Investigation on new material for high density recording media with high thermal stability

1787.5 The exchange bias field Hex and interface domain wall energy σ dependence on the magnetic properties Hc of the CoDy layer .... 1877.6 Magnetic behaviors of exchange coupled CoDy/

THEORETICAL AND EXPERIMENTAL STUDY

OF EXCHANGE COUPLED MEDIA

REN HANBIAO

NATIONAL UNIVERSITY OF SINGAPORE

2006

THEORETICAL AND EXPERIMENTAL STUDY

OF EXCHANGE COUPLED MEDIA

REN HANBIAO (B ENG TSINGHUA UNIV.)

A THESIS SUBMITTED FOR THE DEGREE OF PH D OF PHILOSOPHY DATA STORAGE INSTITUTE, A-STAR, SINGAPORE ELECTRICAL AND COMPUTER ENGINEERING

DEPARTMENT NATIONAL UNIVERSITY OF SINGAPORE

2006

I would like to express my sincere gratitude to my supervisors, A/Prof Wang Jian Ping and Prof Chong Tow Chong, for offering me the chance to study and do my research at DSI, and for their invaluable advice and patient guidance throughout all my work done there

I am truly indebted to Dr Shan Zhensheng, Dr Zhou Tiejin, from whom I have gained

much theoretic knowledge and invaluable advice Dr Shan also gave me helpful support during VSM and AGM measurement

I am especially thankful to Dr Chen Jinshen, Lim Boon Chow, and Pock, who aided

me greatly in the trouble-shooting of the sputtering machine They were always so patient whenever I encountered any problem

My thanks also go to:

Mr Soo Eng Wei, Ms Pang Siew It, Ms Chow Shiaw Kee, Mr Hee Ching Hian, and all other staffs and fellow scholars of the Media and Materials Group from the Data Storage Institute, Sun Chengjun, Jiang Weiwei, Shi Xiao, Zheng Yufang, Hu Jiangfeng, Zhao Yan and Lü Meihua, who were extremely helpful with their assistance and friendship I particularly appreciate the opportunity to spend four years with my fellow scholars

I also would like to thank the National University of Singapore for its financial support and the Data Storage Institute for supplying me with an excellent research environment Last, but not least, I am especially grateful to my wife Tan Qiuyan, my daughter, Ren Jiayue and my family for their encouragement, care, and support

ACKNOWLEDGEMENTS i

Abstract ix

List of Figures and Illustrations xi

List of Tables xx

1 Introduction 1

1.1 Overview of magnetic recording industry 1

1.2 Brief overview of magnetic recording 3

1.3 Thin film media 5

1.3.1 Substrate 6

1.3.2 Underlayer and seedlayer 6

1.3.3 Intermediate layer 7

1.3.4 Magnetic layer 8

1.3.5 Overcoat and lubricant 9

1.4 Key performance indices for current medium 10

1.4.1 Areal density 11

1.4.2 Signal to noise ratio 12

1.4.3 Thermal stability factor 14

1.5 Issues with and resolutions for further increasing the areal density 16

1.5.1 Traditional medium structure 17

1.5.2 IBM and DSI proposed medium (AFC and LAC) 17

1.5.3 Fujitsu proposed medium (SFM) 18

1.5.4 Newly proposed medium in this work 18

1.6 Objective of this research 19

2 Literature review 24

2.1 The origin of exchange coupling 24

2.1.1 The concept of exchange interaction 24

2.1.2 Direct exchange interaction 25

2.1.3 Indirect exchange interaction 26

2.1.4 RKKY interaction 26

2.1.5 Superexchange interaction 27

2.1.6 Double exchange interaction 29

2.2 The interlayer exchange coupling and its effect on magnetic properties 30

2.2.1 Exchange anisotropy discovered by Meiklejohn and Bean 30

2.2.2 RKKY interlayer exchange coupling 35

2.3 The measurement of exchange coupling 35

2.3.1 Magnetization measurement 36

2.3.2 Torque measurements 37

2.3.3 Ferromagnetic resonance measurements 38

2.3.4 Neutron diffraction measurements 40

2.3.5 Magnetoresistance measurements 41

2.3.6 AC-susceptibility measurements 42

2.3.7 Domain observation measurements 43

2.3.8 Brillouin scattering measurements 43

2.3.9 Mössbauer effect measurements 44

2.4 Types of exchange coupled system 45

2.4.1 Small particles 45

2.4.2 Coated antiferromagnetic single crystals 46

2.4.4 Oxide thin film AFMs 47

2.4.5 Metallic thin film AFMs 48

2.4.6 Other thin film AFMs 48

2.4.7 Ferrimagnets 49

2.5 Factors that affect exchange coupling and the unidirectional anisotropy 49 2.5.1 Thickness 49

2.5.2 AFM orientation 53

2.5.3 Interface roughness 54

2.5.4 Block temperature 55

2.5.5 Training effect 56

2.5.6 Perpendicular coupling 57

2.6 The theoretical research in exchange coupling 58

2.6.1 Néel domain model 60

2.6.2 Early random interface model 61

2.6.3 AF domain wall models 62

2.6.4 Orthogonal F and AF magnetic lattices 63

2.6.5 Random interface field models 64

2.6.6 The frozen interface model 66

2.6.7 Local pinning field variation 69

2.7 Perpendicular exchange coupled system 70

2.7.1 Perpendicular ferromagnetic layer thickness dependence 71

2.7.2 AFM layer thickness dependence 72

2.7.3 Interface between the FM and AFM layers 75

2.7.4 Material of FM layers 76

2.8 Summary 76

3 Experimental procedures 90

3.1 Introduction 90

3.2 Sputter deposition 91

3.2.1 Plasma generation and DC diode plasma generation 91

3.2.2 Condensation on the substrate to form the film on the surface 94

3.2.3 Magnetron DC diode plasma generation 95

3.2.4 DSI home designed UHV sputtering machine 96

3.3 Vibrating sample magnetometer (VSM) 97

3.4 Alternating gradient field magnetometer (AGFM) 99

3.5 Hysteresis loop 100

3.6 Thermal stability factor (SF) measurement 102

3.7 X-ray diffraction 103

3.8 Transmission electron microscopy (TEM) 105

4 Analytical model of S-W Particle 110

4.1 Introduction 110

4.2 Stoner-Wohlfarth Model: Uniform Rotation 111

4.3 The switching field hk 113

4.4 Hysteresis loop 115

4.5 Energy barrier 118

4.6 Summary 121

5 Theoretical analysis on exchange coupled media 123

5.1 Introduction 123

5.3.1 Case I: No conditional local minimum energy state exists 129

5.3.2 Case II: Conditional local minimum energy states exist 130

5.4 Results and discussion in the case with external field applied in the direction of the easy axis 135

5.4.1 Hysteresis loop: 135

5.4.2 Coercivity and switching field calculation and exchange bias 140

5.4.3 Energy barrier calculation 141

5.5 Results and discussion in the case with the field applied with an angle of the easy axis 142

5.5.1 Hysteresis loop 142

5.5.2 Switching field 143

5.6 Special issue: estimation of J considering different easy axis directions 147 5.6.1 Analytical Model 148

5.6.2 Results and discussion 149

5.6.3 θ =0 case 149

5.6.4 θ ≠0 case 153

5.7 Results and discussion in the case with strong exchange coupling 157

5.8 Summary 158

6 Effects of Dy-doping on magnetic and reversal properties of recording media 161

6.1 Introduction 161

6.2 Experimental 161

6.3 Results and discussion 162

6.3.1 Effect of doping Dy into the Cr underlayer 162

6.4 Summary 166

7 Effects of perpendicular exchange coupling on the magnetic properties of CoCrPtB-CoDy bilayers 168

7.1 Introduction 1687.2 Literature review of amorphous rare earth-transition metal alloy 168

7.2.1 Magnetization for homogeneous A (RE) – B (TM) alloys [1,2] 1697.2.2 The compensate point and magnetization curve change with the

temperature 1727.2.3 Curie temperature change and its effect on composition 1747.2.4 Perpendicular anisotropy dependence on the Co content of CoDy

amorphous film 1747.3 Experimental 1757.4 Exchange coupling effect of a CoDy toplayer to a perpendicular CoCrPtB

magnetic layer 1787.5 The exchange bias field (Hex) and interface domain wall energy (σ)

dependence on the magnetic properties (Hc) of the CoDy layer 1877.6 Magnetic behaviors of exchange coupled CoDy/CoCrPtB bilayers

dependence on the Co content of CoDy layer 1887.7 Magnetic behaviors of exchange-coupled CoDy/CoCrPtB bilayers’

dependence on the thickness of a CoDy layer 2097.8 Interface domain wall energy estimation by the different magnetization

reversals 2247.9 Training effect of the magnetization and exchange bias for the

CoDy/CoCrPtB coupled bilayers 227

8 Conclusion 235

Thermal instability due to the superparamagnetism is one of key concerns for future magnetic recording media Indirect exchange coupling through an interlayer has been successfully used in recently proposed and commercialized antiferromagnetically coupled medium (AFC) to solve the thermal stability issue for longitudinal recording media This study proposed and investigated for the first time a novel exchange-coupled bilayers structure that used an antiferromagnetic layer to directly exchange-couple with a magnetic layer (recording layer), which was not only meaningful for longitudinal media but also important for future perpendicular media and/or heat assisted magnetic recording media A theoretical model for the proposed exchange-coupled bilayers structure was built up for the first time Magnetization switching behaviors of such medium structure were calculated based on this model The theoretical calculation was well consistent with the experimental results To implement this new medium structure, an antiferromagnetic CoDy layer was chosen and sputtered

on top of a CoCrPtB magnetic layer with perpendicular anisotropy The exchange coupling in such medium structure was shown to be much higher than that in the AFC medium The composition and thickness effects of CoDy layer in such medium structure were investigated systematically The exchange coupling effect between CoDy layer and CoCrPtB layer was made clear by analyzing the shape of major and minor hysteresis loops, DC demagnetizing remanence curves (DCD curves), and switching field distribution curves It was found that the magnetic properties of the CoCrPtB layer were changed greatly through the exchange coupling with the CoDy layer The improvement of coercivity (Hc) and thermal stability factor (SF) was mainly due to the exchange coupling between the two layers and was much dependent on the

found for the exchange-coupled bilayers system, which was a new fundamental finding

on the exchange-coupled bilayers structure In summary this study has for the first time proposed, implemented experimentally and demonstrated successfully a new magnetic bilayers structure consisted of an antiferromagnetic layer (CoDy in this study) and a ferromagnetic layer (perpendicular CoCrPtB in this study) through a direct exchange coupling This novel structure showed an improved thermal stability and will be one of candidates for future extremely high areal density recording media

Figure 1.1 The rate of evolution of rigid disk technology represented as a graph of

areal density versus production year [4]

Figure 1.2 Comparison of HDD and DRAM recording areal density improvement

over the years [4]

Figure 1.3 Schematic diagrams showing the basic principle of the magnetic

recording system: (a) Longitudinal magnetic recording, (b) Perpendicular recording using a probe head and a soft underlayer in the medium, and (c) Perpendicular recording using a ring head and no soft underlayer [5]

Figure 1.4 Structure of typical thin film medium of current longitudinal medium Figure 1.5 Schematic representation of the evolution of actual bit size in terms of

areal density The bit width is given in kilobits per square inch (kbpi), and the bit length is given in terms of tracks per inch (tpi)

Figure 1.6 Structure of magnetic layer of IBM and DSI proposed medium

Figure 1.7 Structure of magnetic layer of Fujitsu designed medium

Figure 1.8 Structure of magnetic layer of the designed medium in this work

Figure 2.1 Function form of RKKY interaction

Figure 2.2 Crystal and magnetic structure of MnO

Figure 2.3 The p-orbit of the O2-ion through which exchange interaction acts

between the spins on magnetic ions M1and M2 Figure 2.4 (a) Hysteresis loops at 77 K of partially oxidized Co particles Curve (1)

shows the resulting loop after cooling the compact in a 10 kOe field Curve (2) shows the loop when cooled in zero field (b) Torque curves

on partially oxidized Co particles cooled in a field to 77 K, where h is the angle between the cooling field axis and the direction of the measuring field Curves a and b in (b) are for counterclockwise and clockwise rotations, respectively Refs [1,2]

Figure 2.5 Schematic of an ideal FM/AFM interface The FM and AFM layers are

single crystal and epitaxial with an atomically smooth interface The interfacial AFM spin plane is fully uncompensated For this ideal interface, the calculated value of the full interfacial energy density is about two orders of magnitude larger than the experimentally observed values

figure The X marks identify the frustrated exchange bonds (i.e., the interfacial spins that are coupled antiferromagnetically) The interfacial region can have a high degree of stress since metals and oxides often have different lattice parameters Dislocations (represented by the dashed line) can form during film growth to relieve the stress

Figure 2.7 Hysteresis loop for NiFe/NiO with G =15 Oe/mm using AGFM

Figure 2.8 (a) Torque magnetization, Γ, and (b) rotational hysteresis, WR, for an

oxidized Co film at T = 77 K after field cooling [64]

Figure 2.9 Perpendicular FMR spectrum of trilayer Si (100) / WTi (6.7 nm) / NiFe

(30.5 nm) / FeMn (13.6 nm) / NiFe (10.1 nm)/WTi(6.7 nm), The absorption fields and line widths values are, respectively, 13130 and 47

Oe for the main uniform mode a and 12 709 and 79 Oe for the second mode b [65]

Figure 2.10 Parallel FMR spectra of trilayer Si (100) / WTi (6.7 nm) / NiFe (30.5

nm) / FeMn (13.6 nm) / NiFe(10.1 nm)/WTi(6.7 nm), In-plane applied static field (a) parallel, (b) perpendicular, and (c) antiparallel to the direction of the exchange bias field

Figure 2.11 Neutron diffraction patterns of CoAl0.1Fe1.9O4 at various temperatures

[71]

Figure 2.12 (a) Schematic diagram of a spin valve device (b) Hysteresis loop, m(H),

and (c) magnetoresistance, ∆R/R(H), of Fe20Ni80/Cu/Fe20Ni80/FeMn GMR spin valve at room temperature[80]

Figure 2.13 Mössbauer spectra of CoAl0.2Fe1.8O4 at various temperatures.[71]

Figure 2.14 The dependence of exchange bias and coercivity on the thickness of

Ni80Fe20 for a Ni80Fe20/Fe50Mn50 bilayer [126]

Figure 2.15 Dependence of exchange bias HE (square symbols) and coercivity Hc

(triangular symbols) with the AFM layer thickness for Fe80Ni20/FeMn at

a fixed tFM = 7 nm [42] Note that 80 A/m =1 Oe

Figure 2.16 The values of exchange field HE measured at 80K and 250 K for 300 Å

NiFe/x Å CoO/300 Å Cu as a function of the CoO layer thickness The dashed line for the data at 80 K is 1/tAF

Figure 2.17 Schematic spin diagram for a (a) compensated and (b) uncompensated

AFM surface

Figure 2.18 Schematic diagrams of spin–spin coupling states at the interface of: (a)

FI/FI and (b) AFM/FM bilayers [128]

Notice that the wavelength of the roughness also varies from 18 to 39

nm with increasing rms roughness.[128]

Figure 2.20 Blocking temperature TB (solid squares) of 300 Å NiFe/x Å CoO/ 300

Å Cu and Néel temperature TN (open circle) of CoO/SiO2 multilayers,

as a function of the CoO layer thickness. [129]

Figure 2.21 (a) Representative hysteresis loops of training effect of mixed FexO and

α-Fe2O3 (50 nm)/NiFe (5 nm) film M–H loop cycles 1, 2, 4, 8, 12, 20 are shown (b) H , H sw+ c ,Heb , H sw− plotted as a function of the number

of the loop cycles (c) Representative hysteresis loops of training effect

of mixed FexO and FexO and α-Fe2O3 (70 nm)/CoFe (5 nm) film M–H loop cycles 1, 2, 4, 8, 12, 20 are shown (d) H , H sw+ c ,Heb , H sw− plotted

as a function of the number of the loop cycles [130]

Figure 2.22 Schematic diagram of angles involved in an exchange bias system Note

that the AFM and FM anisotropy axes are assumed collinear and that the AFM sublattice magnetization MAFM has two opposite directions Figure 2.23 AF rough interface with frustrated interactions marked by full dots The

dashed line marks the boundary between the F and the AF layer

Figure 2.24 Illustration of the perpendicular F and AF magnetic interface

configuration, with spin canting in the first AF layer

Figure 2.25 Exchange bias field (Heb) and coercivity (Hc) vs top Co thickness

Figure 2.26 Exchange bias field (Heb) and coercivity (Hc) vs the number of bilayers Figure 2.27 (a) Dependence of the exchange bias field on the IrMn layer thickness

for [Pt(2 nm)/Co(0.5 nm)]3 / IrMn (t nm) /Pt (2 nm) multilayers on arrays with different particle sizes and flat SiO2 substrates (b) Exchange bias field as a function of the inverse IrMn layer thickness for films on plain SiO2 substrates The solid line is a linear fit to the data

[165]

Figure 2.28 Dependence of the exchange bias field on the inverse particle diameter

for [Pt (2 nm)/ Co (5 nm)]3 / IrMn (t nm) / Pt (2 nm) caps with different IrMn layer thickness t The solid line is a linear fit to the data for t ≥10

nm. [165]

Figure 3.1 Schematic diagram of a DC-sputtering system

Figure 3.2 (a) Structure of the glow discharge of a DC diode system (b) The charge

particle concentration of a glow discharge (c) The voltage variation of the glow discharge

Figure 3.4 The species arriving at and interacting with the substrate in a sputtering

system

Figure 3.5 Cross section of circular magnetron sputter source, with permanent bar

magnets behind the target, and depicting magnetic field, race track, and electron orbits

Figure 3.6 Schematic drawing of a multi-function UHV sputtering system (DSI

home-designed)

Figure 3.7 Schematic diagram of a Vibrating Sample Magnetometer (VSM) with

computer control

Figure 3.8 (a) Overall system of an Alternate Gradient Field Magnetometer

(AGFM) configuration (b) The structure of bimorph, extension and sample

Figure 3.9 Princeton MicroMagTM AGFM system

Figure 3.10 Typical hysteresis loop and parameters measured by VSM

Figure 3.11 Time-dependence measurement for obtaining thermal stability factor,

KuV*/kBT

Figure 3.12 The lattice plane formed in the atoms and the basics of Bragg's law Figure 3.13 The XRD data of series films glass/Ti/CoCrPtB/Ti deposited at

different temperatures with Ti (002) peak and Co (002) peak

Figure 3.14 The schematic diagram of a transmission electron microscope (TEM) Figure 4.1 Diagram of a single-domain particle with uniaxial anisotropy K and

applied field H

Figure 4.2 Normalized energy vs the magnetization direction α with different

applied field The external field applied along the easy axis

Figure 4.3 The normalized switching field hk dependence on the external applied

field direction angle θ

Figure 4.4 The hysteresis loop of S-W particle with the different applied field

angle θ

Figure 4.5 Normalized coercivity of S-W model particle dependence on the

external applied field direction angle θ

Figure 4.6 Normalized energy barrier e vs the external applied field h The

external field is applied along the direction of the easy axis

applied field and that of the easy axis

Figure 5.1 (a) Schematic of antiferromagnetically coupled medium (AFC); (b)

Configuration of magnetization M1, M2, and applied H-field with

respect to the energy easy axis

Figure 5.2 Hysteresis loop for the case with high exchange coupling constant J

The external field applied along the easy axis

Figure 5.3 Energy change with the magnetization angle α and β in case I No

external field was applied on the AFC system

Figure 5.4 Hysteresis loop for the case with lower exchange coupling constant J

The external field is applied along the easy axis

Figure 5.5 Energy change with the magnetization angle α and β in case II No

external field was applied on the system

Figure 5.6 Energy barriers (P3->P1, P3->P2) of an antiferromagnetically coupled

medium with initial parallel configuration

Figure 5.7 Energy barriers of bottom layer switching (P3->P1) relation with the

exchange bias j

Figure 5.8 Hysteresis loop of the antiferromagnetically coupled system The

external field applied along the easy axis, fk = fm =0.2, j = 0.01

Figure 5.9 Hysteresis loop of the antiferromagnetically coupled system The

external field applied along the easy axis, fk = fm =0.2, j = 0.1

Figure 5.10 Hysteresis loop of the antiferromagnetically coupled system The

external field applied along the easy axis, fk = fm =0.2, j = 0.4

Figure 5.11 Normalized energy barriers changes with the external applied field h

The external field is applied on the easy axis

Figure 5.12 A typical hysteresis loop for a magnetically exchange coupled system

In this calculation, K1 = K2 = 1×106 erg/cm3, M1 = M2 = 350 emu/cm3, t1

= 15 nm, t2 = 5 nm

Figure 5.13 Calculated hysteresis loop for magnetically exchange coupled system

with low exchange coupling constant Solid line is the major loop for

M1 layer Dot line is the minor loop for M2 layer

Figure 5.14 Top-layer (M1) switching field (h1n) dependence on the external applied

field angle θ, hn is switching field of single layer, h1n-hn is the increment

of switching field due to the effect of antiferromagnetic coupling

Figure5.15 The dependence of the increment of switching field (h1n-hn) of top layer

increment of switching field due to the effect of antiferromagnetic coupling

Figure 5.17 The dependence of the increment of switching field (h2n-hn) of bottom

layer (M2) on the antiferromagnetic coupling constant J

Figure 5.18 Normalized h ex dependence on the normalized exchange coupling

constant j (a) in the case that fm=0.3; (b)in the case that fm=0 6

Figure 5.19 Normalized exchange bias hex dependence on the external applied field

angle (fk=1/3 fm=1/3) (a) in the case that j =0.033; (b) in the case that j

=0.167

Figure 5.20 a dependence on the orientation ratio (OR) of Mr fk=1/3 fm=1/3

Figure 6.1 Dependence of Hc, SF on the Dy content in CrDy underlayer

Figure 6.2 XRD diffraction spectra of the films with different Dy content in CrDy

underlayer

Figure 6.3 Dependence of Hc, SF on the Dy content in CrDy toplayer

Figure 6.4 XRD diffraction spectra of the films with different Dy content of CrDy

toplayer

Figure 6.5 Hysteresis loop for the medium with CrDy toplayer (10% Dy)

Figure 7.1 The Co concentration dependence of magnetization for Co-Dy alloys,

the total magnetizationσ , Dy-subnetwork magnetizationσDy, and subnetwork magnetization σCo

Co-Figure 7.2 Temperature dependence of the saturation magnetization for amorphous

Dy-Co alloys The dashed lines were calculated from the mean-field theory

Figure 7.3 Thermal-magnetic curves of Co51Dy49 with different thickness in an

applied field of 100 Oe [4]

Figure 7.4 Compositional variation of the compensation temperature for

amorphous RE-TM alloys with RE = Gd, Tb, Dy, Ho for TM = Co Figure 7.5 Compositional variation of the Curie temperature for amorphous RE-Co

alloys with RE = Gd, Tb, Dy, Ho, Here Dy-Co (circles)

Figure 7.6 The anisotropy of CoDy film dependence on Co content of CoDy

Figure 7.7 The magnetic properties of samples deposited with different

temperatures

Figure 7.9 Hc and thermal stability factor (SF) dependence on the Co content of

CoDy layer

Figure 7.10 Switching field (Hs) dependence on the Co content of CoDy layer

Figure 7.11 The saturated magnetization dependence on the Co content of CoDy

Figure 7.15 The exchange coupling strength (erg/cm2) dependence on the Co

content of CoDy layers

Figure 7.16 The magnetization and coercivity of minor loop dependence on the Co

content of CoDy layers

Figure 7.17 The exchange bias Hex and the exchange coupling energy dependence

on the coercivity of CoDy layer with different Co content

Figure 7.18 Hysteresis loop of the sample with the structure of glass / Ti (40 nm) /

CoCrPtB (35 nm) / Ti (4 nm)

Figure 7.19 Hysteresis loop of the sample with the structure of glass / Ti (40 nm) /

CoCrPtB (35 nm) /Co81Dy19 (20 nm) / Ti (4 nm) toplayer

Figure 7.20 Hysteresis loop of the sample with the structure of glass / Ti (40 nm) /

Figure 7.27 The switching field distribution of the sample with the structure of glass

Figure 7.44 The DCD curve of the sample with the structure of glass / Ti (40 nm) /

Figure 7.53 The exchange coupling strength J estimated by the minor loop

Figure 7.54 The exchange coupling strength estimation by the DCD curve

Figure 7.55 The hysteresis loop dependence on the field cycle number for sample

310 The number of hysteresis loop goes from 1 to 50

Figure 7.56 The training effect of the coercivity of CoDy layer of minor hysteresis

loop for sample 310

Figure 7.57 The training effect of the exchange bias field of CoDy layer of minor

hysteresis loop for sample 310

Figure 7.58 The training effect of Hc of the hysteresis loop of CoDy layer: the left

and right switching field of CoDy layer of minor hysteresis loop for sample 310

Figure 7.59 The training effect of the magnetization of the sample 310

Table 2.1 Summary of exchange bias and related properties for different small

particle systems Note that the given loop shifts, HE, are at T = 4–10 K Table 5.1 The magnetic parameters (h11, h12, h21) of exchange coupled system with

different exchange coupling constant j, the fk and fm are chosen as 0.2 Table 5.2 The comparison of energy barrier between experimental and analytical

work The magnetic parameters for this table are, fk = 0.5, hex = 0.036, fm = 0.5

Table 7.1 The experimental condition and structure for sputtering CoCrPtB layers t

is the sputtering temperature of substrate and is set to 200 ◦C, 250 ◦C and

300 ◦C

Table 7.2 The magnetic properties of samples for optimizing the CoCrPtB layer

sputtering temperature

Table 7.3 The film structure for samples to study the CoDy coupling effect with

different Co content of CoDy layer

Table 7.4 Magnetic properties of samples with different Co content of CoDy layer Table 7.5 Magnetic properties of samples with different Co content of CoDy layer Table 7.6 The exchange bias Hex and interface wall energy σ dependent on the

coercivity of CoDy layer with different Co content σ =2H ex M CoDy t

1 Introduction

1.1 Overview of magnetic recording industry

As the field of information technology rapidly advances, the need for larger storing capacity accordingly increases Among data storage methods, magnetic storage is the most economical one The hard disk drive provides nearly half of all computer storage This includes the traditional desktop computer, which usually uses a 95 mm disk-drive containing one or more disks, where the substrate generally used is AlMg alloy and glass coated with a number of underlayers and seedlayers to produce the required magnetic properties on the resulting disk Lap-top computers usually contain a much smaller disk-drive (65 mm or 45mm), where the substrate material is usually glass or glass ceramic High-end servers for electronic mail applications require large capacity, high data-rate disk-drives to cope with the increasing amount of information exchanged via this medium

Since the first magnetic disk drive (RAMAC 350), which had an areal density of 2 Kbits/in2,was invented at IBM in 1955, the technology of magnetic recording has advanced very rapidly The areal density for information storage on magnetic media, especially hard disks, has been increasing at an astonishing rate over the last three decades The oft-quoted density growth rates are 30% per year for 1970-1990 and 60% per year since 1990 [1] Significant improvement came in 1992 with the introduction

of smoother sputter-deposited thin film media to replace the binder-based particulate media, the magneto resistive (MR) head and giant magneto resistive (GMR) head playback transducers After demonstration at 20 Gbits/in2 in 1999, the areal densities achieved in products have grown at a rate approaching 100% per year due to the

introduction of giant magnetoresistive (GMR) spin-valve heads by IBM in 1998 [2] The evolution of hard disk storage in terms of areal density is even faster than that of semiconductor DRAM memory Areal densities of over 100 Gbits/in2 have now been demonstrated [3] Figure 1.1 shows areal density improvement of hard disk over the years, while Figure 1.2 compares the storage capacities of hard disk and DRAM memory The areal density improvement of hard disks is greater than that of the DRAM The principle of information storage in hard disk is magnetic recording My study focuses on the magnetic materials used for magnetic recording Brief magnetic recording and disk structure are explained in the subsequent paragraphs

Figure 1.1 The rate of evolution of rigid disk technology represented as a graph of areal density versus production year [4].

Figure 1.2 Comparison of HDD and DRAM recording areal density improvement over the years [4]

1.2 Brief overview of magnetic recording

Magnetic recording is a process in which data is stored in a magnetic medium by a magnetic recording head (write head) The recording medium consists of a substrate coated with a magnetic layer that can be permanently magnetized, thus permitting information to be stored magnetically The write head is fed with a current containing the information to be recorded Upon moving the write head at a constant speed relative to the medium, the fringing fields from the head gap permanently magnetize the magnetic layer of the medium, and the information is stored At replay, the medium

is again moved past a read head (which can be the same as the writing head or can be a different part), and the flux emanating from the medium and entering the read head gives rise to a read-back signal Today, the magnetic states on the medium are sensed

either by an inductive head or by a flux-sensitive head (magneto resistive head and giant magneto resistive) The output voltage developed in an inductive head is proportional to the rate of change of the flux induced in the core Flux-sensitive heads are sensitive to the magnetic flux induced within them, rather than to its rate of change (the magnetoresistive head is the most widely studied of the flux-sensitive heads) Using recording heads that can establish well-defined directions of magnetization in smaller regions of a magnetic layer is one step towards packing more data bits into a recording medium

Figure 1.3 Schematic diagrams showing the basic principle of the magnetic recording system: (a) Longitudinal magnetic recording, (b) Perpendicular recording using a probe head and a soft underlayer in the medium, and (c) Perpendicular recording using

a ring head and no soft underlayer [5]

The two basic recording modes are longitudinal and perpendicular In the longitudinal recording mode, the magnetic layer of the medium is magnetized along the film plane;

magnetized normal to the film plane The basic principle of the magnetic recording system is shown in Figure 1.3

The longitutinal medium was widely used as the current commercial hard disk because

it was efficiently produced and had well controlled magnetic properties However, the superparamatic limitation has been reached, as the recording bit has shrunk to several hundred nanometers Perpendicular media have better thermal stablity than longtitudinal media because of less demagentic field, so they are their most promising replacements

With the development of magnetic recording, another recording mode was proposed: tilted recording [6-10] In the titled recording mode, the magnetic layer is magnetized tiltled to the film plane It has the advantage of lower requirement for the write head, higher orientation, and better thermal stability compared to the other two recording modes

The magnetic medium that stores information is very important if the hard disk is to accommodate more data Nowadays, multilayer thin film media have replaced particulate media as the dominant rigid disk technology, due to high coercivity, large magnetization, and smooth surfaces, which are basic requirements for high-density recording

1.3 Thin film media

As illustrated in Fig 1.4, a typical thin film medium for recording is composed of substrate, seedlayer, underlayer, intermediate layer, magnetic layer, overcoat, and lubricant

Figure 1.4 Structure of typical thin film medium of current longitudinal medium

1.3.1 Substrate

Substrates are the foundation on which all rigid magnetic recording disks are constructed The important properties for a substrate are high hardness and low density for shock resistance Conventionally, for longitudinal media, mechanically textured Al-Mg substrates are used, and NiP amorphous layers are electrolessly plated on the surface to increase hardness Glass substrate, an attractive alternative to Al-Mg substrates, has also been widely used in small size disk (27 mm to 65 mm), due to its increased rigidity, impact resistance, and lighter weight

1.3.2 Underlayer and seedlayer

The function of the underlayer is to control the crystallographic texture, grain size, and morphology of the magnetic layer Some of the underlayer properties will be inherited

by the magnetic layers For longitudinal recording media, bcc (body centered cubic) structure Cr-based materials are used as an underlayer, due to an epitaxial match of the

Cr planes to preferred ones in the Co-based alloys in the magnetic layer Several papers discuss the methods of using the underlayer control to improve the magnetic properties of the magnetic layer [11-13] For perpendicular recording media, the soft

Intermediate Layer

8 nm 1-2 nm

Substrate Seedlayer Underlayer Magnetic Layer 1

Overcoat

Magnetic Layer 2

Antiferro Interlayer

process, the soft underlayer(SUL) needs to guide the magnetic flux from the write pole

to the collector pole with low reluctance Therefore, materials of choice have high permeability, high saturation magnetization, and low coercivity Even though high saturation materials are used, the thickness of the SUL ranges between 100 and 400 nm This thickness is much larger than that of layers in current longitudinal recording media and poses a considerable challenge to deposition tools In addition, surface roughness of the SUL, which tends to increase with layer thickness, needs to be kept small to allow the head to fly close to the recording medium The total spacing between the write pole and SUL must be minimized for optimal write field efficiency Magnetic domains must be avoided in the SUL, as their presence leads to spike noise

in the readback signal [14] Candidate materials for SULs are soft magnetic materials, such as NiFe, CoNbB, FeAlSi, CoFeB, FeTaN, FeTaC and CoFe Many of these materials have domains in their as-deposited state

The seedlayer is used to induce certain crystallographic texture of the underlayer/magnetic layer or to reduce grain size, thus improving the magnetic properties of the thin film media Currently, NiAl material is used as seedlayer because

it introduces small grain size of the magnetic layer

1.3.3 Intermediate layer

The intermediate layer may have significant influence on both the crystallographic texture and magnetic properties of the magnetic layer Magnetic performance can be significantly improved by inserting a very thin intermediate layer between the magnetic layer and Cr-alloy underlayer This improvement is due to improved lattice matching, promoting Cr inter-diffusion Currently about a CoCrPt layer of 2 ~ 3 nm is

used as an intermediate layer For perpendicular media, a thin non-magnetic interlayer

is used to break up the exchange coupling between the magnetic layer and soft underlayer

1.3.4 Magnetic layer

The magnetic layer is the most important layer on which information is stored The material for this layer has several requirements, including high magnetocrystalline anisotropy to provide high coercivity and thermal stability, as well as low remnant moment-thickness (Mrt) to provide sharp transition to get high areal density Currently, the material for this layer is cobalt-based metal alloys CoCrPtB because of their high uniaxial magnetocrystalline anisotropies Adopting Cr in Co based alloy can help segregate Co-rich grains from Cr-rich boundaries This segregation reduces both the exchange-coupling between grains and the transition noise, which is the dominant noise source of thin film media The addition of Pt has been shown to increase magnetocrystalline anisotropy [15,16], whereas most of the elements added to decrease noise tend to decrease the magnetocrystalline anisotropy [15,17] The addition of B has been shown to increase coercivity when added to Co-alloys containing Pt [18, 19] The solubility of B in Co is very low, but the larger substitutional Pt atom creates an enlarged interstitial site that can be occupied by B The lattice expansion observed with the addition of B to Co-alloys containing Pt, as well as the maximum coercivity for equal atomic concentrations of Pt and B, suggests that B occupies the octahedral interstices in the lattice The addition of B to CoCrPt alloys decreases the stacking fault density The appearance of large numbers of stacking faults in the Co-alloys has been correlated to low coercivity It is believed that the presence of stacking faults and extended fcc regions in the hcp structure can lower the magnetocrystalline anisotropy

The mechanism for increased coercivity with the addition of B could be related to the improved crystalline perfection of the lattice, as well as to the electronic effects associated with the expanded lattice

For perpendicular media, the following material was extensively studied: (1) CoCrPt based granular medium with various non magnetic boundary elements such as B, SiO2, etc; 2) Co/Pd and Co/Pt multilayers; 3) multilayers with high thermal stability and lack

of exchange, decoupled in the film plane, or granular medium with a lack of thermal stability but well-decoupled between the grains Therefore, the granular-continuous coupled medium was introduced It combined the high thermal stability of the multilayer medium with the low–noise properties of the granular perpendicular medium [20]

1.3.5 Overcoat and lubricant

The primary function of the disk overcoat is to prevent wear of the magnetic layer and subsequent data loss, especially during intermittent start/stop contact with the ceramic The natural candidates for overcoat would be hard transition-metal oxides, carbides, and nitrides The disk overcoat is also a support for lubricant Lubricant is the final layer in the finished thin-film disk Its function is very similar to that of the overcoat, which is to protect the disk from friction and wear It brings additional stability to the head-disk interface by providing the recording heads with a smooth transition from a region of dragging to flying and by absorbing some of the energy generated by intermittent contact between the head and disk during flying [21] The typical material for rigid disk lubricant is perfluoropolyether organic polymer Texture, overcoat, and lubricant are the three primary parameters affecting disk durability

1.4 Key performance indices for current medium

There are many indices to guide the design of the thin film medium For my study, Some of them are important and will be discussed in the following section Mainly three factors are used to evaluate the performances of current medium

1 Areal density, which is required for current information technology development The development of the density of the medium has been shown above

2 Signal to noise ratio, which is required for the head to pick up the correct information from the medium As the areal density of the magnetic medium is increased, the bit size has to be decreased; thus, the signal that the reading head picks from the bit is also decreased To identify the information correctly, the signal to noise ratio index is used

3 Thermal stability factor, which is required by the storage duration Current medium is still facing the challenges of simultaneously achieving high thermal stability with high signal amplitude and high SNR for low bit-error-rate

Figure 1.5 Schematic representation of the evolution of actual bit size in terms of areal density The bit width is given in terms of kilobits per inch (kbpi), and the bit length is given in terms of tracks per inch (tpi)

1.4.1 Areal density

The areal density of the medium has increased very rapidly from 2k bit/in2 to 100G bit/in2 The dramatic increase of the density sharply decreases the bit size of the medium Figure 1.5 shows the bit size changes for different media The transition

length (a) between two bits is introduced to determine the areal density a-parameter

value is determined by the spatial variation of the (normalized) head field within the

recording medium (head field gradient),

dx

dH H

d Q

CR

2/

δ

+

response of the medium, which depends on the head, the media parameters, and the head-media spacing The so-called Williams-Comstock transition width parameter [22]

can be formulated as a function of head and media parameters as follows [23], assuming a continuous variation of the magnetization and demagnetization fields and a finite switching field distribution (SFD):

M Q

d S Q

δπ

0.3 µm 0.015 µm

Here Q: normalized head field gradient (≅ 0.8);δ : thickness of the magnetic layer in

nm; M r : remnant magnetization in A/m [(kA/m) = (emu/cm3)]; M rδ : areal moment

density in mA[(mA) =(memu/cm2) × 10)]; H CR: write coercivity during recording in A/m [(A/m) = (Oe) × 103/4π )]; d: physical spacing between pole tip and magnetic layer, in nm; I: transition parameter (1 for arctan, 0.691 for tanh and 2/ π for erf transition); and S*: loop squareness defined by dM dH loop H c =M r /(H c(1−S*) ) and

(S* = 1- SFD ≅ 0.75 to 0.95)

Higher linear densities have been achieved by reducing the head-media spacing (y

height) d and the magnetic spacing d+δ/2, and by lowering the ratio of areal moment density M rδ and remnant coercivity H CR (demagnetization ratio M rδ H CR ) This ratio M rδ H CR has been reduced from about 800 nm to about 10 to 15 nm, almost

two orders of magnitude, since the introduction of thin film disk media

1.4.2 Signal to noise ratio

There are two main noise mechanisms relevant to current and future media developments: transition noise and dc-noise Dc-noise is a direct consequence of the granularity of the media It produces a variance of the magnetization and signal amplitude via grain diameter and grain orientation variations It is the dominant noise source in particulate systems Transition noise is at a minimum when the magnetic regions are composed of small, well isolated magnetic grains However, this kind of noise is the main noise source for thin film media The onset of supra-linear noise at high linear densities is due to large correlated magnetic clusters or zig-zag regions caused by exchange and magnetostatic interactions The onset of this noise can be delayed to higher linear recording densities by improving grain isolation and reducing

r

M It is generally necessary to discuss the recording properties in terms of

correlated magnetic cluster sizes or correlation lengths x, rather than physical grain sizes D, because interactions are important We distinguish between downtrack and crosstrack correlation lengths, xdown and xcross, respectively These correlation lengths are further differentiated by their occurrence within the transitions (transition noise regime, tr) or outside the transition regions (dc-noise regime, dc), respectively

For transition noise-dominated systems, in the linear regime B ¸ B≥Bmin ≅π.athe following power-SNR relation has been derived by a number of authors [24,25,26]

W const

SNR= 0 is the ratio of the voltage pulse amplitude (S0) of an isolated

transition and the total integrated root means square (rms) transition noise (Nrms) Wread

is the read width, a is the Williams-Comstock transition width parameter, tr

M

M PW

W const

dc cross

dc cross

read dc

The above relations have been derived for granular, in-plane random anisotropy media,

in which the crystallographic grain orientation sets the anisotropy and, therefore, the magnetization axis Equations 1.2 and 1.3 may be expanded to include finite grain size distributions by adding a factor (1+σA) in the denominator in Equations 1.2 and 1.3, where σAis the geometrical width of a (log normal) grain area distribution A maximal SNR gain of ∆SNR=10log(≈1.8)≅2.5 dB arises if the grain area distribution can be

cannot be smaller than the (average) grain diameter D In the limit of negligible

interactions, assuming a fixed ratio of linear resolution to transition spacing, PW50/B,

scaling relations SNR( )dc ∝BW read / D 2 and SNR( )tr ∝B2W read / D 3 for dc and transition noise limited systems, respectively, have been proposed These relations emphasize that grain size scaling (reduction) is the best way to achieve higher areal density, AD∝1/BWread In transition noise dominated systems, it is also

advantageous to operate at reduced BAR, as revealed by Equation 1.2 A strong inverse

relationship of the areal density and the required SNR has been demonstrated From the media perspective, it is very important to control the microstructure, especially the grain size, the grain size distribution, and chemical isolation to break exchange and keep the media noise within acceptable bounds [27]

1.4.3 Thermal stability factor

To maintain both linear resolution and media SNR requirements, the grain size must be scaled down and eventually impose a limit on the achievable areal density because of the onset of thermal instabilities Especially for the high density medium, the magnetic energy of grain is too small to compete with the thermal energy Representing a recording medium by an assembly of independent switching units with reversal

barriers, E B, the probability of not switching (retaining the magnetization) is obtained from Boltzmann statistics and N´eel-Arrhenius switching rates:

rt

e

T k

gives the switching rate or inverse time constant, which, via the thermal attempt

frequency f0, is connected to the switching barrier E B The attempt frequency depends

on the details of the switching potential and is typically taken to be f0 =109 Hz, k B = 1.3807 ×10-20 J/K, (1.3807 × 10-16 erg/K, 0.8619 × 10-4 eV/K,) which is Boltzmann’s constant, and T, which is the absolute temperature in K Solving Equation 1.5 for E B

yields an observation time-dependent stability ratio:

In typical laboratory experiments, for example, a time constant of τ 100 s yields =

ln( f0τ ) = 25.3 In magnetic data storage, time constants of τ =10 years yield ln( f0τ )

= 40 The latter sets a minimum requirement for stability

The switching barrier E B depends on the switching mechanism, which may involve non-uniform magnetization states and incoherent processes For systems with weak interactions and small grains, a simple Stoner-Wohlfarth model may be used The material is assumed to be uniformly magnetic and to possess uniaxial anisotropy If a magnetic field H is applied at angle θ to the anisotropy axis, the system energy density is given by the following equations:

αθ

u K

H

H V K

HM h

M

V K

2,

2

For quantitative estimates of the stability barrier in the presence of demagnetization

fields H d , the ratio |H d |/H0 is important as revealed by the following approximation:

( )

m d u

B

H

H V

Here the exponent m may be estimated from the so-called Pfeiffer approximation,

H0 = K the angle-dependent switching field (S-W asteroid) For aligned

cases, θ =0and 180o, one obtains H0 =H K , and the exponent in Equation 1.9

becomes m = 2 For a two-dimensional, random noninteracting particle system with

°

= 21

θ , one obtains H0 =0.567H K[F(21°)=0.567], and m≅3/2, respectively

H K is the anisotropy field defined as H K =2K u /M s In the above example, the ratio

04

thermal unstable I have addressed the problems in the following manner

1.5 Issues with and resolutions for further increasing the areal density

Increasing areal density of the medium will decrease its grain size The anisotropy energy related to the volume of grain will also decrease; ultimately, it will compete with the thermal energy The superparamagnetic phenomena will occur To decrease the grain size further while keeping the grain thermal stable, the exchange coupling between two films is utilized to provide external energy to help the grain stabilize New media structures have been provided that realize this idea The thermal stability

of them was improved The traditional structure and new structures are shown in the following sections

1.5.1 Traditional medium structure

This structure is shown in Figure 1.4 It consists of a substrate, an underlayer, a magnetic layer, a overcoat layer, and a lubricant layer Information is stored in the magnetic layer Other than the magnetic layer, the rest layers of new medium structure are the same as that of the traditional medium structure In the new medium structures, the magnetic layer consists of more layers to introduce external stable energy

1.5.2 IBM and DSI proposed medium (AFC and LAC)

In INTERMAG 2000, IBM proposed a medium structure to extend the thermal limitation of the current medium The magnetic layer of this medium structure is shown in Figure 1.6 It consists of three layers: two magnetic layers and one non-

magnetic spacing layer The top magnetic layer, or the recording layer, is the thicker layer; it has the same function as the magnetic layer in the traditional medium The bottom magnetic layer, or the stabilizing layer, is a thinner layer It has two functions: (1) to decrease the total Mrt and (2) to supply extra stable energy to the recording layer

by antiferromagnetic coupling with the recording layer when the system is in the free external applied field The spacing layer (currently use Ru), which is between the two magnetic layers, introduces antiferromagnetic coupling between the two magnetic layers

Figure 1.6 Structure of magnetic layer of IBM and DSI proposed medium

1.5.3 Fujitsu proposed medium (SFM)

This design introduces another set of spacing layer and stabilizing layer to increase the antiferromagnetic coupling effect and compensate the Mrt loss for stabilizing layer of AFC medium Therefore the Mrt of the system (Figure 1.7) kept unchanged For its latest 100Gbit/in2 demonstration [30], Fujitsu used the two layer structure and the SFM didn’t decrease the Mrt value

Figure 1.7 Structure of magnetic layer of Fujitsu designed medium (SFC)

1.5.4 Newly proposed medium in this work

It is well known that strong exchange coupling exists between the ferromagnetic material and antiferromagnetic layer This interface exchange coupling can be used as the external stable energy for the magnetic layer Based on this idea, we designed a