as the buffer layer The structures and magnetic properties of the Cr-buffered

CMS films were studied Inter-diffusion of component elements as well as thin films

roughness relative to CMS in-situ annealing temperature was examined Cr buffer layer

in-situ annealing effect on the Cr/CMS interface morphology was investigated

The work presented in this thesis is an original work on the structures and magnetic

properties of Co2MnSi (CMS) Heusler alloy thin films for spintronic application It would

provide guidance on the understanding of structural and magnetic properties of CMS Heusler

alloys and its further application into the field of spin-electronics

1.6 Outline of thesis

This thesis was organized into 5 chapters Chapter 1 gave an introduction of HDDs read

head technology development and basic principles involved in GMR and TMR read head, and

summarized current status of full-Heusler alloy Co2MnSi as ferromagnetic layer of CPP-GMR

read head from both materials and device aspects Chapter 2 gave the outline of the experimental

techniques with regards to sample fabrication, characterization and their corresponding working

principles In chapter 3, the structures and magnetic properties of CMS thin films on MgO single

crystal substrates and MgO-buffered MgO single crystal substrates were studied In chapter 4,

the effects of Cr buffer layer on the structural and magnetic properties of CMS films were

investigated Inter-diffusion of Cr element as well as Cr/CMS interface roughness were

examined Conclusion and Future work of the thesis was given in chapter 5

Chapter 2: Experimental methodologies

In this Chapter, we introduced the experimental techniques used in this thesis for sample

fabrication and characterization For sample fabrication, thin films (MgO, Co2MnSi, Cr, Pt) were

all deposited on MgO substrates in high vacuum Magnetron sputtering system X-ray Diffraction

(XRD) and Transmission electron microscopy (TEM) were used to analyze the structural

properties of thin films Alternating gradient force magnetometry (AGFM) was used for the

measurement of magnetic properties Atomic force microscopy (AFM) was used to study surface

morphologies Line-scan energy-dispersive X-ray spectroscopy (EDX) and X-Ray photoelectron

spectroscopy (XPS) depth profile analysis were used to identify distribution of elements in thin

films

2.1 Sample fabrication by magnetron sputtering

Sputtering technique is widely used in the magnetic recording industries for thin films

deposition There are two kinds of collision processes involved in the plasma, elastic and

inelastic process:

Elastic scattering: e + Ar → e + Ar

Inelastic Ionization (conversely, recombination): e + Ar → 2e + Ar+

Inelastic Excitation (conversely, relaxation): e + Ar → e + Ar*

where Ar* represents the excited state of Ar atom Elastic collision involves the interchange of

kinetic energy only; while inelastic collision involves exchange of internal energies

DC sputter deposition is suitable for the deposition of Co2MnSi, Pt, Cr, but not MgO with

non-conducting property Initiation of plasma is difficult when applying DC voltage to an

insulating MgO target To avoid this problem, a high frequency alternating voltage is used in

place of DC voltage The RF voltages can be coupled capacitively through the insulating target

to initiate the plasma Detailed discussion can be found in text books on sputtering processes [66,

67]

2.2 Alternating gradient force magnetometer

Alternating gradient force magnetometer (AGFM) is commonly used for characterization

of magnetic materials, such as hysteresis loop and initial magnetization curves It has a high

sensitivity of 10-8 emu and small sample size of 3×3 mm2 The working system used in this study

was Model 2900 MicroMagTM system The working principles are summarized below

The testing sample is attached on a fragile glass rod and mounted to a piezoelectric

transducer which oscillates as the external magnetic field The alternating field gradient can

apply a force on the sample and this force is proportional to the magnetic moment [68] AGFM

can be used to measure the hysteresis loop including initial magnetization curve, as shown in Fig