NANOMAGNETISM
SPINTR&NICS
Fabrication, Materials, Characterization
and Applications
Farzad Nasirpouri & Alain Nogaret w«w
Y p W o r l d Scientific
This page is intentionally left blank
NANOMAGNETISM
AND
SPINTRONICS
Fabrication, Materials, Characterization
and Applications
7281tp.weimei.2.10.ls.indd 1 2/24/10 4:01:58 PM
NANOMAGNETISM
AND
SPINTRONICS
Fabrication, Materials, Characterization
and Applications
Editors
Farzad Nasirpouri
Sahand University of Technology, Iran
Alain Nogaret
University of Bath, UK
World Scientific
NEW JERSEY • LONDON • SINGAPORE • BEIJING • SHANGHAI • HONG KONG • TA I P E I • CHENNAI
7281tp.weimei.2.10.ls.indd 2 2/24/10 4:01:59 PM
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library.
NANOMAGNETISM AND SPINTRONICS
Fabrication, Materials, Characterization and Applications
Copyright © 2011 by World Scientific Publishing Co. Pte. Ltd.
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
ISBN-13 978-981-4273-05-3
ISBN-10 981-4273-05-8
Printed in Singapore.
YHwa - Nanomagnetism and Spintronics.pmd 1 11/19/2010, 9:11 AM
PREFACE
The goal of spintronics is to manipulate individual magnetic moments to
integrate logic functions and non-volatile information storage on the
same platform. As is often the case in condensed matter science, advances
are made through the synthesis of novel materials, high quality materials
bringing new physics. Giant magnetoresistance and dilute magnetic
semiconductors are two examples at hand. The remarkable potential of
spintronics for quantum computation faces major challenges when it comes
to controlling simultaneously several qbits encoded in magnetic moments.
After giving a brief introduction to concepts in Nanomagnetism and
Spintronics, the present book reviews recent techniques and their
achievements in the synthesis and fabrication of magnetic nanostructures
in part two. The methods presented here emphasize bottom up or top
down approaches for nanodots, nanowires and thin films. They include
focused ion beam irradiation, electron beam induced chemical vapour
deposition, chemical, and electrochemical methods together. The third
part of the book entitled Materials and Characterisation reviews
magnetoelectric materials, the giant magnetoresistance in magnetic
superlattices, dynamics effects in spin transfer torque oscillators, dilute
magnetic oxides, rare earth nitrides together with nuclear resonance
scattering and Mössbauer spectroscopy in spintronics. Finally, the last part
of this book discusses applications to magnetic storage and bio-magnetism.
The book will be useful to graduate students for whom the
introductory chapter is intended, to researchers in the field of
nanoscience and engineers.
Spring 2009
Farzad Nasirpouri
(Sahand University of Technology, Iran)
Alain Nogaret
(University of Bath, UK)
v
This page is intentionally left blank
CONTENTS
Preface v
PART I. INTRODUCTION
1 Concepts in Nanomagnetism and Spintronics 1
Farzad Nasirpouri and Alain Nogaret
1.1. Nanoscale Science and Technology 1
1.2. Nanomagnetism 2
1.2.1. Magnetic ordering on the nanoscale 3
1.2.2. Magnetization reversal 5
1.2.3. Dimensionality in magnetism 7
1.2.3.1. Thin magnetic films 7
1.2.3.2. Nanowires or one dimensional
magnets 8
1.2.3.3. Nanodots and superparamagnetism 11
1.3. Spintronics 12
References 14
PART II. FABRICATION AND GROWTH
2 Artificial Magnetic Domain Structures Realised by
Focussed Ion Beam Irradiation 19
Simon Bending, Simon Crampin and Atif Aziz
2.1. Introduction 19
2.1.1. Controlling magnetic anisotropy by irradiation 21
2.1.2. Intrinsic domain wall resistivity 22
2.2. Fabrication of Artificial Domain Structures 24
vii
viii Contents
2.3. Magnetic Properties of Artificial Domain Structures 26
2.4. Angle-Dependent Domain Wall Resistivity
Measurements 33
2.5. Conclusions and Outlook 41
References 43
3 Fabrication of Magnetic Nanostructures by Electron
Beam Induced Deposition 45
Masaki Takeguchi and Masayuki Shimojo
3.1. Introduction 45
3.2. EBID Fabrication 47
3.3. Fabrication of Iron-Containing Nanostructures 48
3.4. Post-Deposition Heat Treatment: Fabrication of Alpha
Iron Nanostructures 54
3.5. EBID with Fe(CO)5 and Water Vapor: Fabrication of
Magnetite Nanostructures 57
3.6. Summary 59
References 59
4 Preparation of Magnetic Nanoparticles Using Chemical
Route and Functionalization for Medical Applications 63
Yuko Ichiyanagi
4.1. Introduction 63
4.2. Synthesis and Characterization of Magnetic
Nanoparticles 65
4.3. Magnetic Properties of 3d Metal Hydroxide and
Metal Oxide Nanoparticles 66
4.3.1. Magnetic properties of metal hydroxide
nanoparticles 66
4.3.2. Metal oxide nanoparticles 71
4.4. Pluralistic Ferrite Nanoparticles 73
4.4.1. Ni-Zn ferrite nanoparticles 73
4.4.2. Mg ferrite nanoparticles 78
4.5. Functionalization of Magnetic Nanoparticles 80
4.5.1. Amino-silane coupling 80
Contents ix
4.5.2. Development for cell selective magnetic
nanoparticles 81
4.6. Conclusions and Outlook 84
References 85
5 Electrodeposition as a Fabrication Method of Magnetic
Nanostructures 89
László Péter and Imre Bakonyi
5.1. Introduction 89
5.2. Electrodeposition: A General Overview 90
5.2.1. Definitions and major principles 90
5.2.2. Electrodeposition of magnetic elements 94
5.2.3. Electrodeposition of magnetic alloys 94
5.2.4. Non-metallic deposits obtained with
electrochemistry 96
5.3. Electrodeposition: A Route Toward Magnetic
Nanostructures 96
5.3.1. Electrodeposition of ultrathin magnetic films 96
5.3.2. Nanocrystalline magnetic deposits 99
5.3.3. Deposition of metastable precursor alloys and
their treatment for obtaining granular magnetic
alloys 101
5.3.4. Electrodeposition of magnetic/non-magnetic
multilayer films with nanometer-scale
periodicity 102
5.3.5. Deposition of nanostructures at preferred
nucleation sites 105
5.3.6. Electrodeposition into templates 107
5.3.7. Electrodeposition on surfaces modified by
self-assembly of colloids 110
5.3.8. Suspension plating with magnetic particles 112
5.3.9. Formation of suspended magnetic particles
by electrochemistry 113
5.4 Summary 113
References 115
x Contents
PART III. MATERIALS AND CHARACTERISATION
6 Magnetoelectric Materials for Spintronics 121
Faik Mikailzade
6.1. GMR and Spintronics 121
6.2. History and Invention of Magnetoelectricity 123
6.3. Linear Magnetoelectric Effect 124
6.4. Multiferroics 126
6.5. Magnetoelectric Composites 129
6.6. Conclusions and Outlook 132
References 133
7 GMR in Electrodeposited Superlattices 139
Gholamreza Nabiyouni
7.1. Introduction 139
7.2. Electrodeposition 142
7.3. Electrodeposition of Metals and Alloys 144
7.4. Electrodeposition of Multilayers and Superlattices 146
7.4.1. Dual bath electrodeposition 147
7.4.2. Single bath electrodeposition 148
7.4.3. Electrodeposition of metallic thin films onto
semiconductor substrates 151
7.5. Resistivity in Metals 153
7.6. Magnetoresistance 153
7.6.1. Ordinary magnetoresistance 154
7.6.2. Anisotropic magnetoresistance 154
7.7. Giant Magnetoresistance (GMR) 155
7.8. Oscillatory GMR in Superlattices 158
7.9. Research on GMR 161
7.10. Superparamagnetism Contribution to GMR in the
Electrodeposited Superlattices 164
7.11. General Remarks on Electrodeposited Superlattices 166
References 167
Contents xi
8 Introduction to Spin Transfer Torque 173
C. Baraduc, M. Chshiev and U. Ebels
8.1. Introduction 173
8.2. Spin Transfer Torque 174
8.3. A Microscopic Picture 176
8.4. Transverse Spin Transfer Torque 179
8.5. Magnetization Dynamics 183
8.5.1. Conservative dynamics 184
8.5.2. Damped dynamics 185
8.5.3. Spin transfer torque induced dynamics 186
8.5.3.1. Static states 186
8.5.3.2. Stability 187
8.5.3.3. Dynamic states 187
8.6. State Diagram 188
8.6.1. Planar polarizer 189
8.6.2. Perpendicular polarizer 190
8.7. Conclusions 190
References 191
9 Spintronics Potential of Rare-Earth Nitrides 193
Ben J. Ruck
9.1. Introduction 193
9.2. Rare-Earth Nitride Preparation 195
9.3. Electronic Structure 199
9.3.1. Band structure calculations 199
9.3.2. Experiment 205
9.4. Magnetic Properties 211
9.5. Device Prospects and Future Challenges 214
9.6. Conclusions 218
References 218
10 Dilute Magnetic Oxides: Current Status and Prospects 223
Karen Yates
10.1. Introduction 223
10.2. Impurities 225
xii Contents
10.2.1. Types of impurity in DMS systems 225
10.2.1.1. Extrinsic impurities 225
10.2.1.2. Clusters 226
10.2.1.3. Solubility 229
10.2.1.4. Spinels as secondary phases 231
10.2.1.5. Other secondary phases 233
10.3. Intrinsic Mechanisms for Magnetic Behaviour 235
10.3.1. Insulating regime 236
10.3.1.1. Theoretical treatments 237
10.3.1.2. Experimental results 238
10.3.1.3. “d0” ferromagnetism 240
10.3.2. Magnetism at high carrier concentrations 242
10.3.2.1. Theoretical review 243
10.3.2.2. Experimental results 244
10.4. Devices Already Made with DMS DMO and DMD
Materials 250
10.5. Outlook 253
References 254
11 Mössbauer Spectroscopy and Its Applications in
Spintronics 267
Saeed Kamali
11.1. Introduction 267
11.2. Mössbauer Spectroscopy: The Basics 268
11.2.1. Electric monopole interaction 270
11.2.1.1. Isomer shift 270
11.2.1.2. Second order Doppler shift 271
11.2.1.3. Centroid shift 271
11.2.2. Electrical quadrupole interaction 272
11.2.3. Magnetic hyperfine interaction 273
11.2.4. Combined electric and magnetic hyperfine
interaction 275
11.2.5. Transmission vs. conversion electron
Mössbauer spectroscopy 275
11.2.6. Relative intensities of resonance lines 276
Contents xiii
11.3. Superlattices, Thin Films 277
11.3.1. Fe/Co superlattices 278
11.3.1.1. Magnetic hyperfine field 278
11.3.1.2. Magnetic anisotropy energy 280
11.3.2. Fe/Cr 280
11.3.3. Fe/V superlattices 284
11.3.4. Exchange spring magnets 287
References 292
12 Nuclear Resonance Scattering and Its Applications in
Spintronics 297
Saeed Kamali
12.1. Introduction 297
12.2. Synchrotron Radiation 297
12.3. Nuclear Resonance Scattering 299
12.4. Exchange Spring Magnets 303
12.5. Magnetic Tunnel Junctions 306
12.6. Conclusions 311
References 311
PART IV. APPLICATIONS
13 Bionanomagnetism 315
Peter Svedlindh, Klas Gunnarsson, Mattias Strömberg and
Sven Oscarsson
13.1. Introduction 316
13.2. Properties and Biofunctionalisation of Magnetic Beads 318
13.2.1. Magnetic beads 318
13.2.2. Biofunctionalisation of magnetic beads – the
SPDP coupling chemistry 319
13.3. An Example of a Recently Developed Magnetic
Biosensor Scheme — The Volume-Amplified
Magnetic Nanobead Detection Assay 321
13.3.1. Dynamic magnetic properties and relaxation
mechanisms of magnetic beads 321
xiv Contents
13.3.2.
Brief overview of the volume-amplified
nanobead detection assay 322
13.4. Transportation and Release of Biomolecules Using
Magnetic Beads 327
13.5. Conclusions and Outlook 334
13.6. Abbreviations and Acronyms of Chapter 13 336
References 338
14 Domain Walls for Logic and Data Storage Applications 343
Colm C. Faulkner
14.1. Introduction 343
14.2. Theory 344
14.3. Wire Switching 345
14.4. Domain Wall Propagation 347
14.5. Domain Wall Injection 348
14.6. Rounded Corner Structures 351
14.7. Domain Wall Localisation/Trapping/Point Contacts 352
14.8. Domain Wall Protrusion 354
14.9. Domain Wall Chirality 355
14.10. Domain Wall Dynamics 356
14.11. DW Velocity Enhancements 358
14.11.1. Transverse field 358
14.11.2. Roughness 358
14.11.3. Current assisted 359
14.11.4. Ion irradiation 359
14.11.5. Out of plane field 359
14.12. Spin Torque 359
14.13. Domain Wall Mediated Data Storage 361
14.14. DW Racetrack Memory 361
14.15. Domain Wall Logic 362
14.16. NOT 363
14.17. AND/OR 365
14.18. Fanout/Cloning 365
14.19. Crossover 366
14.20. Data Input 368
Contents xv
14.21. DW Diode 369
14.22. Outlook 370
References 370
Author Index 375
Subject Index 377
Chapter 1
CONCEPTS IN NANOMAGNETISM AND SPINTRONICS
Farzad Nasirpouri
Department of Materials Engineering
Sahand University of Technology
Tabriz, Iran
E-mail: [email protected]
Alain Nogaret
Department of Physics, University of Bath, Bath, UK
E-mail: [email protected]
Reduced size and dimensions reveal novel phenomena in magnetism
and electronics. We aim to give a brief description of fundamentals of
Nanomagnetism and Spintronics to introduce concepts used in the
following chapters.
1.1. Nanoscale Science and Technology
Since the initial proposal of Feynman,1 much progress has been
made in the understanding of Physics on the nanoscale. Based on
these advances, an increasing number of technological applications,
from magnetic read heads to automotive sensors, are now reaching
the commercial market. Nanoscience commonly refers to physical
phenomena in ultra-small structures which have one or more
dimensions below 100 nm.2 “Nanomagnetism and Spintronics” form an
interdisciplinary sub-field linking ferromagnetism and electronics
which holds wide scientific and technological prospects for the future.3
This chapter gives a brief introduction to fundamental ideas in
nanomagnetism and spintronics. Review articles on more specific aspects
have appeared in Refs. 4–18.
1
2 F. Nasirpouri and A. Nogaret
1.2. Nanomagnetism
Magnetic ordering in ferromagnetic materials19 is a complex
phenomenon involving competing energies over different length scales.
On the scale of the interatomic distance, exchange interaction is
responsible for the ordering of individual spins. In transition metals,
ferromagnetic or antiferromagnetic spin alignment is determined by the
symmetry of the orbital wavefunction and the position of the Fermi level
relative to the centre of the d-band. On a larger scale, 10–100 nm, the
finite dimensions of the sample cause the formation of magnetic poles
which increase magnetic energy. This dipolar energy and the associated
demagnetizing field are minimized by the formation of the magnetic
domain structure. In rare earth based magnets, high atomic numbers
cause preferential orientation of the magnetization due to spin orbit
interaction. These effects explain the high magnetocrystalline
anisotropy of rare earth magnets and their compounds. The interaction
of the 4f orbitals with the lattice gives oblate (saucer-shaped) or prolate
(cigar-shaped) orbitals which sets the magnetic easy axis in the
direction that maximizes the angular momentum. This is the direction
perpendicular to equatorial plane in oblate systems and in the plane
of the equator in prolate ellipsoids. Magnetocrystalline anisotropy
competes with exchange energy to set the width of magnetic domain
walls. A domain wall consists of a spin chain that progressively tilts
across the domain wall linking the magnetization in one domain to the
magnetization in the next. Exchange interaction conspires to make
the domain wall wider to minimize exchange energy. Conversely, the
energy of magnetocrystalline anisotropy is minimized in vanishingly
small domain walls. In the limit of a domain wall made of antiparallel
spins, the energy of anisotropy is minimum because both spins align with
the easy axis. The domain wall width corresponds to the optimum
distance that minimizes the total energy of the domain wall. As a result,
anisotropic magnets have thin domain walls, down to the interatomic
distance, while magnets lacking anisotropy have domain walls as large as
a few tens of nanometres. The length scales of magnetic ordering are
shown in Fig. 1.1. The following section will explain the principles of
magnetism which leads to explore nanomagnetic properties and devices.
Concepts in Nanomagnetism and Spintronics 3
Fig. 1.1. Critical length scales in nanomagnetism and spintronics.
1.2.1. Magnetic ordering on the nanoscale
Ferromagnetic alignment is determined by Heisenberg exchange theory
when J is positive in:
E = −2 Jexchange S1 . S2 (1)
4 F. Nasirpouri and A. Nogaret
Fig. 1.2. Schematic of density of states as a function of energy in (A) ferromagnet and
(B) a normal metal.
The consequence of the spin alignment due to the exchange
interaction is that the material has a spontaneous magnetization giving
a net magnetic moment per unit volume (magnetization). The
magnetization of ferromagnets corresponds to the filling of incomplete
energy shells: the 3d shell for transition metals (Fe, Ni, Co) and the
4f shell for lanthanides. These form the elemental ferromagnets.
Magnetism in these metals arises from the filling of the 3d and 4f
electron shells, respectively.20
In Fig. 1.2, the density of states of a typical 3d ferromagnet (a) is
compared to a non-magnetic 3d metal (normal metal) (b), referred to as a
Pauli paramagnet. At zero field the normal metal has an equal number of
spin up and down electrons. Since the 4s orbitals are extended in space,
the wavefunctions of neighbouring atoms strongly overlap giving a wide
4s-band (15–20 eV). In contrast, the 3d orbitals are more localized on
Concepts in Nanomagnetism and Spintronics 5
the atomic sites giving a comparatively narrow energy band (4–7 eV). In
the solid state, due to the strong hybridization of these two energy bands,
one cannot make a clear distinction between the 3d and the 4s character.
The 3d and 4s electrons are itinerant electrons, the former being less
mobile21 carry less current. Electrons fill the band up to the Fermi level
according to the Pauli Exclusion Principle. Switching on exchange
coupling transfers electrons from the spin down band to the spin up band.
More parallel spins lowers the exchange energy of the system. The Pauli
Exclusion Principle causes the transferred electrons to fill empty states
above the Fermi level which has the effect of increasing the overall
kinetic energy. Transition metals present the peculiarity of having a high
density of states at the Fermi level due to the 3d electron band. As a
result, the increase in kinetic energy is small comparatively to exchange
interaction which has the net effect to reduce the total energy of the
system and stabilize ferromagnetism. This is formulated by the Stoner
criterion,
Jexchange . Z ( EF ) > 1 (2)
that requires either a large exchange integral, J, or a high Fermi density
of states for ferromagnetism to be stable.
1.2.2. Magnetization reversal
The magnetization of ferromagnets is characterized by a magnetic
hysteresis loop whose most important parameters are the coercive field
( H c ) and the remanent magnetization ( M r ) — see Fig. 1.3. The energy
dissipated in one cycle corresponds to the M-H area enclosed by the
hysteresis loop.
Magnetization reversal follows different processes depending on
size. In small magnetic clusters, a few tens of nanometres in diameter,
dipolar energy is smaller than other relevant energies. All magnetic
microscopic moments bound by exchange interaction rotate coherently
when the applied magnetic field is reversed. The energy barrier to
magnetization reversal is due to anisotropy.
Clusters are very important to magnetic applications because the
coercive field can be set precisely by engineering the magnetic
anisotropy constant K. For this reason, a preferred method for
6 F. Nasirpouri and A. Nogaret
engineering magnetic materials is to sinter magnetic powders instead of
using bulk magnetic alloys.
On the mesoscopic scale, dipolar energy becomes comparable to the
energy of formation of domain walls. Microscopic magnetic moments
tilt parallel to interfaces lowering dipolar energy to the detriment of
exchange energies. The magnetic structure exhibits complex patterns
characterized by the formation of vortices, and is highly dependent on
the geometry. Magnetization reversal occurs through the twisting or
buckling of these vortices and lead to delicate steps in the magnetization
curves.
Fig. 1.3. Magnetic properties of materials as defined on the M-H plane of magnetization
M versus magnetic field H. These include coercivity H c , remanence M r , initial
permeability µin , maximum differential permeability µmax and saturation magnetization
M s . (Virgin curve represents the response of the magnetic moments on the first sweep of
magnetic field).
Concepts in Nanomagnetism and Spintronics 7
On the macroscopic scale, magnetization reversal is determined by
the nucleation and propagation of magnetic domain walls in the bulk.
One distinguishes the case of ideal crystals from dirty metals. In the
former case magnetic domains nucleate at sites where exchange
interaction is weakest. Subsequently, antiparallel magnetic domains
rapidly expand allowing the magnetization to change sign. The coercive
field in this case is controlled by the nucleation of magnetic domains. In
dirty metals, defects pin magnetic domains as they expand. The coercive
field is the field that enables magnetic domains to grow and overcome
the pinning potential.
1.2.3. Dimensionality in magnetism
Dimensionality is a general term which is used to describe the effect of
size confinement on the physical properties of matter. It becomes very
important, when the properties of matter are discussed on the nanometre
scale. In general, one divides nanostructures into three different groups
including: thin films (2D structures), nanowires (1D structure), and
nanodots or nanoparticles (0D structures). The magnetization of these
different groups is significantly influenced by confinement.
1.2.3.1. Thin magnetic films
The electronic structure at the surface of a given material differs
considerably from the bulk. Early studies of nickel monolayers on
copper substrates show significant changes in the density of states as a
function of surface energy. The different coordination number of atoms
and the number of incomplete bonds at the surface modifies the local
density of states. The d orbital is more localized and interacts more
strongly with spin magnetic moments giving a higher net magnetization
at the surface. This is an example of size effect in one dimension.
The magnetization reversal in thin ferromagnetic films is affected by
such size confinement. The nucleation and propagation of magnetic
domains which involve Bloch walls in the bulk is more likely to involve
Néel walls when the thickness decreases. Comparison between the
energy and thickness of Bloch wall and Néel wall as a function of film
8 F. Nasirpouri and A. Nogaret
thickness confirms the tendency of the formation of Néel walls in the
thinner film. In thin films with uniaxial in-plane anisotropy, magnetic
domains form stripes and are separated by Néel walls for thicknesses
below 50 nm. In thin films with perpendicular anisotropy, circular
magnetic domains form which are highly mobile. These magnetic
‘bubbles’ hold potential for magnetic storage technology.22
1.2.3.2. Nanowires or one dimensional magnets
Fundamental interest in arrays of ferromagnetic nanowires lies in the
emergence of novel magnetic and transport properties as the dimension
approaches the length scale of a few nanometres to a few tens of
nanometres. Current interest in research on ferromagnetic nanowires is
stimulated by the potential application to future ultra-high-density
magnetic recording media.23
The controlled production of magnetic nanowire arrays with
outstanding characteristics is important to control the magnetization
process. Free standing nanowires can be fabricated by different methods
such as lithography and template electrodeposition.24–28 The modes of
magnetization reversal have been evaluated using a variety of ‘nanowire’
arrays fabricated by electrodeposition into lithographically made patterns
with different aspect ratios of up to 3 and diameters below 180 nm.
The smaller diameter wires have high out-of-plane remanence while
larger diameter wires have low remanence. This is consistent with the
predictions of a micromagnetic model, which indicates a change from a
‘flower’ (or single domain) to a ‘vortex’ remanent state with increasing
diameter. The flower–vortex transition occurs at a diameter of 3.5 times
of exchange length for cylinders without any magnetocrystalline
anisotropy.29 However, it is important to note that the remanence only
decreases slowly as the vortex develops, and high aspect ratio vortex-
state particles can still have significant remanence. Figure 1.4 depicts the
change of magnetization modes for different geometries.
Previous studies30–33 have indicated for particles with different aspect
ratios that the reversal of magnetization is categorized into two different
modes including coherent-rotation and curling. There exists a critical
radius Rc : If R < Rc , reversal occurs by coherent-rotation, and, if
Concepts in Nanomagnetism and Spintronics 9
R > Rc , the reversal occurs by curling. The critical radius Rc is given
by:
kA
Rc = (3)
µo M s2
where k is a constant depending on the length/diameter ratio of the
particles, ranging from 1.08 for an infinite cylinder to 1.38 for a sphere,
A is the exchange constant in erg/cm and µo M s is the magnetization at
the saturation.34
The main factor which controls magnetization reversal in elongated
ferromagnetic nanostructures is the domain wall. The nucleation and
propagation of domain wall between opposing magnetic domains is of
Fig. 1.4. The axial remanence of cylindrical particles as a function of diameter, for three
different aspect ratios of R = 1.5, 2 and 3. Below 3.5 times of exchange length the vortex
develops and the remanence gradually decreases, more slowly at higher aspect ratios.
Reprinted with kind permission from C.A. Ross.29
10 F. Nasirpouri and A. Nogaret
basic interest and critical to the performance of spintronic and
magnetic storage devices. A series of studies has been carried out on
lithographically fabricated ferromagnetic nanowires35,36 to investigate
the magnetic domain wall dynamics for applications to logic spintronics
devices which will be discussed in Chapter 14. In the other side,
electrodeposited nanowire arrays stimulate great interest because they
provide a relatively simple and inexpensive way to study the magnetic
properties of this one dimensional structure. This type of fabrication
method is also useful for making multilayered nanowires exhibiting
giant magnetoresistance.27,28 Figure 1.5 shows two well known templates
in which magnetic nanowires are grown using electrochemical
deposition.
To increase the density of magnetic storage, the spacing between
magnetic domains (or single nanowires) must be minimized and the
cross-talk between adjacent magnets must be avoided. Nanomagnets
can be obtained by electrodepositing nanowires, whose typical
dimensions may vary routinely from 500 nm down to 30 nm or even
smaller. Generally, the factors determining the final magnetic response
of nanowire arrays are, (i) the magnetic nature of individual nanowires,
preferably having strong longitudinal magnetic anisotropy for data
storage purposes, and (ii) the characteristics of the geometrical
arrangement of the nanopore/nanowire array, which determines
the strength of the magnetostatic interaction among neighbouring
nanowires.33
Details of the magnetic properties of nickel and iron nanowires with
diameters less than the domain wall have been explained by Kroll et al.39
It is confirmed that magnetization reversal in nanowires indicates two
stable orientations of the magnetic moments, one pointing parallel and
one antiparallel to the wire axis. These two orientations are separated by
an energy barrier giving a slight hysteresis in the in-plane magnetization
curves. For Fe and Ni, shape anisotropy predominates giving rise to easy
axes in the direction of antiparalel orientation, while the case for Co
nanowires arrays is more complicated due to temperature-dependent and
size-dependent magnetocrystalline anisotropy.37–44
Concepts in Nanomagnetism and Spintronics 11
(a)
(b)
Fig. 1.5. (a) Top view of a polyester track-etched nanoporous membrane (Osmonics) and
(b) naturally ordered anodic aluminium oxide membrane for template electrodeposition
of magnetic nanowires in the pores.28
1.2.3.3. Nanodots and superparamagnetism
Competition between exchange and anisotropy constants45 induces
a critical size of nanodots. The state of lowest free energy of a
ferromagnetic particle is one of uniform magnetization below a certain
critical size. For larger particles, the magnetization is non-uniform.
These two modes are known to correspond to single domain and vortex-
states which have been studied extensively over last decade.46,47
Measurement of magnetization of single dot presents technical
difficulties, though of high interest. Arrays of magnetic nanodots have
been fabricated mainly by nanolithography techniques (i.e. electron beam,
X-ray, etc.) and have been measured using small scale magneto-optical
12 F. Nasirpouri and A. Nogaret
Kerr effect. The magnetization curves obtained for the arrays of circular
dots exhibiting the two reversal modes. The effect of diameter and
thickness is quite important. As the size approaches the critical radius,
the single domain mode is predominant.48 Nanodots smaller than the
critical radius, tend to give up ferromagnetic alignment, becoming
superparamagnetic. This behaviour takes place when the fluctuations of
magnetic moments caused by the thermal energy distort the intrinsic
alignment of the ferromagnet. Considering a single domain particle
with two distinct magnetic states, when the thermal energy becomes
comparable with the energy barrier between the two states,
superparamagnetism is established. This process obeys the Arrhenius-
Née1 equation:
kuV
K T
f = fo . e B
(4)
where ku is the anisotropy constant, V is the volume of particle, K B
is the thermal energy, f is the frequency of switching between
magnetization states.49
1.3. Spintronics
The idea of a resistance induced by spin scattering has first been put
forward by Mott to explain the resistance drop observed in transition
metals below their Curie temperature. Mott suggested that the electric
current in the metal had a spin up and a spin down component. These
formed independent conduction channels which only mixed via spin
scattering on antiparallel layers. In the paramagnetic phase the
resistance was therefore expected to be higher than in the ordered
ferromagnetic phase, leading to the observed resistance drop. The
development of metal epitaxy in the mid 1980s has allowed growing
atomically thin layers of crystalline quality. This has opened the
possibility of synthesizing sandwich structures such as Fe/Cr/Fe and
Co/Cu/Co of width smaller than the electron mean free path. On the
scale of the structure, a current flowing across it has two independent
components corresponding to each spin polarization which mix
depending on the relative orientation of the magnetizations. The
Concepts in Nanomagnetism and Spintronics 13
ferro/normal/ferro multilayers thus formed the first man made device
demonstrating the interplay between magnetic and electronic properties.
Giant magnetoresistance was first demonstrated in 198850,51 in
sputtered Fe/Cr multilayers. A spin-electronic current was injected in
the plane of a superlattice in the so-called current in-plane (CIP)
configuration. The current perpendicular to plane (CPP) was investigated
afterwards which showed the giant magnetoresistance to be isotropic.
GMR takes place because of spin-dependent electron scattering. The
width of the trilayer structure is smaller than the mean free path so that
momentum scattering has marginal effects compared to spin scattering.
The band structure of the transition metal ferromagnet (Fig. 1.2) can
explain the mechanism of GMR based on the properties of electrons with
energies close to the Fermi energy (EF). The most mobile electrons are
in the s-band while the spin polarized electrons are in the d-band. Current
GMR models assume that s electrons mainly carry the current, whereas
the d states are those responsible for spin scattering. From Fig. 1.2 it
becomes obvious that electrons with spin down are scattered more than
those with spin up.52
Giant magnetoresistance has been implemented in spin valves which
are used in magnetic read heads. These are GMR layers in which one
layer is pinned while the other is free to move. A large current driven
through a spin valve induced high frequency oscillations of the
magnetization through the spin torque effect. Namely the pinned layer
polarizes a spin current which impinges on the free layer with opposite
magnetization. The polarized spin current applies effective magnetic
field
ℏI
Beff = (5)
(2eMAl )n
to the free layer. Here I is the drive current, M is the free layer
magnetization, l is the thickness of the free layer and A is the junction
area. This effective magnetic field tends to align the free layer
magnetization with the magnetization of the pinned layer. This system
becomes unstable as the resistance drops and the effective magnetic
field increases further increasing the frequency of oscillations of the
14 F. Nasirpouri and A. Nogaret
magnetization. Although small, the power emitted by these nano-
oscillators presents an interest for on-chip wireless communications.
An attractive way to control the spin phase is to use an electric field
via spin orbit coupling. The spin orbit coupling Hamiltonian is obtained
in the classical limit of Dirac’s equation. Special relavity stipulates
that
a charged particle moving at momentum p through electric field E will
be subject to magnetic field:
p∧E
B=− (6)
mc 2
This magnetic field induces spin precession proportional to the
electric field. Dephasing induced by spin orbit coupling has been
observed in the vacuum albeit under very large electric fields because the
mc2 term is large. In semiconductors, the energy difference between the
conduction and valence bands replaces the Dirac gap 2mc 2 . The smaller
semiconductor band gap allows spin orbit to induce sizable spin
dephasing, an effect at the core of the proposal for a spin transistor by
Datta and Das53 and used in the Aharonov-Casher effect. A spin
polarized current injected through the plane of a quantum well and
subjected to a transverse electric field will change spin polarization at a
rate controlled by the magnitude of the electric field. By choosing the
electric field the electron spin polarization will either be transmitted of
not, which is the basis of the spin transistor action.
References
1. R. P. Feynman, Miniaturization (Reinholds, New York, 1961).
2. G. M. Whitesides, Nanotechnology: Art of the Possible, Technology Review
(Technology Review Inc., Cambridge, MA, Nov/Dec 1998).
3. Y.-H. Wu, Encylopedia of Nanoscience and Nanotechnology, Vol. 7 (American
Scientific Pub., 2003), pp. 493–544.
4. J. Nogues, I. K. Schuller, J. Magn. Magn. Mater., 192 (1999) 203.
5. R. P. Cowburn, J. Magn. Magn. Mater., 242–245 (2002) 505.
6. S. D. Bader, Rev. Mod. Phys., 78 (2006) 1, also see: S. Bader, K. S. Buchanan,
S.-H. Chung, K. Y. Guslienko, A. Hoffmann, Y. Ji, V. Novosad, Superlattices and
Microstructures, 41 (2007) 72.
Concepts in Nanomagnetism and Spintronics 15
7. C. L. Dennis, R. P. Borges, L. D. Buda, U. Ebel, J. F. Gregg, M. Hehn, E. Jouguelet,
K. Ounadjel, I. Petej, I. L. Prejbeanu, M. J. Thornton, J. Phys.: Condens. Matter, 14
(2002) R1175.
8. G. Srajera, L. H. Lewisb, S. D. Bader, A. J. Epstein, C. S. Fadley, E. E. Fullerton,
A. Hoffmann, J. B. Kortright, K. M. Krishnan, S. A. Majetich, T. S. Rahman, C. A.
Ross, M. B. Salamon, I. K. Schuller, T. C. Schulthess, J. Z. Su, J. Magn. Magn.
Mater., 307 (2006) 1.
9. Y.-B. Xu, Current Opinion in Solid State and Materials Science, 10 (2006) 81.
10. C. Chappert, A. Fert, F. Nguyen Van Dau, Nature Materials, 6 (2007) 813.
11. M. R. J. Gibbs, Current Opinion in Solid State and Materials Science, 7 (2003) 83.
12. J. F. Gregg, I. Petej, E. Jouguelet, C. Dennis, J. Phys. D: Appl. Phys., 35 (2002)
R121.
13. S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Daughton, S. Molnar,
M. L. Roukes, A. Y. Chtchelkanova, D. M. Treger, Science, 294 (2001) 1488.
14. J. I. Martın, J. Nogues, K. Liu, J. L. Vicent, I. K. Schuller, J. Magn. Magn. Mater.,
256 (2003) 449.
15. M. R. Fitzsimmons, S. D. Bader, J. A. Borchers, G. P. Felcher, J. K. Furdynad,
A. Hoffmann, J. B. Kortright, I. K. Schuller, T. C. Schulthess, S. K. Sinha, M. F.
Toney, D. Weller, S. Wolf, J. Magn. Magn. Mater., 271 (2004) 103.
16. S. D. Bader, Surface Science, 500 (2002) 172.
17. R. Skomski, J. Phys.: Condens. Matter, 15 (2003) R841.
18. C. A. Ross, Annu. Rev. Mater. Res., 31 (2001) 203.
19. E. Wolf, Nanophysics and Nanotechnology (Wiley, 2004).
20. B. D. Cullity, Introduction to Magnetic Materials (Adison-Wesley, Reading, Mass.,
1972).
21. Scientific Background on the Nobel Prize in Physics 2007. The Discovery of Giant
Magnetoresistance Compiled by the Class for Physics of the Royal Swedish
Academy of Sciences, October 2007.
22. R. O’Handley, Modern Magnetic Materials (Wiley, 2000).
23. H. Zeng, R. Skomski, L. Menon, Y. Liu, S. Bandyopadhyay, D. J. Sellmyer, Phys.
Rev. B, 65 (2000) 134426.
24. M. Ghorbani, F. Nasirpouri, A. Iraji Zad, A. Saedi, Materials and Design, 27 (2006)
983.
25. F. Nasirpouri, M. Abdollahzadeh, M. J. Almasi, N. Parvini-Ahmadi, Curr. Appl.
Phys., 9 (2009) S91.
26. F. Nasirpouri, P. Southern, M. Ghorbani, A. I. Zad, W. Schwarzacher, J. Magn.
Magn. Mater., 308 (2007) 35.
27. F. Nasirpouri, Electrodeposition of magnetic nanowire arrays, in ed. A. El-Nemr,
New Developments in Electrodeposition and Pitting Research (Research Signpost
Pub., 2007), pp. 55–92.
28. W. Schwarzacher, D. Lashmore, IEEE Trans. Magnetics, 32 (1996) 3113.
16 F. Nasirpouri and A. Nogaret
29. C. A. Ross, M. Hwang, M. Shima, J.-Y. Cheng, M. Farhoud, T. A. Savas, H. I.
Smith, W. Schwarzacher, F. M. Ross, M. Redjdal, F. B. Humphrey, Phys. Rev. B, 67
(2002) 144417, also see: C. A. Ross, M. Hwang, M. Shima, H. I. Smith, M. Farhoud,
T. A. Savas, W. Schwarzacher, J. Parrochon, W. Escoffier, H. Neal Bertram,
F. B. Humphrey, M. Redjdal, J. Magn. Magn. Mater., 249 (2002) 200.
30. Y. Peng, T. H. Shen, B. Ashworth, J. Appl. Phys., 93 (2003) 7050.
31. L. Cheng-Zhang, J. C. Lodder, J. Magn. Magn. Mater., 88 (1990) 236.
32. G. T. A. Huysmans, J. C. Lodder, J. Appl. Phys., 64 (1988) 2016.
33. R. Skomski, J. M. D. Coey, Permanent Magnetism (Institute of Physics Pub., Bristol,
1999).
34. M. Zheng, L. Menon, H. Zheng, Y. Liu, S. Bandyopadhyay, R. D. Kirby, D. J.
Sellmyer, Phys. Rev. B, 62 (2000) 12282.
35. D. Atkinson, D. Allwood, G. Xiong, M. Cooke, C. C. Faulkner, R. P. Cowburn,
Nature Materials, 2 (2003) 85.
36. D. Atkinson, D. A. Allwood, C. Faulkner, G. Xiong, M. D. Cooke, R. P. Cowburn,
IEEE Trans. Magnetics, 39 (2003) 2663.
37. K. Ounadjela, R. Ferré, L. Louail, J. M. George, J. L. Maurice, L. Piraux, S. Dubois,
J. Appl. Phys., 81 (1997) 5455.
38. M. Hernandez-Velez, Thin Solid Films, 495 (2006) 51.
39. M. Kroll, W. J. Blau, D. Grandjean, E. R. Benfield, F. Luis, P. M. Paulus, L. J. de
Jongh, J. Magn. Magn. Mater., 249 (2002) 241.
40. P. M. Paulus, F. Luis, M. Kroll, G. Schmid, L. J. de Jongh, J. Magn. Magn. Mater.,
224 (2001) 180.
41. J. M. Garcıa, A. Asenjo, J. Velazquez, D. Garcıa, M. Vazquez, P. Aranda, E. Ruiz-
Hitzky, J. Appl. Phys., 85 (1999) 5480.
42. G. J. Strijkers, J. H. J. Dalderop, M. A. A. Broeksteeg, H. J. M. Swagten, W. J. M.
Jonge, J. Appl. Phys., 86 (1999) 5141.
43. H. Zeng, M. Zheng, R. Skomski, D. J. Sellmyer, Y. Liu, L. Menon,
S. Bandyopadhyay, J. Appl. Phys., 87 (2000) 4718.
44. H. R. Khan, K. Petrikowski, J. Magn. Magn. Mater., 249 (2000) 458.
45. W. E. Brown Jr, J. Appl. Phys., 39 (1968) 993.
46. R. P. Cowburn, D. K. Koltsov, A. O. Adeyeye, M. E. Welland, Phys. Rev. Lett., 83
(1999) 1042.
47. K. Y. Guslienko, V. Novosad, Y. Otani, H. Shima, K. Fukamichi, Phys. Rev. B, 65
(2001) 24414.
48. F. Nasirpouri, A. Nogaret, D. Atkinson, M. Ghorbani, A. I. Zad, J. Magn. Magn.
Mater., 299 (2006) 356.
49. M. J. Bonder, Y. Huang, G. C. Hadjipanayis, in Advanced Magnetic Nanostructures,
eds. D. Sellmyer, R. Skomski (Springer, 2006), p. 185.
50. M. N. Baibich, J. M. Broto, A. Fert, V. D. F. Nguyen, F. Petroff, Phys. Rev. Lett., 61
(1988) 2472.
Concepts in Nanomagnetism and Spintronics 17
51. G. Binasch, P. Grunberg, F. Saurenbach, W. Zinn, Phys. Rev. B, 39 (1989) 4828.
52. P. Grutberg, in Magnetic Multilayers and Giant Magnetoresistance, ed. U. Hartmann
(Springer, 2000).
53. S. Datta, B. Das, Appl. Phys. Lett., 56 (1990) 665.
This page is intentionally left blank
Chapter 2
ARTIFICIAL MAGNETIC DOMAIN STRUCTURES REALIZED
BY FOCUSSED ION BEAM IRRADIATION
Simon Bending* and Simon Crampin
Department of Physics, University of Bath
Claverton Down, Bath BA2 7AY, UK
*E-mail:
[email protected] Atif Aziz
Department of Materials Science and Metallurgy
University of Cambridge, Cambridge CB2 3QZ, UK
We describe how artificial ferromagnetic domain patterns can be
realized by focussed ion beam irradiation of Pt/Co/Pt trilayers with
perpendicular magnetic anisotropy. This approach allows exquisite
control of the coercive fields of ‘patterned’ regions with very high
spatial resolution. Systematic extraordinary Hall effect studies of
magnetic switching in artificial magnetic domains as a function of
irradiation dose and size are described and interpreted. The potential
of such structures is then illustrated in an investigation of the angle-
dependent intrinsic domain wall resistivity in artificial superlattice
structures. We find excellent agreement between these experimental
results and a theory of domain wall resistivity based on spin-dependent
potentials and scattering rates for majority and minority spin electrons
due to Levy and Zhang.
2.1. Introduction
Intense efforts are currently being devoted worldwide to the goal of
developing magnetic devices based on spin-polarised electronic currents.
These so-called spintronic devices will combine the advantageous
19
20 S. Bending, S. Crampin and A. Aziz
properties of magnetic and semiconductor materials, and are expected to
be fast, non-volatile and versatile, capable of simultaneous data storage
and processing, while at the same time consuming less energy. The first
spintronic structures to be developed were magnetic sensors based
on the giant magnetoresistance (GMR) effect. These devices rely on
spin-dependent electronic transport in sandwiches composed of two
ferromagnetic films separated by a very thin non-magnetic spacer. Spin
scattering asymmetry of up and down spins in ferromagnetic metals leads
to much lower scattering rates (lower resistance) when the magnetisation
in the two layers is parallel as opposed to anti-parallel. GMR devices,
and related structures based on tunnelling magnetoresistance, are now
found in the read heads of nearly all magnetic hard drives and have been
key in maintaining the trend for rapidly increasing data storage
capacities. The discovery of GMR was also recognised by the award of
the 2007 Physics Nobel prize. Magnetic random access memory
(MRAM) is an ‘active’ second generation development in spintronics
consisting of magnetic tunnel junctions where the magnetisation
direction of one electrode is switched by the stray fields from current-
carrying ‘write’ lines. These structures represent switchable non-volatile
memory cells, which can be read via the resistance state of the junction.
However, the magnetic switching used in MRAM technology is
relatively inefficient and leads to unwanted interactions between
neighbouring cells, and attention has now turned to third generation
current-switched devices based on spin transfer torque due to spin-
polarised current injection across a domain wall (DW) (see Refs. 1–6).
A key feature of all these sensor and memory cell technologies is the
ability to prepare, pin and manipulate well-defined and reproducible
domain wall structures. Traditionally shape anisotropy has been used in
nanostructured thin films to control stable domain configurations, and
domain walls have either been trapped at ‘notches’ (constrictions) or in
curved wire segments. This reliance on naturally occurring domain
structures imposes severe restrictions on the design of spintronic devices.
Recently an entirely different approach to the generation and trapping of
domain walls has began to be explored based on artificial domain
structures created by low dose focussed ion beam (FIB) irradiation. This
technique allows the local magnetic anisotropy and coercive fields of
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 21
magnetic multilayers to be precisely tuned with nanoscale (≥10 nm)
resolution. This allows far more flexibility in the generation and
trapping of domain walls, and careful optimisation of the magnetic
anisotropy in adjacent domains could potentially lead to much lower
critical current densities in current-switched devices. The study of
artificial domain structures in multilayer films with perpendicular
magnetic anisotropy is currently most mature, and will be the focus
of this article. There is, however, a growing body of literature on
the influence of FIB irradiation on in-plane magnetised multilayer
materials.
We will demonstrate how the exquisite control offered by
artificial domain structures can be exploited to make an unambiguous
quantitative investigation of intrinsic domain wall resistivity in magnetic
“superlattices”.
2.1.1. Controlling magnetic anisotropy by irradiation
Masked irradiation of ferromagnetic multilayers is becoming a widely
used technique to realize planar structures with ‘patterned’ magnetic
coercivity.7 This approach raises the exciting prospect of directly
imposing a desired ‘artificial’ domain structure onto ferromagnetic
multilayers and removes the reliance on ‘natural’ magnetic domain
structures that can be difficult to initialise or control.
Here we focus on the use of ion beam irradiation to control the
magnetic behaviour of technologically relevant out-of-plane magnetic
anisotropy systems. Co/Pt multilayers have perpendicular magnetic
anisotropy (PMA) over a wide range of layer thicknesses.8,9 The interface
structure, namely the strain, roughness and degree of intermixing/alloy
formation, plays a key role in the strength of the PMA, and this feature
was exploited by Chappert et al.9 to systematically reduce the coercive
field of Pt/Co/Pt trilayers using 30 keV He+ irradiation (the system reverts
to in-plane magnetisation at very high doses for certain parameters). The
induced changes arise from mixing/roughening at the Co/Pt interfaces
combined with a relief of interface strain.10 Remarkably the same
magnetic control was demonstrated using a 28 keV Ga+ FIB with
extremely high spatial resolution (≥10 nm).11 Moreover, most of the
22 S. Bending, S. Crampin and A. Aziz
incident Ga+ ions pass through the ultra-thin films used and implant into
the substrate leading to minimal chemical contamination.
2.1.2. Intrinsic domain wall resistivity
Since the first measurements almost thirty years ago12,13 of the resistance
of single crystal Fe whiskers, which showed large changes as the multi-
domain ferromagnetic state was eliminated in a small applied field, there
has been considerable interest in the domain wall contribution to the
resistivity of ferromagnetic materials. The topic enjoyed a dramatic
revival with the discovery of giant magnetoresistance (GMR) in
ferromagnetic/non-magnetic layered thin films14,15 in which the relative
orientation of the magnetisation in the thin ferromagnetic layers
(effectively the domain structure) determines the spin-dependent
electronic structure and scattering rates. Resistance changes as large as
35% have been observed16 in Co/Cu multilayers at room temperature in
an applied field of ~10 mT and, only ten years after the phenomenon was
discovered, GMR sensors are now routinely used as read heads for data
storage media. In GMR materials the non-magnetic layers break up the
exchange interaction between the magnetic layers, while the chemical
interfaces are almost atomically sharp. In contrast a conventional
domain wall is chemically homogeneous, but the magnetisation varies
gradually on the length scale of the domain wall width (typically
~1–100 nm). Despite the superficial differences between these two
situations they share the common feature that the magnetisation changes
on crossing a magnetic interface, and similar theoretical models have
been applied to both.
Recent theories of domain wall scattering have been based on spin-
dependent potentials and scattering rates for majority and minority spin
electrons in ferromagnets.17 Typically, in a uniformly magnetised
material a large fraction of the electrical current is carried by one spin
channel (i.e. a spin ‘short circuit’). Where the current crosses a domain
wall the spin channels are mixed leading to a partial suppression of the
‘short circuit’ effect and an increase in the resistance which depends
quadratically on the ratio of the precession time for a Fermi surface
electron in the exchange field to the time the electron takes to
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 23
ballistically traverse the domain wall. It has also been shown
theoretically that domain walls can reduce the resistance, either due to
the suppression of weak localisation corrections to the conductivity18 or
due to the redistribution of charge among the majority and minority spin
bands in the domain wall when its electronic structure is accounted for
semiclassically.19 While some experimental observations can be quite
well-described within one or other of these different theoretical
frameworks, it is probably true to say that, until recently, the overall
understanding of domain wall resistance was poor.
Recent advances in the epitaxial growth of metals have allowed the
preparation of a range of thin film samples for domain wall resistance
investigations where the domain structure can be optimised by a
judicious choice of magnetocrystalline anisotropy, film thickness and
orientation and sample geometry (shape anisotropy). Many experiments
have addressed the issue of domain wall resistivity in 4-probe
measurements on micron or sub-micron wire width samples (e.g.
Refs. 20–24). Typical values of the total magnetoresistance in
ferromagnetic thin films are small, a few percent, and the domain wall
contribution is generally much smaller still as the wall width is typically
only a small fraction of the domain size. A major complication for
such measurements arises from the fact that the domain wall
magnetoresistance is readily masked by the extrinsic magnetoresistance
associated with the domains, and an accurate picture of the magnetic
microstructure is vital if the domain wall contribution is to be correctly
extracted. There are two main contributions to ferromagnetic resistivity
anisotropy in measurements well below the Curie temperature;25
anisotropic magnetoresistance due to spin-orbit coupling and Lorentz
magnetoresistance, which depends on the angle between the current
density and the internal magnetic induction. The Hall effect can also lead
to an increase in resistivity as the Hall field changes direction in
alternating magnetic domains,26 while diamagnetic effects influence the
resistivity near a domain wall due to the modification of electron
trajectories there.27
Full control of the properties of magnetic domains can, however, be
achieved provided local anisotropies are changed in a systematic fashion.
This can be realized in perpendicularly magnetised Pt/Co/Pt systems by
24 S. Bending, S. Crampin and A. Aziz
local irradiation with a gallium FIB, which reduces the film coercivity
due to relaxation and mixing of the interfaces28 (see Sec. 2.2). Using Ga
FIB irradiation it is now possible to define artificial static magnetic
domains of arbitrary shapes and sizes in Pt/Co/Pt sandwich or multilayer
structures with deep sub-micron precision. Unlike natural domains,
artificial domains are fabricated by engineering the intrinsic spin-
dependent behaviour of the multilayers and promise reproducible
magnetic and magnetotransport properties. In this article we attempt to
demonstrate the strong potential of this approach. In Sec. 2.3, the
magnetic switching and area-scaling of the dosed regions are studied
using the extraordinary Hall effect (EHE). Artificial domains can also be
an important tool for studying spin transport properties across domain
walls and in Sec. 2.4 we describe a quantitative investigation of DW
resistance in a lateral magnetic ‘superlattice’.
2.2. Fabrication of Artificial Domain Structures
A magnetic thin film sandwich structure (Pt(3.5 nm)/Co(0.5 nm)/
Pt(1.6 nm)) was deposited using dc magnetron sputtering at room
temperature and an Ar pressure of 2.7 mTorr. After deposition, the rms
roughness of the Pt capping layer, as measured by AFM, was 0.4 nm.
Exceptionally thin (1.6 nm) Pt capping layers have been used in these
experiments to minimise the effect of current shunting out of the Co film
in the magnetotransport structures. In addition, the very thin (0.5 nm) Co
layer used in our experiment is probably not chemically continuous, and
will contain some pin holes. However, since the Pt develops an induced
magnetic moment in the proximity of Co, it will be magnetically
continuous. The magnetic thin films were then patterned into
conventional Hall transport structures based on the intersection of 2 µ m
wide ‘wires’ using optical lithography and reactive ion etching with a 1:1
mixture of SF6 and Ar. The very thin Pt capping layers render the
structures extremely sensitive to irradiation, and so thin SiO2 layers with
thicknesses in the range 0–24 nm have been deposited on the top to
reduce the FIB Ga-ion energy and dose. Focussed ion beam irradiation
was performed on the completed structure with a commercial FIB (FEI
Strata 201). The properties of the ions after transmission through the
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 25
SiO2 layer have been estimated using SRIM (The Stopping and Range
of Ions in Matter) software.29 The average energies (% of Ga-ions
transmitted) of the Gallium ions after passing through 0, 8 nm, 16 nm
and 24 nm SiO2 are 30 keV (100%), 19.8 keV (99.6%), 11.7 keV
(88.7%) and 7.1 keV (57%) respectively.
The incident energy of the Ga-ions was 30 keV and a 1 pA current
was used to lightly dose the required area of the sample. The beam
diameter was about 10 nm, the distance between neighbouring pixels
7.4 nm, and irradiation was performed at a magnification of 10,000×,
yielding a write field of 30.4 × 28.5 µ m2. Ion doses of 0.004 pC/µ m2,
0.007 pC/µ m2, 0.011 pC/µ m2 and 0.015 pC/µ m2 were achieved using
pixel dwell times of 0.2 µ s, 0.4 µ s, 0.6 µ s and 0.8 µ s respectively.
Typical AFM and MFM images of a Ga-ion dosed region are shown in
Fig. 2.1, where the test Pt (3.5 nm)/Co (0.5 nm)/Pt (1.6 nm)/SiO2 (8 nm)
sandwich structure has been irradiated with a 0.007 pC/µ m2 Ga-ion dose
in a rectangular strip on the left hand side of the field of view. Under the
measurement conditions (see caption) the dosed region switches, but not
the undosed one, and the MFM image exhibits strong black contrast in
Fig. 2.1. (Left) AFM image of a test structure. (Right) MFM image of the same region
with 50 nm tip lift-off. The perpendicular applied field was reversed to −50 Oe after
saturation in a positive field. The dark area is the dosed region and the light regions
undosed.30
26 S. Bending, S. Crampin and A. Aziz
the irradiated region while no detectable change in the surface
topography is observed in the AFM image.
The magnetic properties of the dosed regions have been investigated
using the Extraordinary Hall Effect (EHE), with the external magnetic
field applied perpendicular to the plane of the film and the in-plane
voltage measured perpendicular to the current.30 In a perpendicular
external magnetic field H, the Hall resistivity ρH can be written as31:
ρ H = Ro H + Re 4π M , (1)
where the first term describes the contribution of the ordinary Hall effect,
Ro is the ordinary Hall coefficient and H is the external magnetic field.
The second term describes the Extraordinary Hall term where M is the
perpendicular component of the magnetisation and Re is the EHE
coefficient, which can be expressed as a function of resistivity ρ :
Re = aρ + bρ 2. Here the first term represents the skew scattering process,
and the second the side jump.
2.3. Magnetic Properties of Artificial Domain Structures
Figure 2.2 shows a family of hysteresis loops of the extraordinary Hall
Effect (EHE) measured at 300 K on four separate probes fabricated with
different thickness SiO2 overlayers and various Ga-ion doses. For all
samples an area of 12 µ m × 2 µ m has been irradiated with the Ga-ion
beam. The EHE yields information about the perpendicular component
of the magnetisation of the Pt/Co/Pt sandwich structures and, in each
case, the EHE of the dosed element (solid line) is compared with a
neighbouring undosed element (dotted line), which is 4 µ m away and
fabricated on the same 2 µ m wide wire. Since the perpendicular
anisotropy is very sensitive to the thickness of the Co film, this
comparison helps to minimise errors introduced by small thickness
inhomogeneities. All the undosed samples have very square hysteresis
loops which indicate that they are all magnetically homogeneous.
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 27
2 2 2
0.004 pC/µ m 0.007 pC/µ m 0.011 pC/µ m 0.015 pC/µ m2
6
4
V (µ V)
0 nm
2
0
−2 1a 1b 1c 1d
−4
−6
6
4
V (µ V)
8 nm
2
0
−2 2a 2b 2c 2d
−4
−6
6
4
16 nm
V (µ V)
2
0
−2 3a 3b 3c 3d
−4
−6
6
4
V (µ V)
24 nm
2
0
−2 4a 4b 4c 4d
−4
−6
−80−40 0 40 80 −80−40 0 40 80 −80−40 0 40 80 −80−40 0 40 80
B (Oe) B (Oe) B (Oe) B (Oe)
Fig. 2.2. EHE measurements of undosed (dotted lines) and dosed (solid lines) elements
with different thickness SiO2 overlayers at 300 K. The Ga-ion beam dose increases from
left to right (see values at top edge of figure) and the SiO2 thickness increases from top to
bottom (see values at right edge of figure). From Ref. 30.
The Ga-ion beam dose increases from left to right (a to d) and the
SiO2 overlayer thickness increases (i.e. the effective energy of the Ga-ion
beam decreases) from top to bottom (1 to 4). These results clearly show
that the perpendicular anisotropy of the Pt/Co/Pt layer can be controlled
not only by varying the Ga-ion dose, but also by varying the thickness of
SiO2 overlayer. The perpendicular magnetic anisotropy arises due to
surface or interface anisotropy, which is a consequence of the reduced
symmetry of the interface atoms, and is very sensitive to the interface
roughness.32 Ga-ion irradiation can induce three main changes in the
Pt/Co/Pt sandwich structure: (1) It increases the roughness at the Pt/Co
interface; (2) Due to the 10% lattice mismatch between Co and Pt, the
thin Co layer is under tensile strain33 which can relax after Ga-ion
irradiation; (3) Ga-ion irradiation can mix the Co/Pt layers at the
interface and result in the formation of a Co-Pt alloy.34 Alloy formation
28 S. Bending, S. Crampin and A. Aziz
requires a relatively high Ga-ion dose as compared to the first two
processes. Since the Co film used in our experiments is very thin
(0.5 nm) Co-Pt alloy formation would result in a significant reduction
in the interfacial anisotropy. Therefore surface roughness and strain
relaxation are believed to be the dominant phenomena that reduce the
coercive field of our samples, whereas Co-Pt alloy formation destroys
the perpendicular anisotropy entirely (see samples 1c, 1d and 2d). When
no SiO2 overlayer is present, the thin (1.6 nm) Pt capping layer allows
the Ga-ion energy to be directly transferred to the Pt and Co atoms at the
Pt/Co interface. Therefore the process of Pt/Co mixing can start at very
low doses as observed in sample 1. When a SiO2 overlayer is present,
Fig. 2.3. EHE measurements of the central Hall cross (M) of the structure sketched in the
inset which has a 3 µ m × 0.5 µ m dosed region at its centre. Step A corresponds to the
switching field of the dosed region whereas steps B and C correspond to switching of the
left and the right undosed regions respectively. Minor hysteresis loops 1 and 2 (gray)
were obtained after saturating the sample in positive and negative magnetic fields
respectively.35
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 29
however, it reduces both the energy absorbed by Pt and Co atoms at the
Pt/Co interface and distributes the recoil energy more uniformly to the
Pt/Co interface through collision cascades, and the mixing process occurs
at a relatively high external dose allowing much better control of the
Pt/Co interface roughness. Samples which have been irradiated with low
effective ion energies have relatively square hysteresis loops indicating
that they are homogeneously magnetised and have high irradiation
homogeneity and low pinning site density.
The switching properties of perpendicularly magnetised dosed
regions have been investigated in more detail using the extraordinary
Hall effect.35 Rectangular regions with dimensions x × 3 µ m2 (where
x = 250 nm, 500 nm, 1 µ m, 2 µ m and 4 µ m) were lightly dosed
(0.007 pC/µ m2) using FIB in the middle of a Hall cross. A sketch of a
typical device is shown in the inset of Fig. 2.3, where the ac current
(I = 10 µ A, f = 29 Hz) flows between contacts on the left and right, and
L, M and R are labels for three adjacent Hall crosses. Figure 2.3 shows
the EHE voltage measured at the middle Hall cross (M) (containing a
500 nm wide dosed stripe) which exhibits three steps. Changes in EHE
voltage are directly related to the out-of-plane component of the
magnetisation of the magnetic film in the Hall cross region. Step A
corresponds to the switching of the dosed region whereas steps B and C
are due to the undosed regions on the left and right sides. The minor
loops 1 and 2 (gray lines) are obtained after saturating the sample in
positive and negative magnetic fields respectively. These are shown in
more detail in Fig. 2.4 which illustrates how the form of minor loops of
identically dosed regions scales with their area after saturation at positive
fields (the sweep rate for all minor loops was 2.6 Oe/s). The width of the
dosed region was varied from 500 nm to 4 µ m while the length was kept
constant at 3 µ m. (A 250 nm wide sample was also measured but its
switching field lay very close to that of the undosed region making it
difficult to measure a minor loop.) The following four features are
clearly identifiable in Fig. 2.4. (a) Minor loops of dosed regions of width
smaller than 2 µ m are strongly asymmetric and (b) the asymmetry
increases with the decreasing width of the dosed regions. (c) For data in
the lower half of the figure, when H is swept in a positive sense,
switching occurs at around 20 Oe, i.e. it is nearly independent of the
30 S. Bending, S. Crampin and A. Aziz
width of the dosed region. (d) Discrete Barkhausen steps are also
observable in the lower half of the minor loops.
The asymmetry of the minor loops shown in Fig. 2.4 indicates that
the switching mechanism is different when the magnetic field is varied in
a positive or negative sense. When all regions (dosed and undosed) are
magnetised in the same direction (upper half), dosed regions switch
sharply and the switching field increases strongly as the irradiated area
decreases (c.f. Fig. 2.5). It shows the logical trend that when the area
approaches zero, Hcd / Hcud → 1. Since this switching field is greater than
the switching fields observed for the lower halves of the minor loops, it
seems probable that reversal is governed by domain nucleation followed
by domain wall motion. The increased switching field for small domains
Fig. 2.4. EHE measurements at 300 K of minor loops of irradiated regions with widths
varying from 500 nm up to 4 µ m. Samples were first saturated in positive fields, and the
external field then swept down until the dosed region reversed, at which point it was
swept back up again. Adapted from Ref. 35.
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 31
Fig. 2.5. Ratio of the coercive field of the dosed region (Hcd) and the average coercive
field of the undosed regions either side (Hcud) as a function of the area of the dosed region
(0.007 pC/µ m2 FIB dose). Adapted from Ref. 35.
(<2 µ m) may be due to the decrease in number of available nucleation
sites. When an external magnetic field is applied in this way, such that it
switches the dosed but not the undosed region, a magnetic domain will
form whose size is determined by the irradiated region, with a domain
wall at its border. We observe that switching back to uniform
magnetisation from this state (lower half) does not depend strongly on
the width of the dosed region, suggesting that reversal is now controlled
by DW motion alone.
In addition to the switching mechanism, minor loops also shed light
on the local pinning potentials in the dosed region. Switching in the
lower half of the minor loops is not as sharp as in the upper half and has
many plateaux-like features, which can be seen particularly clearly in the
1 µ m dosed sample. When the external field is increased from negative
values up to zero, a gradual change in EHE voltage is observed in the
32 S. Bending, S. Crampin and A. Aziz
narrower structures indicating that magnetisation of the irradiated region
is already beginning to relax. Using the Hall effect to quantify this we
estimate that areas of ~0.1 µ m2, ~0.5 µ m2, ~0.7 µ m2 and ~0.7 µ m2 are
relaxed for 4 µ m, 2 µ m, 1 µ m and 0.5 µ m wide regions respectively at
H = 0 (i.e. domains with width <4 µ m have significantly less than 100%
remanence after these minor loops).
For positively increasing fields the reversal of the 1 µ m wide
irradiated stripe (area 3 µ m2) occurs predominantly in six approximately
equal sized steps, which can be associated with local pinning sites
present in the Ga-ion dosed region. We estimate that an area A ≈ 0.4 µ m2
reverses in each step, which must be correlated to large scale interface
irregularities. For a Co film of thickness 0.5 nm, the magnetisation
volume reversed during each step is 2 × 10-16 cm3 and, assuming that
interface irregularities are distributed uniformly, the characteristic
separation of strong pinning sites is estimated as A ≈ 600 nm for the
Ga-ion irradiated region. This is fairly consistent with the observation
that artificial domains of width ≤500 nm generally reverse in a single
step after the relaxation phase (H < 0).
Fig. 2.6. (a1) and (b1) show MFM images of 90° and 45° domain ‘superlattices’ created
by Ga-FIB irradiation at H = 0; (a2) and (b2) show MFM images of the same regions
after saturation of the magnetisation at 300 Oe. Adapted from Ref. 36.
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 33
2.4. Angle-Dependent Domain Wall Resistivity Measurements
We now describe how unique insights into intrinsic domain wall
resistivity can be obtained by studying the magnetotransport properties
of artificial magnetic domain superlattice structures, formed in
perpendicularly magnetised Pt/Co/Pt trilayers irradiated with a Ga
focussed ion beam to locally modify the anisotropy.34 As described
above the deposition of thin SiO2 capping layers prior to Ga-FIB
irradiation can been used to precisely control the coercive fields of the
artificial domains.30,35,36 In this way we have been able to define magnetic
domain structures of arbitrary shape with <10 nm spatial resolution, in
which the domain walls lie at precisely controlled angles, as exemplified
in Figs. 2.6(a) and (b). Other attractive features of this system which
encourage its use for studying domain wall magnetotransport properties
include the fact that closure domains, which can be a major source
of anisotropic magnetoresistance,37 are not energetically favoured in
ultra-thin (0.6 nm) Co films with strong perpendicular anisotropy, whilst
the relatively high film resistivity (~21 µ Ω cm at 300 K) leads to short
scattering times, τ, and ωcτ ~ 10-4 << 1 (where ωc is the cyclotron
frequency), and allow Lorentz magnetoresistance and other effects
associated with the wiggling of current lines at domain walls to be
completely neglected.
Devices used in our transport studies were fabricated from a
Pt(3.5 nm)/Co(0.6 nm)/Pt(1.6 nm) sandwich structure, which was deposited
on a Si/SiO2 substrate at 300 K using dc magnetron sputtering. Optical
lithography and reactive ion etching were used to pattern the magnetic
films in a Wheatstone bridge geometry based on 1 µ m wide ‘active’ wires
(Fig. 2.7(a)). After deposition of an additional 8 nm SiO2 capping layer,
the desired domain pattern was irradiated with a 30 keV Ga-ion dose of
0.009 pC/µ m2 in two asymmetric arms of the bridge (Fig. 2.7(b)) using
an FEI Strata 201 Ga-FIB. To eliminate any extraordinary Hall effect
that may originate from the Pt/Co/Pt underneath the voltage contacts, the
Pt capping layer and Co layer were etched away in an Ar plasma after
photolithography, and before evaporation of Ti/Au electrodes. It is
possible that a very thin film of Co remained unetched after this step. The
magnetic force microscopy (MFM) images shown in Figs. 2.6(a) and (b),
34 S. Bending, S. Crampin and A. Aziz
Fig. 2.7. (a) Micrograph of the Wheatstone bridge device consisting of four Pt/Co/Pt
wires each 16 µ m long and 1 µ m wide. (b) Schematic of the top pair of Pt/Co/Pt wires of
the Wheatstone bridge and sketch of the layer structure used. From Ref. 35.
demonstrate the precision with which ‘superlattice’ artificial domains
can-be defined. Images a1 and b1 were captured at H ⊥ = 50 Oe when
the magnetisation of the irradiated regions was reversed with respect to
unirradiated regions, while a2 and b2 show the same regions after
the magnetisation has been saturated everywhere at H ⊥ = 300 Oe. The
very weak contrast in these latter images indicates that Ga-FIB
irradiation causes a very small change in the saturation magnetisation
due to Co/Pt mixing, and the resistivity of these regions also increases
very slightly by ~2.5%. Measurements of our bridge devices were
performed with a 10 µ A ac current (f = 29 Hz) between current contacts
A and B, and the output voltage measured with a lock-in amplifier
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 35
between V1 and V2. An external magnetic field was applied
perpendicular to the plane of the device with a solenoid.
Fig. 2.8. (a) Magnetoresistance (offset vertically for clarity) of bridge structures
containing 0, 2, 4 and 8 irradiated strips in CPW geometry at 300 K. (b) Maximum
resistance change as a function of the number of irradiated strips of various widths, d.36
36 S. Bending, S. Crampin and A. Aziz
Figure 2.8(a) illustrates a set of magnetoresistance measurements
recorded at 300 K on bridges with stripe domain ‘superlattice’ structures
containing 0, 2, 4 or 8 artificial domains (all 500 nm wide in CPW
geometry). Each measurement has been performed on a different, but
otherwise identical, bridge structure fabricated on the same chip. For
H > +150 Oe, both the irradiated and unirradiated domains are magnetised
in the same direction and the resistance is at a low value. If the field is
now reversed to H ≈ −50 Oe, the magnetisation of the irradiated stripe
domains, which have a smaller coercive field,30,35 begin to reverse and
the resistance increases to a maximum when domain walls are fully
developed between irradiated and unirradiated regions. When the field is
made more negative still it eventually reaches the coercive field of the
unirradiated stripes (~100 Oe), which in turn start to reverse and the
resistance drops back down to a minimum. The same behaviour is
observed if the field is now swept back up in a positive sense. The
weakly varying background signal is probably due to magnetoresistance
arising from residual Co left under the voltage contacts after etching as
described above. A separate Hall bar device was patterned in the same
Pt/Co/Pt trilayer structure with Hall ‘crosses’ containing unirradiated
and irradiated (with the same Ga-FIB dose) strips of the same width
for EHE investigations.30,35 The coercive fields inferred from these
EHE measurements agree well with the values estimated from the
magnetoresistance data. Figure 2.8(b) shows that the magnetoresistance
of such domain structures increases linearly with the number of domain
walls. Also included are data for structures with different irradiated
domain widths, which confirm that the resistance is independent of the
width of artificial domains, exactly as expected for intrinsic domain wall
magnetoresistance at the border between the irradiated and unirradiated
regions.
Figure 2.8(b) establishes the intrinsic domain wall origin of the
magnetoresistance. Magnetic domains were then patterned at various
angles with respect to the current flow in order to probe the
perpendicular and parallel domain wall resistivities. Multiple artificial
magnetic domains were defined, as described above, with the normal to
the wall subtending an angle θ in the range 0°–60° with respect to the
current direction. The geometry is shown in the inset of Fig. 2.10.
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 37
Fig. 2.9. 300 K magnetoresistance (offset vertically for clarity) of bridge structures
containing irradiated strips patterned at different angles relative to the current flow.
Adapted from Ref. 36.
For all samples the channel width was 1 µ m, and the width of the
irradiated artificial domains were kept invariant at 1 µ m. The 16 µ m
long Pt/Co/Pt wires imposed geometrical constraints on the number of
superlattice repeats that could be used. For θ = 0°, 20°, and 30° six
irradiated stripes were patterned, whereas for θ = 40°, 50° and 60° only
three irradiated stripes were possible. Therefore the magnetoresistance is
due to twelve domain walls for the former set of angles, and six for the
latter. Figure 2.9 illustrates the 300 K magnetoresistance of our bridge
devices with different tilt angles. A steady decrease in the maximum
resistance change is observed as the angle, θ, is increased. Also
noticeable is a systematic broadening of the resistance peaks at larger
angles. Since the domain width is the same in each case this can
only arise from changes in the reversal dynamics due to angle-dependent
38 S. Bending, S. Crampin and A. Aziz
Fig. 2.10. Calculated potential profile along the middle of the device channel containing a
single patterned strip 1 µ m wide inclined at different angles (lower inset). Also shown
are typical static eddy currents induced by the domain wall – a 30 nm region containing
the domain wall has been expanded horizontally to enable the current lines there to be
resolved. Adapted from Ref. 36.
changes in the magnetostatic interaction at the sharp and blunt corners of
tilted domains. This will, for example, lead to a distribution of de-
pinning fields for a domain wall trapped at the interface between an
irradiated and an unirradiated domain.
To understand the angular dependence of the magnetoresistance, we
model the current flow within the device channel, treating the problem as
being two-dimensional. In steady state, the continuity equation requires
that div J = 0, and from Maxwell’s equations curl E = 0, where the current
density J and electric field E are related via the resistivity tensor E = ρ J.
On account of the high resistivity of our transport devices we neglect off-
diagonal terms of ρ within each region, which is either (i) unirradiated,
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 39
with ρ xx = ρ yy = ρ 0 (we take x to be along the device channel), (ii)
irradiated, with ρ xx = ρ yy = ρ I, or (iii) a domain wall, for which
cos θ sin θ ρ CPW 0 cosθ − sin θ
ρ DW = , (2)
− sin θ cosθ 0 ρ CIW sin θ cos θ
with ρ CPW and ρ CIW the resistivities for current flow perpendicular and
parallel to the domain wall respectively.17 We solve the equations
numerically in terms of a stream function, ψ , with J = (∂ψ /∂y, −∂ψ /∂x),
assuming uniform current J = J0 = (J0,0) at each end of the device channel
and no current flow through the sides. The equations are discretised on a
piecewise uniform finite difference grid, and solved iteratively using
multigrid relaxation with continuity of the normal current and tangential
electric field boundary conditions at each interface. Iterations are
terminated when the residual is less that 10-10, and the potential V along
the device channel then calculated from E = −gradV.
Figure 2.10 shows typical results for the change in the potential
along the device channel created by the presence of domain walls, for
different angles of the irradiated region. As well as a sharp increase
across the domain wall itself, the potential also displays an increase on
both sides, which is due to the formation of static eddy currents, shown
in the inset. The size and range of this contribution increases with the
angle θ, but our simulations confirm that for the structures investigated
here the potential change is fully contained within the device channel and
not responsible for the changes in Fig. 2.9.
To understand the longitudinal channel resistance, we note that
the additional resistivity due to the domain wall, δρ CPW = ρ CPW − ρ 0,
δρ CIW = ρ CIW − ρ 0, is small. Neglecting the difference in the resistivity
of the irradiated and unirradiated regions (which changes the final
result by at most a few percent) the electric field can be written as
E = ( ρ0 + δρ)(J0 + δJ ) so that to first order δEx = δρ xx J0 + ρ 0δJx. Since
δJx integrates to zero along the channel, the additional resistance may be
attributed to δρ xx and hence reflects the angular variation of the
resistivity tensor in Eq. (2), which in general contains both CPW and
CIW contributions:
40 S. Bending, S. Crampin and A. Aziz
Fig. 2.11. Angular dependence of the total domain wall resistivity; diamonds: measured
values from (a); circles: calculations such as in (b), using δρ CPW = 23 × 10-3 µ Ω-cm,
δρ CIW = 3.5 × 10-3 µ Ω-cm, DW width 15 nm; line: Eq. (3) using the same values.
Adapted from Ref. 36.
δρ DW = δρ xx = (δρ CPW − δρ CIW ) cos 2 θ + δρ CIW . (3)
In Fig. 2.11 we plot the domain wall resistivity, derived from curves
of the type shown in Fig. 2.9, against cos2θ. Also plotted are the results
of our numerical simulations. Both measured and simulated results are
accounted for very well by Eq. (3). In order to derive δρ DW from our
measurements we have taken the domain wall width in our 0.6 nm thick
Co layer to be 15 nm,37 and assume a parallel resistor model for our
trilayer films with resistivities uniformly scaled from the bulk values to
give the correct total resistance.
Unlike previous attempts to study the anisotropy of domain wall
resistivity,38,39 which used naturally occurring reasonably well-ordered
striped domain structures, these are the first measurements to exploit
precise artificial control of domain wall orientation. This eliminates
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 41
much of the uncertainty involved in analysis of the observed angular
dependence. For example, it is not necessary to estimate the relative
proportion of domain walls aligned perpendicular to the current.39 As
expected from Eq. (3) and the theory of Levy and Zhang, which predicts
δρ CPW > δρ CIW, we find that the domain wall resistivity remains positive
for all angles, with values for the perpendicular and parallel domain wall
resistivities that may be inferred from our measurements of δρ CPW =
23.1 ± 1.1 × 10-3 µ Ω-cm and δρ CIW = 3.5 ± 0.7 × 10-3 µ Ω-cm respectively.
We note that the results in Fig. 2.11 also limit the magnitude of any non-
Hall-like contribution to the domain wall resistivity tensor – which has
been assumed to be zero in Eq. (2) but which, if present, would make a
contribution varying like sin 2θ – to less than 0.5 × 10-3 µ Ω-cm. The
magnetoresistivity ratio MRCPW = δρ CPW/ρ 0 is 0.1%, which is comparable
to values deduced for domain walls in other metallic systems.38 The ratio
of CPW to CIW resistivities is δρ CPW/δρ CIW ≈ 6.6 ± 1.2; according to LZ
theory this is 3 + 10 ρ ↓ / ρ ↑ / ( ρ ↓ / ρ ↑ + 1) where ρ↓/ρ ↑ is the ratio of
minority to majority spin carrier resistivities. The value ρ↓/ρ ↑ ~ 5.5 which
we deduce is greater than in bulk Co (ρ↓/ρ ↑ ~ 4),41 but consistent with
the Fermi level density of states ratio that we find in band structure
calculations of a Pt/3ML Co/Pt sandwich structure, where an
enhancement of the minority-spin density of states arises due to band-
narrowing.42
2.5. Conclusions and Outlook
We have described the magnetic properties of a new generation of
magnetic devices in which artificial magnetic domains are introduced
into perpendicularly magnetised Pt (3.5 nm)/Co (0.5 nm)/Pt (1.6 nm)
trilayer structures by local Ga focussed ion beam irradiation. The
switching properties and size-scaling of the dosed regions has been
studied using the extraordinary Hall effect, allowing probable reversal
mechanisms to be established. Discrete steps are observed in the minor
loops of dosed regions which allow a direct estimate of the Barkhausen
volume. Artificial domain walls are created at the border of dosed and
undosed regions when an external magnetic field switches the former but
not the latter, and this property has been used to create stripe-like
42 S. Bending, S. Crampin and A. Aziz
domains. This approached has been exploited in a quantitative study of
angle-dependent intrinsic domain wall resistivity in artificial magnetic
superlattice structures. For the first time we are able to show that our
experimental results are in truly quantitative agreement with a theory of
domain wall resistivity based on spin-dependent potentials and scattering
rates for majority and minority spin electrons due to Levy and Zhang.17
This model system also provides exciting opportunities for studying
and exploiting domain wall physics, such as current-induced domain wall
motion. The ability to control the exact domain structure that forms in
nanowires and suppress extrinsic magnetoresistance effects eliminates
difficulties that have plagued the domain wall resistance field for many
years. These measurements have demonstrated that an unambiguous
smooth mapping of the domain wall resistivity, from that for current
flow perpendicular to the wall to that for current flow parallel, can
be observed in carefully designed samples containing tilted domains.
A natural development will be to study whether the magnitude
of magnetoresistive effects can be enhanced by fabricating
nanoconstrictions within the nanowire, or whether there are important
effects arising from reducing the wire width/domain width ratio for tilted
domains. Multiple-terminal “mesoscopic” measurements to probe
transverse resistance within tilted domains will provide further insight
into the validity of microscopic theories of domain wall scattering.
Finally, our room temperature measurements point to applications in
practical devices – one can envisage, for example, using the domain wall
angle as a control parameter in multiple-state logic devices based upon
spin-torque transfer.
Acknowledgments
This work was supported by the Leverhulme Trust, through grant no.
F/00 351/F. We are grateful to Dr Peter Heard for giving us access to the
Ga-FIB in the Interface Analysis Centre, University of Bristol, and for
assistance with sample irradiation. We would also like to thank Dr Chris
Marrows at the School of Physics and Astronomy, University of Leeds,
for supplying the sputtered Pt/Co/Pt trilayers for this work.
Artificial Magnetic Domain Structures Realized by Focussed Ion Beam Irradiation 43
References
1. L. Berger, J. Appl. Phys. 55, 1954 (1984); L. Berger, J. Appl. Phys. 71, 2721 (1992);
L. Berger, J. Phys. Chem. Solids 35, 947 (1974).
2. J. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996).
3. J. Grollier et al., Appl. Phys. Lett. 78 (2001); J. Grollier et al., Appl. Phys. Lett. 83,
509 (2003).
4. M. Klaui et al., Appl. Phys. Lett. 83, 105 (2003).
5. A. Yamaguchi et al., Phys. Rev. Lett. 92, 077205 (2004).
6. M. Hayashi et al., Phys. Rev. Lett. 96, 197207 (2006); M. Hayashi et al., Phys. Rev.
Lett. 97, 207205 (2006); L. Thomas et al., Nature 443, 197 (2006); M. Hayashi
et al., Nature Physics 3, 21 (2007); M. Hayashi et al., Phys. Rev. Lett. 98, 037204
(2007).
7. J. Fassbender et al., J. Phys D: Appl. Phys. 37, R179 (2004).
8. S. Hashimoto et al., J. Appl. Phys. 67, 4429 (1990).
9. C. Chappert et al., Science 280, 1919 (1998); T. Devolder et al., Appl. Phys. Lett. 74,
3383 (1999).
10. T. Devolder, Phys. Rev. B 62, 5794 (2000).
11. T. Devolder et al., C. R. Acad. Sci. Paris, t. 327 (Ser. II b), 915 (1999).
12. G. G. Cabrera and L. M. Falicov, Phys. Status Solidi (b) 61, 59 (1974); Phys. Status
Solidi (b) 62, 217 (1974).
13. L. Berger, J. Appl. Phys. 49, 2146 (1978).
14. S. S. P. Parkin, Phys. Rev. Lett. 71, 1641 (1993).
15. T. Lalet and A. Fert, Phys. Rev. B 48, 7099 (1993).
16. S. S. P. Parkin in Magnetic Surfaces, Thin Films and Multilayers, p. 211 (1992)
(ed. S. S. P. Parkin, J.-P. Renard, T. Shinjo and W. Zinn).
17. P. M. Levy and S. Zhang, Phys. Rev. Lett. 79, 5110 (1997).
18. G. Tatara and H. Fukuyama, Phys. Rev. Lett. 78, 3773 (1997).
19. R. P. van Gorkom, A. Brataas, and G. E. W. Bauer, Phys. Rev. Lett. 83, 4401 (1999).
20. K. Hong and N. Giordano, Phys. Rev. B 51, 9855 (1995).
21. J. F. Gregg et al., Phys. Rev. Lett. 77, 1580 (1996).
22. Y. B. Xu et al., Phys. Rev. B 61, R14901 (2000).
23. A. D. Kent, J. Yu, U. Rüdiger, and S. S. P. Parkin, J. Phys.: Condens. Matter 13,
R461 (2001).
24. R. Danneau et al., Phys. Rev. Lett. 88, 157201 (2002).
25. I. A. Campbell and A. Fert in Ferromagnetic Materials, Vol 3 (1992) (ed. E. P.
Wohlfarth, North-Holland, Amsterdam).
26. L. Berger, Phys. Rev. B 2, 4559 (1970).
27. Y. I. Mankov, Sov. Phys.-Solid State 14, 62 (1972).
28. C. Vieu, J. Gierak, H. Launois, T. Aign, P. Meyer, J. P. Jamet, J. Ferré, C. Chappert,
T. Devolder, V. Mathet, and H. Bernas, J. Appl. Phys. 91, 3103 (2002).
44 S. Bending, S. Crampin and A. Aziz
29. J. F. Zielgler, J. Biersack, and U. Littmark, computer software SRIM (2003).
Available from www.srim.org.
30. Aziz, S. J. Bending, H. G. Roberts, S. Crampin, P. J. Heard, and C. H. Marrows,
J. Appl. Phys. 98, 124102 (2005).
31. L. Canedy, X. W. Li, and G. Xiao, J. Appl. Phys. 81, 5367 (1997).
32. H. J. G. Draaisma, W. J. M. de Jonge, and F. J. A. Broeder, J. Magn. Magn. Mater.
66, 351 (1987).
33. V. Mathet, T. Devolder, C. Chappert, J. Ferre, S. Lemerle, L. Belliard, and
G. Guentherodt, J. Magn. Magn. Mater. 260, 295 (2003).
34. T. Devolder, Phys. Rev. B 62, 5794 (2000).
35. A. Aziz, S. J. Bending, H. G. Roberts, S. Crampin, P. J. Heard, and C. H. Marrows,
J. Appl. Phys. 99, 08C504 (2006).
36. A. Aziz, S. J. Bending, H. G. Roberts, S. Crampin, P. J. Heard, and C. H. Marrows,
Phys. Rev. Lett. 97, 206602 (2006).
37. I. Knittel and U. Hartmann, J. Magn. Magn. Mater. 294, 16 (2005); U. Rudiger, J.
Yu, L. Thomas, S. S. P. Parkin, and A. D. Kent, Phys. Rev. B 59, 11914 (1999).
38. L. Klein et al., Phys. Rev. Lett. 84, 6090 (2000); M. Feigenson, L. Klein, J. W.
Reiner, and M. R. Beasley, Phys. Rev. B 67, 134436 (2003).
39. M. Viret et al., Phys. Rev. Lett. 85, 3962 (2000).
40. C. H. Marrows, Adv. Phys. 54, 585 (2005); A. D. Kent, J. Yu, U. Rudiger, and
S. S. P. Parkin, J. Phys.: Condens. Matter 13, R461 (2001).
41. D. A. Papaconstantopoulos, Handbook of The Band Structure of Elemental Solids
(Plenum Press, New York, 1986).
42. H. G. Roberts, PhD Thesis, University of Bath (2008).
Chapter 3
FABRICATION OF MAGNETIC NANOSTRUCTURES BY
ELECTRON BEAM INDUCED DEPOSITION
Masaki Takeguchi
Nano-Characterization Center, National Institute for Materials Science
3-13 Sakura, Tsukuba 305-0003, Japan
E-mail: [email protected]
Masayuki Shimojo
Advanced Science Laboratory, Saitama Institute of Technology
1690 Fusaiji, Fukaya 369-0293, Japan
E-mail: [email protected]
Electron beam induced deposition (EBID) is one of the promising
techniques for fabrication of nanometer-sized structures with desired
shapes at desired positions. Constitution of the nanostructures formed
by EBID can be controlled by choosing the type of precursor element.
This chapter describes the recent progress of fabrication of magnetic
nanostructures using EBID.
3.1. Introduction
Magnetic nanostructures have attracted much attention because of
interesting properties of such nanometer-sized magnetic materials and
potential of application to not only ultrahigh density magnetic recording
but also new nanometer-sized magnetic devices such as read heads and
memory cells. The magnetic properties of the nanostructures strongly
depend on their shape and size, and therefore fundamental study of
magnetic properties of magnetic nanostructures with the desired shape is
a significant issue.1 For their application to the advanced magnetic
devices, an understanding of the switching behavior of the magnetic
nanostructures is important from the viewpoint of development of high
45
46 M. Takeguchi and M. Shimojo
density magnetic recording and magnetic random access memory. Their
spatial distribution is also crucial because magnetic interaction between
densely packed nanometer-sized magnetic elements such as rings and
dots via magneto-static stray fields influences the switching behavior.2
Thus, for both fundamental and application researches of magnetic
nanostructures the fabrication of various desired shaped magnetic
nanostructures at the desired positions has been required.3–6
Electron beam induced deposition (EBID) has become one of the
most promising methods for such size- and position-controllable
nanofabrication tools in point of view of high resolution, abundant types
of deposit elements, and capability of three dimensional fabrications.7–14
Constitution of the nanostructures formed by EBID can be controlled by
selecting the type of precursor element. When magnetic elements such
as iron and cobalt are employed as precursor in EBID, magnetic
nanostructures are fabricated. Now, applications of EBID to fabrication
of not only interconnections in semiconductor IC circuits and quantum
dot- and wire-arrays in advanced electronic but also magnetic devices
have been increasingly expected.
In EBID, metal-organic precursor gases are decomposed mainly by
secondary electrons from a substrate surface and resultantly metal
deposits are formed on the substrate surface. The formed deposits are
composed of an amorphous phase or a mixture of nanocrystals and an
amorphous phase. The amorphous phase contains a considerable amount
of carbon in addition to metal elements.15–17 This unfavorable carbon
comes from ligands of the dissociated organic precursors. Decreasing the
carbon concentration is a significant issue because the carbon rich
nanostructures often degrade their electric and magnetic properties. For
instance, the resistance of gold (or platinum)-containing nanowires
formed by EBID is quite larger than that of pure gold (or platinum)
wires.18,19 In the case of magnetic materials, although iron- and cobalt-
containing nanostructures were successfully produced by EBID with
iron and cobalt carbonyl gases (Fe(CO)5 and Co2(CO)8)16,20–25 and by
scanning tunneling microscope-assisted CVD with ferrocene gas
(Fe(C5H5)2),26 quantitative magnetic characterization of individual
nanostructures has rarely been carried out so far.
Fabrication of Magnetic Nanostructures by Electron Beam Induced Deposition 47
A postdeposition heat treatment is effective in the reduction of the
carbon content of the EBID-fabricated nanostructures. For instance, the
postdeposition heat treatment for the nanostructures formed with
Fe(CO)5 has demonstrated the transformation from a mixture of
amorphous carbon and iron phases to crystalline alpha iron without
significant shape change.22,25 Che et al. have reported that ferromagnetic
FePt nanostructures were obtained by EBID with a mixture gas of
Fe(CO)5 and (CH3)3(C5H5)Pt followed by the heat treatment.27 Mixing
water vapor with metal-organic precursors is also effective to decrease
the carbon content,28,29 and recently carbon free magnetite (Fe3O4)
nanostructures were successfully obtained by using a mixture gas of
Fe(CO)5 and water vapor.30
3.2. EBID Fabrication
EBID has been carried out using scanning electron microscopes (SEMs)
and transmission electron microscopes (TEMs). The deposition is
produced mainly by low-energy electrons, such as secondary electrons,
and their spatial distribution becomes wider as the primary electron
energy increases. It was thought that the resulting deposit would be
larger for higher energy electrons, although the probe size of the electron
beams decreases. Tanaka et al.14 and Van Drop et al.9 succeeded in
fabrication of nanodots with a few nanometers in diameter using a
Fig. 3.1. Schematic drawing of a 30 kV FEG-SEM equipped with an introduction system
for fabrication of the nanostructures by EBID.
48 M. Takeguchi and M. Shimojo
200 keV electron beam in transmission electron microscope equipped
with a field emission gun.
A 30 kV ultrahigh vacuum scanning electron microscope equipped
with a field emission gun (JEOL JSM-7800FV), of which base pressure
is about 2 × 10−6 Pa, was employed to carry out the EBID fabrication in
the present work. A schematic drawing of the deposition experiment is
shown in Fig. 3.1. A gas introduction system installed into the SEM is
combined with two sets of gas source reservoirs and pipelines through
leak valves to control the partial pressure of precursor gases. The
precursor gases are introduced into a specimen chamber of the SEM
through a nozzle with an inner diameter about 0.2 mm. Two kinds of
gases are introduced simultaneously and their partial pressures can be
independently controlled. The nozzle tip is about 1 mm from the electron
beam position. In the present work, the nanostructures were fabricated on
edges of carbon microgrid films, silicon films and molybdenum films at
room temperature. The silicon and molybdenum thin areas were prepared
by using a focused ion beam thinning system (JEOL JEM-9310FIB). The
nominal electron beam current was about 0.8 nA with a beam diameter
of 4 nm. The beam position was controlled by an external deflector
voltage input using a computer with digital-analog converters. In order to
fabricate free-standing nanostructures, the electron beam position was
moved from the substrate into an open space.
3.3. Fabrication of Iron-Containing Nanostructures
Figure 3.2(a) shows a transmission electron microscopy (TEM) image of
nanodots and free-standing nanorods formed on a carbon microgrid film
at room temperature. The nanorods were fabricated by moving the beam
from the carbon film edge to an open space at a speed of 2 nm/s.
Arbitrary-shaped nanostructures were also fabricated by the subsequent
electron beam patterning on the preliminarily formed nanorods.
Figure 3.2(b) shows a TEM image of a square frame structure. The
linewidths of the rods and frames are 30–50 nm. Higher-magnification
TEM observation of the free-standing rods and frames showed that the
surface was covered with nanocrystals and that their inside was
amorphous. The diffraction pattern and electron energy loss spectroscopy
Fabrication of Magnetic Nanostructures by Electron Beam Induced Deposition 49
Fig. 3.2. TEM images of (a) nanodots and nanorods and (b) a free-standing square frame
fabricated by EBID with Fe(CO)5. They were fabricated on a carbon microgrid film using
a 30 keV fine electron probe of 4 nm diameter in FEG-SEM. (c) Electron diffraction
pattern taken from area indicated by circle in (b). (d) EELS spectrum from the same area
specified by a circle in (b). Reprinted with permission from Ref. 17 by the Japan Society
of Applied Physics.
(EELS) result from the region in the circle in Fig. 3.2(b) are indicated in
Figs. 3.2(c) and 3.2(d), respectively.
The EELS spectrum indicates that the free-standing rods and frames
were composed of iron, carbon and oxygen. Diffraction indices and
corresponding ring positions for alpha-iron and possible iron oxides are
inserted in Fig. 3.2(c). Their positions are in agreement with the observed
ring positions. Shimojo et al. have reported that a carbon K-edge showed
a characteristic of amorphous carbon but not of iron carbide in their
detailed EELS analysis.16 Hence, it is deduced that some faint
polycrystalline-like rings originated from the iron oxide nanocrystals
existing near the surface and that broad rings were from the amorphous
phases in their body. The existence of surface oxides may be mainly due
50 M. Takeguchi and M. Shimojo
to the exposure of the specimen to air during the transportation from
SEM to TEM machine.
With increasing the scanning speed, the self-standing nanorods
became more slender. The deposit shape is determined by secondary
electrons emitted from the substrate. After the incident electron beam is
moved away from the substrate edge, the self-standing nanorod can be
grown by secondary electrons generated from the as-formed nanorod
itself. Resultantly, the self-standing nanorod maintains an almost
constant width throughout its length. It should be noted that the aspect
ratio (e.g., thickness to linewidth) of the nanostructures formed by EBID
must be considerably large.31 In the case of the nanostructures in Fig. 3.2,
the thickness was estimated to be 150–250 nm.
It has been reported that the as-deposited nanostructures on a carbon
film contained a large amount of carbon (iron and carbon was
approximately equivalent) and transformed to iron carbide by the heating
process.16,17 On the other hand, the carbon concentration of the
nanostructures on Si substrates was rather smaller than that on the carbon
films.17 Figure 3.3 shows a TEM image of free-standing nanostructures
formed on a Si substrate. EELS analysis showed that the carbon
concentration in the nanorods is less than 50%.
It is known that electron holography is a powerful tool to observe
electric and magnetic field inside and outside of the materials
quantitatively with high spatial resolution and applicable to
characterization of magnetic fields of the arrays and chains of magnetic
nanostructures. Electron holography was carried out to observe magnetic
fields around the nanostructure in Fig. 3.3 using a JEOL JEM-3000F
Fig. 3.3. TEM image of the desired-shape self-standing nanostructures on a thin silicon
substrate edge formed by EBID with Fe(CO)5 at room temperature. Reprinted with
permission from Ref. 11 by IOP Publishing Ltd.
Fabrication of Magnetic Nanostructures by Electron Beam Induced Deposition 51
equipped with an electron biprism consisting of a centered wire electrode
and grounded electrodes at both the sides. The electron wave passing
through the sample and vacuum regions is deflected by the electrostatic
potential around the wire electrode so that an interference pattern is
formed in an image plane. This interference pattern is called “electron
hologram”, in which phase information of the object passing through the
sample can be recorded. By reconstructing the electron hologram
digitally by a computer, a phase image was obtained in the standard
manner described in literature.36 Figure 3.4(a) shows an electron
hologram of the nanostructures. Figure 3.4(b) shows a phase image
reconstructed from Fig. 3.4(a), displayed as a COS intensity, where the
phase was amplified by a factor of 4. This phase image is so called
interference micrograph, and one can recognize dark lines as magnetic
fluxes. The transition between the adjacent dark lines corresponds to a
phase shift of π/2 (1.57) rad.
The phase difference of the electron wave passing through both sides
of the nanorod is described as
e
∆ϕ = 2π BS (1)
h
where S is a cross section area of the nanorod. The external magnetic
stray field above and under the nanorod oriented in opposite direction to
the internal magnetic field should be taken into account to estimate the
Fig. 3.4. (a) Electron hologram and (b) reconstructed phase image displayed as a COS
intensity, so-called “interference micrograph”, where the phase was amplified by a factor
of 4.
52 M. Takeguchi and M. Shimojo
accurate magnetic flux in the nanorod,37 because the observed phase
difference between the both sides of the nanorod is actually the sum of
the phase shift due to the internal magnetic field and the vertical
integration of the external magnetic stray fields at the nanorod position.
The phase shift due to the stray fields at the nanorod position was
extrapolated approximately from the phase distribution near the nanorod.
As shown in Fig. 3.5, the phase difference between both the sides of the
nanorod with a ring-head, indicated by an arrow in Fig. 3.3, was
measured to be about 7.3 rad. The phase change due to the external
magnetic stray field at the nanorod position was extrapolated
approximately from the phase inclination near the nanorod, and found to
be about 1.5 rad. Therefore, the phase shift due to the internal magnetic
field was estimated to be about 8.8 rad, which corresponds to the
magnetic flux of 5.75 × 10−15 Wb. The shape of the cross-section of the
nanorod was supposed to be elliptic. The thickness and width of the
nanorod was estimated to be about 250 nm and 40 nm, respectively.
Thus, the elliptical-shape cross-section of the nanorod was about
7.85 × 10−15 m−2, and finally the remanent magnetic flux density Br was
calculated to be about 0.7 T.
The compositional ratio of metal to carbon in EBID nanostructures
significantly influences magnetic properties. Nanostructures with the
different compositional ratio of iron to carbon were formed using a
mixture gas of Fe(CO)5 and ferrocene, Fe(C5H5)2. Stoichometrically, the
ratio of iron to carbon of Fe(CO)5 is twice as many as that of Fe(C5H5)2.
Fig. 3.5. (a) Electron holograms, (b) corresponding COS-intensity-displayed phase
images and (c) line profiles of the phase distribution across the nanostructure, which is
indicated by an arrow in Fig. 3.3. Reprinted with permission from Ref. 11 by IOP
Publishing Ltd.
Fabrication of Magnetic Nanostructures by Electron Beam Induced Deposition 53
Fig. 3.6. (a) SEM image of the EBID fabricated nanorods formed on an edge of a
molybdenum thinned area. (b) The result of EDS analysis, showing Fe-L intensity for
each nanorod. Reprinted with permission from Ref. 24 by Springer.
The iron to carbon compositional ratio in the nanostructures was
controlled by changing the partial pressure of these gases. Figure 3.6(a)
shows a SEM image of nanorods (numbered from 1–10) formed by
EBID with various partial pressures of iron carbonyl and ferrocene gases
on an edge of a molybdenum thinned area. First, the partial pressure of
ferrocene gas was maintained at 1 × 10−5 Pa. Then, carbonyl gas was
introduced so that the total pressure was increased from 3 × 10−5 Pa to
12 × 10−5 Pa by 1 × 10−5 Pa for each EBID fabrication of nanorod (totally
10 nanorods were formed). Figure 3.6(b) shows iron-L peak intensity for
each nanorod acquired by EDS analysis, indicating that the iron
concentration was successfully increased depending on the change of the
partial pressure of carbonyl. Carbon-K peak was hardly detected because
of insensitivity and poor resolution of the current EDS detector. Previously
reported EELS analysis showed that iron, carbon and oxygen contained
in the nanorod formed by EBID with only iron carbonyl on the same
beam condition as the present experiment was about 70, 10 and 20 at%,
respectively.17 Roughly assuming that the iron concentration of No. 8–10
nanorods is 70 at% and the thickness of all the nanorods is the same, the
iron concentration of No. 1 nanorod was approximated to be 30%.
54 M. Takeguchi and M. Shimojo
Fig. 3.7. (a) Electron hologram of the No. 3–6 nanorods. (b) Phase image (interference
micrograph) reconstructed from (a) with an amplification factor of 4. (c) Lline profile of
the phase distribution across the line AA’ in the No. 6 nanorod. (d) The remanent
magnetic flux density Br for the No. 1–10 nanorods. Reprinted with permission from
Ref. 24 by Springer.
The remanent magnetic flux density Br for each nanorod was
evaluated by electron holography. Figures 3.7(a) and 3.7(b) show an
electron hologram of the No. 3–6 nanorods and the corresponding
reconstructed phase image, respectively. A line profile of the phase
across the line AA’ in the No. 6 nanorod is shown in Fig. 3.7(c).
Figure 3.7(d) shows the Br values for the No. 1–10 nanorods, indicating
that a fitting curve (solid line) described in the figure is quite similar to a
plotted curve (iron concentration) in Fig. 3.6(b), showing that the
remanent magnetic flux density Br of the EBID nanostructures containing
iron and carbon is related to the iron concentration.
3.4. Post-Deposition Heat Treatment: Fabrication of Alpha Iron
Nanostructures
It has been reported that substrate heating during deposition results in an
enhancement of the decomposition rate, and the nanostructures
Fabrication of Magnetic Nanostructures by Electron Beam Induced Deposition 55
composed of nanocrystals could be formed (namely, metal concentration
and crystallinity could be improved).20,32,33 Post-deposition heat treatment
for chromium-, tungsten- and rhenium-containing nanostructures has
been reported; after annealing at about 800°C, chromium- and tungsten-
containing nanorods showed the appearance of nanocrystals within them
without change in their original shape, while rhenium-containing
nanorods transformed to crystalline structures with a large grain size but
with change in their shape.34 Here, we attempt the post-deposition heat
treatment to iron-containing nanostructures. The original as-deposited
iron and carbon compositional ratio are crucial for the stoichiometry of
the productions after the heat treatment. By heating in vacuum, carbon
reacts with oxygen to become CO2 molecules, and the CO2 molecules are
desorbed. If the remaining carbon content is large, carbon reacts with
iron to form iron carbides and a suitable phase appears according to a
phase diagram. It has been reported that the as-deposited nanostructures
on a carbon film contains a large amount of carbon (iron and carbon are
approximately equivalent) and therefore transforms to iron carbide
during the heating process.16,17 On the other hand, the carbon
Fig. 3.8. TEM images of the free-standing nanostructures on a silicon substrate before (a)
and after (b) a heat treatment at about 600°C for 2 h. Reprinted with permission from
Ref. 11 by IOP Publishing Ltd.
56 M. Takeguchi and M. Shimojo
Fig. 3.9. (a) Dark-field TEM image of the nanostructure labeled by B in Fig. 8(b), taken
using a (110) diffraction spot at the incidence of <111>a-Fe. (b) Selected area diffraction
pattern from the nanostructure in (a). (c) A high-resolution TEM image of the apex part
of the nanostructure in (a). Reprinted with permission from Ref. 11 by IOP Publishing
Ltd.
concentration of the nanostructures on the Si substrate is rather smaller
than that on the carbon film, resulting in the formation of alpha iron.17
Figures 3.8(a) and 3.8(b) show TEM images of nanostructures on a
silicon film edge before and after the postdeposition heating at about 600°C
for 2 h, respectively. On heating, the shape of the nanostructures was
maintained except for the following: the linewidth near the root became
larger and the inner diameter of the rings was smaller. The tilting of the
nanostructures is supposed to be due to the bending of the thin silicon
film edge. All the nanostructures in this sample transformed into an
alpha-iron phase in their entire body or a part of the body by the heating.
Figure 3.9(a) show a dark-field TEM image of the nanostructure labelled
by B in Fig. 3.8(b), taken using a (110) diffraction spot at the incidence
of <111>a−Fe. The selected area diffraction pattern from the nanostructure
shown in Fig. 3.9(a) is indicated in Fig. 3.9(b), proving that the whole of
the body was a single crystal. Figure 3.9(d) shows a high-resolution
TEM image of the apex part of the nanostructure.
Fabrication of Magnetic Nanostructures by Electron Beam Induced Deposition 57
Fig. 3.10. (a) SEM and (b) HRTEM images of the nanorod formed by EBID with a
mixture of Fe(CO)5 and water vapor. Nano beam electron diffraction is inserted in (b).
(c,d) EELS spectra of nanorods formed withoutand with water vapor, respectively.
Reprinted with permission from Ref. 40 by IOP Publishing Ltd.
3.5. EBID with Fe(CO)5 and Water Vapor: Fabrication of
Magnetite Nanostructures
Iron oxide, especially magnetite (Fe3O4), nanostructures including
nanowires and nanodots are also important in spin electronic devices
because of their half-metallic ferrimagnetic nature.39 We have succeeded
in the direct formation of Fe3O4 nanostructres at room temperature by
EBID with a mixture of Fe(CO)5 and water vapor as precursor.40,41
Water vapor was precisely mixed with an iron carbonyl gas and the
mixture was introduced in an EBID SEM chamber as well as the
experiment of EBID with a mixture of Fe(CO)5 and ferrocene.
Figure 3.10(a) shows a SEM image of a nanorod formed on an edge
of a thinned molybdenum substrate using an Fe(CO)5 precursor mixed
with water vapor. The nanorod was about 300 nm in length and 20–30 nm
in width. Figure 3.10(b) shows a HRTEM image of the middle part
of the nanorod, in which nanobeam electron diffraction pattern is
inserted. These indicate the high crystalinity of the deposit. The nanorod
58 M. Takeguchi and M. Shimojo
Fig. 3.11. (a) SEM image of a double-ring structure, showing the flexibility in shaping of
EBID and (b) HRTEM image of a part of the ring. Reprinted with permission from
Ref. 40 by IOP Publishing Ltd.
was polycrystalline and the grain size was relatively large comparing
with the width of the nanorod. Figures 3.10(c) and 3.10(d) show EELS
spectra of the nanorods with and without mixing water vapor. Iron,
oxygen and carbon edges are observed in the spectrum of the nanorod
fabricated without water vapor, while no carbon edges are seen in that
with water vapor.
EELS analysis for the nanorods formed by various [H2O]/[Fe(CO)5]
partial pressure ratio showed that oxygen content increases and carbon
content decreases with increasing the [H2O]/[Fe(CO)5] partial pressure
ratio. The microstructure still consisted mainly of amorphous carbon at
the partial pressure ratio of 0.4, and carbon content reached substantially
naught and atomic ratio of Fe and O was 1: 1.3 ± 0.2 (slightly fluctuated
at positions) in the deposit when the partial pressure ratio was larger than
1.0.
Comparing the diffraction pattern shown in Fig. 3.10(b) with those of
known iron oxides, it was identified to be Fe3O4 (magnetite, lattice
constant a = 0.84 nm, space group of Fd3m) with an incident beam
direction of [1, 2, −1]. (Note that diffraction spots from another crystal
are overlapped.) This is consistent with the compositional analysis,
described above, which was Fe : O = 1 : 1.30.
Fabrication of Magnetic Nanostructures by Electron Beam Induced Deposition 59
Figure 3.11(a) shows a SEM image of a double-ring nanostructure
formed at the partial pressure ratio of 1.0. The electron beam position,
which was controlled using a computer, was moved from an edge of the
substrate into space to make a rod, followed by being circled to make
a ring. This procedure was repeated to make a double ring structure.
Figure 3.11(b) shows a HRTEM image of a part of the ring. This was of
polycrystalline but the grain size was a few nanometers, which was
smaller than that in the straight rod shown in Fig. 3.10. This implies that
some grains having a preferential growth direction along with the beam
moving direction, but no grains can grow larger when the electron beam
traces on a curved line.
3.6. Summary
It is demonstrated that magnetic nanostructures with the desired shapes at
the desired positions can be fabricated by EBID technique. We believe
that the EBID can be one of the most promising methods for
nanofabrication tools in point of view of high resolution, abundant types
of deposit elements, and capability of three dimensional fabrications.
Acknowledgments
A part of this work was performed with Drs. K. Mitsuishi, M. Tanaka
and K. Furuya of National Institute for Materials Science. This work was
partly supported by KAKENHI 18510103 and 20560028.
References
1. D. J. Smith, R. E. Dunin-Borkowski, M. R. McCartney, B. Kardynal and M. R.
Scheinfein, J. Appl. Phys. 87 (2000) 7400.
2. U. Welp, V. K. Vlasko-Vlasov, G. W. Crabtree, J. Hiller and N. Zaluzec, J. Appl.
Phys. 93 (2003) 7056.
3. S. P. Li, M. Natali, A. Lebib, A. Pepin, Y. Chen and Y. B. Xu, J. Magn. Magn.
Mater. 241 (2002) 447.
4. Y. Otani, B. Pannetier, J. P. Nozieres and D. Givord, J. Magn. Magn. Mater. 126
(1993) 622.
60 M. Takeguchi and M. Shimojo
5. A. Sugawara, T. Coyle, G. G. Hembree and M. R. Scheinfein, Appl. Phys. Lett. 70
(1997) 1043.
6. K. Liu, J. Nogues, C. Leighton, H. Masuda, K. Nishio, I. V. Roshchin and I. K.
Schuller, Appl. Phys. Lett. 81 (2002) 4434.
7. H. W. P. Koops, J. Kretz, M. Rudolph, M. Weber, G. Dahm and K. L. Lee, Jpn. J.
Appl. Phys. 33 (1994) 7099.
8. I. Utke, A. Luisier, P. Hoffmann, D. Laub and P. A. Buffar, Appl. Phys. Lett. 81
(2002) 3245.
9. W. F. Van Dorp, B. Van Someron, C. W. Hagen, P. Kruit and P. A. Crozier, Nano
Lett. 5 (2005) 1303.
10. K. Mitsuishi, M. Shimojo, M. Han and K. Furuya, Appl. Phys. Lett. 83 (2003) 2064.
11. M. Takeguchi, M. Shimojo and K. Furuya, Nanotechnol. 16 (2005) 1321.
12. M. Shimojo, K. Mitsuishi, A. Tameike and K. Furuya, J. Vac. Sci. & Technol. B 22
(2004) 742.
13. K. Mitsuishi, Z. Q. Liu, M. Shimojo, M. Han and K. Furuya, Ultramicrosc. 103
(2005) 17.
14. M. Tanaka, M. Shimojo, K. Mitsuishi and K. Furuya, Appl. Phys. A 78 (2004) 543.
15. M. Han, K. Mitsuishi, M. Shimojo and K. Furuya, Philos. Mag. 84 (2004) 1281.
16. M. Shimojo, M. Takeguchi, M. Tanaka, K. Mitsuishi and K. Furuya, Appl. Phys. A
79 (2004) 1869.
17. M. Takeguchi, M. Shimojo and K. Furuya, Jpn. J. Appl. Phys. 44 (2005) 5631.
18. H. W. P. Koops, A. Kaya and J. J. Webe, Vac. Sci. Technol. B 13 (1995) 2400.
19. I. Utke, P. Hoffmann, B. Dwir, K. Leifer, E. Kapon and P. Doppelt, J. Vac. Sci.
Technol. B 18 (2000) 3168.
20. R. R. Kunz and T. M. Mayer, Appl. Phys. Lett. (1987) 50 962.
21. I. Utke, P. Hoffmann, R. Berger and L. Scandella, Appl. Phys. Lett. 80 (2002) 4792.
22. M. Takeguchi, M. Shimojo and K. Furuya, Nanotechnol. 16 (2005) 1321.
23. M. Takeguchi, M. Shimojo, R. Che and K. Furuya, J. Mater. Sci. 41 (2006) 2627.
24. M. Takeguchi, M. Shimojo, K. Mitsuishi, M. Tanaka, R. Che and K. Furuya,
J. Mater. Sci. 41 (2006) 4532.
25. M. Takeguchi, M. Shimojo and K. Furuya, Jpn. J. Appl. Phys. 44 (2005) 5631.
26. W. W. Pai, J. Zhang and I. F. Wendelken, J. Vac. Sci. Technol. B 15 (1997) 785.
27. R. Che, M. Takeguchi, M. Shimojo and K. Furuya Appl. Phys. Lett. 87 (2005)
223109.
28. A. Folch, J. Tejada, C. H. Peter and M. S. Wrighton, Appl. Phys. Lett. 66 (1995)
2080.
29. K. Molhave, D. N. Madsen, A. M. Rasmussen, A. Carlsson, C. C. Appel,
M. Brorson, C. J. H. Jacobsen and P. Boggild, Nano. Lett. 3 (2003) 1499.
30. M. Shimojo, M. Takeguchi and K. Furuya, Nanotechnol. 17 (2006) 3637.
31. Z. Q. Liu, K. Mitsuishi and K. Furuya, J. Appl. Phys. 96 (2004) 619.
32. A. D. Della Ratta, J. Melngailis and C. V. Thompson, J. Vac. Sci. & Technol. B 11
(1993) 2195.
Fabrication of Magnetic Nanostructures by Electron Beam Induced Deposition 61
33. H. W. P. Koops, C. Schossler, A. Kaya and M. Weber, J. Vac. Sci. & Technol. B 14
(1996) 4105.
34. N. A. Kislov, I. I. Khodos, E. D. Ivanov and J. Barthel, Scanning 18 (1996) 114.
35. S. Yamamuro, D. F. Farrell and S. A. Majetich, Phys. Rev. B 65 (2002) 224431.
36. P. A. Midgley, Micron 32 (2001) 167.
37. J. Biskupek, U. Kaiser, H. Lichte, A. Lenk, T. Gemming, G. Pasold and
W. Witthuhn, J. Magn. Magn. Mater. 293 (2005) 924.
38. M. R. McCartney and M. Gajdardziska-Josifovska, Ultramicrosc. 53 (1994) 283.
39. M. P. Nikiforov, A. A. Vertegel, M. G. Shumsky and J. A. Switzer, Adv. Mater. 12
(2000) 1351.
40. M. Shimojo, M. Takeguchi and K. Furuya, Nanotechnol. 17 (2006) 3637.
41. M. Shimojo, M. Takeguchi, K. Mitsuishi, M. Tanaka and K. Furuya, Jpn. J. Appl.
Phys. 46 (2007) 6247.
This page is intentionally left blank
Chapter 4
PREPARATION OF MAGNETIC
NANOPARTICLES USING CHEMICAL ROUTE AND
FUNCTIONALIZATION FOR MEDICAL APPLICATIONS
Yuko Ichiyanagi
Department of Physics, Yokohama National University
Hodogaya, Yokohama 240-8501, Japan
E-mail: [email protected]
Magnetic nanoparticles encapsulated in a silica cage were produced
in one step by a wet chemical method. Particles were prepared using
a mixture of metal chloride (MCl2 ·nH2O) aqueous solutions and
metasilicate nonahydrate (Na2SiO3 ·mH2O) aqueous solutions, in which
the metals chosen were Fe, Co, Ni, Mn, Zn, Mg, Ba, Ti and their plural
compounds. The obtained particles were monodispersive and the
diameters of these particles were estimated in a range of 2 to 34 nm.
The magnetic properties observed were ferromagnetic, paramagnetic,
antiferromagnetic and superparamagnetic depending on the composition
and annealing conditions. We functionalized these magnetic particles
by modifiying an amino group. The magnetic nanoparticles present
are surrounded by amorphous SiO2, so that Si ions were located on
the outer surface. This characteristic structure enables amino-silane
coupling. These functional magnetic nanoparticles are useful in medical
applications such as drug-delivery systems, MRI imaging and
hyperthermia treatment.
4.1. Introduction
Ever since the National Nanotechnology Initiative (NII) was declared in
the United States, interest in nanotechnology has been growing all over
the world. In particular, magnetism in nanoscopic systems has drawn a
great deal of attention due to the unique magnetic properties, as well as
63
64 Y. Ichiyanagi
the technological applications, of such systems. Magnetic nanoparticles
show peculiar characteristics, such as quantum size effects and quantum
magnetic tunneling.1–6 These interesting phenomena are observed
because a particle consists of a very small number of atoms, much
smaller than the Avogadro constant NA (NA = 6.02252 × 1023). On the
other hand, an observation and evaluation system for nano-sized
materials remains difficult to achieve and has not been established yet,
therefore the evaluation method needs to be improved.
Previously, we obtained monodispersive magnetic nanoparticles
(MNPs) with characteristic magnetic properties by an original wet
chemical method, and reported their magnetic, structural and thermal
properties.7–27 Magnetic nanoclusters surrounded by the host network of
amorphous SiO2 were produced by mixing aqueous solutions of
3d transition-metal chlorides (MCl2 ·nH2O) and sodium metasilicate
nonahydrate (Na2SiO3 ·9H2O) to obtain monodisperse nanoparticles with
diameters of less than 10 nm in a single step. The particle size can be
controlled by adjusting the annealing temperature and pH value of
the solutions. The obtained particle sizes were between 2 and 34 nm.
Magnetic particles below a critical diameter cannot support more than
one domain, and form a single domain. This critical diameter was usually
observed for typical materials between 10 and 100 nm.28–31 A single
domain of ferromagnetic particles has a temporary large coercive force at
a certain domain size, however, for the smaller particles it is not stable
because the magnetic spins of the particle are believed to fluctuate
against the thermal energy.
We also prepared MNPs functionalized with surface amino groups
by means of an amino-silane coupling procedure to covalently bond
target molecules to the particles.32–36 Due to the extremely small size of
the particles, they can be introduced into cells without a cationic coating,
and could even have attached to them molecules such as amino group,
antibodies or drugs. We proposed that MNPs can be localized in living
tissue under the influence of an external magnetic field. In addition, cell
selective functional MNPs were also prepared, being modified with folic
acid.34 These MNPs have successfully been internalized in human cancer
cells. A cellular recognition system based on functional MNPs could be
developed for cell-specific delivery systems.
Preparation of Magnetic Nanoparticles Using Chemical Route 65
In this chapter, preparation, characterization and magnetic properties
of several kinds of nanoparticles will be introduced and development for
biomedical applications will be mentioned.
4.2. Synthesis and Characterization of Magnetic
Nanoparticles
We prepared magnetic nanoparticles (MNPs) surrounded by the host
network of amorphous SiO2 by mixing aqueous solutions of 3d
transition-metal chlorides (MCl2 ·nH2O) and sodium metasilicate
nonahydrate (Na2SiO3 ·9H2O). The 3d transition metals M chosen
were Fe, Ni, Co, Mn, Cu, Mg, Zn, Ba, Ti and their plural compounds.
Metal chlorides aqueous solution mixed with a sodium hydroxide
(NaOH) aqueous solution without a metasilicate nonahydrate, were
also partially prepared. The resulting precipitates were washed
several times with distilled water and dried at around 350 K. The
obtained glassy solids were pulverized in a mortar and calcined in an
electric furnace between 573–1373 K in air. The particle sizes were
controlled by means of the annealing temperature, pH values, and
concentration of the silicates. Particle sizes between 2 and 34 nm were
obtained.
The prepared samples were examined by means of Cu Kα X-ray
(λ = 0.145 nm) powder diffraction at room temperature (Rigaku, miniflexII).
They were additionally examined by transmission electron microscopy
(TEM) imaging, X-ray absorption fine-structure (XAFS) experiments,
thermogravimetric and differential thermal analysis (TG-DTA, Rigaku
TAS-100), and Fourier transform infrared spectroscopy (FT-IR; Horiba
FT-720 and Thermo Scientific NICOLET 380). The morphology and
diameter distribution of the particles were investigated by means of a
transmission electron microscopy (TEM; JEM-1230, JEOL, Japan).
Magnetization was measured by a SQUID magnetometer (Quantum
Design, MPMS) under a ±50 kOe magnetic field, at temperatures from
2 to 300 K.
66 Y. Ichiyanagi
4.3. Magnetic Properties of 3d Metal Hydroxide and
Metal Oxide Nanoparticles
4.3.1. Magnetic properties of metal hydroxide nanoparticles
Nickel hydroxide (Ni(OH)2) nanoparticles encapsulated in amorphous
SiO2 were produced by the wet method described above. MCl2 ·nH2O and
Na2SiO3 ·9H2O were taken as starting chemical reagents.
Figure 4.1 shows the X-ray powder diffraction patterns of samples
measuring 2.4 nm and bulk crystal. The top pattern is the prepared
sample and revealed the formation of a monolayer nanocluster (MNC),
as the diffraction patterns showed only two-dimensional (2D) broad
peaks of (1 0) and (1 1) reflections, corresponding to those of (1 0 l) and
(1 1 l) of the bulk crystal. In addition, noticing the extinction of (0 0 l)-
reflection, and the splitting of (1 0 l)-, (1 1 l)-, and (2 0 l)-reflections,
which appear only in the case of a three-dimensional (3D) structure, the
obtained sample was explained as a monolayer. The bottom illustration
of Fig. 4.2 shows a typical
_
3D Ni(OH)2, which has CdI2-type hexagonal
structure, space group c3m, with lattice constants of a3D = 0.3117 nm and
c3D = 0.4593 nm from our experimental data as illustrated in Fig. 4.2.
This is a layered compound composed of a Ni layer sandwiched between
two O layers. A curve fitting analysis of the X-ray diffraction patterns
based on Warren’s formula37 revealed that the 2D lattice constant a2D of
Ni(OH)2-MNC is 0.308 ± 0.003 nm, which is obviously shorter than that
of the bulk crystal. Nickel hydroxide 3D crystal can be approximated to
the infinite 2D ferromagnetic system, and the monolayer nanocluster can
be regarded as a finite ferromagnetic system.
In order to make out the possibility of the monolayer Ni(OH)2,
further structural analyses for these species were conducted to
understand the correlation between the geometric and magnetic structure.
X-ray absorption fine-structure (XAFS) experiments were performed to
clarify the microscopic structure of bulk crystal and MNC of Ni(OH)2.
Ni K-edge XAFS spectra were taken in the transmission mode at the
Beamline 10B38 of the Photon Factory at the Institute of Materials
Structure Science (KEK-PF: typical operation energy: 2.5 GeV, stored
current: 450–300 mA), using a channel-cut Si(3 1 1) monochromator.
Measured temperatures were 20 K for X-ray absorption near-edge fine-
Preparation of Magnetic Nanoparticles Using Chemical Route 67
structure (XANES), and room temperature for extended X-ray absorption
fine-structure (EXAFS). XANES measurements and DFT calculations39
suggested that the O-H bonds stand upright on the layer. From the results
of EXAFS measurements, the oscillation functions k3χ(k) were obtained
by standard analysis procedures,40 then Fourier transforms were operated
as shown as Fig. 4.3. From this figure it is obvious that Ni-O2,O3 peak
intensity and Ni-Ni3 peak intensity were reduced drastically in MNC in
comparison with the bulk curve, in addition the peaks were shifted
toward a shorter distance. Analyses of EXAFS spectra exhibited that a
significant intralayer lattice contraction takes place with the displacement
of oxygen atoms outward along the perpendicular direction.26
It is known that Ni(OH)2 is antiferromagnetic, with a Neel
temperature of 25.75 K.41 Magnetization measurements were performed
with a SQUID magnetometer (Quantum Design MPMS) in external
fields between −50 and 50 kOe at a temperature range from 5 K to 300 K.
Figure 4.4 shows typical temperature dependence of both field-cooled (FC)
and zero-field-cooled (ZFC) of DC magnetic susceptibilities for Ni(OH)2
MNC samples under a 100 Oe field. The magnetic susceptibilities of FC and
ZFC diverge below TC ≈ 10 K which agrees with the magnetic transition
temperature of Ni(OH)2 MNC determined by AC magnetic susceptibility.10
Ni-MNC
(10)
a-SiO2 (11)
(20)
Intensity / arb.unit
(001) Ni(OH)2 (bulk)
(101)
(100)
(110)
(102) (111) (201)
(200)
2θ /deg
Fig. 4.1. Powder X-ray diffraction patterns of Ni(OH)2 nanoparticles (upper) and bulk
crystal (lower). (Reference 24, permission from The Surface Science Society of Japan,
Copyright 2008.)
68 Y. Ichiyanagi
Fig. 4.2. Top: Schematic structure of Ni(OH)2 based on the X-ray diffraction analysis.
Large and small spheres represent O and Ni atoms, respectively. Bulk Ni(OH)2
is composed of stacked sheets of the Ni(OH)2 unit. Each arrow on a small sphere
indicates direction of spin. Bottom: Schematic view of a monolayer sheet of Ni(OH)2.
Coordination shells are indicated with respect to the central Ni atom denoted by Ni*. The
third-nearest neighbor oxygen atom (O3) is situated in the adjacent sheet (not shown).
(Reference 26, permission from Elsevier, Copyright 2008.)
Particle size dependence of transition temperature of Ni(OH)2 was
summarized in Fig. 4.5.10,41,42 A drastic change of TC is observed below
several nanometer of the particle size, although the current system
consists of a two-dimensional system and reflects the effect of
fluctuation. This could be one of the finite size effects of extremely small
particles.
Preparation of Magnetic Nanoparticles Using Chemical Route 69
Fig. 4.3. Fourier transforms of Ni K-EXAFS from bulk Ni(OH)2 and Ni(OH)2-MNC.
Main contributions are noted above the peaks. See Fig. 4.2 for the numbering of atoms.
(Reference 26, permission form Elsevier, Copyright 2008.)
10
8 Ni-MNC FC
χ / emu/mol-Ni
ZFC
6
4
T
c c
0
0 10 20 30 40 50
T/K
Fig. 4.4. Temperature dependence of DC magnetization of Ni-MNC. (Reference 24,
permission from The Surface Science Society of Japan, Copyright 2008.)
70 Y. Ichiyanagi
By observation of magnetic relaxation time, we can approach a
possibility of quantum magnetic tunneling (QMT) of a nanoscopic
system. To investigate the magnetic viscosity effect directly, time
dependence of the remanent magnetization Mr was measured after
switching off the cooling magnetic field of 100 Oe at the lowest
temperature, between 5 and 11 K. Mr changes linearly with the logarithm
of time t in the above temperature region. From this result, the following
expression can be adopted for Mr ,4,5
M r (t ) = M r (0)[1 − S ln(t )] , (1)
where S is the decaying rate of Mr and is called magnetic viscosity. The
temperature dependence of S given in Fig. 4.6 changes drastically near
TC ≈ 10 K, then becomes nearly constant below about 10 K. This
phenomenon suggests quantum magnetic tunneling in this system. If we
define QMT-temperature as TQ, magnetic transition temperature TC is
nearly equal to TQ.
30
20 Sorai et al. (1969)
Enoki et al. (1978)
Tc / K
10
present work
0 0 1 2 3 4
10 10 10 10 10
particle size / nm
Fig. 4.5. Particle size dependence of transition temperature TC of Ni(OH)2. (Reference 24,
permission from The Surface Science Society of Japan, Copyright 2008.)
Preparation of Magnetic Nanoparticles Using Chemical Route 71
0.20
0.15
0.10
S
0.05
0.00
0 5 10 15 20
T/K
Fig. 4.6. Magnetic viscosity S as a functions of temperature for Ni(OH)2 MNC for
measuring the field of 100 Oe. (Reference 13, permission from Elsevier, Copyright
2008.)
As mentioned above, the magnetic structure of bulk Ni(OH)2, the
magnetic moments of the Ni atoms in a layer are all directed to the c-axis,
but they are antiparallel with those in the adjacent Ni layers, resulting in
antiferromagnetism. On the other hand, MNC is composed of a single Ni
layer sandwiched by two O layers, and it could exhibit a ferromagnetic
character. We observed, in fact, ferromagnetic behavior below 10 K.11 It
is expected that the neighboring of TC and TQ in 2D Ising ferromagnetic
nanoscopic system can be explained by theoretical studies.
4.3.2. Metal oxide nanoparticles
Using FeCl2 ·4H2O as a starting reagent precursor of metal chloride, two
types, of iron oxides of α-Fe2O3 (Hematite) and γ-Fe2O3 (Maghemite)
nanoparticles ranging from 1.8 to 23 nm in size were obtained depending
on the annealing temperature and annealing time. Fe3O4 was also
obtained by using Na(OH) in place of Na2SiO3 ·9H2O. In this Fe-O
system, it is very interesting that a relative large coercivity of 1 kOe was
observed for 5 nm particles even at room temperature, as shown in
Fig. 4.7. These results were considered to be due to the coexistence of
γ- and α-Fe2O3 phase as evidenced by X-ray diffraction measurements.
72 Y. Ichiyanagi
0.3
1023 K, 10 h ann.
0.2
) / Fe-ion
0.1
0
−1
Β
-0.1
(M µ
-0.2
-0.3
-50 -25 0 25 50
H / kOe
(a)
.
0 .0 4
0 .0 3
Q
LR
0 .0 2
H
) 0 .0 1
0
%
˩
- 0 .0 1
0
- 0 .0 2
- 0 .0 3
- 0 .0 4
-1 0 -5 0 5 10
+ N 2 H
(b)
Fig. 4.7. Magnetization curve of Fe-O system of 5 nm (a-top panel) and enlargement of
the origin (b-bottom panel). (Reference 24, permission from The Surface Science Society
of Japan, Copyright 2008.)
A gamma phase has an inverse spinel structure and is ferrimagnetic
at room temperature, and an alpha phase has a corundum structure
and weak ferromagnetism, however, it reorders at 260 K into
antiferromagnetism, in the case of bulk crystal. Figure 4.8 shows the
temperature dependence of both field-cooled and zero-field-cooled
magnetization for various annealing temperature. If we define the
bifurcated temperature of FC and ZFC as blocking temperature TB, above
Preparation of Magnetic Nanoparticles Using Chemical Route 73
0.25
0.2
923K,10h ann. ZFC
T
B FC
0.15 1023K,10h ann.ZFC
FC
T
B
0.1
0.05
H = 5 kOe
0
0 50 100 150 200 250 300
T/K
Fig. 4.8. Temperature dependence of magnetization of Fe-O nanoparticles system (original).
the blocking temperature, the magnetic spin of the particles is supposed
to fluctuate against thermal energy and behave superparamagnetically.
Ferromagnetic behavior can be observed below TB. The large coercive
force described above was observed even in the superparamagnetic region.
In addition to iron oxides, we prepared Co3O4, NiO, Mn3O4
nanoparticles and reported their magnetic properties.16,17,20,23
4.4. Pluralistic Ferrite Nanoparticles
In Secs. 4.3.1 and 4.3.2, I introduced magnetic particles composed of a
single atom. In order to approach a practical use, it is important to
research the possibilities of ferromagnetic nanoparticles. Here I will
summarize pluralistic ferrite nanoparticles.
4.4.1. Ni-Zn ferrite nanoparticles
Ni1-xZnxFe2O4 (0 ≤ x ≤ 1.0) (Ni-Zn ferrite) nanoparticles encapsulated in
amorphous SiO2 were prepared by the method mentioned above by
74 Y. Ichiyanagi
mixing solutions of nickel chloride, zinc chloride, iron chloride and
sodium metasilicate nonahydrate aqueous solutions. The mole ratio of
metal and Si was arranged to be 1:1. The diameters of these particles
were estimated from X-ray diffraction patterns as ranging from 2 to
34 nm. The particle size dependence on saturation magnetization is
shown in Fig. 4.9. As the particle size decreases, magnetization decreases
gradually and abruptly changes at around 6 nm.
Ni ferrite bulk crystal has an inverse spinel structure, where the A
sublattice contains half of the Fe3+ ions, and the B sublattice contains the
other half, together with all the Ni2+ ions. Here, A sites are tetrahedral
sites, and B sites are octahedral sites. In a mixed Ni-Zn ferrite, the Zn2+
ions are expected to remain in the A sites, and Ni2+ ion in the B sites. It is
a very curious and interesting phenomenon that when Zn ions are added
to spinel ferrites, their magnetization increases according to the Zn
content, even though Zn ions are nonmagnetic.43 Such a phenomenon is
attributable to the magnetic spin locate counter direction in the A site and
the B site. In order to observe the concentration dependence of zinc ion,
particle size was controlled at about 6 nm by adjusting the annealing
temperature. The particle sizes and lattice constants are summarized in
Table 4.1. It is known that the lattice constant of Ni-Zn ferrite
nanoparticles are larger than that of bulk crystal,44 and our data reach
agreement on that point.
The magnetic parameters of permeability µ and coercivity Hc, varying
with Zn content are summarized in Fig. 4.10. The highest permeability
was observed at a Zn content of x = 0.6, where the lowest coercivity was
also found. Saturation magnetizations Ms, determined from the M-H
curves at each Zn concentration, are shown in Fig. 4.11. Reported Ms
values of bulk crystal are also shown in Fig. 4.11.43 As an overall
tendency, the Ms value varies with the Zn concentration, and changes for
the bulk crystal in a similar trend, having a maximum value near x = 0.7.
But the difference between the nanoparticles and bulk crystal is that for
the nanoparticles when x = 0.7, Ms value has the maximum, while for the
bulk crystal when x = 0.5, the Ms value is the highest. On the other hand,
no difference between nanoparticles and bulk crystal was found at higher
Preparation of Magnetic Nanoparticles Using Chemical Route 75
concentration of Zn ion above x = 0.7. The maximum value of the
saturation magnetization Ms of the nanoparticles tends to be 30% lower
than that of bulk crystal. This phenomenon could be due to the canting of
the surface spin of nanoparticles at a definite angle. The canting angle θ
can be roughly estimated as θ = 45º.45,46 Another possibility is that ion
distribution is achieved during the annealing of the particles. The cation
distribution in extremely small particles could be different from that in
bulk material. For bulk crystal, Ni2+ ions are expected to remain in the B
sites, however, in the nanoparticles system some Ni ions might to be in
A sites, thus resulting in a decrease of the magnetic moment.47 Spin
canting or ion distribution should be determined by Mössbauer
spectroscopy in the future.
Table 4.1. Particle size of the samples at various concentrations, annealing temperatures
with lattice constants estimated by X-ray diffraction.
x ann. temp. d / nm a / nm
0.0 (NiFe2O4) 1173 K 5.5 0.834
0.1 1173 K 5.6 0.836
0.2 1173 K 5.6 0.837
0.3 1173 K 5.9 0.837
0.4 1148 K 6.1 0.838
0.5 1173 K 6.2 0.839
0.6 1123 K 5.6 0.841
0.7 1073 K 5.6 0.843
0.8 1073 K 6.2 0.843
0.9 1073 K 6.3 0.845
1.0(ZnFe2O4) 1073 K 6.4 0.845
76 Y. Ichiyanagi
([SHULPHQW
Ms / µB
Diameter / nm
Fig. 4.9. The particle size dependence of the saturation magnetization Ms. (Reference 21,
permission from LEXICIA Springer, Copyright 2008.)
H
P 2
ᨿ
F
+
+
µ
NiFe2O4 Zn content x ZnFe2O4
Fig. 4.10. Concentration dependence of permeability (u) and coercivity (Hc) at 5 K under
±50 kOe. (Reference 19, permission from WILEY-VCH, Copyright 2008.)
Preparation of Magnetic Nanoparticles Using Chemical Route 77
* R U W H U
SUHVHQW
MS / µ B
1L)H 2 = [
Q F R QW HQW = Q) H 2
2
= Q F R QW HQW [
Fig. 4.11. Saturation moment in Bohr magnetons of various Zn concentrations. Closed
squares show our experiment and open squares indicate value of bulk crystal reported by
Gorter. (Reference 19, permission from WILEY-VCH, Copyright 2008.)
T=5K D
E
F
G
H
I
µB
M
H kOe
Fig. 4.12. Magnetization curves for particles of 6.1 nm (a, b); 4.9 nm (c, d) and 3.1 nm
(e, f) quenched by different methods. Closed marks – Method 1: placing the sample on a
copper plate cooled by liquid nitrogen. Open marks – Method 2: quenching to room
temperature over a period of 20 h. (Reference 25, permission from LEXICIA Springer,
Copyright 2008.)
78 Y. Ichiyanagi
4.4.2. Mg ferrite nanoparticles
MgFe2O4 (Mg-ferrite) is a very useful soft magnetic material, and is
expected to be useful for sensor and catalysis applications.47–49 It is very
interesting that Mg-ferrite shows magnetization depending on the ion
distribution, despite the fact that Mg2+ ions have nonmagnetic properties.
It is known that Mg-ferrite bulk crystal has an inverse spinel structure,
where Mg2+ ions prefer to exist in the octahedral B sites. When subjected
to heat treatment (annealing or quenching), Mg ions tend to position
themselves in the tetrahedral A sites in some cases.50,51 In this section,
MgFe2O4 (Mg-ferrite) nanoparticles were produced, and their magnetic
properties and heat-treatment effects were investigated. The starting
reagents chosen were MgCl2 ·6H2O, FeCl2 ·4H2O and Na2SiO3 ·9H2O.
Magnetization curves were observed after different quenching
methods: Method 1: the samples were placed on a copper plate cooled by
liquid nitrogen. Method 2: they were quenched to room temperature over
a period of 20 hours. Method 3: they were quenched for 4 hours
(normally annealed). The M-H curves for each particle size measured at a
temperature of 5 K under a ±50 kOe field are given in Fig. 4.12. The
closed marks (letters a, c, e) correspond to quenching Method 1, and the
open marks (letters b, d, f) correspond to quenching Method 2. The
magnetization value increased as the quenching time decreased for all the
samples, and the maximum saturation magnetization value, Ms, was
found to be 1.8 µB per molecule at 5 K. These results indicate that some
of the Mg ions were distributed more in the A sites as a result of
quenching.
In order to clarify the formation process and heat-treatment
effect, differential thermal analysis and thermogravimetric (TG-DTA)
measurements were carried out using a Rigaku TAS-100 under air. In
the TG curve, mass loss by dehydration was observed at temperatures
above 370 K, after which the sample mass gradually decreased as the
temperature increased up to 1500 K, as shown in Fig. 4.13. Figure 4.13
also shows the DTA curve for the as-prepared samples of Mg-ferrite
from room temperature to 1500 K. As can be seen in this figure, there is
a deep endothermic peak at around 370 K, corresponding to the above
Preparation of Magnetic Nanoparticles Using Chemical Route 79
dehydration (Fig. 4.13a). Two exothermic peaks were observed at around
980 and 1280 K. The former small peak could be considered to have
been caused by the crystallization, at that temperature, of the amorphous
as-prepared sample (Fig. 4.13b), and this explanation is consistent with
the powder X-ray diffraction patterns. The slight shoulder in the latter
could be due to the sudden aggregation of nanoparticles in this system.
The existence of an exothermic peak at 1500 K is unknown at present,
however, it is considered that Mg-ferrite was decomposed into other
compounds such as α-Fe2O3 and MgSiO3 (Fig. 4.13c). The diffraction
patterns of those compounds were observed by X-ray measurement after
heating to 1500 K.
Thus, it was clarified that some of the Mg ions positioned
themselves in the A sites of the spinel structure as a result of quenching.
The magnetization values varied with the heat treatment. However, the
precise ratio of site distribution was not clear. Mössbauer measurements
are needed to determine the ion distribution in this system.
E (
7*
[R
'7$
Mass loss / mg
F
∆ E / µV
. ᨺ .
. ᨺ .
(
. P LQ
QG
R
D
T/K
Fig. 4.13. TG-DTA curves of as-prepared MgFe2O4 nanoparticles under air in the
temperature range between room temperature and 1500 K. (Reference 25, permission
from LEXICIA Springer, Copyright 2008.)
80 Y. Ichiyanagi
4.5. Functionalization of Magnetic Nanoparticles
As mentioned above, we have prepared different types of interesting
magnetic nanoparticles. They are, of course, useful for high-density
magnetic recording media and other magnetic materials. However,
recently, we are endeavoring to use the magnetic nanoparticles for
biomedical applications. Our magnetic nanoparticles are extremely small,
so that the particles can be introduced into cells without a cationic
coating, and could even have some molecules attached to them such as
amino group, antibodies or drugs. Considering an application for a drug-
delivery system or hyperthermia treatment, it is important to be
functionalized by modifying amino groups. So we tried to conjugate
amino groups with magnetic nanoparticles.27,32–34
4.5.1. Amino-silane coupling
Functionalized magnetic nanoparticles (f-MNPs; d = 3.5 nm) were
prepared by means of a silanization procedure using γ-Fe2O3 as
introduced in Sec. 4.3.2 and 3-aminopropyl-triethoxysilane (γ-APTES).
The mixture of γ-Fe2O3 and γ-APTES was stirred for 10 min, heated to
403 K, and then washed several times with pure water. The f-MNPs were
labeled with rhodamine, which is a fluorescent material. The labeled
MNPs were introduced into PtK2 cells, as well as into the ear of a
mouse, wherein they were detected in the ear by use of a button magnet
with a surface flux density of 240 mT.
The silanized particles bear surface amino groups, which can be
used to covalently attach various materials to the particle surface. The
presence of the amino groups on the MNPs was confirmed by Fourier
transform infrared spectroscopy (FT-IR). In Fig. 4.14, the typical O–H
peak (3660–2991 cm–1), the C–H peak (2972–2842 cm–1), and the C–N
peak (1583–1481 cm–1) were detected in the spectrum of the f-MNPs
(Fig. 4.14b), whereas only the O–H (3660–2991 cm–1) peak was observed
in the spectrum of the unfunctionalized MNPs.
The X-ray diffraction pattern of the f-MNPs indicated that the
functionalized particles retained the spinel γ-Fe2O3 structure of the
unfunctionalized MNPs. The diameters of the amino-MNPs and the
Preparation of Magnetic Nanoparticles Using Chemical Route 81
MNPs were independently estimated at 3 nm from the half-width of the
X-ray diffraction peaks for both f-MNPs and MNPs. TEM images
showed no significant difference between the shape of the f-MNPs and
that of the unfunctionalized MNPs. The average diameter of the amino-
MNPs was determined to be about 3 nm, which is in good agreement
with the X-ray diffraction experimental value.
The magnetization data indicated that the both the f-MNPs and the
unfunctionalized MNPs increased linearly with an increasing magnetic
field; that is, both types of particles showed paramagnetic behavior. The
MNP precipitates could be easily attracted in water solution by means of
an external magnetic field of 240 mT. These results indicate that the
structure and magnetism were basically unchanged by functionalization
with the amino groups.
To evaluate the internalization of the MNPs. the PtK2 cells were
cultured after addition of the labeled MNPs. After the cells were cultured
for 24 hours, fluorescence was observed from inside the cells even after
washing as shown in Fig. 4.15a. On the other hand, no fluorescence was
observed from the untreated cells as control. Toxicity was also checked
by dividing the cultured cells. The PtK2 cells with MNPs continued to
grow even after 5 days.
These results indicate that the PtK2 cell division was not inhibited
by the MNPs. We investigated the intracellular space of the living PtK2
cells by TEM to confirm the introduction of the labeled MNPs into the
cells. The TEM image of the cells into which MNPs had been introduced
showed high-density spots (dashed circles) as shown in Fig. 4.15b.
Thus we have successfully introduced magnetic particles into the
living cells without cationic coating as usual done. Localization of MNPs
in a mouse’s ear by means of an external magnetic field was also
observed.32
4.5.2. Development for cell selective magnetic nanoparticles
If functionalized magnetic nanoparticles could be introduced into cancer
cells selectively, it would be possible to realize drug-delivery without
side effects.
82 Y. Ichiyanagi
Fig. 4.14. FT-IR spectra of unmodified nanoparticles (a) and amino-nanoparticles (b).
(Reference 32, permission form The Surface Science Society of Japan, Copyright 2008.)
In pursuit of this, we developed a cell-specific delivery system that
makes use of MNPs (d = 3 nm) conjugated with folic acid (FA) and a
coumarin fluorophore (CF) for recognition by folate receptors on the cell
surface.49 The modified MNPs were internalized by human pharyngeal
Preparation of Magnetic Nanoparticles Using Chemical Route 83
cancer cells (KB cells) after an incubation time that was short compared
with the time required for internalization of MNPs without folic acid.
Cellular recognition of MNPs may lead to the development of other cell-
specific delivery systems.
The amino-MNPs were modified with FA and a CF. The FT-IR
spectra of the FA-CF-MNPs showed a peak at 1419 cm–1 corresponding
to the p-amino benzoic acid moiety of FA, and no peaks for the amino-
MNPs were observed. The presence of CF moiety on the MNPs was
confirmed by fluorescence microscopy. Transmission electron microscopy
showed that the shape of the FA-CF-MNPs did not differ markedly from
the shape of the unmodified MNPs. The number-average diameter of the
FA-CF-MNPs was determined to be about 3 nm.
We evaluated the cellular uptake of the FA-CF-MNPs in three
dimensions. After KB cells had been cultured for 3 hours, fluorescence
was observed from the cells treated with the MNPs. The lateral image
clearly shows that the aggregated particles were present inside the cells.
In contrast, no fluorescence was observed from inside the cells treated
with CF-MNPs, although weak fluorescence was observed from the cell
surface and hollow region, respectively, owing to nonspecific adsorption
Fig. 4.15. PtK2 cells incubated with Rhodamine labeled MNPs for 24 hours (a) and TEM
image of PtK2 cells incubated with labeled MNPs (b). (Reference 32, permission form
The Surface Science Society of Japan, Copyright 2008.)
84 Y. Ichiyanagi
of the particles. Untreated KB cells showed no fluorescence, which
indicates that the fluorescence observed from the KB cells treated with
the FA-CF-MNPs was not autofluorescence originating from within the
cells themselves. Normalized fluorescence intensities (F.I.)/um2 were
estimated. All fluorescence intensities were normalized with respect to
the fluorescence intensity of the untreated control cells. The ratio of the
fluorescence intensity of the FA-CF-MNPs internalized in the cells to the
intensity of the cells treated with CF-MNPs to the intensity of the
untreated cells was 158:2:1. In addition, no fluorescence was observed in
cells treated with FA-CF-MNPs and excess free FA at the same time. As
another control experiment, FA-CF-MNPs were added to rat kangaroo
kidney epithelium (PtK2) cells. No fluorescence was observed from
either the inside cells or the surficial cells treated with FA-CF-MNPs.
These results indicate that FA-CF-MNPs were internalized in the
KB cells via folate receptors mediated endocytosis (ligand–receptor
interaction).
4.6. Conclusions and Outlook
In summary, preparation and characterization of magnetic nanoparticles
were introduced. We found a very interesting 2D Ising ferromagnetic
nanoscopic system for Ni(OH)2 monolayered nanocluster. In a Fe-O
system, it is very interesting that a relative large coercivity of 1 kOe was
observed for 5 nm particles, even at room temperature. Pluralistic ferrite
nanoparticles are interesting because their magnetization values vary
accordeing to the cation distribution in the spinel structure. In addition, it
is a very curious and interesting phenomenon that when Zn ions are
added to spinel ferrites, their magnetization increases according to the Zn
content, even though Zn ions are nonmagnetic. Quantum size effects and
quantum magnetic tunneling were suggested for our magnetic
nanoparticles.
For biomedical applications, we studied the feasibility of in vivo
introduction of functionalized MNPs. The particles could be incorporated
into the subcutaneous tissue of a mouse’s ear by means of a magnetic
field. We demonstrated that Folic Acid Coumarin Fluorophore MNPs
could recognize a specific cell type and be internalized in tumor cells. It
Preparation of Magnetic Nanoparticles Using Chemical Route 85
therefore would be possible to realize ideal medical treatment, for
example, to deliver therapeutic agents to diseased tissue by application of
a magnetic field and thus achieve selective delivery at the cellular level
using our MNPs. Furthermore, hyperthermia treatment could be also
applied after delivery into the cells.
Acknowledgments
This study was partially supported by Japan Society for the Promotion
of Science Grant-in Aid for Scientific Research (No. 12650008,
No. 15510089, No. 18510090), JST Innovation Bridge, Potentiality
Verification Stage, of the Japan Science and Technology Agency and
Precursory Research for Embryonic Science and Technology of Japan
Science and Technology Agency. I would like to thank the members
of the Mitsubishi Kagaku Institute of Life Sciences (MITILS) and
Dr S. Taira of JAIST for biological experiments, and I would like to
express special thanks to the students of my laboratory.
References
1. A. F. Berkowitz and W. J. Schuele, J. Appl. Phys. 30 (1959).
2. R. H. Kodama, A. E. Berkowitz, E. J. Jr McNiff and S. Foner, Phys. Rev. Lett. 77
(1996) 394.
3. J. R. Friedmann, M. P. Sarachik, J. Tajada and R. Ziolo, Phys. Rev. Lett. 76 (1996)
3830.
4. E. M. Chudnovsky and L. Gunther, Phys. Rev. Lett. 60 (1988) 661.
5. M. J. O’Shea and P. Perera, J. Appl. Phys. 76 (1994).
6. E. M. Chudnovsky, Science 274 (1996).
7. Y. Ichiyanagi, Y. Kimishima, K. Iga and N. Miyata, Physica B 194-196 (1994) 297.
8. Y. Ichiyanagi and Y. Kimishima, J. Magn. Magn. Mater. 140-144 (1995) 1629.
9. Y. Ichiyanagi and Y. Kimishima, Jpn. J. Appl. Phys. 35 (1996) 215.
10. Y. Ichiyanagi and Y. Kimishima, Mat. Sci. Eng. A 217 (1996) 358.
11. Y. Ichiyanagi and Y. Kimishima, J. Magn. Magn. Mater. 177-181 (1998) 964.
12. Y. Kimishima and Y. Ichiyanagi, J. Magn. Magn. Mater. 189 (1998) 62.
13. Y. Ichiyanagi and Y. Kimishima, J. Magn. Magn. Mater. 198-199 (1998) 197.
14. Y. Ichiyanagi and Y. Kimishima, Trans. Mater. Res. Soc. Jpn. 26 (2001) 1097.
15. Y. Ichiyanagi and Y. Kimishima, J. Ther. Anal. Calor. 69 (2002) 919.
86 Y. Ichiyanagi
16. Y. Ichiyanagi, N. Wakabayashi, J. Yamazaki, S. Yamada, Y. Kimishima, E. Komatsu
and H. Tajima, Physica B 329-333 (2003) 862.
17. Y. Ichiyanagi, Y. Kimishima and S. Yamada, J. Magn. Magn. Mater. 272-276
(2004) 1245.
18. Y. Ichiyanagi, J. Yamazaki, Y. Kimishima and Y. Tachibana, Trans. Mater. Res.
Soc. Jpn. 29 (2004) 1651.
19. Y. Ichiyanagi, T. Uehashi and S. Yamada, Phys. Stat. Sol. (c) 1 (2004) 3485.
20. Y. Ichiyanagi and S. Yamada, Polyhedron 24 (2005) 2813.
21. Y. Ichiyanagi, T. Uehashi, S. Yamada, Y. Kanazawa and T. Yamada, J. Ther. Anal.
Calor. 81 (2005) 541.
22. Y. Ichiyanagi, M. Kubota, S. Moritake, Y. Kanazawa, T. Yamada and T. Uehashi,
J. Magn. Magn. Mater. 310 (2007) 2378.
23. Y. Ichiyanagi, T. Yamada, Y. Kanazawa and T. Uehashi, AIP Conference
Proceeding 850 (2006) 1155.
24. Y. Ichiyanagi, Hyoumen Kagaku 27-2 (2006) 95.
25. M. Kubota, Y. Kanazawa, K. Nasu, S. Moritake, H. Kawaji, T. Atake and
Y. Ichiyanagi, J. Therm. Anal. Cal. 92-2 (2008) 461.
26. Y. Ichiyanagi, K. Okamoto, Nagai, H. Kondo, Yokoyama, T. Ohta, Chem. Phys.
Lett. 379 (2003).
27. Y. Ichiyanagi, S. Moritake, S. Taira and M. Setou, J. Magn. Magn. Mater. 310
(2007) 2877.
28. C. KIttel, Phys. Rev. 70 (1946) 965.
29. C. Néel, Compt. Rend. 224 (1947) 1488.
30. E. C. Stoner and E. P. Wohlfarth, Nature 160 (1947) 650.
31. D. J. Dunlop, Philos. Mag. Ser. 19 (1969) 329.
32. S. Moritake, S. Taira, Y. Ichiyanagi, N. Morone, S. Song, T. Hatanaka, S. Yuasa and
M. Setou, J. Nanosci. Nanotech. 7 (2007) 937.
33. S. Moritake, S. Taira, T. Hatanaka, M. Setou and Y. Ichiyanagi, J. Surf Sci.
Nanotech. 5 (2007) 60.
34. S. Taira, S. Moritake, Y. Kai, T. Hatanaka, Y. Ichiyanagi and M. Setou, J. Surf. Sci.
Nanotech. 5 (2007) 23-28.
35. S. Moritake, S. Taira, Y. Sugiura, M. Setou and Y. Ichiyanagi, J. Nanosci. Nanotech.
8 (2008) 1.
36. S. S. Taira, Y. Sugiura, S. Moritake, Y. Ichiyanagi and M. Setou, Anal. Chem. 80
(2008) 4761.
37. B. E. Warren, Phys. Rev. 59 (1941) 693.
38. M. Nomura, A. Koyama, KEK Rep. 89 (1989) 16.
39. H. Orita, private communication.
40. D. C. Koningsberger, R. Prins (Eds.), X-Ray Absorption: Principles, Applications,
Techniques of EXAFS, SEXAFS and XANES, Wiley, New York, 1988.
41. T. Enoki and Tsujikawa, J. Phys. Soc. Jpn. 45 (1978) 1515.
42. M. Sorai, A. Kosaki, H. Suga and S. Seki, J. Chem. Thermodynamics 1 (1969) 119.
Preparation of Magnetic Nanoparticles Using Chemical Route 87
43. E. W. Gorter, Nature 165 (1950) 798.
44. L. Wang and F. S. Li, J. Magn. Magn. Mater. 223 (2001) 233.
45. J. M. D. Coey, Phys. Rev. Lett. 27 (1971) 1140.
46. A. H. Morrish and K. Haneda, J. Magn. Magn. Mater. 35 (1983) 105.
47. C. N. Chinnasamy, A. Narayanasamy, N. Pondandian and R. Justin Joseyphus,
J. Magn. Magn. Mater. 238 (2002) 281.
48. N. Ma, Y. Yue, W. Hua and Z. Gao, Appl. Catal. A: General 251 (2003) 39.
49. C. Xiong, Q. Chen, W. Lu, H. Gao and Z. Gao, Catal. Lett. 69 (2000) 231.
50. V. Sepelak, D. Schultze, G. Krumeich, U. Steinike and K. D. Becker, Sol. Stat.
Ionics 141-142 (2001) 677.
51. M. Singh, J. Magn. Magn. Mater. 299 (2006) 397.
This page is intentionally left blank
Chapter 5
ELECTRODEPOSITION AS A FABRICATION METHOD
OF MAGNETIC NANOSTRUCTURES
László Péter and Imre Bakonyi
Metals Research Department
Research Institute for Solid State Physics and Optics
Hungarian Academy of Sciences
P.O. Box 49, H-1525 Budapest, Hungary
E-mail:
[email protected] The first part of this chapter gives a general survey on electrochemistry,
with a special emphasis on electrodeposition of magnetic metals. The
second part comprises a discussion of the electrodeposition of magnetic
nanostructures in various forms. This overview is meant to yield an
easy-to-read guidance amongst electrodeposited nanostructures even
for those who are not involved in this field. Thin magnetic films,
nanocrystalline deposits and multilayers are discussed as well as other
methods that can be used to prepare a precursor material for magnetic
nanostructure formation by a follow-up heat treatment. Template-based
methods are listed in a separate section because electrodeposition is a
unique method to achieve structures with a very high aspect ratio such
as homogeneous and multi-segment nanowires.
5.1. Introduction
Electrodeposition or electroplating has been widely used in many fields
of technology. Traditional applications such as hard chromium plating
for anti-corrosion and decorative purposes have now been complemented
with the incorporation of electrodeposition procedures into micro-
electronic procedures. The most important method in the latter group is
the formation of Cu interconnects in printed circuit boards by using the
so-called damascene process. This is already a part of nanotechnology
since the width of Cu interconnects are nowadays in the range of 60 nm.
89
90 L. Péter and I. Bakonyi
Electrodeposition of magnetic nanostructures has long been regarded as a
possible route of manufacturing individual bit elements for magnetic
storage media.
In this chapter, a short overview is given about electrochemistry. The
authors’ intention was to make the reader familiar with the most
important principles in a schematic way but without any overburdening
by complicated formulas. The detailed description of the specific
methods follows the same line, keeping in mind that the purpose is rather
the visualization of the methods than their deep discussion. For subtle
details, the reader will be referred to specific reviews.
5.2. Electrodeposition: A General Overview
5.2.1. Definitions and major principles
Electrochemistry1–5 deals with chemical reactions where a charge
transfer takes place. In heterogeneous electrochemical reactions, the
charge transfer occurs at a phase boundary, and at least one of the phases
involved is an ionic conductor (a melt or an electrolyte solution) whereas
the other phase can also be an electron conductor. For electrodeposi-
tion,6–9 one needs a solid metal or a semiconductor substrate on the
surface of which the charge transfer takes place and where the deposit is
accumulated.
The boundary layer between an ionic and an electron conductor is
termed as an electrode. The driving force of the electrode reactions is
often ranked by the order of standard electrode potentials where the
standpoint of referencing is the standard hydrogen electrode (SHE), and
the more positive standard potential indicates a more noble behaviour.
The dependence of the equilibrium potential (E) of a metal electrode on
the metal ion concentration [Mz+] in the electrolyte is described with the
well-renowned Nernst equation:
RT
E = E0 + ln M z + (1)
zF
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 91
where E0 is the standard potential of the M/Mz+ redox system, F is the
Faraday constant, and the other symbols have the usual meaning.
Another thermodynamic approach to electrodeposition is the
construction of the potential–pH diagrams10 in which the relative stability
of many possible phases involving the metal, its ion(s), possible oxide
phases and the solvent can be displayed simultaneously. This approach
describes the thermodynamic prerequisites of the electrodeposition only,
while it can account for neither the kinetic conditions of the reactions nor
the accessibility of the specific electrode potential in a given system.
Electrochemistry as a fabrication method is unique in the sense that
either the driving force of a reaction or the reaction rate can be easily
regulated. These operation modes correspond to the potential control
(potentiostatic mode, P) and the current control (galvanostatic mode, G),
respectively. While the application of the potential control is
indispensable for the investigation of the mechanism of a reaction, large-
scale industrial processes take advantage of the simplicity of the current
control. In the basic research of electrodeposition of nanostructures, both
control methods occur, but they are not always equivalent. For the
appropriate choice of the control parameter, the behaviour of the specific
system should be known very well.
Instead of relying on the equilibrium potentials, an electrochemical
reaction taking place with a nonzero rate should be described with the
appropriate kinetic equations. For a one-step reversible electrochemical
reaction, O + e = R (O and R are the oxidized and reduced form of the
same species), the dependence of the reaction rate (or current density, i)
on the electrode potential can be written as
α zFE (1 − α ) zFE
i = k AcR exp − kC cO exp − (2)
RT RT
where k is the rate constant of the anode or cathode reaction as shown by
the lower index, c is the concentration of the appropriate reactant, α is
the transfer parameter, E is the electrode potential with respect to a
reference electrode and z is the number of electron(s) transferred.
Equation (2) is known as the Butler–Volmer equation,11 and it describes
the polarization curve when the charge transfer is by far more hindered
than any serial reaction step (i.e., diffusion, adsorption, crystallization
92 L. Péter and I. Bakonyi
etc.). Polarization is the general term used for the difference between the
actual potential of an electrode and its rest potential.
Although Eq. (2) defines an exponential potential dependence of the
reaction rate (current density), one has to notice that for complex
electrode reactions the actual relationship of the current density and the
potential is much more complicated. For practical cases of electro-
deposition, at least the transport of the components in the electrolyte/melt
must also be taken into account. A key concept is the limiting current
density (see Fig. 5.1) which means that the electrochemical reaction takes
place at a maximum rate allowed by the transport of the reactant in the
electrode potential / a.u.
current density / a.u.
A
B
D C
Concentration of the metal ion is:
high low
(c'') (c')
c'' c'
metal ion concentration
A
metal ion concentration
A
B
C, D
0 0
dN 0 dN
0
distance from the cathode surface distance from the cathode surface
Fig. 5.1. Top: Comparison of the typical polarization curves of metal deposition at low
and high metal ion concentrations. At low concentration, the flat section of the
polarization curve corresponds to the limiting current region. Bottom: Steady-state metal
ion concentrations for high (left) and low (right) bulk metal ion concentrations as a
function of the distance from the cathode surface. Capital letters correspond to the
potentials shown on the polarization curves. The distance dN is the so-called Nernstian
diffusion layer thickness.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 93
electrolyte solution. Under limiting current condition, the electrolyte
boundary layer close the electrode surface is completely depleted, and
the current can no longer be increased as a result of the increase of the
polarization.
Although the theory of electrochemistry was developed with the help
of aqueous systems, nonaqueous solvents and ionic liquids have also
been used recently. Whereas the basic principles remain valid for the
latter two media, too, startpoints of the electrode potential scales are yet
to be defined due to the lack of appropriate reference electrodes.
Besides the precursor compounds of the metallic deposits, plating
baths almost always have several other components. These can be
electrochemically inactive salts to increase the conductivity of the bath
(often termed as supporting electrolytes), complexing agents to increase
the solubility of the metal ions, buffering agents to regulate the pH, and
other (mostly organic) compounds called additives.12 The latter family of
bath components exhibiting various functional groups adsorbs more or
less reversibly on the cathode and helps to achieve the desired
appearance, brightness, low surface roughness, high consistency with
reduced porosity, an appropriate crystal size, coatings with low internal
stress etc. The formulation of a bath for a specific purpose requires the
knowledge of lots of empirical information that can be found in the
technical literature.
The dependence of the crystallite size on the parameters of the
electrodeposition is of particular interest if nanostructures are aimed at.
The general finding is13,14 that the decrease in polarization (or current
density) or the increase of the surface concentration of the metal ions by
any means (increase in bulk concentration, bath agitation, temperature
elevation) leads to the enlargement of the crystals deposited.
Nevertheless, the increase of the crystallite size with the increase of the
current density was also reported,15,16 and the explanation is likely to be
the varying intensity of the side reaction (hydrogen evolution). Nano-
crystalline deposits can be obtained at highly inhibited deposition
conditions, and the application of pulsed current is also useful.17
94 L. Péter and I. Bakonyi
5.2.2. Electrodeposition of magnetic elements
Among magnetic elements, cobalt and nickel can be easily deposited in
elemental form from aqueous media. The standard electrode potentials
for the Ni/Ni2+ and Co/Co2+ redox systems are − 0.257 V and
− 0.280 V vs. SHE, respectively. These values are out of the stability
range of water; hence, hydrogen evolution as a side reaction is
unavoidable. Iron exhibits a more negative standard potential (− 0.447 V).
For iron, the difficulty of obtaining pure deposits stems from various
factors such as the sensitivity of iron to water and oxygen, the high
corrosion rate as compared to nickel and cobalt, the instability of the
electrolyte due to the formation of Fe3+, etc. Therefore, pure iron is
seldom obtained via electrodeposition. The standard electrode potential
of other magnetic elements (e.g., Gd) is so negative that their deposition
from aqueous electrolyte solutions is practically impossible.
5.2.3. Electrodeposition of magnetic alloys
The magnetic properties of an alloy can be tuned with the composition of
the deposit. Therefore, the codeposition modes of various pairs of metals
should be known.
Figure 5.2 serves as a general overview of the most common
codeposition modes. Equilibrium codeposition means that the
composition data all lie on the reference line y = x. This codeposition
mode is very scarce and cannot be found among magnetic alloys. Normal
codeposition is obtained if the deposition is regulated mostly by the
transport of the ions and is unaffected by any specific interaction of the
ions in the electrolyte with the cathode surface. In the normal
codeposition, the deposit is composed exclusively from the more noble
metal as long as the transport rate of its ions within the electrolyte is
sufficient to account for all current passed; in other words, the more
noble metal ions have the priority to be discharged. After achieving the
limiting current of the more noble metal, the codeposition of the less
noble metal takes place to such an extent that the total current passed can
be maintained. This deposition mode is observed for many magnetic–
non-magnetic pairs like Ni-Cu, Co-Cu, Co-Ru, Co-Ag or Co-Pt. The
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 95
yLN = nLN/(nLN+nMN) (in deposit)
1
o us
al
mo
an
m
r iu
lib
ui
eq
ar
ul
eg
ir r
al
rm
0 no
0 1
cLN/(cLN+cMN) (in electrolyte)
Fig. 5.2. A possible representation of the basic features of the most common codeposition
modes. yLN is the molar fraction of the less noble metal in the deposit, while the x axis
shows a similar ratio of the elements calculated with the concentration of the ions of the
alloy constituent metals in the electrolyte (the index MN refers to the more noble
element). All other technical parameters (temperature, total metal ion concentration in the
electrolyte, hydrodynamic conditions, current density) are constant.
irregular codeposition means that the ions of the less noble element are
discharged with the more noble metal at all concentration ratios, while
the deposition of the more noble metal is predominant. This is observed,
e.g., for the Zn-Cu pair. The anomalous codeposition is very common
among magnetic metals since the electrocrystallization of any
combination of Fe, Co Ni and Zn from aqueous electrolytes belongs to
this group. In this case, the less noble element, which can be present in a
minor concentration in the electrolyte, is deposited preferentially so that
its molar fraction in the deposit exceeds its electrolyte concentration ratio
at all compositions. This is why electrolytes used for Fe-Co-Ni plating
employ the precursor metal salts in concentrations increasing with the
standard potential of the metals.
Another codeposition mode not displayed in Fig. 5.2 is the so-called
induced codeposition. In this case, one element (for instance, Ni or Co)
can be deposited alone easily. However, another metal (e.g., Mo) cannot
be fully reduced from its higher-valency ions (MoO42−) in the absence of
the former metal. The reason is that the process stops at an intermediate
valency oxide (MoO2) whose further electroreduction is kinetically
96 L. Péter and I. Bakonyi
hindered. However, when the ions of both elements are present, the alloy
formation by electrodeposition can take place. This deposition mode is
general for any pairs of the (Fe, Co, Ni)-(Mo, W) combination. Although
it is not yet described in detail, the codeposition of Gd with Co can also
be classified as induced codeposition18 since the electrodeposition of
pure Gd is not feasible from aqueous solutions.
The mutual deposition preferences are often retained when more
than two elements are deposited from the same bath to form a multi-
component alloy.
5.2.4. Non-metallic deposits obtained with electrochemistry19
Some magnetic oxides can be obtained with a cathode process. Similarly
to the production of couprous/coupric oxides and zinc oxide, the key
process is the hydrogen evolution and/or the oxoanion reduction at the
cathode and the resulting increase in pH, leading to the precipitation of
the oxide via a chemical reaction. Here, charge transfer between the
magnetic precursor species and the cathode does not always take place.
Instead, the electric current changes the media, hence creating the
condition of an ordinary chemical reaction.
Some anodic reactions can also be used to produce magnetic oxides.
In these reactions, the precursor compound is indeed oxidized. The
solubility of the corresponding metal salt is much lower than that of the
starting form, and the metal ions of higher valency form an oxide. In this
way, higher-valency oxides of iron,20 cobalt and nickel can be obtained.
The kinetics of both cathode and anode reactions leading to oxide
formation is usually complicated and yet to be explored.
5.3. Electrodeposition: A Route Toward Magnetic Nanostructures
5.3.1. Electrodeposition of ultrathin magnetic films21
For the purpose of this section, a single electrodeposited layer shall be
called “ultrathin” if its total thickness does not exceed 2 nm. In this
thickness range, the growth of the deposit is often quantified in the
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 97
relevant literature by the number of equivalent monolayers (ML) instead
of nanometers.
The interest in ultrathin magnetic layers22,23 stems from various
features of these layers: (i) Magnetization can be observed only after the
accumulation of several atomic layers. The threshold thickness where the
ferromagnetic character can be observed depends on the chemical
composition of the layers deposited, the morphology of the deposit and
the substrate quality (composition, crystallite size and orientation).
(ii) There is a thickness range for many magnetic deposits where the
strength of the perpendicular magnetic anisotropy (PMA) exceeds the
shape anisotropy, and it leads to a perpendicular magnetization. PMA has
recently gained particularly strong interest due to the possibility of high-
density magnetic data storage. The characteristic thickness where PMA
is lost depends on the quality of both interfaces of the ultrathin magnetic
layer. Therefore, it may be a function of the electrolyte composition, too,
when an in-situ study is performed.
Ultrathin magnetic films can be deposited onto various substrates. In
most cases, highly textured noble metal surfaces of well-defined crystal
faces are used,24 Au(111) surface being the most common substrate. The
application of a specific crystal face of a noble metal has the advantage
that the nucleation pattern of the layer can be studied in-situ by using
Fig. 5.3. Sequence of in situ STM images (140 nm x 120 nm) showing the initial stages
of Fe/Au(111) growth at −1.5 V vs. saturated mercury sulfate electrode.25 Vertical arrows
show in which order the horizontal linescans were recorded, and numbers indicate the
local deposit thickness (in ML). Parts (a) and (b) show subsequently recorded pictures for
the same area. Reprinted from Physica B: Condensed Matter Vol. 354, Gündel et al.,
Electrodeposition of Fe/Au(111) ultrathin layers with perpendicular magnetic anisotropy,
page 243, Copyright (2004), with permission from Elsevier.
98 L. Péter and I. Bakonyi
scanning probe microscopy (see Fig. 5.3). Surface reconstruction during
the deposition of the metal adlayers is commonly observed by STM. The
nobility of the substrate makes it possible to study the behaviour of the
layer under anodic conditions, too (dissolution).
The concentration of the electroactive species in the electrolyte has to
be very low (typically in the range of 1–10 mM) in order to keep the
number of atomic layers well controlled with the deposition time at
the applied electrode potential. It has to be noted that the nominal
thickness of the deposit does not determine the magnetic properties
unambiguously, since the growth morphology of the deposit may also
depend on the electrode potential applied.
Magnetization of the ultrathin layers can be studied with various
methods. Basically, all types of magnetic measurements with high
enough sensitivity are viable for thin layers, too. Nevertheless, ex-situ
measurements are very cumbersome due to the necessity of the
protection of the sample surface from oxidation and the difficulty of the
transfer of the sample from the electrolyte to an inert environment.
Therefore, in-situ methods like magneto-optical Kerr effect (MOKE) and
Table 1. Summary of the studies of magnetic properties of electrodeposited ultrathin
magnetic films. tON : onset thickness of ferromagnetism; t*: transient thickness to
in-plane magnetization (in contact with the media given).
Substrate Deposit tON (ML) t* (ML) Reference(s)
Ni 6 --- 21
1.6 6.2-7.2 (Cu capped) 21, 26, 27
1.6 (SO42–- solution)
Au (111) Co 4 (SCN– solution) 27
20 (solution, pH = 8.5)
--- function of the 28
deposition potential
(Cu capped)
--- > 8 (Au capped) 29
Fe 1 2 ( SO42–- solution) 21, 25, 30
--- (Cu capped)
Ag (poly) Co < 5.5 < 15 31
Co --- 1.5 (SO42–- solution) 32
Cu (111) 1.5 2 (SO42–- solution) 33
Fe 2 8 (SO42–- solution) 33
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 99
alternating gradient field magnetometry are used most often. The in-situ
measurement of the Mössbauer spectrum of the deposit is also
informative on both cluster formation and the magnetic interaction of the
atoms deposited.31
The magnetic properties of electrodeposited ultrathin magnetic films
are summarized in Table 1.
5.3.2. Nanocrystalline magnetic deposits34,35
The crystallite size of a deposit is regulated by the relative importance of
the nucleation of new crystals and by the growth of existing ones.
Nanocrystalline or even amorphous deposits can be obtained if the
growth of the existing crystals is suppressed, but the rate of nucleation is
high.
The suppression of the growth of the existing crystals can be easily
achieved in electrodeposition. If the bath contains additives12 whose
preferred binding site at the metal–electrolyte interface is the same as the
entry position of the new metal atoms into the crystal, then the growth is
blocked or inhibited. These preferred sites are the step edges and kink
positions of the metal. Of course, the adsorption of the additive should be
kinetically reversible to avoid the significant incorporation of non-
metallic components.
Although the inhibition intensity14 is never measured exactly, it is
commonly used for describing bath or additive properties. Roughly
speaking, the inhibition intensity increases with surface coverage by the
additive and with bond strength between the additive and the metal
surface (hence, with additive concentration until a certain saturation
level). Electrolytes developed for the deposition of nanocrystalline
metals almost always contain some components that are capable of
adsorbing at the metal surface.
It is worthwhile to mention that the larger the number of components
of a deposit, the more likely is the formation of a nanocrystalline deposit
since the non-identical alloying atoms hinder the long-range ordering in
the same way as the temporarily adsorbed molecules at the preferred
growth sites.
100 L. Péter and I. Bakonyi
Table 2. Electrodeposited nanocrystalline magnetic metals
(where magnetic properties were of particular interest).
Component(s) Reference(s) Component(s) Reference(s)
Fe 36, 37 Fe-Pd 39, 50
Ni 38–40 Co-Ni 51
Co 41 Co-Fe 51
Ni-Cu 40 Co-Ni-P 52
Ni-Fe 36, 42–46 Co-Fe-P 53
Co-Pd 47, 48 Co-Fe-Ni 51, 54–56
Co-Pt 49
The rate of nucleation can be enlarged if the surface concentration of
the adatoms increases. However, by using a d.c. current, the maximum
deposition rate (and hence the maximum surface concentration of the
adatoms) is determined by the mass transport limited current density.
This limitation can be overcome if pulsed current17,57–59 is used. If a short
but very high cathodic current pulse is applied, the local metal ion
concentration in the neighbourhood of the cathode is much larger than
under d.c. conditions. Hence, the surface concentration of the adatoms
becomes very high, but temporarily only. The high current pulse needs to
be followed by an off-time period during which the electrolyte can relax
and the bulk metal ion concentration is nearly regenerated in the vicinity
of the cathode. The actual crystallite size is determined by the entire
deposition process (on-time, off-time, duty cycle, and electrolyte
composition). More complicated current waveforms including anodic
pulses in order to remove the hydrogen produced have also been
reported.17 From ionic liquids, the application of d.c. current leads to
nanocrystalline deposits more often than for aqueous solutions37 because
the chemical properties of the solvent automatically provide some
inhibition.
Literature on the electrodeposited nanocrystalline magnetic materials
is summarized in Table 2.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 101
5.3.3. Deposition of metastable precursor alloys and
their treatment for obtaining granular magnetic alloys
A common method to obtain magnetic nanoparticles dispersed in a non-
magnetic metallic matrix is the following. A metastable alloy is obtained
by the codeposition of a magnetic element with a non-magnetic one, the
latter being the majority (matrix-forming) component. Metastability
originates from the low miscibility of the constituents. Upon a heat
treatment, the activation of the diffusion of the atoms leads to the
precipitation of small crystals of the magnetic metal. Figure 5.4
demonstrates how the phase segregation during annealing can be
detected by the evolution of the diffraction pattern with time, and the
random distribution of the magnetic particles can also be seen in the
elemental map.
The electrochemical deposition of such systems is analogous to the
physical deposition60,61 (parallel sputtering or evaporation of the compo-
nents), and the purpose of the subsequent annealing is exactly the same.
The size distribution of the magnetic grains obtained by annealing is
usually lognormal, regardless of the preparation method of the precursor
Fig. 5.4. Left: Evolution of the X-ray diffractogram of an electrodeposited metastable
Co-Cu alloy with annealing time.62 The segregation of Co is indicated by the emerging
hcp Co line and the shift of the diffraction angle of the Cu peaks. Right: Elemental map
of a Co-Cu alloy annealed for 1.5 hours. The light spots are the small Co crystals in the
Cu matrix. Reproduced from Journal of Materials Science 39, 5701–5709 (2004);
T. Cohen-Hyams et al., Microstructural dependence of giant-magnetoresistance in
electrodeposited Cu-Co alloys, Figs. 5.1 and 5.4.d. With kind permission of Springer
Science and Business Media.
102 L. Péter and I. Bakonyi
alloy, and the mean particle size and interparticle distance can be regu-
lated by the initial concentration and the conditions of the heat treatment
(temperature, duration). Several reports are available on magnetic and/or
magnetotransport properties of electrodeposited Co-Cu,62–76 Co-Ni-Cu77
and Co-Ag65 systems.
From the point of view of the electrolysis, it is necessary to promote
the alloy formation, despite the constituent elements cannot form a stable
alloy. In this case, the selection of the bath constituents needs to follow
the general rule that the deposition potential of the metals should be
brought close to each other. This usually makes it necessary to use some
kind of complexing agent that shifts the deposition potential of the more
noble element to the cathodic direction to a larger extent than the
deposition potential of the less noble one. The application of the
complexing agent is likely to lead to the formation of a nanocrystalline
precursor alloy, although the crystallite size is usually not a subject of
investigation prior to the heat treatment. The recrystallization certainly
leads to a grain coarsening of the matrix forming non-magnetic metal.
If the alloy formation is not forced by the appropriate choice of the
bath formulation, the codeposition of immiscible metals may lead to the
formation of a granular alloy in one step.
5.3.4. Electrodeposition of magnetic/non-magnetic multilayer films
with nanometer-scale periodicity
Electrodeposition of metallic multilayers has been the subject of research
for a long time.78 However, the significance of the magnetic/non-
magnetic multilayers comes from the giant magnetoresistance (GMR)
effect which was discovered only in the late 1980’s79,80 and was also
awarded with the Nobel Prize in Physics two decades later. A few years
after the discovery of the GMR it was shown81 that such nanostructured
materials can be produced by electrochemical deposition, too. An
example for a multilayer structure obtained by electrodeposition is
presented in Fig. 5.5, and typical magnetoresistance curves are shown in
Fig. 5.6. The details of the deposition method82 and literature data on
electrodeposited multilayers with GMR83–85 are summarized in various
reviews (see also Part III- Chapter 7 of this book).
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 103
100 nm
Fig. 5.5. Cross-sectional TEM image of an electrodeposited magnetic/non-magnetic
multilayer sample with alternating thick and thin magnetic layers86 (pseudo-spin valve
structure). The Cu layers appear in dark. The repeating structure elements are:
NiCo(6.0 nm)/Cu(3.6 nm)/NiCo(2.0 nm)/Cu(3.6 nm). The arrow indicates the growth
direction of the sample.
Metallic multilayers can be deposited by contacting the substrate
alternatingly with two different electrolytes,87 each of them capable of
yielding one type of layer only. This method is rather cumbersome for
producing nanolayers, and the deterioration of the sample surface during
the electrolyte change is difficult to eliminate. Hence, nanoscale multi-
layers are usually obtained with the single bath method.78
The key feature of the single bath method is that the salt of the more
noble non-magnetic metal is applied in a fairly low concentration, while
the concentration of the salt of the magnetic metal is often close to the
limit defined by its solubility. The normal codeposition mode is a
prerequisite; otherwise the non-magnetic metal cannot be obtained in the
pure form. For pairs like (Ni,Co,Fe)/(Cu,Ag) this condition is fulfilled.
At low cathodic current density or at moderately negative potential the
non-magnetic “spacer” layer is deposited with no magnetic impurity. At
much larger cathodic current density or at a more negative potential, the
deposit obtained contains the magnetic metal as the major component
with some non-magnetic impurity. If the molar fraction of the non-
magnetic component in the magnetic layer is not very high, it does not
impact significantly the magnetic properties of the magnetic layer.
It is clear from the above train of thoughts that the deposition
potentials of the magnetic and non-magnetic metals should not be too
close, and there must be a potential window where the non-magnetic
metal can be deposited alone. Hence, for multilayer deposition, one
104 L. Péter and I. Bakonyi
should not strive for a good alloy formation, but rather a sufficient
separation of the potential regimes of the deposition of various metals is
favourable. Multilayers with high magnetoresistance are usually obtained
from baths containing no additives, brighteners and complexing agents
because such bath components are deleterious for GMR.88,89
Both the magnetic and the non-magnetic layers can be obtained by
either a current pulse or a potential pulse. However, the potential control
for the deposition of the non-magnetic layer is much better to obtain a
good layer structure since the spontaneous exchange reaction between
the metals (e.g., Co + Cu2+ = Cu + Co2+) can then be excluded.
The deposition potential of the more noble metal has to be tuned
precisely so that neither the dissolution nor the codeposition of the
[Co(2.5nm)/Cu(0.5nm)] X160
Magnetoresistance: ∆R/R (%)
longitudinal discontinuous Cu layer
0 transverse
-2 longitudinal
transverse
-4 [Co(2.5nm)/Cu(3.0nm)] X91
high Cu concent in the Co layer
-6
longitudinal
-8
transverse [Co(0.8nm)/Cu(3.0nm)] X130
-10
very low Cu concent in the Co layer
-8 -4 0 4 8
H (kOe)
Fig. 5.6. Typical magnetoresistance curves of various electrodeposited Co/Cu multi-
layers.90 Pinholes in the magnetic layer result in anisotropic magnetoresistance due to the
direct ferromagnetic coupling of the magnetic layers (top curves). A high Cu content in
the Co layer leads to a superparamagnetic behaviour because of the magnetic
fragmentation of the Co layer (middle). A sample with a satisfactory separation of the Co
layers with low Cu content exhibits giant magnetoresistance with a dominant
ferromagnetic contribution (bottom). Longitudinal and transverse magnetoresistance
means that the external magnetic field was parallel and perpendicular to the measuring
current, respectively.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 105
magnetic metals can take place. One of the optimization methods is
based on the analysis of the current transients during the less negative
deposition pulse,91 while another one relies on the weight change of the
substrate by using an electrochemical quartz crystal microbalance.92 The
current transient analysis can be regarded as being more sensitive since
the data acquisition frequency can be higher and it measures an actual
reaction rate instead of an integral parameter determined from the total
weight change.
Besides the appropriate technical realization, a general requirement of
the formation of subsequent layers at fairly large areas is that the
nucleation of the alternating layers onto each other should take place
easily. The smallest possible difference in the lattice parameters in the
pure forms of the individual layers favours the growth of a deposit whose
composition is modulated in the growth direction only (this is the case
for Ni/Cu and Co/Cu pairs). However, a large difference in the lattice
parameters prevents the continuous coverage of the previous layer and
may result in an island-like growth, yielding a magnetic structure similar
to the granular alloys.
Apart from metal/metal type magnetic multilayers, some attempts
were also made to deposit metal/insulator type nanoscale iron/iron-oxide
multilayers by using a cathodic/anodic pulse sequence.93
5.3.5. Deposition of nanostructures at preferred nucleation sites
When a metal ion is discharged on the surface of the parent metal, the
adatom tends to occupy the so-called half crystal position,94 especially if
the deposition conditions are not very far from the equilibrium. The
special feature of the half-crystal or kink position is that either the
removal or the incorporation of an atom creates a position with identical
properties as the initial one.94 Hence, these kink positions are the
locations where the step edge can propagate in the case of either the
deposition or the dissolution of a metal.
A single adatom on the single-crystal surface of the parent metal has
higher energy than that of an atom in the kink position. However, this
energy difference is quite small when it is compared with the energy
difference of a metal adatom at similar crystal positions of a non-metallic
106 L. Péter and I. Bakonyi
surface. The energy of a metallic adatom on a carbon or silicon surface is
so high that the nucleation of a metallic deposit cannot take place
anywhere else but on an active site or along a step edge of the surface.
After the nucleation, which is mostly instantaneous, further electro-
deposition takes place on these grains only, and the deposit remains
discontinuous until the nominal (average) coverage achieves a few
hundred nanometers.
The deposit formation at the step edges of highly oriented
pyrolitic graphite (commonly known as HOPG) is often termed as
electrochemical step-edge decoration.95–97 After the initial nucleation and
grain growth, the coalescence of the grains results in nanowires, this is
why the appearance of the wire is often similar to a string of pearls, as
seen in Fig. 5.7. The diameter can be finely tuned by applying an either
cathodic current (increasing wire diameter) or anodic current (wire
thinning) after the wire formation. The deposition of Ni96 and Ni-Pd98
wires has been reported so far, though the successful deposition was not
accompanied with the study of magnetic properties.
If the nucleation of the metal takes place at a high-energy single
crystal surface (e.g., Si), the nucleation is also instantaneous,99 which
leads to the growth of particles with a narrow size distribution if the
deposition does not last until the coalescence of the grains. Interestingly,
this deposition mode often leads to the formation of core-shell magnetic
particles.100,101
Fig. 5.7. Scanning electron micrograph of a Ni nanowire system obtained by electro-
deposition onto HOPG. The wires were formed along the step edges of the HOPG
surface. Reprinted in part with permission from Ref. 96. Copyright (2002) American
Chemical Society.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 107
From technical point of view, the preparation of nanoparticles or
nanowires usually comprises several steps. A preliminary current pulse is
often applied to achieve appropriate conditions for the nucleation, and
then a short high-potential cathodic nucleation pulse is applied, followed
by the growth pulse with a low or moderate overpotential. The
concentration of the electroactive species is usually much lower than in
normal plating electrolytes.
5.3.6. Electrodeposition into templates
The word “template” is used here as the general term for the solid spatial
confinement at the electrode surface that regulates the shape of the
deposit being formed. Hence, the deposit will not grow in a bulk form
but will fill up the empty space defined by the template. The templates
prepared with lithographic methods102 have now been replaced with
etched self-organized templates. The emerging importance of this field is
well explained by the fact that objects with an aspect ratio of as high as
103–104 and with an areal pore density of around 1011 cm-2 can be
obtained with a fairly simple method, making electrodeposition an
indispensable tool of nanotechnology.
A very simple template can be a piece of sintered glass disk covered
with a conducting coating at one side and filled up with the electrolyte.
Therefore, the deposit fills up the pores within the sintered glass disk.
Such pore structures are random or irregular.
A more regular template is a plastic (mostly polycarbonate)
membrane that is bombarded with heavy ions perpendicular to the
surface.103 Along the ion tracks the bond structure of the polymer is
destroyed, and the track can be etched, typically with a strong alkaline
solution to widen the columnar pores. This type of template is regular in
the sense that the pores are perpendicular to the surface at least to the
same extent as the trajectory of the bombarding ions is unchanged by the
subsequent collisions. The pore density is defined by the ion dose, while
the pore diameter can be tuned by the etching time. Hence, the relative
position of the pores is random, so this system is still laterally disordered.
However, this type of template is suitable for the preparation of a single-
ion track pore and for the study of an individual wire deposited within.104
108 L. Péter and I. Bakonyi
Fig. 5.8. Left: Scanning electron micrograph of an etched porous alumina template
(top view). Middle: AFM picture of the top of the pore structure. Right: Cross-sectional
view of the AAO template. Anodization conditions: 1 M phosphoric acid at 195 V.
Anodization was repeated 3 times by dissolving the oxide layer produced after the 1st and
2nd step with 0.2 M chromic acid. By courtesy of J. Gong and G. Zangari.105
The most ordered type of template up to now is the anodic aluminum
oxide (AAO).106,107 The preparation method is that the aluminum foil is
immersed in an acidic solution (phosphoric, oxalic or sulfuric acid), then
it is anodized at a high voltage (20–195 V). This anodization leads to the
growth of a porous oxide whose surface is still disordered, but the
bottom part (i.e., the metal/oxide boundary) becomes ordered upon
extended anodization. Then the initial oxide is dissolved, and the
anodization is repeated. Therefore, a laterally ordered hexagonal pore
system can be produced (see Fig. 5.8) in which the ordered areas form
domains with some disorder within the domain walls. The disadvantage
of this method is the lengthy multistep preparation of the template.
The available pore size for AAO templates is 50–500 nm, but the ratio
of the pore diameter to the pore axis distance is practically fixed, and it
can be modified only by a pore widening during a subsequent chemical
etching. The degree of ordering can be increased by a mechanical
initialization of the pore formation by using an imprinting method. If the
aluminum is fully oxidized, one side of the resulting alumina honeycomb
structure can be coated with a metal by evaporation to ensure the
conductivity, or the deposition can be started by making use of the
remaining aluminum.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 109
Fig. 5.9. Typical magnetization curves of metallic nanowires electrodeposited into the
pores of AAO templates.111 Deposit structure: 30 nm diameter and 700 nm long Ni wires.
Continuous curves show SQUID magnetization data for the field direction shown in the
figure, and filled circles refer to magnetic force microscopy measurements. The
hysteresis loops are determined by the shape anisotropy of the wires. Reprinted from
Journal of Magnetism and Magnetic Materials, Vol. 249, K. Nielsch et al., High density
hexagonal nickel nanowire array, page 236, Copyright (2002), with permission from
Elsevier.
Recently, porous templates based on anodized titania108 or porous
silicon109 have also become available. The deposition on silicon
substrates patterned with submicrometer holes by using focused ion
beam110 can be very significant since any regular (and hence not only
hexagonal) surface structure is easily available by this method.
There are two basic growth modes in a template. One of them is the
growth of a wire. In this case, the growing column fills the pore entirely.
However, it may happen that the plating in the vicinity of the walls is
accelerated as compared to the center of the cavity. This leads to a fast
depletion of the electrolyte within the metal channel being formed, and
the growth mode becomes tubular. It was found112 that a low current
efficiency and the resulting electrolyte flow pattern induced by the
release of hydrogen bubbles in the center of the template channel can
promote the tubular growth mode.
There is a great variety of magnetic nanostructures obtained by
electrodeposition in templates (see, e.g., Fig. 5.9). Besides homogeneous
magnetic nano-wires, multisectioned nanowires can be obtained; here,
the length of each section in the wire is much larger than the wire
110 L. Péter and I. Bakonyi
diameter. The structure is, however, called a multilayered nanowire
when the layer thickness is much smaller than wire diameter. The
guidelines of the electrodeposition of each type of structure are identical
to those discussed for their deposition without any template. In very
special (and unpredictable) cases a core-shell structure can also form
spontaneously.113–115
After the removal of the template, a second electrolysis step can
provide a coated nanowire system116 in which either the internal wire
core or the coating can be a magnetic metal. The field of template-based
electrodeposited magnetic nanostructures is expanding very fast, and
hence the reader is advised to perform an own literature search for the
topic of interest.
5.3.7. Electrodeposition on surfaces modified by self-assembly of
colloids
Self-assembly of adsorbing molecules at a metal surface has been well
known for a long time. However, entities much larger than a single
molecule are also capable of self-assembly, forming a structure that can
be used similarly to solid templates for electrodeposition.
The self-ordering of non-conducting (e.g., polystyrene, poly(methyl-
metacrylate) etc.) colloidal particles on the surface of a metal is similar
to the crystal growth that leads to ordered domains with some domain
boundaries in between. Depending on the amount of colloid particles
applied, a single adlayer structure or a multilayer structure can be
produced. The empty space between the ordered particles is suitable for a
fill-up process by electrodeposition. In this case, the magnetic nano-
structure will be the inverse of the non-conducting particle system,
exhibiting a typical periodicity of some hundreds of nanometers. After
the electrolysis, the nanostructure produced can be studied as it is or after
the removal of the colloids by dissolving them in an appropriate organic
solvent. Besides structural properties117 and magnetoresistance,118
coercivity behaviour of the system as a function of the colloid particle
size119–121 has deserved particular attention so far.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 111
Similarly to the nanotube formation in the porous solid templates, the
specific interaction of the metal to be deposited with the structure-
forming colloid can lead to an incomplete void filling, hence leading to a
foam-like magnetic nanostructure (see Fig. 5.10).122 If the empty space
between the colloids is first filled up with a conductive polymer, then the
colloids are removed, the inversion of the ordinary template is suitable
Fig. 5.10. Top: SEM image of structured Ni film prepared using 0.5 µm polystyrene
spheres (a) and cross-sectional view of a thick film after cleaving (b). In both cases the
white marker corresponds to 1 µm. Bottom: Coercivity of various magnetic hollow
structures as a function of the deposit thickness normalized to the diameter of the colloid
particles. Reprinted with permission from Ref. 119. Copyright (2006), American Institute
of Physics.
112 L. Péter and I. Bakonyi
for the electrodeposition of an ordered system of magnetic
nanospheres.123 A similar method can be used with silica spher es and a
polymer filling, and the silica spheres can be removed with hydrogen
fluoride.124 The key factor is the chemical dissimilarity of the self-
ordered nanospheres and the material filling the empty space between
them.
Another mean of the self-assembly is the spontaneous structure
formation of the two-phase mixture of a lyotropic liquid crystal with
water. Here, the hydrophobic and hydrophilic regions form separated
zones with cylindrical hydrophilic columns within a practically
insulating matrix. The electrolyte formation by dissolving various salts in
the water-rich phase usually does not modify significantly the miscibility
properties of the system. The columnar regions with the aqueous
electrolyte solution behave similarly to the pores in a solid template
while the non-conducting liquid crystal matrix serves as a wall. After the
removal of the liquid crystal, one can obtain an ordered system of
nanorods attached to the substrate. Nanostructured cobalt films have
been obtained with this method.125
5.3.8. Suspension plating with magnetic particles
Suspension plating126–128 was originally developed for the codeposition of
inert particles in order to enhance tribological, wear resistance and
corrosion properties. The host metal and the particles embedded (Al2O3,
ZrO2, SiO2, SiC, Si3N4, diamond etc.) provide the desired properties
together. While the particle to be incorporated is never involved in an
electrochemical reaction, its surface charge and adsorption properties can
strongly influence the entire codeposition process.
Reports on codeposited magnetic particles in electroplating are
scarce. Incorporation of Ni particles in Zn deposit129 and barium ferrite
particles in various magnetic metals,130 even combining with a
lithographic surface patterning,131 has been reported. However, the
particle size was in both cases in the order of micrometers. With the
application of nanosized particles, this field is expected to develop
significantly in the near future.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 113
5.3.9. Formation of suspended magnetic particles by electrochemistry
Particles produced by electrodeposition sometimes do not attach to the
substrate surface well enough to make a continuous layer or an ordered
system. This statement is particularly valid for non-metallic particles.
Hence, the particles can leave the vicinity of the electrode, and in the
case when the particles within the electrolyte are not stabilized by any
mean, the product can be precipitated.132 However, the stabilization of
the particles as a suspension can be promoted by the application of
various surfactants.133–135 The surfactant also has a role in the regulation
of the particle size. Magnetic particles with a typical diameter of
5–50 nm can be synthesized in this manner.
5.4. Summary
The sections above presented a variety of electrodeposited nanostructures
following the logical order of the deposition modes and conditions.
Figure 5.11 gives another type of overview, showing that principally
different deposition modes may result in similar structures, although
there is very little commonality in the methods themselves.
Electrodeposition of nanostructures is a fast-expanding field, and the
versatility of available electrodeposited nanostructures is almost
unlimited. From the point of view of technical application, electro-
deposition of magnetic nanostructures is very promising in fields where
it can contribute to the development of information storage technology
(magnetic nanodots, deposition in regular templates). It can be
anticipated that electrodeposition in templates may gain some important
role in manufacturing of magnetic storage media in the near future.
Acknowledgment
This work was supported by the Hungarian Scientific Research Fund
(OTKA) via the grant # K-60821.
114 L. Péter and I. Bakonyi
Ultrathin magnetic deposit Nanocrystalline deposit
deposit (magnetic element)
substrate (non-magnetic)
< 200 nm
Heat-treatment of a Granular alloy Suspension plating
metastable deposit
magnetic grains
non-magnetic matrix
Multilayer film Nanoscale deposits at the step edges
magnetic layer
non-magnetic layer
template with nanograin nanowire
channels (coalescence of particles) step edge
Nanotubes self-assembly of colloid
particles on the electrode
electrodeposition
homogeneous multilayered
Nanowires
template
multi-sectioned removal “Inverse” magnetic
structure with voids
repeated
coating
“Magnetic foam”
Fig. 5.11. A schematic summary of the main types of magnetic nanostructures
that can be obtained with electrodeposition.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 115
References
1. A. J. Bard, L. M. Faulkner, Electrochemical Methods. Fundamentals and
Application, 2nd Edition (John Wiley & Sons, Inc., New York, 2001).
2. C. H. Hamann, A. Hamnet, W. Vielstich, Electrochemistry (WILEY-VCH,
Weinheim, Gemany, 2007).
3. C. M. A. Brett, A. M. O. Brett, Electrochemistry. Principles, Methods, and
Application (Oxford University Press, Oxford, UK, 1993).
4. J. O’M. Bockris, A. K. N. Reddy, Modern Electrochemistry, Second Edition
(Kluwer Academic Publishers, New York, USA, 1998).
5. W. Plieth, Electrochemistry for Materials Science (Elsevier, Amsterdam, 2007).
6. A. Brenner, Electrodeposition of Alloys I-II (Academic Press, New York, USA
1963).
7. W. H. Safranek, The Properties of Electrodeposited Metals and Alloys –
A Handbook (American Elsevier Publishing, New York, 1974).
8. J. W. Dini, Electrodeposition (Noyes Publications, Park Ridge, New Jersey, USA,
1993).
9. M. Paunovich, M. Schlesinger, Fundamentals of Electrochemical Deposition (John
Wiley & Sons, Inc., New York, USA, 1998).
10. M. Pourbaix, Atlas of Electrochemical Equilibra in Aqueous Solutions (National
Association of Corrosion Engineers, Houston, Texas, USA, 1974).
11. T. Erdey-Grúz, M. Volmer, Z. Phys. Chem. A 150 (1930) 203.
12. L. Oniciu, L. Muresan, J. Appl. Electrochem. 21 (1991) 565.
13. J. W. Dini, Plat. Surf. Fin. 75 (1988) 10.
14. R. Winand, Electrochim. Acta 39 (1994) 1091.
15. I. Bakonyi, E. Tóth-Kádár, L. Pogány, Á. Cziráki, I. Gerıcs, K. Varga-Josepovits,
B. Arnold, K. Wetzig, Surf. Coat. Technol. 78 (1996) 124.
16. E. Tóth-Kádár, I. Bakonyi, L. Pogány, Á. Cziráki, Surf. Coat. Technol. 88 (1997)
57.
17. J. C. Puippe, F. Leaman eds., Theory and Practice of Pulse Plating (American
Electroplaters and Surface Finishers Society, Orlando, Florida, 1986).
18. R. Mishra, E. J. Podlaha, Electrochem. Solid-State Lett. 9 (2006) C199.
19. G. H. A. Therese, P. V. Kamath, Chem. Mater. 12 (2000) 1195.
20. T. A. Sorenson, S. A. Morton, G. D. Waddill, J. A. Switzer, J. Am. Chem. Soc. 124
(2002) 7604.
21. P. Allongue, F. Maroun, in Electrocrystallization in Nanotechnology, ed. G. Staikov
(Wiley-VCH, Weinheim, Germany, 2007), p. 217.
22. J. A. C. Bland, B. Heinrich, Ultrathin Magnetic Structures I-IV (Springer-Verlag,
Berlin, 2005).
23. C. A. F. Vaz, J. A. C. Bland, G. Lauhoff, Rep. Prog. Phys. 71 (2008) 056501.
24. P. Allongue, F. Maroun, Curr. Opin. Solid State Mater. Sci. 10 (2006) 173.
116 L. Péter and I. Bakonyi
25. A. Gündel, T. Devolder, C. Chappert, J. E. Schmidt, R. Cortès, P. Allongue,
Physica B 354 (2004) 282.
26. L. Cagnon, A. Gündel, T. Devolder, A. Morrone, C. Chappert, J. E. Schmidt,
P. Allongue, Appl. Surf. Sci. 164 (2000) 22.
27. J. G. Borges, P. Prod’homme, F. Maroun, R. Cortès, J. Geshev, J. E. Schmidt,
P. Allongue, Physica B 384 (2006) 138.
28. J. L. Bubendorff, E. Beaurepaire, C. Mény, J. P. Bucher, J. Appl. Phys. 83 (1998)
7043.
29. P. Prod’homme, F. Maroun, R. Cortès, P. Allongue, J. Hamrle, J. Ferré, J. P. Jamet,
N. Vernier, J. Magn. Magn. Mater. 315 (2007) 26.
30. A. Gundel, A. Morrone, J. E. Schmidt, L. Cagnon, P. Allongue, J. Magn. Magn.
Mater. 226–230 (2001) 1616.
31. I. Dézsi, Cs. Fetzer, Electrochem. Commun. 9 (2007) 1846.
32. W. Schindler, J. Kirschner, Phys. Rev. B 55 (1997) R1989.
33. W. Schindler, O. Schneider, J. Kirschner, J. Appl. Phys. 81 (1997) 3915.
34. T. Watanabe, Nano-Plating (Elsevier, Amsterdam, 2004).
35. U. Erb, K. T. Aust, G. Palumbo, in: Nanostruct. Mater., ed. C. C. Koch (Noyes
Publications / William Andrew Publishing, Norwich, New York, USA, 2002),
p. 179.
36. M. L. Trudeau, Nanostruct. Mater. 12 (1999) 55.
37. H. Natter, M. Bukowski, R. Hempelmann, S. Zein El Abedin, E. M. Moustafa,
F. Endres, Z. Phys. Chem. 220 (2006) 1275.
38. M. J. Aus, B. Szpunar, A. M. El-Sherik, U. Erb, G. Palumbo, K. T. Aus, Scr. Met.
Mater. 27 (1992) 1639.
39. S. Lauer, Z. Guan, H. Wolf, H. Natter, M. Schmelzer, R. Hempelmann, T. Wichert,
Nanostruct. Mater. 12 (1999) 955.
40. H. Wolf, Z. Guan, X. Li, T. Wichert, Hyperfine Interact. 136–137 (2001) 281.
41. M. J. Aus, C. Cheung, B. Szpunar, U. Erb, J. Mater. Sci. Lett. 17 (1998) 1952.
42. H. L. Seet, X. P. Li, Z. J. Zhao, Y. K. Kong, H. M. Zheng, W. C. Ng, J. Appl. Phys.
97 (2005) 10N304.
43. L. J. Gao, G. W. Anderson, P. R. Norton, Z. H. Lu, J. P. McCaffrey, M. J. Graham,
J. Appl. Phys. 78 (1995) 5795.
44. C. Cheung, G. Palumbo, U. Erb, Scr. Met. Mater. 31 (1994) 735.
45. F. Czerwinski, J. A. Szpunar, U. Erb, J. Mater. Sci. – Mater. El. 11 (2000) 243.
46. H. Li, F. Ebrahimi, Mater. Sci. Eng. A 347 (2003) 93.
47. R. D. Noce, N. Barelli, P. T. A. Sumodjo, D. R. Cornejo, A. V. Benedetti, J. Magn.
Magn. Mater. 306 (2006) 199.
48. R. D. Noce, N. Barelli, R. F. C. Marques, P. T. A. Sumodjo, A. V. Benedetti, Surf.
Coat. Technol. 202 (2007) 107.
49. F. Wang, K. Hosoiri, S. Doi, N. Okamoto, T. Kuzushima, T. Totsuka, T. Watanabe,
Electrochem. Commun. 6 (2004) 1149.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 117
50. F. M. Takata, G. Pattanaik, W. A. Soffa, P. T. A. Sumodjo, G. Zangari,
Electrochem. Commun. 10 (2008) 568.
51. D. Kim, D. Y. Park, B. Y. Yoo, P. T. A. Sumodjo, N. V. Myung, Electrochim. Acta
48 (2003) 819.
52. D. Y. Park, N. V. Myung, M. Schwartz, K. Nobe, Electrochim. Acta 47 (2002)
2893.
53. E. E. Kalu, J. Solid State Electrochem. 11 (2007) 1145.
54. T. Nakanishi, M. Ozaki, H. S. Nam, T. Yokoshima, T. Osaka, J. Electrochem. Soc.
148 (2001) C627.
55. E. H. du Marchie van Voorthuysen, F. T. ten Broek, N. G. Chechenin, D. O.
Boerma, J. Magn. Magn. Mater. 266 (2003) 251.
56. Y. Zhang, D. G. Ivey, Chem. Mater. 16 (2004) 1189.
57. G. Devaraj, S. Guruviah, S. K. Seshdari, Mater. Chem. Phys. 25 (1990) 439.
58. N. V. Mandrich, Met. Fin. 100 (2002) 359.
59. M. S. Chandrasekar, M. Pushpavanam, Electrochim. Acta 53 (2008) 3313.
60. J. Q. Xiao, J. S. Jiang, C. L. Chien, Phys. Rev. Lett. 68 (1992) 3749.
61. A. E. Berkowitz, J. R. Mitchell, M. J. Carey, A. P. Young, S. Zhang, F. E. Spada,
F. T. Parker, A. Hutten, G. Thomas, Phys. Rev. Lett. 68 (1992) 3745.
62. T. Cohen-Hyams, J. M. Plitzko, C. J. D. Hetherington, J. L. Hutchinson,
J. Yahalom, W. D. Kaplan, J. Mater. Sci. 39 (2004) 5701.
63. Y. Ueda, M. Ito, Jpn. J. Appl. Phys. 33 (1994) L1403.
64. V. M. Fedosyuk, O. I. Kasyutich, D. Ravinder, H. J. Blythe, J. Magn. Magn. Mater.
156 (1996) 345.
65. H. Zaman, A. Yamada, H. Fukuda, Y. Ueda, J. Electrochem. Soc. 145 (1998) 565.
66. G. R. Pattanaik, D. K. Pandya, S. C. Kashyap, J. Magn. Magn. Mater. 219 (2000)
309.
67. S. Ge, H. Li, C. Li, L. Xi, W. Li, J. Chi, J. Phys.: Condens. Matter 12 (2000) 5905.
68. G. R. Pattanaik, D. K. Pandya, S. C. Kashyap, J. Magn. Magn. Mater. 234 (2001)
294.
69. G. R. Pattanaik, D. K. Pandya, S. C. Kashyap, J. Alloy. Comp. 326 (2001) 260.
70. T. A. Tochitskii, G. A. Jones, H. J. Blythe, V. M. Fedosyuk, J. Castro, J. Magn.
Magn. Mater. 224 (2001) 221.
71. G. R. Pattanaik, D. K. Pandya, S. C. Kashyap, J. Electrochem. Soc. 149 (2002)
C363.
72. S. Kainuma, K. Takayanagi, K. Hisatake, T. Watanabe, J. Magn. Magn. Mater. 246
(2002) 207.
73. G. R. Pattanaik, D. K. Pandya, S. C. Kashyap, Thin Solid Films 433 (2003) 247.
74. R. López Anton, M. L. Fdez-Gubieda, G. Kurlyandskaya, A. García-Arribas,
J. Magn. Magn. Mater. 254–255 (2003) 85.
75. E. Gómez, A. Labarta, A. Llorente, E. Vallés, J. Electrochem. Soc. 151 (2004)
C731.
76. S. Yao, H. Wu, W. Zhang, H. Wang, J. Mater. Sci. Technol. (China) 21 (2005) 307.
118 L. Péter and I. Bakonyi
77. S. Pané, E. Gómez, E. Vallés, J. Electroanal. Chem. 596 (2006) 87.
78. C. A. Ross, Annu. Rev. Mater. Sci. 24 (1994) 159.
79. M. N. Baibich, J. M. Broto, A. Fert, F. Nguyen Van Dau, F. Petroff, P. Etienne,
G. Creuzet, A. Friederich, J. Chazelas, Phys. Rev. Lett. 61 (1988) 2472.
80. G. Binasch, P. Grünberg, F. Saurenbach, W. Zinn, Phys. Rev. B 39 (1989) 4828.
81. M. Alper, K. Attenborough, R. Hart, S. J. Lane, D. S. Lashmore, C. Younes,
W. Schwarzacher, Appl. Phys. Lett. 63 (1993) 2144.
82. L. Péter, I. Bakonyi, in Electrocrystallization in Nanotechnology, ed. G. Staikov
(Wiley-VCH, Weinheim, Germany, 2007), p. 242.
83. W. Schwarzacher, D. S. Lashmore, IEEE Trans. Magn. 32 (1996) 3133.
84. I. Bakonyi, L. Péter, in: Magnetic Materials, Processes and Devices VIII, The
Electrochemical Society PV 2004-23, Eds. S. Krongelb et al. (The Electrochemical
Society, Pennington, New Jersey, USA, 2006) p. 227.
85. I. Bakonyi, L. Péter, Progr. Mater. Sci., in preparation.
86. A. Bartók, L. Péter, I. Bakonyi, J. Pádár, P. B. Barna, unpublished work.
87. L. M. Goldman, C. A. Ross, W. Ohasi, D. Wu, F. Spaepen, Appl. Phys. Lett. 55
(1989) 2182.
88. E. Chassaing, J. Electrochem. Soc. 148 (2001) C690.
89. S. K. J. Lenczowski, C. Schönenberger, M. A. M. Gijs, W. J. M. de Jonge, J. Magn.
Magn. Mater. 148 (1995) 455.
90. L. Péter, I. Bakonyi, unpublished results.
91. L. Péter, Q. X. Liu, Zs. Kerner, I. Bakonyi, Electrochim. Acta 49 (2004) 1513.
92. S. K. Gosh, A. K. Grover, P. Chowdhury, S. K. Gupta, G. Ravikumar, D. K. Aswal,
M. Senthil Kumar, R. O. Dusane, Appl. Phys. Lett. 89 (2006) 132507.
93. M. Lakatos-Varsányi, A. Mikó, L. K. Varga, E. Kálmán, Corr. Sci. 47 (2005) 681.
94. E. Budewski, G. Staikov, W. J. Lorentz, Electrochemical Phase Formation and
Growth (VCH, Weinheim, 1996).
95. R. M. Penner, in: Electrocrystallization in Nanotechnology, ed. G. Staikov (Wiley-
VCH, Weinheim, Germany, 2007), p. 195.
96. E. C. Walter, B. J. Murray, F. Favier, G. Kaltenpoth, M. Grunze, R. M. Penner,
J Phys. Chem B 106 (2002) 11407.
97. R. M. Penner, J. Phys. Chem. B 106 (2002) 3339.
98. Y. Xiao, G. Yu, J. Yuan, J. Wang, Z. Chen, Electrochim. Acta 51 (2006) 4218.
99. K. H. Kim, J. D. Lee, J. J. Lee, B. Y. Ahn, H. S. Kim, Y. W. Shin, Thin Solid Films
483 (2005) 74.
100. H. W. Wang, H. C. Lin, Y. C. Yeh, C. H. Kuo, Appl. Phys. Lett. 88 (2006) 033111.
101. L. Y. Zhao, H. Jalili, N. Panjwani, T. Chan, Z. H. He, N. F. Heinig, K. T. Leung,
Electrochem. Solid-State Lett. 10 (2007) K47.
102. C. A. Ross, M. Hwang, M. Shima, H. I. Smith, M. Farhoud, T. A. Savas,
W. Schwarzacher, J. Parrochon, W. Escoffier, H. Neal Bertram, F. B. Humphrey,
M. Redjdal, J. Magn. Magn. Mater. 249 (2002) 200.
103. P. Apel, Radiat. Meas. 34 (2001) 559.
Electrodeposition as a Fabrication Method of Magnetic Nanostructures 119
104. M. Daub, I. Enculescu, R. Neuman, R. Spohr, J. Optoel. Adv. Mater. 7 (2005) 865.
105. J. Gong, G. Zangari, unpublished results.
106. H. Masuda, K. Fukuda, Science 268 (1995) 1466.
107. H. Asoh, S. Ono, in Electrocrystallization in Nanotechnology, ed. G. Staikov
(Wiley-VCH, Weinheim, Germany, 2007), p. 138.
108. Y. Kang, J. Zhao, J. Tao, X. Wang, Y. Li, Appl. Sur. Sci. 254 (2008) 3935.
109. S. Aravamudhan, K. Luongo, P. Poddar, H. Srikanth, S. Bhansali, Appl. Phys. A 87
(2007) 773.
110. M. V. Rastei, S. Colis, R. Meckenstock, O. Ersen, J. P. Bucher, Surf. Sci. 600
(2006) 2178.
111. K. Nielsch, R. B. Wehrspohn, J. Barthel, J. Kirschner, S. F. Fischer,
H. Kronmüuller, T. Schweinböck, D. Weiss, U. Gösele, J. Magn. Magn. Mater. 249
(2002) 234.
112. D. M. Davis, E. J. Podlaha, Electrochem. Solid-State Lett. 8 (2005) D1.
113. Q. Wang, G. Wang, X. Han, X. Wang, J. Hou, J. Phys. Chem B 109 (2005) 23326.
114. Z. Liu, L. Guo, C. L. Chien, P. C. Searson, J. Electrochem. Soc. 155 (2008) D569.
115. Z. Liu, G. Xia, F. Zhu, S. Kim, N. Markovic, C. L. Chien, P. C. Searson, J. Appl.
Phys. 103 (2008) 064313.
116. P. R. Evans, W. R. Hendren, R. Atkinson, R. J. Pollard, J. Electrochem. Soc. 154
(2007) K79.
117. Y. W. Chung, I. C. Leu, J. H. Lee, J. H. Yen, M. H. Hon, J. Electrochem. Soc. 154
(2007) E77.
118. M. E. Kiziroglou, X. Li, D. C. Gonzalez, C. H. de Groot, A. A. Zhukov, P. A. J. de
Groot, P. N. Bartlett, J. Appl. Phys. 100 (2006) 113720.
119. A. A. Zhukov, A. V. Goncharov, P. A. J. de Groot, M. A. Ghanem, P. N. Bartlett,
R. Boardman, H. Fangohr, V. Novosad, G. Karapetrov, Appl. Phys. Lett. 88 (2006)
062511.
120. A. A. Zhukov, A. V. Goncharov, P. A. J. de Groot, M. A. Ghanem, I. S. El-Hallag,
P. N. Bartlett, R. Boardman, H. Fangohr, V. Novosad, G. Karapetrov, J. Appl.
Phys. 97 (2005) 10J701.
121. A. A. Zhukov, M. A. Ghanem, A. V. Goncharov, P. A. J. de Groot, I. S. El-Hallag,
P. N. Bartlett, R. Boardman, H. Fangohr, J. Magn. Magn. Mater. 272–276 (2004)
1621.
122. Y. W. Chung, I. C. Leu, J. H. Lee, J. H. Yen, M. H. Hon, Electrochim. Acta 53,
(2007) 1703.
123. M. A. Ghanem, P. N. Bartlett, P. de Groot, A. Zhukov, Electrochem. Commun. 6,
(2004) 447.
124. L. Xu, L. D. Tung, L. Spinu, Adv. Mater. 15 (2003) 1562.
125. P. N. Bartlett, P. N. Birkin, M. A. Ghanem, P. de Groot, M. Sawicki,
J. Electrochem. Soc. 148 (2001) C119.
126. A. Hovestad, L. J. J. Janssen, J. Appl. Electrochem. 25 (1995) 519.
120 L. Péter and I. Bakonyi
127. J. L. Stojak, J. Fransaer, J. B. Talbot, in Advances in Electrochemical Science and
Engineering, Vol. 7, eds. R. C. Alkire, D. M. Kolb (Wiley-VCH, Weinheim,
Germany, 2002), p. 193.
128. C. T. J. Low, R. G. A. Wills, F. C. Walsh, Surf. Coat. Technol. 201 (2006) 371.
129. R. A. Tacken, P. Jiskoot, L. J. J. Janssen, J. Appl. Electrochem. 26 (1996) 129.
130. S. Guan, B. J. Nelson, K. Vollmers, J. Electrochem. Soc. 151 (2004) C545.
131. S. Guan, B. J. Nelson, J. Microelectromech. Syst. 15 (2006) 330.
132. R. Amigó, J. Asenjo, E. Krotenko, F. Torres, J. Tejada, Chem. Mater. 12 (2000)
573.
133. C. Pascal, J. L. Pascal, F. Favier, Chem. Mater. 11 (1999) 141.
134. F. J. Santos, L. C. Varanda, L. C. Ferracin, M. Jafelicci, Jr., J. Phys. Chem. C 112
(2008) 5301.
135. L. Cabrera, S. Gutierrez, N. Menendez, M. P. Morales, P. Herrasti, Electrochim.
Acta 53 (2008) 3436.
Chapter 6
MAGNETOELECTRIC MATERIALS FOR SPINTRONICS
Faik Mikailzade
Department of Physics, Gebze Institute of Technology
P.O. Box 141,41400 Gebze, Kocaeli, Turkey
E-mail: [email protected]
Recent developments in the fields of multiferroic and magnetoelectric
materials are reviewed. The investigations of magnetoelectric effects
in multiferroic crystals and magnetoelectric composites are considered
in a frame of their possible applications in spintronics and
magnetoelectronics.
6.1. GMR and Spintronics
Intensive investigations in the field of the nanoscale materials promote to
the great progress in various kinds of the storage media and memory
applications. The technical progress of last years in the preparation of
multilayer thin films promote to discovering the Giant Magnetoresistance
(GMR) phenomena,1,2 consisting in extraordinary changing of
resistivity/impedance of the material while applying external magnetic
field. Nobel Prize in Physics in 2007 is awarded to Albert Fert and Peter
Grünberg for their discovery of GMR. Applications of this phenomenon
have revolutionized techniques for retrieving data from hard disks. The
discovery also plays a major role in various magnetic sensors as well as
for the development of a new generation of electronics. The use of Giant
Magnetoresistance can be regarded as one of the first major applications
of nanotechnology. The GMR- materials have already found applications
as sensors of low magnetic field, computer hard disk heads,
magnetoresistive RAM chips etc. The “read” heads for magnetic hard
121
122 F. Mikailzade
disk drives (HDD) have allowed to increase the storage density on a disk
drive from 1 to 20 Gbit per square inch, merely by the incorporation of
the new GMR materials.3,4 The market only in the field of GMR-
nanotechnology is estimated over $100 billion annually. In addition to
hard disk drives, the technologies developed have also been applied to
magnetic random access memories (MRAMs).5–7 Further advance in
these fields is the key to realizing terabits/in2 hard disk drives and gigabit
nonvolatile memories within this decade.
Magnetic recording has dominated the area of peripheral information
storage ever since the beginning of the computer era a half century ago,
with tapes and disks representing the two main application areas. The
embodiment of the technology involves a relatively thin magnetic layer
supported by a flexible or rigid substrate, which can be magnetized by an
external magnetic field and which retains its magnetization after the field
is removed. Information is recorded in the form of oppositely magnetized
regions in the surface layer of the medium utilizing the fringing field
of an inductive transducer. In this frame, half-metallic dioxide films
are expected to be the key element of Magnetic Tunnel Junction
Magnetoresistive Random Access Memories (MTJ MRAM), prototypes
of which, based on conventional metallic ferromagnets, have already
been tested under laboratory conditions. Use of these materials with two
orders of magnitude discrimination between the “zero”-state and the
“one”-state current essentially simplifies the control electronics and
increases the operational speed. They are also considered as spin
aligner/filters in FET, LED, RTD devices and optical switches. Another
class of the oxide-based spintronics materials is magnetically doped
transition metal oxides (TiO2, SnO2, ZnO, etc.). Among other techniques,
laser ablation, co-sputtering and sol-gel methods have so far been
primarily used to synthesize these unique oxides doped with iron-group
metals.
Evolution beyond passive magneto-electronic components is
envisioned in the next generation of spintronics devices, which should
combine memory and logical functions in order to set new standards in
future information technology.8 Recently, there has been growing interest
in studying magnetization reversal involving spin transfer from a spin-
polarized current injected into the device as an alternative to stray
Magnetoelectric Materials for Spintronics 123
magnetic fields for switching the magnetic configurations in GMR or
TMR devices.9,10 The required huge current densities hamper the
technical realization of this attractive concept. One can use the electric
field as alternative means for controlling the magnetic configuration of
magnetoresistive systems. So, the quest for higher data density in
information storage is motivating investigations into approaches for
manipulating magnetization without using magnetic field. This is also
evidenced by the recent boom in magnetoelectronics and spintronics,11
where phenomena such as carrier effects in magnetic semiconductors12
and high-correlation effects in colossal magnetoresistive compounds13
are studied for their device potential.
6.2. History and Invention of Magnetoelectricity
The magnetoelectric effect – the induction of polarization by a magnetic
field and of magnetization by an electric field – provides another route
for linking magnetic and electric properties. Hence, one expects the
direct coupling between the magnetic and dielectric properties and their
control by the application of magnetic and/or electric fields.
Although the ME effect was prophetically predicted by Curie early in
1894 on the basis of crystal symmetry consideration14 no further work
was done until 1958 when Landau and Lifshitz proved the feasibility of
the ME effect in certain crystals. Subsequently, the symmetry argument
was applied by Dzyaloshinskii15 to antiferromagnetic Cr2O3 and it was
suggested that the ME effect could appear in Cr2O3. This was followed
by experimental confirmation.16 The ME effect was observed in
antiferromagnetic Cr2O3 in 196116,17 and later some single phase crystal
families, e.g., perovskite-type BiFeO3 BiMnO3, and TbMnO3, hexagonal
(RE)MnO3 (RE = rare earths), and the rare-earth molybdates, were found
to have the ME effect.18–22 Note that most of these compounds display an
antiferromagnetic behavior.
As it has just been mentioned, these materials can display a magneto-
electric effect in which change in magnetization is induced by an electric
field and change in electric polarization is induced by an applied
magnetic field. Hence, one expects the direct coupling between the
magnetic and dielectric properties and their control by the application of
124 F. Mikailzade
magnetic and/or electric fields. In multiferroic materials, in which
ferroelectric and ferromagnetic ordering occur simultaneously, magnetic
domains can be tuned by the application of an external electric field, and
likewise electric domains are switched by magnetic field, which is
unachievable separately in either ferroelectric or magnetic materials. So,
multiferroics, characterized by simultaneous ferroelectric and magnetic
ordering, may exhibit a larger magnetoelectricity exhibiting potential in a
wide range of applications, including the emerging field of spintronics,
data-storage media, and multiple-state memories, information storage,
sensors, transducers, actuators, storage devices etc.23
Besides application potential, the fundamental physics of multiferroic
materials is rich and fascinating. This nontrivial spin-lattice coupling in
the magnetoelectrics has been manifested through various forms, such as
linear and bilinear magnetoelectric effects, polarization change through
field-induced phase transition, magneto-dielectric effect, and dielectric
anomalies at magnetic transition temperatures.
6.3. Linear Magnetoelectric Effect
The magnetoelectric (ME) effect in its most general definition
denominates the coupling between electric and magnetic fields in matter.
A systematic progression of contributions to the ME effect is obtained
from the expansion of the free energy ( F ) of a material, i.e.24
1 1
( )
F E , H = F0 − Pi s Ei − M is H i − ε 0ε ij Eı E j − µ 0 µij H i H j
2 2
1 1
− α ij Ei H j − β ijk Ei H j H k − γ ijk H i E j E k − ... ,
2 2
where Ei and H i denote the components of electric and magnetic fields.
Differentiation leads to the polarization
∂F
( )
Pi E , H = −
∂Ei
1
= Pi s + ε 0ε ij E j + α ij H j + β ijk H j H k + γ ijk H j Ek + ...
2
Magnetoelectric Materials for Spintronics 125
and to the magnetization
∂F
( )
M i E, H = −
∂H i
1
= M is + µ 0 µij E j + α ij E j + β ijk E j H k + γ ijk E j Ek + ... ,
2
where Pi s and M is denote the spontaneous polarization and
magnetization, whereas ε and µ are the electric and magnetic
susceptibilities. The tensor α corresponds to induction of polarization
by a magnetic field or of magnetization by an electric field which is
designated as the linear ME effect. It is supplemented by higher-order
ME effects like those parametrized by the tensors β and γ . The vast
majority of research on the ME effect is devoted to the linear ME effect
and it is generally acceptable to omit the prefix ‘linear’ and simply to
refer to the linear manifestation as the ‘ME effect’.
As it was mentioned by Fiebig,24 in order to estimate the magnitude of
the coupling coefficients, it is necessary to understand the microscopic
mechanism driving ME behaviour. Hornreich and Shtrikman25 presented
a model which explains the temperature dependence of both the parallel
and the perpendicular ME effects in Cr2O3. The generalization of this
approach with a compilation of magnetic interaction mechanisms, their
modification by an electric field and the consequences for different types
of magnetic ordering are published in Ref. 26. According to these works,
the microscopic mechanism of ME behavior in Cr2O3 can be explained as
described in Fig. 6.1. As it is seen from the figure, the movement of the
Cr3+ ions in an electric field shows that the A1,2 ions move closer to the
small O2− triangle while the B1,2 ions move away from it. This breaks the
equivalence of the ferromagnetic Cr3+ sublattices, introducing a ME
magnetization. Thus, the figure shows that in differently magnetized
sublattices the sign or magnitude of the described effects may be
different. This is a very common source for emergence of a ME net
magnetization in a compensated antiferromagnet subjected to an electric
field.
126 F. Mikailzade
Fig. 6.1. Microscopic sources of ME behaviour in Cr2O3.25,26
It was shown that the ME response is limited by the relation27
α ij2 < ε ii µ jj .
So, one can conclude that the ME effect can only be large in ferroelectric
and/or ferromagnetic materials, which takes place in multiferroics.
6.4. Multiferroics
The term multiferroic was first used by H. Schmid28 in 1994. His
definition referred to multiferroics as single phase materials which
simultaneously posses two or more primary ferroic properties. Today the
term multiferroic has been expanded to include materials, which exhibit
any type of long range magnetic ordering, spontaneous electric
Magnetoelectric Materials for Spintronics 127
polarization, and/or ferroelasticity. Working under this expanded
definition the history of magnetoelectric multiferroics can be traced back
to the 1960’s.29 In the most general sense the field of multiferroics was
born from studies of magnetoelectric systems.
Multiferroics were grown for the first time in 1958, when
magnetically active 3d ions were used to substitute ions with a noble gas
shell in ferroelectrically distorted perovskite lattices.28,29 This leads us to
ferroelectric antiferromagnetic compositions like PbFe0.5Nb0.5O3 and
PbFe0.5Ta0.5O3.30 Up to now more than 80 single-phase multiferroics
were grown either as a discrete composition or as a solid solution.
Multiferroics have been the topic of various review papers with a focus
on structure and materials science,31,32 compounds,33 phase diagrams,34
symmetries29,35,36 and theory.28
Currently four major crystallographic types of multiferroics can be
distinguished as: 1) compounds with perovskite structure – BiFeO3,
BiMnO3, TbMnO3 and others;37–39 2) compounds with hexagonal
structure, which include ferroelectric antiferromagnetic rare-earth
manganites RMnO3 with R = Sc, Y, In, Ho, Er, Tm, Yb, Lu;40–48 3)
Boracite compounds with the general formula M3B7O13X (M = Cr, Mn,
Fe, Co, Cu, Ni) which are ferroelectric ferroelastic antiferromagnets;49–51
4) Orthorhombic BaMF4 compounds (M = Mg, Mn, Fe, Co, Ni, Zn).52,53
Aside from these major types a large number of multiferroics with
different structures are known. Specific examples are discussed in the
before mentioned review articles.28,31,33 A systematic classification of
symmetries, related types of ferroic ordering and compounds can be
found in Refs. 35, 36.
Generally, magnetically driven multiferroics are insulating materials,
mostly oxides, in which macroscopic electric polarization is induced by
magnetic long-range order. A necessary but not sufficient condition for
the appearance of spontaneous electric polarization is the absence of
inversion symmetry. In these materials inversion symmetry is broken by
magnetic ordering. Such a symmetry breaking often occurs in so-called
frustrated magnets, where competing interactions between spins favor
unconventional magnetic orders. The microscopic mechanisms of
magnetically induced ferroelectricity involve the polarization of
128 F. Mikailzade
electronic orbitals and relative displacement of ions in response to
magnetic ordering.
However, finding a choice for multiferroic material is very limited
because a few materials exhibit coexistence of ferroelectric and
ferromagnetic properties at room temperature. There are very few
natural multiferroic magnetoelectrics with both magnetic and electric
polarizations existing or present in nature or synthesized in the
laboratory, on the other hand, the observed ME effect for most such
materials is too weak to be applicable.54 The mechanisms and reasons for
the existence of a very few magnetoelectric systems has been discussed
in a number of works.18,24,35,55–57 According to these works, in perovskite
compounds the transition metal ion can trigger two types of Jahn–Teller
distortion, which is responsible for appearing ferroelectric polarization.
A first-order Jahn–Teller distortion occurs in the case of partially filled 3d
orbitals, which retains the centre of symmetry, e.g. by elongation of the
octahedron of ligands (as in the case of LaMnO3 and YTiO3). A second-
order Jahn–Teller distortion, which is weaker than the first order one,
requires an empty 3d shell for which a first-order Jahn–Teller distortion
cannot occur, and breaks the centrosymmetry by off-centre movement of
the transition-metal ion (examples are BaTiO3 and PZT). Since only a
partially filled 3d orbital can lead to magnetic ordering, whereas the
breaking of centrosymmetry is a necessary condition for the formation of
a spontaneous polarization, the conditions for ferroelectricity and (anti-)
ferromagnetism are mutually exclusive. In an alternative approach56 the
incompatibility was explained by Hund’s rule coupling that tends to keep
the 3d spins parallel to one another. This mechanism breaks the strong
covalent bonds that are necessary for ferroelectricity. In turn magnetic
ferroelectrics must be materials in which the double well potential
provoking the ferroelectric distortion is not caused by the hybridization
of transition-metal ions in a noble gas configuration.
So, all these reasons and mechanisms led to presence of little number
of single phase multiferroic materials and a low value of magnetoelectric
coefficient in these materials, which made them unsuitable for possible
applications.
Magnetoelectric Materials for Spintronics 129
6.5. Magnetoelectric Composites
Researchers, therefore, attempt to bring the desired effects by growing
heterostructures of ferroelectric and ferromagnetic materials.58 However,
it is a difficult task to arrive at suitable compositions, which would give
desired properties. Consequently, an additional electronic or structural
driving force must be present for magnetic and ferroelectric ordering to
occur simultaneously. A strong ME effect, however, could be realized
in the composite consisting of magnetorestrictive and piezoelectric
effects.59 A much higher ME effect has been identified in specially
designed composites in which the magnetostrictive phase is combined
with the piezoelectric one so that an efficient magnetomechanical–
piezoelectric coupling between the two phases is achieved.59 The ME
effect in composite materials of magnetostrictive and piezoelectric
phases originates from the elastic interaction between the magnetostrictive
and piezoelectric subsystems. In a magnetic field, the magnetostriction
in the magnetostrictive phase gives rise to mechanical stresses that
are transferred into the piezoelectric phase, owing to the piezoelectric
effect, resulting in an electric polarization of the piezoelectric
phase.
Van Suchtelen and co-workers59–62 made the first artificial
magnetoelectric material by combining a ferroelectric (piezoelectric)
material BaTiO3 (BTO) and a ferromagnetic (piezomagnetic) material
CoFe2O4 (CFO) in an eutectic sintered composite. Most studies in the
past focused exclusively on ferrite - PbZrTiO3 /BaTiO3 composites,60–64
although the experimental ME yield was only about 1%–2% of the
theoretical prediction.
There also have been reports of magnetoelectric effects in layered
composites such as (PbZr1-xTixO3 (PZT)–Tb0.3Dy0.7Fe1.92 (Terfenol-D),
PZT-NiFeO4, polyvinylidenefluoride–Terfenol-D and laminate
PbMgNbO–PbTiO–Terfenol-D. (PZT)–Tb0.3Dy0.7Fe1.92 (Terfenol-D),
polyvinylidenefluoride-Terfenol-D and PbMg1/3Nb2/3O3–PbTiO3–
Terfenol-D).65–70 All these composite materials have been exclusively
studied in bulk form. Interestingly, the materials made in the form of
superlattice structure yielded unusual transport properties that cannot be
obtained by classical solid-state chemistry route. Thus it is possible to
130 F. Mikailzade
construct superlattices whose structure consists of alternating FE and FM
layers, and one can investigate the coupling between the two properties.
The CFO-PZT (Pb(Zr0.52Ti0.48)O3) multilayered structures prepared by
Harshe et al.64 show a ME-voltage coefficient α E of 75 mV/cm Oe, also
much lower than the predicted value. The α E for a NFO ~nickel ferrite -
PZT multilayered structure prepared using the tape casting technique
was; 1500 mV/cm Oe.65,66 A giant ME effect was measured in a
sandwiched composite structure stacked alternatively by large
magnetostrictive material - Terfenol-D and PZT disks,67,68 and in the
Terfenol-D/PVDF laminate.69 Since then, in particular, Dong et al.70–90
have reported the giant ME effect in a number of laminate composites of
Terfenol-D and various piezoelectric materials including PZT ceramics,
Pb(Mg1/3Nb2/3O3)–PbTiO3 (PMN-PT) or Pb(Zn1/3Nb2/3O)3–TiO3 (PZN-
PT) single crystal, or electroactive PVDF copolymers.
Previous investigations of ME laminates have focused on
piezoelectric and magnetostrictive layers that were, respectively, poled
and magnetized along their thickness (transverse geometry – T-T mode)
directions. For these previous laminate designs, relatively large ME
coefficients were only observed under a high dc magnetic bias.
Relatively large ME voltage coefficients of α E = 4.8 V/cm Oe was
reported by Ryu et al.68 under dc magnetic field bias of Hdc ≥ 4000 Oe;
although later, other investigators repeated the actual value to be
α E = 1.3 V/cm Oe.91,92 Figure 6.2 shows these later experimental results
for laminate in T-T mode, where a maximum ME voltage of 66 mV/Oe
or 1.32 V/cm Oe was observed under Hdc ≈ 4000 Oe.
The main problem with the transverse (T-T) mode laminates is
that a quite high dc magnetic bias Hdc is required to obtain a maximum
value of α E . This high Hdc is caused by a large demagnetization
factor in transversely magnetized Terfenol-D layers. To reduce the
demagnetization factor effect, a long-type configuration that uses a
longitudinal magnetization was designed.70–72 This dramatic decrease in
the demagnetization factor results in a large reduction in the Hdc, required
to achieve the maximum ME coefficient.51 Long rectangular shaped
Terfenol-D/PZT/Terfenol-D and Terfenol-D/PMNPT/Terfenol-D three
layer laminates with a longitudinal magnetization and transverse
polarization (L-T mode) were then reported70–72 based on this
Magnetoelectric Materials for Spintronics 131
consideration. Experimental results confirmed that at low magnetic
biases of Hdc < 500 Oe much large values of α E could be obtained for L-T
laminates relative to T-T ones. It has been established that the induced
ME voltage for the L-T Terfenol-D/PZT laminates under Hdc = 500 Oe
was 0.085 V/Oe (or α E = 1.7 V/cm Oe and for L-T Terfenol-D/PMN-PT
ones ~0.11 VOe ( α E = 2.2 V/cm Oe); whereas that for the T-T mode of
Terfenol-D/PZT laminate under Hdc = 500 Oe was only ~0.015 V/Oe
(α E = 0.3 V/cm Oe). Clearly, long type L-T laminates have significantly
higher ME voltage coefficients than T-T ones under modest magnetic
biases.
To achieve a high output voltage, a L-L mode (longitudinal
magnetization and longitudinal polarization) of Terfenol-
D/PZT/Terfenol-D or Terfenol-D/PMN-PT/Terfenol-D laminate is a
good choice due to its large dielectric displacement along the length
(longitudinal) direction.73,88,92 For a long-type ME laminate, the length of
the piezoelectric layer is much larger than its thickness, in addition, the
longitudinal electromechanical coupling coefficient and piezoelectric
voltage constant are higher than the corresponding transverse quantities.
So, the Terfenol-D/PZT laminate operated in an L-L mode should have a
Fig. 6.2. ME voltage coefficient of Terfenol-D/PZT laminate (T-T mode) as a function of
the applied dc magnetic field bias Hdc (Dong et al.92).
132 F. Mikailzade
Fig. 6.3. Induced ME voltage in the L-L Terfenol-D/PZT/Terfenol-D laminate (Dong
et al.88).
much higher induced voltage under magnetic field excitation. For
example, Fig. 6.3 shows measurements for an L-L mode terfenol-D/PZT
laminate, which exhibits a maximum VME = 3.5 V/Oe at 1 kHz (or
α E = 2.4 V/cm Oe) at Hdc = 500 Oe.
Thus, magnetostrictive-piezoelectric laminate composites have been
found to have higher ME coefficients than that of single-phase materials
or particle composites. A very comprehensive review on magnetoelectric
composites have been presented recently in Ref. 93.
6.6. Conclusions and Outlook
Hence, in normal practice, desired magnetoelectric effect is achieved by
growing heterostructures of ferroelectric and magnetic materials.
Realization of heterostructures with desired properties is not only
difficult but also involves complicated lengthy procedures. Recently,
attention to ME materials has been gradually drawn toward composite
thin films. Compared to bulk composites, ME composite thin films
exhibit unique advantages. Their composition and connectivity could be
Magnetoelectric Materials for Spintronics 133
modulated at the microscopic scale, and the artificial thin film
heterostructures can thus be achieved, which have potential applications
in all kinds of microdevices and integrated units such as microsensors,
MEMS devices, and high density information storage devices. It
has been identified that a number of magnetostrictive-piezoelectric
composite structures show significant ME output qualified for potential
applications.
Recently, efforts have been made to fabricate the artificial layers and
tailor their structures for the suitability of multiferroics.94–98 These
materials are in the form of either composites, superlattices, or
multilayers. Interestingly, the material made in the form of superlattices,
whose structure consists of alternating ferroelectric and ferromagnetic
layers, yielded unusual electrical and magnetic transport properties that
cannot be obtained in either of their constituents. As a result of these
studies of sandwich devices, practical applications of magnetoelectric
effects now appear more feasible. Nevertheless, sandwich devices are
intrinsically limited in the feature size and difficult to miniaturize.
Therefore, further advances in new magnetoelectric composites are still
desirable.
Acknowledgments
The author is indebted to Research Projects Commission of Gebze
Institute of Technology for supporting this work by the Grant No. 2007-
A-14 and to The Scientific & Technological Research Council of Turkey
(TÜBĐTAK) for supporting by Project. No. 106M540.
References
1. M. N. Baibich, J. M. Broto, A. Fert, F. Nguyen van Dau, F. Petroff, P. Etienne,
G. Creuzet, A. Friederich, and J. Chazelas, Phys.Rev. Lett. 61, 2472 (1988).
2. G. Binasch, P. Grünberg, F. Saurenbach, and W. Zinn, Phys. Rev. B 39 4828 (1989).
3. G. A. Prinz, Sci. 282, 1660 (1998).
4. P. Grünberg, Acta Mater. 48, 239 (2000).
5. J. M. Dauton, Thin Solid Films 216, 162 (1992); J. M. Dauton, J. Appl. Phys. 81,
3758 (1997).
134 F. Mikailzade
6. S. S. P. Parkin, K. P. Roche, M. G. Samant, P. M. Rice, R. B. Beyers, R. E.
Scheurlein, E. J. O’Sullivan, S. L. Brown, J. Bucchigana, D. W. Abraham, Y. Lu,
M. Rooks, P. L. Trouilloud, R. A. Wanner, and W. J. Gallagher, J. Appl. Phys. 85,
5828 (1999).
7. S. Tehrani, E. Chen, M. Durlam, M. Deherrera, J. M. Slaughter, J. Shi, and
G. Kerszkowski, J. Appl. Phys. 85, 5822 (1999).
8. A. Ney, C. Pampuch, R. Koch, and K. H. Ploog, Nature Londond 425, 485 (2003).
9. J. Slonczewski, J. Magn. Magn. Mater. 159, L1 (1996); 195, 261 (1999).
10. L. Berger, Phys. Rev. B 54, 9353 (1996).
11. G. A. Prinz, J. Magn. Magn. Mater. 200, 57–68 (1999).
12. H. Ohno et al., Nature 408, 944–946 (2000).
13. A. Asamitsu, Y. Tomioka, H. Kuwahara, and Y. Tokura, Nature 388, 50–52 (1997).
14. P. Curie, J. Physique 3, 393 (1894).
15. I. E. Dzyaloshinskii, Zh. Eksp. Teor. Fiz. 37, 881 (1959), Sov. Phys. JETP 10, 628
(1960).
16. D. N. Astrov, Sov. Phys. JETP 13, 729 (1961).
17. G. T. Rado and V. J. Folen, Phys. Rev. Lett. 7, 310 (1961).
18. N. A. Hill, J. Phys. Chem. B 104, 6694 (2000).
19. N. Hur, S. Park, P. A. Sharma, J. S. Ahn, S. Guha, and S.-W. Cheong, Nature 429,
392 (2004).
20. T. Kimura, S. Kawamoto, I. Yamada, M. Azuma, M. Takano, and Y. Tokura,
Phys. Rev. B 67, 180401R (2003); T. Kimura, T. Goto, K. Thizaka, T. Arima, and
Y. Tokura, Nature 426, 55 (2003).
21. H. Zheng, J. Wang, S. E. Lofland, Z. Ma, L. M. Ardabili, T. Zhao, L. S. Riba, S. R.
Shinde, S. B. Ogale, F. Bai, D. Viehland, Y. Jia, D. G. Schlom, M. Wutting,
A. Roytburd, and R. Ramesh, Sci. 303, 661 (2004).
22. P. Murugavel, D. Saurel, W. Prellier, Ch. Simon, and B. Raveau, Appl. Phys. Lett.
85, 4424 (2004).
23. G. R. Slemon, Magnetoelectric Devices Transducers, Transformers, and Machines,
Wiley, New York, 1966.
24. M. Fiebig, J. Phys. D: Appl. Phys. 38, R123 (2005).
25. R. M. Hornreich and S. Shtrikman, Phys. Rev. 161, 506 (1967).
26. G. A. Gehring, Ferroelectrics 161, 275 (1994).
27. T. H. O’Dell, Phil. Mag. 8, 411 (1963).
28. H. Schmid, Ferroelectrics 162, 317 (1994).
29. G. A. Smolenskii and I. E. Chupis, Sov. Phys.—Usp. 25, 475 (1982).
30. G. A. Smolenskii, A. I. Agranovskaya, and V. A. Isupov, Sov. Phys. Solid State 1,
149, (1959).
31. Y. N. Venevtsev, V. V. Gagulin, and I. D. Zhitomirsky, Ferroelectrics 73, 221
(1987).
32. G. A. Smolenskii, V. A. Bokov, V. A. Isupov, N. N. Krainik, and G. H. Nedlin,
Helv. Phys. Acta 41, 1187 (1968).
Magnetoelectric Materials for Spintronics 135
33. Y. N. Venevtsev and V. V. Gagulin, Ferroelectrics 162, 23 (1994).
34. S. M. Skinner, IEEE Trans. Parts, Mater. Packaging 6, 68 (1970).
35. H. Schmid, Int. J. Magn. 4, 337 (1973).
36. K. Aizu, Phys. Rev. B 2, 754 (1970).
37. S. M. Skinner, IEEE Trans. Parts, Mater. Packaging 6, 68 (1970).
38. F. Kubel and H. Schmid, Acta Crystallogr. B 46, 698 (1990).
39. S. V. Kizelev, R. P. Ozerov, and G. S. Zhdanov, Sov. Phys.—Dokl. 145, 1255
(1962).
40. A. V. Kovalev and G. T. Andreeva, C. R. Acad. Sci. 256, 1958 (1963).
41. K. H. Hellwege and A. M. Hellwege (ed), Numerical Data and Functional
Relationships (Landolt-Bornstein, New Series) Group III, vol 16a (Berlin, Springer,
1981).
42. B. B. van Aken, Structural Response to Electronic Transitions in Hexagonal and
Ortho-Manganites PhD Thesis University of Groningen, 2001.
43. E. F. Bertaut and M. Mercier, Phys. Lett. 5, 27 (1963).
44. H. Sugie, N. Iwata, and K. Kohn, J. Phys. Soc. Japan 71, 1558 (2002).
45. C. Moure, M. Villegas, J. F. Fernandez, J. Tartaj, and P. Duran, J. Mater. Sci. 34,
2565 (1999).
46. B. B. van Aken, J. W. G. Bos, R. A. de Groot, and T. T. M. Palstra, Phys. Rev. B 63,
125127 (2001).
47. M. Bieringer, J. E. Greedan, and A. S. Wills, Appl. Phys. A 74, S601 (2002).
48. N. Fujimura, H. Sakata, D. Ito, T. Yoshimura, T. Yokota, and T. Ito, J. Appl. Phys.
93, 6990 (2003).
49. Y. Y. Tomashpol’ski, Y. N. Venevtsev, and V. N. Beznozdrev, Fiz. Tverd. Tela 7,
2763 (1965).
50. E. Ascher, H. Schmid, and D. Tar, Solid State Commun. 2, 45 (1964).
51. H. Schmid, H. Rieder, and E. Ascher, Solid State Commun. 3, 327 (1965).
52. J. F. Scott, Ferroelectrics 24, 127 (1980).
53. M. Eibschütz and H. J. Guggenheim, Solid State Commun. 6, 737 (1968).
54. J. Wang, J. B. Neaton, H. Zheng, V. Nagarajan, S. B. Ogale, B. Liu, D. Viehland,
V. Vaithyanathan, D. G. Schlom, U. V. Waghmare, N. A. Spaldin, K. M. Rade,
M. Wuttig, and R. Ramesh, Sci. 299, 1719 (2003).
55. B. B. van Aken, T. T. M. Palstra, A. Filippetti, and N. A. Spaldin, Nat. Mater. 3, 164
(2004).
56. D. I. Khomskii, Bull. Am. Phys. Soc. C 21.002 (2001).
57. N. A. Hill and A. Filipetti, J. Magn. Magn. Mater. 242–245, 976 (2002).
58. K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett. 75, 555 (1999).
59. J. van Suchtelen, Philips Res. Rep. 27, 28 (1972).
60. J. van den Boomgaard, D. R. Terrell, and R. A. J. Born, J. Mater. Sci. 9, 1705
(1974).
61. J. van den Boomgaard, A. M. J. G. van Run, and J. van Suchtelen, Ferroelectrics 14,
727 (1976).
136 F. Mikailzade
62. J. van den Boomgaard, A. M. J. G. van Run, and J. van Suchtelen, Ferroelectrics 10,
295 (1976).
63. J. van den Boomgaard and R. A. J. Born, J. Mater. Sci. 13, 1538 (1978).
64. G. Harshe, J. P. Dougherty, and R. E. Newnham, Int. J. Appl. Electro-magn. Mater.
4, 161 (1993).
65. G. Srinivasan, E. T. Rasmussen, J. Gallegos, R. Srinivasan, Yu. I. Bokhan, and
V. M. Laletin, Phys. Rev. B 64, 214408 (2001).
66. G. Srinivasan, E. T. Rasmussen, B. J. Levin, and R. Hayer, Phys. Rev. B 65, 134402
(2002).
67. J. Ryu, S. Priya, A. V. Carazo, and K. Uchino, J. Am. Ceram. Soc. 84, 2905 (2001).
68. J. Ryu, A. V. Carazo, K. Uchino, and H. E. Kim, Jpn. J. Appl. Phys. Part 1, 40, 4948
(2001).
69. K. Mori and M. Wuttig, Appl. Phys. Lett. 81, 100 (2002).
70. S. X. Dong, J. F. Li, and D. Viehland, Appl. Phys. Lett. 83, 2265 (2003).
71. S. X. Dong, J. F. Li, and D. Viehland, IEEE Trans. Ultrason. Ferroelectr. Freq.
Control 50, 1236 (2003).
72. S. X. Dong, J. F. Li, and D. Viehland, J. Appl. Phys. 95, 2625 (2004).
73. S. X. Dong, J. F. Li, and D. Viehland, Appl. Phys. Lett. 85, 2307 (2004).
74. S. X. Dong, J. F. Li, and D. Viehland, J. Appl. Phys. 96, 3382 (2004).
75. S. X. Dong, J. F. Li, and D. Viehland, Appl. Phys. Lett. 85, 2307 (2004).
76. S. X. Dong, J. F. Li, and D. Viehland, Appl. Phys. Lett. 85, 3534 (2004).
77. S. X. Dong, J. F. Li, and D. Viehland, Appl. Phys. Lett. 84, 4188 (2004).
78. S. X. Dong, J. Y. Zhai, Z. P. Xing, J. F. Li, and D. Viehland, Appl. Phys. Lett. 86,
102901 (2005).
79. S. X. Dong, J. G. Bai, J. Y. Zhai, J. F. Li, G. Q. Lu, D. Viehland, S. J. Zhang, and
T. R. Shrout, Appl. Phys. Lett. 86, 182506 (2005).
80. S. X. Dong, J. Y. Zhai, F. M. Bai, J. F. Li, D. Viehland, and T. A. Lograsso, J. Appl.
Phys. 97, 103902 (2005).
81. S. X. Dong, J. Y. Zhai, N. G. Wang, F. M. Bai, J. F. Li, D. Viehland, and T. A.
Lograsso, Appl. Phys. Lett. 87, 222504 (2005).
82. S. X. Dong, J. Zhai, J. F. Li, and D. Viehland, Appl. Phys. Lett. 88, 082907 (2006).
83. S. X. Dong, J. Y. Zhai, J. F. Li, and D. Viehland, Appl. Phys. Lett. 89, 252904
(2006).
84. S. X. Dong, J. Y. Zhai, J. F. Li, and D. Viehland, Appl. Phys. Lett. 89, 122903
(2006).
85. S. X. Dong, J. F. Li, and D. Viehland, J. Appl. Phys. 100, 124108 (2006).
86. S. X. Dong, J. Y. Zhai, J. F. Li, D. Viehland, and M. I. Bichurin, Appl. Phys. Lett.
89, 243512 (2006).
87. J. Y. Zhai, J. F. Li, S. X. Dong, D. Viehland, and M. I. Bichurin, J. Appl. Phys. 100,
124509 (2006).
88. S. X. Dong, J. F. Li, and D. Viehland, IEEE Trans. Ultrason. Ferroelectr. Freq.
Control 50, 1253 (2003).
Magnetoelectric Materials for Spintronics 137
89. S. X. Dong, J. F. Li, and D. Viehland, J. Mater. Sci. 41, 97 (2006).
90. S. X. Dong, J. R. Cheng, J. F. Li, and D. Viehland, Appl. Phys. Lett. 83, 4812 (2003).
91. J. Y. Zhai, Z. Xing, S. X. Dong, J. F. Li, and D. Viehland, Appl. Phys. Lett. 88,
062510 (2006).
92. S. X. Dong, J. F. Li, and D. Viehland, IEEE Trans. Ultrason. Ferroelectr. Freq.
Control 51, 794 (2004).
93. C.-W. Nan, M. I. Bichurin, S. Dong, D. Viehland, and G. Srinivasan, J. Appl. Phys.
103, 031101 (2008).
94. K. S. Wang et al., Appl. Phys. Lett. 84, 3091 (2004).
95. H. Zheng et al., Sci. 303, 661 (2004).
96. C.-W. Nan, N. Cai, L. Liu, J. Zhai, Y. Ye, and Y. Lin, J. Appl. Phys. 94, 5930
(2003).
97. G. Liu, C.-W. Nan, N. Cai, and Y. Lin, ibid. 95, 2660 (2004).
98. P. Murugavel, D. Saurel, W. Prellier, Ch. Simon, and B. Raveau, Appl. Phys. Lett.
85, 4424 (2004).
This page is intentionally left blank
Chapter 7
GMR IN ELECTRODEPOSITED SUPERLATTICES
Gholamreza Nabiyouni
Department of Physics, University of Arak
Shahid Beheshti Avenue, P.O. Box 38156, Arak, Iran
E-mail: [email protected]
Following the discovery of GMR in 1988, and since the very first
observation of GMR in electrodeposited superlattices, there has been an
extensive interest in electrodeposition of magnetic thin films and
multilayers. Among the different multilayer systems, electrodeposition
of magnetic/nonmagnetic superlattices attracted an especial interest
because of their magnetoresistive properties and their applications in
the modern data storage and information technology. After giving a
brief introduction we introduce electrodeposition as a versatile
technique to fabricate GMR materials. The chapter then deals with
origin of magnetoresistance, and ends with dealing the parameters
involved in the GMR in electrodeposited superlattices.
7.1. Introduction
Technological applicability of metallic nano-structures has generated
tremendous interest in recent years because of their wide range of
applications, especially in computer and microelectronic industries.
Progress in the growth of high quality thin films, be they metals,
semiconductors or insulators, has allowed the realisation of many
modern electronic, magnetic and optical devices. The study of magnetic
films, consisting of alternating magnetic and non-magnetic layers is of
interest as they exhibit new magnetic phenomena with the potential for
application in data storage technology, and also for obtaining a better
understanding of the magnetic properties of materials. The properties of
multilayers strongly depend on the layer thicknesses and effects of the
boundaries between magnetic and non-magnetic layers. Multilayer thin
139
140 G. Nabiyouni
films usually consist of layers with thicknesses from a few angstroms
up to ~100Å. The mechanical, transport and magnetic properties of
multilayer thin films are often radically different from the bulk properties
of their constituents. With the introduction of magnetoresistance (MR)
heads, the data reading function of the head is now performed by a MR
sensor. Thus, with an increase in performance parameters, a demand for
materials with improved properties, both magnetic properties and
corrosion resistance, is desired for the recording head.
Co-rich FeCoNi thin film alloys are good candidates as soft magnetic
materials.1,2 Layering the Co-rich alloy with copper can increase the
alloy resistivity, and avoiding eddy currents and make them alternative
for the traditional permalloy NiFe write materials.
Much attention has been paid to metallic superlattices consisting of
alternating layers of a ferromagnetic metal e.g. Ni, Co, Fe or their alloys
and a non-magnetic transition or noble metal e.g. Cu, Ag, Pt, Au and
Ru since they can exhibit special properties such as the magneto-optic
Kerr effect (MOKE), anisotropic magnetoresistance (AMR) or giant
magnetoresistance (GMR).3–14
A range of methods are available for the fabrication of thin films.
Multilayer and superlattice structures are most commonly prepared by
vacuum based techniques such as evaporation, sputtering and molecular
beam epitaxy (MBE). In these techniques a metal is evaporated or
sputtered and deposited on a substrate to form a thin film. In MBE under
certain conditions, it is possible to grow one atomic layer after another,
forming a nearly perfect single crystal. However the required equipment
is complicated and expensive. Sputtering which is less complicated and
less expensive has been widely used to produce thin films and
superlattices although the produced materials are not perfect single
crystals and do not always have very sharp interfaces.
Electrodeposition, or electroplating, as an alternative deposition
technique to grow thin films has been little used compared to vacuum
deposition till last two decades. This may be because electrodeposition
was not expected to produce pure and high quality thin films.
Nevertheless Brenner15 reported that in the right conditions, it was
possible to electrodeposit material that had impurity levels as low as 100
GMR in Electrodeposited Superlattices 141
ppm. Electrodeposition has long been used for depositing decorative and
protective coatings and is now hugely popular as it is the method of
choice for metal deposition in the fabrication of integrated circuits16 and
magnetic recording heads for computer hard discs.2
Electrodeposition has some advantages over vacuum deposition
techniques including a high deposition rate, simplicity in the
experimental set-up, deposition at nearly room temperature and low cost.
It is also a very useful technique to fabricate metal deposits into recessed
and curved areas.
Moreover the properties of the electrodeposited thin films can be
easily controlled by changing the deposition parameters such as applying
the voltage and current density, electrolyte concentration and pH,
and temperature. Electrodeposited alloys are used widely in the
microelectronics field, due to the cost efficiency, and the applicability to
irregular geometries. It is also an advantageous method compare to other
vapour technique when depositing alloys of elements with extremely
different vapour pressures.
Electrodeposition, usually refers to deposition of a metal or an alloy
from an electrolyte by passing a charge between two electrodes located
in the electrolyte. It dates back to Volta’s discovery of the production of
electricity by chemical reactions. It is a method by which a relatively
pure metal or alloy may be deposited from a solution. Electrodeposition
is based on Faraday’s law i.e. the amount of chemical reaction caused
by the flow of current is proportional to the amount of electricity
passed.
The first electrodeposition of alloys, brass and bronze, probably took
place at the same time. However up to 1900 the application of alloy
plating was very limited and quite empirical. The first systematic work
on alloy deposition was that of Fritz Spitzer17 published in 1905, which
discussed the electrodeposition of brass. To date the electrodeposition of
many metals and alloys has been reported, and the conditions for their
deposition may vary considerably. Modern electrodeposition in the sense
of the deposition of thin and dense metal films for the purposes of
decoration and protection took place after the discovery of the value of
cyanides for plating baths.
142 G. Nabiyouni
Artificially layered materials consisting of alternating layers of two
different metals or alloys are widely used to study their mechanical,
optical and magnetic properties which might be different from those of
bulk materials. Electrodeposition was possibly the earliest deposition
technique for metal multilayers, as Blum deposited Cu/Ni and other
multilayer systems by alternate deposition from two different electrolytes
as early as 1921.18 Electrodeposition of multilayers was then developed
and reported on by a number of groups in different metal (alloy)/metal
systems.13,19–22 A few years after the discovery of GMR in MBE grown
Fe/Cr superlattices,23 GMR was observed in electrodeposited Co-Ni-
Cu/Cu superlattices.24
We first deal with the electrodeposition of metallic thin films and
superlattices, which has not been studied as widely as vacuum-based
deposition techniques, despite its capability for making superlattices or
precision microstructures. Some fundamental concepts of electrochemistry
are described in this section before dealing with the electrodeposition
of metals and superlattices. The electrodeposition of nanostructure
superlattice thin films and study their magnetic properties will then be
described. We then focus of the magnetic nanostructures with GMR
properties.
7.2. Electrodeposition
Electrodeposition usually takes place in a single or double cell containing
a solution (electrolyte) and two or three electrodes. The electrode is a
conducting material, usually a metal or a semiconductor, in which charge
is carried by electron movement and it may be a solid or a liquid. The
electrolyte is a conducting medium, usually a liquid or a fused salt
solution, in which charge is carried by the movement of ions.
An electrode reaction is a chemical process involving the transfer of
one or more electrons to or from the electrode surface, for example
M (aq) n + + ne −
→ M (s )
(1)
→ M n + + ne −
M
GMR in Electrodeposited Superlattices 143
where M indicates a metal atom. The former reaction occurs at the
cathode while the latter occurs at the anode.
A cathode is an electrode at which reduction occurs and positive ions
are discharged or negative ions formed. An anode is an electrode at
which oxidation occurs. Current which consists of electrons flowing
from an electrode to the electrolyte solution is called a reduction or
cathodic current. When there is no current flowing through the cell, the
cathode potential eventually reaches a value which is known as the
equilibrium potential (Veq), given by the Nernst equation:
R'T C O
Veq = Vs + ln (2)
nF C R
where Vs is the standard potential of O and R ions, (O reduced to R in the
solution), R' is the ideal gas constant (J K-1 mol-1), T is the absolute
temperature, F is Faraday’s constant (96485 coulomb/mol) and CO and
CR are the concentrations of the species O and R respectively.
The standard electrode potential is defined as the potential of the
metal in equilibrium with a one molar solution of its ions at 25°C,
measured relative to the standard hydrogen electrode (SHE). From a
practical point of view the SHE is not convenient; so in routine
experiments, an alternative reference electrode is used.
Often it is necessary to apply an external potential (Vext) to drive a
reaction. Consider a reaction at the cathode:
Vext = Veq + η + iR . (3)
In this equation Veq is the equilibrium potential of the cathode and i is
the current flowing between cathode and anode in the solution, so that iR
is the potential drop between the two electrodes due to the solution
resistance R. η is the overpotential which is defined as the deviation of
the potential from its equilibrium value.
η = V - Veq . (4)
Both iR and η vary with the current but in different ways. In order to
achieve a situation in which the variation of i with V is characteristic of
144 G. Nabiyouni
the processes at only one electrode (known as the working electrode), a
reference electrode (RE), should be used in addition to an electrode
used to complete the circuit (known as the secondary electrode). The
reference electrode should be placed in the cell quite close to the working
electrode (WE). A potentiostat applies a potential difference between the
WE and the secondary electrode (SE) such that the potential difference
between the WE and RE is kept constant at a specific voltage. This can
be done using a DC power supply and a feedback circuit. Since no
current is passed through the RE, its potential is constant and it is not
polarised during the experiment.
The more positive the standard electrode potential of a metal, the
more easily it is reduced and the more noble the metal. For example
Cu2+ has a more positive standard electrode potential than Co2+ or Ni2+ so
it has a greater tendency to absorb two electrons, be reduced and deposit
on the cathode.
For electrodeposition the electrolyte is a conducting solution
containing ions of the metals which are to be deposited on the working
electrode. It can also contain other ions to decrease the resistance of the
solution between the working and secondary electrodes and thereby
decrease the potential drop iR, reduce unwanted Joule heating and make
the potential distribution between working and secondary electrodes
more uniform.
7.3. Electrodeposition of Metals and Alloys
In order to understand the multilayer electrodeposition process it is easier
to start by describing the deposition of a single metal from a solution
under potentiostatic control. Figure 7.1 shows a cell containing an
electrolyte and three electrodes connected to a potentiostat.
The potentiostat applies a potential difference between the WE and
the RE. Since the RE is at fixed potential relative to the solution, any
change in the potential applied to the working electrode changes the WE
potential relative to the solution by the same amount.
The working electrode is the substrate (a metal or a semiconductor)
on which the desired material is subsequently deposited. The choice of
GMR in Electrodeposited Superlattices 145
Fig. 7.1. Schematic of a simple electrodeposition cell.
electrode material is determined by the nature of the material deposited.
In order to obtain good electrodeposited films, it is essential that the WE
is polished (mechanically, chemically or electrochemically) and cleaned
to avoid absorbed impurities.
The secondary electrode supplies the current required by the WE. It
can be any nonpolarisable metal or other conductor that has no chemical
reaction with the electrolyte. It is located far from the WE to gives as
uniform a current distribution as possible across the cell in order to give
uniform deposition.
The metal ions are reduced to metal atoms and deposit on the WE.
Noble metals are reduced at less negative electrode potentials than other
metals. Metal atoms in the working electrode may be oxidised and can
dissolve in the solution.
The mass of metal (m) deposited on or dissolved from the WE can be
calculated using Faraday’s law:
qM
m= (5)
nF
where q is the charge passed between WE and SE, F is Faraday’s
constant and n and M are the valence and atomic mass of the metal
respectively.
146 G. Nabiyouni
It is quite possible to electrodeposit more than one metal at a time in
the form of an alloy by selecting an appropriate electrolyte containing the
different metal ions and applying a suitable potential. The concentrations
of the metals in the electrodeposited alloy can be controlled by selecting
different ion concentrations in the solution.
The quality of the electrodeposited material is affected by factors
including the pH, temperature, current density and external voltage
applied between the electrodes.
7.4. Electrodeposition of Multilayers and Superlattices
One approach in developing materials with new physical properties is to
artificially structure the materials; and an easy way to do this is by
layering them and making a sandwich or multilayer structure rather than
the homogenous alloys.
Multilayer thin films consist of successive layers of two or more
different materials deposited on a substrate. The thickness of each layer
is typically between a few angstrom and hundreds of angstrom and the
multilayer may contain up to several hundred layers. A superlattice is a
periodic multilayer. Superlattices have been widely studied because their
properties may be different from bulk materials.
MBE, sputtering and evaporation are currently the main deposition
techniques for making superlattices. Electrodeposition also offers
advantages however, including cheapness, a high deposition rate and the
simplicity of the apparatus. Electrodeposition was used as a method for
fabricating multilayers with thin layers by Brenner in 1939 who grew
Cu/Bi multilayers with a bilayer thickness of about 100 nm from a single
bath by varying the current between two suitable values.13 So far most
of the research on electrodeposited multilayers has concentrated on
metallic multilayers in which magnetic metal layers are sandwiched
between two non-magnetic layers, because they exhibit interesting
magnetic properties as well as interesting mechanical properties.
Tench and White18 electrodeposited Ni/Cu superlattices from a
single bath, with a period of tens of nanometers which was one order
of magnitude less than Brenner’s films. Lashmore20 grew Ni/Cu
GMR in Electrodeposited Superlattices 147
superlattices from a single bath on (100), (111) and (110) oriented Cu
single crystal substrates, and they were highly textured. They also had
relatively sharp interfaces, and magnetic measurements confirmed that
the quality of these electrodeposited superlattices compared well with
films grown by vacuum deposition methods. However there has also
been a significant amount of work on other metallic and ceramic
multilayers. For example Cohen et al.19 electrodeposited Ag/Pd
multilayers and Switzer et al.25 electrodeposited Pb-Ti oxide ceramic
superlattices.
Electrodeposition of superlattices can be divided in two major
categories, using either the dual bath technique or the single bath
technique. Each technique has its own advantages and suffers from some
limitations.
7.4.1. Dual bath electrodeposition
In dual bath electrodeposition the substrate is transferred between two
different baths containing different ions in solution. The idea is to deposit
only one metal, either a single element or a particular alloy composition
from each electrolyte.
Clearly the substrate should be cleaned before the deposition of each
layer in order to avoid contamination of one electrolyte by the other.
This can be done by rinsing the substrate in distilled water. However it
is hard to keep the substrate absolutely clean and to protect it from
undesirable surface reactions occurring during the substrate transfer from
one bath to the other. For instance, the thickness of any oxide layers
formed during transfer could be significant if the multilayer consists of
very thin layers.
Dual bath electrodeposition also requires relatively complex
mechanical apparatus which could limit its application. Apart from these
restrictions, the dual bath deposition method may be used more in the
future because of its flexibility and ability to deposit two different pure
metals.
148 G. Nabiyouni
7.4.2. Single bath electrodeposition
In the single bath deposition technique, a superlattice structure is formed
by periodically varying either the deposition voltage (potentiostatic
control) or deposition current (galvanostatic control) between two
suitable values. All of the metal ions to be deposited are present in the
single electrolyte used. At a less negative potential the more noble metal
present in the solution (say metal A) is deposited on the working
electrode, forming layer A, while at a more negative potential (depending
on the equilibrium potentials of the metal ions present in the solution) the
other metal (say metal B) is deposited and layer B formed. Thus it is
possible to electrodeposit metallic superlattices by switching the
potential between two suitable values. It should be noted that the more
noble metal will be deposited as well as the less noble metal at the more
negative potential, so that the real composition of layer B is AxB1-x
instead of B and the superlattice structure will consisting alternative layer
of A, AxB1-x, A, AxB1-x …. Although it is not possible to deposit a layer
of pure metal B, the concentration of metal A in the AxB1-x layer
may be kept low if the concentration of metal A in the solution is
sufficiently low. Figure 7.2 show a schematic diagram for a superlattice
structure.
The single bath technique has been widely used to electrodeposit
superlattices containing different metals, but most work has been done
on the Ni-Cu/Cu system.20,22 This systems used for mechanical, electrical
and magnetic studies, since Ni and Cu, one a ferromagnet and the other
non-magnetic, have the same crystal structure (fcc) and very similar
lattice parameters. Electrodeposition of Ni-Cu/Cu multilayers is
straightforward, Ni often becoming passive in the electrolyte so that it
does not dissolve during the Cu deposition. Electrodeposited Ni-Cu/Cu
multilayers have very well-defined layer structures. Nevertheless, this
system is not expected to have a GMR as large as that observed in the
Co/Cu systems.
Other superlattice systems prepared using single bath method include
Co-Cu/Cu and Co-Ni-Cu/Cu. Figure 7.3 shows the high angle X-ray
diffraction pattern of an electrodeposited Ni-Cu/Cu superlattice.26
GMR in Electrodeposited Superlattices 149
Fig. 7.2. A schematic diagram for a superlattice structure.
Fig. 7.3. High angle X-ray diffraction pattern measured using Cu Kα radiation for a
100x(Cu0.19Ni0.81 6nm/Cu0.79Ni0.21 2nm) electrodeposited multilayer. The position of the
central peak labelled ‘0’ is determined by the average lattice spacing in the <111>
direction of the film, while the positions of the satellite peaks labelled –3 (third order) to
+2 (second order) are determined by the period of the multilayer.26 [Reproduced by kind
permision of I. Kazeminezhad].
The satellite peaks surrounding the central <111> peak correspond to
the period of the composition modulation. A low angle X-ray diffraction
pattern (using synchrotron radiation and choosing a wavelength of
150 G. Nabiyouni
1.610Å) of an electrodeposited Co-Ni-Cu/Cu superlattice grown on
Cu(100) single crystal27 is shown in Fig. 7.4. Observation of up to two
orders of superlattice peaks indicates the existence of well defined
superlattice structure.
Dissolution of the less noble metal during the deposition of the more
noble metal is a problem which can restrict the applicability of the single
bath method and make it difficult to control the thickness of each layer.
This can be minimised by choice of solution and deposition waveform.
One major reason for growing the magnetic/nonmagnetic superlattices
is to observe the GMR effect. While GMR in Ni/Cu or Ni-Cu/Cu
superlattices is small (only a few percent, and hardly reaches to 5%), in
Co/Cu or Co-Cu/Cu superlattices, the effect could be considerably larger.
By adding Co ions to the Ni-Cu electrolyte and electrodepositing of Ni-
Co-Cu/Cu superlattices, a GMR magnitude of 25% at room temperature
was achieved by Nabiyouni et al.,28 the result is shown in Fig. 7.5.
Fig. 7.4. Low angle X-ray diffraction from an electrodeposited super lattice.28
GMR in Electrodeposited Superlattices 151
Fig. 7.5. Transverse MR for a superlattice consisting of 100 repeats of nominally Co-Ni-
Cu(30Å)/Cu(7Å) grown on a polycrystalline Cu substrate.28
The problem with growing pure Co/Cu superlattices by
electrodeposition is that the deposition of Co is a more reversible
reaction than that of Ni. The consequence of this is that at the potential
for the deposition of the Cu layer the Co in the Co-Cu layer will dissolve
back into the electrolyte giving a rough and uneven interface between
layers. By depositing the Co as a Ni-Co-Cu alloy this dissolution reduces
and has less effect on the quality of the layers. This effect of one metal
depositing preferentially to another even though they have similar
standard potentials is known as anomalous co-deposition and is thought
to be related to an increase in hydroxide concentration at the electrode.
7.4.3. Electrodeposition of metallic thin films onto semiconductor
substrates
A problem that may have limited the utilisation of electrodeposition in
modern devices is the requirement that the substrate has to be conducting
which in most cases means metallic. Metals, are not, however the
generally preferred substrate for thin film growth due to their physical
and electrical properties. Physically they are soft which makes it very
152 G. Nabiyouni
difficult to prepare a high quality defect free flat surface for deposition,
while if the electronic transport properties of the film are of interest these
will be short-circuited by the low resistance of the substrate.
A semiconducting substrate has the advantage of allowing resistivity
measurements to be made without having to remove the film, as the
semiconductor has a much higher resistivity than the metal as well as
being partly isolated from the film by the Schottky barrier formed at the
semiconductor/metal interface. Semiconductors can also be prepared
with flatter surfaces than metals and results achieved would be directly
comparable to those for vacuum techniques which also tend to employ
semiconducting substrates.
The choice for choosing a semiconductor as the substrate is usually
between GaAs and Si, both of which can be readily obtained as high
quality polished wafers as they are commonly used in the electronics
industry. Si has the advantage of lower cost but the problem of removing
the insulating SiO2 layer from the surface before deposition made it
unattractive. GaAs also has native oxides but these are easier to remove
and it had been used in the majority of previous studies of metal
electrodeposition on semiconductors. It is also a commonly used
substrate in vacuum deposition of multilayers.
The requirement that the substrate must conduct does not preclude
intrinsic semiconductors. By using doped semiconductors it is possible
to pass enough current through the bulk of the material to allow
electrodeposition, although the mechanisms for charge transfer at the
solid/electrolyte interface are different between semiconductors and
metals. This is especially the case for p-type material where illumination
is often required to generate enough minority carrier electrons at the
surface to reduce the ions in solution. A detailed discussion of the
consequences of using a semiconductor instead of a metal as a substrate
for electrodeposition is given by R. Hart.29
The growth of epitaxial single crystal thin films on matching
semiconductor substrates has been an area of great interest especially
where the thin film is ferromagnetic. A number of devices have been
proposed and are currently being developed which employ integrated
magnetic and semiconductor structures.
GMR in Electrodeposited Superlattices 153
7.5. Resistivity in Metals
The electrical resistance of most metals at room temperature (~300 K) is
caused by conduction electrons being scattered from lattice phonons, and
at liquid helium temperature (~4 K) it is caused by scattering of the
conduction electrons from impurity atoms and lattice imperfections.30 A
pure and perfect (ideal) crystal has infinite conductivity (zero resistivity),
but when the atoms are displaced from their mean positions due to
thermal motion, impurities and vacancies the electrons are scattered and
the metal has a finite resistance. The resistance of a metal is proportional
to the number of times per second an electron is scattered from one state
to another state. In transition metals, resistivity is caused by electrons
being scattered from the 4s to the 3d band, because the density of states
in the d band is large and therefore electrons will jump more frequently
from the s to the d band than from one s band state to another. Therefore
all of the transition metals are relatively poor conductors compared to the
elements that follow them in the periodic table. For example the
conductivity of nickel and palladium (with unoccupied d bands) is much
less than the conductivity of the neighbouring elements copper and silver
which have occupied d bands. In transition metals the relaxation time is
also shorter and the conductivity is smaller than for noble metals in
which only s-s transitions can take place.31
7.6. Magnetoresistance
Magnetoresistance is defined as a fractional change in the electrical
resistance of a material when subjected to a magnetic field,
∆R R o − R s
MR = = (6)
R Rs
where Ro is the resistance at H = 0 and Rs is the resistance at saturation
field Hs. At the saturation field, the magnetization of the sample will be
saturated. Usually it is more convenient to measure the percentage
change in the resistance instead of the resistance change.
154 G. Nabiyouni
∆R Ro − Rs
MR(%) = × 100 = × 100(%) (7)
R Rs
If the applied field is not large enough to saturate the magnetization, the
resistance at maximum applied field RHmax, is used instead of Rs.
7.6.1. Ordinary magnetoresistance
When an electric field (E) is applied to a metallic specimen, a current
density (j) flows parallel to the electric field, from which the normal
electrical resistivity (ρ) can be defined as (ρ = E/j). Now if a magnetic
field H is applied in addition to the electric field, the conduction
electrons are forced to have helical instead of linear paths between two
collisions and the resistivity of the specimen will be increased. This
phenomenon is called ordinary magnetoresistance. If the magnetic field
is parallel to the current density, the fractional change in resistance that
occurs is called the longitudinal magnetoresistance, and if the applied
field is perpendicular to the current density, the fractional change in
resistance is called the transverse magnetoresistance.
All non-magnetic metals exhibit an increase in their electric resistivities
as a magnetic field is applied, and the transverse magnetoresistance is
larger than the longitudinal one. In the first approximation the increase
in the resistivity is proportional to H 2, but this can be more complicated
when a high magnetic field is applied or at low temperature. The simple
free electron model can not predict magnetoresistance quantitatively, and
to obtain this it is necessary to consider more complicated models. The
magnetoresistance itself is a second order effect compared to the normal
resistance. The ordinary magnetoresistance has been treated theoretically
by a number of authors.32,33
7.6.2. Anisotropic magnetoresistance
It was about 160 years ago that William Thomson discovered that
the resistivity of nickel and iron changes when they are magnetised.34
In ferromagnetic materials there is a decrease in the transverse
GMR in Electrodeposited Superlattices 155
magnetoresistance (current perpendicular to magnetic field), and an
increase in the longitudinal magnetoresistance (current parallel to magnetic
field), hence it is known as anisotropic magnetoresistance (AMR).
The magnitude of the change in resistance caused by magnetization is
usually a few per cent and rarely reaches 2% at room temperature.
For a ferromagnetic specimen which is subjected to a magnetic field
the resistivity ρ(ξ) is given by35:
ρ(θ) = ρ⊥ + ∆ρ cos2(θ) (8)
where θ is the angle between the magnetization M and current density j,
and ρ⊥ and ρ|| are the resistivity of the specimen when M is perpendicular
and parallel to j respectively. The initial difference ∆ρ is the magnitude
of the anisotropic magnetoresistance;
∆ρ = ρ|| - ρ⊥ (9)
AMR has been observed in thin films as well as in bulk materials.
The magnitude of the AMR in thin films depends on the thickness, grain
size and film surface conditions.36
Large AMR values (up to 8.2% at 4.2 K in MBE-grown Co/Ni
multilayers) have been reported in Co/Ni (ferromagnetic/ferromagnetic)
multilayer thin films.37 A review of AMR in bulk and thin film
ferromagnetic 3d alloys is given by McGuire and Potter.35
7.7. Giant Magnetoresistance (GMR)
The phenomenon of giant magnetoresistance (GMR) has been observed
in many multilayer structures of the form n[(F(tF Å)/NM(tNM Å)] in
which n is the number of bilayers, F is a ferromagnetic transition metal
such as Fe, Co, Ni or one of their alloys, NM is a non-magnetic transition
or noble metal such as Cr, Cu, Ag, Au, Pt and tF and tNM refer to the
thickness of the magnetic and non-magnetic layer (in Å) respectively.
The first observation of GMR was reported by Baibich et al. who found
the resistivity of (001)Fe/(001)Cr superlattices prepared by molecular
beam epitaxy (MBE) on (001)GaAs decreased as an external magnetic
field was applied. The percentage change in resistivity increased with
decreasing temperature, so that at 4.2 K the resistivity decreased by
156 G. Nabiyouni
almost a factor of 2 in a magnetic field of 2T. Nearly at the same time,
and independently, Binasch et al.38 found that the resistivity of Fe-Cr-Fe
sandwiches prepared by MBE on (110)GaAs decreased when a magnetic
field was applied parallel to the film planes. The effect was much
stronger than the usual anisotropic magnetoresistance, measured for a
single Fe layer with the same thickness, (25 nm) which was prepared in
similar growth conditions in order to enable comparison.
In contrast to AMR where the transverse MR decreases and the
longitudinal MR increases as a magnetic field is applied, in GMR both
the transverse and longitudinal MR decrease with increasing magnetic
field. In most cases the magnitude of the transverse MR is larger than
the longitudinal one, due to an anisotropic component in addition to the
GMR.
It was noted earlier that the resistivity of metals is caused by
scattering of conduction electrons. The amount of scattering and hence
the resistivity is related to whether the scattered electrons are spin moment
up (majority spin) or spin moment down (minority spin) i.e. whether
their spin moments are parallel or antiparallel to the magnetisation.
Electric current is carried by both spin moment up and spin moment
down electrons which may be considered as moving in two different
channels with associated resistivity of ρ↑ and ρ↓ respectively. The total
current is therefore the sum of the currents carried by spin moment up
and spin moment down electrons. In non-magnetic metals, the two
currents and hence the resistivities of the two channels are equal, but in
ferromagnetic metals and magnetic multilayers they are different.39,40
The effective resistivity ρ, can be determined by adding the spin
moment up resistivity ρ↑ and the spin moment down resistivity ρ↓ in
parallel
ρ ↑ρ↓
ρ= . (10)
ρ↑ + ρ ↓
It is generally accepted that GMR arises from spin dependent scattering
with electrons of one spin moment scattered more strongly than those of
the other. The fact that for the ferromagnetic transition metals the spin
moment down electrons (minority spins) are usually scattered more
strongly than the spin moment up electrons (majority spins), can be
GMR in Electrodeposited Superlattices 157
understood in term of the density of states at the Fermi level. In non-
magnetic metals and in the absence of an external magnetic field, the
densities of states for spin moment up and spin moment down electrons
are equal. Application of a magnetic field splits the spin moment up and
spin moment down bands. In ferromagnetic metals, however, the spin
moment up and spin moment down bands are split due to the local
magnetization, and the result is an asymmetry between the scattering of
majority and minority spin conduction electrons. Figure 7.6 shows a
schematic diagram of the density of states of the s and d bands in a
ferromagnetic transition metal. The spin moment down electrons have
more unoccupied d states available to be scattered into, than the spin
moment up electrons, and hence they are scattered more strongly than the
spin moment up electrons.
Fig. 7.6. A schematic diagram of the density of state for s and d bands in a transition
metal (redrawn from Ref. 28).
158 G. Nabiyouni
The resistivity of a magnetic multilayer for each conduction channel
(spin moment up or spin moment down) depends on the relative
alignment of the magnetizations of neighbouring layers. If the
magnetizations of successive magnetic layers are antiparallel Fig. 7.7(a),
each spin moment direction is alternately strongly scattered when the
spin moment direction and the magnetization are antiparallel, and weakly
scattered when the spin moment direction and the magnetization are
parallel. The effective resistivity of two successive ferromagnetic layers
in an antiparallel configuration can be written as:
1 1 1
= + (11)
ρ ρ↑ + ρ↓ ρ↓ + ρ ↑
1
so ρ ~ ρ ↓ (assuming ρ↑ << ρ↓) which is still large. If the magnetizations
2
of successive layers in the multilayer specimen are parallel, (due to
application of a magnetic field) spin moment up electrons are hardly
scattered whereas spin moment down electrons are very strongly
scattered Fig. 7.7(b). Thus the effective resistivity of successive layers in
parallel alignment may be written as
1 1 1
= + (12)
ρ ρ↑ + ρ ↑ ρ↓ + ρ ↓
so ρ ~ 2ρ↑ which is quite small.
Thus in magnetic multilayers, if the magnetizations of the magnetic
layers are initially antiparallel to each other, application of an external
magnetic field can rearrange them to be parallel to the field and decrease
the resistivity of the sample.
This is the origin of GMR, as first proposed by Baibich et al. The
discussion is only correct if the thicknesses of the individual layers are
much less than the mean free path of the conduction electrons.
7.8. Oscillatory GMR in Superlattices
After the discovery of GMR and the attribution of the effect to the
relative arrangement of the magnetization in the different magnetic
layers, Parkin et al.41,42 discovered that the magnetic coupling in
the Co/Cu, Co/Ru Co/Cr and Fe/Cr systems oscillated between
GMR in Electrodeposited Superlattices 159
Fig. 7.7. Alignment of the magnetization in superlattices (a) in the antiparallel
configuration and (b) in the parallel configuration. In each magnetic layer some of the
electrons are strongly scattered (high resistance), and some are weakly scattered (low
resistance).28
antiferromagnetic and ferromagnetic as the non-magnetic layer thickness
was increased. GMR in Co/Cu superlattices also oscillates as a function
of the Cu layer thickness. The GMR had a maximum value when the
magnetic layers were antiferromagnetically coupled and was reduced
(and even vanished) for ferromagnetic coupling.
Oscillations in the maximum percentage change in MR were
observed later in many different multilayer and sandwich systems,
grown by MBE or sputtering. For example, oscillatory GMR was then
observed in different systems such as Co/Cu,43 Fe/Mo,44 Fe/Cr/Fe,45
160 G. Nabiyouni
Fe/Cu,46 Co/Ag,47,48 Ni81Fe19/Cu,49 Fe/Cr,50,51 Ni-Co/Cu,52,53 and
NiFe/Cu/CoFe/Cu.54 However, while sputtered Co/Cu superlattices
grown on (100) and (111)-orientation substrates exhibit oscillatory
behaviour, the GMR in MBE grown superlattices on (111)-oriented
substrates decreases continuously as the spacer layer thickness is
increased.49,55
Oscillatory behaviour in electrodeposited superlattices was first
reported by Lashmore et al.56 who observed at least two peaks at Cu
layer thicknesses of about 8 and 24Å. Hua et al.57 reported the
observation of an oscillation in the GMR for electrodeposited Co-Ni-
Cu/Cu superlattices. The first peak that was centred at a copper thickness
of ~10Å has been attributed to AFM coupling analogous to that observed
in many Co/Cu and Co-Ni/Cu sputtered superlattices.52,58,59
Alper et al.,60 who first reported the observation of GMR in
electrodeposited superlattices, did not observe the oscillatory behaviour
in their Co-Ni-Cu/Cu superlattices. They attributed the lack of oscillation
in their samples to fluctuations in the space layer thickness which could
lead to partial contact of successive ferromagnetic layers, and therefore
lead to their being ferromagnetically coupled.
Electrodeposition of Co-Cu/Cu superlattices was first reported by
Bird and Schlesinger,61 who observed a remarkable 55% change in MR
for a superlattice with a Cu layer thickness of ~8Å at a magnetic field of
0.45T. They also reported observation of oscillatory GMR in their Co-
Cu/Cu superlattices, with three peaks observed in the graph of GMR as a
function of Cu layer thickness. Bird and Schlesinger also reported a
maximum percentage change in GMR as large as 7% for a series of
electrodeposited Ni-Cu/Cu superlattices, with varying Cu layer
thickness.
This result is surprising because it is nearly two times larger than
that found by Lashmore et al.56 Oscillatory GMR with three peaks in
good agreement (in peak positions) with those reported by Lashmore
et al.,56 was observed in this series of superlattices. Nabiyouni and
Schwarzacher5 observed oscillatory behaviour for electrodeposited Co-
Ni-Cu/Cu superlattices grown on Cu(100) single crystal substrate, but
they did not observe any oscillatory behaviour for the superlattices grown
GMR in Electrodeposited Superlattices 161
Fig. 7.8. Dependence of the maximum percentage change in ‘transverse’ (current
predominantly perpendicular to the applied field) magnetoresistance on Cu layer
•
thickness for electrodeposited Co-Ni-Cu/Cu superlattices grown on Cu (100) ( ) and Cu
(111) (+). The Co-Ni-Cu layers had a nominal thickness of 30Å. The solid line is a guide
to the eye.28
on (111)-oriented single crystal and (100)-textured polycrystalline
copper substrates. The result is illustrated in Fig. 7.8.
7.9. Research on GMR
Since 1988 when GMR was discovered, (Babich et al. and Binasch et al.)
thousands of papers have been published on GMR in different multilayer,
superlattice and sandwich systems. Araki and Shinjo47 soon confirmed
Baibich’s results by growing Fe/Cr multilayers on MgO and glass
substrates. The significant step of the observation of GMR in different
multilayer systems was made by Parkin et al.,41 who introduced two new
metallic superlattice systems that exhibited the GMR effect, Co/Cr and
Co/Ru. In these systems, as well as in Fe/Cr superlattices, the coupling
between the ferromagnetic layers oscillates between antiferromagnetic
and ferromagnetic as a function of the spacer layer thickness. The
magnitude of the saturation MR is also found to oscillate with the spacer
162 G. Nabiyouni
layer thickness. Parkin et al.47 also demonstrated that the GMR
amplitude in sputtered superlattices could be even larger than in MBE
grown multilayers.
Two different multilayer and sandwich systems, the Fe/Cr42,62,63 and
Co/Cu64–69 systems have been widely studied because they exhibit
a large room temperature GMR as well as the oscillatory behaviour
which caused much scientific interest in the origin of GMR. Co/Cu
multilayers show the largest saturation MR value at room temperature.
Parkin et al.70 sputtered Co(10Å)/Cu(9Å) multilayers with a maximum
room temperature GMR of 65% at 10 kOe. These multilayers exhibit a
maximum of 115% GMR at 4.2 K. However the largest MR value at
liquid helium temperature was reported by Fullerton et al.,71 for sputtered
Fe(14Å)/Cr(8Å) superlattices, with a maximum GMR of 150% at
20 kOe.
Although Co/Cu and Fe/Cr superlattices exhibit large GMR,
a large magnetic field (usually greater than 1 kOe) is required. The
antiferromagnetic coupling between the magnetic layers is so strong that
only a large magnetic field can realign the moments from antiparallel to
parallel. The coupling is stronger when the non-magnetic spacer layer is
ultrathin. There is particular interest, however, in multilayer systems for
which GMR is observed at low field, because of their potential
applications in read heads for magnetic recording. The field range for a
magnetoresistive head is typically from 0–20 Oe. The field sensitivity of
a MR system is defined as the fractional change of resistivity per unit
change in the applied magnetic field i.e. (∆ρ/ρ∆H) × 100% where ∆H is
the field required to change the resistivity by the amount of ∆ρ.
The Co/Cu superlattices with the largest room temperature GMR
exhibit a sensitivity as low as 0.01% Oe-1, which is not high enough
for this particular application. Using soft magnetic materials as the
ferromagnetic layers and increasing the non-magnetic layer thicknesses,
it is possible to fabricate multilayer structures which, although they do
not exhibit a very large GMR value, are sensitive to small magnetic
fields. Ni80Co20 and permalloy Ni80Fe20 are good candidates for soft
magnetic multilayers. Jimbo et al.72 observed a saturation field of as low
as 20 or 30 Oe for Ni66Fe16Co16(15Å)/Cu(22Å) multilayers with
maximum 12% GMR. Relatively thick Cu layers in these multilayers
GMR in Electrodeposited Superlattices 163
prevent the magnetic layers from being strongly antiferromagnetically
coupled and enhance the sensitivity. Further work has been done on soft
magnetic multilayers. A sensitivity of 2.5% Oe-1 has been achieved
by Jimbo et al.73 in spin valve structures. The term ‘spin valve’ was
introduced by Dieny et al.74 In spin valve multilayer systems two
uncoupled ferromagnetic layers with in-plane anisotropy are separated by
a non-magnetic metal. The first layer is an antiferromagnetic material
and there is an exchange coupling between this layer and the first
magnetic layer. Dieny et al. showed that, the spin valve effect is related
to the relative angle between the magnetizations of the two ferromagnetic
layers. Electrodeposition of spin valve multilayers has been reported by
a number of research groups.75,76
GMR was also observed in non-multilayer, but still magnetically
heterogeneous media. It is well known that GMR can be observed in
both antiferromagnetically coupled and uncoupled layer structures. In
uncoupled layer structures, at the coercive field, the magnetizations are
randomly oriented. Granular materials consist of ferromagnetic metallic
particles dispersed in a non magnetic (metallic or non-metallic) matrix.
Berkowitz et al.77 prepared granular Cu-Co alloys in which the
uncoupled layers were replaced by single domain magnetic particles in a
non-magnetic matrix and thereby showed that GMR is not restricted only
to multilayer systems. GMR then was observed in electrodeposited
granular systems.78,79 Fine grains play an important role in obtaining low
coercivity for electrodeposited CoNiFe thin films and other soft
magnetic materials.2,80,81 Ross et al.82 showed that the phosphorous
content in electrodeposited Ni/NiP multilayers dictated the crystalline
sizes. They also showed that in the case of Ni/Cu systems large grain size
were observed when the copper layer was deposited well below the
copper limiting current density, while nano-size grains were reported
when the depositing copper current density was closed to its limiting
value. GMR in granular magnetic systems has been invesigated
theoretically by Sheng et al.,83 Wiser84 and Hichey et al.85
After GMR was discovered in MBE-grown superlattices,23 a number
of groups started to investigate whether it is possible to fabricate
superlattices showing GMR using simpler and cheaper techniques. The
advantages of electrodeposition over these vacuum deposition techniques
164 G. Nabiyouni
encouraged a number of groups to grow superlattices by this method.
Alper et al.60 reported the first observation of GMR in electrodeposited
superlattices. Their Co-Ni-Cu/Cu superlattices showed 15% GMR at
room temperature in a magnetic field of up to 8 kOe. The largest GMR
value in electrodeposited superlattices was reported by Bird and
Schlesinger61 who observed a maximum of 55% GMR in Co-Cu/Cu
multilayers in a field of 0.45T.
GMR magnitude in the electrodeposited superlattices is highly related
to electrochemical growth conditions such as electrolyte pH and
temperature, cell geometry and electrodeposition mode (potentiostatic or
galvanostatic control or a combination on both potentiostatic and
galvanostatic controls). Alper et al.86,87 studied the effect of electrolyte
pH on structural and magnetoresistive properties of Co-Ni/Cu and Ni-
Cu/Cu superlattices. They found that for superlattices with the same
bilayer and total thicknesses, the GMR magnitude decreases as the
electrolyte pH increases. They also found that superlattices grown at high
pH, contain much more Cu compare to those deposited at low pH.
Two comprehensive review articles on GMR in electrodeposited
films have been given by Schwarzacher et al.,88 and, Bakonyi et al.89
7.10. Superparamagnetism Contribution to GMR in the
Electrodeposited Superlattices
One the most important multilayer systems exhibiting GMR effect
is Co/Cu superlattices. As it was explained before, it is not possible
to electrodeposit pure cobalt/pure copper multilayers using an
electrochemical single bath, but the grown material is Co-Cu/Cu instead.
However, by keeping the Cu concentration in the electrolyte as low as
possible one would expect to be able electrodeposite Co-rich/pure Cu
superlattices. This system is appropriate for studying an imperfect
(and in some cases discontinuous) multilayer with the rough
magnetic/nonmagnetic interfaces. Under equilibrium condition Co and
Cu atoms do not expect to mix and make a homogenous alloy, instead,
some separation takes place when Cu is co-deposited during Co
deposition, and when Co is dissoluted during Cu deposition. This is the
case in all of the electrodeposition modes i.e. potentiostatic,90
GMR in Electrodeposited Superlattices 165
galvanostatic91 and two pulse plating92 modes. In spite of consisting
magnetic atoms, in some of Co-Cu alloy the Co particle sizes are too
small to exhibit ferromagnetic behaviour with a constant direction
at the field applied. This region is called superparamagnetic. The
superparamagnetism behaviour is generally related to small
ferromagnetic entities for which the orientation of the magnetization
rapidly fluctuates.93 In such areas, anisotropy energy is less than thermal
energy, so that in the absence of any external magnetic field the
magnetization orientation can be randomly fluctuated. Since GMR is
known to be related to spin dependent scattering, the scattering of
electrons from (or to) the superparamagnetic region as well as scattering
of electrons from (or to) the ferromagnetic layer has a contribution to
GMR magnitude in the electrodeposited Co-Cu/Cu superlattices.
According to the Wiser-Hickey theory,84,85 for a given temperature there
is a distribution of magnetic particle sizes with some particle being in
the superparamagnetic regime and the rest in the ferromagnetic regime.
With the simultaneous presence of both ferromagnetic (FM) and
superparamagnetic (SPM) regions the GMR contain three contributions:
(i) scattering of electrons from one superparamagnetic region to the next
in the successive magnetic layer (SPM SMP), (ii) scattering of
electrons from one superparamagnetic region to the successive
ferromagnetic layer and vice versa (SPM FM = FM SPM), and
(iii) scattering of electrons from one ferromagnetic layer to the
superparamagnetic region in the next successive ferromagnetic layer
(FM FM). The above contributions have different effects on the
magnetoresistance. While FM to FM contribution leads to a
magnetoresitance curve saturating in a magnetic field as small as 1 kOe,
much stronger magnetic field (typically larger than 10 kOe) is required to
saturate the contribution of SPM to SPM in the magnetoresistance curve.
Attempt to describe the magnetoresistance of electrodeposited Co-Ni-
Cu/Cu superlattices was made by Nabiyouni et al.90 Based on the Wiser
and Hickey theory on the effect of superparamagnetism on GMR in the
granular metals, Bakonyi and Peter94–98 showed that in electrodeposited
Co-Cu/Cu superlattices the field dependence of the magnetoresistance at
sufficiently high magnetic field can be successfully described by the
Langvin function L(µ H/kT), with µ denoting to the average magnetic
166 G. Nabiyouni
moment of superparamagnetic regions. Bakonyi and Peter described the
field dependent of the measured magnetoresistance MR(H) for H > Hs in
the two separated terms:
MR = MRFM + GMRSPM × L(µ H/kT) (13)
where;
MRFM = GMRFM + AMR (14)
Liu et al.95 found that, while the increase in the Cu content of the
magnetic layer significantly increases contribution of the GMRSPM in
the magnetoresistance, MRFM contribution is a function of Cu layer
thickness. Regarding to the Eq. (13), while MRFM is not a very sensitive
function of temperature, both GMRSPM and Langvin function L(µ H/kT),
vary as a function of temperature. Peter et al.99 made a comprehensive
study on the effect of temperature on the GMR and magnetic properties
of electrodeposited Co-Cu/Cu multilayers. They found that both GMR
magnitude and coercivity increase as temperature decreases from room
temperature down to liquid helium temperature. They attributed this
result to the influence of superparamagnetism regime on the magnetic
properties and in particular on the GMR value in electrodeposited
superlattices.
7.11. General Remarks on Electrodeposited Superlattices
While very large GMR value has been reported in the multilayer and
sandwich systems prepared by physical deposition techniques, the GMR
magnitude in the electrodeposited superlattices is not high. One of the
largest MR value at liquid helium temperature was reported by Fullerton
et al.,71 for sputtered Fe(14Å)/Cr(8Å) superlattices, with a maximum
GMR of 150% at 20 kOe. On the best knowledge of author the largest
GMR value in electrodeposited superlattices was reported by Bird and
Schlesinger61 who observed a maximum of 55% GMR in Co-Cu/Cu
multilayers in a field of 0.45T. The fact that the magnitude of GMR in
electrodeposited superlattices is almost by a factor of two smaller than
which obtained for similar superlattice system prepared by physical
deposition methods, can be attributed to the lack of complete
antiferromagnetic coupling between two successive magnetic layers.
GMR in Electrodeposited Superlattices 167
Very few research groups have observed oscillatory GMR as a
function of spacer layer thickness. Nabiyouni and Schwarzacher5
observed oscillatory behaviour for electrodeposited Co-Ni-Cu/Cu
superlattices grown on Cu(100) single crystal substrate, but none was
observed for the superlattices grown on (111)-oriented single crystal and
(100)-textured polycrystalline copper substrates.
In most cases, the magnetoresistance curves of electrodeposited
superlattices, do not saturate at low magnetic fields. They usually require
magnetic fields of up to 10 kOe to saturate, while for superlattices grown
by vacuum base techniques, only half of this magnetic field is enough to
saturate.
Despite some limitations in using electrodeposition technique to
fabricate superlattices with GMR properties, still electrodeposition
promises a cheap, simple and more importantly a reliable method for
preparation of thin films and superlattices.
Acknowledgments
I would like to thank those who to gave me permission to refer to their
results. Special thanks to W. Schwarzacher and I. Bakonyi whose review
articles provide me a good source of materials in electrodeposition of
metallic multilayers. I would also like to thank my PhD student
K. Hedayati who carefully and penitently typed, corrected and justified
the manuscript to the right format of this book.
References
1. P. C. Andricacos and N. Robertson, IBM J. Res. Dev. 42, 671 (1998).
2. T. Osaka, Electrochimica Acta 45, 3311 (2000).
3. M. Alper, K. Attenborough, V. Baryshev, R. Hart, D. S. Lashmore, and
W. Schwarzacher, J. Appl. Phys. 75, 6543-6545 (1994).
4. K. Attenborough, R. Hart, S. J. Lane, M. Alper, and W. S. Schwarzacher, J. Magn.
Magn. Mater. 148, 335-336 (1995).
5. G. Nabiyouni and W. Schwarzacher, J. Magn. Magn. Mater. 156, 355-356 (1996).
6. I. Bakonyi, J. Tóth, L. Goualou, T. Becsei, E. Tóth-Kádár, W. Schwarzacher, and
G. Nabiyouni, J. Electrochem. Soc. 149, C195-C201 (2002).
7. A. Yamada, T. Houga, and Y. Ueda, J. Magn. Magn. Mater. 239, 272-275 (2002).
168 G. Nabiyouni
8. A. P. O’Keeffe, O. I. Kasyutich, W. Schwarzacher, L. S. de Oliveira, and A. A. Pasa,
Appl. Phys. Lett. 73, 1002-1004 (1998).
9. E. M. Kakuno, R. C. da Silva, N. Mattoso, W. H. Schreiner, D. H. Mosca, and S. R.
Teixeira, J. Phys. D: Appl. Phys. 32, 1209-1213 (1999).
10. H. Zaman, S. Ikeda, and Y. Ueda, IEEE Trans. Magn. 33, 3517-3519 (1997).
11. I. Bakonyi, E. Tóth Kádár, Á. Cziráki, J. Tóth, L. F. Kiss, C. Ulhaq-Bouillet,
V. Pierron-Bohnes, A. Dinia, B. Arnold, K. Wetzig, P. Santiago, and M.-J.
Yacamán, J. Electrochem. Soc. 149, C469-C473 (2002).
12. C. L. S. Rizal, A. Yamada, Y. Hori, S. Ishida, M. Matsuda, and Y. Ueda, Phys. Stat.
Sol. (c) 1, 1756-1759 (2004).
13. L. Péter, Z. Kupay, J. Pádár, Á. Cziráki, Zs. Kerner, and I. Bakonyi, Electrochim.
Acta 49, 3613-3621 (2004).
14. S. M. S. I. Dulal, E. A. Charles, and S. Roy, Electrochim. Acta 49, 2041-2049
(2004).
15. A. Brenner, Electrodeposition of Alloys: Principles and Practise. Academic Press,
New York (1963).
16. P. C. Andricacos, C. Uzoh, J. O. Horkans, and H. Deligianni, IBM J. Res. Dev. 42,
567 (1998).
17. F. Spitzer, Z. Elektrochem. 11, 345 (1905).
18. W. Blum, Trans. Amer. Electrochem. Soc. 40, 307 (1921).
19. U. Cohen, F. B. Koch, and R. Sard, J. Electrochem. Soc. 130, 1987 (1983).
20. D. Tench and J. White, Metall. Trans. A. 15A, 2039 (1984).
21. J. Yahalom, D. F. Tessier, R. S. Timsit, A. M. Rosenfeld, D. F. Mitchell, and P. T.
Robinson, J. Mater. Res. 4, 755 (1989).
22. D. S. Lashmore and M. P. Dariel, J. Electrochem. Soc. 135, 1288 (1988).
23. M. N. Baibich, J. M. Broto, A. Fert, F. Nguyen Van Dau, F. Petroff, P. Etienne,
G. Creuzet, A. Friedrich, and J. Chazelas, Phys. Rev. Lett. 61, 2472 (1988).
24. M. Alper, K. Attenborough, R. Hart, J. S. Lane, D. S. Lashmore, C. Younes, and
W. Schwarzacher, Appl. Phys. Lett. 63, 2144 (1993).
25. J. A. Switzer, R. P. Raffaelle, R. J. Phillips, C. J. Hung, and T. D. Golden, Sci. 252,
1918 (1992).
26. I. Kazeminezhad, PhD Thesis, University of Bristol (2001).
27. G. Nabiyouni and W. Schwarzacher, J. of Crystal Growth e275, 1259 (2005).
28. G. Nabiyouni, PhD Thesis, University of Bristol (1997).
29. R. Hart, PhD thesis, University of Bristol (1996).
30. C. Kittel, Introduction to Solid State Physics, Wily Eastern, Eight Edition (2004).
31. N. F. Mott, Proc. Phys. Soc. 47, 571 (1936).
32. J. P. Jan, Solid states Physics, Vol 5, F. Seitz and D. Turnlbull, Eds. New York,
Academic Press (1957).
33. R. S. Allgaier, Phys. Rev. B 8, 4470 (1973).
34. W. Thomson, Proc. Roy. Soc. 8, 546 (1857).
35. T. R. Mc Guire and R. T. Potter, IEEE. Trans. Magn. 11 (4), 1081 (1975).
GMR in Electrodeposited Superlattices 169
36. M. Tondra, D. K. Lottis, K. T. Riggs, Y. Chen, and E. D. Dahlberg, J. Appl. Phys.
73, 6393 (1993).
37. J. M. Gallego, D. Lederman, T. J. Moran, and I. K. Schuller, Appl. Phys. Lett. 64,
2590 (1994).
38. G. Binasch, P. Grünberg, F. Saurenbach, and W. Zinn, Phys. Rev. B 39, 4828 (1989).
39. P. M. Levy, J. Magn. Magn. Mater. 140-144, 485 (1995).
40. A. Barthélémy, A. Fert, R. Morel, and L. Steren, Physics World 34 (1994).
41. S. S. P. Parkin, S. Fan, and N. More, J. Appl. Phys. 67, 5931 (1990).
42. S. S. P. Parkin, G. Z. Li, and D. J. Smith, Appl. Phys. Lett. 58, 2710 (1991).
43. D. H. Mosca, F. Petroff, A. Fert, P. A. Schroeder, W. P. Pratt, and R. Laloee,
J. Magn. Magn. Mater. 94, L1 (1991).
44. M. E. Brubaker, J. E. Mattson, C. H. Sowers, and S. D. Bader, Appl. Phys. Lett. 58,
2306 (1991).
45. J. Unguris, R. J. Celotta, and D. T. Piercfe, Phys. Rev. Lett. 67, 140 (1991).
46. F. Petroff, A. Barthélemy, A. Hamzic, A. Fert, P. Etienne, S. Lequien, and
G. Creuzet, J. Magn. Magn. Mater. 93, 95 (1991).
47. S. Araki and T. Shinjo, Jpn. J. Appl. Phys. 29, L621 (1990).
48. L. Wu, T. Shinjo, and N. Nakayama, J. Magn. Magn. Mater. 125, L14 (1993).
49. S. S. P. Parkin, R. F. Marks, R. F. C. Farrow, G. R. Harp, Q. H. Lam, and R. J.
Savoy, Phys. Rev. B. 46, 9262 (1992). S. S. P. Parkin, Appl. Phys. Lett. 60, 512
(1992).
50. P. Grunberg, S. Demokritov, A. Fuss, R. Schreiber, J. A. Wolf, and S. T. Purcell,
J. Magn. Magn. Mater. 104-107, 1734 (1992).
51. M. A. M. Gijs and M. Okada, J. Magn. Magn. Mater. 113, 105 (1992).
52. X. Bian, J. Q. Ström-Olsen, Z. Altounian, Y. Huai, and R. W. Cochrane, Appl. Phys.
Lett. 62, 3525 (1993).
53. H. Kubota, S. Ishio, and T. Miyazaki, J. Magn. Magn. Maert. 126, 463 (1993).
54. M. T. Kife, J. Bresowar, and Q. Leng, J. Appl. Phys. 79, 4766 (1996).
55. M. J. Hall, B. J. Hickey, M. A. Howson, M. J. Walker, J. Xu, D. Greig, and
N. Wiser, Phys. Rev. B 47, 12785 (1993).
56. D. S. Lashmore, Y. Zhang, S. Hua, M. P. Dariel, L. Swartzendruber, and
L. Salamanca-Riba, Magnetic Materials, Processes and Devices, Edited by L. T.
Romankiw and D. A. Herman, Electrochem. Soc. Symp. Proc. 94-6, 205 (1994).
57. S. Hou, C. Harell, L. Trofin, P. Kohli, and C. R. Martin, J. Am. Chem. Soc. 126,
5674 (2004).
58. A. Fert, A. Barthélémy, P. Etienne, S. Lequien, R. Loloee, D. K. Lottis, D. H.
Mosca, F. Petroff, W. P. Pratt, and P. A. Schroeder, J. Magn. Magn. Mater. 104-107,
1712 (1992).
59. S. Honda, S. Ohmoto, R. Imada, and M. Nawate, J. Magn. Magn. Mater. 126, 419
(1993).
60. M. Alper, P. S. Aplin, K. Attenborough, D. J. Dingley, R. Hart, S. J. Lane, D. S.
Lashmore, and W. Schwarzacher, J. Magn. Magn. Mater. 126, 8 (1993).
170 G. Nabiyouni
61. K. Bird and M. Schlesinger, J. Electrochem. Soc. 142(4), L65 (1995).
62. G. Oomi, Y. Uwatoko, Y. Obi, K. Takanashi, and K. Fujimori, J. Magn. Magn.
Mater. 126, 513 (1993).
63. R. Schad, C. D. Potter, P. Beliën, G. Verbanck, V. V. Moshchalkov, and
Y. Bruyneraede, Appl. Phys. Lett. 64, 3500 (1994).
64. M. Uhlemann, A. Gebert, M. Herrich, A. Krause, A. Cziráki, and L. Schultz,
Electrochim. Acta 48, 3005-3011 (2003).
65. H. Zhang and W. Wang, Microsyst. Technol. 9, 436-440 (2003).
66. Y. Hayashi, C. G. Lee, B. H. Koo, T. Sato, M. Arita, and M. Masuda, Phys. Stat. Sol.
(a) 201, 1658-1661 (2004).
67. L. Péter, Q. X. Liu, Z. Kerner, and I. Bakonyi, Electrochim. Acta 49, 1513-1526
(2004).
68. H. Sasaki, S. Kainuma, K. Takayanagi, K. Hisatake, and C. O. Kim, J. Magn. Magn.
Mater. 281, 53-57 (2004).
69. D. K. Pandya, P. Gupta, S. C. Kashyap, and S. Chaudhary, J. Magn. Magn. Mater.
doi: 10.1016/j.jmmm. 2008.03.020 (2008).
70. S. S. P. Parkin, R. Bhadra, and K. P. Roche, Phys. Rev. Lett. 66, 2152 (1991).
71. E. E. Fullerton, M. J. Conver, J. E. Mattson, C. H. Sowers, and S. D. Bader, Appl.
Phys. Lett. 63, 1699 (1993).
72. M. Jimbo, S. Tsunashima, T. Kanda, S. Goto, and S. Uchiyama, J. Appl. Phys. 74,
3341 (1993).
73. M. Jimbo, K. Komiyama, and S. Tsunashima, J. Appl. Phys. 79(8), 6237 (1996).
74. B. Dieny, V. S. Speriosu, B. A. Gurney, S. S. P. Parkin, D. R. Wilhoit, K. P. Roche,
S. Metin, D. T. Peterson, and S. Nadimi, J. Magn. Magn. Mater. 93, 101 (1991).
75. Y. Jiang, S. Yao, and W. Zhang, Thin Solid Films 516, 3210-3216 (2008).
76. K. Attenborough, H. Boeve, J. De Boeck, G. Borghs, and J. P. Celies, Appl. Phys.
Lett. 74(15), 2206 (1999).
77. A. E. Berkowitz, J. R. Mitchell, M. J. Caarey, A. P. Young, S. Zhang, F. E. Spada,
F. T. Parker, A. Hutten, and G. Thomas, Phys. Rev. Lett. 68, 3745 (1992).
78. S. Kenane, J. Voiron, N. Benbrahim, E. Chainet, and F. Robaut, J. Magn. Magn.
Mater. 297, 99-106 (2006).
79. G. Nabiyouni and W. Schwarzacher, J. Magn. Magn. Mater. 198, 116-118 (1999).
80. H. Iwasaki, R. Akashi, and Y. Ohsawa, J. of Appl. Phys. 73, 8441 (1993).
81. H. Matsubara and A. Yamada, J. Electrochem. Soc. 141, 2386 (1994).
82. C. Ross, L. M. Goldman, and F. Spaepen, J. Electrochem. Soc. 140, 91 (1993).
83. L. Sheng, R. Y. Gu, D. Y. Xing, Z. D. Wang, and J.-X. Zhu, J. Appl. Phys. 79(8),
6255 (1996).
84. N. Wiser, J. Magn. Magn. Matter. 159, 119 (1996).
85. B. J. Hckey, M. A. Howson, S. O. Musa, and N. Wiser, Phys. Rev B 51, 667 (1995).
86. M. Alper, M. C. Baykul, L. Peter, J. Toth, and I. Bakonyi, J. Appl. Electrochem. 34,
841-848 (2004).
GMR in Electrodeposited Superlattices 171
87. M. Alper, W. Schwarzacher, and Lane S. L. Lane, J. Electrochem. Soc. 144, 2346
(1997).
88. W. Schwarzacher and D. S. Lashmore, IEEE Trans. Magn. 32(4), 3133 (1996).
89. I. Bakonyi and L. Peter, Proc. 8th Int. Symp. On Magnetic Materials, Processes and
Divices (206th Electrochemical Society Meeting, Honalulu, Hawaii, USA, 2004).
90. G. Nabiyouni, W. Schwarzacher, Z. Rolik, and I. Bakonyi, J. Magn. Magn. Mater.
253, 77-85 (2002).
91. Á. Cziráki, J. G. Zheng, A. Michel, Z. S. Czigány, G. Nabiyouni, W. Schwarzacher,
E. Tóth-Kádár, and I. Bakonyi, Z Metallkde 90, 278-283 (1999).
92. Q. X. Liua, L. Peter, J. Toth, L. F. Kiss, A. Cziraki, and I. Bakonyi, J. Magn. Magn.
Mater. 280, 60-74 (2004).
93. B. D. Cultity, Introduction to Magnetic materials, Addison-Wesley, Reading (1972).
94. I. Bakonyi, L. Péter, Z. Rolik, K. Kiss-Szabó, Z. Kupay, J. Tóth, L. F. Kiss, and
J. Pádár, Phys. Rev. B 70, 054427 (2004).
95. Q. X. Liu, L. Péter, J. Pádár, and I. Bakonyi, J. Electrochem. Soc. 152, C316-C323
(2005).
96. Q. X. Liu, L. Peter, J. Toth, L. F. Kiss, A. Cziraki, and I. Bakonyi, J. Magn. Magn.
Mater. 280, 60-74 (2004).
97. I. Bakonyi, J. Tóth, L. F. Kiss, E. Tóth-Kádár, L. Péter, and A. Dinia, J. Magn.
Magn. Mater. 269, 156-167 (2004).
98. L. Péter, V. Weihnacht, J. Tóth, J. Padar, L. Pogany, C. M. Schneider, and
I. Bakonyi, J. Magn. Magn. Mater. 312, 258-265 (2007).
99. L. Péter, Z. Rolik, L. F. Kiss, J. Tóth, V. Weihnacht, C. M. Schneider, and
I. Bakonyi, Phys. Rev. B 73, 174401/1-10 (2006).
This page is intentionally left blank
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Chapter 8
INTRODUCTION TO SPIN TRANSFER TORQUE
C. Baraduc, M. Chshiev∗ and U. Ebels†
SPINTEC, CEA/CNRS/UJF/INPG
CEA/INAC, 17 rue des martyrs
38054 Grenoble cedex 9, France
E-mail:
[email protected] Spin transfer torque corresponds to the interaction of a spin polarized
electronic current with the local magnetization: magnetic moment is
transferred from the conduction electrons to the magnetization, resulting
in a change of the magnetization orientation.
This chapter introduces the physics of spin transfer torque in mag-
netic devices. It is meant as a tutorial and provides an elementary
discussion of the basic concepts. It is not a review of the extensive the-
oretical and experimental work done on this subject in these last years.
8.1. Introduction
In 1996, it was theoretically demonstrated both by Berger1 and Slon-
czewski2 that a spin polarized electronic current can transfer magnetic
moment to the local magnetization of a ferromagnet. In particular, this
spin transfer can induce steady-state precession or reversal of the magne-
tization. In addition to its intrinsic scientific interest, the control of the
magnetization state by an electrical current, instead of an applied field,
may lead to several possible applications.
In order to appreciate this breakthrough, let us say a few words on
spin electronics, also called “spintronics”. Spintronic devices are usually
composed of two ferromagnetic layers separated by a non magnetic spacer.
Depending on the nature of the spacer, one consider two broad classes:
spin valves when the spacer is a normal metal layer, and magnetic tun-
nel junctions when a spacer is an insulator (or semiconductor). In these
∗
[email protected] †
[email protected] 173
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
174 C. Baraduc, M. Chshiev and U. Ebels
structures, one ferromagnetic layer is called the fixed or the reference layer,
since its magnetization is pinned by magnetic coupling to an antiferro-
magnet, and the other is the free layer: the free layer magnetization can
be switched to obtain either parallel or antiparallel alignment of the two
magnetizations, which corresponds to two different resistance states of the
device. This magnetoresistance has led to various applications such as field
sensors (in particular read-heads for hard-disk) or two-state devices like
magnetic memory (MRAM). Concerning MRAMs, significant progress can
be achieved thanks to the spin transfer torque effect, since commutation
between the two states of the memory can be directly driven by an electrical
current.
Since it was theoretically predicted, the concept of spin transfer torque
attracted a considerable interest and was tested on a large variety of devices,
either metallic multilayer or magnetic tunnel junctions, with different ge-
ometries: mechanical or lithographically defined point contact, nanopillars,
electrochemically grown nanowires etc. All these devices present the same
essential characteristic: they have nanometric cross-sectional area with typ-
ical dimension about 100 nm. Such a small size is necessary since spin
transfer torque can only be observed for very large current density of the
order of 106 to 108 A.cm−2 . Soon after its theoretical prediction, a pioneer
work showed experimentally the existence of spin transfer torque.3 Then
unambiguous demonstrations of current induced magnetization switching
between the parallel and antiparallel states were performed on metallic
systems during the years 2000-2001.4,5 Similar results were later obtained
on magnetic tunnel junctions.6,7 Moreover, spectral measurements showed
that spin transfer torque induces also steady-state precession of the mag-
netization at frequencies of a few GHz,8,9 resulting in oscillations between
the high and the low resistance level.
In the following, we will first introduce the concept of spin transfer
torque. Then we will get more physical insight by considering this phe-
nomenon at the microscopic level. Finally, we will address the influence of
spin transfer torque on magnetization dynamics.
8.2. Spin Transfer Torque
In this section, we present a simple and intuitive approach of the physical
underlying mechanism of spin transfer torque using a macrospin descrip-
tion. Let us consider two ferromagnetic layers F1 and F2 , separated by a
thin spacer (either a metallic layer or a tunnel barrier) with their magne-
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 175
tizations misaligned by an angle θ (see Fig. 8.1(a)). An electrical current
passes through the structure, perpendicularly to the interfaces, with elec-
trons flowing from F1 to F2 . Electrons flowing in F1 are spin polarized
along the direction of the magnetization M 1 of this ferromagnetic layer,
whereas electrons flowing in F2 are spin polarized along the direction of
M 2 . So in the time span between crossing the imaginary surfaces A and B,
the spin polarization of electrons changes. At this stage, we should distin-
guish between spin polarization and magnetic moment: the electron mag-
netic moment is opposite to its spin-angular momentum µ = ge µB S/ with
ge ≈ −2, where S is the spin, and likewise in transition-metal ferromagnets
the magnetization is generally opposite to the spin density M = gµB s/
where g is typically in the range −2.1 to −2.2.
Let us consider the fictitious box defined by the surfaces A and B:
incoming electrons have a magnetic moment µ1 parallel to M 1 and outgo-
ing electrons a magnetic moment µ2 parallel to M 2 . Since the incoming
and outgoing magnetic moment is not the same, some magnetic moment
is transferred to the system per time unit. However magnetization vec-
tors have a fixed magnitude, so any temporal variation of M has to be
perpendicular to M :
|2
d|M · dM = 0 .
= 2M (8.1)
dt dt
Consequently only the component perpendicular to M of the electrons
magnetic moment can be transferred to the bulk magnetization. The to-
tal transferred moment T is then decomposed in two parts T1 and T2
1 and the other to M
(Fig. 8.1(b)). One part is transferred to M 2 . The
a) b)
T2
F1 F2 T1
T µ2
µ1 T
m̂2
A B m̂1
Electron
Electron flow flow mˆ 1 u mˆ 2
Fig. 8.1. a) The studied device composed of two ferromagnetic layers F1 and F2 sepa-
rated by a spacer. The virtual planes A and B are far enough from the N/F interface so
that the current is perfectly spin polarized along the ferromagnetic layer magnetization.
b) Schematics of the ingoing (µ1 ), outgoing (µ2 ) and transferred (T ) magnetic moments,
drawn in the plane that contains the magnetizations of the two layers. In the particular
case studied here, this plane is parallel to the N/F interfaces.
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
176 C. Baraduc, M. Chshiev and U. Ebels
magnetic moment transferred to M 1 is then a vector perpendicular to
M1 which lies in the plane defined by (M 1, M
2 ). Since the unit vector
a
normal to this plane is m̂1 × m̂2 , T1 and T2 are necessarily written as:
T1,2 ≡ T1,2 m̂1,2 ×(m̂1 × m̂2 ). By analogy to magnetization dynamics, where
the time derivative of the magnetization is equal to the torque exerted by
the magnetic field, this transferred moment per time unit was originally
called “pseudo-torque” by Slonczewski.10 It is now universally called spin
transfer torque. Nevertheless we must keep in mind that this term is not
exactly a torque but stems directly from the flow of spin momentum into
the considered volume element.
In the case of a two-magnetic-layers device composed of a fixed and
a free layer, the spin torque on the free-layer magnetization due to the
misalignment with the fixed layer magnetization is:
aj
T = γ0 M × (M × p̂) (8.2)
Ms
where Ms is the saturation magnetization and p̂ the magnetic polarization
of the fixed layer. The prefactor aj was calculated in the case of a metallic
multilayer.11 The subscript recalls that the torque is parallel to the plane
that contains the two magnetization vectors.
8.3. A Microscopic Picture
We will try now to get more insight on the physics of spin transfer torque
at the microscopic level. Again we consider the device composed of two
ferromagnetic layers F1 and F2 separated by a normal metal spacer N ,
with their magnetizations misaligned by an angle θ. Let us consider a single
electron, initially polarized by F1 along zˆ , arriving at the N/F2 interface.
The incoming electron is described as a plane wave with a wave-vector equal
to k. With respect to the quantization axis of the second ferromagnet F2
(ẑ), the electron wave-function is a superposition of spin-up and spin-down
states:
eikx
ψin = √ [cos(θ/2) |↑ + sin(θ/2) |↓] . (8.3)
Ω
This decomposition corresponds to a geometrical projection on an orthogo-
nal basis, taking into account the fact that |↑ and |↓ define an orthogonal
basis in spin-space, but with a π angle in geometrical space. Geometrical
angles must then be divided by 2 when dealing with spin-space.
a Where i for i = 1, 2.
m̂i is a unit vector along M
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 177
Following the Stoner model, spin-up and spin-down electrons are con-
sidered as free electrons but with a different energy potential: V+ for spin-
up electrons, and V− = V+ + ∆ for spin-down electrons, where ∆ is the
exchange energy. So, depending on its spin, an incoming electron will ex-
perience a different potential step at the N/F2 interface (Fig. 8.2). For
simplicity, we assume that V+ = 0. In that case, a spin-up electron is al-
ways transmitted whereas a spin-down electron may be either partially or
fully reflected depending whether its energy is larger or smaller than ∆.
Now let us consider an incident electron polarized along zˆ with an en-
ergy larger than ∆. When entering into F2 , the spin-up and spin-down
parts of its wave-function are submitted to a different potential, so their
velocities will be different. We note k↑ (resp. k↓ ) the wave-vector of spin-up
(resp. spin-down) electrons. By matching wave-functions and their deriva-
tives at the interface, it is easy to calculate transmitted and reflected wave
functions:
eik↑ x eik↓ x 2k
ψtrans = √ cos(θ/2) |↑ + √ sin(θ/2) |↓
Ω Ω k + k↓
e−ikx k − k↓
ψref l = √ sin(θ/2) |↓ . (8.4)
Ω k + k↓
2
The spin flux Φ = (Φx , Φy , Φz ) is defined by Φ = 2m Im(ψ ∗ σ ⊗ ∇ψ)12 or
2
equivalently by:
2 dψ↑∗ dψ↓
Φ+ = Φx + iΦy = i ψ↓ − ψ↑∗
2m dx dx
2
dψ↑ dψ↓
Φz = Im ψ↑∗ − ψ↓∗ . (8.5)
2m dx dx
F1 N F2
& &
M1 Incoming el. M2
Transmitted el.
Reflected el.
ẑ ' V- ẑ
ŷ ' x̂' ŷ x̂
V+= 0
Fig. 8.2. Illustration of the quantum mechanics calculation of the spin transfer torque.
An incident electron polarized along zˆ , with an energy larger than the Stoner potential
step is partially transmitted as a precessing spin and partially reflected as a spin-down
electron.
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
178 C. Baraduc, M. Chshiev and U. Ebels
So using the wave-function defined in Eq. (8.3) and (8.4), the spin flux is
determined:
2
Φin = (k sin θx̂ + k cos θẑ)
2mΩ
2
Φtrans = k sin θ (cos [(k↑ − k↓ )x] x̂ − sin [(k↑ − k↓ )x] ŷ)
2mΩ 2
2 2 2k 2
+ k cos (θ/2) − k↓ sin (θ/2) ẑ
2mΩ k + k↓
2
2 k − k↓
Φref l = k sin2 (θ/2)ẑ . (8.6)
2mΩ k + k↓
Some points of physics can be illustrated by this example. First, when
the energy of the incident electron is larger than V− , spin flux is continuous
at the interface: Φin + Φref l = Φtrans at x = 0.
The second point is that when the incoming electron penetrates into
the ferromagnet, its spin-up and spin-down parts do not travel at the same
speed due to the difference between k↑ and k↓ . This induces a phase shift
between the spin-up and spin-down components of the wave-function. In
other words the electron spin precesses around the local field with a very
small spatial precession period (a few atomic spacings). So in metallic sam-
ples, when many electrons arrive at the interface from different directions,
they follow different paths into the ferromagnet and the spin precessions
of all these electrons rapidly average to zero due to classical dephasing.13
Then within about 1 nm, the spin flux transverse to ẑ has disappeared and
has been transferred to the local magnetization. In magnetic tunnel junc-
tions however, only the directions of incidence close to the perpendicular
are selected, since the effective barrier thickness is much larger for grazing
incidence. In that case, the classical dephasing between precessing electrons
is less efficient. In a pure ballistic case, a decaying oscillating spin flux is
expected.14
Finally, as shown in Fig. 8.2, it is important to notice that there is a
reflected spin current. In our case, it is due to the spin-down part of the
wave-function that is partially reflected on the potential step.b When these
reflected electrons will enter the first ferromagnetic layer F1 , they will start
to precess along the local field and finally transfer their transverse moment
to the local magnetization. So reflected electrons are responsible for the
torque on the first ferromagnetic layer F1 .
b Spin-down are traveling backwards, which is equivalent to spin-up traveling forwards.
So the positive sign of the reflected spin flux is correct.
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 179
We may summarize the situation as the following: the incoming elec-
trons arriving in F2 tend to tilt the magnetization toward their own spin
moment direction which is the direction of the magnetization of F1 . By
contrast, the reflected electrons which are mostly polarized antiparallel to
M 2 tend to tilt the F1 magnetization toward their own spin moment and
therefore away from the F2 magnetization. So the spin torque in both layers
has the same direction.
8.4. Transverse Spin Transfer Torque
In a ferromagnetic structure F1 /N/F2 , there is a magnetic coupling be-
tween the two ferromagnets that is mediated by the conduction electrons,
in absence of any bias current. This RKKY coupling is due to conduction
electrons below the Fermi energy which travel through the structure along
both directions.15,16 The coupling energy is Eex = J M 1 · M
2 . This energy
corresponds to a field hi = ∂Eex /∂ M
i ∝ M j (i = j) that exerts a torque
Ti⊥ on the magnetization Mi , in each ferromagnetic layer i = 1, 2:
i × hi ∝ −M
Ti⊥ = −γ0 M i × M
j (i = j) . (8.7)
Here the subscript ⊥ emphasizes the fact that this torque is transverse
to the plane that contains M 2 , contrary to the longitudinal spin
1 and M
transfer torque described in the previous sections. Let us also notice that
the torques on both magnetizations are equal and opposite: T1⊥ = −T2⊥ .
The same type of coupling also exists in magnetic tunnel junctions and
is called interlayer exchange coupling or conservative exchange coupling.17
This transverse torque, which exists in absence of bias current, can never-
theless be related to a spin current: electrons traveling from the right and
from the left electrodes produce no net charge current but a non-zero spin
current as soon as the magnetizations of the two ferromagnetic layers are
misaligned. When the structure is biased, this transverse torque increases.
The transverse torque exerted on the free layer magnetization is equal to:
× p̂
T⊥ = γ0 bj M (8.8)
where p̂ is the polarization of the fixed layer.
In the following, we present a few hints to understand the origin of this
transverse torque at the microscopic level. First, we reconsider the problem
we studied in the previous section of a spin polarized electron impinging
the N/F2 interface. Now we choose a more realistic description of the
ferromagnet, with two non-zero Stoner potential steps V+ and V− .
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
180 C. Baraduc, M. Chshiev and U. Ebels
F1 N F2 TA
Incoming el. T//
& &
M1 M2
Transmitted el.
ẑ
ẑ '
x̂' ŷ ' Reflected el.
x̂ ŷ
V’- V-
V’+ V+
Fig. 8.3. Illustration of a spin-polarized incident electron with an energy between the
two Stoner potential steps V+ and V− . The spin-down part of the wave function can only
penetrate into the ferromagnet as an evanescent wave; it is fully reflected. This process
induces a phase shift between the spin-up and spin-down part of the wave-function which
is equivalent to a rotation of the spin moment. Note that V+ (resp. V− ) is the Stoner
potential of spin-up (resp. spin-down) electrons defined relatively to the quantization
axis zˆ .
Let us consider an incident electron with an energy comprised between
V+ and V− . The spin-polarized electron wave function is a superposition
of a spin-up and a spin-down states. The spin-up part experiences a small
potential step so it is partially transmitted and partially reflected. The
spin-down part is fully reflected at the interface since the electron energy
cannot overcome the V− step. Only an evanescent spin-down wave enters
the ferromagnet. It can be demonstrated with simple quantum mechan-
ics that this process induces a phase shift between the spin-up and the
spin-down part of the reflected electron wave-function. This phase shift
corresponds to a rotation of the spin around ẑ-axis. It may be viewed as
a precession of the spin that enters the ferromagnet as a vanishing wave
before being reflected. Since the reflected electron gained some magnetic
moment along ŷ, it is responsible for a transverse component Φy of the re-
flected spin flux. Therefore, in a F1 /N/F2 structure, electrons indefinitely
reflected by the two N/F interfaces contribute to a transverse torque on
both ferromagnets. In the case studied here, let us note that the spin flux
may be considered as discontinuous when the evanescent contributions to
the wave function are neglected.
From the previous section we already know that electrons with an energy
larger than V− enter the ferromagnet with their spins precessing around
the local field. This induces an oscillatory spin flux on both x̂ and ŷ axis.
Due to the various directions of the electrons motion, classical dephasing
rapidly averages Φx and Φy close to the interface. A small amount of
average transverse flux may remain and also contribute to the transverse
torque. In metallic systems the transverse torque T⊥ does not exceed a
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 181
few percent of the longitudinal spin transfer torque T . In tunnel junctions
where wave vector close to the perpendicular incidence are selected, the
transverse torque is much larger.
In the particular case of magnetic tunnel junctions, spin currents in
the complete structure F1 /barrier/F2 can be calculated, thus leading to a
better understanding of spin transfer torque compared to the description
of a single interface. In the limit of a thick barrier and with use of WKB
approximation inside the latter, the spin torques can be written as follows:
↑ ↓ ↑2 ↓2 ↑ ↓
q0 qa (kL − kL )(kR − kR )(q02 − kL kL ) −2 a q(y)dy
T ∼ sin θ e 0 (fL − fR )
| Den |2
↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓
× (qa2 + kR kR ) sin[(kR − kR )y ] + qa (kR − kR ) cos[(kR − kR )y ]
T⊥ = T⊥0 + T⊥1
with
↑ ↓ ↑2 ↓2 a
q0 qa (kL −kL )(kR −kR ) −2
T⊥0 ∼ − sin θ |Den|2
e 2fR0
q(y)dy
↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓
× q0 qa (kL + kL )(kR + kR ) − (q02 − kL kL )(qa2 − kR kR ) cos[(kR + kR )y ]
↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓
− q0 (kL + kL )(qa2 − kR kR ) + qa (kR + kR )(q02 − kL kL ) sin[(kR + kR )y ]
↑ ↓ ↑2 ↓2 ↑ ↓
q0 qa (kL − kL )(kR − kR )(q02 − kL kL ) −2 a q(y)dy
T⊥1 ∼ − sin θ 2
e 0 (fL − fR )
| Den |
↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓
× qa2 + kR kR cos[(kR − kR )y ] − qa (kR − kR ) sin[(kR − kR )y ]
where
↑2 ↓2 ↑2 ↓2
| Den |2 = (q02 + kL )(q02 + kL )(qa2 + kR )(qa2 + kR );
q0 and qa are the values of the evanescent wave vector inside the barrier
↑
q(y) at the left (y = 0) and right interfaces (y = a) and y = y − a; kR
↓ ↑ ↓
and kR (resp. kL and kL ) are the wave-vectors of spin-up and spin-down
electrons in the right (resp. left) electrode. Again, these expressions are
valid when the exponential factors are small. It is interesting to note that
unlike T and T⊥1 which are proportional to a difference between the left
and right Fermi-Dirac functions (fL − fR ), T⊥0 exists even in the absence of
applied voltage and gives origin to the exchange coupling between the FM
electrodes.17 This term corresponds to the already mentioned interlayer
exchange coupling. Some more physical insights can be gained from those
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
182 C. Baraduc, M. Chshiev and U. Ebels
equations. As already observed in the single interface model, both T and
↑ ↓
T⊥1 spatially oscillate with an inverse period kR − kR , which corresponds
to the spin precession around the exchange field in the ferromagnetic layer.
This oscillation is due to interferences between transmitted (or equivalently
reflected) majority and minority wave functions. In contrast with that, T⊥0
↑ ↓
oscillates with a smaller period inversely proportional to (kR + kR ) which
originates from the interference between original waves incident from the
right and their reflected parts.14
Finally let us conclude on two remarkable properties of the transverse
torque which are detailed in a review paper by Ralph and Stiles.12 First,
the transverse torque is always equal and opposite on the two ferromagnets.
It can easily be understood by considering the incoming and outgoing mag-
netic moment in an imaginary box around the structure (see Fig. 8.4). Its
borders A and B are far from the interfaces with the spacer, so that at these
locations, the electrical current is perfectly spin polarized along the bulk
magnetization. Thus, there is no transverse moment coming in or going
out. Due to the conservation of angular momentum, the total transverse
torque must be zero: T1⊥ + T2⊥ = 0. The second interesting property is
that the transverse torque does not change sign when reversing the bias
current in a symmetric structure. The picture represents the transverse
torque when electrons flow from left to right (positive flow). The effect of
a flow from right to left on F2 can be imagined by rotating the system: it
is the same as the effect of the positive flow on F1 ; the torque still points
outwards. So, the transverse torque does not depend on the sign of the
bias in a symmetric structure. It is the reason why the transverse torque
is expected to depend only on V 2 in symmetric magnetic tunnel junctions.
However, deviations from such a behavior are expected when asymmetry
T1A T2A
F1 F2
A B
Electron flow
Fig. 8.4. The transverse torque is equal and opposite in the two ferromagnetic layers.
It does not depend on the current direction in the case of a symmetric device. The
torques here exist only on the surface of these thick ferromagnetic layers. Nevertheless
the same considerations still apply for thin ferromagnetic layers in a N/F1 /N/F2 /N
structure (cf 12 ).
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 183
is introduced which may be caused either by using different thicknesses
and/or type of FM layers14,18 or by disorder effects.12,19
8.5. Magnetization Dynamics
In this section we describe the effects of spin transfer torque on magneti-
zation dynamics in the macrospin approximation. The magnetization dy-
namics is described by the Landau-Lifschitz Gilbert (LLG) equation that in-
cludes the precession and the damping torques (first two terms of Eq. (8.9)).
In order to take the effect of spin transfer into account, the longitudinal
spin torque T is included to the LLG equation (third term of Eq. (8.9)),
thus leading to the Landau-Lifschitz-Gilbert-Slonczewski (LLGS) equation:
dM
= −γ0 M ef f + α M
×H × dM + γ0 aj M
× (M
× p) . (8.9)
dt Ms dt Ms
Here γ0 is the gyromagnetic ratio, α the natural damping constant, Ms
the saturation magnetization and H ef f the effective field which is given by
the variational derivative of the energy density with respect to M ,H ef f =
−∂E/∂ M. Including spin torque will change the static and dynamic states
of a given system. However it does not change the fundamental properties of
the magnetization dynamics. In order to see this, we will discuss the LLGS
equation in three steps: 1) Conservative dynamics including the precession
term only; 2) Damped dynamics including the precession and damping
term; 3) Spin transfer induced dynamics for the full LLGS equation.
Here, we will consider a thin film platelet (see Fig. 8.5), characterized
by a uniaxial shape anisotropy field H u and subjected to an external bias
field Hb , applied parallel to the shape anisotropy easy axis. The magnetic
energy of this system can be written as the sum of the anisotropy, exchange
and demagnetizating energies:
M s Hu
E= 1 − m2x − M H b + 2πMs2 m2z (8.10)
2
M
m
= = (mx , my , mz ) = (cos θ cos φ, cos θ sin φ, sin θ) .
Ms
In most cases the general solution of the LLG and LGGS equations
cannot be given analytically, in particular if one considers micromagnetic
configurations. Solution of LLG(S) then requires numerical approaches.13,20
However, in order to bring out the essential features of the magnetization
dynamics we will assume a homogenous magnetization configuration.
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
184 C. Baraduc, M. Chshiev and U. Ebels
=RXW SODQH
M
<
θ
ϕ ;
M
+X+E
Fig. 8.5. Definition of the (x, y, z) coordinate system and the spherical angles θ, φ.
Right: the free layer considered is an in-plane magnetized thin film platelet.
8.5.1. Conservative dynamics
The precession term is a conservative term that describes the precessional
motion of the magnetization along constant energy trajectories. More gen-
erally, the static and periodic solutions of the precession term are directly re-
lated to the underlying energy surface which is plotted in Fig. 8.6 as a func-
tion of the spherical coordinates (θ, φ) for the energy given by Eq. (8.10).
Obviously, there are three static states (defined by ∂ M /∂t = 0) which
are the in-plane energy minimum (along the in-plane easy axis), the out-
of-plane energy maximum (representing the demagnetization energy cost
when rotating M perpendicular to the film plane) and the saddle point (in-
plane hard axis). Preparing the system initially in one of these three static
states, the system does not change and remain in its initial state (constant
energy). When perturbing the system, the magnetization will follow period
orbits around either the energy minimum or the energy maximum as indi-
cated by the colored lines in Fig. 8.6(a). These periodic orbits are better
visualized in spherical coordinates, shown in Fig. 8.6(b). The precession
around the energy minimum is generally called in-plane precession (IPP)
since the energy minimum of a thin platelet is in-plane and the precession
around the out-of-plane energy maximum is called out-of-plane precession
(OPP).8,13,20,21
These trajectories require a few comments. 1) The IPP and OPP peri-
odic orbits (or constant energy trajectories) are completely defined by their
initial energy E0 or equivalently by their initial state (θ0 , φ0 ). 2) With
increasing initial energy, the amplitude of the clamshell shaped IPP orbits
increases. At the saddle point, the IPP orbit transforms into an OPP orbit
whose amplitude, i.e. the out-of-plane magnetization component increases
while at the same time the diameter decreases. 3) Due to the non-linearity
of the equation of motion, there is a strong dependence of the precession
frequency on the precession amplitude, which is different for the IPP and
OPP orbits.8,13 For IPP orbits, the frequency decreases with increasing
amplitude since the length of the orbit increases faster than the local ve-
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 185
(a) (b) In-Plane Out of Plane
Energy Maximum
Energy maximum Precession IPP Precession OPP
max max
Energy (a.u.)
Energy Minimum
Energy minimum E
OPP Eo
o
Saddle point mz
Saddle Point mz
my
m⊥ min
IPP FMR m min
my
x mx
90° Mo M Oscillation around Oscillation around
WKHWD
(c) energy minimum energy maximum
θ
IPP OPP
Frequency (GHz)
80° Spiral Orbit
° 1
0SKL 30
Limit Cycle 0° ° ϕ Damped case
-180 FMR
Under STT 20 I = γ π0 V P]
Stable periodic orbits
Conservative case 10
0,0 0,2 0,4 0,6 0,8 1,0
Emz
o/Emaxmax
Fig. 8.6. (a) Energy surface as a function of (θ, φ) for a thin film platelet. In this exam-
ple, the bias field amplitude Hb is larger than the uniaxial in-plane anisotropy field Hu ,
so that there is only one in-plane energy minimum at 0. The three stationary states (en-
ergy minimum, maximum and saddle point) are indicated. The IPP and OPP constant
energy trajectories are indicated by the colored lines on top of the energy surface. The
white spiral indicates the damped motion. Below the energy surface from left to right
are schematically indicated a limit cycle, a constant energy trajectory (stable orbit) and
a spiral orbit. (b) The IPP and OPP constant energy trajectories as a function of the
magnetization components mx , my , mz . (c) The frequency vs. precession amplitude for
the IPP and OPP orbits.
locity. Close to the saddle point, the local velocity goes to zero, explaining
the zero frequency at the transition from IPP to OPP. For the OPP orbit
the frequency is given by the out-of-plane demagnetization field which is
proportional to the out-of-plane magnetization component mz .
8.5.2. Damped dynamics
Adding the damping term to the precession term will change the stability
of the three static solutions. Now, only the energy minimum is a stable
solution, while the energy maximum and the saddle point correspond to
unstable solutions. This means that they are not stable against small per-
turbations, which will relax the magnetization on a spiral orbit away from
the energy maximum or the saddle point towards the stable energy min-
imum. On the other hand, perturbation of the system around the stable
energy minimum leads to small amplitude damped oscillations of the form
m⊥ = m0 e−λt e−iωt where m⊥ is a small transverse magnetization com-
ponent, ω is the oscillation frequency and λ the damping rate (λ > 0).
These small amplitude oscillations correspond to solutions of the linearized
LLG equation around the static state M 0 using M
=M 0 +m ⊥ . These os-
cillations can be typically excited in ferromagnetic resonance experiments
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
186 C. Baraduc, M. Chshiev and U. Ebels
J>0
(a)
planar perpendicular (b)
Free Layer
&X &X Spin Torque
mz
mz
Polarizer
Humin//
, Hb
ϕο
θo=90° ϕο
Hu, Hb
y
my
m
mm
x
Precession Torque
x
Fig. 8.7. (a) Schematics of a spin valve structure with a planar free layer and a planar
(left) or perpendicular (right) polarizer. (b) In-plane rotation φ0 of the magnetization
for the perpendicular polarizer configuration.
(FMR) using a small radiofrequency pumping field.22 Let us note that the
same stability analysis around the energy maximum would give a negative
damping rate λ < 0, indicating instability.
8.5.3. Spin transfer torque induced dynamics
The static and dynamic properties of the free layer magnetization including
the spin torque term will be discussed here for two different configurations
of the fixed layer, called planar and perpendicular polarizer, see Fig. 8.7(a).
8.5.3.1. Static states
As compared to the conservative and the damped case, the static (stable
or unstable) states are not necessarily identical to the stationary states
defined in Fig. 8.6. Following the LLGS equation, a static state is now ob-
tained when the precession torque is equal and opposite to the spin torque,
by contrast to previous cases where the precession torque is zero. Fortu-
nately, in the planar case, the spin polarization p̂ is collinear to the effective
field H ef f of the free layer energy minimum. Thus the precession and spin
torques are both zero when the magnetization is directed along its energy
minimum (p̂//H ). In the case of the perpendicular polarizer, p̂ and
ef f //M
ef f are not collinear. Spin torque tends to push the magnetization M
H par-
allel to p̂, i.e. out-of-plane. A static state is obtained when the precession
torque counterbalances this out-of-plane spin torque: the magnetization
then rotates in-plane away from the energy minimum, see Fig. 8.7(b). In
this case, one can easily verify that M ×H ef f is pointing out-of-plane since
Hef f = Hb + Hu is oriented parallel to the x-axis. The corresponding ro-
tation angle φ0 increases with current and is negative for negative current
and positive for positive current.23
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 187
8.5.3.2. Stability
We may now wonder whether the above mentioned static states are sta-
ble. To answer this question, the dynamics equation is linearized around
the calculated static states. This procedure leads to solutions of the kind
m⊥ = m0 e−λt e−iωt .21,23–26 The damping rate depends now on the current
density, λ = λ(J) and can be either positive or negative depending on J.
This means that the motion is either damped (λ(J) > 0) and the static state
is stable,c or excited away from the static state (λ(J) < 0) and the static
state is unstable. Thus upon increasing the current density from zero, an
initially stable state can become unstable. The current value for which this
occurs is called the critical current Jc and is given by λ(Jc ) = 0. When the
initial static state becomes unstable, the magnetization can transit either
into another static state or into a dynamic state. For the planar and per-
pendicular polarizers the instability criteria describing the transition from
the static state to a dynamic state are quite different. In the following we
illustrate some of the general features of the dynamic state. For simplicity
we will discuss in more detail only the planar configuration.
8.5.3.3. Dynamic states
At zero or low current, when λ(J) > 0, a damped motion of magnetization
is observed under spin torque. In fact, the energy loss through the natu-
ral damping is only partially compensated by the energy gained from spin
transfer. Once the current is strong enough, the energy gain through the
spin transfer overcomes the energy loss. Magnetization then reaches a limit
cycle where the energy gained from the spin torque during each cycle of
precession is balanced by the energy lost to damping.13,25 A schematic of
such a limit cycle is shown at the bottom in Fig. 8.6(a). It is important
to point out here, that the stabilization of the limit cycle at finite am-
plitude is a consequence of the non-linear dynamical system described by
LLGS.21,26,28 Non-linear means here that the spin torque and the damping
torque depend on the precession amplitude. If this were not the case, then
once the spin torque overcomes the damping torque at the static state, no
condition for a stable periodic orbit could be realized.
Due to the compensation of the damping losses by spin torque, it seems
at hand to suggest that the limit cycles are identical to the (zero current,
c Inthat case, the equilibrium state at zero temperature is unchanged compared to the
zero current case. However, at finite temperature, a change of the thermal magnetization
fluctuations is observed due to the evolution of the damping rate.27
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
188 C. Baraduc, M. Chshiev and U. Ebels
zero damping) constant energy trajectories shown in Fig. 8.6(a). This sug-
gestion is supported by the fact that both the spin torque and the damping
torque are small compared to the precession torque and should thus be
treated as perturbations.13,21,25 It is fortunate that this suggestion holds
well for the case of the planar and the perpendicular polarizer configura-
tion, due to their special symmetry with respect to the energy minimum
and maximum. However, in the general case for an arbitrary spin polariza-
tion direction, limit cycles can be stabilized that are quite different from
constant energy trajectories. Furthermore, the amplitude of a constant en-
ergy trajectory is given by its initial condition (i.e. initial energy), while
the amplitude of a limit cycle is defined by the balance between damping
and spin torque (energy feedback) and thus by the current amplitude. Note
also that the limit cycle is independent of its initial condition. As shown in
the schematics of Fig. 8.6(a), preparing the system at any point inside (or
outside) the limit cycle will produce a transient spiral motion of the mag-
netization until it reaches the stable limit cycle. To conclude, the planar
polarizer can support limit cycles that are very close to IPP orbits and for
larger current that are close to OPP orbits.8,13,25 In contrast a perpendicu-
lar polarizer only supports limit cycles that are very close to OPP orbits23
since the perpendicular polarizer does not have the symmetry of the energy
minimum or of IPP trajectories and therefore cannot support limit cycles
of this type. Finally, for a conservative trajectory there is no change of the
energy with time, while for the spin current induced limit cycles the energy
oscillates around a constant energy value.20 To conclude, due to this cor-
respondence between limit cycles and IPP and OPP orbits, the frequency
of the limit cycles as a function of current follows the dependence given in
Fig. 8.6(c), where the horizontal axis has to be replaced by the current.
This property is another manifestation of the non-linearity of the LLGS
equation.
8.6. State Diagram
In this part we will describe the state diagrams in the case of the planar
and perpendicular polarizers. State diagrams define the different static and
dynamic states in the current-field plane. As indicated above, the boundary
between a static and a dynamic state can be obtained by finding the zeros
of the damping rate λ(J).
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 189
Jc1 Jc2
OPP
4
3 P IPP
Hb>Hu
OPP
FLFL 3
IPP
2 &X
1 minimum
Pol
Pol
1 Hb<Hu IPP
Hb/Hu
0 3$3 2 minima
-1 P AP
-2 AP J~aJ>0 unstable stable
OPP
IPP
J~aJ<0 stable unstable
-3 $3 &X
-4
-1.5 -1.0 -0.5 0 0.5 1.0 1.5
aJ/(Msα)
Fig. 8.8. Left: State diagram for the planar polarizer as a function of the normalized
bias field Hb /Hu and the current density J. The static states correspond to the parallel
(P) and antiparallel (AP) orientation of the free layer magnetization relatively to the
polarizer. The critical current separating the static state and the IPP limit cycles is given
by Jc1 and the critical current separating IPP from OPP limit cycles is given by Jc2 .
Right: Schematic indication of the different states as a function of the energy vs.in-plane
angle φ.
8.6.1. Planar polarizer
The full diagram for the planar polarizer configuration is shown in
Fig. 8.8.8,13 This diagram is asymmetric as a function of the sign of current
and field. The spin torque acts such that it either enhances the damping
rate λ(J) for one sign of the current and thus stabilizes the initial state, or
for the opposite current sign it counteracts natural damping and decreases
λ(J) thus destabilizing the initial state. In that case the damped oscilla-
tions become less and less damped until λ(J) changes sign at the critical
current Jc1 . The static state then becomes unstable and gives way to IPP
limit cycles. Upon further increasing the current one has to distinguish two
regions for which the bias field Hb is either smaller (two in-plane energy
minima) or larger than the anisotropy field Hu (only one energy minimum).
When Hb < Hu , for increasing current density J, the IPP orbits pass the
energy saddle point (in-plane hard axis) and the magnetization transits into
the AP state. This essentially means that one can switch the magnetization
orientation using a spin polarized current. In the case of Hb > Hu , there
does not exist a second energy minimum and the magnetization therefore
transits from IPP to OPP limit cycles at the saddle point. The correspond-
ing critical current Jc2 is shown in Fig. 8.8. It is calculated by considering
that the total energy gain over one precession period is zero.13,25 It should
be noted that for Hb > Hu (resp. Hb < −Hu ) and very large positive (resp.
negative) currents, the magnetization eventually stabilizes in a static state.
However such high current densities cannot be reached experimentally.
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
190 C. Baraduc, M. Chshiev and U. Ebels
J
Jc2
c2 Jc1
c1
J
Jc2
c2 OPP OPS
IPS 0°
Hb (Oe)
θo
OPP
OPP
OPS
OPS
ϕo≈180°
mz z
mz z
m
m
IPS 180° HH u,
u, H
Hb
IPS b
mmyy
IPS
y
y
m
m
J (1011 A/m²) m
mx m
m
xx
x
Fig. 8.9. State Diagram for the perpendicular polarizer, with three states: IPS (in-plane
stable state), OPP (out-of-plane limit cycle) and OPS (out-of-plane stable state).
8.6.2. Perpendicular polarizer
For completeness, we briefly discuss the diagram for the perpendicular po-
larizer in Fig. 8.9.23 In this case (i) the static states correspond to an
in-plane rotation and (ii) only OPP limit cycles are supported. Here the
critical current Jc1 is reached when the spin torque becomes larger than the
precession torque.23 This transition from the in-plane static (IPS) state to
the OPP limit cycles is accompanied by a relatively abrupt jump of the
out-of-plane magnetization component mz , see Fig. 8.9. By contrast to the
planar case, the regions of OPP limit cycles are symmetric as a function of
current and field. For very large currents, the free layer magnetization M
is rotated either parallel (J < 0) or antiparallel (J > 0) to the polarizer
direction p̂ described as the out-of-plane stable (OPS) state in the diagram
of Fig. 8.9. Finally, let us compare the effects of spin torque in the case
of the planar and perpendicular polarizer. The planar polarizer has the
symmetry of the energy minimum and can thus destabilize the in-plane
static state when the spin torque is larger than the damping torque. For
the perpendicular polarizer, the role of spin torque and damping torque
are inverted, since the polarizer has the symmetry of the energy maximum.
Thus the spin torque stabilizes the out-of-plane state, while it is the damp-
ing torque that destabilizes it. Thus the OPS state becomes unstable when
the current is smaller than a certain critical current Jc2 .23,26
8.7. Conclusions
At large enough current density, spin polarized electrons exert a significant
torque on the local magnetization. This spin transfer torque corresponds
to a direct transfer of magnetic moment from the conduction electrons to
the local magnetization. This phenomenon strongly affects the magnetiza-
tion dynamics. When the applied current it large enough, spin torque is
responsible either for magnetization reversal or steady-state precession.
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
Introduction to Spin Transfer Torque 191
Acknowledgments
It is a pleasure to acknowledge our colleagues, S. Petit, N. de Mestier, D.
Houssameddine, C. Thirion, I. Firastrau, D. Gusakova, L. Buda-Prejbeanu,
B. Diény, A. Vedyayev. One of us (C.B.) thanks also M. Stiles for helpful
discussions. This work has been supported by the French National Research
Agency ANR (SpinChat, MagICO) and the by OSEO/Anvar.
References
1. L. Berger, Emission of spin waves by a magnetic multilayer traversed by a
current, Phys. Rev. B. 54(13), 9353–9358 (Oct., 1996).
2. J. C. Slonczewski, Current-driven excitation of magnetic multilayers, J.
Magn. Magn. Mater. 159, 1–7 (Jun., 1996).
3. M. Tsoi, A. G. M. Jansen, J. Bass, W. C. Chiang, M. Seck, V. Tsoi, and
P. Wyder, Excitation of a magnetic multilayer by an electric current, Phys.
Rev. Lett. 80(19), 4281–4284 (May, 1998).
4. J. A. Katine, F. J. Albert, R. A. Buhrman, E. B. Myers, and D. C. Ralph,
Current-driven magnetization reversal and spin-wave excitations in co/cu/co
pillars, Phys. Rev. Lett. 84(14), 3149–3152 (Apr., 2000).
5. J. Grollier, V. Cros, A. Hamzic, J. M. George, H. Jaffres, A. Fert, G. Faini,
J. B. Youssef, and H. Legall, Spin-polarized current induced switching in
co/cu/co pillars, Appl. Phys. Lett. 78(23), 3663–3665 (Jun., 2001).
6. Y. Huai, F. Albert, P. Nguyen, M. Pakala, and T. Valet, Observation of
spin-transfer switching in deep submicron-sized and low-resistance magnetic
tunnel junctions, Appl. Phys. Lett. 84(16), 3118–3120 (Apr., 2004).
7. G. D. Fuchs, N. C. Emley, I. N. Krivorotov, P. M. Braganca, E. M. Ryan, S. I.
Kiselev, J. C. Sankey, D. C. Ralph, R. A. Buhrman, and J. A. Katine, Spin-
transfer effects in nanoscale magnetic tunnel junctions, Appl. Phys. Lett. 85
(7), 1205–1207 (Aug., 2004).
8. S. I. Kiselev, J. C. Sankey, I. N. Krivorotov, N. C. Emley, R. J. Schoelkopf,
R. A. Buhrman, and D. C. Ralph, Microwave oscillations of a nanomagnet
driven by a spin-polarized current, Nature. 425(6956), 380–383 (2003).
9. W. H. Rippard, M. R. Pufall, S. Kaka, S. E. Russek, and T. J. Silva, Direct-
current induced dynamics in co90 f e10 /ni80 f e20 point contacts, Phys. Rev.
Lett. 92(2), 027201 (Jan., 2004).
10. J. C. Slonczewski, Currents, torques, and polarization factors in magnetic
tunnel junctions, Phys. Rev. B. 71(2), 024411 (2005).
11. J. C. Slonczewski, Currents and torques in metallic magnetic multilayers,
J. Magn. Magn. Mater. 247(3), 324–338 (Jun., 2002).
12. D. Ralph and M. Stiles, Spin transfer torques, J. Magn. Magn. Mater. 320
(7), 1190–1216 (Apr., 2008).
13. M. Stiles and J. Miltat, Spin transfer torque and dynamics. vol. Spin Dynam-
ics in Confined Magnetic Structures III (Springer Berlin/Heidelberg, 2006).
March 26, 2010 18:36 World Scientific Review Volume - 9in x 6in chapter8
192 C. Baraduc, M. Chshiev and U. Ebels
14. A. Manchon, N. Ryzhanova, A. Vedyayev, M. Chschiev, and B. Dieny, De-
scription of current-driven torques in magnetic tunnel junctions, J. Phys.:
Condensed Matter. 20(14), 145208 (2008). ISSN 0953-8984.
15. M. D. Stiles, Exchange coupling in magnetic heterostructures, Phys. Rev. B.
48(10), 7238 (Sept., 1993).
16. J. C. Slonczewski, Overview of interlayer exchange theory, J. Magn. Magn.
Mater. 150(1), 13–24 (Sept., 1995).
17. J. C. Slonczewski, Conductance and exchange coupling of two ferromagnets
separated by a tunneling barrier, Phys. Rev. B. 39(10), 6995–7002 (Apr.,
1989).
18. M. Wilczynski, J. Barnas, and R. Swirkowicz, Free-electron model of current-
induced spin-transfer torque in magnetic tunnel junctions, Phys. Rev. B. 77
(5), 054434–9 (Feb., 2008).
19. Z. Li, S. Zhang, Z. Diao, Y. Ding, X. Tang, D. M. Apalkov, Z. Yang, K. Kawa-
bata, and Y. Huai, Perpendicular spin torques in magnetic tunnel junctions,
Phys. Rev. Lett. 100(24), 246602–4 (Jun., 2008).
20. Z. Li and S. Zhang, Magnetization dynamics with a spin-transfer torque,
Phys. Rev. B. 68(2), 024404 (Jul., 2003).
21. A. N. Slavin and V. S. Tiberkevich, Current-induced bistability and dynamic
range of microwave generation in magnetic nanostructures, Phys. Rev. B. 72
(9), 094428–5 (Sept., 2005).
22. L. Baselgia, M. Warden, F. Waldner, S. L. Hutton, J. E. Drumheller, Y. Q.
He, P. E. Wigen, and M. Maryko, Derivation of the resonance frequency from
the free energy of ferromagnets, Phys. Rev. B. 38(4), 2237 (Aug., 1988).
23. U. Ebels, D. Houssameddine, I. Firastrau, D. Gusakova, C. Thirion, B. Di-
eny, and L. D. Buda-Prejbeanu, Macrospin description of the perpendicu-
lar polarizer-planar free-layer spin-torque oscillator, Phys. Rev. B. 78(2),
024436–16 (Jul., 2008).
24. J. Grollier, V. Cros, H. Jaffrès, A. Hamzic, J. M. George, G. Faini,
J. Ben Youssef, H. Le Gall, and A. Fert, Field dependence of magnetiza-
tion reversal by spin transfer, Phys. Rev. B. 67(17), 174402 (May, 2003).
25. G. Bertotti, C. Serpico, I. D. Mayergoyz, A. Magni, M. d’Aquino, and
R. Bonin, Magnetization switching and microwave oscillations in nanomag-
nets driven by spin-polarized currents, Phys. Rev. Lett. 94(12), 127206–4
(Apr., 2005).
26. A. Slavin and V. Tiberkevich, Excitation of spin waves by spin-polarized
current in magnetic nano-structures, IEEE Trans. Magn. 44(7), 1916–1927
(2008). ISSN 0018-9464.
27. S. Petit, C. Baraduc, C. Thirion, U. Ebels, Y. Liu, M. Li, P. Wang, and B. Di-
eny, Spin-torque influence on the high-frequency magnetization fluctuations
in magnetic tunnel junctions, Phys. Rev. Lett. 98(7), 077203 (2007).
28. S. M. Rezende, F. M. de Aguiar, and A. Azevedo, Spin-wave theory for the
dynamics induced by direct currents in magnetic multilayers, Phys. Rev. Lett.
94(3), 037202–4 (Jan., 2005).
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Chapter 9
SPINTRONICS POTENTIAL OF
RARE-EARTH NITRIDES
Ben J. Ruck
School of Chemical and Physical Sciences
Victoria University of Wellington
P.O. Box 600
Wellington, New Zealand
E-mail: [email protected]∗
The rare-earth nitrides are an interesting class of materials that have re-
cently received attention due to their strong coupling between spin and
charge degrees of freedom. This Chapter presents a review of our cur-
rent understanding of these materials. Various means have been devised
to produce rare-earth nitrides in both bulk and thin film form, but at
present the materials still suffer from rather large defect concentrations
that inhibit the study of their fundamental properties. Their electronic
and magnetic properties have been studied both theoretically and exper-
imentally. It is now clear that some members of the series are intrinsic
ferromagnetic semiconductors, while others may be half-metals in their
ground state. Their magnetic interactions remain less well understood
and present a fascinating avenue for further study. The present state
of knowledge is used to formulate a discussion of the rare-earth nitrides
potential for use in spintronics devices.
9.1. Introduction
The rapidly developing field of spintronics, in which the spin degree of free-
dom is utilised as well as the electric charge, has generated substantial in-
terest in materials that show strong coupling between their electronic and
magnetic properties.1–3 The most striking example is the magnetic read
head found in all modern hard disk drives, which is now based on the tun-
neling magnetoresistance phenomenon,4 and which has lead to enormous
∗ Also at MacDiarmid Institute for Advanced Materials and Nanotechnology, Victoria
University of Wellington.
193
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
194 B. J. Ruck
increases in magnetic storage capacity. But there also exists the possibil-
ity of even more advanced spintronics devices. Numerous proposals have
been put forward for creating magnetic random access memories with rapid
read/write times and low power consumption,2,3 in which the information
is retained even after the device is powered down. The first such devices
are now appearing on the market.5 Other possibilities include devices in
which the electron spin plays either an active or a passive role, such as spin
field effect transistors, spin-logic devices, novel polarised photon sources, or
even quantum computers.3,6
The giant magnetoresistance and tunneling magnetoresistance devices
in hard disks are based around conventional materials such as iron and mag-
nesium oxide. Rapid uptake of spintronics technology will be aided if more
novel materials can be developed that show the strong spin-charge coupling
required to effectively manipulate the electron spins. The most intensely
studied class of novel materials for spintronics are dilute magnetic semi-
conductors (DMS) in which magnetic elements are added to conventional
semiconductors to induce ferromagnetism.7 GaAs doped with a few percent
manganese is known to become ferromagnetic, but only at temperatures
below8 about 170 K. There are numerous claims of room temperature fer-
romagnetism in other DMS systems, such as Gax Mn1−x N,9 Gax Gd1−x N10
or transition metal doped ZnO,11 but these claims are complicated by the
lack of a strong theoretical underpinning and issues associated with nano-
sized impurity phases. Furthermore, the application of these materials is
somewhat limited because the magnetic elements heavily dope the semi-
conductor preventing independent control of the carrier concentration, and
their main proposed use is as spin injection layers coupled to conventional
semiconductor devices.
Against this background it is clearly interesting to consider materials
that are both intrinsically semiconducting and magnetic. In principle it is
then possible to control the carrier concentration through doping indepen-
dent of the magnetism, opening new possibilities for spintronics applications
and for fundamental spin-transport research. At present only a few intrin-
sic ferromagnetic semiconductors are known. They include the first fer-
romagnetic semiconductor discovered, CrBr3 , the europium chalcogenides
EuO and EuS, which are perhaps the most studied, a few chalcogenide
spinels some of which have Curie temperatures near room temperature
(e.g., CdCr2 S4 with Tc of 97 K or CuCr2 Te3 I with Tc of 294 K), halides
such as DyF3 or GdCl3 , and perovskite oxides such as EuLiH3 .12,13 The
parent compounds of the manganite perovskites that show colossal mag-
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 195
netoresistance (CMR) are semiconducting, but they are antiferromagnetic,
only displaying ferromagnetism and the CMR effect upon doping to degen-
eracy.13
A particularly interesting class of materials that contain intrinsic ferro-
magnetic semiconductors are the rare-earth nitrides (RE-Ns), binary alloys
of a rare-earth element and nitrogen with formula RN (R = rare-earth)
which crystallise in the simple rock-salt structure. Early investigations of
these compounds were performed in the 1960s and 1970s,14,15 but with-
out significant focus on their spin-charge coupling. The advent of mod-
ern band-structure calculation techniques has seen a renewal of interest in
the RE-Ns, driven by predictions that the series includes both magneti-
cally ordered half-metals and semiconductors, with potential application in
spintronics.16–19 This has been coupled with renewed experimental atten-
tion aimed at addressing contradictions amongst the theoretical predictions.
Many outstanding questions remain regarding the electronic structure and
the nature of the magnetic interactions, but the combination of theory and
experiment is beginning to present a clear picture of some of the RE-Ns
more fundamental properties.
This review will focus on recent developments in the study of rare-
earth nitrides, especially the experimental aspects. An excellent review
covering more of the theoretical aspects can be found in Ref. 19. The
preparation of rare-earth nitrides will be discussed in Section 9.2, followed
by a review of the theoretical and experimental investigations of their band
structures in Sections 9.3.1 and 9.3.2. Section 9.4 describes their observed
magnetic ordering and models of the exchange mechanisms responsible for
the ordering. The chapter is concluded in Section 9.5 with a discussion of
the potential for these materials to be used in spintronics devices.
9.2. Rare-Earth Nitride Preparation
The difficulty of preparing rare-earth nitride samples with the correct stoi-
chiometry, low defect density, and low impurity concentration is one of the
most significant challenges that must be overcome if a full understanding
of their properties is to be achieved. In particular, their rapid reaction
with atmosphere means that any ex-situ studies must be performed with
the samples protected from exposure to air. A range of techniques have
been applied to rare-earth nitride preparation, resulting in varying levels of
sample quality.
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
196 B. J. Ruck
A significant amount of work was performed in the 1960s and 1970s
aimed at determining the bulk magnetic properties of the RE-Ns, for which
samples were prepared in both thin film and bulk form. Bulk samples can
be prepared by high temperature reaction of the lanthanide metal with
either nitrogen gas or ammonia, or through chemical pathways such as the
decomposition of lanthanide amides (e.g., RE(NH2 )3 ). Early attempts were
also made at thin film growth, with reported epitaxial layers prepared on
heated substrates by decomposition of a volatile lanthanide chloride with
ammonia. Films were also prepared by sublimation of fine powders. A
summary of these early preparation techniques can be found in Ref. 14.
Perhaps the highest quality samples studied to date have been a series of
YbN single crystals grown by Degiorgi et al.20 in 1990. Considerable care
was taken to avoid exposing the crystals to atmosphere, with all experimen-
tal characterisations being carried out after transferring the samples only
via an argon-filled glove box. This allowed them to undertake the most com-
prehensive study of well-characterised RE-N samples yet reported. Various
stoichiometries were achieved, ranging from YbN0.96 to fully stoichiomet-
ric YbN (within ±0.5%) as determined by a micro Kjeldahl method. The
stoichiometric sample had a lattice constant of 4.781 Å, which is the low-
est value reported for this material. The YbN0.96 sample had a slightly
expanded lattice constant of 4.784 Å, with the increase being comparable
to the 0.005 Å increase for 5% off-stoichiometry reported as typical for
the RE-Ns.21 The stoichiometric material has a lower conductivity than
off-stoichiometric material which is consistent with the expectation that
nitrogen vacanices add free electrons to the system (YbN was found here
to be semi-metallic).
More recent attempts to prepare rare-earth nitride samples for study
have returned focus to thin film growth, which is the likely form that would
be required for technological applications of these materials. This has been
encouraged by the rapid improvements in nitride growth technology that
have accompanied the rise of gallium nitride as a wide-band-gap electronic
material, including the development of nitrogen plasma sources for molec-
ular beam epitaxy growth.
Leuenberger et al. have successfully prepared GdN films on silicon [100],
sapphire [1120], kapton, or mylar substrates using reactive ion-beam sput-
tering in a partial pressure of 2 − 3 × 10−5 mbar N2 , with the substrate
chosen to suit the subsequent characterisation method.22 Reactive nitrogen
was supplied by an ECR plasma source, and the growth temperature was
varied from room temperature to 700◦C. For ex situ studies capping layers
of 50 Å Al or W were deposited, and a thin buffer layer of Cr or W was also
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 197
deposited onto the silicon substrates prior to film growth to prevent reaction
of the film with the substrate. Films grown above 400◦ C are single phase
polycrystalline GdN with crystallite diameter of approximately 10 nm and
a lattice constant of 5.05 Å, slightly larger than the bulk value. They are
stoichiometric and free of oxygen to within the uncertainty of the resonant
nuclear reaction analysis and X-ray photoelectron spectroscopy techniques
used to characterise them.
One of the simplest preparation techniques involves evaporation (either
thermal or electron beam) of the rare-earth element in a background partial
pressure of approximately 10−4 mbar pure N2 gas onto substrates held at
room temperature.23 Surprisingly, even in the absence of any excitation
source for the nitrogen, the depositing rare-earth film reacts with the N2
to form a nitride. After growth the films were capped with a layer of GaN
grown by ion-assisted deposition. Thus far this has been demonstrated for
GdN,23 SmN, DyN,24 ErN, and LuN,25 using substrates of sapphire [100],
silicon [100] or [111], or YSZ [100]. The stoichiometry of films thus prepared
was investigated using Rutherford backscattering spectroscopy (RBS) and
X-ray diffraction. RBS results indicated the films consist of equal atomic
concentrations of rare-earth and nitrogen to within the uncertainty of a
few percent, while the lattice constants measured by XRD are in agree-
ment with previously reported values. A Scherrer analysis of the xrd peak
widths indicates a small crystallite size of ≈ 10 nm, with evidence of weak
[111] texturing in all materials except for SmN which appears to be fully
randomly oriented. Further evidence for the stoichiometric nature of the
films is obtained from the value of 70 K measured for the Curie temperature
of GdN, which is in agreement with other reported values for stoichiometric
GdN. Growth of GdN under similar conditions but using nitrogen excited
by an ion-source produced films with smaller crystallites (≈ 3 nm) and with
a lattice constant expanded by 2.4%.26
Clearly some minimum nitrogen pressure must be supplied to grow RE-
Ns using this technique. The pressure dependence of the films’ stoichiome-
try has been investigated by measuring the resistance of films during growth
on substrates onto which electrical contacts had first been deposited.23 For
growth at 10−4 mbar the resistance falls in inverse proportion to the film
thickness, as expected for a uniform layer, with a resistivity of 0.25 µΩcm
characteristic of a semiconducting material (see Section 9.3 below for fur-
ther discussion of the electronic structure). However, when the nitrogen
pressure was steadily lowered during growth the conductivity steadily in-
creased. The overall conductance S of the film is related to the conductivity
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
198 B. J. Ruck
Fig. 9.1. The conductivity of a GdN film measured as a function of N2 partial pressure
during growth. Below about 10−5 mbar the GdN becomes metallic, while at higher
pressures the film acts as a heavily doped semiconductor, with nitrogen vacancies the
likely dopants (from Ref. 23).
σ(z) at depth z by
W
S= σ(z)dz,
L
L dS
σ(z) = ,
W dz
where W is the width of the film and L is the length between the contact
pads. The conductivity thus extracted is plotted as a function of growth
pressure in Fig. 9.1. The clear implication is that with decreasing nitrogen
pressure the number of nitrogen vacancies increases, with evidence for a
transition to a metallic state for growth pressures below ≈ 10−5 mbar.
The simplicity of this technique has been exploited to perform in situ
X-ray absorption and emission studies of SmN and DyN as described be-
low, where the films were grown in a preparation chamber attached to the
synchrotron beamline and then transferred to the measurement chamber
under vacuum without exposing them to atmosphere.24
More recently Gerlach et al.27 have demonstrated epitaxial growth of
GdN by molecular beam epitaxy onto [100] MgO substrates at a temper-
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 199
ature of 750◦C. A hollow anode plasma-beam source was used to provide
the nitrogen. Reflection high-energy electron diffraction proved the epitax-
ial nature of the films, which remained smooth for thicknesses up to 17 nm,
but which became three-dimensional at greater thicknesses. The crystalline
quality of these films, as determined by XRD, was found to be higher than
for any other reported films, although further optimisation of the growth
parameters is still required to reduce the degree of twinning and mosaic
spread. Epitaxial films of GdN, SmN, and LuN have also been produced
recently using pulsed laser deposition, although these tend to suffer from
the presence of metallic nanoparticulates produced by the laser ablation.28
An optimised growth temperature window around 500◦ C was found for
SmN, while for GdN and LuN no clear temperature dependence of the film
quality was found between 500 and 850◦C. This is presumably related to
the much higher vapour pressure of Sm metal at these temperatures.
Given the dramatic enhancements in film quality that have been
achieved with more conventional nitrides in recent years the prospects for
improving rare-earth nitride film growth appear very strong. A more de-
tailed exploration of the parameter space, including choice and prepara-
tion of substrate material, growth temperature, the nature of the nitrogen
source, and the rare-earth to nitrogen flux ratio, should enable significant
improvements in crystal quality and a reduction in defect density. Alter-
native growth techniques, such as chemical vapour deposition, may also
prove fruitful in the search for better quality films. Of particular impor-
tance will be an improved understanding of the dynamics of the nitrogen
incorporation, as nitrogen vacancies are an important source of free carriers
that influence the material properties. At present no deliberate attempts to
dope RE-N films have been made, but as we show below the band structure
of these materials allows the possibility of doping both n- and p-type if the
residual doping levels can be reduced sufficiently.
9.3. Electronic Structure
9.3.1. Band structure calculations
Until recently there was little reliable experimental information regarding
the electronic structure of the rare-earth nitrides, and fundamental ques-
tions such as whether they are metals or insulators in their ground state
remained unanswered. In a simple ionic model one would expect the rare-
earth ion to be in the +3 charge state (with the exception of tetravalent
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
200 B. J. Ruck
Table 9.1. Physical and magnetic properties of the rare-earth nitrides. Both anti-
ferromagnetic and ferromagnetic ground states have been attributed to SmN, TbN,
and DyN, as indicated by the inclusion of both measured Curie (Tc ) and measured
Néel (TN ) temperatures. The quoted conductivities are for the ground state, where
SM = semimetallic, SC = semiconducting, HM = half metallic, M = metallic. Ra-
dioactive promethium (Pm) is omitted.
RE-N r Magn. order Tc /TN Conduct. Conduct.
(Å) (expt.) (K) (theory) (expt.)
LaN 5.30 none15 - SC17,29 SC14,30
CeN 4.87 mixed val.15 - SM16,29 -
SC17
PrN 5.17 unknown - HM16,17,29 M14,30
NdN 5.15 FM15 Tc = 27.6 − 35 HM16,29 M14
SC17
SmN 5.04 AFM15 /FM31 TN = 18 HM16,17,29 SC24
Tc = 26 − 30
EuN 5.00 unknown - HM16,17,29,32 M14,30
GdN 4.98 FM15,22,23 Tc = 58 − 69 HM16,18,33–35 SC23
SC17,36,37
TbN 4.92 FM/AFM Tc = 34 − 42 SC16,17 -
TN = 31
DyN 4.90 FM31 /AFM15 Tc = 17 − 26 SC16,17 SC24
TN = 15
HoN 4.87 FM15 Tc = 11 − 18 SC16,17 -
ErN 4.83 FM15,38 Tc = 3.4 − 7 SM16 SC38
SC17
TmN 4.80 none - SM16 -
SC17
YbN 4.78 AFM15,20 Tc = 0.75 SM16 SM20
SC17
LuN 4.76 none - SC17 M14,30
SC25
Ce), having donated the two 6s electrons and either a 5d or a 4f electron to
leave nitrogen in a −3 state. The resulting electronic structure would have
a valence band derived from the occupied nitrogen 2p states and a conduc-
tion band made from the empty rare-earth states. In this picture the 4f
levels remain partially occupied, and thus one may naively expect them to
lie in a narrow band at the Fermi level resulting in a metallic state. Indeed
basic calculations within the local density approximation (LDA) to density
functional theory (DFT) do result in such a band structure.19 However,
the picture is made considerably more complex when the strong correla-
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 201
tions amongst the 4f electrons are taken into account, something which
LDA fails to do. Furthermore, for materials possessing a band gap there is
the well-known failure of the LDA to adequately account for the screening
of the long-ranged Coulomb repulsion between the band electrons, which
results in underestimation of the band gap in semiconductors.39 Finally,
the strong exchange interaction between the 4f and the band electrons re-
quires the two different spin orientations to be treated separately. Thus
more advanced treatments are required, and the rare-earth nitrides have
become an ideal testing ground for techniques such as the the local spin
density approximation with a self-interaction correction (LSDA-SIC),40,41
LSDA with a Hubbard-U correction (LSDA+U ),29,34,35,42–45 advances on
the GW correction to DFT,36,46 or other means of accounting for many-
body effects.37,47
The relatively simple half-filled 4f shell of Gd in GdN has led band
structure calculations to focus the most attention on this RE-N. One of the
first band structure calculations was performed by Hasegawa and Yanase,48
who predicted a semiconducting state with a band gap of ∼ 1 eV, but who
noted the possibility that exchange splitting of the bands could conceivably
close the gap and make GdN metallic. In this calculation they treated the
4f levels as core states. Determining the best way to treat the 4f s and an-
swering the question of whether the ground state is metallic, half-metallic,
semi-metallic, or semiconducting, has dominated subsequent calculation at-
tempts. A series of papers by Lambrecht et al. has done a great deal to
elucidate the physics of the GdN band structure. An LSDA calculation
published in 1996 gave the first prediction that GdN is half-metallic,49 but
a follow-up paper in 2000 which used LSDA with an approximate GW cor-
rection found instead a semiconducting ground state.50 This was confirmed
by an LSDA+U calculation that included a Hubbard-like Uf to shift the
4f levels away from the Fermi level, as well as a quasiparticle correction to
the d levels that accounts for the usual underestimation of band gaps by
LSDA.51 The Uf values were determined by extrapolating photoemission
and inverse photoemission data from GdP, GdAs, GdSb, and GdBi.52 The
quasiparticle correction was included using a Ud value in a similar fashion
to the Hubbard Uf , but it was stressed that it actually represents quite
different physics, being a correction for long-ranged Coulomb effects rather
than the localised 4f correlations. The importance of including the Ud
correction is highlighted by the fact that without it the band structure is
half-metallic.
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
202 B. J. Ruck
Fig. 9.2. (a) Spin-resolved band structures of (a) GdN (from Ref. 17) and (b) EuO
(from Ref. 53) calculated by LSDA+U . The majority spin states are solid lines and the
minority states in dashed lines.
The resulting calculated band structure of GdN is reproduced in
Fig. 9.2(a). The occupied majority spin 4f levels, whose location is de-
termined by Uf , are seen as the very flat bands about 8 eV below the
Fermi level (defined to be the top of the valence band). The minority spin
4f levels, located 5 eV above the Fermi level, show slightly more dispersion
due to greater hybridisation with the conduction band states. Both the con-
duction and valence bands show strong exchange splitting of approximately
0.5 eV, but it is important to note that the splitting is of opposite sign, so
that the upper valence band and lower conduction band states are both of
majority spin character. The resulting band gap is indirect. The smallest
indirect gaps from Γ to X for majority and minority spins are 0.09 and
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 203
Fig. 9.3. The dependence of the band structure of GdN on lattice parameter. As the
lattice constant increases from (a) 4.92 Å through (b) 5.16 Å to (c) 5.63 Å the band
structure evolves from half-metallic to semiconducting (from Ref. 18).
0.82 eV, respectively, while the corresponding smallest direct gaps at X are
0.50 and 1.41 eV (note that these values have since been refined slightly
in light of new experimental data, as will be described in Section 9.3.2).
For comparison to experimental data it is important to note that these
values are appropriate for the ferromagnetic ground state. In the room
temperature paramagnetic state the band structure will be approximately
an average of the majority and minority spin bands.
The subtleties involved in obtaining the correct band structure near
the Fermi level are illustrated by contrasting the above results with an
LSDA-SIC calculation by Aerts et al. which found a half-metallic ground
state.16 A very interesting transition from a half-metallic to a semicon-
ducting state with increasing lattice constant was predicted by Duan et
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
204 B. J. Ruck
al.18 using LSDA+U , with the main results reproduced in Fig. 9.3. At
the equilibrium lattice constant the orbital overlap is large enough that the
Gd 5d majority spin conduction bands cross the majority spin N 2p va-
lence bands, giving a half-metal. As the lattice expands the orbital overlap
decreases and a band gap opens up. The exchange parameters were also
found to depend on the lattice constant, implying that both the conduction
state and the Curie temperature could be tuned by applying stress.
Possibly the most advanced band structure calculation has been pro-
vided by Chantis et al.36 using a recently developed quasiparticle self-
consistent GW correction to LSDA. This approach is able to cope with
both the localised nature of the 4f levels and the band-gap underestima-
tion issues. The band structure of GdN was found to be semiconducting,
and in good agreement with the LSDA+Uf +Ud results,51 but the prox-
imity to the critical point of a (half-) metal to insulator transition was
emphasised.
It is interesting to compare the band structure of GdN in Fig. 9.2(a) to
that of EuO shown in Fig. 9.2(b), which was calculated within the LSDA+U
framework, with U shifts applied to both the 4f and 5d electrons.53 Like
GdN, EuO is calculated to be semiconducting with a (spin-averaged) band
gap of 0.97 eV, and there is a strong exchange splitting of the Eu 5d-derived
conduction band of 0.66 eV. However, EuO and GdN differ significantly in
the location of the majority 4f levels. As described above, for GdN these
are well below the N 2p-derived valence band, but for EuO the occupied
4f levels form a rather flat band just below the Fermi level with the O
2p-derived valence band states lying approximately 2 eV further down.
Furthermore, the sense of the exchange splitting of the p levels is opposite
in EuO to that in GdN, with the minority O p states lying above the
majority states.
There have also been attempts to calculate the electronic structures of
other members of the rare-earth nitride series, including SmN,54 CeN,55
ErN,56 and EuN32,57 using similar theoretical frameworks to those applied
to GdN. The first systematic attempt to calculate the band structure across
the entire RE-N series was provided in 2004 by Aerts et al.,16 using LSDA-
SIC. This work predicted ground states ranging from metallic to insulating,
and including half-metallic for the RE-Ns from PrN to GdN, which height-
ened interest in the RE-Ns and their alloys as potential spintronic materials.
Following the focus on refining the calculation techniques that was applied
to GdN, Larson et al.17 used their LSDA+U framework to calculate the
band structures of all of the RE-Ns. This work emphasised the importance
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 205
of allowing the 4f orbitals to break the crystal field symmetry, and they
are generally found instead to obey Hund’s rules (the exceptions being EuN
and YbN). PrN, PmN, EuN, and SmN are found to be half-metallic, while
the other RE-Ns possess small indirect band gaps that tend to increase as
the rare-earth atomic number increases. EuN proved the most difficult to
study from a theoretical perspective, partly because divalent and trivalent
solutions have similar energy. A review article by Duan et al. has pro-
vided an excellent summary of the theoretical progress, including a more
detailed discussion of the calculation methods and their relative strengths
and weaknesses.19
9.3.2. Experiment
The rare-earth nitrides are presently in the unusual position that there
have been more attempts to calculate their band structure than there have
been experimental investigations to guide the theory. There were early at-
tempts to measure their room-temperature optical transmission, with the
majority of the series exhibiting an absorption minimum near 1 eV that
was interpreted as a band-gap.14,58 In some cases this was supported by
resistivity data that appeared to show semiconducting behaviour, albeit
with very large carrier concentrations. However, it is far from clear from
these measurements that the rare-earth nitrides are in fact semiconductors.
First, it should be remembered that the optical absorption edge generally
measures the direct band gap, but in RE-Ns elementary symmetry consid-
erations show that the indirect gap is smaller and may in fact be negative20
(see Section 9.3.1 above). The measured transmission minimum is rather
weak, so it was not completely clear whether the low-frequency absorption
implied that the materials were heavily doped semiconductors or in fact
semi-metals. Furthermore, as described above, a room temperature band
gap does not imply the existence of a band gap in the ground state of
ferromagnetic RE-Ns, as the exchange splitting tends to narrow the gap.
Finally, the quality of the early samples was compromised by the difficulties
achieving stoichiometry and in protecting the samples from oxidation, so it
was clear that a definitive determination of the conduction states required
improved samples.
The most comprehensive experimental investigation of any member of
the RE-N series was carried out using single crystals of YbN,20 which at
the time was motivated by suggestions that YbN might be a heavy fermion
system. This was found not to be the case, but the main result was that
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
206 B. J. Ruck
Fig. 9.4. Reflectivity of a YbN single crystal across a broad energy range. Also shown
is the reflectivity from a non-stoichiometric sample where the resulting increased carrier
concentration shifts the plasma edge to higher frequencies (from Ref. 20).
YbN is a self-compensated semimetal in which the ytterbium d states cross
below the nitrogen p states giving rise to a moderate carrier concentration
of about 1020 cm−3 . This conclusion was drawn from a study of the resis-
tivity and Hall effect, which show metallic behaviour, and measurements
of the optical constants covering a very broad energy range from 1 meV
to 12 eV (see Fig. 9.4). The reflectivity of a stoichiometric sample shows
a clear Drude edge at about 0.1 eV, and this moves out to about 0.2 eV
in a nitrogen deficient sample due to an increased carrier concentration.
Direct interband transitions dominate at higher energies, and they were
able to identify the feature at 0.2 eV as arising from transitions between
Yb d sates at the Fermi level and the unoccupied 4f 14 state. The close
proximity of the the empty 4f 14 level to Ef is in agreement with the band
structure calculation in Ref. 17, although there the calculation predicts a
semiconducting ground state.
This comprehensive study of YbN is extremely useful as a guide to the-
ory, but YbN becomes antiferromagnetic only below 0.75 K. Thus, some of
the interesting aspects of the RE-N band structures, such as the exchange
splitting, are not accessible. As with theoretical work, the more recent
experimental investigations have been largely focussed on GdN. Figure 9.5
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 207
Fig. 9.5. Temperature dependent resistance and magnetoresistance of a GdN film (from
Ref. 22).
Fig. 9.6. Temperature dependent resistivity of a GdN film demonstrating a semicon-
ducting ground state in this material (from Ref. 23). This film had fewer nitrogen
vacancies than the sample described in Fig. 9.5.
shows the temperature dependent resistivity and magnetoresistance of GdN
thin films grown by reactive-ion sputtering as measured by Leuenberger et
al.22 The resistivity goes through a peak at the Curie temperature (Tc )
of about 59 K, and a rather strong negative magnetoresistance is observed
in the vicinity of Tc . They attribute the behaviour to a transition from
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
208 B. J. Ruck
a degenerately doped semiconductor above Tc to a metallic state below
Tc . The nature of the metallic state is similar to that observed in non-
stoichiometric EuO,59 namely the exchange splitting does not fully close
the band gap but instead it drives the bottom of the conduction band
down below a rather large density of nitrogen vacancy induced donor states,
which lie in the gap for temperatures above Tc . Further work by Granville
et al.23 on samples with higher resistivity, and presumably fewer nitrogen
vacancies, found a stronger peak at Tc (which was 69 K in their films) and
a semiconducting ground state at low temperature (see Fig. 9.6). These
findings are in agreement with the band structure calculations in Refs. 36,
51 and highlight the importance of adequately treating the long-range
Coulomb interactions.
More direct evidence for the presence of a band gap in the ground state
of GdN was obtained from the temperature dependent optical transmit-
tance data of Trodahl et al.60 shown in Fig. 9.7. At room temperature
there is a relatively clear absorption edge at around 1.3 eV, corresponding
to direct transitions at X in the band structure, and there is little change
as the temperature is lowered to 75 K. However, at 6 K there is a clear red
shift of the absorption edge to 0.9 eV, which is exactly as expected when
exchange splitting narrows the majority spin band gap in the ferromagnetic
Fig. 9.7. Optical transmittance of a GdN thin film showing a red shift of the absorp-
tion edge at low temperature due to narrowing of the majority spin band gap in the
ferromagnetic phase (from Ref. 60).
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 209
phase. These data were used to retune the Ud parameter in the calculated
LSDA+U band structure and thereby obtain excellent agreement between
theory and experiment. Recently, calculations of the full optical conductiv-
ity of GdN have been presented,35,61 and it will be interesting in future to
compare quality experimental measurements to these calculations.
Transport and optical studies address the band structure near to the
Fermi level. This can be complemented by synchrotron-based X-ray ab-
sorption and emission spectroscopy62 (XAS and XES), which measure the
nitrogen p-projected partial density of empty and occupied states, respec-
tively. XAS and XES studies have been performed for GdN by Leuenberger
et al.22,63 and for SmN and DyN by Preston et al.24 Results from the latter
study are shown in Fig. 9.8 where they are compared to the appropriate
partial densities of states (PDOS) obtained from LSDA+U band structure
calculations performed according to the method described in Ref. 17. The
XES data show a relatively narrow nitrogen-derived valence band which is
in excellent agreement with the calculated PDOS. For the XAS data the
presence of the N 1s core hole in the final state must be considered, which
is achieved using the so-called final state rule.64,65 Thus the XAS data
are in excellent agreement across a broad energy range with a calculation
that includes a core hole on every fourth nitrogen site. The data for SmN
and DyN are similar, with only subtle differences between relative peak
locations and intensities, highlighting the similarity between the materials
electronic structures (more striking differences would be observed using a
technique that directly probed the 4f PDOS). For both materials there is
a gap of around 1 eV between the filled and empty states, but due to the
effects of the core hole and the absence of strong N p character at the bot-
tom of the conduction band this cannot be taken as direct evidence for the
presence of a room temperature band gap. However, resistivity data from
these two films both showed a semiconductor-like temperature dependence
with a peak at the magnetic ordering temperature similar to that shown in
Fig. 9.6 for GdN, providing evidence that these two materials are indeed
semiconductors both at room temperature and at low temperature.
There is clearly a reasonable agreement between the available experi-
mental data and the most recent band structure calculations for the RE-N
series. However, some issues still require addressing, such as the disagree-
ment between the calculated and measured conduction states of SmN and
YbN, and the absence of any systematic experimental data that provides
information about the location of the 4f orbitals. In the case of the former
it will be interesting to see whether further tuning of the Ud parameter
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
210 B. J. Ruck
Fig. 9.8. Measured X-ray absorption (thick black line) and emission (thin black line)
for (a) DyN and (b) SmN, compared to the calculated nitrogen PDOS. The calculations
are performed both with (solid lines) and without (dotted lines) including the core hole
(from Ref. 24).
in LSDA+U will be sufficient to describe the band structures of all of the
RE-Ns near the Fermi level, or whether perhaps further new physics must
be considered.
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 211
9.4. Magnetic Properties
The magnetic properties of rare-earth ions have remained at the forefront
of condensed matter physics for nearly 80 years, for reasons of both fun-
damental and technological interest. The highly localised nature of the 4f
electrons is in contrast to the more itinerant nature of the d electrons in
transition metal-based magnetic materials, and thus provides an interest-
ing contrast to those systems. The atomic picture generally provides an
excellent starting point for a description of their magnetic moments, but
the lack of direct orbital overlap between the 4f electrons on neighbouring
ions means that the exchange interactions that lead to magnetic ordering
can be quite complex.
Much of the focus of the early work on the RE-Ns was on determining
their magnetic ground states and transition temperatures, with the results
summarised in Table 9.1. As with the electronic properties, issues with
sample stoichiometry and oxygen contamination have contributed to con-
siderable uncertainty in the magnetic transition temperatures, and even
in the exact nature of the magnetic ordering. For example, it has been
noted that quantities of oxygen greater than about 5% can destroy fer-
romagnetism and give rise to antiferromagnetic correlations in GdN,66,67
and it has even been reported that pure GdN with no nitrogen vacancies
is metamagnetic.68 Inclusion of carbon has been reported to raise Tc to
as much as 190 K.67 More recent experimental work has found ferromag-
netism in GdN samples with low levels of oxygen impurities and nitrogen
vacancies, although there is still some variation in the measured Curie tem-
peratures.22,23,26,63
The strong coupling between the electronic and magnetic properties of
the RE-Ns has stimulated renewed interest in their magnetism. Numerous
attempts have been made to model the exchange mechanism in GdN69–72
starting from band-structure calculations (there are no recent predictions
of the transition temperatures of the other RE-Ns). The focus on GdN is
in part prompted by the simplifying effect of having a half-filled 4f shell
and hence no orbital component to the 4f magnetism. It is generally ac-
cepted that competing exchange interactions exist,72 with an antiferromag-
netic Gd-Gd superexchange contribution mediated through the nitrogen
ions and a ferromagnetic Gd-Gd interaction mediated by the Gd d or-
bitals. The first interaction is reasonably well understood, and calculations
are able to reproduce the measured Néel temperatures of the related pnic-
tides GdP, GdAs, GdSb, and GdBi for which the same antiferromagnetic
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
212 B. J. Ruck
contribution is the dominant exchange mechanism.69 The ferromagnetic
contribution has proven more difficult to interpret. It has been assumed
that a Ruderman-Kittel-Kasuya-Yosida-type carrier mediated exchange is
important in Gd-monopnictides,19 but this is unlikely to be relevant for
semiconducting GdN at the experimentally observed carrier concentrations,
and higher order mechanisms have also been explored.71
Some insight into the magnetic interactions in GdN has been provided
by recent experiments. Leuenberger et al. found a surprisingly strong x-
ray magnetic circular dichroism (XMCD) signal at the nitrogen K-edge
in GdN, which could only be understood based on the presence of p − d
exchange between the Gd and N sites which induces a magnetic moment
on the nitrogen. Two groups have modelled the XMCD with some success,
although their calculations are both based on LSDA+U band structure
calculations that predict half-metallic behaviour in GdN.33,34 The influence
on Tc of an expanded lattice constant has also been investigated using
XMCD63 and using ferromagnetic resonance.26 In both cases a reduced
Tc was observed, in agreement with expectations of a reduction in the
ferromagnetic exchange interaction as the lattice constant increases.18
A very detailed theoretical analysis has recently been provided by Mi-
tra and Lambrecht.69 Their crucial finding is that the Gd 5d and N 2p
states carry equal and opposite moments, so the material is essentially an
antiferromagnet. This coupling can be directly interpreted in terms of the
band structure, where the Gd 5d – N 2p hybridisation is greater for the
majority spin states as these lie closer together (see Fig. 9.2(a)), with the
result being a transfer of majority spin to the d levels leaving a net minority
spin on the nitrogen site. However, the p and d moments are small, and
the ferromagnetic f − d coupling on the Gd site ensures that the strong 4f
moments are all aligned with the 5d moments giving overall ferromagnetism
amongst the 4f spins.
Although the intuitive nature of this model is highly appealing, when the
extracted exchange parameters are used to calculate the Curie temperature
the resulting value of 11 K is significantly lower than the experimental
values (see Table 9.1). It may be that this discrepancy is related to the
changes to the band structure induced by defects such as nitrogen vacancies,
which could alter the rather subtle antiferromagnetic p − d exchange. This
again is evidence of the sensitivity of the RE-N electronic structure to
perturbations, as was also shown in Fig. 9.3. Clarifying the exact nature of
the exchange interaction may have to wait for further advances in sample
preparation to enable systematic studies of the role of defects.
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 213
Fig. 9.9. Magnetic hysteresis loops from a SmN thin film taken after cooling in zero
field (filled circles) or after cooling in 6 T (open squares). At 2 K the coercive field is
larger than the maximum applied field of 6 T, so no hysteresis is evident. The curves
on the right have been corrected for a paramagnetic contribution from the GaN capping
layer and Si substrate (from Ref. 31).
The two rare-earth ions with one or two electrons less than a half-
filled 4f shell, Sm and Eu, are also particularly interesting. According to
Hund’s rules Eu3+ has a ground state with J = 0 and hence it should be
nonmagnetic (i.e., the magnetic moment should have zero projection along
an applied field). However, Johannes and Pickett32 have pointed out that
this does not mean that EuN lacks long-ranged spin or orbital correlations.
For EuN they have raised the possibility of a “hidden ferromagnetic order”,
which provides a fascinating experimental challenge to detect.
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
214 B. J. Ruck
Sm3+ has a 4f 5 configuration and according to Hund’s rules a total
orbital angular momentum L = 5, a spin angular momentum S = 5/2,
and in the ground state these oppose each other giving J = |L − S| =
5/2. The magnetic moment per ion is given by |M | = µB |(L
+ 2S)|
so for
SmN the orbital and spin contributions to the magnetic moment are then
expected to almost cancel. This has been highlighted in the band structure
calculations of Ref. 17, and recently confirmed experimentally by Meyer
et al.31 Magnetisation data measured on a thin film of SmN are shown in
Fig. 9.9. Clear hysteresis is observed at temperatures below about 30 K
indicative of ferromagnetism. However, below about 10 K the hysteresis
loops close, and loops taken after cooling in a 6 T applied field are shifted
relative to those taken after zero-field cooling. This is understood based
on the very small moment of the Sm3+ ion, which couples weakly to the
applied field such that at low temperatures the magnetic domains cannot be
rotated even in the highest available field of 6 T. Near-vanishing magnetic
moments are not unknown in Sm compounds,73 but the observation of this
in a semiconducting material appears thus far to be unique to SmN.
It should be noted that early experiments concluded that SmN is anti-
ferromagnetic15 below about 20 K, although these were not confirmed by
neutron scattering.74 This is not surprising considering the weak moment
in SmN. A peak in the resistance is observed at the magnetic transition tem-
perature similar to that shown for GdN in Fig. 9.6, giving further support
to the conclusion that SmN is indeed ferromagnetic.
The magnetic properties of rare-earth nitrides offer rich prospects for
future research. It is also interesting to consider the relationship between
RE-Ns and rare-earth doped conventional nitride semiconductors, such as
Gax Gd1−x N, for which remarkably high magnetic moments have been ob-
served.10 Although the local structure around a Gd ion in GdN is different
to that in GaN, an understanding of the exchange mechanism and electronic
states in GdN may provide insight into the magnetism in Gax Gd1−x N, shed-
ding light on aspects such as whether the Gd introduces significant defect
states.
9.5. Device Prospects and Future Challenges
The discussion presented above makes it clear that, although some highly
interesting questions remain, a consensus is emerging about many of the
RE-Ns electronic and magnetic properties which make them of consider-
able interest for use in future spintronics devices. The series contains several
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 215
ferromagnetic semiconductors which can in principle be doped to show very
high spin polarisation, and other members of the series may be half-metallic
in their intrinsic state. This range of electronic structure, combined with
their chemical similarity, makes it possible to imagine tuning their proper-
ties by alloying, or through growth of heterostructures consisting of layers
of different rare-earth nitrides. The possible applications that can be con-
sidered range from simple field sensors to more exotic devices that rely on
the RE-Ns novel band structures.
Fig. 9.10. Demonstration of spin filtering using a thin EuO tunnel barrier between two
metallic electrodes (Al/EuO/Y). (a) The tunneling current decreases significantly below
Tc and is dependent on the applied bias. (b) The tunnel barrier is determined by the
exchange splitting and is thus spin dependent. The measured resistance is best fitted
using a model in which spin polarisation reaches 98% at low temperature (from Ref. 75).
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
216 B. J. Ruck
The results of Section 9.3.2 show that GdN has a large magnetoresis-
tance, especially at temperatures close to the Curie temperature, and it is
quite likely that this can be further enhanced by controlling the defect states
that dominate the electrical conductivity. This material thus becomes of
interest for use in magnetic field sensors. The major requirements for a
resistive field sensor are a large magnetoresistance over a broad field range,
small coercive field, and high transition temperature. GdN satisfies the
first, and as its magnetic moment is almost entirely spin-derived it also has
a small coercive field which can be minimised by improving crystalline qual-
ity. The transition temperature is low, so a GdN-based field sensor would
be restricted to cryogenic operation, performing such tasks as sensing the
field in a superconducting magnet. However, it is possible that steps can be
taken to increase the transition temperature through alloying or strained
growth. As noted above carbon doping of GdN has been reported67 to raise
the Tc to 190 K, and the Tc of the ferromagnetic semiconductor EuO can
be raised from 69 K to 118 K by La doping,76 to 170 K by Gd doping,77 or
to over 200 K under pressure.78
A far more novel proof-of-concept spintronic device relying on the strong
exchange splitting in EuO to produce a highly efficient spin filter has re-
cently been demonstrated by Santos et al.75 Figure 9.10 demonstrates the
device architecture and performance. A thin layer (1–6 nm) of ferromag-
netic EuO is placed between two metallic contacts (Al and Y in the present
example), with the exchange splitting of the EuO conduction band used
to produce a spin-dependent tunneling barrier for electrons passing from
the Al to the Y electrode. The very strong dependence of the tunneling
current on the barrier height produces a tunneling current with 98% spin
polarisation. The rare-earth nitrides exhibit comparable exchange splitting,
so similar devices could be made for example by replacing the EuO with
GdN. Furthermore, GdN also has a spin-split valence band (see Fig. 9.2),
so a GdN-based spin filter could also be used as a source or detector of
spin-polarised holes.
This polarisation of both the valence and conduction bands also raises
the possibility of utilising semiconducting rare-earth nitrides to modify con-
ventional semiconductor devices. If controlled doping can be achieved then
it should be possible to fabricate diodes or transistors in which the cur-
rents are fully spin polarised. While the concept of a so-called spin bipolar
junction transistor has been explored,3 that proposal only includes a spin-
polarised base and does not include polarisation of both carrier types. An
investigation of such a device and an analysis of its efficacy would be both
novel and interesting.
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 217
It was shown in Section 9.4 that SmN is a ferromagnetic semiconductor
with a near-vanishing magnetic moment. As such it can potentially be used
as a ferromagnetic element in spintronics devices in which stray fields are
undesirable. One such specialist application would be as a magnetic coating
on a scanning tunneling microscope tip, and more conventional spintronics
applications could be imagined.
The above discussion lists only a few potential applications of rare-earth
nitrides in spintronics devices, and it is certain that more examples could be
imagined. For uptake of these devices several materials constraints must be
overcome. Room temperature device operation is clearly desirable in most
applications, and here the relatively low Tc s of the RE-Ns are an obvious
limitation. However, as already mentioned there is scope for increasing the
Tc , and there are niche markets for spintronics devices operating at low
temperature. Furthermore, even cryogenic demonstration of highly novel
proof-of-concept devices exploiting the rare-earth nitrides band structures
would clearly represent an exciting development in the field of spintronics,
perhaps paving the way to commercial uptake of similar devices based on
materials with related properties.
Improving the material quality is another immediate area to address. At
present even the highest quality samples have significant densities of defects
such as nitrogen vacancies and grain boundaries, and these are capable of
having a significant impact on the observed properties of the RE-Ns. Fur-
ther investigation of epitaxial growth is required to determine the growth
conditions that produce highly crystalline and stoichiometric material. This
will require systematic studies of the influence of parameters such as sub-
strate temperature, active nitrogen source, and rare-earth to nitrogen flux
ratios. Of particular use would be schemes to grow RE-N films on existing
semiconductors, such as silicon or gallium arsenide, which could allow in-
tegration into existing device architectures. Following optimisation of the
growth it will be possible to move towards deliberate doping, both n- and
p-type, to control the carrier concentration, from which point multilayered
devices could be produced. The situation in this respect very much mirrors
that of GaN in the early 1990s, where a few key breakthroughs paved the
way for commercial applications.79 Further development of robust encapsu-
lation techniques will also be required to prevent degradation of the RE-Ns
in air. The growth studies should be combined with theoretical investiga-
tions of the electronic structure of defects and potential dopants. These
will be extremely useful in guiding the growth optimisation, but are likely
to test the limits of the calculation techniques.
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
218 B. J. Ruck
9.6. Conclusions
This Chapter has provided a summary of the present state of knowledge
concerning the properties of the rare-earth nitrides. They exhibit some
highly interesting physics, with strongly correlated electronic structures
that lie near the boundary between what can and cannot be treated ad-
equately with modern band structure calculation tools. Their number
includes several intrinsically ferromagnetic semiconductors, which can in
principle be doped either n- or p-type, marking a key difference between
the rare-earth nitrides and other commonly studied intrinsic ferromagnetic
semiconductors or dilute magnetic semiconductors. If sample preparation
techniques can be optimised to improve crystal quality and minimise defect
density then their novel properties may one day be utilised in spintronics
devices.
Acknowledgments
The author gratefully acknowledges useful discussions with Prof. H. Joe
Trodahl, Prof. Claire Meyer, Prof. Walter R. L. Lambrecht, Prof. Kevin
Smith, Dr. James Downes, Dr. Jan Richter, Andrew Preston, and Bart
Ludbrook.
References
1. I. Z̆utic̆, J. Fabian, and S. Das Sarma, Rev. Mod. Phys. 76, 323, (2003).
2. D. D. Awschalom and M. E. Flatte, Nat. Phys. 3, 153, (2007).
3. S. Bandyopadhyay and M. Cahay, Introduction to Spintronics. (CRC Press,
2008).
4. S. S. P. Parkin, C. Kaiser, A. Panchula, P. M. Rice, B. Hughes, M. Samant,
and S.-H. Yang, Nat. Mat. 3, 862, (1994).
5. http://www.freescale.com; http://www.everspin.com.
6. I. Z̆utic̆, J. Fabian, and S. C. Erwin, J. Phys.: Condens. Matter. 19, 165219,
(2007).
7. T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, Sci. 287, 1019,
(2000).
8. T. Jungwirth, K. Y. Wang, J. Masek, K. W. Edmonds, J. Konig, J. Sinova,
M. Polini, N. A. Goncharuk, A. H. MacDonald, M. Sawicki, A. W. Rushforth,
R. P. Campion, L. X. Zhao, C. T. Foxon, and B. L. Gallagher, Phys. Rev. B.
72, 165204, (2005).
9. G. T. Thaler, M. E. Overberg, B. Gila, R. Frazier, C. R. Abernathy, S. J.
Paerton, J. S. Lee, S. Y. Lee, Y. D. Park, Z. G. Khim, J. kim, and F. Ren,
Appl. Phys. Lett. 80(3964), (2002).
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 219
10. S. Dhar, O. Brandt, M. Ramsteiner, V. F. Sapega, and K. H. Ploog, Phys.
Rev. Lett. 94(037205), (2005).
11. M. Venkatesan, C. B. Fitzgerald, J. G. Lunney, and J. M. D. Coey, Phys.
Rev. Lett. 93(177206), (2004).
12. E. L. Nagaev and M. Samokhvalov (Translator), Physics of Magnetic Semi-
conductors. (Mir Publishers, 1983).
13. E. L. Nagaev, Colossal Magnetoresistance Materials and Other Magnetic
Semiconductors. (World Scientific Publishing Company, 2002).
14. F. Hulliger, Handbook on the Physics and Chemistry of Rare Earths, vol. 4,
chapter edited by Karl A. Gschneider Jr., and LeRoy Eyring, pp. 153–236.
(North-Holland Physics Publishing, New York, 1979).
15. O. Vogt and K. Mattenberger, Handbook on the Physics and Chemistry of
Rare Earths, vol. 17, chapter edited by K. A. Gschneider Jr., L. Eyring, G.
H. Lander and G. R. Choppin, pp. 301–407. (Elsevier Science Publishers,
New York, 1993).
16. C. M. Aerts, P. Strange, M. Horne, W. M. Temmerman, Z. Szotek, and
A. Svane, Phys. Rev. B. 69, 045115, (2004).
17. P. Larson, W. R. L. Lambrecht, A. Chantis, and M. van Schilfgaarde, Phys.
Rev. B. 75, 045114, (2007).
18. C.-G. Duan, R. F. Sabiryanov, J. Liu, W. N. Mei, P. A. Dowben, and J. R.
Hardy, Phys. Rev. Lett. 94, 237201, (2005).
19. C.-G. Duan, R. F. Sabiryanov, W. N. Mei, P. A. Dowben, S. S. Jaswal, and
E. Y. Tsymbal, J. Phys.: Condens. Matter. 19, 315220, (2007).
20. L. Degiorgi, W. Bacsa, and P. Wachter, Phys. Rev. B. 42, 530, (1990).
21. R. Hauger, E. Kaldis, G. von Schulthess, P. Wachter, and C. Zürcher, J.
Magn. Magn. Mater. 103, 3, (1976).
22. F. Leuenberger, A. Parge, W. Felsch, K. Fauth, and M. Hessler, Phys. Rev.
B. 72, 014427, (2005).
23. S. Granville, B. J. Ruck, F. Budde, A. Koo, D. Pringle, F. Kuchler, A. Bittar,
G. V. M. Williams, and H. J. Trodahl, Phys. Rev. B. 73, 235335, (2006).
24. A. R. H. Preston, S. Granville, D. H. Housden, B. Ludbrook, B. J. Ruck, H. J.
Trodahl, A. Bittar, G. V. M. Williams, J. E. Downes, A. DeMasi, Y. Zhang,
K. E. Smith, and W. R. L. Lambrecht, Phys. Rev. B. 76, 245120, (2007).
25. A. R. H. Preston, unpublished. (2008).
26. K. Khazen, H. J. von Bardeleben, J. L. Cantin, A. Bittar, S. Granville, H. J.
Trodahl, and B. J. Ruck, Phys. Rev. B. 74, 245330, (2006).
27. J. W. Gerlach, J. Mennig, and B. Rauschenbach, Appl. Phys. Lett. 90,
061919, (2007).
28. B. Ludbrook, unpublished. (2008).
29. E. Burzo, N. Bucur, H. Allmaier, and L. Chioncel, J. Optoelectron. Adv.
Mater. 10, 389, (2008).
30. W. Stutius, Phys. Kondens. Mater. 10, 152, (1969).
31. C. Meyer, B. J. Ruck, J. Zhong, S. Granville, A. R. H. Preston, G. V. M.
Williams, and H. J. Trodahl, Phys. Rev. B. 78, 174406, (2008).
32. M. D. Johannes and W. E. Pickett, Phys. Rev. B. 72, 195116, (2005).
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
220 B. J. Ruck
33. V. N. Antonov, B. N. Harmon, A. N. Yaresko, and A. P. Shpak, Phys. Rev.
B. 75, 184422, (2007).
34. S. Abdelouahed and M. Alouani, Phys. Rev. B. 76, 214409, (2007).
35. D. B. Ghosh, M. De, and S. K. De, Phys. Rev. B. 72, 045140, (2005).
36. A. N. Chantis, M. van Schilfgaarde, and T. Kotani, Phys. Rev. B. 76, 165126,
(2007).
37. A. Sharma and W. Nolting, J. Phys.: Cond. Matt. 18, 7337, (2006).
38. C. Meyer, unpublished. (2008).
39. R. P. Martin, Electronic Structure: Basic Theory and Practical Methods.
(Cambridge University Press, 2004).
40. J. P. Perdew and A. Zunger, Phys. Rev. B. 23, 5048, (1981).
41. W. M. Temmerman, A. Svane, Z. Szotec, and H. Winter, Electronic Density
Functional Theory: Recent Progress and New Directions. (Plenum, New York,
1997).
42. V. I. Anisimov, J. Zaanen, and O. K. Anderson, Phys. Rev. B. 44, 943,
(1991).
43. V. I. Anisimov, F. Aryasetiawan, and A. I. Lichtenstein, J. Phys.: Condens.
Matter. 9, 767, (1997).
44. A. I. Lichtenstein, V. I. Anisimov, and J. Zaanen, Phys. Rev. B. 52, R5476,
(1995).
45. G. Kotliar, S. Y. Savrasov, K. Haule, V. S. Oudovenko, O. Parcollet, and
C. A. Marianetti, Rev. Mod. Phys. 78, 865, (2006).
46. M. van Schilfgaarde, T. Kotani, and S. Faleev, Phys. Rev. Lett. 96, 226402,
(2006).
47. S. Bhattacharjee and S. M. Jaya, Eur. Phys. J. B. 49, 305, (2006).
48. A. Hasegawa and A. Yanase, J. Phys. Soc. Jpn. 42, 492, (1977).
49. A. G. Petukhov, W. R. L. Lambrecht, and B. Segall, Phys. Rev. B. 53, 4324,
(1996).
50. W. R. L. Lambrecht, Phys. Rev. B. 62, 13538, (2000).
51. P. Larson and W. R. L. Lambrecht, Phys. Rev. B. 74, 085108, (2006).
52. H. Yamada, T. Fukawa, T. Muro, Y. Tanaka, S. Imada, S. Suga, D. X. Li,
and T. Suzuki, J. Phys. Soc. Jpn. 65, 1000, (1996).
53. P. Larson and W. R. L. Lambrecht, J. Phys.: Condens. Matt. 18, 11333,
(2006).
54. A. Svane, V. Kanchana, G. Vaitheeswaran, G. Santi, W. M. Temmerman,
Z. Szotek, P. Strange, and L. Petit, Phys. Rev. B. 71, 045119, (2005).
55. M. S. S. Brooks, J. Magn. Magn. Mater. 47&48, 260, (1985).
56. T. Komesu, H.-K. Jeong, J. Choi, C. N. Borca, P. A. Dowben, A. G.
Petukhov, B. D. Schultz, and C. J. Palmström, Phys. Rev. B. 67, 035104,
(2003).
57. M. Horne, P. Strange, W. M. Temmerman, Z. Szotek, A. Svane, and H. Win-
ter, J. Phys.: Condens. Matter. 16, 5061, (2004).
58. G. Busch, G. E. Kaldis, E. Schaufelberger-Teker, and P. Wachter. In Coll.
Int. CNRS No. 180, vol. 1, p. 359, Paris-Grenoble, (1970).
59. Y. Shapira, S. Foner, and T. B. Reed, Phys. Rev. B. 8, 2299, (1973).
March 26, 2010 18:39 World Scientific Review Volume - 9in x 6in chapter9
Spintronics Potential of Rare-Earth Nitrides 221
60. H. J. Trodahl, A. R. H. Preston, J. Zhong, B. J. Ruck, N. Strickland,
C. Mitra, and W. R. L. Lambrecht, Phys. Rev. B. 76, 085211, (2007).
61. C. Mitra and W. R. L. Lambrecht, Phys. Rev. B. 78, 195203, (2008).
62. J. Stöhr, NEXAFS Spectroscopy. (Springer, 2003).
63. F. Leuenberger, A. Parge, W. Felsch, F. Baudelet, C. Giorgetti, E. Dartyge,
and F. Wilhelm, Phys. Rev. B. 73, 214430, (2006).
64. U. von Barth and G. Grossmann, Phys. Rev. B. 25, 5150, (1982).
65. G. D. Mahan, Phys. Rev. B. 21, 1421, (1980).
66. R. J. Gambino, T. R. McGuire, H. A. Alperin, and S. J. Pickart, J. Appl.
Phys. 41, 933, (1970).
67. M. Kuznietz, J. Appl. Phys. 42, 1470, (1971).
68. R. A. Cutler and A. W. Lawson, J. Appl. Phys. 46, 6, (1976).
69. C. Mitra and W. R. L. Lambrecht, Phys. Rev. B. 78, 134421, (2008).
70. K. Doll, J. Phys.: Condens. Matter. 20, 075214, (2008).
71. T. Kasuya and D. X. Li, J. Magn. Magn. Mater. 167, L1, (1997).
72. C.-G. Duan, R. F. Sabiryanov, W. N. Mei, P. A. Dowben, S. S. Jaswal, and
E. Y. Tsymbal, Appl. Phys. Lett. 88, 182505, (2006).
73. H. Adachi and H. Ino, Nature. 401, 148, (1999).
74. R. M. Moon and W. C. Koehler, J. Magn. Magn. Mater. 14, 265, (1979).
75. T. S. Santos, J. S. Moodera, K. V. Raman, E. Negusse, J. Holroyd, J. Dvorak,
M. Liberati, Y. U. Idzerda, and E. Arenholz, Phys. Rev. Lett. 101, 147201,
(2008).
76. A. Schmehl, V. Vaithyanathan, A. Herrnberger, S. Thiel, C. Richter,
M. Liberati, T. Heeg, M. Rockerath, L. F. Kourkoutis, S. Muhlbauer, P. Boni,
D. Muller, Y. Barash, J. Schubert, Y. Idzerda, J. Mannhart, and D. Schlom,
Nat. Mater. 6, 882, (2007).
77. H. Ott, S. J. Heise, R. Sutarto, Z. Hu, C. F. Chang, H. H. Hsieh, H. J. Lin,
C. T. Chen, and L. H. Tjeng, Phys. Rev. B. 73, 094407, (2006).
78. D. DiMarzio, M. Croft, N. Sakai, and M. W. Schafer, Phys. Rev. B. 35, 8891,
(1987).
79. S. Nakamura, G. Fasol, and S. J. Pearton, The Blue Laser Diode: The Com-
plete Story. 2nd edition, (Springer, 2000).
This page is intentionally left blank
Chapter 10
DILUTE MAGNETIC OXIDES:
CURRENT STATUS AND PROSPECTS
Karen Yates
The Blackett Laboratory, Physics Department
Imperial College London, SW7 2AZ, UK
E-mail: [email protected]
Over the past ten years, a new field of dilute magnetic semiconductors
and dilute magnetic oxides has emerged. These materials, in which a
host semiconductor such as ZnO is doped with a transtion metal ion in
order to produce ferromagnetic behaviour, have been extremely
controversial. Numerous reports have claimed intrinsic ferromagnetism
above room temperature while a large number of others attribute these
room temperature effects to impurities. Certainly, the observation of
room temperature ferromagnetism in insulating samples is a challenge
both theoretically and experimentally. Recent developments have
shown that some of these dilute magnetic oxides carry a spin polarised
transport current. These materials hold significant potential for
spintronics devices if the ferromagnetism can be understood and
controlled.
10.1. Introduction
Dilute magnetic semiconductors (DMS) are materials such as GaAs,
AlN, InSb that become ferromagnetic (FM) when doped with transition
metal (TM) ions such as Mn, Fe, Co etc. In recent years, the field of
dilute magnetic semiconductors has developed a sub-field, “dilute
magnetic oxides” (DMO) in which the TM ions are doped into
semiconducting oxides (eg. ZnO, SnO2, TiO2) with, or without,
additional doping in order to introduce carriers. Many of these oxides
have been reported to show FM behaviour above room temperature and a
significant portion of these FM oxides are highly resistive. This has
223
224 K. Yates
caused some authors to introduce a further classification “Dilute
magnetic dielectrics” (DMD) when FM is observed in insulating
samples. Research into DMS, DMO and DMD materials has, in recent
years, been extraordinarily active. The volume of papers published in the
field (more than 290 on the magnetic properties of ZnO in the past year
alone) has meant that there are, by necessity, frequent reviews of the
subject. In the past few years there have been excellent reviews on
growth conditions, theoretical models and device applications and the
reader is referred to these for in depth discussion of these points.1–5 A
room temperature (or above) ferromagnetic spin injector that could easily
be incorporated into existing electronic architecture, opens up many
device applications for spintronics and has been a motivating factor in
this research drive.1 Furthermore, the prospect of transparent DMOs
opens the door to a whole new field of magneto-optical devices.6 Early
research on the carrier mediated dilute magnetic semiconductor
(Ga,Mn)As quickly led to devices being made from this DMS material.7,8
However, despite intensive effort to raise the Curie temperature, Tc,
the low TC, of (Ga,Mn)As (172 K) limits this material’s applications
potential. When calculations by Dietl et al., suggested that p-type Mn
doped ZnO would be ferromagnetic at temperatures up to or above
room temperature9 a frenzy of activity followed, where numerous
semiconductors were investigated for their room temperature DMS
potential.
DMS materials typically show very small magnetic moments relative
to their sample size, of the order of ~10-6 emu. The magnitude of these
magnetic signals has inevitably led to conflicting results. One difficulty
is that such magnetic signals can easily arise from impurity sources,
either extrinsic (eg. stainless steel tweezers10) or intrinsic (eg. clusters11–13).
Additionally, even in thoroughly characterised and carefully handled
samples, results are highly variable and contradictory.2,14–62 It is
becoming understood that defects are necessary for the observation of
room temperature ferromagnetism, particularly in the insulating DMD
materials.4,54 The implicit irreproducibility of defects may then go some
way to explaining the amount of disagreement in the literature. Recently,
it has even been suggested that under certain conditions, some undoped
Dilute Magnetic Oxides: Current Status and Prospects 225
materials become ferromagnetic10,63–67 and that adding the magnetic
dopant, Co, into ZnO actually reduces the natural magnetic moment of
the host material.66 These results are often explained within the
developing theoretical context of d0 ferromagnetism.64,68,69 However,
before assigning the inconsistencies in the literature to intrinsic defects or
to carrier concentration, it is important to rule out the potential impurities
that can contribute to the magnetic signal observed. It should be
anticipated that some of the papers which report intrinsic ferromagnetism
may contain impurities yet limitations in detection technique available
prevent their observation.70 In this field of DMS, Carl Sagan’s much
used quote holds more than ever: “absence of evidence is not evidence
of absence”.
10.2. Impurities
A major problem of DMS research is that the moments typically
measured in thin film samples are at the detection limit of the
magnetometers employed. This becomes particularly important when
these small FM signals are superimposed onto a large diamagnetic signal
which originates from the substrate. Small amounts of extrinsic
magnetic impurity can easily show magnetic moments ~5 µ emu.31
Thorough characterisation of the samples for potential impurities is thus
key to producing intrinsically FM samples. The low concentration of
TM ions and therefore impurity phases means that traditional detection
techniques such as X-ray diffraction may not be adequate to detect
impurities that cause measureable magnetic properties. In this section,
expected impurities and ways to detect them are reviewed.
10.2.1. Types of impurity in DMS systems
10.2.1.1. Extrinsic impurities
Different types of impurities can contribute to the magnetic properties of
DMS candidate systems. At the simplest level, impurities have been
found to originate from the type of tweezers used to handle the thin films
226 K. Yates
as they were grown.10 Abraham et al.,10 undertook a systematic study of
the growth of HfO2 thin films. It has recently been suggested that this
material is intrinsically magnetic if oxygen deficient.64,67,68 Abraham
et al., found no ferromagnetic signal in any of the films that they grew
when Teflon tweezers were used to handle the films during and after
growth. However, stainless steel tweezers used during film handling led
to the observation of a magnetic signal of the same order of magnitude as
that found in the material itself by Venkatesan et al.67 Similarly, by
subjecting a series of blank substrates to the same thermal treatment as
used in thin film growth of DMS, Golmar et al.,71 and Belghazi et al.,72
showed that magnetic signals were present which were due to extrinsic
impurties originating from the substrate holders. These facts emphasise
the caution with which the DMS materials must be handled during
growth and characterisation when the magnetic moments observed are
close to the detection limit. Nonetheless, even while avoiding extrinsic
impurities, the DMS/DMO materials are host to a minefield of potential
magnetic impurities which need to be considered and eliminated as
potential sources of ferromagnetism in any claimed DMS/DMO
material.3
10.2.1.2. Clusters
The most common method of obtaining a DMS or DMO
material has been the incorporation of TM ions into a host
lattice.2,11,13,15,16,18,23,27,28,31,33–37,39,41–43,46–48,52,54,58,62,66,73–128 Nanoclusters of
the TM ion, which is frequently a magnetic element (eg. Co, Fe)
therefore form the most obvious candidates for impurity sources of
magnetism. If these clusters are small enough, the clustering results
in superparamagnetic behaviour which can be easily detected by
temperature dependent magnetisation measurements70,129 although if
formed from Co, the clusters only need to be between 4–8 nm for the
blocking temperature to be higher than room temperature.11 It may be
suggested that an easier dopant to work with would be one that could not
form ferromagnetic impurities such as Mn. However, even this dopant
has been suggested to cluster and be a source of spurious magnetic
moments in thin films of DMS and DMO material.12,130–132
Dilute Magnetic Oxides: Current Status and Prospects 227
Ferromagnetic coupling has been observed in very small clusters of Mn
below 68 K,131 although it has been argued as to whether Mn clustering
would result in an increased, or decreased Tc in DMS.130,132 The low level
of cluster impurity expected limits the ability of techniques such as XRD
to detect them. Nonetheless, owing to the extreme likelihood of TM
cluster formation, it is worth taking long XRD scans to determine
whether such clustering exists.70 In a recent study Zhou et al., showed
that clusters of certain TM ions were easily detected with conventional
XRD owing to their tendency to grow orientated with the host
lattice,70,133 this is shown in Fig. 10.1. For example, Co clusters can
form both hcp and fcc structures. Zhou et al., found that in ZnO,
Co clusters orientated with the ZnO host matrix such that hcp-
Co(0001)[1100]//ZnO(0001)[1100] and as a consequence, Co cluster
formation within ZnO was detectable by conventional XRD with the
hcp-Co(0002) diffraction peak being present at ~44° (Fig. 10.2). A
similar relation was found for Ni implanted ZnO133 indicating that both
Co and Ni can be detected by conventional XRD if the scan time is long
enough.
The situation however is quite different for Co implanted TiO2 where
the Co clusters are orientated at 49.82° compared with the crystal axis,
corresponding to (b) in Fig. 10.1. In this case conventional XRD was not
sufficient to detect the Co clusters and Zhou et al. needed synchrotron
XRD to detect the presence of the clusters. For Mn into Si (forming
MnSi1.7) and Fe into ZnO (bcc structure), the magnetic clusters had
random orientation with respect to the host lattice ((c) in Fig. 10.1).
Fig. 10.1. Schematic of the orientation of clusters in DMS systems with respect to the
substrate. (a) orientated, typical of Co and Ni in ZnO, (b) orientated but not with
substrate, typical of Co in TiO2, (c) random orientation, typical of Fe in ZnO, Reprinted
with permission from Ref. 70 (Copyright 2008, American Institute of Physics).
228 K. Yates
Fig. 10.2. (a) XRD diffraction of Co implanted ZnO as a function of ion implantation
dosage. (b) Syncrotron XRD of Co doped ZnO. Reprinted with permission from Ref. 70
(Copyright 2008, American Institute of Physics).
Again, conventional XRD failed to detect the clusters. However for all of
these samples in which cluster formation was observed, the temperature
dependent magnetic properties indicated superparamagnetic behaviour.
Therefore, even if the clusters were not indicated by conventional XRD,
the temperature dependent magnetisation curves should signify that a
degree of caution is necessary.
Transmission electron microscope (TEM) results are invaluable
for cluster identification. Figure 10.3 shows a high resolution TEM
image of a Mn5Ge3 cluster within a GeMn thin film. This system is of
significant interest because, although room temperature ferromagnetism
has been observed in nanowires of Mn doped Ge, FM is not observed
in thin films of similar composition.134 Kazakova et al.,134 showed
that in thin films, nanoclusters of Mn5Ge3 were formed (<10 nm
diameter) which were absent in the corresponding nanowires. In
Co doped ZnO, Ivill et al., used TEM to show that Co cluster
formation tended to occur near the film-substrate interface as shown
in Fig. 10.3(b).
Dilute Magnetic Oxides: Current Status and Prospects 229
(b)
Fig. 10.3. High resolution TEM of a Ge:Mn film showing Mn5Ge3 clusters. Reprinted
with permission from Ref. 134 (Copyright (2008) American Physical Society) (b) Co
clusters forming near the substrate-film boundary in Co doped ZnO, reproduced with
permission from Ref. 92, (copyright (2008) Institute of Physics Publishing).
However, observation of clusters by TEM requires the cross section
made by the TEM to be coincident with a region of nanoclusters. The
dilution of TM ions used to dope these DMS materials means that it is
quite conceivable that the cross section does not contain any such
impurities. Therefore, again, absence of evidence from TEM should not
be taken as evidence of absence.113,135
10.2.1.3. Solubility
Although the concentration of TM ions doped into host materials for
DMS/DMO research is very low, the solubility limit of the ions in the
semiconductor host can be even lower.136 These materials readily phase
segregate into a variety of possible magnetic phases that are challenging
to detect by techniques such as XRD. A summary of the more common
230 K. Yates
types of magnetic impurity and the detection methods used to isolate
them is given in Table 10.1.
Table 10.1. Dopants and possible impurities in DMO systems.
Dopant Impurity Magnetic properties Detection method Ref.
Co Co clusters FM, ~1373 K XRD, RHEED, 13, 98, 99,
Auger, TEM, 137
XAS, XMCD
Co metal at FM TEM 116
interfaces
ZnCo2O4 AFM if n-type Raman, XRD 138, 139
FM if p-type
(low Temp.)
Co3O4 AFM, Raman, HRTEM 53
TN 30–40 K140
CoO AFM XRD at high conc 34
CoAl2O4 AFM 92
Mn Mn clusters Variable 131, 132, 141
Mn2-xZnxO3-δ FM at rt Measurements as a 142
function of O
content
MnxN Predicted FM 12, 143, 144
Mn4N Ferrimagnetic, 145
Tc = 745 K
MnSi (Si from Complex Synchrotron 70
substrate) radiation XRD
ZnMnO3 Spin glass 100
Ni Ni clusters FM XRD, Synchrotron 70
radiation XRD
Fe Fe clusters FM Synchrotron 40, 70
radiation XRD,
Mössbauer
Fe3O4 Ferrimagnetic Mössbauer 146
FexO Variable XRD at Fe 34
concentrations
> 0.1
(ZnFe)3O4 FM, Tc = 440 K XRD at Fe 34
concentrations
> 0.1
Dilute Magnetic Oxides: Current Status and Prospects 231
Therefore, even if magnetic cluster formation is avoided, there
is still an intrinsic difficulty that many of the transition metal
ions used to dope these materials magnetically have a low solubility
threshold.30,34,41,47
Preparation methods and off-equilibrium growth techniques such as
pulsed laser deposition (PLD) may result in a greater incorporation of
magnetic dopant ions than equilibrium growth methods. Nonetheless, the
solubility of TM ions into the host materials is still relatively low and
close to that used in many studies. For bulk samples, solubility levels of
<10% have been reported.47,100 Riyadi et al., monitored the lattice
parameter shift with increasing Mn content to show that the solubility of
Mn into ZnO could even be as low as 6%.47 In bulk materials it has
also been shown that the ferrimagnetic phase (Mn,Zn)Mn2O4 is easily
formed.28 In contrast in thin films, Jin et al., reported that although the
equilibrium solubility of Mn in ZnO was ~20%30 it was up to 36% in
laser assisted processes far off equilibrium.23,30 For Co doping the
situation is fairly similar with very low solubility limits reported.53,147,148
Raman spectroscopy was used to show that the solubility of Co into ZnO
was only ~5% in films deposited by metal-organic decomposition,53 10%
in bulk material139 and up to ~15% in nanoalloys prepared by the
solvothermal technique147 or PLD,149 Prellier et al., showed that the
c-lattice parameter in Co doped ZnO obeyed Vegard’s law until Co
doping of 10% while Kim et al., measured a solubility of Co in ZnO
of 40%.
10.2.1.4. Spinels as secondary phases
A key phase that is formed beyond the solubility phase are the spinels
such as ZnCo2O4.150 This phase is known to be magnetic under certain
oxygenation conditions138 and so is a significant problem phase for DMS
materials doped with Co. The magnetic properties of ZnCo2O4 as a
function of oxygenation is shown in Fig. 10.4.
232 K. Yates
Fig. 10.4. As the oxygen conditions in the deposition chamber are varied ZnCo2O4 goes
from n type to p type and an appreciable magnetic moment is observed in the films at low
temperatures. Inset shows the Arrott plot for the p-type sample, taken with permission
from Ref. 138 (Copyright (2004) Wiley-VCH Verlag GmbH & Co. KGaA).
Again, the low concentration of the spinel limits the ability of
many detection methods to reliably confirm or eliminate its existence in
a film.92,151,152 It has been shown that for the spinel phase, Raman
spectroscopy is a very sensitive probe for small quantities of impurity
elements and detects this particular impurity form at concentrations
much less than that detectable by XRD.139,149,151 For example, Wang
et al., found that in Co doping into ZnO, the impurity spinel phase
ZnyCo3-yO4 was found by XRD at a dopant concentration of Co ≥ 0.391.
In contrast, from Raman spectroscopy, it was clear that this phase was
already present at Co concentrations of x ≥ 0.155.147 Figure 10.5(a)
shows the effect on the Raman spectra of increasing the Co dopant
concentration into ZnO in PLD grown thin films. Above a critical level
of doping (5% in Ref. 149, 10% in Ref. 139), Raman modes owing to
impurity phases can be observed. Figure 10.5(b) compares the Raman
spectrum of a PLD target of Co doped ZnO and the spinel ZnCo2O4
showing clear indications that the spinel is present in the target
material.139,151 This makes the Raman technique very attractive for
Dilute Magnetic Oxides: Current Status and Prospects 233
Fig. 10.5. (a) Raman spectra of Co doped ZnO as a function of Co concentration showing
development of impurity modes (labeled AM) (b) Raman spectrum of a Co doped ZnO
PLD target and the spinel ZnCo2O4 showing clear evidence of this impurity phase in
the target. Taken with permission from Ref. 139 (Copyright (2006) American Physical
Society).
detecting-possible spinel accumulation. For the n-type materials where
this spinel phase is FM, this may prove particularly useful.150 Raman
spectroscopy has also been used to search for the existence of the
magnetic impurity Co3O4.53,147,152
Although the similarity in Raman spectra between the spinel Co3O4
and the spinel ZnyCo3-yO4 has been noted.147,152 In all of these studies,
XRD had shown the samples to be single phase at low dopant
concentrations.
10.2.1.5. Other secondary phases
Even in the absence of specific precipitates such as Mn clusters, Co
clusters or the spinel phase, it has been reported that metastable phases
exist and are responsible for the observed magnetic phases. Such a
metastable phase is oxygen vacancy stabilised Mn2-xZnxO3-d which can
form at the interface of thin films.142 The ferromagnetism observed is
then not due to an intrinsic carrier mechanism but because of this
234 K. Yates
metastable phase.142 Similarly, uncompensated spins at the surfaces and
interfaces of thin films can result in the observation of a ferromagnetic
moment.110 Additional use of common experimental tools can indicate
the presence of metastable or impurity phases. Temperature dependent
magnetisation measurements indicate the presence (or absence) of
superparamagnetic metallic clusters while time dependent magnetisation
measurements can reveal information regarding the metastable phases.110,153
Furthermore, growing films of the same composition but varying the
thickness (or Ar-ion milling a film and making measurements as a
function of film thickness) can indicate whether interface effects and
defects are important to the ferromagnetic coupling mechanism.36,154–157
Other characterisation tools such as transport measurements can reveal
details regarding defects in the grains and grain boundaries,158 however,
the Anomalous Hall Effect (AHE) is not necessarily a good indicator of
intrinsic ferromagnetic mechanisms in these oxides,4 because AHE has
been observed in non-magnetic semiconductors that contain magnetic
secondary phases.114,159
Many other techniques such as magnetic circular dichromism (MCD),
X-ray absorption (emission) spectroscopy (XA(E)S), X-ray magnetic
circular dichromism (XMCD) and XANES give crucial information
regarding the valency and environment of the TM dopant ions and the
spin splitting of the band structure. XAS has recently been used to show
that in magnetic Co or Mn doped ZnO, the TM ion is in the 2+ valency16
while XMCD was used in TiO2, to detect the presence of Co metal
present in clusters.4,99 Also by XMCD, Tietze et al., were able to pinpoint
oxygen vacancies as the source of ferromagnetic behaviour in their
films.54
These results indicate that a combination of detection methods, are
required before a new material can be considered an intrinsic dilute
magnetic semiconductor and some crucial experimental checks (such as
cycling the substrate without depositing the film) need to be undertaken
routinely.160
Dilute Magnetic Oxides: Current Status and Prospects 235
10.3. Intrinsic Mechanisms for Magnetic Behaviour
Once all impurities have been accounted for, there remains a large
number of conflicting reports either of ferromagnetism1,4 or the absence
of ferromagnetism.11,34,47,73,100,122,153 Many studies have indicated that
oxide DMS were only magnetic in the absence of carriers, that is, when
the films were insulating.83,161 Theoretical predictions however had
indicated that carriers were necessary for FM behaviour and that p-type
ZnO in particular would be ferromagnetic at room temperature when
further doped with Mn.150 In the absence of p-type carriers, neither Co or
Mn doped ZnO were expected to be ferromagnetic.6 Nonetheless,
attempts to introduce carriers into the material were often unsuccessful,
partly due to the intrinsic difficulty of doping p-type carriers into ZnO6
and often resulted in non-magnetic films.31 Attempts at systematic
doping of ZnO with both p and n-type carriers did produce results that
agreed with theory.32 Nonetheless reports were contradictory, and
ferromagnetism at room temperature has now been observed in hole
doped and electron doped systems18,37,54,59,85,90,92,102,104,162–165 as well as
being absent in similarly doped and prepared materials.15,31 Results
therefore seem contradictory. In DMO materials, oxygen stoichiometry
has been shown to be of enormous importance in producing n-type
carriers.125 A partial explanation for the wide range of behaviours
observed on doping with carriers, in addition to TM ions, was suggested
by the results of Behan et al.15 By deliberately introducing Al as an
electron dopant with oxygen defects, Behan et al., showed that there
exist three distinct regions of magnetic behaviour as a function of carrier
concentration, Fig. 10.6.15 At low carrier concentrations (<2 x 1018 cm-3),
appreciable magnetic moments were observed in films doped with Co.
As the concentration of carriers increased, the magnetic moment
decreased until no magnetism was observed at room temperature
at intermediate carrier concentrations. As the carrier concentration
increased to beyond the Mott limit (>1 x 1020 cm-3), ferromagnetism at
room temperature was regained. Surprisingly, the results were similar for
films codoped with Al and Mn, as well as Al and Co, despite the fact
that n-type Mn doped ZnO is not expected to be ferromagnetic at
room temperature.6,9,150 However, as seen by the scatter on the data in
236 K. Yates
Fig. 10.6. For ZnO doped with Al, there are three regimes of magnetic behaviour as a
function of carrier concentration. Arrows indicate the magnetisation at low temperature
of two films. Taken with permission from Ref. 15 (Copyright (2008) American Physical
Society).
Fig. 10.6 and the contradictory reports of magnetism (or not) in similarly
prepared films, the situation is more complex.
10.3.1. Insulating regime
Both insulating DMS and DMOs have been reported. The absence of
free carriers, in these materials means that conventional carrier mediated
models for ferromagnetic exchange such as the Ruderman-Kittel-
Kasuya-Yosida (RKKY) interaction cannot be adopted to explain the
results.96 It was proposed that the bound magnetic polaron (BMP) model
could be used to explain the magnetic behaviour of the materials.160
Nonetheless, the Tc calculated with this model was far below that
observed. Recent experimental4,152,158 and theoretical6,83,161,166 works
have emphasised the role that defects play in mediating the FM exchange
in these DMD materials. The role of defects is further suggested by
the fact that, well characterised samples, or films which were better
Dilute Magnetic Oxides: Current Status and Prospects 237
structurally, showed either AFM or weaker FM coupling compared to
more disordered films.97,167 The magnetic properties are highly sensitive
to the growth conditions.
10.3.1.1. Theoretical treatments
The Bound Magnetic Polaron (BMP) model was first proposed to explain
the magnetic behaviour of the insulating material EuO168 and has since
been extended to deal with the specific situation of the dilute magnetic
semiconductors.160,169,170 In the theory, a shallow donor defect (such
as an oxygen vacancy) provides an electron which is confined to
a hydrogenic orbital of radius rm = εr(m/m*)a0 where εr is the high
frequency dielectric constant, m is the electron mass and m* the effective
mass of the donor electron. a0 is the Bohr radius.160 The exact εr will be
dependent on growth parameters but assuming reasonable values of εr,
the polaron size is estimated to be 7.6Å in ZnO, 4.8Å in TiO2 and 8.6Å
in SnO2.160 As the polaron density increases, bulk FM behaviour is
observed when the polarons are large enough and plentiful enough to
overlap. Although qualitatively there is experimental support for the
BMP model,53,120 quantitatively, the Tc predicted by the BMP model is
below that experimentally observed in these materials.
Theories have been developed that incorporate the intrinsic
irreproducibility of these materials to defects. The associated sensitivity
of the magnetism to materials preparation has even been suggested as
being integral to the ferromagnetic state of the DMS/DMO/DMD
materials.83,161,171,172 In these models, an oxygen vacancy-TM ion
complex allows a vacancy mediated indirect exchange mechanism
to develop that is long range and strong enough to provide room
temperature ferromagnetism.83,161 With only a 1% vacancy-TM ion
complex, the extended defect states lead to a band of donor levels. As in
the BMP model, the effective radius of the magnetic interaction is
governed by the extended defect states. However, in this case, as the
impurity band becomes full, the vacancy mediated exchange weakens
and so a maximum should be observed in both the Tc and the magnetic
saturation with increasing vacancy band occupation.83,161 Although the
238 K. Yates
theory was developed for TiO2, experimental support for the theory was
claimed in ZnO.125
10.3.1.2. Experimental results
In the insulating regime, where host semiconductors are doped only with
a single TM ion, the maximum saturation magnetisation is observed at
different doping levels for different TM ions.4 Specifically, the most
robust saturation magnetic moments in ZnO occur at 3–5% Co; 2 at%
Mn; 1 at% Cr, Ni or Cu doping.4 The differences between the ions were
associated with differences in the solubility of the TM ion. For Mn
doping into ZnO, low concentration led to above room temperature
ferromagnetism4 whereas higher concentrations led to FM ordering
below room temperature (around 45 K),4,173 presumably due to the
formation of magnetic impurities.
Qualitative agreement with the BMP model comes from a number of
studies16,17,40,52,66,95,167,174 and the low carrier concentration region of the
results of Ref. 15. Various studies have used carrier introduction as a
means to test the BMP theory.18 Chakraborti et al., showed that as the
carrier concentration increased, the resistivity decreased, as did the room
temperature FM.18 Similarly Ji et al., doped ZnO with Mn and the p-type
dopant Sb.95 All of these films were insulating at all temperatures and so
it was suggested that in accordance with theoretical expectations, p-type
carriers localised on the TM site form BMPs that result in the observed
room temperature ferromagnetism. Support for defect based models
comes from the fact that films of good crystallinity, with narrow XRD
rocking curves and in which there is evidence of TM ion substitution into
the lattice are frequently non-magnetic, antiferromagnetic, or at least less
magnetic than their disordered counterparts.4,39,97,158,175,176 Although
defects have been considered to be essential to the observation of
ferromagnetism in TM doped thin films4 the nature of these defects is
debated. Many studies have shown that room temperature magnetic
moments can be induced in previously non-magnetic films when the
film is subjected to a reducing anneal leading to the conclusion that
oxygen vacancies are a prime candidate defect for the development
Dilute Magnetic Oxides: Current Status and Prospects 239
of a ferromagnetic exchange.36,44,53,54,83,87,117,152,161 Evidence for oxygen
vacancies being of critical importance for FM in these DMDs also comes
from XMCD measurements. Tietze et al., used XMCD to show that in
Co doped ZnO, there was no FM signal on Co or O elements and
therefore that oxygen vacancies were the origin for the room temperature
ferromagnetism observed in their films.54 Similar results were observed
by Sudakar et al., using Raman spectroscopy.152 Sudakar et al., studied
the Raman spectra and magnetic behaviour of a 1.8% Co doped ZnO
film as a function of a series of vacuum anneals. The spectra are shown
in Fig. 10.7. A mode at ~690 cm-1, which is attributed to disorder
activated Raman scattering is alternatively switched on (off) when
annealed in air (vacuum). The activation of this mode with air annealing
suggests that it may be linked with oxygen vacancies. The films that
were air annealed (few oxygen vacancies) did not show room
temperature ferromagnetism, whereas the vacuum annealed films (many
oxygen vacancies) showed room temperature ferromagnetism with a
maximum value of 0.3 µ B/Co for Zn0.988Co0.012O at 300 K.152 Hydrogen
interstitials have been proposed as an alternative to oxygen defects,48,66
although other studies have indicated that oxygen vacancies of
themselves are not sufficient to activate the FM in Cr doped TiO2.97
Defects such as Zn interstitials have also been proposed to be activated
on annealing the samples.93,177,178 Defects that are produced not on
annealing the samples, but in the growth process have also been shown
to be of importance in determining which films show FM behaviour.
Song et al., studied the influence of varying the substrate on the magnetic
behaviour of Co doped ZnO.155 They found that the orientation of the
substrate could have considerable influence on the room temperature FM
properties indicating that the Co-O bond length was important in the
manifestation of room temperature FM. Similarly strain was shown to be
important by Liu et al., who showed that the films subjected to the
greatest tensile strain showed the largest saturation magnetizations.179
These results are consistent with theories suggesting oxygen vacancy-
TM ion complexes as a mediator in FM exchange.83,161
The role of defects in general to FM behaviour has been indicated by
a variety of other measurement techniques. XRD was used to monitor
240 K. Yates
structural defects and correlate these with magnetic behaviour and
growth conditions,115 while Impedance spectroscopy was used by
Hsu et al., to show that when films were annealed in Ar, defects
occurred mainly at the grain boundaries, whereas films annealed
in Ar-H introduced defects to both grains and grain boundaries.158
It was only films with defects in both grains and grain boundaries
that were strongly magnetic at room temperature.158 Raman spectroscopy
has also been used to correlate defect modes in the Raman
spectra with magnetic behaviour. Duan et al., showed that in ZnO
nanoparticles co-doped with Co and Mn the room temperature
ferromagnetism observed in the film was associated with a high
ratio of two particular defect modes to the main ZnO wurtzite mode
at 436 cm-1,180 suggesting a possible detection method for defect
mediated FM.
The results of Griffin et al., suggest an alternative reason for the
absence of FM in samples of good crystallinity.176 Using high resolution
TEM, Griffin et al., showed that Co doped TiO2 films of high
crystallinity had a tendency towards Co enrichment near surfaces (grain
boundaries, interfaces etc). In contrast, lower crystalline quality films
had a more homogenous Co distribution.176 These results indicate that
while DMD materials do show a marked dependency of magnetic
behaviour on precise details of the growth conditions, the defects
produced, and the effect that these have on the films magnetic character,
are far from understood.
10.3.1.3. “d 0” ferromagnetism
The observation of ferromagnetism in undoped, insulating oxide thin
films such as HfO2, TiO2 and In2O3 suggests the possibility of a new
form of magnetism, so called d0 magnetism.64 It is so called because of
the absence of d or f electrons which are traditionally thought to be
essential for ferromagnetism.
Dilute Magnetic Oxides: Current Status and Prospects 241
Fig. 10.7. The Raman spectra of Zn0.982Co0.018O films subjected to repeated annealing
in air (A) or vacuum (HV). * indicates substrate peaks while the arrow indicates the
presence of a mode attributed to disorder induced Raman scattering, reproduced with
permission from Ref. 152 (Copyright (2007) Institute of Physics Publishing).
Ferromagnetic behaviour at room temperature has now also been
observed in nanocrystalline MgO,181 GaN,182,183 CeO265 CdS, TiO2-d,
SnO2 and ZnO63,76,182 and thin films and nanocrystals of HfO210,64,67,68
and a number theoretical models have been proposed.184–187 The low
magnetic moment observed in these materials also means that caution
should be applied.10,71,72 Nonetheless, the study of Hong et al.,68 where
FM was observed at room temperature in In2O3 grown on MgO but not
242 K. Yates
on Al2O3 does suggest that some form of intrinsic, but defect mediated
ferromagnetism can occur in such wide gap semiconducting materials.
Again, there is debate about the role of oxygen deficiency in these
materials. Hong et al., and Coey et al., showed that oxygen deficiency
was crucial in HfO2 thin films and nanoparticles respectively and that
after annealing in oxygen the magnetic moment at room temperature was
reduced. Other, contradictory results have been observed by Liu et al.,
where it was size effects that were proposed to be imperative to
observing FM behaviour in nanoparticles of CeO2.65 For thin films of
these undoped materials, the defects were found to originate at the film-
substrate interface as the magnetic moment increased as a function of
decreasing film thickness in In2O3.68 Although of course such apparent
enhancement of the magnetic moment could result from extrinsic sources
of magnetic impurity as discussed earlier.3,71,72
10.3.2. Magnetism at high carrier concentrations
In order for DMO materials to be useful in spintronics devices it is
desirable that they can be used as spin injectors or sources of spin
polarised transport carriers,1 the absence of spin polarised carriers could
limit the spintronics applications of these materials.188 A carrier mediated
Double Exchange or RKKY type exchange was originally proposed
to explain the ferromagnetism in DMS and DMOs.37,150,157,160,189,190
As with data in the insulating regime, there is a large amount of
irreproducibility and variability in the data for films that are grown
with high carrier concentrations and lower resistivity. N-type oxides
are relatively easy to grow and electron dopants have included
oxygen vacancies32,125,176,191–194 and/or co-doping with H,48 Sn,90–92,195
Al,15,38,106,107 Ga,127 Ni12 and P.90 P would normally be considered a
p-type dopant, but was measured to be n-type in this case.90 Hole type
carriers are more difficult to introduce to materials such as ZnO6,109 but
p-type carriers have been introduced by nitrogen capping nanoparticles
or films,32,90,191 or doping with Cu,88 As102–104 or P.196 It has been
observed that the dopant ions are often not sufficient in themselves to
produce the carrier concentrations observed194 and additional carriers are
Dilute Magnetic Oxides: Current Status and Prospects 243
present which are thought to originate from oxygen defiency, Zn
interstitials or hydrogen donors.4 Thus structural defects are intrinsically
linked to carrier concentration and attempts to increase the latter
necessarily induce the former.4,108 It has been suggested that it would be
naïve to attribute to carrier concentration, magnetic effects that may be
more properly associated with defects.4 Experiments to distinguish
between the former and the latter have been attempted but have been
limited by the fact that not all studies of the effect of carrier
concentration on FM properties measure the carrier concentration. As
Behan et al., have shown,15 introducing only a small number of carriers
can actually decrease the resistivity and the magnetisation and it is only
when the carrier concentration reaches extremely high levels that room
temperature magnetisation is restored, Fig. 10.6. Even in this regime, the
scatter in the experimental data and the sensitivity of the properties
to the growth conditions implies that other effects must also play a
role.15,31,88,107 Nonetheless, evidence is emerging of carrier mediated
ferromagnetism in some of these materials. The observation of such a
carrier mediated mechanism and particularly the observation of a spin
polarised transport current would be critical for these material’s use in
spintronics applications where spin injection is required.
10.3.2.1. Theoretical review
It is beyond the scope of this review to cover basic concepts associated
with Double Exchange, RKKY etc. and the reader is referred to more
theoretical reviews of the subject for further information.5 There are
many theoretical models which are based on a need to introduce
carriers into a DMO in order that the DMO is FM and not AFM or spin-
glass.6,96,101,167 Hole type carriers are favoured in many studies.6,9,37,150,167
Risbud et al., found that in Co doped ZnO, the FM state was only
stabilised by hole doping (introduced by Zn intersitials) while electron
doping (oxygen vacancies) led to an AFM ground state.167 Carrier based
FM exchange dependent on electron doping has also been proposed96,101
where the ferromagnetism was attributed to a Double Exchange
mechanism. Lee et al., even suggested that the RKKY was inappropriate
in these materials101 and suggested a possible cause for the large
244 K. Yates
variability in FM properties. They proposed that to observe FM in Co
doped ZnO, a high concentration of Co is needed. Increasing Co content
decreases the electron-Co ratio. The FM properties are therefore a fine
balance between these two effects. The authors suggested that this, rather
than a defect mediated mechanism4 was the cause of the large amount of
variability in the reported experimental results.
Other studies have focussed on the means by which carriers can be
introduced to DMOs. It has been thought that oxygen vacancies produce
shallow donor defects in the host material that provide n-type carriers.
There are many experimental studies in which non-magnetic films are
subjected to a reducing anneal (to introduce n-type carriers) after which
the films become conducting and ferromagnetic.36,125 On annealing in an
oxygen atmosphere (decreasing the free n-type carrier concentration), the
films lose their ferromagnetism and become highly resistive thus the
ferromagnetism is claimed to be linked to carrier concentration.192
However, recent calculations have shown that in materials such as ZnO,
the oxygen vacancy is a deep donor, not a shallow donor.166,197–200
Janotti et al., proposed that, rather than the oxygen vacancies being
themselves responsible for the introduction of n-type carriers, instead,
the oxgen vacancies introduced by the reducing anneal, led to highly
reactive, free Zn bonds, which chemically bond to hydrogen.166,197,198
Their ab-initio calculations of H substitution into ZnO showed that
H substitution would form a shallow donor site in both ZnO and
MgO.166,197,198 Thus hydrogen substitution could explain the carrier
concentration dependence on partial pressure of oxygen during growth.
These theoretical calculations have recently received experimental
support by optical transmission measurements.199
10.3.2.2. Experimental results
Hole doping
Initially, the results of Kittilsveld et al., on carrier mediated FM
exchange in Mn or Co doped ZnO191 appeared to agree with the
theoretical predictions of Dietl et al.9 In this study, nitrogen capping
Dilute Magnetic Oxides: Current Status and Prospects 245
was used to introduce p-type carriers into TM doped ZnO. For ZnO thin
films doped with Mn, Kittilsveld et al., found that nitrogen capping
(p-type doping) was a necessary condition for observing room
temperature ferromagnetism. Conversely, Co doped ZnO showed the
opposite behaviour: room temperature ferromagnetism was only
observed in the presence of n-type carriers. There are now several
studies which have indicated that p-type carriers are necessary for
ferromagnetism to be observed in conducting Mn doped ZnO thin
films104,126,196 and that there is a correlation between the FM and the
carrier concentration.38,59,104,125,126,146,191,195,196 Lee et al., varied the p-type
carrier concentration in Mn doped ZnO by As implantation to obtain
carrier concentrations between 1 x 1017 cm-3 and 2.7 x 1018 cm-3.102,103
They found a direct correlation between carrier concentration and
room temperature magnetism suggesting a carrier mediated exchange.
However Hong et al., used Cu doping at 0% and 5% to introduce hole
carriers into Mn doped ZnO thin films grown by PLD.88 They found that
although hole doping had some effect on the room temperature magnetic
behaviour, the greater effect was obtained from defects eg. oxygen
vacancies, hence it was defects rather than carrier concentration that
enhanced the ferromagnetic exchange.
Nonetheless, although these studies have indicated that carrier
mediated ferromagnetism is possible, perhaps even preferred in Mn
doped ZnO, hole doping studies have also provided an insight into the
time dependence of the room temperature FM observed in these films.
Wan196 showed a clear correlation between the saturation magnetisation
at room temperature and the carrier concentration (which was up to
2.4 x 1016 cm-3) in P doped, p-type, Mn doped ZnO thin films. In this
study they showed that a lower oxygen environment was preferable for
the observation of ferromagnetism at room temperature as was a lower
growth temperature, although the lower growth temperature resulted in a
poorer structural quality. The maximum saturation magnetisation
observed was 0.35 µ B/Mn at room temperature. Over ten days, the
films were re-measured and the saturation magnetisation and carrier
concentration were shown to decrease while the resistivity, increased.
After ten days, the films were no longer ferromagnetic. A similar time
246 K. Yates
dependent behaviour was observed in Mn doped Indium tin oxide
(ITO)11 and in Ni implanted ZnO, where the FM was attributed to
defects.133
Electron doping
In order to make meaningful statements concerning the relation between
carrier density and magnetic behaviour it is instructive, indeed necessary,
to measure both. Claims in the literature regarding carrier concentration
are frequently made based on resistivity behaviour alone. Moreover, as
has been pointed out by Pan et al.,4 the ratio of the carrier density to the
dopant density can give important information concerning the defect
density. A measure of the carrier concentration is therefore of extreme
importance. N-type carriers can easily be introduced to materials such as
ZnO by a reducing anneal or using low oxygen partial pressure during
film growth.15,125,192,194 To introduce enough carriers into the lattice for
FM behaviour to be re-introduced, co-doping is required.15,107,194 It has
been shown that the n-type carrier concentration produced by doping
with Al produces more carriers than would be expected from ionisation
of the Al.194 So even in this case, low oxygen partial pressures are
required during the growth process in order to produce films with high
carrier concentrations. At carrier concentrations between ~3 x 1018 cm-3
and 1 x 1020 cm-3, the so-called intermediate regime, the relation between
the n-type carrier concentration and the magnetisation is non-
trivial15,90,91,194 and necessitates accurate determination of the carrier
concentration. At n-type carrier concentrations of <1020 cm-3, the
magnetisation may even decrease with increasing doping as shown by
Ivill et al., in Sn doped Mn:ZnO.90,91 The results in this regime have been
explained in terms of the bound magnetic polaron model.15,90,91 In the
theory, the ferromagnetic ordering between Mn moments is mediated by
holes. Adding Sn to the ZnO in order to dope electrons into the structure
cancels the free holes that can participate in this mediated exchange and
thereby reduces the ferromagnetic exchange.90–92 Similar behaviour was
found by Behan et al.15 whereby, in a certain region, introducing
n-type carriers into the material acts against the BMP ferromagnetic
Dilute Magnetic Oxides: Current Status and Prospects 247
mechanism and decreased the observed saturation magnetisation at room
temperature. The ferromagnetic state could be induced at low
temperature by freezing out the carriers as shown by the blue data points
in Fig. 10.6.
At high carrier concentrations (>1020 cm-3), room temperature FM is
restored in thin films of ZnO suggesting that a different, carrier mediated
FM exchange mechanism is at work.15 Several authors have indicated
that at high carrier concentrations, a FM phase is stable in n-type
films15,38,125,126,195,201 both for Co doped and Mn doped ZnO. Defects need
to be considered too and in a study of codoping ZnO with Al and Co,
Liu et al.107 showed that while the saturation magnetisation did increase
with carrier concentration (up to 2.2 x 1020 cm-3), the defects, as measured
by the ratio of carrier concentration to Al concentration also correlated
with the saturation magnetization.107 The authors suggested that the
ferromagnetic ordering was correlated with the structural defects rather
than the electron concentration. The situation was quite different for
Behan et al., who suggested that the dependence of the saturation
magnetisation at room temperature on the ratio of the free carriers to the
TM dopants was reminiscent of that observed in GaMnAs.202 Although
the scatter in the data indicated that growth conditions were critical in the
formation of a room temperature FM state, the clear indication was of a
carrier mediated, rather than defect assisted mechanism. An intrinsic
carrier mediated mechanism was also suggested by Yang et al.,59 in Mn
doped ZnO films codoped with Ga. In this case, the carrier concentration
was very high and the saturation magnetisation was directly correlated
with the carrier concentration. Intriguingly, for films of such a high
carrier concentration, the saturation magnetisation shows very little
temperature dependence between 4.2 K and 300 K, suggesting a very
high Tc (Fig. 10.8).
A further problem when measuring co-doped films, or films with a
high level of oxygen defects is the tendency to phase segregation.11,203 In
these studies, although there was an inverse correlation between the
magnetic behaviour of the films and the resistivity, this was due to
secondary phase formation rather than inherent ferromagnetism.
248 K. Yates
Fig. 10.8. Magnetisation loops at 300 K and 10 K in Mn doped ZnO co-doped with
Ga, inset shows a temperature independent saturation magnetisation, Reprinted with
permission from Ref. 59 (Copyright (2008), American Institute of Physics).
Measurements of spin polarisation, can these materials be used as spin
injectors?
One key question which will determine the suitability of these materials
for applications purposes will be whether the transport current in the
conducting DMO materials is spin polarised.1 There are various ways of
determining the spin polarisation of the transport current. Xu et al.,
measured the magnetoresistance of tunnel junctions made of Zn-CoO
and inferred a spin polarised current was injected from the Co doped
ZnO electrode.204 While Ji et al., found that the current that was
spin injected from Co doped ZnO was polarised by 7% at room
temperature.24,205 These devices suggest that, under certain conditions,
Co doped into ZnO carries a spin polarised current. A more direct
method of measuring the transport spin polarisation is to use point
Dilute Magnetic Oxides: Current Status and Prospects 249
contact Andreev reflection.206–208 Point contact Andreev reflection
(PCAR) spectroscopy uses the fact that in a spin polarised material, the
conductance across a junction between a sharpened superconducting
tip and the material is completely suppressed at voltages below the
superconducting gap voltage.206 The degree of suppression of the
conductance gives an indication of the transport spin polarisation in any
given material. For DMS, DMO and DMD research, the technique
suffers from two major drawbacks. Firstly the fact that it requires a
voltage drop across the contact between the superconductor tip and the
film, means that it is restricted to highly conducting films. Secondly, the
use of a superconducting tip necessitates that the measurements are
conducted below the superconducting transition temperature of the tip
material. Typically tips are Nb, Sn or Pb superconductors which means
the measurements have to be taken below 10 K. Nonetheless, the ability
of the technique to measure the actual transport spin polarisation and not
the spin polarisation of the density of states has meant that it has been
employed for a number of DMS materials. A spin polarised transport
current of 52 ± 3% was observed in InMnSb but not in InBeSb.209
Similarly, in GaMnSb a spin polarisation of 57 ± 5% was observed.210
Spin polarisation has also been detected using the Andreev technique in
GaMnAs.211 Very recently, point contact results have been obtained on
highly n-type ZnO codoped with Mn.60 In this technique, due to the
requirement for the film to have a low resistivity, PCAR measurements
are restricted to films in the high carrier regime of Fig. 10.6.
Nonetheless, in a highly conducting film (carrier concentration
= 1.32 x 1021 cm-3) and co-doped with Al and Mn, it was shown that this
material can support a spin polarised current at 4.2 K.60 A transport spin
polarisation of up to 55% was observed in contacts in which a surface
barrier had been punctured with the tip, Fig. 10.9.
In this figure, the quantity “1/Rn”, is a measure of the pressure
between the superconducting point contact and the DMS film. At low
pressures the observed transport spin polarisation is low but as the tip
250 K. Yates
60
Polarisation (%)
40
20
0
0.010 0.015 0.020 0.025 0.030
-1
1/RN (Ω )
Fig. 10.9. Polarisation as a function of tip pressure (1/Rn) in 2% Mn, 1% Al doped ZnO
(●) and 1% Al doped ZnO (●,▲), the line is a guide for the eye.
pressure increases (and a surface barrier is punctured), the polarisation
observed increases up to 55 ± 1%. In contrast, a non-magnetic film of
ZnO that had been doped only with Al showed no dependence of the
polarisation on tip pressure. These results suggest that the bulk of the
thin film of Mn and Al co-doped ZnO supports a spin polarised transport
current and therefore suggests a carrier mediated FM exchange
mechanism operates in the high carrier concentration regime. The
observation of a spin polarised transport current in highly n-type Mn
doped ZnO opens this material up as a potential spin injector for
spintronics devices, especially owing to the ease with which ZnO can be
incorporated into the existing semiconducting architecture.
10.4. Devices Already Made with DMS DMO and DMD Materials
It has long been an aim of the spintronics research community to obtain
spin injection into semiconducting materials.8 In order to use DMS and
Dilute Magnetic Oxides: Current Status and Prospects 251
DMO in device structures, it is desirable that they can act as spin
injectors and that they have a long spin diffusion length. Recent results
have suggested that spin polarised transport is supported in highly n-type
ZnO co-doped with a TM ion.25,26,60 While Ghosh et al., showed that the
spin lifetime of carriers in Ga doped (n-type) ZnO was long even at room
temperature.25,26 The spin lifetime was also shown to increase with
increasing electric field (Fig. 10.10) which has important consequences
for devices made out of ZnO as it would allow spin manipulation by
electrical fields rather than magnetic fields.26
The ability to make magnetic tunnel junctions (MTJs) with room
temperature DMO materials is also being investigated.24,115,204,205 Such
devices have been very successful using low temperature DMS materials
such as (Ga,Mn)As and reviews on spintronics devices based on
these materials are already appearing.8 Nonetheless, it is preferable
technologically for devices to operate at, or close to room temperature.
Using dc and rf magnetron sputtering Song et al., grew all-oxide
Fig. 10.10. Measurement of spin lifetime as a function of electric field and temperature in
Ga doped ZnO, reprinted with permission from Ref. 26 (Copyright (2008), American
Institute of Physics).
252 K. Yates
heterostructures of (Co doped ZnO)/ZnO/(Co doped ZnO) in order to
obtain a tunnel magnetoresistance through the ZnO barrier, Fig. 10.11.115
Song et al., obtained a TMR of 20.8% at 4 K and in 2T. Although a TMR
of 180% was observed at room temperature in Fe/MgO/Fe single crystal
structures,212 the results on ZnO structures are promising. Song et al.,
attributed the large TMR observed in their structures to interface control
which led to improved the temperature dependence of the TMR and
indeed they were still able to observe a TMR at 300 K, although the
magnitude of this effect had diminished to 0.35%. Xu et al. recently
used MTJs to show that spin polarised transport existed and could be
injected from Co doped ZnO at 5 K204 in agreement with PCAR results
on n-type Mn doped ZnO. Room temperature spin polarised injection
into ZnO has been theoretically predicted213 and experimentally
measured24,205 via spin valves made with Co doped ZnO as one of the
electrodes. Ji et al., showed that at 300 K the spin injection from Co
doped ZnO into ZnO thin films was of the order of 7% (compared with
11.7% at 90 K).24,205 There are a large number of devices now made with
DMO based structures,24,121,204,205,214,215 suggesting that there is hope
for significant improvement in room temperature spintronics devices
utilising these DMO materials.
Fig. 10.11. Device structures made out of an all ZnO epitaxial heterostructure, (a)
schematic, (b) SEM image of an array of MTJs. Reprinted with permission from
Ref. 115 (Copyright (2007) American Institute of Physics).
Dilute Magnetic Oxides: Current Status and Prospects 253
ZnO itself also has applications in photovoltaics,216 UV LEDs,217
piezoelectrics218 and electrochemical biosensors.219 The crossover
between the use of ZnO as a piezoelectric and doping it to be a
ferromagnetic DMD suggests that doped ZnO and similar DMDs may
play a role in the emerging field of multiferroics.220 Indeed, claims
of multiferroic behaviour (the coexistence of both ferroelectric and
antiferromagnetic or ferromagnetic behaviour) have already been made
by Lin et al., in Li and Co co-doped ZnO105 and Yang et al., in Cr doped
ZnO.221
10.5. Outlook
Although there has been a considerable amount of debate, there is
growing evidence to support the argument that subtle changes in growth
conditions can lead to large changes in magnetic behaviour in DMS and
DMO based materials. The role played by defects in mediating the
ferromagnetism is gradually being established, both when the material
is insulating and when it contains free carriers, although much work
remains to be done. There are strong suggestions that two types of FM
exchange exist in these materials as a function of carrier concentration.
At low and intermediate carrier concentrations, some form of BMP
model appears to adequately explain the experimental results, in
combination with the effects on the BMP model of defects. At high
carrier concentrations, a carrier mediated FM exchange appears likely
although the role of growth conditions and defects in this regime still
needs to be established. In particular, spin polarised transport current has
been observed in highly n-type FM ZnO both by direct measurements
with point contact Andreev reflection and by inference from magnetic
tunnel junction structures. These results mean that this material has
enormous potential in spintronic devices. Even in the DMD regime, the
combination of ferromagnetic behaviour with optical devices and/or
piezoelectric properties means that these DMD materials also have
high device potential. Nonetheless, measurements with a variety of
characterisation techniques, especially including measuring the number
and type of free carriers (in dilute magnetic conducting materials) will be
254 K. Yates
invaluable in further elucidating the nature of the ferromagnetic
exchange in these fascinating systems and in making it controllable so
that it can be more easily incorporated into device structures.
Acknowledgements
The author is grateful for helpful discussions with colleagues and in
particular to K. Morrison, W. R. Branford and L. F. Cohen for careful
reading of the manuscript.
References
1. S. J. Pearton, C. R. Abernathy, D. P. Norton, et al., Mater. Sci.
Eng.: R: Reports 40 (2003) 137.
2. R. Janisch, P. Gopal, and N. A. Spaldin, J. Phys.: Condens. Matter
17 (2005) R657.
3. J. M. D. Coey, Curr. Opin. Sol. St. 10 (2006) 83.
4. F. Pan, C. Song, X. J. Liu, et al., Mater. Sci. Eng.: R: Reports 62
(2008) 1.
5. T. Dietl, J. Phys.: Condens. Matter 19 (2007) 165204.
6. N. A. Spaldin, Phys. Rev. B 69 (2004) 125201.
7. A. M. Nazmul, S. Sugahara, and M. Tanaka, Phys. Rev. B 67
(2003) 241308.
8. C. K. P. Gould, G. Schmidt, L. W. Molenkamp, Adv. Mater. 19
(2007) 323.
9. T. Dietl, H. Ohno, F. Matsukura, et al., Sci. 287 (2000) 1019.
10. D. W. Abraham, M. M. Frank, S. Guha, Appl. Phys. Lett. 87 (2005)
252502.
11. M. Venkatesan, P. Stamenov, L. S. Dorneles, et al., Appl. Phys.
Lett. 90 (2007) 242508.
12. L. M. Sandratskii, P. Bruno, and S. Mirbt, Phys. Rev. B (Condens.
Matter Mater. Phys.) 71 (2005) 045210.
13. S. A. Chambers, T. Droubay, C. M. Wang, et al., Appl. Phys. Lett.
82 (2003) 1257.
14. A. J. Behan, J. R. Neal, R. M. Ibrahim, et al., J. Magn. Magn.
Mater. 310 (2007) 2158.
Dilute Magnetic Oxides: Current Status and Prospects 255
15. A. J. Behan, A. Mokhtari, H. J. Blythe, et al., Phys. Rev. Lett. 100
(2008) 047206.
16. F. Bondino, K. B. Garg, E. Magnano, et al., J. Phys.: Condens.
Matter 20 (2008) 275205.
17. E. Cespedes, G. R. Castro, et al., J. Phys.: Condens. Matter 20
(2008) 095207.
18. D. Chakraborti, S. Ramachandran, G. Trichy, et al., J. Appl. Phys.
101 (2007) 053918.
19. W. M. Chen, I. A. Buyanova, A. Murayama, et al., Appl. Phys.
Lett. 92 (2008) 092103.
20. H. Frenzel, H. V. Wenckstern, A. Weber, et al., Physics of
Semiconductors, Pts A and B, AIP Conference Proceedings, 893
(2007) 301.
21. H. Frenzel, H. von Wenckstern, A. Weber, et al., Phys. Rev. B 76
(2007) 035214.
22. H. Frenzel, A. Lajn, M. Brandt, et al., Appl. Phys. Lett. 92 (2008)
92108.
23. T. Fukumura, Z. Jin, M. Kawasaki, et al., Appl. Phys. Lett. 78
(2001) 958.
24. J. Gang, S.-S. Yan, Y.-X. Chen, et al., Chin. Phys. Lett. 23 (2006)
446.
25. S. Ghosh, V. Sih, W. H. Lau, et al., Appl. Phys. Lett. 86 (2005)
232507.
26. S. Ghosh, D. W. Steuerman, B. Maertz, et al., Appl. Phys. Lett. 92
(2008) 162109.
27. S.-J. Han, J. W. Song, C.-H. Yang, et al., Appl. Phys. Lett. 81
(2002) 4212.
28. S.-J. Han, T.-H. Jang, Y. B. Kim, et al., Appl. Phys. Lett. 83 (2003)
920.
29. Y. Ishida, J. I. Hwang, M. Kobayashi, et al., Phys. B-Condens.
Matter 351 (2004) 304.
30. Z. Jin, T. Fukumura, M. Kawasaki, et al., Appl. Phys. Lett. 78
(2001) 3824.
31. T. C. Kaspar, T. Droubay, S. M. Heald, et al., New J. Phys. 10,
(2008) 055010.
32. K. R. Kittilstved and D. R. Gamelin, J. Appl. Phys. 99 (2006)
08M112.
256 K. Yates
33. M. Kobayashi, Y. Ishida, J. I. Hwang, et al., Phys. Rev. B 72
(2005) 201201.
34. S. Kolesnik, B. Dabrowski, and J. Mais, J. Appl. Phys. 95 (2004)
2582.
35. H.-J. Lee, S.-Y. Jeong, C. R. Cho, et al., Appl. Phys. Lett. 81
(2002) 4020.
36. X.-C. Liu, E.-W. Shi, Z.-Z. Chen, et al., J. Phys.: Condens. Matter
20 (2008) 025208.
37. D. Maouche, P. Ruterana, and L. Louail, Phys. Lett. A 365 (2007)
231.
38. J. R. Neal, A. J. Behan, R. M. Ibrahim, et al., Phys. Rev. Lett. 96
(2006) 197208.
39. K. Nielsen, S. Bauer, M. Lubbe, et al., Phys. Status Solidi a-Appl.
Mater. Sci. 203 (2006) 3581.
40. K. Potzger, S. Zhou, H. Reuther, et al., Appl. Phys. Lett. 88 (2006)
052508.
41. W. Prellier, A. Fouchet, B. Mercey, et al., Appl. Phys. Lett. 82
(2003) 3490.
42. A. Quesada, M. A. García, J. de la Venta, et al., The European
Phys. J. B - Condens. Matter and Complex Systems 59 (2007) 457.
43. S. Ramachandran, A. Tiwari, J. Narayan, et al., Appl. Phys. Lett. 87
(2005) 172502.
44. S. Ramachandran, J. Narayan, and J. T. Prater, Appl. Phys. Lett. 88
(2006) 242503.
45. S. Ramachandran, J. T. Prater, N. Sudhakar, et al., Sol. Stat.
Comm. 145 (2008) 18.
46. C. N. R. Rao and F. L. Deepak, J. Mater. Chem. 15 (2005) 573.
47. S. Riyadi, Muafif, A. A. Nugroho, et al., J. Phys.-Condens. Matter
19 (2007) 476214.
48. B. K. Roberts, A. B. Pakhomov, and K. M. Krishnan, J. Appl.
Phys. 103 (2008) 07D133.
49. B. K. Roberts, A. B. Pakhomov, P. Voll, et al., Appl. Phys. Lett. 92
(2008) 162511.
50. K. Rode, A. Anane, R. Mattana, et al., J. Appl. Phys. 93 (2003)
7676.
51. P. Sharma, A. Gupta, K. V. Rao, et al., Nat. Mater. 2 (2003) 673.
Dilute Magnetic Oxides: Current Status and Prospects 257
52. P. Singh, Deepak, R. N. Goyal, et al., J. Phys.: Condens. Matter 20
(2008) 315005.
53. J. S. Thakur, G. W. Auner, V. M. Naik, et al., J. Appl. Phys. 102
(2007) 093904.
54. T. Tietze, M. Gacic, G. Schütz, et al., New J. Phys. 10 (2008)
055009.
55. K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett. 79 (2001) 988.
56. L. H. Van, J. Ding, M. H. Hong, et al., Surf. Rev. Lett. 15 (2008)
81.
57. X. J. Wang, I. A. Buyanova, W. M. Chen, et al., J. Appl. Phys. 103
(2008) 023712.
58. L. Yan, C. K. Ong, and X. S. Rao, J. Appl. Phys. 96 (2004) 508.
59. Z. Yang, J. L. Liu, M. Biasini, et al., Appl. Phys. Lett. 92 (2008)
042111.
60. K. A. Yates, A. J. Behan, J. R. Neal, et al., Phys. Rev. Lett.
submitted (2008).
61. Y.-Z. Yoo, T. Fukumura, Z. Jin, et al., J. Appl. Phys. 90 (2001)
4246.
62. Y. Zheng, J. C. Boulliard, D. Demaille, et al., J. Cryst. Growth 274
(2005) 156.
63. N. H. Hong, N. Poirot, and J. Sakai, Phys. Rev. B (Condens. Matter
Mater. Phys.) 77 (2008) 033205.
64. J. M. D. Coey, M. Venkatesan, P. Stamenov, et al., Phys. Rev. B
(Condens. Matter Mater. Phys.) 72 (2005) 024450.
65. Y. Liu, Z. Lockman, A. Aziz, et al., J. Phys.: Condens. Matter 20
(2008) 165201.
66. S. Ghoshal and P. S. A. Kumar, J. Phys.: Condens. Matter 20
(2008) 192201.
67. M. Venkatesan, C. B. Fitzgerald, and J. M. D. Coey, Nat. 430
(2004) 630.
68. N. H. Hong, J. Sakai, N. Poirot, et al., Phys. Rev. B (Condens.
Matter Mater. Phys.) 73 (2006) 132404.
69. D. M. Edwards and M. I. Katsnelson, J. Phys.: Condens. Matter 18,
(2006) 7209.
70. S. Zhou, K. Potzger, G. Talut, et al., in Proceedings of the 52nd
Annual Conference on Magnetism and Magnetic Materials (AIP,
2008), Vol. 103, p. 07D530.
258 K. Yates
71. F. Golmar, A. M. M. Navarro, C. E. R. Torres, et al., Appl. Phys.
Lett. 92 (2008) 262503.
72. Y. Belghazi, G. Schmerber, S. Colis, et al., Appl. Phys. Lett. 89
(2006) 122504.
73. J. Alaria, H. Bieber, S. Colis, et al., Appl. Phys. Lett. 88 (2006)
112503.
74. K. Ando, Sci. 312 (2006) 1883.
75. K. Ando, H. Saito, Z. Jin, et al., Appl. Phys. Lett. 78 (2001) 2700.
76. S. Banerjee, K. Rajendran, N. Gayathri, et al., J. Appl. Phys. 104
(2008) 043913.
77. A. Barla, G. Schmerber, E. Beaurepaire, et al., Phys. Rev. B
(Condens. Matter Mater. Phys.) 76 (2007) 125201.
78. K. P. Bhatti, S. Kundu, S. Chaudhary, et al., J. Phys. D: Appl. Phys.
39 (2006) 4909.
79. K. S. Burch, D. B. Shrekenhamer, E. J. Singley, et al., Phys. Rev.
Lett. 97 (2006) 087208.
80. I. A. Buyanova, W. M. Chen, M. P. Ivill, et al., J. Vac. Sci.
Technol. B: Microelectron. Nanometer Struct. 24 (2006) 259.
81. S. A. Chambers, Surf. Sci. Reports 61 (2006) 345.
82. S. A. Chambers, S. Thevuthasan, R. F. C. Farrow, et al., Appl.
Phys. Lett. 79 (2001) 3467.
83. G. Cohen, V. Fleurov, and K. Kikoin, in 10th Joint MMM/Intermag
Conference (AIP, Baltimore, Maryland (USA), 2007), Vol. 101,
p. 09H106.
84. D. R. G. D. A. Schwartz, Adv. Mater. 16 (2004) 2115.
85. P. Dutta, M. S. Seehra, Y. Zhang, et al., J. Appl. Phys. 103 (2008)
07D104.
86. T. Fukumura, H. Toyosaki, K. Ueno, et al., New J. Phys. 10 (2008)
55018.
87. M. Gacic, G. Jakob, C. Herbort, et al., Phys. Rev. B (Condens.
Matter Mater. Phys.) 75 (2007) 205206.
88. N. H. Hong, V. Brize, and J. Sakai, Appl. Phys. Lett. 86 (2005)
082505.
89. G. J. Huang, J. B. Wang, X. L. Zhong, et al., J. Mater. Sci. 42
(2007) 6464.
90. M. Ivill, S. J. Pearton, Y. W. Heo, et al., J. Appl. Phys. 101 (2007)
123909.
Dilute Magnetic Oxides: Current Status and Prospects 259
91. M. Ivill, S. J. Pearton, D. P. Norton, et al., J. Appl. Phys. 97 (2005)
053904.
92. M. Ivill, S. J. Pearton, S. Rawal, et al., New J. Phys. 10 (2008) 10.
93. N. K. J. L. MacManus-Driscoll, Y. Liu, M. E. Vickers, Adv. Mater.
19 (2007) 2925.
94. O. D. Jayakumar, I. K. Gopalakrishnan, K. Shashikala, et al., Appl.
Phys. Lett. 89 (2006) 202507.
95. G.-H. Ji, Z.-B. Gu, M.-H. Lu, et al., J. Phys.: Condens. Matter 20
(2008) 425207.
96. E.-J. Kan, L.-F. Yuan, and J. Yang, J. Appl. Phys. 102 (2007)
033915.
97. T. C. Kaspar, S. M. Heald, C. M. Wang, et al., Phys. Rev. Lett. 95
(2005) 217203.
98. R. J. Kennedy, P. A. Stampe, E. Hu, et al., Appl. Phys. Lett. 84
(2004) 2832.
99. J.-Y. Kim, J.-H. Park, B.-G. Park, et al., Phys. Rev. Lett. 90 (2003)
017401.
100. S. Kolesnik, B. Dabrowski, and J. Mais, J. Supercond. 15 (2002)
251.
101. E.-C. Lee and K. J. Chang, Phys. Rev. B 69 (2004) 085205.
102. S. Lee, D. Y. Kim, Y. Shon, et al., Appl. Phys. Lett. 89 (2006)
022120.
103. S. Lee, Y. Shon, T. W. Kang, et al., Appl. Phys. Lett. 93 (2008)
022113.
104. S. W. Lim, M. C. Jeong, M. H. Ham, et al., Japn. J. Appl. Phys.
Part 2-Lett. 43 (2004) L280.
105. Y.-H. Lin, M. Ying, M. Li, et al., Appl. Phys. Lett. 90 (2007)
222110.
106. X.-C. Liu, E.-W. Shi, Z.-Z. Chen, et al., Appl. Phys. Lett. 88 (2006)
252503.
107. X.-C. Liu, E.-W. Shi, Z.-Z. Chen, et al., J. Cryst. Growth 307
(2007) 14.
108. X. J. Liu, C. Song, F. Zeng, et al., J. Phys.: Condens. Matter 19
(2007) 296208.
109. D. C. Look, D. C. Reynolds, C. W. Litton, et al., Appl. Phys. Lett.
81 (2002) 1830.
260 K. Yates
110. A. Ney, R. Rajaram, S. S. P. Parkin, et al., Phys. Rev. B (Condens.
Matter Mater. Phys.) 76 (2007) 035205.
111. H. H. Nguyen, W. Prellier, J. Sakai, et al., (AIP, 2004), Vol. 95,
p. 7378.
112. U. Ozgur, Y. I. Alivov, C. Liu, et al., J. Appl. Phys. 98 (2005)
041301.
113. S. Ramachandran, A. Tiwari, and J. Narayan, Appl. Phys. Lett. 84
(2004) 5255.
114. S. R. Shinde, S. B. Ogale, S. Das Sarma, et al., Phys. Rev. B 67
(2003) 115211.
115. C. Song, X. J. Liu, F. Zeng, et al., Appl. Phys. Lett. 91 (2007)
042106.
116. P. A. Stampe, R. J. Kennedy, Y. Xin, et al., J. Appl. Phys. 93
(2003) 7864.
117. C. Sudakar, P. Kharel, G. Lawes, et al., in 10th Joint
MMM/Intermag Conference (AIP, Baltimore, Maryland (USA),
2007), Vol. 101, p. 09H118.
118. M. Tay, Y. Wu, G. C. Han, et al., J. Appl. Phys. 100 (2006)
063910.
119. N. Theodoropoulou, V. Misra, J. Philip, et al., J. Magn. Magn.
Mater. 300 (2006) 407.
120. M. Tortosa, M. Mollar, B. Mari, et al., J. Appl. Phys. 104 (2008)
033901.
121. H. Toyosaki, T. Fukumura, K. Ueno, et al., (AIP, 2006), Vol. 99,
p. 08M102.
122. S. C. Wi, J. S. Kang, J. H. Kim, et al., Appl. Phys. Lett. 84 (2004)
4233.
123. A. Wojcik, M. Godlewski, E. Guziewicz, et al., Appl. Phys. Lett.
90 (2007) 082502.
124. H. Y. Xu, Y. C. Liu, C. S. Xu, et al., Appl. Phys. Lett. 88 (2006)
242502.
125. X. H. Xu, H. J. Blythe, M. Ziese, et al., New J. Phys. 8 (2006) 135.
126. W. Yan, Z. Sun, Q. Liu, et al., Appl. Phys. Lett. 90 (2007) 242509.
127. S. G. Yang, T. Li, B. X. Gu, et al., Appl. Phys. Lett. 83 (2003)
3746.
128. S. G. Yang, A. B. Pakhomov, S. T. Hung, et al., IEEE Trans.
Magn. 38 (2002) 2877.
Dilute Magnetic Oxides: Current Status and Prospects 261
129. S. Zhou, K. Potzger, G. Zhang, et al., J. Appl. Phys. 100 (2006)
114304.
130. S. H. Xiong and Y. W. Du, Phys. Lett. A 372 (2008) 2114.
131. M. B. Knickelbein, Phys. Rev. Lett. 86 (2001) 5255-5257.
132. M. R. Pederson, F. Reuse, and S. N. Khanna, Phys. Rev. B 58
(1998) 5632-5636.
133. S. Zhou, K. Potzger, J. von Borany, et al., Phys. Rev. B (Condens.
Matter Mater. Phys.) 77 (2008) 035209.
134. O. Kazakova, R. Morgunov, J. Kulkarni, et al., Phys. Rev. B
(Condens. Matter Mater. Phys.) 77 (2008) 235317.
135. C. Song, K. W. Geng, F. Zeng, et al., Phys. Rev. B (Condens.
Matter Mater. Phys.) 73 (2006) 024405.
136. J. Robin, Ann. Chim. 10 (1955) 400.
137. S. R. Shinde, S. B. Ogale, J. S. Higgins, et al., Phys. Rev. Lett. 92
(2004) 166601.
138. H. Kim, Phys. Stat. Sol. 241 (2004) 1553.
139. K. Samanta, P. Bhattacharya, R. S. Katiyar, et al., Phys. Rev. B
(Condens. Matter Mater. Phys.) 73 (2006) 245213.
140. P. Dutta, M. S. Seehra, S. Thota, et al., J. Phys.: Condens. Matter
20 (2008) 015218.
141. P. Bobadova-Parvanova, K. A. Jackson, S. Srinivas, et al., J. Chem.
Phys. 122 (2005) 014310.
142. D. C. Kundaliya, S. B. Ogale, S. E. Lofland, et al., Nat. Mater. 3
(2004) 709.
143. B. K. Rao and P. Jena, Phys. Rev. Lett. 89 (2002) 185504.
144. M. Zajac, J. Gosk, E. Grzanka, et al., J. Appl. Phys. 93 (2003) 4715.
145. W. J. Takei, R. R. Heikes, and G. Shirane, Phys. Rev. 125 (1962)
1893 LP.
146. M. Venkatesan, R. D. Gunning, P. Stamenov, et al., in Proceedings
of the 52nd Annual Conference on Magnetism and Magnetic
Materials (AIP, Tampa, Florida (USA), 2008), Vol. 103, p. 07D135.
147. X. Wang, J. Xu, X. Yu, et al., Appl. Phys. Lett. 91 (2007) 031908.
148. J. H. Kim, H. Kim, D. Kim, et al., Phys. B-Condens. Matter 327
(2003) 304.
149. H. Harima, J. Phys.: Condens. Matter 16 (2004) S5653.
150. T. Dietl, Phys. E-Low-Dimensional Sys. & Nanostruct. 35 (2006)
293.
262 K. Yates
151. K. Samanta, P. Bhattacharya, and R. S. Katiyar, Phys. Rev. B
(Condens. Matter Mater. Phys.) 75 (2007) 035208.
152. C. Sudakar, P. Kharel, G. Lawes, et al., J. Phys.: Condens. Matter
19 (2007) 026212.
153. A. Ney, R. Rajaram, T. Kammermeier, et al., J. Phys.: Condens.
Matter 20 (2008) 285222.
154. A. Espinosa, E. Cespedes, C. Prieto, et al., in Proceedings of the
52nd Annual Conference on Magnetism and Magnetic Materials
(AIP, Tampa, Florida (USA), 2008), Vol. 103, p. 07D129.
155. C. Song, F. Zeng, K. W. Geng, et al., Phys. Rev. B (Condens.
Matter Mater. Phys.) 76 (2007) 045215.
156. C. Song, S. N. Pan, X. J. Liu, et al., J. Phys.: Condens. Matter 19
(2007) 176229.
157. M. A. Garcia, M. L. Ruiz-Gonzalez, A. Quesada, et al., Phys. Rev.
Lett. 94 (2005) 217206.
158. H. S. Hsu, J. C. A. Huang, S. F. Chen, et al., Appl. Phys. Lett. 90
(2007) 102506.
159. Q. Y. Xu, H. Schmidt, S. Q. Zhou, et al., Appl. Phys. Lett. 92
(2008) 082508.
160. J. M. D. Coey, M. Venkatesan, and C. B. Fitzgerald, Nat. Mater. 4
(2005) 173.
161. K. Kikoin and V. Fleurov, Phys. Rev. B (Condens. Matter Mater.
Phys.) 74 (2006) 174407.
162. S. Halm, P. E. Hohage, E. Neshataeva, et al., Phys. Stat. Solidi a-
Appl. Mater. Sci. 204 (2007) 191.
163. H. Toyosaki, T. Fukumura, Y. Yamada, et al., Nat. Mater. 3 (2004)
221.
164. Z. L. Lu, G. Q. Yan, S. Wang, et al., J. Appl. Phys. 104 (2008)
033919.
165. T. Schallenberg and H. Munekata, Appl. Phys. Lett. 89 (2006)
042507.
166. A. Janotti and C. G. Van de Walle, Nat. Mater. 6 (2007) 44.
167. A. S. Risbud, N. A. Spaldin, Z. Q. Chen, et al., Phys. Rev. B 68
(2003) 205202.
168. J. B. Torrance, M. W. Shafer, and T. R. McGuire, Phys. Rev. Lett.
29 (1972) 1168 LP.
169. D. E. Angelescu and R. N. Bhatt, Phys. Rev. B 65 (2002) 075211.
Dilute Magnetic Oxides: Current Status and Prospects 263
170. A. C. Durst, R. N. Bhatt, and P. A. Wolff, Phys. Rev. B 65 (2002)
235205.
171. B. W. Wessels, New J. Phys. 10 (2008) 055008.
172. T. Chanier, I. Opahle, M. Sargolzaei, et al., Phys. Rev. Lett. 100
(2008) 026405.
173. S. W. Jung, S.-J. An, G.-C. Yi, et al., Appl. Phys. Lett. 80 (2002)
4561.
174. K. G. Roberts, M. Varela, S. Rashkeev, et al., Phys. Rev. B
(Condens. Matter Mater. Phys.) 78 (2008) 014409.
175. R. Seshadri, Curr. Opin. Sol. Stat. Mater. Sci. 9 (2005) 1.
176. K. A. Griffin, M. Varela, S. J. Pennycook, et al., (AIP, 2006),
Vol. 99, p. 08M114.
177. G. Weyer, H. P. Gunnlaugsson, R. Mantovan, et al., J. Appl. Phys.
102 (2007) 113915.
178. N. Khare, M. J. Kappers, M. Wei, et al., Adv. Mater. 18 (2006)
1449.
179. X. J. Liu, C. Song, F. Zeng, et al., J. Appl. Phys. 103 (2008)
093911.
180. L. B. Duan, G. H. Rao, Y. C. Wang, et al., J. Appl. Phys. 104
(2008) 013909.
181. J. Hu, Z. Zhang, M. Zhao, et al., Appl. Phys. Lett. 93 (2008)
192503.
182. C. Madhu, A. Sundaresan, and C. N. R. Rao, Phys. Rev. B
(Condens. Matter Mater. Phys.) 77 (2008) 201306.
183. J. Hong, J. Appl. Phys. 103 (2008) 063907.
184. P. Dev, Y. Xue, and P. Zhang, Phys. Rev. Lett. 100 (2008) 117204.
185. D.-Y. Cho, J.-M. Lee, S.-J. Oh, et al., Phys. Rev. B (Condens.
Matter Mater. Phys.) 76 (2007) 165411.
186. P. Larson and S. Satpathy, Phys. Rev. B (Condens. Matter Mater.
Phys.) 76 (2007) 245205.
187. V. Pardo and W. E. Pickett, Phys. Rev. B (Condens. Matter Mater.
Phys.) 78 (2008) 134427.
188. P. Stamenov, M. Venkatesan, L. S. Dorneles, et al., in 50th Annual
Conference on Magnetism and Magnetic Materials (AIP, San Jose,
California (USA), 2006), Vol. 99, p. 08M124.
189. K. Sato, Jpn. J. Appl. Phys. 39 (2000) L555.
264 K. Yates
190. A. F. Jalbout, H. Chen, and S. L. Whittenburg, Appl. Phys. Lett. 81
(2002) 2217.
191. K. R. Kittilstved, N. S. Norberg, and D. R. Gamelin, Phys. Rev.
Lett. 94 (2005) 147209.
192. A. C. Tuan, J. D. Bryan, A. B. Pakhomov, et al., Phys. Rev. B 70
(2004) 054424.
193. Y. W. Heo, M. P. Ivill, K. Ip, et al., Appl. Phys. Lett. 84 (2004)
2292.
194. A. Mokhtari, New J. Phys. submitted (2008).
195. D. P. Norton, S. J. Pearton, A. F. Hebard, et al., Appl. Phys. Lett.
82 (2003) 239.
196. Q. Wan, Appl. Phys. Lett. 89 (2006) 082515.
197. A. Janotti and C. G. V. de Walle, Phys. Rev. B 75 (2007) 121201.
198. A. Janotti and C. G. V. de Walle, Phys. Rev. B 76 (2007) 165202.
199. F. A. Selim, M. H. Weber, D. Solodovnikov, et al., Phys. Rev. Lett.
99 (2007) 085502.
200. L. S. Vlasenko and G. D. Watkins, Phys. B-Condens. Matter 376
(2006) 677.
201. B. Yang, A. Kumar, P. Feng, et al., Appl. Phys. Lett. 92 (2008)
233112.
202. A. Chattopadhyay, S. Das Sarma, and A. J. Millis, Phys. Rev. Lett.
87 (2001) 227202.
203. S. Lee, H. S. Lee, S. J. Hwang, et al., J. Cryst. Growth 286 (2006)
223.
204. Q. Xu, L. Hartmann, S. Zhou, et al., Phys. Rev. Lett. 101 (2008)
076601.
205. G. Ji, S. S. Yan, Y. X. Chen, et al., J. Mater. Sci. Technol. 24
(2008) 415.
206. R. J. Soulen, J. M. Byers, M. S. Osofsky, et al., Sci. 282 (1998) 85.
207. G. E. Blonder, M. Tinkham, and T. M. Klapwijk, Phys. Rev. B 25
(1982) 4515 LP.
208. I. I. Mazin, A. A. Golubov, and B. Nadgorny, J. Appl. Phys. 89
(2001) 7576.
209. R. P. Panguluri, B. Nadgorny, T. Wojtowicz, et al., Appl. Phys.
Lett. 84 (2004) 4947.
210. R. P. Panguluri, B. Nadgorny, T. Wojtowicz, et al., Appl. Phys.
Lett. 91 (2007) 252502.
Dilute Magnetic Oxides: Current Status and Prospects 265
211. R. P. Panguluri, K. C. Ku, T. Wojtowicz, et al., Phys. Rev. B 72
(2005) 054510.
212. S. Yuasa, T. Nagahama, A. Fukushima, et al., Nat. Mater. 3 (2004)
868.
213. L. Zhang, N. Deng, M. Ren, et al., Superlatt. Microstruct. 42 (2007)
222.
214. W. G. Wang, C. Ni, T. Moriyama, et al., Appl. Phys. Lett. 88
(2006) 202501.
215. S. J. May, P. J. Phillips, and B. W. Wessels, J. Appl. Phys. 100
(2006) 053912.
216. Z.-M. Liao, J. Xu, J.-M. Zhang, et al., Appl. Phys. Lett. 93 (2008)
023111.
217. M. Yoshikawa, K. Inoue, T. Nakagawa, et al., Appl. Phys. Lett. 92
(2008) 113115.
218. F. J. von Preissig, H. Zeng, and E. S. Kim, Smart Mater. Struct. 7
(1998) 396.
219. X. Lu, H. Zhang, Y. Ni, et al., Biosens. Bioelectron. 24 (2008) 93.
220. W. Eerenstein, N. D. Mathur, and J. F. Scott, Nat. 442 (2006) 759.
221. Y. C. Yang, C. F. Zhong, X. H. Wang, et al., J. Appl. Phys. 104
(2008) 064102.
This page is intentionally left blank
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Chapter 11
MÖSSBAUER SPECTROSCOPY AND ITS
APPLICATIONS IN SPINTRONICS
Saeed Kamali
Department of Applied Science, University of California
Davis, California 95616
Physical Biosciences Division, Lawrence Berkeley National Laboratory
Berkeley, California 94720
E-mail: [email protected]
There are a number of spectroscopic techniques available for characteri-
zation of condensed matter. Among these methods, the so-called nuclear
probe techniques provide information about the samples under study at
atomic level. Mössbauer spectroscopy, due to a very high energy resolu-
tion, which is in the neV range, is an outstanding technique providing
valuable information about the electrical, magnetic and structural prop-
erties of condensed materials.
11.1. Introduction
In this chapter, an unique method for characterization of nano-structures,
which are the basic structures in spintronics, will be introduced. Mössbauer
spectroscopy (MS) is an excellent probe technique for the characterization
of magnetic and electronic properties of materials at atomic level. It pro-
vides (via the hyperfine interaction) local (atomistic) information about the
magnitude and direction of the atomic moment, the atomic environment
and its symmetry around the probed atom, and also about the s-electron
density at the probed nucleus in the sample. A German Ph.D. student,
Rudolf Mössbauer, in 1958–59 discovered the effect, which is based on re-
coilless emission and absorption of γ-rays by nuclei. It is worth to mention
that this effect is intimately related to the theory for neutron capture by
atoms in a crystal, which was developed by Lamb in 1930’s.1 Dicke2 studied
the theory of the effect in early 50’s. The powerfulness of the effect is more
obvious, when one realizes that it can be used in so different fields such as
267
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
268 S. Kamali
solid state physics, chemistry, geology and even in biology. The first exper-
iment was done on 191 Ir, but later on different isotopes were used, among
which 57 Fe is the most used isotope. Mössbauer was awarded the 1961 No-
bel prize in physics for this discovery. By applying this technique, one is
able to study structural, electrical, magnetic and dynamical properties of
materials.
To be able to understand the information one can extract from the
samples under study by Mössbauer spectroscopy, an introduction of this
technique is necessary. In the following, we will give a brief description
of the technique. Since the discovery of the technique, a number of excel-
lent books has been published, which describe Mössbauer spectroscopy and
its applications in different fields.3–10 The Mössbauer community, consist-
ing of several thousands of researchers in different fields gather together in
the International Conference on the Applications of the Mössbauer Effect
(ICAME) every second year and also in other international as well as re-
gional conferences. The web site, belonging to the International Board on
the Applications of the Mössbauer Effect (IBAME),11 as well as other web
sites,12–14 provide valuable information concerning the technique and other
activities.
The outline of this chapter will be as follows. First, we will describe
the basics of the technique. Then, we will present some applications of the
technique in studying different kinds of superlattices.
11.2. Mössbauer Spectroscopy: The Basics
The main feature of Mössbauer spectroscopy is emission of γ-rays from
excited nuclei and absorption of these γ-rays by identical nuclei in an ab-
sorber, i.e. the sample, whose properties are supposed to be studied. The
process is schematically shown in Fig. 11.1.
However, if we take into account the conservation of momentum and
energy, there will be some recoil of both the emitting and absorbing nuclei
in the emission as well as absorption processes. Hence, there will be some
loss of energy. After some simple calculations, the expression for the recoil
energy will be as follows:
Eγ 2
ER = (11.1)
M c2
where Eγ is the γ energy, M is the mass of the nucleus and c is the velocity
of light. In case of free atoms, the recoil energy is enough to destroy the
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 269
Excited State Excited State
Ground State Ground State
Fig. 11.1. Emission and absorption of γ-rays by identical nuclei.
resonance, i.e. no absorption will occur. It is why the free nuclei cannot
experience the Mössbauer effect. On the other hand, in case of condensed
matter, the nucleus in question is strongly bound to the crystal. In this
case the whole crystal will be recoiled as a whole, and the mass in the
Eq. (11.1) will be the mass of the whole crystal, hence the recoil energy can
be ignored.
The so-called hyperfine interactions, which are the interactions between
electronic and nuclear magnetic moments, not only change the energy of
the nuclear states but also remove, partially or completely, the degeneracies
of these states. The parameters, extracted from these interactions, give
valuable information about oxidation states, bonding properties, covalency,
electronegativity, structural and magnetic properties of the sample under
study. The following three interactions are of paramount importance:
(1) Electric Monopole Interaction.
(2) Electric Quadrupole Interaction.
(3) Magnetic Dipole Interaction
It is worth mentioning that the dimension of all parameters extracted
from these interactions is mm/s, because the vibrator carrying the radioac-
tive source vibrates at such velocities to make resonances possible.
In most cases the radioactive sources emit a single line, i.e. the emitted
γ-quanta have just one energy. On the other hand, due to the hyperfine
interactions, the nuclei under study may have several energy transitions. In
order to be able to study these transitions, one has to tune the energy of the
γ-quanta to match these transitions. As is seen in Fig. 11.2, the radioactive
source is put on a vibrator, hence the energy of the incoming γ-quanta is
changed due to the Doppler effect. In this way, the whole energy range
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
270 S. Kamali
Radioactive source on a vibrator Sample under study Detector
J-quantum
Electronics
Vibrator
PC
Fig. 11.2. A basic schematic picture of a Mössbauer set up in a transmission mode.
By putting the radioactive source on a vibrator, vibrating with a constant acceleration,
different velocities are achieved. Hence, due to the Doppler effect, a range of different
energies is scanned.
under study can be scanned and all different transitions can be observed in
a Mössbauer spectrum.
In the following parts, these interactions and the physical as well as
chemical information extracted from each, are shortly described.
11.2.1. Electric monopole interaction
The simplest form of a Mössbauer spectrum is a single absorption line, also
called singlet, where the changes in nuclear state energies are measured
against a reference. There are two sources for this difference, isomer shift
(IS) and second order Doppler shift (SOD).
11.2.1.1. Isomer shift
In a simple but real picture an atom consists of a nucleus with finite size and
a positive charge immersed in a cloud of electrons. Hence, the electrostatic
energy, due to the presence of the nucleus in the electrical potential, V(r),
caused by the surrounding electrons, will be as follows:
Eel = ρn V (r)dτ (11.2)
all space
where ρn is the charge density of the nucleus and dτ = dxdydz is the volume
element.
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 271
Subtracting the overall potential, taking into account the fact that a
nucleus has different radius in the excited state, Re , and the ground state,
Rg , then the isomer shift (IS), which is the difference between the nuclear
energy levels in the emitter or the radioactive source (S) and the absorber
(A), can be defined as:
2 2 2
IS = −S(Z) Ze(Re − Rg ) [ρSe (0) − ρA
e (0)] (11.3)
100
where ρe is the electron density at the nucleus and 0 is the permeability
constant. S(Z) is the relativistic correction factor due to the fact that in a
relativistic calculation, in addition to s-electrons there is some probability
for the presence of p-electrons in the nucleus. As is seen from Eq. (11.3),
isomer shift is a function of electron concentration, then one can extract
oxidation states, covalency, electronegativity and bonding properties of the
probe atom by studying isomer shift.
11.2.1.2. Second order Doppler shift
Because the nuclei in a crystal are not frozen and vibrate around their
equilibrium positions, the frequency of the emitted and absorbed γ-rays
are changed due to the Doppler effect, which results in the change of γ-rays
energy. At non-zero temperatures, the nuclei oscillate about their mean
position in the crystal with a frequency of the order of 1012 Hz. Though,
the averaged displacement during Mössbauer life time (140 ns) is then zero,
the second term in the Doppler shift, which is non-zero gives rise to the
second order Doppler shift (SOD) with energy change as:
δEγ v2
=− 2 . (11.4)
Eγ 2c
In the case of Fe, SOD = +0.07 mm/s as temperature is decreased by
100 K.8
11.2.1.3. Centroid shift
The measured energy shift, which is simply the position of the singlet rela-
tive to the reference is the sum of the mentioned quantities, i.e. the IS and
SOD, and called the centroid shift, CS:
CS = IS + SOD . (11.5)
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
272 S. Kamali
Intensity [A.U.]
CS 2
Excited
1 CS
Ground
2
-3 -2 -1 0 1 2 3
Emitter Absorber Velocity [mm/s]
Fig. 11.3. The energy levels in the presence of just electric monopole interaction (left),
which gives rise to a singlet (right).
The reference is generally α-Fe with CS positioned at zero mm/s. In
Fig. 11.3 the Mössbauer spectrum, a singlet, is depicted together with the
change in CS between the source and the absorber.
11.2.2. Electrical quadrupole interaction
A nucleus with non-spherical charge distribution, possesses a quadrupole
moment, which can be described by a second rank tensor with nine ele-
ments. In a principal coordinate system, the off-diagonal elements of the
tensor are all zero. The electrical quadrupole moment of a nucleus with a
charge distribution of cylindrical symmetry and with the z-axis as the axis
of quantization, can be expressed as:
1 2 2 1
Q= ρn (r)(3z − r )dτ = ρn (r)r2 (3 cos2 (θ) − 1)dτ (11.6)
e e
where dτ is volume element and θ is the angle between the vector from the
origin to a charge element in the nucleus, − →r and the symmetry axis z.
The degeneracy of the nuclear levels with spin I > 12 of a nucleus pos-
sessing a quadrupole moment in a nonzero Electric Field Gradient (EFG),
which is the second derivative of the potential caused by the surrounding
electrons, will be lifted. This is shown in Fig. 11.4.
The EFG, described by a second rank tensor, can also be presented in a
principal coordinate system, which results in diagonalizing the tensor. Due
to Laplace’s equation:
Vii = 0 (11.7)
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 273
Intensity [A. U.]
QS
-6 -4 -2 0 2 4 6
Velocity [mm/s]
Fig. 11.4. The excited state with I = ± 32 level in presence of quadrupole splitting is
split (left), which results in a doublet in the Mössbauer spectrum (right).
with i = x, y and z, there will remain only two independent elements; Vzz
and η, which is the asymmetry parameter, defined as:
Vyy − Vxx
η= . (11.8)
Vzz
If the principal coordinate system is chosen in a way implying Vxx ≤
Vyy ≤ Vzz then 0 ≤ η ≤ 1 .
The eigenvalues of the Hamiltonian of the quadrupole interaction will
then be written as:
1/2
eqVzz η2
EQ = [3m2I − I(I + 1)] 1 + (11.9)
4I(2I − 1) 3
where mI is the nuclear magnetic spin quantum number with values rang-
ing, -I, -I+1, ..., I-1, I, resulting in total (2I+1) different values. In the case
of Fe, the ground state with I = 12 will still be degenerate. On the other
hand, the degeneracy of the excited state with I = 32 will be lifted and
will be split into ± 12 and ± 23 . As an example, the so-called Quadrupole
Splitting in an axially symmetric EFG, where η = 0, will be:
3 1 eqVzz
∆EQ = EQ ± − EQ ± = . (11.10)
2 2 2
11.2.3. Magnetic hyperfine interaction
A nucleus with spin quantum number I > 0 possesses a magnetic dipole
moment −→
µ:
→
− →
−
µ = gN βN I (11.11)
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
274 S. Kamali
e
where gN is the nuclear Landé factor and βN = 2Mc , with M as nuclear
mass, is the nuclear magneton. The Hamiltonian for the −
→µ in a magnetic
→
−
field at the nucleus, H , is given by:
→
−
H = −−→
µ .H . (11.12)
The eigenvalues are given as:
EM = −gN βN HmI (11.13)
where mI is again the nuclear magnetic spin quantum number with values
ranging, -I,-I+1,...,I-1,I, resulting in total (2I+1) values. The ground state
is now split into two sublevels and the excited state into 4 levels as shown
in Fig. 11.5.
Of the eight possible transitions, only six transitions are allowed for
pure magnetic dipole transitions due to the selection rules, i.e. ∆I = 1 and
∆m = 0, ±1. Therefore, the transition from 32 level in the excited state
(I = 32 and m = 32 ) to - 12 level in the ground state (I = 12 and m = − 12 )
and the transition from - 32 level in the excited state (I = 32 and m = − 32 )
to 12 level in the ground state (I = 12 and m = 12 ) are forbidden because
∆m = −2 in the first case and ∆m = 2 in the second case, respectively.
Though in some cases one can have external magnetic field during the
measurement, the magnetic field at the nucleus originates mainly from the
atomic electrons. There are different contributions to this field:
(1) The Fermi contact field H c which is the result of the spin-polarization
of inner filled s-shells, i.e. when the spin-up and spin-down electron
Intensity [A. U.]
3
2
E6 1
2
E3 E5 1
CS
Excited E2 E4 2
v3 v4
3
2
E1
1 v2 v5
2 Magnetic Splitting
Ground v1 v6
Absorber 1 -8 -6 -4 -2 0 2 4 6 8
Emitter in a MHF 2 Velocity [mm/s]
Fig. 11.5. The energy levels in presence of magnetic interaction (left) and the related
Mössbauer spectrum.
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 275
densities are not equal. It is caused by spin-polarized partially filled
outer d-shells.
(2) The orbital field H L caused by orbital motion of valence electrons.
(3) The spin-dipolar field H d , caused by atomic electron spins.
11.2.4. Combined electric and magnetic hyperfine
interaction
Beside the electric monopole interaction, which is present in all cases, in
many cases there is a combination of electric quadrupole and magnetic
dipole interactions. Three cases are distinguished, EQ << EM , EQ >> EM
or EQ ≈ EM . In the first cases, there is one dominant interaction and
the other one is just a perturbation. When both interaction have equal
strength, the full Hamiltonian should be used. Häggström15 was the first,
who solved this complicated case analytically. Due to our emphasis on
magnetic materials, we consider here the case where EQ << EM , i.e. when
the magnetic interaction is the main component and quadrupole interaction
is just a perturbation. The EFG is assumed to be axially symmetric. The
eigenvalues can be written as:
1 eQVzz
EM,Q = −gN βN HmI + (−1)|mI |+ 2 (3 cos2 β − 1) (11.14)
8
where β is the angle between the axis of the magnetic field and the z-
component of the the EFG’s principal axis. If Vzz is positive then the
± 32 levels are shifted to higher energy by an amount which is equal to the
quadrupole energy and the ± 21 levels are shifted to lower energy by the
same amount. The shifts are reversed if Vzz is negative as can be seen in
Fig. 11.6.
11.2.5. Transmission vs. conversion electron Mössbauer
spectroscopy
In transmission Mössbauer spectroscopy, the absorber, i.e. the sample un-
der study, will be placed between the radioactive source and the detector.
The γ-rays, passed the absorber, will then be counted as a function of the
velocity of the vibrator. In this way, the γ-rays, which are matched with
a resonance energy, will be absorbed by the sample and the spectrum will
consist of dips as for example shown in the right panel of Fig. 11.5. It is
also possible to perform Mössbauer spectroscopy in a scattering geometry.
In this mode, the excited Mössbauer nuclei in the sample will de-excite
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
276 S. Kamali
%
% %
Intensity [A. U.]
%
%
v3 v4
%
v2 v5
v1 v6
-8 -6 -4 -2 0 2 4 6 8
Velocity [mm/s]
Fig. 11.6. The energy levels in the presence of magnetic interaction with quadrupole
interaction as perturbation (left), which results in asymmetry of the line positions in the
Mössbauer spectrum (right).
by emission of conversion electrons from K shell, hence, the name Conver-
sion Electron Mössbauer Spectroscopy (CEMS). It is worth noting that the
probability of de-excitation via this mode is over 90%. A conventional gas
flow detector (He + 10%CH4 ), which is transparent to γ-rays and placed
between the radioactive source and the sample, is then used to detect emit-
ted electrons as a function of velocity. The spectra will then consist of
peaks. Due to the scattering of the conversion electrons in the sample, only
electrons from approximately 1000 Åunder the surface have enough energy
to escape the surface and be detected by the detector. This makes CEMS
a surface sensitive technique. Nevertheless, due to this fact that thin films
and superlattices are always deposited on a thick substrate, the study of
these systems, such as the examples presented in this chapter, are always
performed in this mode.
11.2.6. Relative intensities of resonance lines
In spintronic devices, not only the magnetic moments, but also their di-
rections play important roles. In Mössbauer spectroscopy, one can get
information concerning the magnetic anisotropy of the sample under study
by inspecting the line intensities. The relative intensities of the resonance
lines are not only very sensitive to the properties of the nuclear transitions,
e.g. spin and parity of the different states, but also to the geometrical con-
figuration, i.e. they are functions of the angle between the direction of the
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 277
photon and the quantization axis, which can be the magnetic field direc-
tion or the z-axis in the principal coordinate system of the EFG. They are
also proportional to the Clebsch-Gordan coefficients. The relative intensity
ratios between the six lines in a 1/2 - 3/2 transition are 3: x :1:1: x :3, where
x can vary between 0 and 4. The ratio of the second (fifth) to the third
(forth) is:
A2,5 sin2 θ
=4 (11.15)
A3,4 1 + cos2 θ
As an example, in the case of powder sample, where all angles have
the same probability, the ratios are 3:2:1:1:2:3. In the case of magnetic
multilayers, if the magnetic field direction is in the plane, i.e. perpendicular
to the direction of the photon, the ratios are 3:4:1:1:4:3, whereas in case
of perpendicular magnetization, where the the magnetic field direction is
perpendicular to the sample surface and parallel to the incoming γ-rays,
the second and fifth lines vanish, i.e. the ratios will be 3:0:1:1:0:3. Thus,
Mössbauer spectroscopy is a very good tool for studying the anisotropy in
magnetic thin films and superlattices.
11.3. Superlattices, Thin Films
Thin films are defined as materials shrunk to nano scale in one dimension.
These can be magnetic multilayers, which can be constructed artificially
in the laboratory world by different techniques. Growing successive layers
of different atomic species on a well characterized substrate will produce
a multilayer. If layers of different elements are forced to have the same
lattice constant one effectively has a single crystal and the multilayer is
called a superlattice. The layers can be all magnetic, or magnetic layers
can be separated by non-magnetic ones, the so-called spacer layers. The
samples are not thermodynamically in their lowest energy state, but they
are metastable. This is why the growing conditions are crucial factors for
the resulting samples. Multilayers have differerent and unique properties
compared to the bulk ones. They are, therefore, materials with properties
that can be tailored to different applications. As an example one can men-
tion growing an element in a crystallographic phase, which can not be found
in nature because of thermodynamical laws. For example, body centered
cubic (bcc) Co and Ni, which are hcp and fcc in bulk form, respectively.
Other properties like, Magnetic Interlayer Coupling, which is the interac-
tion between the magnetic layers via non-magnetic spacers, Giant Magnetic
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
278 S. Kamali
Resistance (GMR) and Spin Polarized Tunneling (SPT) make such sup-
perlattices very nice devices in many sophisticated industrial applications.
Another important task is Interface Magnetism because interfaces in mul-
tilayers have a very important role. In interfaces the symmetry is broken
and the atomic environment is completely different from the interior part.
It is then very important to understand the interactions occurring there.
Again, because Mössbauer spectroscopy is a unique nuclear method, and
furthermore, because it is sensitive enough to allow the study of monolay-
ers by using isotopically enriched material, it may give in a fingerprint-type
of way information on the presence of different magnetic and/or crystal-
lographic phases in the sample. Moreover, because of the local character
of the technique, it is a valuable tool to distinguish different properties at
the different areas in a superlattice by introducing down to a monolayer of
57
Fe at those areas and measure the spectra. Here there are some studies
described related to superlattices.
11.3.1. Fe/Co superlattices
11.3.1.1. Magnetic hyperfine field
The systems containing iron and cobalt are still the most interesting sys-
tems in materials research, especially in the field of applied magnetism. The
very large magnetic moment per atom, and macroscopic magnetization, ob-
tained in Fe-Co alloys with approximately 30 at% Co, at the maximum of
the Slater-Pauling curve16 are some of the reasons for the intense investi-
gations of these systems. The potentiality of Fe-Co alloys in information
storage applications has been studied in theoretical papers.17 Bulk Fe-Co
alloys have been studied for a long time and almost all their structural and
magnetic properties are revealed by specially local-probe techniques.18–25
Superlattices of these elements have also been studied intensively ever since
the possibilities to grow such system by different deposition techniques have
been available experimentally26–29 as well as theoretically.30,31 Mössbauer
spectroscopy has also been a suitable technique to study such iron-based
system,32–37 because there is no need for introducing impurities in such
system, due to the present of Fe, which is a Mössbauer isotope. In most
of these works, experimental as well as theoretical, it was inferred that
the Fe hyperfine field and the magnetic moment of Fe increase at the inter-
faces. In contrary to these studies, in a experiment performed with Nuclear
Resonance Scattering (NRS), which is a synchrotron based Mössbauer spec-
troscopy technique, Lindgren et al.33 reported that the Bhf is largest in
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 279
the middle of the Fe layer in Fe(6 ML)/Co(3 ML) and Fe(5 ML)/Co(5 ML)
supperlattices, where ML stands for monolayers. This was a puzzling re-
sult. This discrepancy was resolved partially in the study by Kamali et
al., where a systematic study was performed on such superlattices.37 They
measured a series of 57 Fe(x ML)/Co(7 ML) by means of Conversion Elec-
tron Mössbauer Spectroscopy (CEMS), where x was 2, 3, 5, 9 and 14 MLs.
The CEMS together with the Bhf distributions are plotted in Fig. 11.7.
2/7
2/7
3/7
3/7
Intensity [Arbitrary Units]
Intensity [Normalized]
5/7
5/7
9/7
9/7
14/7
14/7
–8 –6 –4 –2 0 2 4 6 8 28 30 32 34 36 38 40
Velocity [mm/s] Magnetic hyperfine field [T]
Fig. 11.7. In the left panel, CEM spectra of all Fe/Co samples denoted by x/y, where
x and y stand for the nominal number of monolayers of Fe and Co, respectively, are
illustrated. In the right panel, the correspondent Fe magnetic hyperfine field distributions
are plotted. The line indicates the magnetic hyperfine field for Fe atoms in α-Fe, which
is 33.0 T. Reprinted with permission from Kamali et al.,37 J. Phys.: Condens. Matter
18 (2006) 5807. c 2006 IOP Publishing Ltd.
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
280 S. Kamali
They concluded that although the Bhf and hence the magnetic moment
of Fe atoms at interfaces increase compared with bulk Fe, the highest val-
ues are reached approximately in the third layer in the Fe layers due to
intermixing of Fe and Co atoms with a lower Co concentration than 50%
as in the first interface layer. The Bhf will then decrease for inner parts of
the Fe layers and reaches the bulk value.
11.3.1.2. Magnetic anisotropy energy
Another important issue for such superlattices is the orientation of the
Bhf . The out of plane orientation can be easily extracted by studying
the intensity ratio of second (fifth) to third (forth) lines, in a measurement
geometry, where the incoming γ-ray direction is perpendicular to the sample
surface. However, the direction of the Bhf in plane cannot be resolved
by this geometry since all directions have 90◦ angle with respect to the
incoming γ-rays. Hägström et al.36 by using a geometry similar to the one
shown in Fig. 11.10, were able to extract information about the magnetic
anisotropy in that system. They concluded that the magnetic moment of
the system lay in the [110] directions.
11.3.2. Fe/Cr
During the last few decades, there has been a huge interest in investigating
the Fe/Cr superlattices, due to the fundamental and practical interests,
which will be described shortly. One of the interesting effects discovered
first in Fe/Cr systems, which are artificial ferromagnet/antiferromagnet
superlattices, was the existence of the antiferromagnetic coupling38,39 be-
tween the Fe layers separated by Cr layers. Soon after this discovery,
an important related phenomenon, the so-called giant magnetoresistance
(GMR) effect was discovered both in Fe/Cr superlattices40 and Fe/Cr/Fe
trilayers.41 In this phenomenon, electrons with different spin directions
experience different potential barriers in passing from one ferromagnet to
another via the spacer layer. In case of parallel alignment of the magnetic
layers, spin-up electrons have high conductivity and spin-down electrons
have low conductivity. In case of anti-parallel alignment, electrons with
both spin directions will experience high barrier potential and the conduc-
tivity will be low. While in zero external magnetic field, the ferromagnetic
Fe layers couple antiferromagnetically to each other, they will couple fer-
romagnetically under an external magnetic field with sufficient strength.
GMR has many applications in spintronics, such as magnetic sensors and
magnetic recording technology, and has been recognized as an important
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 281
effect, which resulted in the awarding of the Nobel prize for physics to Fert
and Gründberg in 2007. Fe/Cr was also one of the systems, in which the os-
cillatory magnetic exchange coupling was observed.42 This effect has been
discussed as a consequence of Ruderman-Kittel-Kasuya-Yosida (RKKY)
interaction.43
Still the current investigations reveal a lot of surprising effects in mag-
netization of such multilayers. Nevertheless, the interface structure in these
systems is of paramount role in determining magnetic properties in such sys-
tems. Moreover, other new phenomena with industrial applications, such as
non-collinear magnetic exchange interaction between Fe layers,44,45 depend
on the defects, steps, roughness and intermixing at the interfaces. It is why
the understanding of the interfaces is very important in designing the spin-
tronic devices. Techniques such as X-ray diffraction are just able to give
an average roughness in the whole sample and most of the other techniques
may in the best case give indirect information on the atomic scale about
the interfaces. Once again, due to the monolayer resolution, Mössbauer
spectroscopy, has been a powerful technique to study interfaces in Fe/Cr
superlattices.46–55 Landes et al.46,47 studied the interfaces in Fe/Cr super-
lattices by introducing 2 ML 57 Fe at the interfaces and successively moving
them from interfaces to the Fe layers. In such a way, an oscillatory behavior
of the Bhf was demonstrated. It means that, while the Bhf on the Fe is
reduced drastically at the interface, there are some increases at the second
and the third layers of Fe inside the Fe layer, before the Bhf reaches the
Fe bulk value for Fe layers very far from interfaces. Furthermore, in these
studies, by measuring the Bhf as a function of temperature, a T 3/2 spin
wave law was obtained:
Bhf (T ) = Bhf (0)(1 − bT 3/2 ) (11.16)
where b, the spin wave parameter and the slope of the function, is inversely
proportional to the exchange interaction, J, i.e. the higher the value of b
the weaker the coupling at the interface. The strength of this parameter at
the interface compared to the bulk is as high as double.56,57
In analogy with alloys of Fe and Cr,58 the effect of Cr atoms on the Fe
Bhf in the interfaces has been described phenomenologically by Klinkham-
mer et al.:50
Bhf = Bhf (bulk) + n1 B1 + n2 B2 + ∆B + ∆Bi=2 (11.17)
where n1 and n2 are the number of the nearest neighbour (nn) and next
nearest neighbour (nnn) Cr atoms and B1 and B2 are the contributions from
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
282 S. Kamali
one Cr atom in the first and second shell on the Fe Bhf . Starting with Fe
Bhf (bulk)=-33.3 T, the fitted parameters were B1 =2.5 T and B2 =2.05 T,
which means that the Bhf is decreased as the number of nn and nnn Cr
is increased. The constant term, ∆B=-1.75 T, which is related to the
broken translational symmetry at the interface, is in accordance with the
observed enhancement of the Bhf . The last term ∆Bi=2 =-1.2 T is related
to the Bhf at the second Fe layer at the interface, where n1 =0 and n2 =1.
This term vanishes for all other Fe sites. This procedure is in accordance
with the oscillatory behaviour of the Bhf at the interface as mentioned
before. It is worth to mention that the oscillatory behaviour of the Bhf has
been observed only in molecular beam epitaxially growth samples, where
smooth interfaces exist, and not for sputtered samples. These experimental
results have been combined with band structure calculations59 to give a
clear picture of such systems. Kazansky and Uzdin,60 using a method
of self-consistent calculation of magnetic moments, studied the magnetic
structure of Fe/Cr interfaces with different degrees of disordered roughness.
Uzdin et al.54 performed CEMS measurements on Fe/Cr superlattices with
57
Fe ML at the different interfaces with the following compositions:
• MgO/Cr(50 Å)/[57 Fe(3 ML)/nat Fe(8 ML)/57 Fe(3 ML)/Cr(8 ML)] ×
10 (sample 1)
• MgO/Cr(50 Å)/[57 Fe(0.7 ML)/nat Fe(8 ML)/Cr(8 ML)] × 40 (sample
2)
• MgO/Cr(50 Å)/[57 Fe(0.7 ML)/nat Fe(8 ML)/Cr(8 ML)] × 200 (sample
3)
• MgO/Cr(50 Å)/[57 Fe(0.7 ML)/56 Fe(8 ML)/Cr(8 ML)] × 200 (sample
4)
The CEMS spectra together with the Bhf distributions are plotted in
Fig. 11.8.
As can be seen, there are six distinct maxima, located at Bhf =33.1, 30.6,
28.0, 25.2, 22.7, and 19.6 T for sample 1, and with very small deviations
for samples 2-4. There is an additional peak at 16.9 T for samples 2-4
only, which has been explained due to the thin 57 Fe probe layer in these
samples. The highest value (33.1 T in sample 1 and 33.2 T in other samples)
emanates probably from the bulk-like environment and, in agreement with
Eq. (11.17), from the Fe atoms in the second layer under the surface, due to
the existence of a distribution of Bhf , which includes higher values than the
bulk value. Other values are attributed to Fe atoms with different numbers
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 283
Fig. 11.8. (Left) CEMS spectra of all Fe/Cr measured at room temperature. (a) Sample
1, (b) Sample 2, (c) Sample 3, and (d) Sample 4. (Right) Magnetic hyperfine field
distributions of the same sample in the same order as in the left panel. Reprinted with
permission from Uzdin et al.,54 Phys. Rev. B 63 (2001) 1044071. c 2001 The American
Physical Society.
of Cr as nn and nnn. It is worth to mention that according to Landes et
al.,46,47 the 19.6 T should be attributed to a flat surface, i.e. n1 =4 and
n1 =1, while according to Klinkhammer et al.,50 the value 22.7 T should
be attributed to the flat surface. Furthermore, Uzdin et al.54 used these
Bhf values and the results from the self-consistent calculation of magnetic
moments for the same superlattice structures, to correlate the magnetic
moments to magnetic hyperfine fields. In the later study,55 by using the
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
284 S. Kamali
same procedure, they investigated the differences between Fe lower and
upper interfaces, as a function of annealing temperature.
11.3.3. Fe/V superlattices
Although the interesting effects discussed in the previous section, such as
the existence of the antiferromagnetic coupling,38,39 giant magnetoresis-
tance (GMR) effect40,41 and the oscillatory magnetic exchange coupling,42
were all discovered in Fe/Cr systems, the same phenomena also exist in
systems of magnetic layers separated by nonmagnetic metals in form of su-
perlattices such as Fe/V.61,62 It is also possible to tune interlayer exchange
coupling by hydrogen absorption in V layers.63,64 In the periodic table, V
stands between the non-magnetic and magnetic elements in the 3d tran-
sition metals. Though it is a non-magnetic element, when dissolved in Fe
it gets an induced magnetic moment.65 In case of Fe/V thin films66 or
superlattices,67 the induced magnetic moment in the V layers at the inter-
face couple antiferromagnetically to Fe magnetic moments. The exchange
coupling in Fe/V and Fe/VCr superlattice was studied using first princi-
ple method.68,69 In a later study, Schwickert et al. combined augmented
spherical wave method calculations with the experimental techniques to
investigate magnetic properties of Fe/V systems with interdiffused inter-
faces.61 They found that V magnetic moment in an interdiffused interface
is enhanced compared with an ideal interface. They also showed that the
induced V magnetic moments, which couple ferromagneticlly to each other,
decay monotonically away from Fe interface. The decaying extent, which
is 4 ML (6 Å), was termed as a “transient ferromagnetic” state by them.
Furthermore, they observed three antiferromagnetic coupling peaks for V
thickness at 22, 32 and 42 Å. They expected to observe another peak at
12 Å V thickness, but they explained that the probable reason not to ob-
serve it was due to the transient ferromagnetic behavior of V. In an ideal
Fe/V superlattice, an increase of Fe magnetic moment compared to the
bulk value, i.e. 2.2 µB was predicted by Niklasson et al.30 This is in ac-
cordance with the results from dilute FeV alloys.70 In the latter study,70
the authors suggested that substitution of every nn and nnn Fe atom by a
V atom will decrease Bhf by 3 T. Tage et al.,71 by applying first principle
theory, were able to show that by geometrical tuning of superlattices with
appropriate choice of material, interesting magnetic configuration can be
produced. They demonstrated that the magnetization moment in every
second layer in Fe/V/Co superlattice can be stabilized in an out of plane
direction and perpendicular to the magnetization direction of the adjacent
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 285
magnetic layers. A magnetic configuration that can be used in sensor ap-
plications.
As it has been mentioned before, the interface configurations have cru-
cial role in the above mentioned effects and hence in the construction of the
spintronic devices. Then it is an interesting subject to study interfaces in
Fe/V superlattices to understand how the V atoms influence the neighbor-
ing Fe atoms at the interfaces and deeper in the Fe layer. CEMS has been
extensively used to study Fe/V superlattices. It is obvious that Bhf and the
magnetic moments of Fe atom decreases as the surrounding Fe atoms are
replaced by non-magnetic atoms such as V. By depositing a monolayer of
57
Fe at Fe(10 ML)/V(10 ML) interface and at different distances from the
interface, using 15 T/µB conversion factor, the average magnetic moment
of Fe was estimated to be 1.1 µB at the interface, while it was increased
gradually to 2.2 µB , the characteristic value for bulk Fe, for Fe atoms deep
inside the Fe layer.72 In a series of studies, Wäppling et al. used CEMS
to investigate Fe hyperfine field distribution in Fe(x ML)/V(y ML) super-
lattices as function of x and y.73–77 Some spectra from these studies are
presented in the left panels in Fig. 11.9. Bhf distributions from these spec-
tra and some other related Bhf distributions are indicated in right panels
in Fig. 11.9.
The samples with Fe layer less than 5 ML showed no magnetic splitting
down to 133 K. The average magnetic hyperfine field was strongly depen-
dent on the Fe layer thickness. Nevertheless, the bulk value, i.e. 33 T
was not found even in a 10 ML thick Fe layer at room temperature, which
according to their argument, was due to the lower Curie temperature com-
pared to bulk value. Furthermore, they concluded that Fe Bhf depends not
only on the Fe layer thickness but also on the V layer thickness. For a fixed
Fe layer thickness, the increase of V layer thickness caused the decrease
of Bhf . They explained that interlayer coupling, which is stronger for the
thinner spacer layer, contributes to the magnetization of the non-magnetic
part, hence there will be positive influence on the magnetic hyperfine field
distribution. By depositing 57 Fe at the different interfaces, i.e. Fe on
V and V on Fe, they74 studied different interfaces in Fe/V superlattices.
They showed that the average Bhf is higher for the upper interface of Fe
than the lower interface, which means that intermixing is less in the Fe
upper interface, and thus, that the Fe layer is smoother than the V layer.
They also studied the interface quality as a function of growth temper-
ature.78 While the XRD data suggested that the temperature growth is
optimum for Tg = 300−330◦C, the CEM spectra were almost identical for
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
286 S. Kamali
Fig. 11.9. (Left) CEMS spectra from Fe/V superlattices with varying Fe and V thick-
nesses. (Right) Magnetic hyperfine field distributions for the samples depicted in the left
panels and some other related superlattices. Reprinted with permission from Nordsröm et
al.,76 Hyperfine Interact. 141/142 (2002) 465. c 2002 Kluwer Academic Publishers.
Tg = 230−330◦C. This means that the short-range order is similar in the
samples grown in this range and the growth temperature has minor effect
if it is kept under the optimum temperature, i.e. 330◦ C.
In a more recent study, Uzdin and Häggström79 used Mössbauer spec-
troscopy together with the theoretical modeling and self-consistent calcu-
lations of magnetic moments to investigate the magnetic structure of Fe/V
systems with different intermixings at the interface. They used the so called
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 287
floating model55 to describe the intermixing at the interface. In this model,
it is supposed that during the growth a fraction, ψ, of the atoms of the
deposited monolayer exchanges position with the atoms in the “substrate”,
i.e. the layer under. The extreme values, ψ=0 and ψ=1 mean ideal interface
and complete interchange of the two monolayers in question, respectively.
By using Bhf distribution, it is then possible to extract a value for the
intermixing parameter, ψ. Furthermore, they found a linear relationship
between calculated Fe magnetic moment and Fe Bhf from CEMS, with the
proportionality factor of 15 T/µB .
11.3.4. Exchange spring magnets
Magnetization reversal process is one of the fundamental questions of mag-
netism. The evolution of a magnetic structure as a response of the system
on an external magnetic field can lead to a rich variety of behaviours at the
nanoscale. The investigation of such processes is important to elucidate the
internal nature of the magnetic interactions, as well as their modification
upon confined geometry. It is quite relevant also in technological applica-
tions for the improvement of the functionality of magnetic memory media
and spintronic devices. The development of nanotechnology has pushed the
study and utilization of non-collinear magnetic ordering in single magnetic
elements. One of such systems where non-collinear magnetic structure can
be reversibly varied by magnetic field is the system consisting of exchange
coupled of hard and soft magnetic phases, the so-called spring magnets. In
spring magnets the hard and the soft magnetic bilayers or multilayers are
coupled to each other ferromagnetically by an exchange field, Hex , which
is a function of magnetic parameters and the geometry of the soft layers.
For external applied magnetic field, Happ , less than Hex antiparallel to the
magnetization direction of the system, the magnetic moment of the soft
layers will remain parallel to the magnetic moment of the hard layers. As
the Happ is increased behind this limit, the magnetization reversal occurs
via the development of a non-collinear spiral structure in the soft material,
but if the field is then removed, the spiral structure will spring back to the
original direction, i.e. the easy axis of the hard magnet, hence the system
is called spring magnet.80 For further increase of Happ , the magnetization
of the hard layer is irreversibly switched at the so-called nucleation field,
Hnu , which depends on magnetic parameters and the geometry of both
hard and soft layers. The variation of the external magnetic field leads
to the reversible change of the magnetic moment direction mainly in the
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
288 S. Kamali
soft magnetic layers. Goto et al. were the first who studied the magne-
tization reversal process in the soft layer. Later on, Skomski and Coey,81
and Coey82 studied the exchange coupling theoretically. A model of one-
dimensional chain of classical spins83 were used to simulate characteristic
behaviour of such systems. If one divide the whole magnetic system, i.e.
Fe layer plus FePt layer, into N sublayers, the total energy of this system
can be written as:
N −1 N N
Ai,i+1
E=− 2
cos(φi − φi+1 ) − Ki cos2 (φi ) − HMi cos(φi − φH )
d
i=1 i=1 i=1
(11.18)
which is the sum of the magnetic exchange and anisotropy energies and
the dipolar interaction with the external field. Here Ai,i+1 is the exchange
constant between sublayers i and i+1, d = (dsof t + dhard )/N is the sublayer
thickness, φ is the angle between the sublayer magnetization relative to the
remanent magnetization of the hard layer and H is the external magnetic
field. Ki and Mi are the the anisotropy constant and magnetization of the
ith sublayer, respectively. To achieve the equilibrium spin configuration, one
has to minimize the energy with respect to each of φi . The combination
of the high magnetization of the soft phase with the high coercivity of the
hard one, makes these systems suitable for building permanent magnets,
with large energy product (magnetic saturations times coercivity).84,85 An
energy product as high as 1090 kJ/m3 is predicted to be achieved by suit-
able nanostructured composites such as Sm2 Fe17 N3 /Fe65 Co35 . Exchange
spring media are also promising systems for magnetic recording media.86–90
Various experimental methods have been used for the investigation of mag-
netic ordering in spring magnets, which in turn became a model system for
benchmarking different methods of non-collinear magnetism study. Large
efforts were undertaken to characterize the angular distribution of magnetic
moments directions across the soft magnetic layer. For example, Kuncser
et al.91 used angle dependent Mössbauer spectroscopy to study the spin
structure of Sm-Co(10nm)/Fe(15nm) bilayers. Two samples were studied.
In the first sample (A), the whole Fe layer was homogeneously enriched in
57
Fe, whereas in the second sample (B) 2 nm layer of 57 Fe was deposited at
the interface. It means that, while the Mössbauer signal is from the whole
Fe layer in the first sample, it reveals just the spin direction at the inter-
face in the second sample. The first measurements in zero external field
and with perpendicular geometry, i.e. when γ-rays have the same direction
as the normal direction of the samples, the 3:4:1:1:4:3 line ratios of the
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 289
Fig. 11.10. (a) Schematic geometrical arrangement of the CEMS measurement. The
external magnetic field is varied along y axis, which is the easy axis of the hard magnetic
layer. (b) Schematic representation of a spin spiral distribution in the plane of the soft
magnetic layer. Reprinted with permission from Kuncser et al.,91 Phys. Rev. B 68
(2003) 0644161. c 2003 the American Physical Society.
Mössbauer spectra, revealed that all spins lay in the plane of the samples.
With the geometry depicted in Fig. 11.10, the second line intensity in the
Mössbauer spectra is changed as the applied magnetic field is increased.
Figure 11.11 shows the CEMS spectra of both samples as a function
of the external magnetic field. The intensity ratio R23 in both samples
undergoes characteristic changes as the external magnetic field is varied.
The intensity ratio has a minimum value of about 1.5 at a field intensity of
0.4 T.
Because the Fe spin moments in an applied external field will form a
spiral structure, the angle between the incident γ-ray and the spin moment
will not be constant but will have a certain angular distribution P(φ) with:
2π
P (ϕ)dϕ = 1 (11.19)
0
The intensity ratio will then be written as:
2π
1 − cos2 φ cos2 ϕ
R23 = 4 P (ϕ)dϕ (11.20)
0 1 + cos2 φ cos2 ϕ
Taking into account that the incoming γ-rays are perpendicular to the
easy axis of the hard magnetic layer and also that the conversion electrons
from different depths have different probabilities to escape the sample sur-
face, Eq. (11.20) can be expressed as:
θ0 +∆θ
4 1 − cos2 φ sin2 θ 0.14 · (θ − θ0 )
R23 = × (0.60 + )dθ
0.67 · ∆θ θ0 1 + cos2 φ sin2 θ ∆θ
(11.21)
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
290 S. Kamali
Fig. 11.11. CEMS spectra at room temperature as a function of external magnetic
field. (a) Sample A with an incident angle of γ radiation, φ, of 20◦ (b) Sample B with
an incident angle of γ-rays, φ, of 30◦ . Reprinted with permission from Kuncser et al.,91
Phys. Rev. B 68 (2003) 0644161. c 2003 The American Physical Society.
where θ0 is the angle between the magnetic direction of the first Fe layer
and the easy axis of the hard layer, and ∆θ is the moment aperture, i.e. the
angle between the Fe moment at the interface and the Fe magnetic moment
at the surface of the layer. Now, by having intensity ratios, R23 from a series
of measurements, one is able to extract the distribution parameters, which
has been the aim of this study. In combination with the intensity ratio
from the second sample, which gave the angle between the Fe moment
and the easy axis of the hard magnetic layer at the interface, they were
able to infer that there exists a uniform spin spiral structure in the Fe
layer. Furthermore, they concluded that for the highest value of the applied
magnetic field before the irreversible switching of the hard magnet occurs,
the Fe moment direction reaches a maximum of 40◦ relative the easy axis of
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 291
Fig. 11.12. (Left) Calculated intensity ratios R23 according to Eq. (11.21) for sample
A as a function of moment aperture, ∆θ, with θ0 as a parameter. Large circles indicate
experimental data points and their corresponding applied fields values for positive fields
and negative fields < 0.25 T and ellipses indicate the same for negative fields > 0.30 T.
(Right) Schematic illustration of the spiral structure of spins in the soft magnetic layer
in case of an applied magnetic field of 0.58 T. Reprinted with permission from Kuncser et
al.,91 Phys. Rev. B 68 (2003) 0644161. c 2003 the American Physical Society.
the Sm-Co layer and the moment aperture was 140◦ . These are summarized
in Fig. 11.12.
For utilization of exchange spring magnets in electronic applications it
is very important to know how the properties of the system depend on
the characteristics of the hard and the soft magnetic layers. One of the
ways which allows tuning the exchange spring behaviour is the variation of
soft layer composition. Recently, a method to improve the exchange-spring
nanocomposite permanent magnets was suggested.92–94 In this method,
by annealing or depositing Sm-Co/Fe bilayers at high temperatures in-
duced intermixing at the interface occurs, which results in an improvement
of the energy product. This result was in contrast to the earlier belief
that an ideal interface should give the optimal exchange coupling. The
hysteresis loop becomes more single phase-like, though the magnetization
remains fully reversible. The interpretation of this phenomenon is based
on a simple phenomenological model, which considers the spring magnet
as one-dimensional chain of classical spins.83 Micromagnetic simulation
requires the knowledge about the concentration profile near the interface
as well as the dependence on parameters (uniaxial anisotropy, saturation
magnetization and exchange constant) and on the concentration in alloy.
It cannot give information about real magnetic structure at the atomic
scale. Only recently an atomic scale quantum-mechanical theory based
on a non-collinear Tight-Binding formalism for self-consistent calculations
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
292 S. Kamali
of electronic and magnetic structure of d-metallic systems in an external
magnetic field has been developed.95,96 The theory was applied to the de-
scription of the magnetic reversal process in exchange spring magnets with
Fe as the soft magnetic material. The calculations reproduce all the main
features of the exchange spring behavior. The theory is suitable for the
study of more complex systems with interface roughness, interdiffusion and
non-coplanar magnetic ordering. This approach can be used for the in-
terpretation of magnetization reversal processes in systems with different
soft layer composition. Based on the theoretical approach, magnetic struc-
tures measured by Mössbauer spectroscopy in external magnetic field can
be interpreted to give a complete picture, which will be highly appreciated
not only by the scientists in this field, but also by the industrial research
centers, which can use these results in very sophisticated devices.
Acknowledgments
I sincerely thank my colleagues, L. Häggström, V. Uzdin, H. Raanaei, E.
Goikolea, J. Åkerman, for suggestions and comments, and especially R.
Wäppling for reading the manuscript. Special thanks go to my colleagues,
whose permissions for reusing their results, make this work possible. I
would also like to thank my colleagues at the Mössbauer Effect Data Cen-
ter13 (MEDC), J. Stevens, N. Hall and A. Khasanov, for a kind collabora-
tion. Japanese society for promotion of science (JSPS) is acknowledged for
financial support.
References
1. W. E. J. Lamb, Phys. Rev. 55, 190, (1939).
2. R. H. Dicke, Phys. Rev. 89, 472, (1953).
3. H. Frauenfelder, The Mössbauer Effect. (W. A. Benjamin, Inc, New York,
1963).
4. G. K. Wertheim, Mössbauer Effect: Principles and Applications. (Academic
Press, New York, 1964).
5. V. I. Goldanskii and R. H. Herber, Chemical Applications of Mössbauer Spec-
troscopy. (Academic Press, New York, 1968).
6. N. N. Greenwood and T. C. Gibb, Mössbauer Spectroscopy. (Chapman and
Hall, London, 1971).
7. L. May, An Introduction to Mössbauer Spectroscopy. (Plenum, New York,
1971).
8. T. C. Gibb, Principles of Mössbauer Spectroscopy. (Chapman and Hall, Lon-
don, 1977).
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 293
9. P. Gutlich, R. Link, and A. Trautwein, Mössbauer Spectroscopy and Transi-
tion Metal Chemistry. (Springer-Verlag, Berlin, 1978).
10. G. Long, Ed., Mössbauer Spectroscopy Applied to Inorganic Chemistry, Vol-
ume I. (Plenum Press, New York and London, 1984).
11. http://www.ibame.org/index.html.
12. http://www.mossbauer.org/index.html.
13. http://orgs.unca.edu/medc/.
14. http://pecbip2.univ-lemans.fr/~moss/webibame/.
15. L. Häggström. Report UUIP-851, University of Uppsala, Sweeden. Technical
report, Physics Institute, (1974).
16. C. Kittel, Introduction to Solid State Physics. (John Wiley & Sons, Inc.,
1996).
17. T. Burkert, L. Nordström, O. Eriksson, and O. Heinonen, Phys. Rev. Lett.
93(2), 027203, (2004).
18. M. F. Collins and J. B. Forsyth, Phil. Mag. 8, 401, (1963).
19. C. E. Johnson, M. S. Ridout, and T. E. Cranshaw, Proc. Phys. Soc. 81,
1079, (1963).
20. T. R. McGuire and D. I. Bardos, J. Appl. Phys. 40, 1371, (1969).
21. I. Vincze, I. A. Campbell, and A. J. Meyer, Solid State Commun. 15, 1495,
(1974).
22. A. Narayanasamy, T. Nagarajan, P. Muthukumarasamy, and T. S. Radhakr-
ishnan, J. Phys. F: Metal Phys. 9(11), 2261, (1979).
23. B. deMayo, Phys. Rev. B. 24, 6503, (1981).
24. B. Fultz, Hyperfine Interact. 41, 607, (1988).
25. H. H. Hamdeh, B. Fultz, and D. H. Pearson, Phys. Rev. B. 39, 11233, (1989).
26. J. Dekoster, E. Jedryka, M. Wójcik, and G. Langouche, J. Magn. Magn.
Mater. 126, 12, (1993).
27. P. Blomqvist, R. Wäppling, A. Broddefalk, P. Nordblad, S. G. E. Velthuis,
and G. P. Felcher, J. Magn. Magn. Mater. 248, 75, (2002).
28. P. Blomqvist and R. Wäppling, J. Cryst. Growth. 252, 120, (2003).
29. M. Björck, G. Andersson, B. Lindgren, R. Wäppling, V. Stanciu, and
P. Nordblad, J. Magn. Magn. Mater. 284, 273, (2004).
30. A. M. N. Niklasson, B. Johansson, and H. L. Skriver, Phys. Rev. B. 59(9),
6373, (1999).
31. O. Eriksson, L. Bergkvist, E. Holmström, A. Bergman, O. LeBacq, S. Frota-
Pessoa, B. Hjövarsson, and L. Nordström, J. Phys.: Condens. Mater. 15(5),
599, (2003).
32. J. Dekoster, H. Bemelmans, S. Degroote, J. DeWachter, E. Jedryka,
R. Moons, A. Vantomme, and G. Langouche, Hyperfine Interact. 95, 191,
(1995).
33. B. Lindgren, M. Andreeva, L. Häggström, V. S. B. Kalska, A. Chumakov,
O. Leupold, and R. Rüffer, Hyperfine Interact. 136/137, 439, (2001).
34. B. Kalska, P. Blomquist, L. Häggström, and R. Wäppling, J. Phys.: Condens.
Mater. 13, 2963, (2001).
35. B. Kalska, P. Blomquist, L. Häggström, and R. Wäppling, J. Magn. Magn.
Mater. 226-230, 1773, (2001).
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
294 S. Kamali
36. L. Häggström, B. Kalska, E. Nordström, P. Blomqvist, and R. Wäppling, J.
Alloys and Compounds. 347, 252, (2002).
37. S. Kamali-M, A. Bergman, G. Andersson, V. Stanciu, and L. Häggström, J.
Phys.: Condens. Mater. 18, 5807, (2006).
38. P. Grünberg, R. Schreiber, Y. Pang, M. B. Brodsky, and H. Sowers, Phys.
Rev. Lett. 57(19), 2442, (1986).
39. C. Carbone and S. F. Alvarado, Phys. Rev. B. 36(4), 2433, (1987).
40. M. N. Baibich, J. M. Broto, A. Fert, F. N. V. Dau, F. Petroff, P. Eitenne,
G. Creuzet, A. Friederich, and J. Chazelas, Phys. Rev. Lett. 61(21), 2472,
(1988).
41. G. Binasch, P. Gründberg, F. Saurenbach, and W. Zinn, Phys. Rev. B. 39
(7), 4828, (1989).
42. S. S. P. Parkin, N. More, and K. P. Roche, Phys. Rev. Lett. 64(19), 2304,
(1990).
43. M. A. Ruderman and C. Kittel, Phys. Rev. 96, 99, (1954).
44. J. C. Slonczewski, J. Magn. Magn. Mater. 150, 13, (1995).
45. A. Schreyer, J. F. Ankner, T. Zeidler, H. Zabel, C. F. Majkrzak, M. Schäfer,
and P. Grünberg, Europhys. Lett. 32, 595, (1995).
46. J. Landes, C. Sauer, R. A. Brand, W. Zinn, and Z. Kajcsos, Hyperfine Inter-
act. 57, 1941, (1990).
47. J. Landes, C. Sauer, R. A. Brand, W. Zinn, S. Mantl, and Z. Kajcsos, J.
Magn. Magn. Mater. 86, 71, (1990).
48. D. L. Williamson, B. M. Lairson, A. P. Payne, N. M. Rensing, and B. M.
Clemens, Hyperfine Interact. 92, 1271, (1994).
49. J. Zukrowski, G. Liu, H. Fritzsche, and U. Gradmann, J. Magn. Magn. Mater.
145, 57, (1995).
50. F. Klinkhammer, C. Sauer, E. Y. Tsymbal, S. Handschuh, Q. Leng, and
W. Zinn, J. Magn. Magn. Mater. 161, 49, (1996).
51. R. Schad, P. Belien, G. Verbanck, K. Temst, V. V. Moshchalkov, Y. Bruynser-
aede, B. Bahr, J. Falta, J. Dekoster, and G. Langouche, Europhys. Lett. 44,
379, (1998).
52. P. B. R. Schad, G. Verbanck, K. Temst, H. Fischer, S. Lefebvre, M. Bessiere,
D. Bahr, J. Falta, J. Dekoster, L. Langouche, V. V. Moshchalkov, and
Y. Bruynseraede, J. Magn. Magn. Mater. 198-199, 104, (1999).
53. M. Kopcewicz, T. Lucinski, F. Stobiecki, and G. Reiss, J. Appl. Phys. 85(8),
5039, (1999).
54. V. M. Uzdin, Phys. Rev. B. 63, 104407, (2001).
55. V. Uzdin, W. Keune, and M. Walterfang, J. Magn. Magn. Mater. 240, 504,
(2002).
56. C. T. Rado and T. R. Weertman, J. Phys. Chem. Solids. 11, 315, (1959).
57. D. L. Mills and A. A. Maradudin, J. Phys. Chem. Solids. 28, 1855, (1967).
58. S. M. Dubiel and J. Zukrowski, J. Magn. Magn. Mater. 23, 214, (1981).
59. A. J. Freeman and C. L. Fu, J. Appl. Phys. 61, 3356, (1987).
60. A. K. Kazansky and V. M. Uzdin, Phys. Rev. B. 52, 9477, (1995).
61. M. M. Schwickert, R. Coehoorn, M. A. Tomaz, E. Mayo, D. Lederman, and
W. L. O’Brien, Phys. Rev. B. 57(21), 13681, (1998).
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
Mössbauer Spectroscopy and Its Applications in Spintronics 295
62. R. M. A. Broddefalk, P. Nordblad, P. Blomqvist, R. Wäppling, J. Lu, and
E. Olsson, Phys. Rev. B. 65, 214430, (2002).
63. B. Hjörvarsson, J. A. Dura, P. Isberg, T. Watanabe, T. J. Udovic, G. An-
dersson, and C. F. Majkrzak, Phys. Rev. Lett. 79(5), 901, (1997).
64. V. Leiner, K. Westerholt, A. M. B. H. Zabel, and B. Hjörvarsson, Phys. Rev.
Lett. 91(3), 037202, (2003).
65. I. Mirebeau, G. Parette, and J. W. Cable, J. Phys. F: Met. Phys. 17, 191,
(1987).
66. T. G. Walker and H. Hopster, Phys. Rev. B. 49(11), 7687, (1994).
67. M. A. Tomaz, W. J. A. Jr., W. L. O’Brien, and G. R. Harp, J. Phys.:
Condens. Matter. 9, L179, (1997).
68. M. van Schifgaarde and F. Herman, Phys. Rev. Lett. 71(12), 1923, (1993).
69. M. van Schifgaarde, F. Herman, S. P. Parkin, and J. Kudrnovský, Phys. Rev.
Lett. 74(20), 4063, (1995).
70. J. C. Krause, J. Schaf, and J. M. I. da Costa, Phys. Rev. B. 61(9), 6196,
(2000).
71. A. Tage, L. Nordström, P. James, B. Johansson, and O. Eriksson, Nature.
406, 280, (2000).
72. L. Häggström, P. Isberg, M. Wikner, B. Degroote, and R. Wäppling, Hyper-
fine Interact. C 3, 401, (1998).
73. B. Kalska, L. Häggström, P. Blomquist, and R. Wäppling, J. Phys.: Condens.
Matter. 12, 539, (2000).
74. B. Kalska, P. Blomquist, L. Häggström, and R. Wäppling, Europhys. Lett.
53(3), 395, (2001).
75. B. Kalska, L. Häggström, P. Blomquist, and R. Wäppling, J. Magn. Magn.
Mater. 226-230, 1782, (2001).
76. E. Nordström, B. Kalska, L. Häggström, P. Blomqvist, and R. Wäppling,
Hyperfine Interact. 141/142, 465, (2002).
77. E. Nordström, B. Kalska, L. Häggström, P. Blomqvist, and R. Wäppling, J.
Alloys Comp. 348, 208, (2003).
78. G. Andersson, E. Nordström, and R. Wäppling, Europhys. Lett. 60(5), 731,
(2002).
79. V. M. Uzdin and L. Häggström, Phys. Rev. B. 72, 024407, (2005).
80. E. F. Kneller and R. Hawig, IEEE Trans. Magn. 27(4), 3588, (1991).
81. R. Skomski and J. M. D. Coey, Phys. Rev. B. 48(21), 15812, (1993).
82. J. M. D. Coey, Solid State Commun. 102, 101, (1997).
83. E. E. Fullerton, J. S. Jiang, M. Grimsditch, C. H. Sowers, and S. D. Bader,
Phys. Rev. B. 58(18), 12193, (1998).
84. R. Skomski, J. Appl. Phys. 76(10), 7059, (1994).
85. S. D. Bader, Rev. Mod. Phys. 78, 1, (2006).
86. R. H. Victora and X. Shen, IEEE Trans. Magn. 41(2), 537, (2005).
87. R. H. Victora and X. Shen, IEEE Trans. Magn. 41(10), 2828, (2005).
88. D. Suess, T. Schrefl, S. Fähler, M. Kirschner, G. Hrkac, F. Dorfbauer, and
J. Fidler, Appl. Phys. Lett. 89, 0125041, (2006).
89. D. Suess, Appl. Phys. Lett. 89, 1131051, (2006).
90. J. P. Wang, W. Shen, and J. Bai, IEEE Trans. Magn. 41(10), 3181, (2005).
April 1, 2010 10:27 World Scientific Review Volume - 9in x 6in chapter11
296 S. Kamali
91. V. E. Kuncser, M. Doi, W. Keune, M. Askin, H. Spies, J. S. Jiang, A. Ino-
mata, and S. D. Bader, Phys. Rev. B. 68, 064416, (2003).
92. J. S. Jiang, J. E. Pearson, Z. Y. Liu, B. Kabius, S. Trasobares, D. J. Miller,
S. D. Bader, D. R. Lee, D. Haskel, G. Srajer, and J. P. Liu, Appl. Phys. Lett.
85(22), 5293, (2004).
93. J. S. Jiang, J. E. Pearson, Z. Y. Liu, B. Kabius, S. Trasobares, D. J. Miller,
S. D. Bader, D. R. Lee, D. Haskel, G. Srajer, and J. P. Liu, J. Appl. Phys.
97, 10K3111, (2005).
94. Y. Choi, J. S. Jiang, Y. Ding, R. A. Rosenberg, J. E. Pearson, S. D. Bader,
A. Zambano, M. Murakami, I. Takeuchi, Z. L. Wang, and J. P. Liu, Phys.
Rev. B. 75, 104432, (2007).
95. V. M. Uzdin and A. Vega, Phys. Rev. B. 77, 134446, (2008).
96. V. M. Uzdin and A. Vega, Nanotechnol. 19, 315401, (2008).
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Chapter 12
NUCLEAR RESONANCE SCATTERING AND ITS
APPLICATIONS IN SPINTRONICS
Saeed Kamali
Department of Applied Science, University of California
Davis, California 95616
Physical Biosciences Division, Lawrence Berkeley National Laboratory
Berkeley, California 94720
E-mail: [email protected]
Parallel to the development of synchrotron radiation facilities, in addi-
tion to the use of such radiation in conventional techniques, new meth-
ods have been developed. One of these methods is synchrotron radiation
based Mössbauer spectroscopy, which is also called Nuclear Resonance
Scattering. Nuclear Resonance Scattering is now a very powerful tech-
nique for characterization of nanostructured materials.
12.1. Introduction
Nuclear Resonance Scattering (NRS) is the Synchrotron Radiation based
Mössbauer spectroscopy. As a continuation of the previous chapter, NRS
will be described here. NRS is now a routine technique for studying many
different kinds of samples, especially in extreme conditions such as high
pressure. By introducing some examples, the use of NRS in spintronics will
be presented.
Because the source for this technique is synchrotron radiation, a short
description of it will be given before introducing the technique and its
applications.
12.2. Synchrotron Radiation
Synchrotron radiation (SR), which once was an unwanted radiation for
high energy physicists, now has become a standard source for almost all
297
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
298 S. Kamali
conventional techniques, especially in the third generation synchrotron fa-
cilities, where insertion devices such as undulators and wigglers are used.
Its applications cover all natural sciences, ranging from material science to
biophysics. Synchrotron radiation, which is based on the emission of elec-
tromagnetic radiation by relativistic electrons in magnetic fields, has several
advantages over conventional X-ray sources, some of which are listed below:
• Brilliance: It is brighter than the conventional sources by several
orders of magnitude. Shorter times are needed for measurements,
especially in the case of the nanostructures.
• Polarization: Different kinds of polarizations. Linearly or circularly
polarized can be arranged thanks to the conservation of angular
momentum law. It also results in an increase in sensitivity and a
decrease in background noise.
• Collimation: The beam can be collimated to sub-micron scale.
• Continuous Spectrum: From infra red to hard X-rays are produced.
• Pulsation: It can be powerful in dynamical studies.
The photon energy from such a source ranges from a few tens eV up to sev-
eral hundreds keV. This means that all the conventional techniques based
on the interaction of X-rays with different kinds of materials can be per-
formed using such radiation. Parallel to the development of Synchrotron
radiation, the interest for nanotechnology and its applications has increased
drastically during the last decade. Fabrication of all kinds of nanostruc-
tured materials is now possible by various techniques due to the recent
progress in this field. Due to the increase of the ratio of surface atoms with
respect to the total number of atoms of the particle and also, to the changes
of the electronic structure of nanomaterials, there are significant differences
between properties of such materials and bulk properties. Moreover, there
are some phenomena, which are characteristic just for nanostructured mate-
rials. Such phenomena make nanostructured materials applicable for many
devices. Then, understanding the physical properties and electronic struc-
ture of nanomaterials is of crucial importance.
Synchrotron radiation, due to its novel properties, is a proper source
for studying nanostructured materials. The development of different syn-
chrotron based techniques enables us to perform measurements, which have
not been possible before. There are now several synchrotron radiation fa-
cilities all over the world, among them SPring-81 in Japan is the biggest
with a circumference of 1436 m and an electron energy of 8 GeV. Advanced
Photon Source (APS)2 in USA with 7 GeV electron energy and European
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Nuclear Resonance Scattering and Its Applications in Spintronics 299
Synchrotron Radiation Facility (ESRF)3 in France with 6 GeV are the other
largest synchrotron facilities in the world.
For a detailed description of synchrotron radiation and its properties,
the interested readers are referred to a number of excellent books in this
field.4–6
12.3. Nuclear Resonance Scattering
In 1974 Ruby7 suggested to used synchrotron radiation as the source for
Mössbauer spectroscopy. Gerdau et al.8 were the first who monochroma-
tized the SR down to 10−8 eV at an energy of 14.4 keV by double nuclear
Bragg diffraction from yttrium iron garnet (YIP) single crystal enriched
with 57 Fe. Later on, they performed the first NRS experiment on YIG
sample in 1985 at HASYLAB (Hamburger Synchrotron Strahlungslabor)
and observed the so-called quantum beats in the time domain.9 Since
then, NRS has been developed very rapidly to establish itself as a powerful
technique. One can refer to some interesting articles and reviews10–19 and
books20–22 for a detailed description of the technique.
In addition to coherent elastic NRS, i.e. a spectroscopy based on ana-
lyzing beat patterns in the temporal evolution of the nuclear decay, there is
also inelastic NRS (coherent and incoherent), which is a suitable technique
for studying phonon density of states. This technique was performed for
the first time by Seto et al.23 Here we will just consider coherent elastic
NRS, denoted by NRS from now on. For studying inelastic NRS one can,
for example, refer to the interesting book by Röhlsberger.22
A main component in NRS is the detector. The basic requirements for a
detector in NRS experiments are, for example, high quantum efficiency, high
counting rate, low background noise and high time resolution. A detector
fulfilling these criteria is the Avalanche Photodiode (APD) detector.24,25
APDs are made of thin silicon wafer with a depletion region of 50-100 mm
and are the most efficient detectors in nuclear resonance scattering experi-
ments due to their better count-rate capability (up to few GHz), lower noise
(around 0.01 Hz) and small size (0.5 cm3 ). APDs have an effective gain
in the range of 100-1000 (below the breakthrough voltage), and thus, low
energy photons (below 10 keV) are also detectable, which is an important
task in inelastic nuclear scattering experiments.
Due to the broad bandwidth of the synchrotron radiation, which cur-
rently can be reduced to sub-meV in the best case, all nuclear states are
excited simultaneously. After de-excitation of these states, which occurs
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
300 S. Kamali
Fig. 12.1. (a) A schematic experimental set-up for NRS. Black dots present the electron
bunches in the storage ring. (b) Time structure of the detected radiation. The black
peaks correspond to the exciting pulses with 176 ns in between in 16-bunch mode at
ESRF. (c) NRS time patterns for nuclei with just one transition. (d) NRS time patterns
for the case, where there are two nuclear transitions. The forward scattering process from
two close lying nuclear levels gives rise to the modulation of the time pattern, which is
due to the emission of γ-rays with slightly different energies. The dashed lines in (c)
and (d) correspond to the response of a single nucleus, and the full lines result from
multiple scattering in a thick sample. The dotted lines in (d) correspond to the intensity
pattern of the unsplit nuclear levels that determine the envelope of the quantum beat
pattern. Reprinted with permission from Leupold,18 Hyperfine Interact. 144/145 (2002)
21. c 2002 Kluwer Academic Publishers.
after e.g. 100 ns in case of 57 Fe, the emitted photons, which have different
energies will interfere as time goes on, resulting in the quantum beats. Fig-
ure 12.1 shows a schematic set-up, the synchrotron pulses and some time
spectra for absent and present of hyperfine interactions, respectively.
In Fig. 12.1(a), a very schematic experimental set-up for NRS, in a trans-
mission mode, is depicted. Figure 12.1(b) shows the time structure in the
experiment. Figures 12.1(c) and (d) show that, depending on the absence
or the existence of hyperfine interactions, the time spectra will undergo
characteristic changes. In case of a single nucleus with no hyperfine inter-
actions in a very thin sample, where the probability of multiple scattering
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Nuclear Resonance Scattering and Its Applications in Spintronics 301
is negligible, the response will be an exponential decay, as indicated by the
dashed line in Fig. 12.1(c). On the other hand, for thick samples, according
to quantum mechanics, there is the possibility that the incoming photon
will interact with more than one nucleus, the so-called multiple scattering,
the result of which is the modulation of the time pattern depicted by the
full line in Fig. 12.1(c). In case of existence of hyperfine interactions, there
will be more than one transition in a single nucleus. The emitted γ-rays
from these transitions have slightly different energies, the interference of
which results in the modulation of the time pattern as indicated by the
dashed line in Fig. 12.1(d). Again, in the thick samples, where the multiple
scattering exists, there will be response from different nuclei, which results
in the time pattern modulation indicated by full line in Fig. 12.1(d).
Furthermore, because the γ-rays are polarized in NRS, depending on the
direction of the magnetic hyperfine field, different transitions can be excited,
which reflect themselves in the time spectra as indicated in Fig. 12.2.
Figure 12.3 depicts in more detail the relative orientation of incidence
→
−
of photons with k to a sample with magnetization direction − →m.
Figure 12.4 indicates the Mössbauer transitions and the related energies
in case of 57 F e. The six allowed transitions decompose into three different
groups depending on the polarization. ω’s are angular frequencies of the
different transitions, and F−1 , F0 and F+1 are functions describing the scat-
tering of radiation with left-circular, linear, and right-circular polarization,
respectively.22
As an example related to magnetism, Fig. 12.5 shows time spectra for
different orientation of the magnetic hyperfine field for purely σ polarized
incident radiation on a 2 nm thick 57 Fe film on a W substrate.
It should be emphasized that in contrast to conventional Mössbauer
spectroscopy, which is performed in the energy domain, NRS is performed
in the time domain. Fourier analysis will then be applied to extract the
relevant magnetic and electronic parameters. Several computer based fit-
ting programs have been developed during the last decade for analyzing the
NRS spectra in reflection as well as transmission mode.26–30
One of the most interesting cases for application of NRS in spintronics
is the so-called Grazing Incidence mode, the theory of which can be studied
in a series of papers by Hannon et al.31–35 Related to this mode, Grazing
Incidence Mössbauer Spectroscopy (GIMS) or Grazing Incidence Reflectiv-
ity (GIR) mode has now become a standard technique to study thin films
and superlattices.36–39 In this technique, presented in a series of papers
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
302 S. Kamali
Fig. 12.2. NRS spectra of a 3 µm thin 57 Fe foil with different magnetization directions.
E is the polarization of the incident photon (σ polarization) with propagation direction
k. The corresponding transitions are also indicated in the right part of the figure. l and r
denote left and right circular polarization, respectively. Reprinted with permission from
Leupold,18 Hyperfine Interact. 144/145 (2002) 21. c 2002 Kluwer Academic Publishers.
by Irkaev et al.40–42 and Röhlsberger,43 the incoming photons penetrate
the sample with a very small angle relative to the surface of the sample.
By successively increasing the angle, the photons penetrate more and more
in the sample, so one is able to study different parts of the sample in a
systematic way.
In addition to all Mössbauer spectroscopy applications, NRS can be
used in other fields such as dynamical processes thanks to the very fast
data acquisition.
NRS has been used in many kinds of superlattice systems, for example,
the systems that we have already discussed in the previous chapter such
as Fe/Co,44 Fe/Cr45–48 and Fe/V.49 Presenting the results from all these
and other interesting studies done by NRS will be beyond the scope of this
chapter. In the following sections, just a few examples will be presented,
so that the readers get some kind of feeling for the technique.
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Nuclear Resonance Scattering and Its Applications in Spintronics 303
Fig. 12.3. Incidence of photons on a sample with magnetization direction m. Reprinted
with permission from Röhlsberger et al.,19 Phys. Rev. B 67 (2003) 245412.
c 2003 The
American Physical Society.
12.4. Exchange Spring Magnets
As an instructive example, which is also related to one of the former top-
ics discussed earlier, the magnetic spin structure of exchange coupled thin
films, studied by NRS will be presented. Röhlsberger et al.50 did a direct
measurement of the depth-dependent spin structure in a Exchange Spring
Magnet consisting of Fe55 Pt45 (denoted here as FePt) as the hard magnetic
layer and Fe as the soft magnetic layer. The sample was prepared by rf
magnetron sputtering in an Ar atmosphere. First a 30 nm FePt layer was
deposited on a superpolished Si substrate. After annealing, a wedge-shaped
Fe layer with a slope of 0.5 nm/mm was grown on the FePt, on which a
0.7 nm 57 Fe layer was deposited. Another layer of Fe was deposited with
opposite slope. The sample was then coated by 3 nm Ag to prevent the
oxidation. As seen in the right panel of Fig. 12.6, it was then possible to
probe different depths of the sample just by exposing different parts of the
sample to the SR. In the figure, the transverse displacement of the expo-
sure of the SR relative to the edge of the sample, i.e. where the 57 Fe probe
atoms are in contact with the FePt layer, ∆x, and the depth position of
them, D, are indicated. In the left panel of Fig. 12.6, the spin structure
of the system is schematically shown after applying an external magnetic
field perpendicular to the easy axis of the hard layer.
In this way, it has been possible to perform a series of NRS measure-
ments on a single sample to extract magnetic properties through out the
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
304 S. Kamali
Fig. 12.4. The six dipole allowed Mössbauer transitions of 57 Fe decompose into three
different polarization dependencies given by F−1 , F0 and F+1 , which are functions of
ω and describing the scattering of radiation with left-circular, linear, and right-circular
polarization, respectively. Reprinted with permission from Röhlsberger et al.,19 Phys.
Rev. B 67 (2003) 245412. c 2003 The American Physical Society.
soft magnetic layer in form of the spin structure. Figure 12.7 shows the time
spectra of the grazing incidence reflection. In the left panel, the time spec-
tra at various depths under an applied external magnetic field of 160 mT are
depicted. In the right panel, the time spectra for variation of the magnetic
hyperfine field at the center of Fe layer as a function of external magnetic
field are plotted.
As can be seen, both subspectra, as a function of the depth and as
a function of the external magnetic field, undergo characteristic changes,
which are directly related to the rotation of the magnetization. The result
of the fitted spectra is summarized in Fig. 12.8.
The characteristic behaviour of such systems was simulated using a
model of one-dimensional chain of classical spins.51 In this model, where
the whole magnetic system, i.e. soft magnetic layer plus hard magnetic
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Nuclear Resonance Scattering and Its Applications in Spintronics 305
Fig. 12.5. Calculated time spectra for a 2 nm thick 57 Fe film on a W substrate for
different orientations of the magnetic hyperfine field for purely σ polarized incident
radiation. Reprinted with permission from Röhlsberger et al.,19 Phys. Rev. B 67
(2003) 245412. c 2003 The American Physical Society.
layer, is divided into N sublayers, the total energy of the system can be
expressed as:
N −1 N N
Ai,i+1 2
E=− cos(φi − φi+1 ) − K i cos (φi ) − HMi cos(φi − φH ) .
i=1
d2 i=1 i=1
(12.1)
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
306 S. Kamali
scattering
H plane
φ ∆x
D 11 nm
Fe
k0
FePt
M 20 mm 0.7 nm
57 Fe
Fig. 12.6. Left: schematic spin structure in an exchange spring bilayer consisting of a
soft-magnetic layer (Fe) on a hard-magnetic layer (FePt) with uniaxial anisotropy after
applying an external field H perpendicular to the remanent magnetization M of the hard
magnetic layer. Right: The different components of the sample and the direction of the
SR are shown. By adjusting the transverse displacement, ∆x, it is possible to probe the
magnetic properties of the different depths, D, of the sample. Reprinted with permission
from Röhlsberger et al.,50 Phys. Rev. Lett. 89 (2002) 237201. c 2002 The American
Physical Society.
Equation (12.1) was used to simulate such characteristic behaviour.
The parameters used for FePt layer in the simulation were Kh = 4.0 ·
107 erg/cm3, Mh = 1100 emu/cm3 and Ah = 4.0 · 10−6 erg/cm. The pa-
rameters used for Fe layer were Ks = 1.0·103 erg/cm3 , Ms = 1900 emu/cm3
and As = 1.0 · 10−6 erg/cm. The solid lines in Fig. 12.8 are the results from
these simulations. The measured data at the top of the Fe layer deviate
from the simulation. The authors explained it as the diffusion of the oxy-
gen through the capping layer, which results in the formation of an Fe
oxide phase with a reduced exchange coupling. By letting this parame-
ter to be adjusted during the simulation, a reduced exchange coupling of
As = 3.0 · 10−7 erg/cm for top 3 nm Fe layer was achieved, which results
in the dashed lines.
12.5. Magnetic Tunnel Junctions
The development of the applied spintronic technology is one of the hottest
scientific fields in the world. One result of this technology is Magnetore-
sistive Random Access Memory (MRAM) which will have a crucial role in
industry. MRAM is slated for commercial introduction in the very near
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Nuclear Resonance Scattering and Its Applications in Spintronics 307
Fig. 12.7. Time spectra of Fe/FePt exchange spring bilayer, depicted in Fig. 12.6, in
a grazing incidence reflection geometry. Left: As a function of different depths under
influence of a perpendicular magnetic field of B=160 mT. Right: variation of the time
spectra at the center of the Fe layer as a function of external magnetic field. In both
cases the time spectra undergo characteristic changes, which reflect the rotation of the
magnetization direction. The dots are the measured data and the solid lines are theoret-
ical simulations according to Eq. (12.1). Reprinted with permission from Röhlsberger et
al.,50 Phys. Rev. Lett. 89 (2002) 237201. c 2002 The American Physical Society.
future and will greatly simplify memory use in many applications thanks
to its unique attributes such as non-volatility, fast read and write, and
virtually unlimited read and write cycles.52,53 The core element of the
MRAM bit cell is the Magnetic Tunnel Junction (MTJ), which in its sim-
plest form consists of two ferromagnetic films separated by a very thin
(10–15 Å) dielectric tunneling barrier.54,55 The information is stored in the
relative orientation of the magnetization of the two ferromagnetic layers
[parallel(P)/antiparallel(AP)]. The MTJ typically has a Tunneling Magne-
toresistance (TMR) greater than 40%, i.e. the AP state has a 40% greater
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
308 S. Kamali
Fig. 12.8. The spin rotation as a function of depth in the Fe layer for different external
magnetic fields of 160 mT (circles), 240 mT (triangles) and 500 mT (squares). Solid lines
are simulations according to the model described by Eq. (12.1). The dashed lines are
simulations with a reduced exchange coupling in the first 3 nm of the top Fe layer due
to the possible oxidation. This oxidation phase of Fe is depicted in the left inset. The
rotation angle of the magnetization in the center of the Fe layer is plotted as a function
of the external magnetic field in the right inset. Solid and dashed lines are simulation
for the mentioned model and with a reduced exchange coupling, respectively. Reprinted
with permission from Röhlsberger et al.,50 Phys. Rev. Lett. 89 (2002) 237201. c 2002
The American Physical Society.
resistance than the P state. The memory information is read out by apply-
ing a voltage on the MTJ and measuring the resulting current. Figure 12.9
shows schematic representation of tunneling and MTJ’s basic components.
For MRAM and also the read heads of the hard disk drives (HDD)
currently in use, a TMR ratio of about 70% is sufficient, but for next
generation of spintronic devices such as high density MRAM and ultra
density HDD, much higher TMR ratios are necessary. In 2001, Butler et
al.57 and Mathon and Umerski58 used first-principle calculations to study
the mechanism of TMR in epitaxial Fe/MgO/Fe MTJs. A TMR ratio of
over 1000% was then predicted. In 2007, Lee et al.59 reported a TMR
ratio as high as 500% at room temperature and of 1010% at 5 K for a
[Co25 Fe75 ]80 B20 (4 nm)/MgO(2.1 nm)/[Co25 Fe75 ]80 B20 (4.3 nm) MTJ.
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Nuclear Resonance Scattering and Its Applications in Spintronics 309
FM Tunnel FM FM Tunnel FM
electrode 1 barrier electrode 2 electrode 1 barrier electrode 2
e
e e
e e
e
Energy Energy Energy Energy
e e
EF EF EF EF
e e
D1 D1 D2 D2 D1 D1 D2 D2
(a) (b)
Resistance of MTJ ( R )
Upper lead
FM electrode R AP
Cap layer
(free layer)
or
Tunnel barrier
SyF
structure
FM electrode RP
(pinned layer)
AF layer MR ratio
Seed layer (R AP – R P )/ R P
Lower lead
–H C 0 +H C Magnetic field ( H )
(c) (d)
Fig. 12.9. Schematic illustration of the different components of a MTJ and of the TMR
effect. (a) In the parallel alignment of the magnetizations of the two electrodes (P state),
there is a low tunneling resistance, due to the matching of the density of states (DOSs)
at the Fermi level, EF , for spin-up, D1↑ , D2↑ , and spin-down, D1↓ , D2↓ , respectively,
in the electrodes 1 and 2. (b) When the magnetizations of the electrodes are aligned
antiparallel, the tunneling resistance is high, because the DOSs are not matched. (c)
Different components of a typical MTJ. (d) A magnetoresistance curve of a MTJ, where
HC denotes the coercivity field for the free layer of MTJ. The inset shows the definition
of MR ratio. Reprinted with permission from Yuasa,56 J. Phys. Soc. Jpn. 77 (2008)
031001. c 2008 The Physical Society of Japan.
The tunneling barrier is typically formed by first depositing a metal film
(e.g. Al or Mg) and subsequently oxidizing the metal in an oxygen plasma.
Proper oxidation of the MTJ is critical for achieving high TMR, and in
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
310 S. Kamali
this sense optimal oxidation is typically defined as the condition for which
TMR is maximized.60 Under-oxidation leaves unoxidized metal at the bot-
tom electrode, and over-oxidation results in a low-quality magnetic oxide
added to the tunneling barrier. Both scenarios reduce the spin polariza-
tion at the bottom interface and consequently reduce the TMR. In order
to achieve a completely oxidized tunneling barrier it is favorable to over-
oxidize the MTJ at the oxidation step, continue with the deposition of the
top layer, and finally anneal the MTJ material stack to reduce the mag-
netic oxide back to metal and increase the oxidation level of the tunneling
barrier.
The quality of the two interfaces to the tunneling barrier (top and bot-
tom) is also critical for achieving good long-term reliability of the MTJ.61–63
Any remaining magnetic oxide is of low quality, prone to both gradual dete-
rioration and dielectric breakdown, and is consequently detrimental to the
reliability of the MTJ. It has been argued that it is primarily oxidized Fe
that governs the poor reliability of the magnetic oxide.
By depositing an Al or Mg wedge followed by a single oxidation one
will be able to probe an entire range of oxidation conditions from severely
under-oxidized via optimal to severely over-oxidized. Since Fe plays a key
role in MRAM technology as one of the main elements in both the top
and bottom magnetic electrode of the Magnetic Tunnel Junctions (MTJ),
and is believed to be the most critical element for achieving both high
Tunneling Magnetoresistance and long-term reliability, it will be extremely
valuable to use Mössbauer Spectroscopy to extract as much information as
possible about the nature of Fe in MTJs. While the MTJ oxidation pro-
cess and the composition at the bottom interface has been studied in the
past using various spectroscopic techniques, no attempt has been made to
use Mössbauer spectroscopy (or actually NRS) for this purpose, although
Mössbauer spectroscopy, as mentioned before, is a unique method to pro-
vide atomistic information about the magnitude of the Fe atomic moment,
the environment around Fe nuclei at the atomic level and also the symme-
try around the Fe atom. Valuable information, which can be used to design
such devices in an optimum way. Through dusting of the interfaces with
57
Fe and using NRS, Kamali el al.64 were the first to study the magnetic
environment of the 57 Fe atoms and determined how the oxidation state of
Fe evolves with the number of annealing steps, both at the top and the
bottom interface, and how the local interface quality correlates with the
overall MTJ quality.
They proposed fabrication of trilayers of the typical compounds used in
MRAM devices, CoFeB for the top and bottom ferromagnetic electrodes,
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Nuclear Resonance Scattering and Its Applications in Spintronics 311
and MgO for the tunneling barriers. In addition they deposited probing lay-
ers of 57 Fe and used nuclear resonance scattering of this layer to determine
the oxidation state of Fe at the two interfaces as a function of oxidation
conditions and annealing times. It is worth mentioning that MTJs used
in spintronic devices, are trilayers and not multilayers, so there will be
just a few atomic 57 Fe layers in such samples, and conventional Mössbauer
spectroscopy, even with the most strongest sources, will be insufficient for
getting enough statistic. On the other hand, the high brilliance of syn-
chrotron radiation facilities makes it possible to perform NRS on samples
with very low quantity of 57 Fe.
12.6. Conclusions
Synchrotron based Mössbauer Spectroscopy, the so-called Nuclear Reso-
nance Scattering (NRS), is a very sensitive probe technique to study the
magnetic state and environment of Fe atoms and other Mössbauer isotopes.
Thanks to the very high photon intensity available at third generation syn-
chrotron facilities, NRS and other Mössbauer related techniques have al-
ready had and will have a significant scientific contribution to the active
field of spintronic materials and devices.
Acknowledgments
I sincerely thank my colleagues, L. Häggström, R. Wäppling, V. Uzdin,
M. Andreeva, Y. Yoda, Y. Sakurai, E. Goikolea and J. Åkerman for sug-
gestions and comments. I would especially like to thank R. Röhlsberger
for the permission of reusing his work on Exchange Spring Magnets and
also for reading the manuscript. I would also like to thank my colleagues
at the Mössbauer Effect Data Center65 (MEDC), J. Stevens, N. Hall and
A. Khasanov, for a kind collaboration. Japanese society for promotion of
science (JSPS) is acknowledged for financial support.
References
1. http://www.spring8.or.jp.
2. http://www.aps.or.
3. http://www.ESRF.fr.
4. H. Winick, Synchrotron Radiation Sources: A Primer, Vol 1. (World Scien-
tific Publishing Co Pte Ltd, Singapore, 1995).
5. D. Attwood, Soft X-Rays and Extreme Ultraviolet Radiation: Principles and
Applications. (Cambridge University Press, Cambridge CB2 2RU, UK, 1999).
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
312 S. Kamali
6. J. Als-Nielsen, Elements of Modern X-ray Physics. (John Wiley & Sons Ltd,
Chichester West Sussex England, 2001).
7. S. L. Ruby, J. de Physique. 35(C6), 209, (1974).
8. E. Gerdau, R. Rüffer, H. Winkler, W. Tolksdorf, C. P. Klages, and J. P.
Hannon, Phys. Rev. Lett. 54(8), 835, (1985).
9. E. Gerdau, R. Rüffer, R. Hollatz, and J. P. Hannon, Phys. Rev. Lett. 57,
1141, (1986).
10. E. Gerdau, Hyperfine Interact. 90, 301, (1994).
11. W. Sturhahn and E. Gerdau, Phys. Rev. B. 49(14), 9285, (1994).
12. G. V. Smirnov, Hyperfine Interact. 97/98, 551, (1996).
13. G. V. Smirnov, Hyperfine Interact. 123/124, 31, (1999).
14. Y. V. Shvyd’ko, Hyperfine Interact. 123/124, 275, (1999).
15. Y. V. Shvyd’ko, Phys. Rev. B. 59(14), 9132, (1999).
16. G. R. HOY, Hyperfine Interact. 135, 191, (2001).
17. R. Rüffer, Hyperfine Interact. 141/142, 83, (2002).
18. O. Leupold, Hyperfine Interact. 144/145, 21, (2002).
19. R. Röhlsberger, J. Bansmann, V. Senz, K. L. Jonas, A. Bettac, K. H. Meiwes-
Broer, and O. Leupold, Phys. Rev. B. 67, 245412, (2003).
20. E. Gerdau and H. de Waard, Eds., Nuclear Resonant Scattering of Syn-
chrotron Radiation (Part A). (Hyperfine Interact, The Netherlands, 1999).
21. E. Gerdau and H. de Waard, Eds., Nuclear Resonant Scattering of Syn-
chrotron Radiation (Part B). (Hyperfine Interact, The Netherlands, 2000).
22. R. Röhlsberger, Nuclear Condensed Matter Physics with Synchrotron Ra-
diation: Basic Principles, Methodology and Applications. (Springer-Verlag,
Berlin Heidelberg Germany, 2004).
23. M. Seto, Y. Yoda, S. Kikuta, X. W. Zhang, and M. Ando, Phys. Rev. Lett.
74(19), 3828, (1995).
24. S. Kishimoto, Nucl. Instr. Meth. Res. A. 309, 603, (1991).
25. A. Q. R. Baron and S. L. Ruby, Nucl. Instr. Meth. Res. A. 343, 517, (1993).
26. W. Sturhahn, Hyperfine Interact. 125, 149, (2000).
27. Y. V. Shvyd’ko, Hyperfine Interact. 125, 173, (2000).
28. M. Haas, E. Realo, H. Winkler, W. Meyer-Klaucke, and A. Trautwein, Hy-
perfine Interact. 125, 189, (2000).
29. H. Spiering, L. Déak, and L. Bottýan, Hyperfine Interact. 125, 197, (2000).
30. M. Andreeva, B. Lindgren, and V. Panchuk. REFTIM, Version 6.1. http:
//www.esrf.fr/computing/scientific/REFTIM/MAIN.htm.
31. J. Hannon, G. Trammell, M. Mueller, E. Gerdau, H. Winkler, and R. Rüffer,
Phys. Rev. Lett. 43(9), 636, (1979).
32. J. Hannon, N. V. Hung, G. Trammell, E. Gerdau, M. Mueller, R. Rüffer, and
H. Winkler, Phys. Rev. B. 32(8), 5068, (1985).
33. J. Hannon, N. V. Hung, G. Trammell, E. Gerdau, M. Mueller, R. Rüffer, and
H. Winkler, Phys. Rev. B. 32(8), 5081, (1985).
34. J. Hannon, G. Trammell, M. Mueller, E. Gerdau, R. Rüffer, and H. Winkler,
Phys. Rev. B. 32(10), 6363, (1985).
35. J. Hannon, G. Trammell, M. Mueller, E. Gerdau, R. Rüffer, and H. Winkler,
Phys. Rev. B. 32(10), 6374, (1985).
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
Nuclear Resonance Scattering and Its Applications in Spintronics 313
36. A. I. Chumakov, L. Niesen, D. L. Nagy, and E. E. Alp, Hyperfine Interact.
123/124, 427, (1999).
37. D. L. Nagy, L. Bottyán, L. Deák, E. Szilágyi, H. Spiering, J. Dekoster, and
G. Langouche, Hyperfine Interact. 126, 353, (2000).
38. M. A. Andreeva, Hyperfine Interact. 156/157, 595, (2004).
39. M. A. Andreeva and B. Lindgren, Phys. Rev. B. 72, 125422, (2005).
40. S. M. Irkaev, M. A. Andreeva, V. G. Semenov, G. N. Belozerskii, and O. V.
Grishin, Nucl. Instr. and Meth. B. 74, 545, (1993).
41. S. M. Irkaev, M. A. Andreeva, V. G. Semenov, G. N. Belozerskii, and O. V.
Grishin, Nucl. Instr. and Meth. B. 74, 554, (1993).
42. S. M. Irkaev, M. A. Andreeva, V. G. Semenov, G. N. Belozerskii, and O. V.
Grishin, Nucl. Instr. and Meth. B. 103, 351, (1995).
43. R. Röhlsberger, Hyperfine Interact. 123/124, 301, (1999).
44. B. Lindgren, M. Andreeva, L. Häggström, V. S. B. Kalska, A. Chumakov,
O. Leupold, and R. Rüffer, Hyperfine Interact. 136/137, 439, (2001).
45. T. S. Toellner, W. Sturhahn, R. Röhlsberger, E. E. Alp, C. H. Sowers, and
E. E. Fullerton, Phys. Rev. Lett. 74(17), 3475, (1995).
46. T. Ruckert, W. Keune, W. Sturhahn, M. Hu, J. Sutter, T. Toellner, and
E. Alp, Hyperfine Interact. 126, 363, (2000).
47. M. A. Andreeva, V. G. Semenov, B. Lindgren, L. Häggström, B. Kalska, A. I.
Chumakov, O. Leupold, R. Rüffer, K. A. Prokhorov, and N. N. Salashchenko,
Hyperfine Interact. 141/142, 119, (2002).
48. C. L’abbé, J. Meersschaut, W. Sturhahn, J. S. Jiang, T. S. Toellner, E. E.
Alp, and S. D. Bader, Phys. Rev. Lett. 93(3), 037201, (2004).
49. B. Kalska, L. Häggström, B. Lindgren, P. Blomquist, R. Wäppling, M. A.
Andreena, Y. V. Nikitenko, V. V. Proglyado, V. L. Aksenov, V. G. Semenov,
A. I. Chumakov, O. Leupold, and R. Rüffer, Hyperfine Interact. 136/137,
295, (2001).
50. R. Röhlsberger, H. Thomas, K. Schlage, E. Burkel, O. Leupold, and
R. Rüffer, Phys. Rev. Lett. 89(23), 237201, (2002).
51. E. E. Fullerton, J. S. Jiang, M. Grimsditch, C. H. Sowers, and S. D. Bader,
Phys. Rev. B. 58(18), 12193, (1998).
52. J. Åkerman, M. DeHerrera, M. Durlam, B. Engel, J. Janesky, F. Mancoff,
J. Slaughter, and S. Tehrani, Magnetic Tunnel Junction Based Magnetore-
sistive Random Access Memory, In ed. M. Johnson, Magnetoelectronics,
pp. 231–272. (Elsevier Ltd, 2004).
53. J. Åkerman, Sci. 308, 508, (2005).
54. J. S. Moodera, L. R. Kinder, T. M. Wong, and R. Meservey, Phys. Rev. Lett.
74, 3273, (1995).
55. T. Miyazaki and N. Tezuka, J. Magn. Magn. Mater. 139(3), L231, (1995).
56. S. Yuasa, J. Phys. Soc. Jpn. 77(3), 031001, (2008).
57. W. H. Butler, X. G. Zhang, and T. C. Schulthess, Phys. Rev. B. 63, 054416,
(2001).
58. J. Mathon and A. Umerski, Phys. Rev. B. 63, 220403, (2001).
59. Y. M. Lee, J. Hayakawa, S. Ikeda, F. Matsukura, and H. Ohno, Appl. Phys.
Lett. 90, 212507, (2007).
March 31, 2010 17:17 World Scientific Review Volume - 9in x 6in chapter12
314 S. Kamali
60. S. Moodera, E. F. Gallagher, K. Robinson, and J. Nowak, Appl. Phys. Lett.
70, 3050, (1997).
61. J. Åkerman, P. Brown, M. DeHerrera, M. Durlam, E. Fuchs, D. Gajewski,
M. Griswold, J. Janesky, J. Nahas, and S. Tehrani, IEEE Transactions on
Device and Materials Reliability. 4(3), 428, (2004).
62. J. Åkerman, P. Brown, D. Gajewski, M. Griswold, J. Janesky, M. Martin,
H. Mekonnen, J. Nahas, S. Pietambaram, J. Slaughter, and S. Tehrani, IEEE
International Reliability Physics Symposium. 43, 163, (2005).
63. S. Kim and B. Cho, Appl. Phys. Lett. 86, 142106, (2005).
64. S. Kamali and et al., Unpublished.
65. http://orgs.unca.edu/medc/.
Chapter 13
BIONANOMAGNETISM
Peter Svedlindh*, Klas Gunnarsson*, Mattias Strömberg* and Sven Oscarsson†
*
Department of Engineering Sciences, Uppsala University
P.O. Box 534, SE-75121, Sweden
†
Department of Physics and Materials Science, Uppsala University
P.O. Box 530, SE-75121 Uppsala, Sweden
E-mail:
[email protected] Magnetic nanoparticles constantly find new applications in
biotechnology and medicine. In all cases, the response of the
nanoparticles to a magnetic field plays an essential role, either by
creating local magnetic fields or by being attracted to magnetic field
sources. This article will not attempt to review existing and emerging
applications, but will instead focus on two methods with high potential
for future bioassay applications. Firstly, we describe a new principle for
DNA sequence detection using oligonucleotide probe-tagged magnetic
beads. The target-DNA is recognised by hybridisation to a linear DNA-
probe, forming upon ligation a circularised target-probe complex.
Thereafter these complexes are volume-amplified to micrometer-sized
DNA coils, which are detected by the addition of the probe-tagged
beads. Upon incorporation of the beads, the complex magnetisation vs.
frequency spectrum changes dramatically. Secondly, a novel method
for controlled transport and release of proteins immobilised to magnetic
beads is presented. Protein covered beads are added to a substrate
containing micron-sized magnetic elements, which together with a
rotating magnetic field control their movement. Adding a reductive
agent, the immobilised proteins are momentarily released from the
beads. Using this method, transportation and local release of sub-
attomoles of any kind of molecule can be achieved.
315
316 P. Svedlindh et al.
13.1. Introduction
During the last decades, single domain magnetic nanoparticles, often
referred to as magnetic beads, have found numerous applications in
biosciences. A unique property of magnetic beads is their response to a
magnetic field; magnetic beads acquire a sizeable field induced magnetic
moment and are attracted to sources of magnetic field. In medicine,
magnetic nanoparticles are used as magnetic resonance imaging (MRI)
contrast agents.1 MRI is based on the nuclear magnetic resonance signal
from protons in e.g. tissue water. The imaging is carried out by detecting
the relaxation processes of the proton spins after having momentarily
being influenced by a transverse radio-frequency (RF) field; one
distinguishes between a longitudinal relaxation process (T1-decay) when
the proton spins align with a static magnetic field and a transversal de-
phasing relaxation process when the coherent spin precession of the
proton spins is lost with the RF field switched off (T2-decay). It is by
now well established that the magnetic field from nanoparticles will have
a strong effect on the T2-decay,2 the relaxation time of this decay is
shortened, and can thus be used for MRI contrast enhancement. Anti-
body conjugated microparticles carrying iron-oxide nanoparticles have
e.g. been used for in vivo MRI contrast enhancement, probing small-
sized cancers3 and acute brain inflammation.4 Another interesting
application of magnetic nanoparticles in medicine is in hyperthermia
cancer therapy.1,5 Here the therapy is e.g. achieved by using in vivo
administrated stealth magnetic nanoparticles designed for selective
uptake by tumour cells. Once inside a cell, under the influence of an AC
magnetic field, energy is dissipated, either due to Néel relaxation of the
particle magnetic moments or because of rotational Brownian motion of
the nanoparticles.1
Array-based bioassay methods are powerful as tools for analyses
of DNA, RNA, proteins and other biomolecules.6 Today, fluorescence
and chemiluminescence technologies are dominating the read-out of
microarrays owing to their high sensitivity and multiplex capabilities.
However, magnetic beads can be used as labels in bioassays and give
alternatives for detection of biological processes. Magnetic bead based
Bionanomagnetism 317
bioassays involve labelling probe biomolecules with magnetic beads and
upon hybridisation with the target, either static or dynamic magnetic
bead properties can be detected. Furthermore, magnetic biosensor
schemes are either classified as substrate-based (sensor on chip) or
substrate free. In the former biosensor category, if the target is present,
probe functionalised magnetic beads bind directly to the sensor surface,
thereby inducing a signal change. Examples of existing magnetic
sensor based techniques are micro-Hall effect7 and magnetoresistive
micro-biosensor platforms,8 and magnetic force microscopy.9 One
substrate-free method is the Brownian relaxation biosensor scheme,10
which makes use of the decrease in the bead Brownian relaxation
frequency caused by the hydrodynamic size increase upon probe-target
hybridisation.
For bioseparation, the use of bioconjugated magnetic beads is a
well-known and routinely used technique capable of specific separation
of e.g. cells,11,12 bacteria,13 and proteins.14 State-of-the-art procedures
imply that millions of magnetic beads, each bead often carrying
thousands of probe-target complexes, are separated simultaneously as a
group.
A controlled transport of magnetic beads on microchips, on the
single bead level, opens up new perspectives for novel instrumentation
in bioseparation and bioassay technology, where magnetic beads with
individually attached probe molecules, e.g. immunoglobulines, are
allowed to continuously pass through a sample in a controlled and
predictable way. Various methods for magnetic bead control have been
published,15–17 and we will here focus on one method that allows for both
transportation and separation of magnetic beads on the single bead
level.15 In addition, we will describe a substrate-free biosensor method18
combining the Brownian relaxation of magnetic nanobeads with a
method to amplify the size of the target-probe complex, thus also
amplifying the effect on the Brownian relaxation process. Combining the
two here described methods15,18 has to potential to yield a powerful all
magnetic bioassay method.
318 P. Svedlindh et al.
13.2. Properties and Biofunctionalisation of Magnetic Beads
13.2.1. Magnetic beads
Today, a variety of magnetic beads are available where the surfaces of
the beads have been functionalised in order to fit the needs of different
biochemical applications.19 The magnetic material used can be either
ferromagnetic, a magnetic transition metal element or alloy,20 or
ferrimagnetic, e.g. magnetite (Fe3O4).21 For temperatures below 350°C
and particle sizes less than 500 nm, particles of magnetite by time
oxidise into ferrimagnetic maghemite (γ-Fe2O3).22,23 The vast majority of
prepared magnetic bead systems make use of ferrimagnetic iron-oxides
as magnetic material, although ferromagnetic transition metal
nanoparticles offer higher saturation magnetisations. In the applications
described here, magnetic nanoparticles are embedded in a non-magnetic
host material, which defines the bead size and the surface to be
functionalised. The magnetic material consists of grains of γ-Fe2O3,
typically with a grain diameter somewhere in the range 5–15 nm, each
grain being small enough to be in a single domain state. The grains either
form a clustered core or are randomly dispersed in the pores of the non-
magnetic matrix. Beads in suspension will exhibit a superparamagnetic
behaviour, i.e. exhibiting a strong field induced magnetisation but
showing negligible magnetic hysteresis. Important properties of the
beads are e.g. the concentration of beads in solution, the content of
magnetic material, and shape and size of the beads. In most applications,
the beads need to be mono-dispersive, i.e. their diameter should be well
defined.
In the experiments described here, the following types of spherical
beads have been utilised. For all products, the carrying liquid is
deionised water. In Section 13.3, so-called Nanomag® -D beads of
diameters 40 nm, 80 nm, 130 nm and 250 nm, respectively, have been
used.24 In this type of bead, a cluster of maghemite nanoparticles form a
core surrounded by a dextran hull. In Section 13.4, Dynabead® M-27025
or the Micromer® -M24 beads have been used, with diameters of 2.8 µm
and 4.9 µm, respectively. In the former case, maghemite grains are
Bionanomagnetism 319
randomly trapped inside the pores of a polystyrene matrix, while in the
latter case, the nanoparticles are clustered to form the core in a latex
casing.
13.2.2. Biofunctionalisation of magnetic beads – the SPDP coupling
chemistry
Surface modification of magnetic beads is often necessary in order to
achieve ferrofluidic stability, solubility in a certain solvent environment
or in order to introduce molecular entities on the surface serving
either as sites for further functionalisation or as molecular probes
in biotechnological applications. Immobilisation of biomolecules
onto the surface of magnetic beads is a process commonly
denoted biofunctionalisation. There exist numerous publications on
biofunctionalisation of magnetic beads in the scientific literature and by
that also a large number of coupling chemistries. Just to mention a few
examples, Ma et al.26 synthesized bovine serum albumin protein coated
magnetic microspheres, Aurich et al.27 synthesised polyaspartate coated
magnetic nanoparticles for biomedical applications and Gong et al.28
performed immobilisation of monoclonal antibody against human
α-fetoprotein on amino group functionalised silica-coated magnetic
nanoparticles. In particular, conjugation of short single-stranded DNA
molecules (oligonucleotides) onto magnetic beads have been performed
by e.g. Lee et al.29 and Kouassi and Irudayaraj.30
The N-succinimidyl 3-(2-pyridyldithio)propionate (SPDP) molecule
is a heterobifunctional cross linker, which can be used to cross-link an
amino group containing entity and an aliphatic thiol group containing
entity through a disulphide bridge. The SPDP coupling chemistry, first
demonstrated by Carlsson et al.31 in 1978, is a reversible covalent
immobilisation method since the disulphide bond can be reduced with
dithiothreitol (DTT), thereby cleaving the bridge. Figure 13.1 taken from
Strömberg et al.32 schematically illustrates how thiolated single-stranded
oligonucleotides can be conjugated to magnetic beads having amino
groups on the bead surface. For cross-linking other entities, the
procedure is analogous. In step (I), the primary amino groups on the bead
surface are SPDP-activated by reaction with the N-hydroxysuccinimide
320 P. Svedlindh et al.
Fig. 13.1. Schematic illustration of covalent immobilisation of thiolated single-stranded
oligonucleotides on magnetic beads with amino groups using SPDP coupling chemistry.
In step (I), the N-hydroxysuccinimide ester group of the SPDP molecule reacts with the
amino group on the bead and in step (II), the 2-pyridyl disulphide group of the SPDP
molecule reacts with the oligonucleotide thiol group. Reprinted with permission from
Ref. 32. Copyright (2007) IOP Publishing Ltd.
ester group of the SPDP molecule. In step (II), the oligonucleotide
molecules attach to the bead by reaction between the oligonucleotide
thiol group and the 2-pyridyl disulphide group of the SPDP molecule.
Briefly, the practical procedure begins with resuspension of the
beads in PBS buffer (pH ~ 7) followed by addition of SPDP dissolved in
dimethylsulphoxide (DMSO) and the mixture is incubated at room
temperature during ~1 hr. After washing the beads by exchanging the
solvent a number of times, DTT reduced thiolated oligonucleotides are
added followed by incubation over night at room temperature. After
Bionanomagnetism 321
washing, any remaining SPDP-activated amino groups are blocked
by incubation with iodoacetamide at room temperature for ~1 hr,
thereby alkylating these sites. After some final washing steps, the as-
synthesised batch is obtained. In the following sections describing two
bionanomagnetism applications, the SPDP-coupling chemistry was used
for biofunctionalisation of amino group covered magnetic iron oxide beads
with oligonucleotides (Section 13.3) and protein molecules (Section 13.4).
13.3. An Example of a Recently Developed Magnetic Biosensor
Scheme — The Volume-Amplified Magnetic Nanobead
Detection Assay
In this section, a magnetic biosensor principle recently developed by
some of the authors is briefly described, namely the volume-amplified
magnetic nanobead detection assay (VAM-NDA). This bioassay
principle is a powerful combination between changes in dynamic
magnetic properties of magnetic beads and the use of high performance
molecular tools. However, we will begin with a brief theoretical
background relevant for the VAM-NDA method, i.e. frequency
dependent magnetic properties of magnetic beads and magnetic
relaxation mechanisms.
13.3.1. Dynamic magnetic properties and relaxation mechanisms of
magnetic beads
As mentioned in Section 13.2.1, the magnetic beads used in this work
are composed of a clustered core consisting of single domain maghemite
nanoparticles (each having a diameter ~15 nm) held together by a
dextran casing, where the nanoparticle moments are thermally blocked
relative to each other at room temperature.
According to the Debye theory,33 the complex low-field
susceptibility χ (ω ) for an ensemble of beads, reads
χ (ω ) = m(ω ) H AC = ( χ 0 − χ ∞ ) (1 + iωτ ) + χ ∞ (1)
where ω is the angular frequency of the applied AC magnetic field, χ ∞
is the high frequency susceptibility, χ 0 is the low field static
susceptibility, m(ω ) = m ′ (ω ) − im ′′(ω ) is the complex magnetisation,
322 P. Svedlindh et al.
H AC is the amplitude of the applied AC magnetic field and τ is the
characteristic relaxation time of the beads. The magnetic relaxation is
governed by either of two relaxation mechanisms: The Néel relaxation,34
where the magnetic moment rotates within the nanoparticles or the
Brownian relaxation,35 where the entire bead rotates in response to the
AC magnetic field. In the Néel relaxation process, the magnetic moment
needs to overcome an energy barrier Eb = KV p with the relaxation time
expressed as36
τ N = τ 0 exp ( Eb β ) (2)
where K is the magnetic anisotropy constant, V p is the nanoparticle
volume, β = kT is the thermal energy and τ 0 is a weakly temperature
and volume dependent microscopic relaxation time. In the Brownian
relaxation model, the characteristic relaxation time is instead given by
τ B = 3ηVB β . (3)
Here VB is the hydrodynamic volume of the bead and η is the dynamic
viscosity of the carrier liquid. Experimentally, in our case, the relaxation
−1
time is obtained from the Brownian relaxation frequency f B = (2πτ B ) ,
which is the frequency characterising the position of the peak in the m ′′
vs. frequency spectrum.
13.3.2. Brief overview of the volume-amplified nanobead detection
assay
The volume-amplified magnetic nanobead detection assay for detection
of single-stranded DNA targets, first time shown by Strömberg et al.,18
combines the Brownian relaxation biosensor principle10 with high
performance molecular tools, viz. padlock probe target recognition37,38
and rolling circle amplification (RCA).39,40 It is a non-optical magnetic
bioassay principle and is also of lab-on-a-bead type. By this, the VAM-
NDA method offers the possibility to design a new generation of
diagnostic devices, mostly of point-of-care and over-the-counter type, at
a substantially lower cost than those used today. Although the authors
hitherto only have demonstrated detection of DNA targets, other kinds of
biomolecules, viz. RNA and proteins, can be detected using the so-called
proximity ligation technique41,42 that produce RCA products as a
Bionanomagnetism 323
consequence of coincident binding of two or more specific antibodies
equipped with DNA strands.
Figure 13.2, taken from Ref. 18, schematically shows the VAM-
NDA bioassay principle. A collection of single-stranded target-DNA
molecules (black lines) is shown in the left part of the figure. Linear
padlock probe molecules (light grey lines) designed to exactly match the
target-DNA are added, which upon hybridisation with the target-DNA
molecules form circularised probe-target complexes. The ends of the
padlock probes are joined together by ligation. The addition of RCA
polymerase initiates RCA, and the circularised padlock probes are
amplified for a certain time (the RCA-time). After completion of RCA,
the sample contains ~1 µm-sized (for a RCA time ~1 hr) random-coil
single-stranded DNA molecules having a periodic sequence with
the complement of the padlock probe as the repeating motif. The
RCA-coils are detected by addition of single-stranded detection probe
oligonucleotide functionalised magnetic beads (total hydrodynamic
diameter ~150 nm), where the coupled oligonucleotides are
complementary to a region of the repeating motif in the RCA-coils. The
probe-tagged beads and the RCA-coils will – during their diffusive
motion – approach each other and the probe-tagged beads are then
incorporated into the coils by base-pair hybridisation (bead
immobilisation). The hydrodynamic size of the immobilised beads is thus
strongly increased, now essentially corresponding to the diameter of an
RCA-coil. Non-immobilised beads exhibit an unaltered hydrodynamic
diameter. When target-DNA is not present, no circularised padlock
probes and hence no RCA-coils are formed. In this case, all probe-tagged
beads remain free in solution. The lower right part of Fig. 13.2 shows the
m ′′ vs. frequency spectrum for one positive sample and for a negative
control sample. The negative control sample exhibits a well defined fB
value, and the peak magnitude gives a measure of the number of free
probe-tagged beads. The positive sample has essentially two relaxation
frequencies where the low-frequency peak (LFP) mainly corresponds to
single RCA-coils with immobilised beads and the high-frequency peak
(HFP) arises from probe-tagged beads that remain free in solution. The
two relaxation events have been resolved using a Cole-Cole fitting
procedure43 and the bold black curves show the two contributions to the
324 P. Svedlindh et al.
Fig. 13.2. Schematic illustration of the volume amplified magnetic nanobead detection
assay. In a positive sample (left part), single-stranded target DNA molecules (black lines)
are recognised by padlock probes (light grey lines), i.e. the 5′ and 3′ ends of the
padlock probe hybridise to the target strand, thereby forming a circularised probe-target
complex. After ligation and addition of RCA polymerases, each circularised complex is
enzymatically amplified during a certain time (the RCA-time) to a macromolecular DNA
coil (RCA-coil) having a periodic sequence with the complement of the padlock
probe as the repeating motif. The RCA-coils are detected by adding detection probe
oligonucleotide functionalised magnetic beads (filled black circles with dark grey lines)
exhibiting Brownian relaxation behaviour where the oligonucleotides are complementary
to part of the repeating sequence of the RCA-coils. Upon this, a certain amount of the
probe-tagged beads bind to the RCA-coils (bead immobilisation) depending on the RCA-
coil concentration. This in turn, compared to a negative control sample (right part), is
manifested as a decreased m′′ high frequency peak level (turn-off detection) and/or an
increased m′′ low frequency peak level (turn-on detection). Reprinted with permission
from Ref. 18. Copyright (2008) American Chemical Society.
Bionanomagnetism 325
measured positive sample curve. The extracted peak frequency values at
310 K are also indicated. By measuring the decrease in the m ′′ HFP
level relative to the corresponding level of the negative control sample,
quantitative DNA target concentration analysis can be performed.
Figure 13.3, taken from Ref. 18, shows the outcome of a quantitative
detection experiment of single-stranded target DNA using magnetic
beads with a bare diameter of 130 nm. Samples were incubated for
30 min at 343 K, the probe-tagged bead concentration was 1 nM and the
RCA-time was 1 hr. The complex magnetisation spectra were recorded at
310 K by a commercial SQUID magnetometer using an AC magnetic
field amplitude of 2 Oe. In order to achieve similar bead sedimentation
rates and chemical environment in all samples, the negative control
sample contained 100 pM of RCA-coils having a sequence non-
complementary to the detection probes on the beads.
Fig. 13.3. Quantitative turn-off type detection of single-stranded target DNA using
magnetic beads with a bare diameter of 130 nm. (a) shows imaginary part of the complex
magnetisation (normalised with respect to the solid content of beads) vs. frequency at
310 K for samples with different RCA-coil concentrations (c). Samples were incubated
for 30 min at 343 K, the probe-tagged bead concentration was 1 nM and the RCA-time
was 1 hr. In order to achieve similar bead sedimentation rates and chemical environment
in all samples, the negative control sample contained 100 pM of RCA-coils having
a sequence non-complementary to the detection probes on the beads. (b) shows a
more condensed representation of the data in panel (a), viz. ∆m′′ , defined by
∆m′′′ = m′′ (0) − m′′ (c ) at the high frequency peak vs. RCA-coil concentration. Reprinted
with permission from Ref. 18. Copyright (2008) American Chemical Society.
326 P. Svedlindh et al.
A limit of detection of ~3 pM was achieved. In this case, the
detection scheme is of turn-off type since it is based on the decrease in
the m ′′ HFP level due to immobilisation of magnetic beads in the RCA-
coils, i.e. the number of free beads decreases in the carrier liquid. On the
other hand, turn-on type detection relates to the increase in the m ′′ LFP
level when the number of beads immobilised in the coils increases.
Apparently, from Fig. 13.3, only turn-off type detection is possible using
130 nm beads since the m ′′ LFP levels do not correlate well with RCA-
coil concentration even though LFPs are clearly visible.
In Strömberg et al.,44 turn-on and turn-off detection possibilities were
investigated in more detail with respect to bead size. It was found that
when performing quantitative analysis with 40 nm beads as in Fig. 13.3,
both turn-off and turn-on type detection is possible. In the same paper, a
similar investigation was also presented for 250 nm beads showing that
even for RCA-coil concentrations as high as 300 pM, the observed LFPs
were broad and low in magnitude. These observations suggest that the
40 nm beads preferably immobilise such that a collection of more or less
well separated magnetic bead containing RCA-coils are formed (LFP
located at ~1 Hz at 310 K for 1 hr RCA-time). This implies that the
40 nm beads preferably bind to the interior of the coils. On the other
hand, the 250 nm beads immobilise in such a way that coil aggregates are
formed with sizes much larger than the size of a single RCA-coil. This
can only occur if the 250 nm beads cross-link coils by binding to the
exterior of the coils. These coil aggregates exhibit characteristic
frequencies below the low frequency limit (0.5 Hz) of the SQUID
experiments. For 130 nm beads, the situation is intermediate although
only turn-off detection is practically possible.
Also, in Ref. 44, it was found that for a given RCA-coil
concentration, the ratio between the number of immobilised beads and
the number of coils increased with decreasing bead size. This is
reasonable since the available free space inside a coil is limited and a
smaller bead size will therefore allow for a larger number of immobilised
beads. It was also found that smaller bead size yields faster
immobilisation kinetics. Furthermore, a larger number of detection probe
oligonucleotides per bead and a lower concentration of probe-tagged
beads were found to yield more efficient bead immobilisation.
Bionanomagnetism 327
On the one hand, according to the above discussion, from a bioassay
sensitivity point of view, one should use a low concentration of beads
and a small bead size. On the other hand, the magnetic moment per bead
will be small and using a low concentration of beads will yield a small
total magnetic moment of a sample. Since we have used a commercial
SQUID magnetometer, with a comparably low pick-up coil filling factor,
to record the complex magnetisation, we have not yet been able to
improve the detection limit of a few pM of target DNA. However, as we
will discuss further in Section 13.5, our intention is to replace the SQUID
magnetometer with a miniaturised magnetic field sensor with size
adapted to the size of the magnetic beads. This further development is
expected to considerably improve the detection sensitivity of the VAM-
NDA bioassay method.
Finally, as was pointed out in Refs. 18 and 44, the VAM-NDA
method can also be used for turn-off type multiplex analysis by using
different bead sizes for each type of target molecule such that the free
bead relaxation peaks (HPFs) of the respective bead sizes are well
separated in frequency. Presence of one kind of target results in a
decreased m ′′ HFP level for the corresponding bead size, presence of
two targets lead to decreased m ′′ HFP levels for the corresponding bead
sizes, and so on and so forth. It should be mentioned that for each
multiplexed sample, all contributions to the measured m ′′ profile have to
be extracted using a Cole-Cole fit procedure in order to see how each m ′′
HFP level has changed comparing to the corresponding negative control
levels. One also has to choose the largest bead size used in a multiplex
analysis test such that its HFP does not interfere significantly with the
frequency region where one finds the LFP for RCA-coils containing
magnetic beads. A qualitative proof of duoplex analysis was recently
presented in Ref. 44, where simultaneous detection of two kinds of
bacterial DNA sequences, viz. Vibrio cholerae (250 nm beads) and
Vibrio vulnificus (130 nm beads) is described.
13.4. Transportation and Release of Biomolecules Using Magnetic
Beads
Functionalised magnetic beads used in biosciences are usually
manipulated as large ensembles. A system that enables controlled
328 P. Svedlindh et al.
transport and separation of beads one-by-one would open up a large
number of possibilities in bioscience.16,17
In this section, a lab-on-a-chip method for single bead transport is
presented,15 making it possible to direct a particular bead into a specific
position. Proteins immobilised to the surface of the beads can be
successfully transported and released on a micrometer scale.45 The
essential part of the system consists of a substrate containing micron-
sized elements made of a soft magnetic material, the so-called transport
microchip. Exposed to an in-plane rotating magnetic field, an element is
continuously magnetised and demagnetised and will thus exert an
attractive force on beads varying in time as the magnetic field rotates.
The substrate of the bead transport microchip is a 5 × 5 mm2 (100)
silicon substrate with a native silicon oxide layer. The magnetic
structures were fabricated by either e-beam or optical lithography. The
pattern was written in a layer of positive or negative resist spin-coated
onto the substrate. After developing the pattern, the magnetic material
(Permalloy) was deposited by thermal evaporation and the procedure was
completed by a lift-off process. The magnetic properties of the elements
effectively depend on size and shape of the elements. In the case of
Permalloy ellipses, the easy/hard axis of magnetisation is along the
long/short axis of an element.
The magnetic domain structure of the elements has been investigated
using magnetic force microscopy (MFM). Figure 13.4 shows the domain
state of 2 × 6 × 0.1 µm3 ellipses in a weak in-plane magnetic field (field
direction indicated by the arrow). In Fig. 13.4(a), the ellipses in the
bottom row, having the field parallel to their long axes, are magnetised
close to saturation. Consequently, the fringing magnetic field is large at
the short ends of the ellipses. The ellipses in the top row, where the field
is parallel to the short axis, show a distorted multi-domain structure.
Even in this case, however, there is a weak net magnetisation in the
direction of applied field. In Fig. 13.4(c), the situation is reversed
compared to Fig. 13.4(a): Since the field now is parallel to the long axes
of the top row ellipses, these are in a state close to saturation. On rotating
the field, this domain state cycle will be repeated and the direction of net
magnetisation will rotate in phase with the field rotation. The domain
Bionanomagnetism 329
Fig. 13.4. MFM images of the magnetic domain structures of Permalloy ellipses
2 × 6 × 0.1 µm3. (a) corresponds to demagnetised magnetic elements, while (b-d) give the
domain states in an in-plane magnetic field of constant amplitude, 6.3 kA/m; arrows
indicate the directions of the magnetic field. Reprinted with permission from Ref. 15.
Copyright (2005) Wiley-VCH Verlag.
structures mapped by MFM are further verified by results from
computer simulations using the OOMMF micromagnetic solver.46 These
simulations15 indicate that the net magnetisation for a saturated ellipse is
five times larger than for a multi-domain ellipse.
Also triangular Permalloy elements have been prepared. Exposed to
an in-plane magnetic field, these elements show distorted multi-domain
states with a net magnetisation in the direction of the field.45 The
maximum fringing magnetic field will appear at the apex of a triangle
when the field is directed perpendicular to the base.
In conclusion, the net magnetisation of an element follows the
direction of the applied field and the magnitude of magnetisation will
depend on the direction of the magnetic field with respect to the element
geometry.
In the experimental setup, the transport microchip is fixed inside a
fluid cell and is observed using an optical microscope. Two mutually
perpendicular electromagnets generate a uniform in-plane rotating
magnetic field at the position of the elements. When the bead suspension
is injected into the fluid cell, the fringing magnetic fields due to the
Permalloy elements create attractive forces on close enough beads and
these beads will thus move towards the elements. Finally, they will stick
330 P. Svedlindh et al.
Fig. 13.5. Transport of magnetic beads (Dynabead M-270, diameter = 2.8 µm) on a
staircase pattern of Permalloy ellipses ( 2 × 6 × 0.1 µm3) in an in-plane rotating magnetic
field (6.3 kA/m). The arrows indicate the direction of the field. After one complete field
revolution, the beads have moved one step in the pattern as indicated by the white curve
in (f). Reprinted with permission from Ref. 15. Copyright (2005) Wiley-VCH Verlag.
to the rim of the elements at a point defined by the direction of the net
magnetisation.
Figure 13.5 shows two beads (dark circles) being transported in a
staircase pattern of Permalloy ellipses (white structures). The magnetic
particles will move along the perimeter of the ellipses, following the
rotation of the net magnetisation. In Fig. 13.5(d), the beads are attracted
by the strong force from the apex of the now saturated vertical ellipses.
Hence, the beads will jump to the neighbouring vertical ellipses.
Following the pace of the rotating field, the beads can move at a speed in
range 2–20 µm/s. The staircase pattern represents a one-way transport
line; the particles can only jump from a long side of an ellipse to the apex
of a neighbouring ellipse. Further, they follow the rim of ellipses with the
same sense of rotation as the external in-plane magnetic field. An
advantage of this feature is demonstrated in Fig. 13.6, which shows a
junction added to a transport line. When the field is clock-wise rotated,
the beads move around elements with the same sense of rotation and will
therefore enter the loop of elements (Fig. 13.6(a)). Anti-clock-wise
rotation will instead let the particles pass the junction without entering
Bionanomagnetism 331
Fig. 13.6. Separation of magnetic beads (Dynabead M-270, diameter = 2.8 µm) by
changing the sense of magnetic field rotation. The curled arrows in each figure show the
sense of field rotation. (a) Two particles are trapped inside the loop. A third particle is
located at the end position of the long transport line and is not further considered. (b)
When the first particle is approaching the second junction, the field rotation is turned
anti-clockwise. (c) The first particle avoided the loop and has continued moving straight
to the end of the long transport line. The field is then switched back to clockwise rotation
in order to trap the remaining particle in the loop. (d) The remaining particle is kept in the
loop and the two particles are thus separated. Reprinted with permission from Ref. 15.
Copyright (2005) Wiley-VCH Verlag.
the loop. Momentarily switching the sense of magnetic field rotation
(Fig. 13.6(b–d)), makes it possible to separate beads one-by-one.
It is easy to imagine that life-science molecules can be transported
and separated with the method described above. For this, we have used
so-called Micromer-M beads (cf. Section 13.2.1.) with the bead surface
covered by polyethylene glycol with amino function (PEG-NH2).
Proteins were conjugated to the amino groups using SPDP (cf. Section
13.2.2) and the beads were suspended in PBS buffer. In this case, the
major obstacle to a successful bead transport was sticking forces between
the bead proteins and the surface of the transport microchip.
332 P. Svedlindh et al.
Proteins can be classified as advanced polymers with more than 20
different amino acids as building blocks. These amino acids contribute
both with charges and hydrophobic regions in the polymer and since
amino acids are amphiphilic in nature the number of charges vary by pH.
Electrostatic interactions, hydrophobic interactions and hydrogen bonds
can thus be foreseen when proteins interact with surfaces, which makes
the situation quite complex. In aqueous media the contribution of
hydrogen bonding to the adsorption affinity is often marginal,47 since
these forces are constantly reorganising between different elements
where hydrogen bonds can be formed. Moreover, our results show that
the dominating sticking force, which may even inhibit bead
transportation, is the hydrophobic force.45
In our experiment, a bead (radius R = 2.5 ⋅103 nm) covered by IgG
on PEG (total length of organic molecule l ≈ 50 nm) has an area
in contact with the chip surface of approximately 2π R l = 8 ⋅ 105 nm2,
corresponding to 1% of the total bead surface.45 Alsteens et al.48 report
on chemical force microscopy where the interaction between a
hydrophobic tip and a self assembled monolayer of CH3- and OH-
terminated alkanethioles is estimated. Using their result for 50%
coverage of CH3-alkanethioles and taking only the part of the tip in
contact with the monolayer into account, the hydrophobic force becomes
0.02 nN/nm2. Applying this result to our beads, the hydrophobic force
between IgG covered beads and the chip surface will be of order 104 nN.
The possibility to manipulate magnetic beads by magnetic fields is
used in several biotechnological applications, including single molecule
force spectroscopy using magnetic tweezers,49,50 methods where
micromagnets are used to achieve a programmable assembly of magnetic
beads51 and microelectromagnets used for magnetic manipulation in lab-
on-a-chip systems.52 Magnetic forces on a bead fall off rapidly with
displacement between field source and magnetic bead. Consequently,
one can expect appreciable forces only when the bead is in reasonable
proximity to the magnetic field source. Magnetic tweezers, where the
field source is a single-pole electromagnet,50 have been developed with
pulling forces as large as 10 nN on a 4.5 µm bead at 10 µm distance from
the pole piece. Forces on magnetic beads interacting with thin film
micromagnets as well as with neighbouring magnetic beads were studied
Bionanomagnetism 333
in Ref. 51, with the attempt to develop a method for programmable
assembly of beads using magnetic microwell templates. Forces on a
second bead in close proximity to a microwell with one bead already
trapped, varied with the distance between the two beads; applying a large
enough magnetic field, forces close to 1 nN on a 8 µm bead were
predicted with a separation between the two beads equal to the bead
diameter.
Our case is different from the applications described above in that
the magnetic field from a magnetized element will not be uniform over
the magnetic bead volume. This also implies that the force acting on a
bead is non-uniform, in the sense that the part of the bead facing the
magnetic element will experience a larger force than more remote parts
of the bead. To estimate the force exerted by a saturated magnetic
element on a superparamagnetic bead one needs to calculate the
magnetic stray field H s (r) outside the element. The force acting on a
small volume element dV of the bead at r ′ is then given as
dF (r ′) = −∇E , where E = − µ 0 M (r ′) ⋅ (H s (r ′) + H a ) dV is the field
energy, M (r ′) the bead magnetisation and H a the applied magnetic
field assumed to be constant in magnitude over the bead volume. To
obtain the total force acting on the bead, one integrates dF (r ′) over the
bead volume. Calculations have been performed for Permalloy magnetic
elements of rectangular block geometry, with dimensions 2 µm × 6 µm ×
0.1 µm, and magnetic beads with a diameter of 2.8 µm.53 The beads
considered in the calculations were polystyrene beads with iron-oxide
nanoparticles evenly distributed inside the polystyrene matrix. With the
Permalloy magnetisation saturated along the long axis of the element, the
horisontal force component pointing towards the magnetic element varies
in magnitude between a few tenth of nN to some nN as the distance
between the centre of the bead and the short side of the element
decreases from a few µm down to some tenth of a micrometer. At the
same time, the vertical force component grows in magnitude, from a few
tenth of nN to become as large as some tens of nN. Even though the
calculations have been performed for an element geometry and bead size
slightly different from what is used in the present study, the results of the
calculations indicate the order of magnitude of the magnetic force
exerted by a magnetic element on a superparamagnetic bead.
334 P. Svedlindh et al.
In conclusion, the hydrophobic force between a protein covered bead
and the chip surface can be 100–1000 times larger in magnitude
compared to the magnetic force due to a magnetic element.
The effect of hydrophobic forces can be reduced by modifying the
surface with polymer brushes like polyethylene oxide (PEO) or
derivatives of PEO.54 Protein-surface interactions, where the surface is
covered with PEO, have been studied in detail in Ref. 55. Both the
polymer length and the surface density of the polymer are of crucial
importance for good protein resistance. Direct force measurements
between two adsorbed PEO surfaces in an aqueous solvent show that
repulsion forces develop at certain separation distances due to steric
repulsion phenomena.56–58 The physics of terminally attached polymer
surfaces has been studied theoretically by de Gennes.59
In this work, PEG derivatised with end groups of thiols has been
used as a polymer brush. The gold-thiol binding is useful for chemical
modification of a surface with organic molecules.60 The transport
microchip was therefore sputter coated with a 10 nm thick gold layer.
Our experimental results show that the hydrophobic interaction between
the protein conjugated magnetic beads and the surface of the transport
microchip is to a large extent eliminated by chemical grafting of the bead
transport microchip with PEG. Moreover, the bead transport was further
improved by adding a small amount of surfactant (0.5% w/v of Tween-
20) to the suspension. Using this arrangement, we have managed to
transport four different proteins with various properties regarding
isoelectric point and protein size, viz. Lyzozyme, α-Lactalbumin, Human
Serum Albumin (HSA) and Immunoglobulin G (IgG). Protein levels
down to one tenth of an atto mole were attached to the bead surface and
successfully transported on the microchip. By adding a reducing agent
like DTT to the suspension, the proteins are released instantaneously
from the beads.
13.5. Conclusions and Outlook
Summarising, the VAM-NDA bioassay principle,18,44 discussed in
Section 13.3, is a lab-on-a-bead type magnetic biosensor scheme, relying
on changes in Brownian relaxation of magnetic nanobeads10 and the use
Bionanomagnetism 335
of high performance molecular tools, viz. padlock probe target
recognition37,38 and RCA.39,40 It may be applied for detection of several
types of biomolecule targets; DNA (using padlock probes), RNA and
proteins (using proximity probes41,42). Due to its non-optical nature, the
VAM-NDA bioassay principle has the potential for future development
of low-cost diagnostic test devices of over-the-counter and/or point-of-
care type.
So far in the development work of the VAM-NDA method, we
have used a commercial SQUID-magnetometer to record the dynamic
magnetic properties of the samples. The used SQUID does not allow low
enough nanobead concentrations that are necessary for improving the
lower limit of detection of target molecules beyond the pM range.
Changing the read-out format by developing a miniaturised platform
based on a magnetoresistive sensor8,61 or a micro-Hall probe,7,62 is
expected to considerably improve detection sensitivity. Moreover, by
introducing a magnetic field sensor that covers a broader frequency
interval than that of the SQUID, a larger range of bead sizes can be
implemented thus allowing for improved multiplex detection strategies.
In Section 13.4, a method capable of controlled transport and release
of proteins at well defined positions on the chip surface is described.
Although demonstrated for proteins, the method is useful for any kind of
molecule. To our knowledge, this is the first report of controlled
transport of sub-attomole levels of molecules with micrometer precision.
The transport system is flexible with respect to dimensions and can be up
or down scaled by at least a factor of 10.
Potential applications of the system are in the field of separation
technology and as analytical system since during transport on the chip,
the protein covered beads are controlled one-by-one. A real time
analytical system testing e.g. for allergic disease and monoclonal
antibody-antigen interactions can be constructed where beads conjugated
with antigens continuously are transported through a sample containing
region of the transport microchip. With suitable miniaturised magnetic
field sensors added to the chip, counting and numbering of the beads can
be performed. This yields an effective encoding system, where beads
carrying a particular antibody-antigen combination can be directed to a
specific transport line and beads carrying other antibody-antigen
336 P. Svedlindh et al.
combinations to other lines, thus forming the building block for an
effective multiplex analysis system.
It may even be possible to combine the two here described methods.
In a diagnostic system making use of miniaturised magnetic field sensors
to detect the response of magnetic beads to AC and/or DC magnetic
fields, it will be necessary to guide magnetic beads to come in close
proximity to the field sensors. This can be accomplished by using the
here described transport system for bioconjugated magnetic beads. An all
magnetic multiplex analysis system can be constructed where different
bead sizes are used for different target DNA molecules. With magnetic
field sensors capable of single-bead-detection,7,8 different bead sizes can
be separated by the magnitude of the DC field induced bead magnetic
moments and transported to different sites on the chip where the
existence or non-existence of RCA-coils is determined from the
Brownian relaxation behaviour of the beads. Quantitative analysis can be
performed by either measuring the response of magnetic beads one-by-
one as they pass by the magnetic field sensor or by collecting a
comparably large number of beads close to the magnetic field sensor
before measuring the AC magnetic field response of the ensemble of
beads.
13.6. Abbreviations and Acronyms of Chapter 13
AC Alternate Current
DC Direct Current
DMSO DiMethylSulphOxide
DNA DeoxyriboNucleic Acid
DTT DiThioThreitol
HFP High-Frequency Peak
HSA Human Serum Albumin
IgG ImmunoGlobulin G
LFP Low-Frequency Peak
MFM Magnetic Force Microscopy
MRI Magnetic Resonance Imaging
Bionanomagnetism 337
PEG PolyEthylene Glycol
PEO PolyEthylene Oxide
PBS Phosphate Buffered Saline
RF Radio-Frequency
RCA Rolling Circle Amplification
RNA RiboNucleic Acid
SPDP N-succinimidyl 3-(2-
pyridyldithio)propionate
SQUID Superconducting QUantum Interference
Device
VAM-NDA Volume-Amplified Magnetic Nanobead
Detection Assay
T1-decay, T2-decay Relaxation processes in MRI
ω Angular frequency of applied AC
magnetic field
χ (ω ) Complex low-field magnetic susceptibility
χ∞ High-frequency magnetic susceptibility
χ0 Low-field static magnetic susceptibility
m(ω ) = m ′(ω ) − im ′′ (ω ) Complex magnetisation
H AC Amplitude of applied AC magnetic field
τ Characteristic relaxation time of a
magnetic bead
τN Néel relaxation time
Eb Energy barrier (in connection to Néel
relaxation)
K Magnetic anisotropy constant
Vp Magnetic nanoparticle volume
(in connection to Néel relaxation)
β = kT Thermal energy
τ0 Microscopic relaxation time
(in connection to Néel relaxation)
338 P. Svedlindh et al.
τB Brownian relaxation time
VB Hydrodynamic volume of a magnetic bead
η Dynamic viscosity of the carrier liquid of a
ferrofluid
fB Brownian relaxation frequency
R Magnetic bead radius
l Length of organic molecule
(in connection to PEG)
H s (r) Magnetic stray field ouside Permalloy
element
dV Small volume element of a magnetic bead
r′ Position inside a magnetic bead
dF ( r ′ ) Force acting on a small volume element,
dV, of a magnetic bead at r ′
E Field energy
M (r ′) Bead magnetisation
Ha Applied magnetic field
Acknowledgments
We are grateful to M. Strømme, T. Zardan Gomez de la Torre,
M. Nilsson, J. Göransson, L.-E. Johansson and K. Eriksson for
collaboration and valuable discussions. Financial support from the
Swedish Research Council (VR) and the Knut and Alice Wallenberg
Foundation (KAW) is gratefully acknowledged.
References
1. S. Mornet, S. Vasseur, F. Grasset and E. Duguet, J. Mater. Chem. 14, 2161-2175
(2004).
2. J. M. Perez, L. Josephson, T. O´Loughlin, D. Högermann and R. Weissleder, Nature
Biotechnology 20, 816-820 (2002).
3. J.-H. Lee, Y.-M. Huh, Y.-W. Jun, J.-W. Seo, J.-T. Jang, H.-T. Song, S. Kim,
E.-J. Cho, H.-G. Yoon, J.-S. Suh and J. Cheon, Nature Medicine 13, 95-99 (2007).
Bionanomagnetism 339
4. M. A. McAteer, N. R. Sibson, C. von zur Muhlen, J. E. Schneider, A. S. Lowe,
N. Warrick, K. M. Channon, D. C. Anthony and R. P. Choudhury, Nature Medicine
13, 1253-1258 (2007).
5. P. Moriz, S. K. Jones and B. N. Gray, Int. J. Hyperthermia 18, 267-284 (2002).
6. T. Osaka, T. Matsunaga, T. Nakanishi, A. Arakaki, D. Niwa and H. Iida, Anal.
Bioanal. Chem. 384, 593-600 (2006).
7. G. Mihajlovic, P. Xiong, S. von Molár, K. Ohtani, H. Ohno, M. Field and G. J.
Sullivan, Appl. Phys. Lett. 87, 112502 (2005).
8. D. L. Graham, H. A. Ferreira and P. P. Freitas, Trends in Biotechnology 22, 455-462
(2004).
9. Y. Amemiya, T. Tanaka, B. Yoza and T. Matsunaga, J. Biotechnol. 120, 308-314
(2005).
10. J. Connolly and T. G. St Pierre, J. Magn. Magn. Mater. 225, 156-160 (2001).
11. F. Vartdal et al., Tissue Antigens 28, 301-312 (1986).
12. H. H. Oberg, D. Wesch, J. Lenke and D. Kabelitz, Scand. J. Immunol. 64, 353-360
(2006).
13. M. J. Lynch, C. G. Leon-Velarde, S. McEwen and J. A. Odumeru, J. Microbiol.
Methods 58, 285-288 (2004).
14. C. M. Bearden, B. K. Book, R. A. Sidner and M. D. Pescovitz, J. Immunol Methods
300, 192-199 (2005).
15. K. Gunnarsson, P. E. Roy, S. Felton, J. Pihl, P. Svedlindh, S. Berner, H. Lidbaum
and S. Oscarsson, Adv. Mater. 17, 1730-1734 (2005).
16. T. Deng, G. M. Whitesides, M. Radhakrishnan, G. Zabow and M. Prentiss, Appl.
Phys. Lett. 78, 1775-1777 (2001).
17. B. R. Yellen and G. Friedman, Adv. Mater. 16(2), 111 (2004).
18. M. Strömberg, J. Göransson, K. Gunnarsson, M. Nilsson, P. Svedlindh and
M. Strømme, Nano Lett. 8, 816-821 (2008).
19. G. Li, V. Joshi, R. L. White, S. X. Wang, J. T. Kemp, C. Webb, R. W. Davis and
S. Sun, J. Appl. Phys. 93, 7557 (2003).
20. T. Neuberger, B. Schöpf, H. Hofmann, M. Hofmann and B. V. Rechenberg, J. Magn.
Magn. Mater. 293, 483-496 (2005).
21. P. Majewski and B. Thierry, Critical Rev. Solid State Mater. Sci. 32, 203-215
(2007).
22. B. Gillot, A. Rousset and G. Dupre, J. Solid State Chem. 25, 263-271 (1978).
23. G. Hägg, Allmän Och Oorganisk Kemi (Almqvist och Wiksell, Uppsala 1969).
24. Micromod Partikeltechnologie GmbH, www.micromod.de .
25. Dynal Biotech, www.invitrogen.com .
26. Z. Ma, Y. Guan, X. Liu and H. Liu, Polym. Adv. Technol. 16, 554-558 (2005).
27. K. Aurich, M. Schwalbe, J. H. Clement, W. Weitschies and N. Buske, J. Magn.
Magn. Mater. 311, 1-5 (2007).
28. J.-L. Gong, Y. Liang, Y. Huang, J.-W. Chen, J.-H. Jiang, G.-L. Shen and R.-Q. Yu,
Biosensors and Bioelectronics 22, 1501-1507 (2007).
340 P. Svedlindh et al.
29. C. W. Lee, K. T. Huang, P. K. Wei and Y. D. Yao, J. Magn. Magn. Mater. 304,
e412-e414 (2006).
30. G. K. Kouassi and J. Irudayaraj, Anal. Chem. 78, 3234-3241 (2006).
31. J. Carlsson, H. Drevin and R. Axén, Biochem. J. 173, 723-737 (1978).
32. M. Strömberg, K. Gunnarsson, H. Johansson, M. Nilsson P. Svedlindh and
M. Strømme, J. Phys. D: Appl. Phys. 40, 1320-1330 (2007).
33. P. Debye, Polar Molecules (The Chemical Catalogue Company, New York 1929).
34. L. Néel, Ann. Geophys. 5, 99-136 (1949).
35. W. F. Brown Jr., J. Appl. Phys. 34, 1319-1320 (1963).
36. W. F. Brown Jr., Physical Review 130, 1677-1686 (1963).
37. M. Nilsson, H. Malmgren, M. Samiotaki, M. Kwiatkowski, B. P. Chowdhary and
U. Landegren, Sci. 265, 2085-2088 (1994).
38. U. Landegren, F. Dahl, M. Nilsson, S. Fredriksson, J. Banér, M. Gullberg, J. Jarvius,
S. Gustafsdottir, O. Söderberg, O. Ericsson, J. Stenberg and E. Schallmeiner,
Comparative and Functional Genomics 4, 525-530 (2003).
39. A. Fire and S.-Q. Xu, Proceedings of the National Academy of Sciences of U.S.A.
92, 4641-4645 (1995).
40. D. Liu, S. L. Daubendiek, M. A. Zillman, K. Ryan and E. T. Kool, J. Am. Chem.
Soc. 118, 1587-1594 (1996).
41. O. Söderberg, K. J. Leuchowius, M. Kamali-Moghaddam, M. Jarvius, S. Gustafsdottir,
E. Schallmeiner, M. Gullberg, J. Jarvius and U. Landegren, Genetic Engineering 28,
85-93 (2007).
42. O. Söderberg, M. Gullberg, M. Jarvius, K. Ridderstråle, K.-J. Leuchowius,
J. Jarvius, K. Wester, P. Hydbring, F. Bahram, L.-G. Larsson and U. Landegren,
Nature Methods 3, 995-1000 (2006).
43. K. S. Cole and R. H. Cole, J. Chem. Phys. 9, 341-351 (1941).
44. M. Strömberg, T. Zardán Gómez de la Torre, J. Göransson, K. Gunnarsson,
M. Nilsson, M. Strømme and P. Svedlindh, Biosensors and Bioelectronics 24,
696-703 (2008).
45. L.-E. Johansson, K. Gunnarsson, S. Bijelovic, K. Eriksson, A. Surpi, P. Svedlindh
and S. Oscarsson, unpublished (2008).
46. http://math.nist.gov/oommf/software.html.
47. M. D. Joesten and L. J. Schaad, in Hydrogen Bonding (Marcel Dekker, New York
1974).
48. D. Alsteens, E. Dague, P. G. Rouxhet, A. R. Baulard and Y. F. Dufrêne, Langmuir
23, 11977-11979 (2007).
49. K. C. Neuman and A. Nagy, Nature Methods 5, 491-505 (2008).
50. A. R. Bausch, F. Ziemann, A. A. Boulbitch, K. Jacobson and E. Sackmann,
Biophys. J. 75, 2038-2049 (1998).
51. B. B. Yellen and G. Friedman, Langmuir 20, 2553-2559 (2004).
52. K. Smistrup, P. T. Tang, O. Hansen and M. F. Hansen, J. Magn. Magn. Mater. 300,
418-426 (2006).
Bionanomagnetism 341
53. Johan Pihl, 2004, Field Manipulation and Design of Thin-film Structures for
Programmable Motion of a Magnetic Bead, UPTEC FO4 086, diploma work at
Uppsala University.
54. W. Norde, Colloids and Interfaces in Life Sciences (Marcel Dekker, New York
2003).
55. S. I. Jeon, J. H. Lee, J. D. Andrade and P. G. De Gennes, J. Coll. Interf. Sci. 142,
149-158 (1991).
56. J. Klein and P. Luckham, Nature (London) 300, 429 (1982).
57. J. Klein and P. Luckham, Macromol. 17, 1041 (1984).
58. P. Luckham and J. Klein, Macromol. 18, 721 (1985).
59. P. G. de Gennes, Macromol. 13, 1069 (1980); ibid. 14, 1637 (1981); ibid. 15, 492
(1982).
60. R. G. Nuzzo and D. L. Allara, J. Am. Chem. Soc. 105, 4481 (1983).
61. P. P. Freitas, R. Ferreira, S. Cardoso and F. Cardoso, J. Phys.: Condens. Matter 19,
165221 (2007).
62. A. Sandhu, Y. Kumagai, A. Lapicki, S. Sakamoto, M. Abe and H. Handa,
Biosensors and Bioelectronics 22, 2115-2120 (2007).
This page is intentionally left blank
Chapter 14
DOMAIN WALLS FOR LOGIC AND DATA STORAGE
APPLICATIONS
Colm C. Faulkner
Centre for Research on Adaptive Nanostructures and Nanodevices
Trinity College Dublin, Dublin 2, Ireland
E-mail: [email protected]
14.1. Introduction
New properties emerge in magnetic nanostructures when dimensions
become comparable to length scales such as the spin diffusion length, the
magnetic exchange length, and the domain wall width. Additionally as
dimensions are reduced, element geometry, and the proximity of interfaces
strongly influences magnetic properties. Therefore new applications emerge.
Once it is technologically feasible to control the geometry of
ferromagnetic structures on the nanoscale, it is possible to isolate and
control the position of domain walls on these length-scales. Due to
advances in fabrication technology, and improved spatially and
temporally resolved magnetic microscopy techniques, magnetic domain
walls (DWs) have been heavily investigated in recent years. The
predicted breakdown of Moore’s law in the semiconductor industry,1
where the number of transistors on a fixed area of computer circuit board
doubles every two years, means that emerging magnetic materials and
technologies may play an important role in future logic processing
devices, as well as improved data storage applications.
Devices which link both charge and magnetization were introduced
in 1991 by IBM, in the form of hard disk drives using anisotropic
magnetoresistance, where the resistance of a magnetic device changes
according to the orientation of the magnetic layers. Many of these
devices, utilizing the spin degree of freedom of the electron, and labelled
“spintronic” devices may feature magnetic domain walls in the future.
343
344 C. C. Faulkner
Domain walls, which we can consider a well defined entity like a soliton,
can be used to manipulate information using magnetic fields, electric
current, or a combination of both. Therefore a treatment of domain wall
mediated switching and dynamics in patterned magnetic nanostructures
for future device applications is presented herein.
14.2. Theory
A domain wall is a spatially localized mobile boundary between
two regions of opposing magnetization states. The propagation fields,
velocity, and structure of magnetic domain walls were initially
predominately studied in thin films, either experimentally,2,3 or
computationally, following from the classic study of magnetic domain
walls travelling in wires and elongated iron whiskers.4 Some recent
advances in the experimental study of domain walls and domain
configurations include the observation of domain wall motion with sub
nanometre resolution (essentially at an atomic length scale) in garnets by
Hall Probe,5 magnetic force microscopy (MFM) tip stray field induced
domain wall displacement,6 and the direct experimental observation of
vortex core type domain formations in 50 nm thick NiFe elements.7 Spin
polarized scanning tunnelling microscopy, has also been employed to
map the in-plane and out of plane magnetization of Fe nano-islands in
the vortex configuration to sub 10 nm resolution.8
Generalized modes of reversal for a ferromagnetic monodomain
particle, such as the curling, buckling and nucleation modes were treated
by Frei et al. in 1957.9 Generally the wire structures discussed in this
chapter are the soft ferromagnetic alloy Permalloy, Ni80Fe20, chosen for
its low coercivity and magnetostriction, high Curie temperature, and
widespread use in magnetic recording head sensors. Additionally
Permalloy exhibits a strong anisotropic magnetoresistance effect (AMR),
which is useful for point contact measurements. The length scales treated
are typically 3–50 nm for wire thickness, 100–500 nm for wire width,
and wire lengths of the order of microns, unless explicitly stated
otherwise. Structures on these length scales are generally single domain
at room temperature (geometry dependent), and reside above the super
paramagnetic limit at 293 K. Planar wires are treated, as they are
straightforward to fabricate by standard electron beam lithography,
Domain Walls for Logic and Data Storage Applications 345
although much work has been performed on domain walls with a circular
cross section.10
Typically an elongated Permalloy planar nanowire of these
dimensions does not support a DW at remanence. They are bi-stable at
remanence, having 2 allowable magnetization configurations, with spins
aligned parallel to wire sides due to shape anisotropy. It is energetically
more efficient for samples of these dimensions to reside in a ground state
of 1 ferromagnetic domain. However a magnetic domain wall can be
reproducibly situated in such samples, by methods including geometric
modification,11 current carrying strip lines12,13 or an appropriate field
sequence.14
In thin films the domain wall type depends on intrinsic material
properties, but for a nanowire the geometry, as well as applied fields and
currents have an important effect on the wall spin structure. For a given
material, the DW type is governed by the wire geometry,15 and the spin
structure is the result of an energy minimization process. For soft
magnetic materials such as Permalloy, the two basic domain wall types
are transverse where spins rotate in the plane of the wire (either head-to-
head or tail-to-tail), Fig. 14.1(a), and vortex where spins curl clockwise
or counterclockwise around an out of plane vortex core (Fig. 14.1(b)).
Generally a vortex DW occurs in thicker and wider nanowires. (Flux
closure reduces surface charge and demagnetizing fields, leading to a
lower energy ground state, than a transverse wall in a wire of identical
dimensions).
14.3. Wire Switching
Magnetic nanowires are excellent candidates for device applications,
because a very high level of control is exercised over magnetization
reversal mechanisms, switching fields, and increasingly the reversal
rates. First consider high aspect ratio, single domain ferromagnetic
nanowires. Magnetization reversal is generally triggered by a nucleation
event, typically at an imperfection or wire end, followed by DW
propagation, leading to rapid domain expansion and the reversal of the
entire nanowire. Data plotted in Fig. 14.2 is for 6 element arrays of
elongated planar Permalloy nanowires from 60–500 nm in width. For the
346 C. C. Faulkner
wire dimensions studied, internal dipolar fields control the coercivity, Hc,
i.e. it depends on the demagnetizing field of the sample. Hc drops rapidly
with increasing wire width, and excellent control of the switching
properties of the wires is evident from the data plot. However it is noted,
that as the wire width is reduced to dimensions of the order of 100 nm,
desirable for device applications, the coercive field becomes very high –
of the order of 100’s of Oersted. From Stoner-Wohlfarth analysis, an
absolute upper bound of Hc for an elongated particle with uniaxial
anisotropy is given by an analytic expression of the form:17
Fig. 14.1. Micromagnetically determined spin structured of a transverse head-to-head (a)
and vortex domain wall (b) in a thin Permalloy strip. The transverse domain wall resides
in wires of reduced thickness and width, relative to the vortex wall (c).16
Domain Walls for Logic and Data Storage Applications 347
8M s T
Hc = ⋅S (1)
W
where Ms is the saturation magnetization, T the thickness, W the wire
width, and S is a finite length shape anisotropy factor. This is the
magnetostatic energy barrier associated with reversing magnetization of
the entire wire in-plane, parallel to the short axis – reversal by coherent
rotation. The expression is appropriate in the thin film limit, where
sample magnetization is constrained in-plane, and micromagnetic
configurations such as vortices with out of plane spins are not
energetically favoured.
1
( ρ − ρ )
S= . (2)
1+ ρ 2
Here, ρ = L / W is the wire in-plane aspect ratio. The discrepancy
in switching data between the experimentally measured results and the
analytically calculated results shown is attributed to thermal effects and
sample side wall imperfections inherent in the fabrication process. In
short we note that excellent control over ferromagnetic nanowires is
possible, but switching fields are very high as critical dimensions are
reduced, requiring more power for magnetic switching to occur.
However, wire geometries can now be engineered such that one or more
DWs reside in the system in the absence of field. Then magnetization
reversal can be effected reproducibly at low DW propagation field
values, as opposed to the intrinsic nucleation field of a wire.11,18
14.4. Domain Wall Propagation
When an external magnetic field is applied to a static domain wall in a
nanowire, the wall feels a change in potential, and local pinning can be
overcome. In theory a DW in a perfect planar nanowire, with no
roughness or defects, will propagate at a negligible field, and recent
experimental reports demonstrate that the DW can propagate over
reasonable length scales with negligible power dissipation,19 and
propagation fields of circa 10 Oe, an order of magnitude or lower better
348 C. C. Faulkner
600
500
400
H (Oe)
2 Pm
300
c
200
100
0
0 100 200 300 400 500
Wire Width (nm)
Fig. 14.2. Experimentally measured switching field, Hc, [•], for Ni81Fe19 planar
nanowires of the following dimensions: width = (60–500 nm), length = (9 µm), film
thickness = (5 nm). The dashed line (-----) represents an analytic solution of Hc. Inset is a
secondary electron image from a W = 300 nm 6-element planar wire array.
than nucleation fields for these structures, 20 see also Fig. 14.2. We note
also that power gain is a critical consideration for device applications.
It is therefore desirable to isolate a domain wall in a magnetic
nanostructure, and investigate domain wall propagation, as a distinct type
of magnetization reversal process.
14.5. Domain Wall Injection
In the last decade advances in nanolithography have enabled the
fabrication and investigation of magnetic nanostructures with
increasingly complex geometries. It is now possible to engineer the
Domain Walls for Logic and Data Storage Applications 349
dimensions of magnetic structures in 3 dimensions on the nanoscale, and
therefore engineer the switching mechanism with great precision. Using
the giant magnetoresistance effect in a single NiFe/Cu/NiFe nanowire,
Ono and co-workers detected the propagation of a single domain wall in
a high aspect ratio nanowire, at temperatures from 77–160 K.18 The
trapping of a single head-to-head domain wall about a notch in a
submicrometer NiFe(200 Å)/Cu(100 Å)/NiFe(50 Å) wire was described.
In order to separate DW nucleation and propagation fields, and to
control the position of a DW in a nanostructure with deep sub-micron
accuracy, structures are fabricated with large low aspect ratio pads
attached to thin elongated nanowires. A high aspect ratio structure is
ferromagnetically “soft”, having a low switching field. A DW can be
localized at the junction between the ‘soft’ and ‘hard’ regions of the
patterned structure, and then selectively propagated into the thin
nanowire. Nucleation pads of relatively large lateral dimensions (∼µm)
relative to a wire afford a technique to artificially introduce domain walls
into nano-wires for experiment, at applied fields below the wire critical
nucleation field. Several groups have utilized domain wall injection pads,
as a domain wall source to perform transport, propagation, and pinning
studies on magnetic domain walls. One of the first examples of this kind
of experimental geometry is the work of Fournel et al. in samples of
(CoNi/Pt) defined by electron beam lithography.21 The technique of
controlling magnetization reversal of a continuous wire structure, by
field mediated domain wall reversal was extended to 150 nm wide NiFe
wires investigated by GMR,22 with measurements by MFM,23 MOKE,24
and the ballistic Hall effect25 subsequently performed. A variety of pad
geometries (circular, square, diamond) were investigated in the GMR
measurement of trilayer structure switching.26
By tuning the injection pad geometry (rectangular, circular, diamond
shaped, triangular), pad aspect ratio, and the ratio of pad to output wire
thickness, the DW injection field can be very well controlled, at fields
markedly lower than the injection field of a similar padless wire
structure.11 For example in the data presented here (Fig. 14.3), the
elongated DW nucleation pad, aspect ratio 5:1, switched at HC = 44 Oe,
and time averaged magneto-optical Kerr effect hysteresis loops,
measured locally with a 5 µm laser spot diameter, give evidence of
350 C. C. Faulkner
1.1752
MOKE Signal (V)
1.1751
1.175
1.1749
1.1748
1.1747
1.1746
1.1745
-100 -50 0 50 100
Applied Field (Oe)
2 µm
Fig. 14.3. Locally measured time averaged M-H loop on a 5 nm thick, 200 nm wide
Permalloy nanowire, with a 600 nm, 3 µm wide domain wall injection pad at the left end,
and corresponding micrograph of the sample.
reproducible domain wall switching at all points along the wire at 44 Oe.
This controlled injection and propagation of domain walls at low field
values relative to the nucleation field of structures without DW injection
pads has even been demonstrated in wires kinked at 30°.11
These switching values are markedly lower than the nucleation field
of a pad free control structure — 150 Oe. Domain wall motion occurs in
the direction of applied field. Magnetization seeks to align itself with the
applied field, and this re-alignment is affected by the propagation of the
domain wall boundary. By modifying the geometry of wire ends with a
domain wall injection pad, not only can the reversal field and mechanism
for magnetization reversal be engineered, the initial reversal event can be
Domain Walls for Logic and Data Storage Applications 351
localized to one wire end. Thus the direction of magnetization reversal
by field controlled domain wall propagation can be controlled.
14.6. Rounded Corner Structures
Intentionally introducing an asymmetry into magnetic nanostructures
affords more control over the magnetization reversal process, and is
therefore an avenue to future applications. Ferromagnetic nanostructures
with rounded corner features are an attractive experimental system for
fundamental investigations,14,27–29 and possible applications such as
magnetic random recording memory — MRAM.30 Once a DW is present
in a nanowire, the DW propagation field is relatively low, being only
limited by the intrinsic wire coercivity, and not, for example, a DW pad
depinning field from a pinning centre, or a domain wall nucleation field.
In rounded corner structures (Fig. 14.4), domain wall nucleation and
propagation can be easily separated, by applying a series of external
fields in different directions. If a DW can be isolated in a rounded wire,
many of its properties, such as velocity,27 mass,31 or response to spin
polarized current32,33 can be probed. The L-shape,34 or corner geometry is
a geometry in which a domain wall is easily stabilized and isolated.
A rotating magnetic field is used to propagate DWs along wires that
change direction and turn corners.35 The phase of the transverse field,
Hy, and the longitudinal field, Hx, must be separated by π/2, generating a
field vector that rotates in an anti-clockwise direction. A DW may
propagate around a corner structure if the field and corner are of the
same chirality or “handedness”. Care should be taken to match the x and
y field components to the corner geometry, so a domain wall is not
pinned at a corner.
Figure 14.4 depicts a micrograph of a continuous planar Permalloy
wire structure on a Si substrate, fabricated by subtractive focused ion
beam milling. Upon application of a suitable magnetic field in the ‘y’
plane, a domain wall is nucleated in the domain wall injection pad, ‘A’,
and propagates to corner ‘B’. The domain wall is confined within the
corner, until an ‘x’ field propagates a domain wall along the structure
from ‘B’ to ‘E’. With no component of external field in the x-direction, the
wall remains at the corner. A rotating magnetic field of correct chirality
352 C. C. Faulkner
can depin the domain wall from corner ‘B’ at a very low depinning field.
Figure 14.5 is a typical hysteresis loop measured at point ‘C’ showing
domain wall switching at ~10 Oe, time averaged over 1,300 field cycles.
This very reproducible switching, and evidence of very low DW reversal
fields has also been evidenced micromagnetically and proposed for data
storage applications.37 Experiments can be performed in the long output
wire, at lower fields than those possible when using solely a DW
injection pad and output wire (Hinj ∼ 40 Oe).
14.7. Domain Wall Localisation/Trapping/Point Contacts
A domain wall in a nanowire can be considered as a quasi-particle. The
study of DW propagation is closely related to the study of varying energy
landscapes and pinning sites, such as grain boundaries, defects, edge
roughness, sample thickness inhomogeneity, and dislocations. This topic
in nanoscale magnetic systems is of interest for a number of reasons.
Fig. 14.4. Backscattered secondary image of a rounded corner geometry NiFe planar
wire structure [light grey] on a Si substrate [dark gray]. ‘A’ indicates a DW injection pad,
‘B’ a smoothly rounded corner, ‘C’ and ‘E’ are hysteresis loop measurement positions,
and ‘D’ a lithographically defined defect. Inset i-ii show the trapping centre in more
detail.36
Domain Walls for Logic and Data Storage Applications 353
Fig. 14.5. Hysteresis loop evidencing reproducible domain wall propagation from a
corner structure at extremely low field value.36
Firstly, fabrication limits for electron beam lithography and ion
beam milling techniques38 are continuing to improve, making possible
the study of magnetization reversal processes in very complex wire
geometries.39,40 As the dimensions of nanomagnets continue to decrease,
the role of defects, interfaces, and surface effects in these quasi 1-D
systems becomes more important,41 and new physical phenomena may
be detected. Secondly, in the drive towards higher density magnetic
random access memory systems, workers have found that magnets of
reduced lateral dimensions have higher switching fields, costing more
power and dissipating more energy per reversal event. One method of
circumventing this problem is to use a ‘domain wall trap’ system,37 now
extended to the notched ring or ‘doughnut’ system,42 where a domain
354 C. C. Faulkner
wall always resides in the system. Thus magnetization reversal is
achieved over low field values, necessary for depinning and propagating
a domain wall, as opposed to the high field values needed to activate
nucleation of magnetic domains and associated magnetization reversal.
In addition traps afford a means of controlling the position of domain
walls at low field values for the purpose of spin current experiments,43
room temperature magnetoresistance measurements, dynamics experiments,
or research into magnetic tunnelling processes.44
Different geometrical variations generate a potential landscape that
exerts forces on the domain wall. By tuning the geometry, the energy
landscape can be controlled. Domain walls can be localized in a wire by
bends or sharp L-shapes,34 traps, protrusions, injection pads, wire
junctions and intersections.35,45–47 For example a perpendicular wire on a
long Permalloy nanowire, in a T-shaped junction, can act as a DW gate,
either blocking a DW or letting it propagate down the horizontal wire,
based solely on the magnetic orientation of the vertical wire.47 Another
method of creating a potential barrier is by subtractively milling a thin
line from the top of the wire by focused ion beam.48 Figure 14.6 shows
bi-stable DW mediated switching at a trapping site, with sharp 2-step
switching transitions desirable for applications. Figure 14.7 shows the
increase in switching field with trap depth for M-H loops measured post
DW interaction with a trap, for samples similar to Fig. 14.4.
14.8. Domain Wall Protrusion
The concept of a domain wall notch or trap has been extended to
protrusions which extend outwards from the wire side instead of cutting
a section out of the wire sidewall.20,29,52 The energy potential of
these structures was recently investigated and compared to notches.20
Depending on the system characteristics the pinning centre can be either
attractive, or repulsive. The term “anti-notch” has also been coined,29 and
it is noted that a wall may take positions in front of, partially in front of,
or inside the trap.
Domain Walls for Logic and Data Storage Applications 355
Norm alised M OKE Signal H - Domain
1
1
w all enters
trap
H - Domain
2
-1 w all leaves
trap
-80 -40 0 40 80
X Field (Oe)
Fig. 14.6. Hysteresis loop taken at position D, with the MOKE laser spot positioned over
a single nano-trap of depth 125 nm and width 240 nm. Loop was averaged for 60 s – for
1600 field cycles.19
14.9. Domain Wall Chirality
A 180° domain wall, also known as a transverse wall can point up or
down at rest. Applying transverse fields changes the axial field needed
to inject a domain wall from a pad of low coercivity material,49 and
Bryan also observed by single shot MOKE measurements drastically
different switching fields for a wire output under identical field
conditions. Evidence was presented for different domain wall states
being present at junctions between domain wall injection pad and the
output wire, leading to different switching field for the output wire.
Hayashi et al. successfully imaged 4 different wall type at same notch
by magnetic force microscopy, measuring markedly different de-pinning
field for each.50 Subsequently it was proposed that by breaking the 1-D
symmetry of an elongated nanowire with a notch, or other modification,
domain wall chirality can be determined, and leveraged for potential
356 C. C. Faulkner
applications in a memory device.51 A notch poses a different pinning
potential to a transverse domain wall depending on the wall chirality.20
I.e. at a fixed field, walls of a given chirality may propagate through a
notch, and wires of the opposite chirality may be pinned. This chirality
filtering effect applies to vortex domain wall types52 and the head-to-
head or tail-to-tail type found in thinner films.20 Furthermore a domain
wall trap can act as either a potential well or barrier depending on the
chirality of the domain wall.
14.10. Domain Wall Dynamics
The success of many DW devices depends on fast and reproducible DW
mediated magnetic switching, because the processing speed of these
devices depends on the domain wall velocity. The velocity of a DW
driven by an in-plane magnetic field in a nanowire is limited by the
Walker field, Hw. At low fields, below Hw, the DW velocity, v,
increases linearly with applied field, v = (γ0∆/α)H, where γ0 is the
1
D = 125 nm
D = 57 nm
s
D = 45 nm
M /M
0
t
t
-1
-80 -40 0 40 80
H (Oe)
x
Fig. 14.7. Time averaged locally measured M-H loops at measurement position E, for
trap depths from 45 nm–125 nm.
Domain Walls for Logic and Data Storage Applications 357
gyromagnetic ratio multiplied by the permeability of a vacuum µ 0, ∆
indicates the wall width, and α the Gilbert damping parameter. This is
the viscous regime. Above this critical field threshold the domain wall
propagation mode changes, and velocity does not increase linearly
with field. Below Hw the domain wall structure remains stable as it
propagates, above Hw this is not the case. The DW moves in an
oscillatory fashion, periodically stopping and starting in a turbulent
propagation mode, leading to an abrupt decrease in velocity. The domain
wall cants away from the direction of propagation at increasing fields,
and above Hw, the canting angle ф precesses continually. At the same
time the chirality of the domain wall also periodically changes. The wall
then alternates between an in-plane Néel-like structure, and an out of
plane Bloch like vortex structure, and propagates at reduced mobility in
an oscillatory fashion.
In tri-layer structures,18 the GMR effect was utilised to elucidate the
switching mechanism mediated by an individual domain wall in a NiFe
wire at low temperature (100 K), yielding an experimentally determined
damping parameter, α = 0.63. Using a high-resolution longitudinal
magneto-optical Kerr effect (MOKE) system, Atkinson et al.
successfully isolated an individual domain wall in different positions of a
2-corner planar NiFe – wire.27 Domain wall velocities were locally
probed in the structure, by localized single shot Kerr effect laser
measurements.53 Current pulses were applied by micro-strip line, and
wall velocities up to 1500 ms-1 were measured,27 giving a damping
constant, α = 0.053, comparable to extended thin film studies in
Permalloy.2
By time and spatially resolved MOKE,54–56 magnetic transmission
X-ray microscopy,57 or resistance measurements,58 Walker breakdown
has been successfully investigated in the lab, and an enhancement of
domain wall velocity in the presence of both current and external field
has been recorded.55 The oscillations of domain wall position and
chirality can be further elucidated micromagnetically (Fig. 14.8). It is
also suggested that field pulse rise time affects the domain wall
velocity.56 Much focus has been on in-plane magnetized systems, but
investigations into DW dynamics in perpendicular anisotropy systems
have also been performed.59
358 C. C. Faulkner
Fig. 14.8. Micromagnetic simulation of the temporal evolution of domain wall spin
structure and chirality above the Walker field. A 200 nm wide, 10 nm thick Permalloy
wire was simulated, with α = 0.001, and moving boundary conditions to negate any end
effects.58
14.11. DW Velocity Enhancements
14.11.1. Transverse field
The Walker field represents a fundamental limit to the DW propagation
field, and limits the operating frequency of any proposed DW devices. A
number of treatments/techniques have been proposed to increase this
threshold, enhancing the DW velocity. Reducing the thickness of the
nanowire can increase the field at which Walker breakdown occurs.60
The use of a field perpendicular to the direction of domain wall motion61
has also been shown to increase or decrease the domain wall velocity,
and the Walker breakdown field. In addition the direction of transverse
field affects velocity and Hw.
14.11.2. Roughness
Nakatani et al. proposed that edge roughness, on a ~9 nm length scale,
in elongated NiFe wires could enhance domain wall velocity, stabilizing
the domain wall at velocities above the theoretical Walker limit, by
acting as an additional channel for energy dissipation from a system, and
Domain Walls for Logic and Data Storage Applications 359
off-putting the onset of domain wall oscillation, and the formation
of vortex type wall configurations until higher Walker fields.60 This
seemingly counterintuitive claim was well supported by micromagnetic
simulations. For example, Miltat and co-workers demonstrated that
rough edged wires increase HW, showed that the defect sizes affects Hw,
and that the DW velocity above the Walker breakdown field is greater
for wires with edge fluctuations of 6–7 nm length scales.60
14.11.3. Current assisted
When spin polarized currents are injected into a nanowire, the domain
wall velocity can be enhanced or reduced, depending on the direction of
current flow,55,62 by up to 110 ms−1.
14.11.4. Ion irradiation
For perpendicular magnetic anisotropy systems, such as Co/Pt multi-
layers, an enhancement of domain wall velocity with Ga focused ion
beam fluence was demonstrated experimentally.63
14.11.5. Out of plane field
It has been demonstrated micromagnetically that a strongly out of plane
magnetized underlayer, such as the perpendicular magnetic anisotropy
(PMA) material FePt, to a soft magnetic strip leads to fast domain wall
motion for a wide range of driving fields, and can suppress the onset of
the Walker effect.64,65 The PMA underlayer helps prevent the formation
of vortices at wire edges, meaning oscillatory domain wall motion occurs
at higher fields.
14.12. Spin Torque
Berger proposed that electric current should apply a torque to a magnetic
domain wall in 1978.66 As a spin-polarized electron passes through a
magnetic domain wall, its spin direction follows the magnetization of its
local environment, and it therefore rotates from the spin ‘‘up’’ state to
spin ‘‘down’’. Since each quantum of angular momentum, ħ, must be
conserved, angular momentum is transferred to the spins of the domain
360 C. C. Faulkner
wall. This transfer of momentum creates a torque on the wall
magnetization and can apply a force to the wall similar to pressure
applied to a macroscopic object. In this manner, the position of a DW in
a nanowire may be manipulated by electric current.28,33,67,68 Current
controlled displacement of DWs, or current assisted switching of other
magnetic nanostructures, enables localized control of magnetic structures
without effecting neighboring element. No magnetic strip lines are
needed during device fabrication, and fast and reproducible switching
can result. For example Vernier and co-workers demonstrated that a
domain wall can be propagated along a 120 nm wide Permalloy wire,
and even around a corner, solely using current. Generally, above a
critical current, the DW moves in the direction of the flow of electrons.
The actual domain wall spin structure can also be modified by this
application of current pulses.69 Progress must be made in reproducibly
controlling the domain wall propagation at lower critical current
densities [of the order of ~108 A/cm2 for Permalloy], more suitable for
device applications.
Initial investigations on this topic were performed quasi-statically28,33
however for technologically relevant applications, walls must be
moved on short timescales. Amplification of domain wall motion with a
set of current pulses, using a resonance effect has recently been
demonstrated,70,71 making possible the motion of domain walls in a given
potential at lower currents, and at reduced power. The key is to tune the
current pulse to the precessional period of the domain wall. Tuning the
length and separation of current pulses enhanced this effect. In some of
these investigations oscillatory domain wall motion has been observed,
in which the wall moves against the flow of electrons.70
Current causes Joule heating, which can create or annihilate domain
walls, or change domain wall spin structure, e.g. chirality, or domain
wall type.69 Allenspach and co-workers recently demonstrated that
domain wall chirality can be better controlled by spin current, if NiFe
nanowire is fabricated on a Fe underlayer.72 Much work will be directed
on current controlled magnetic switching in the coming years, and
critical current densities will be reduced. New materials and geometries
will be investigated. For example, current induced domain wall motion
has already been demonstrated in promising future semi-conducting
Domain Walls for Logic and Data Storage Applications 361
spintronics material systems like alloys of GaAs and Mn (albeit at low
temperatures).73 A proposed technology that may yield even higher data
storage densities in the future is a storage paradigm using magnetic
vortices.74
14.13. Domain Wall Mediated Data Storage
Magnetic storage of binary information, requires an energy barrier
between two opposing directions of magnetization, which is large
enough to resist thermal effects.25 For applications such as sensors or
data storage37 sharp domain wall depinning transitions, simple switching
mechanisms, and statistically robust magnetic switching are needed.
Ring elements are an attractive system in which to study domain wall
motion, due to the device symmetry, and the ease of positioning a
domain wall in these structures by using an external field. The ring
geometry has been proposed for MRAM devices,30,75,76 analogous to
magnetic memory flux closure devices in the 1950’s. Changing the
ring aspect ratio,76 or adding a notch,77 gives further control over the
magnetization reversal.
14.14. DW Racetrack Memory
Parkin and co-workers have proposed a domain wall racetrack memory
device, where long tracks of 2 or 3-d wires on silicon, each holding up to
100 domain walls act as high performance, reliable, low cost, solid state
memory storage devices.68 A memory storage paradigm with a high
number of vertical storage strips gives a much greater data storage
density than traditional 2-d geometries. Domain walls would be spaced
by either notches, wire width modulations, or wire property modifications.
Domain walls in this paradigm are moved by short pulses of spin
polarized current, and the read in/out is performed by a nanoscale
spintronic device, such as a magnetic tunnel junction. Magnetic fields
cannot be used to transport domain walls in this memory device/shift
register. If a magnetic field were applied, eventually domain walls would
be forced into each other and annihilate. Therefore spin polarized current
is used to displace the domain walls. The direction of motion of the
362 C. C. Faulkner
domain walls is independent of the magnetic charge of the wall, so an
entire sequence of walls can be shifted around the loop, when a critical
current is exceeded. Parkin and workers have already reported on
displacing pairs of domain walls by spin polarized current alone.68 A
high concentration of domain walls in each wire would reduce the cost of
the system. Controlling position of domain walls by short pulses of
current is desirable for efficient operation of this device type, as has been
previously discussed.
Challenges for this architecture include reducing the critical current
density, increasing the density of domain walls in a given wire, and
fabrication challenges related to a 3-d geometry. The current densities
reported in the literature can result in domain wall spin structure change,
domain wall chirality change, and Joule heating up to the Curie
temperature of Permalloy – 850 K.
14.15. Domain Wall Logic
Recently a quantum cellular automata (QCA) Boolean logic paradigm
was proposed by Cowburn and Welland,78 based on propagating
information in chains of 110 nm diameter ferromagnetic dots. These
strings of single domain dots have been successfully combined into logic
gates at room temperature.79 The QCA paradigm evolved into a DW
logic scheme whereby information is transmitted and manipulated in
continuous planar ferromagnetic nanowires at room temperature, by
controlled manipulation of DWs.35,80 This DW logic scheme is analogous
to the bubble logic scheme of the 1970’s.81 Through very precise control
of wire geometries, domain wall mediated switching can be repeatedly
controlled. Wires treated here are typically 5 nm thick and of the order of
200 nm wide, so do not support vortex walls, or 360° domain walls,
which can make switching mechanisms less reproducible.
In the context of a possible logic schema, the reproducible transfer of
information, in the form of magnetization vector orientation in a wire,
over tens of microns mediated by field controlled domain wall
propagation has been demonstrated. Controlled information transfer by
domain wall propagation was demonstrated in elongated wires, and
around corners in the presence of an appropriate external field sequence
Domain Walls for Logic and Data Storage Applications 363
and direction. Single nanowire switching by domain wall can be
controlled not only by magnetic field, but by element geometry by
intentionally introducing artificial trapping centres in wires.
Some advantages are non volatile data retention, radiation hardness,
reduced energy consumption, a relatively simple planar geometry,
compatability with existing spintronics technologies such as MRAM and
the prospect of high integration densities. Disk drives require mechanical
movement, domain wall logic does not. Opposing magnetization
directions in a wire correspond to the logical “0” and “1” states. Logic
operations are performed using wire junctions linked by continuous
planar wires. The 4 principal operations that have been experimentally
demonstrated to date include logical NOT, logical AND, signal fan out
(1 DW is split into 2), and signal crossover (where DW’s cross a wire at
right angles to the direction of DW propagation.) The preferred
magnetization direction of the wire structures presented herein is parallel
to the wire edge.
The two magnetic orientations supported in elongated submicron
wires represent the Boolean 1 and 0. Note that if there is a 180° curve
in a wire, and the magnetization is continuous throughout, then the
absolute direction of magnetization is different before and after the
bend. Therefore Allwood and workers35 assigned logical “1” to wire
magnetization in the same direction as DW propagation, and logical “0”
to magnetization direction opposing DW motion. The extent of DW
propagation is limited by 90° and 180° corners in the nanowire networks.
In this manner the rotating field permits clear representation of different
logic states, and a direction of information propagation in the logic
system is well defined.
14.16. NOT
Signal inversion, and therefore NOT gate functionality was realized
at room temperature, in a 3-terminal continuous wire junction, by
field controlled domain wall motion (Fig. 14.9).35 This operation was
subsequently further investigated and optimized.82,83 Locally probing the
magnetization states in continuous wire loops with 3-terminal junctions,
in a 27 Hz rotating field, workers were able to reproducibly invert the
364 C. C. Faulkner
Fig. 14.9. Schematic of a typical NOT gate geometry. ‘A’ depicts a typical device
geometry and magnetic field directions. ‘B’-‘E’ indicate how the magnetization vector in
the devices respond to an external rotating magnetic field.35
logical state of the device output, over up to 100,000 repetitions. The
rotating field, which is in the plane of the samples, functions as the
power supply and clock. Initial NOT gates were connected in a closed
loop, such that the output is connected back round to the input.
The operation of the devices was subsequently verified and imaged
by magnetic force microscopy.83 Note that excellent geometrical control
is exerted over these patterned thin film structures by a customized FIB
fabrication process,38 and that small changes to the junction geometry
[on ~25 nm length scales] gives reproducible control over the device
operating field values, and the device performance tolerances.
It is possible to link a series of gates together to perform more
complex logic functions. A series of NOT gates were concatenated in a
loop to form a shift register, where complex bit patterns can be stored or
reproducibly transmitted about the circuit. Another useful aspect of this
Domain Walls for Logic and Data Storage Applications 365
device geometry is that the gate is symmetrical, so can be operated in
one of 2 directions. The concept of a domain wall NOT gate has
subsequently been investigated by other workers, and recently NOT gate
functionality was demonstrated by current driven domain wall motion in
a set of nanocontacts fabricated from Invar (a Nickel Steel alloy).84
14.17. AND/OR
Additional logic gates are required to compliment the NOT gate for
computational operations to be performed. A gate with 2 or more inputs is
required. For a logical AND/Or-like function, 2 magnetic nanowires unite
at a junction, with one output wire in a three terminal configuration.45,85
Domain walls interact at junctions. The switching field of the AND gate
output wire depends on whether there is 0, 1, or 2 DWs contained in the
input wires. The switching field at the output wire is markedly higher
when there is one domain wall propagating into the junction, that when 2
domain walls propagate into the junction. The operation of any potential
domain wall device strongly depends on the narrow contact geometry of
the wire junction. Varying wire widths, and wire angles at terminal
junctions affords control over the switching fields and field operating
ranges of any multi-terminal domain wall devices.86 Furthermore 3-
terminal devices of this nature have recently been demonstrated in multi-
layered NiFe/Cu films suitable for GMR-type characterization and
current driven output wire switching has been demonstrated.85
14.18. Fanout/Cloning
It is important to split signals in logic architectures, such as a multi-
terminal domain wall circuit. This process where domain walls are
duplicated at 3-terminal junctions has been dubbed ‘cloning’.87 Consider
a planar nanowire DW conduit that splits into two arms in a tuning fork-
type geometry, as shown schematically in Fig. 14.10. If a DW propagates
into the junction, and the junction nucleation field is high, then the DW
expands across the junction, bubble-like, and injects a DW into both of
the device outputs.80,87 Effectively 1 domain wall is cloned into 2.
By integrating these tuning fork junctions with other gates, it was
demonstrated that this cloning operation was reproducibly effected over
366 C. C. Faulkner
Fig. 14.10. Schematic of the domain wall cloning process.
13,000 field cycles, and that two of these cloning junctions can
be connected in series. The device performance or figure of merit
is well controlled by subtle modifications to the Y shaped device
geometry. For example if the wires are wide the junction nucleation
field is very low, meaning high field cannot be used. If the junction
wires are too narrow, a DW cannot be successfully propagated through
them, due to intrinsic sidewall pinning effects, and the onset of
superparamagnetism.
This concept has been extended by concatenating cloning gates in
series, as shown in Fig. 14.11. The device shown was tested in an anti-
clock-wise rotating field applied in the plane of the samples. A domain
wall is injected from the NOT gate output wire (loop top of Fig. 14.11)
left into cloning junction, C1. From here one domain wall propagates
through 2 corners and domain wall interconnect to C2 where it is
divided. The output wires from C2 are connected to the input wires of
junctions C3 and C4. Therefore 2 walls propagate to gates C3 and
C4, where they are split again. Local MOKE measurements taken
at positions 1-5 verified this domain wall cloning process. One
domain wall in a single nanowire is cloned into 4, giving 4 times digital
amplification.
14.19. Crossover
The signal cross-over element is a 4-terminal device where in the
appropriate field sequence, a domain wall can cross an orthogonal wire.
I.e. 2 nanowires cross over each other in the same plane. See Fig. 14.12,
where the main logic gates, and their analogous counterparts in CMOS
technology are displayed.
Domain Walls for Logic and Data Storage Applications 367
Fig. 14.11. ‘A’ is an electron image of 4 domain wall cloning junctions linked together,
labelled C1-C4. Plots ‘b’-‘f’ show magnetization switching data for MOKE measurement
positions labelled 1-5, and the concomitant magnetic field trace is shown in ‘g’.87
368 C. C. Faulkner
Fig. 14.12. Domain wall logic elements.80
14.20. Data Input
Any practical circuit must be able to receive data from external sources,
and data must also be read out from the system. For the DW logic
architecture, writing of information may be realized by current carrying
wires, or possibly heating,88 but layering of devices leads to further
Domain Walls for Logic and Data Storage Applications 369
fabrication complexity. However a simple planar field writable
method was recently described and experimentally verified,87 using low
coercivity areas of magnetic material, similar to a domain wall injection
pad. DWs can be reproducibly and selectively nucleated into a structure
without compromising the performance of other logic gates.
14.21. DW Diode
In the DW logic paradigm, the direction of information flow is defined
by the handedness of the rotating magnetic field and the orientation of
the corners. It is necessary to have a diode device for selectively filtering
information in one direction only, so direction of logic signals can
be controlled. (For example to prevent the unwanted meeting and
annihilation of 2 domain walls). A domain wall diode was demonstrated
experimentally in triangle like protrusions at the edge of a Permalloy
nanowire.82,90 This diode or ratchet effect was demonstrated for a
number of these protrusions concatenated in series,13,91 and has also been
documented in a simplified asymmetric DW trap geometry.92 In these
structures, in a geometry dependent field range (for fixed temperature),
below the wire nucleation field, a domain wall will propagate through the
feature in one direction only.
Since the feature is asymmetric, Fig. 14.13, the DW is presented with
a different energy landscape, depending on the direction of propagation.
Therefore for a fixed field a DW may propagate through a patterned
triangular feature, for one field parallel to the wire long axis, but not the
other. By varying the device geometry, the field bias switching effect can
be tuned.90,92
Current driven DW motion has been documented in these devices.91
By magnetic force microscopy measurements (MFM), current driven
DW magnetization reversal was evidenced in one direction, but
nucleation of more domains walls, and a higher critical current density
was measured when current was applied in the opposing direction. Since
it is known that due to stray fields from the tip, MFM is an invasive
measurement technique, these devices were also characterized by the
giant magnetoresistance effect.13
370 C. C. Faulkner
Fig. 14.13. ‘A’ shows a geometry where DWs were propagated in different directions,
through a central trap. ‘B’ shows a magnified trap electron scan. ‘C’ elucidates a diode
geometry in more detail.92
14.22. Outlook
Domain wall mediated switching of planar nanowires for device
applications has been discussed. Domain wall traps have been shown to
control domain wall spin structure, position, propagation or depinning
field, and direction of propagation in planar nanowires. Further
investigations into domain wall devices of increasing complexity can be
expected, leading to incorporation in commercially viable devices and
future scientific breakthroughs.
References
1. G. E. Moore, Electronics 38 (1965).
2. S. Konishi, S. Yamada, T. Kusuda, IEEE Trans. Magn. 7, 722 (1971).
Domain Walls for Logic and Data Storage Applications 371
3. C. E. Patton, Z. Frait, C. H. Wilts, J. Appl. Phys. 46, 5002 (1975).
4. K. J. Sixtus, L. Tonks, Phys. Rev. 37, 930 (1931).
5. K. S. Novoselov, A. K. Geim, S. V. Dubonos, E. W. Hill, I. V. Grigorieva, Nat. 426,
812 (2003).
6. X. B. Zhu, P. Grütter, V. Metlushko, B. Ilic, Phys. Rev. B 66, 024423 (2002).
7. T. Shinjo, T. Okuno, R. Hassdorf, K. Shigeto, T. Ono, Sci. 289, 930 (2000).
8. A. Wachowiak, J. Wiebe, M. Bode, O. Pietzsch, M. Morgenstern, R. Wiesendanger,
Sci. 298, 577 (2002).
9. E. H. Frei, S. Shtrikman, D. Treves, Phys. Rev. 106, 446 (1957).
10. W. Wernsdorfer, B. Doudin, D. Mailly, K. Hasselbach, A. Benoit, J. Meier, J.-Ph.
Ansermet, B. Barbara, Phys. Rev. Lett. 77, 1873 (1996).
11. R. P. Cowburn, D. A. Allwood, G. Xiong, M. D. Cooke, J. Appl. Phys. 91, 6949
(2002).
12. A. Himeno, T. Ono, S. Nasu, K. Shigeto, K. Mibu, T. Shinjo, J. Appl. Phys. 93, 8430
(2003).
13. M. Himeno, S. Kasai, T. Ono, J. Appl. Phys. 99, 08G304 (2006).
14. E. Saitoh, H. Miyajima, T. Yamaoka, G. Tatara, Nat. 432, 203 (2004).
15. R. D. McMichael, M. J. Donahue, IEEE Trans. Magn. 33, 4167 (1997).
16. G. S. D. Beach, M. Tsoi, J. L. Erskine, J. Magn. Magn. Mat. 320, 1272 (2008).
17. S. W. Yuan, H. N. Bertram, J. F. Smyth, S. Schultz, IEEE Trans. Magn. 28, 3171
(1992).
18. T. Ono, H. Miyajima, K. Shigeto, K. Mibu, N. Hosoito, T. Shinjo, Sci. 284, 468
(1999).
19. C. C. Faulkner, M. D. Cooke, D. A. Allwood, D. Petit, D. Atkinson, R. P. Cowburn,
J. Appl. Phys. 95, 6717 (2004).
20. D. Petit, A. V. Jausovec, D. Read, R. P. Cowburn, J. Appl. Phys. 103, 114307
(2008).
21. F. Fournal, Y. Chen, F. Carcenac, N. Essaidi, H. Launois, V. Kottler, C. Chappert,
IEEE Trans. Magn. 34, 1027 (1998).
22. K. Shigeto, T. Shinjo, T. Ono, Appl. Phys. Lett. 72, 1116 (1998).
23. Y. Yokoyama, Y. Suzuki, S. Yusaa, K. Ando, K. Shigeto, T. Shinjo, P. Gogol,
J. Miltat, A. Thiaville, T. Ono, T. Kawagoe, J. Appl. Phys. 87, 5618 (2000).
24. W. Y. Lee, C. C. Yao, A. Hirohata, Y. B. Xu, H. T. Leung, S. M. Gardiner,
S. McPhail, B. C. Choi, D. G. Hasko, J. A. C. Bland, J. Appl. Phys. 87, 3032 (2000).
25. P. Lendecke, René Eiselt, Guido Meier, Ulrich Merkt, J. Appl. Phys. 103, 073909
(2008).
26. K. Shigeto, T. Okuno, T. Shinjo, Y. Suzuki, T. Ono, J. Appl. Phys. 88, 6636 (2000).
27. D. Atkinson, D. A. Allwood, G. Xiong, M. D. Cooke, C. C. Faulkner, R. P.
Cowburn, Nat. Mater. 2, 85 (2003).
28. N. Vernier, D. A. Allwood, D. Atkinson, M. D. Cooke, R. P. Cowburn, Europhys.
Lett. 65, 526 (2004).
372 C. C. Faulkner
29. C. W. Sandweg, N. Wiese, D. McGrouther, S. J. Hermsdoerfer, H. Schultheiss,
B. Leven, S. McVitie, B. Hillebrands, J. N. Chapman, J. Appl. Phys. 103, 093906
(2008).
30. M. Kläui, C. A. F. Vaz, J. A. C. Bland, L. J. Heyderman, Appl. Phys. Lett. 86,
032504 (2005).
31. E. Saitoh, H. Miyajima, T. Yamaoka, G. Tatara, Nat. 432, 203 (2004).
32. N. Vernier, D. A. Allwood, D. Atkinson, M. D. Cooke, R. P. Cowburn, Europhys.
Lett. 65, 526 (2004).
33. M. Kläui, C. A. F. Vaz, J. A. C. Bland, W. Wernsdorfer, G. Faini, E. Cambril, L. J.
Heyderman, Appl. Phys. Lett. 83, 105 (2003).
34. D. A. Allwood, N. Vernier, G. Xiong, M. D. Cooke, D. Atkinson, C. C. Faulkner,
R. P. Cowburn, Appl. Phys. Lett. 81, 4005 (2002).
35. D. A. Allwood, G. Xiong, M. D. Cooke, C. C. Faulkner, D. Atkinson, N. Vernier,
R. P. Cowburn, Sci. 296, 2003 (2002).
36. C. C. Faulkner, PhD Thesis, University of Durham, 2005.
37. D. McMichael, J. Eicke, M. J. Donahue, D. G. Porter, J. Appl. Phys. 87, 7058
(2000).
38. G. Xiong, D. A. Allwood, M. D. Cooke, R. P. Cowburn, Appl. Phys. Lett. 80, 1195
(2002).
39. M. Herrmann, S. McVitie, J. N. Chapman, J. Appl. Phys. 87, 2994 (2000).
40. F. Cayssol, D. Ravelosona, C. Chappert, J. Ferré, J. P. Jamet, Phys. Rev. Lett. 92,
107202 (2004).
41. J. Gadbois, J.-G. Zhu, IEEE Trans. Magn. 31, 3802 (1995).
42. M. Kläui, C. A. F. Vaz, J. A. C. Bland, W. Wernsdorfer, G. Faini, E. Cambril,
J. Appl. Phys. 93, 7885 (2003).
43. J. Grollier, P. Boulenc, V. Cros, A. Hamzić, A. Vaurès, A. Fert, G. Faini, Appl.
Phys. Lett. 83, 509 (2003).
44. L. Gunther, B. Barbara, Phys. Rev. B 49, 3926 (1994).
45. C. C. Faulkner, D. A. Allwood, M. D. Cooke, D. Atkinson, R. P. Cowburn, IEEE
Trans. Magn. 39, 2860 (2003).
46. F. Cayassol, D. Ravelosona, J. Wunderlich, C. Chappert, J. P. Jamet, J. Ferré,
J. Magn. Magn. Mat. 240, 30 (2002).
47. D. Petit, A. V. Jausovec, H. T. Zeng, E. Lewis, L. O’Brien, D. Read, R. P. Cowburn,
Appl. Phys. Lett. 93, 163108 (2008).
48. A. W. Holleitner, H. Knotz, R. C. Myers, A. C. Gossard, D. D. Awschalom, Appl.
Phys. Lett. 85, 5622 (2004).
49. M. T. Bryan, D. Atkinson, D. A. Allwood, Appl. Phys. Lett. 88, 032505 (2006).
50. M. Hayashi, L. Thomas, C. Rettner, R. Moriya, X. Jiang, S. S. P. Parkin, Phys. Rev.
Lett. 97, 207205 (2006).
51. L. K. Bogart, D. S. Eastwood, D. Atkinson, J. Appl. Phys. 104, 033904 (2008).
52. K. J. O’Shea, S. McVitie, J. N. Chapman, J. M. R. Weaver, Appl. Phys. Lett. 93,
202505 (2008).
Domain Walls for Logic and Data Storage Applications 373
53. D. A. Allwood, G. Xiong, M. D. Cooke, R. P. Cowburn, J. Phys. D. 36, 2175 (2003).
54. G. S. D. Beach, C. Nistor, C. Knutson, M. Tsoi, J. L. Erskine, Nat. Mater. 4, 741
(2005).
55. G. S. D. Beach, C. Knutson, C. Nistor, M. Tsoi, J. L. Erskine, Phys. Rev. Lett. 97,
057203 (2006).
56. K. Weerts, P. Neutens, L. Lagae, G. Borghs, J. Appl. Phys. 103, 094307 (2008).
57. G. Meier, M. Bolte, R. Eiselt, B. Krüger, D.-H. Kim, P. Fischer, Phys. Rev. Lett. 98,
187292 (2007).
58. M. Hayashi, L. Thomas, C. Rettner, R. Moriya, S. S. P. Parkin, Nat. Phys. 3, 21
(2007).
59. C. Burrowes, D. Ravelosona, C. Chappert, S. Mangin, E. E. Fullerton, J. A. Katine,
B. D. Terris, Appl. Phys. Lett. 93, 172513 (2008).
60. Y. Nakatani, A. Thiaville, J. Miltat, Nat. Mater. 2, 521 (2003).
61. M. T. Bryan, T. Schrefl, Del. Atkinson, D. A. Allwood, J. Appl. Phys. 103, 073906
(2008).
62. M. Hayashi, L. Thomas, Y. B. Bazaliy, C. Rettner, R. Moriya, X. Jiang, S. S. P.
Parkin, Phys. Rev. Lett. 96, 197207 (2006).
63. J. P. Jamet, J. Ferré, P. Meyer, J. Gierak, C. Vieu, F. Rousseaux, C. Chappert,
V. Mathet, IEEE Trans. Magn. 37, 2120 (2001).
64. J.-Y. Lee, K.-S. Lee, S.-K. Kim, Appl. Phys. Lett. 91, 122513 (2007).
65. A. Kunz, S. C. Reiff, Appl. Phys. Lett. 93, 082503 (2008).
66. L. Berger, J. Appl. Phys. 49, 2156 (1978).
67. A. Vanhaverbeke, A. Bischof, R. Allenspach, Phys. Rev. Lett. 101, 107202 (2008).
68. S. S. P. Parkin, M. Hayashi, L. Thomas, Sci. 320, 190 (2008).
69. M. Kläui, P.-O. Jubert, R. Allenspach, A. Bischof, J. A. C. Bland, G. Faini,
U. Rüdiger, C. A. F. Vaz, L. Vila, C. Vouille, Phys. Rev. Lett. 95, 026601 (2005).
70. L. Thomas, M. Hayashi, X. Jiang, R. Moriya, C. Rettner, S. Parkin, Nat. 443, 197
(2006).
71. L. Thomas, M. Hayashi, X. Jiang, R. Moriya, C. Rettner, S. Parkin, Sci. 315, 1553
(2007).
72. A. Vanhaverbeke, A. Bischof, R. Allenspach, Phys. Rev. Lett. 101, 107202 (2008).
73. M. Yamanouchi, D. Chiba, F. Matsukura, H. Ohno, Nat. 428, 539 (2004).
74. S. Kasai, Y. Nakatani, K. Kobayashi, H. Kohno, T. Ono, Phys. Rev. Lett. 97, 107204
(2006).
75. J. G. Zhu, Y. F. Zheng, G. A. Prinz, J. Appl. Phys. 87, 6668 (2000).
76. F. J. Castano, C. A. Ross, A. Eilez, J. Phys. D 36, 2031 (2003).
77. M. Kläui, J. Phys.: Cond. Mat. 20, 313001 (2008).
78. R. P. Cowburn, M. E. Welland, Sci. 287, 1466 (2000).
79. A. Imre, G. Csaba, L. Ji, A. Orlov, G. H. Bernstein, W. Porod, Sci. 331, 205 (2006).
80. D. A. Allwood, G. Xiong, C. C. Faulkner, D. Atkinson, D. Petit, R. P. Cowburn, Sci.
309, 1688 (2005).
81. A. Hubert, R. Schäfer, Magnetic Domains (Springer, Berlin, 1998).
374 C. C. Faulkner
82. D. A. Allwood, G. Xiong, M. D. Cooke, C. C. Faulkner, D. Atkinson, R. P.
Cowburn, J. Appl. Phys. 95, 8264 (2004).
83. X. Zhu, D. A. Allwood, G. Xiong, R. P. Cowburn, P. Grütter, Appl. Phys. Lett. 87,
062503 (2005).
84. P. Xu, K. Xia, C. Z. Gu, L. Tang, H. F. Yang, J. J. Li, Nat. Nanotech. 3, 97 (2008).
85. C. Nam, Y. Jang, K. S. Lee, B. K. Cho, Nanotechnol. 19, 015703 (2008).
86. D. A. Allwood, G. Xiong, M. D. Cooke, C. C. Faulkner, D. Atkinson, R. P.
Cowburn, J. Appl. Phys. 95, 8264 (2004).
87. D. A. Allwood, G. Xiong, R. P. Cowburn, J. Appl. Phys. 101, 024308 (2007).
88. D. A. Allwood, G. Xiong, R. P. Cowburn, J. Appl. Phys. 100, 123908 (2006).
89. D. Atkinson, R. P. Cowburn, Appl. Phys. Lett. 85, 1386 (2004).
90. M. T. Bryan, T. Schrefl, D. A. Allwood, Appl. Phys. Lett. 91, 142502 (2007).
91. M. Himeno, S. Kasai, T. Ono, Appl. Phys. Lett. 87, 243108 (2005).
92. C. C. Faulkner, D. A. Allwood, R. P. Cowburn, J. Appl. Phys. 103, 073914 (2008).
AUTHOR INDEX
Aziz A., 19 Nabiyouni G., 139
Nasirpouri F., 1
Bakonyi I., 89 Nogaret A., 1
Baraduc C., 173
Bending S., 19 Oscarsson S., 315
Chshiev M., 173 Péter L., 89
Crampin S., 19
Ruck B. J., 193
Ebels U., 173
Shimojo M., 45
Faulkner C. C., 343 Strömberg M., 315
Svedlindh P., 315
Gunnarsson K., 315
Takeguchi M., 45
Ichiyanagi Y., 63
Yates K., 223
Kamali S., 267, 297
Mikailzade F., 121
375
This page is intentionally left blank
SUBJECT INDEX
Absorption, 267, 268 Buffering agent, 93
Adatoms, 100, 105 Butler-Volmer equation, 91
Adlayer, 98, 110
Adsorption, 91 Calcinations, 65
Aluminium oxide, 11 Carbon, 46
template, 11, 107 amorphous, 49
Anisotropic magnetoresistance, 154, 344 nanostructures, 46
Anisotropy Cathode, 93
in-plane, 8 Cell Selective Magnetic Nanoparticle,
magneto-crystalline, 2 81
perpendicular, 8, 97 Centroid Shift, 271
Annealing, 63 Charge, 193
Anomalous codeposition, 95 Chemiluminescence, 316
Antiferromagnetic, 63, 72, 123, 162, 166 Cluster, 226
Amino group, 63 Cobalt, 46, 96
Amino-MNP, 83 Nanowires, 8
Amino-silane coupling, 63 Codeposition, 94, 101
Aqueous, 63 Cole-Cole fitting, 323, 327
Solutions, 63, 96 Colloidal particles, 110
Atomic moment, 267 Collimation, 298
Avogadro, 64 Complexing agents, 93
Conversion Electron Mössbauer
Bioassay, 315 Spectroscopy, 276
Biofunctionalisation, 318 Coupling, 14
Biomolecules, 317 antiferromagnetic, 162, 166, 238
Bionanomagnetism, 315 ferromagnetic, 159, 162, 166,
Biotechnology, 315 227, 237
Bioscience, 316 spin-charge, 195
Bloch wall, 7 spin-orbit, 14
Bound magnetic polaron, 236, 237 Crystal, 67
Brilliance, 298 bulk, 66
Brownian motion, 316 lattice constants, 66
Brownian relaxation, 317 Crystallinity, 55, 57
377
378 Subject Index
Crystallite, 93 transverse, 345
Crystallization, 91 velocity, 358
Curie temperature, 23, 216 vortex, 345
Couprous/coupric oxides, 96 Dopant, 230
Current density, 92 Doping, 245, 246
limiting, 92 Doppler effect, 269
Doppler shift, 271
D0-ferromagnetism, 225
Damascene process, 89 Electric Monopole Interaction, 269, 270
Data storage, 20, 343 Electric Quadrupole Interaction, 269
media, 22 Electrochemistry, 89, 90
non-volatile, 20 Fabrication, 91
Delivery system, 64 Electrochemical reaction, 91
Density functional theory, 200 Electrocrystallization, 95
Density of states, 4, 157 Electrode potential, 90
Partial (PDOS), 209 Standard, 90
Dielectric, 123 Electrode reaction, 92, 142
Diffusion, 92 Electrodeposition, 8, 89, 139
Dilute magnetic dielectrics, 224 alloys, 144
Dilute magnetic oxides, 223 metals, 144
Dilute magnetic semiconductors, 194, magnetic elements, 94
223 magnetic alloys, 94
Dimensionality, 7 template, 107
DiMethylSulphOxide, 320 Electrolyte, 90
Dipolar energy, 2 Electron beam induced deposition, 45
DiThioThreitol, 319 Electron energy loss spectroscopy, 48
DNA, 315 Electron hologram, 51
Domain, 30 Electron holography, 50
domain nucleation, 31 Electronics, 1
pinning sites, 32 Electronic structure, 23
single, 11, 64, 316 semi-classical, 23
structure, 31 surface, 7
superlattice’ artificial width, 32 Electroplating, 89
Domain Wall, 9, 19, 20, 343 Electroreduction, 95
artificial, 19, 20, 24 Emission, 269
Bloch wall, 7 Endocytosis, 84
intrinsic pattern, 19 Endothermic peak, 78
logic, 362 Equilibrium potential, 90
motion, 9, 30 Exchange coupling, 179
natural, 21 conservative, 179
Néel wall, 7 interlayer, 179
resistivity, 22, 34 length, 344
Subject Index 379
Exchange Spring Magnets, 287, 303 Grain coarsening, 102
Exothermic peaks, 79 Grazing incidence Mössbauer
Extended X-ray absorption spectroscopy, 301
fine-structure, 66 Grazing incidence reflectivity, 301
Fabrication, 45, 59, 89 Hard disc, 121, 193, 194
focused ion beam, 19, 48 Half crystal position, 105
three dimensional, 59 Hall effect, 19, 23
Faraday’s constant, 143 anomalous, 234
Faraday’s law, 145 ballistic, 349
Fermi level, 5, 157, 200 extraordinary, 24, 26
Fermi-Dirac function, 181 resistivity, 26
Ferrimagnetic, 72, 318 Hall probe, 344
Ferrite Nanoparticles, 73 Heisenberg exchange, 3
Ferroelectric, 124 Heterostructures, 129, 132, 133
Ferromagnetic, 63, 318 Highly oriented pyrolitic graphite
Ferromagnetism, 72 (HOPG), 106
intrinsic, 223 Human Serum Albumin, 334
multi-domain, 22 Hund’s rules, 213
FePt, 47 Hysteresis loop, 5
thin ferromagnetic layers, 22 Hydrogen evolution, 93
Field cooled (FC), 67, 72 Hydrodynamic, 95
Fluorescence, 316 Hyperfine interaction, 267, 300, 301
Fluorescent material, 80
Focused ion beam, 19 Immobalization, 319
dose, 19, 20, 21, 24-32, 36, 41 ImmunoGlobulin, 317, 334
irradiation, 19, 21 Impurity, 225
masked irradiation, 21 extrinsic, 225
Folate receptor, 84 Interaction
Fourier transform, 69 Coulomb, 208
Fourier transform infrared dipole, 269
Spectroscopy, 65, 80 exchange, 2
Functionalization, 63, 80 spin-orbit, 1, 13
inverse spinel structure, 74
Galvanostatic, 91, 148 Inhibition intensity, 99
Giant magnetoresistance, 13, 20, 121, Interconnect, 89
140, 155, 194, 277, 349 copper, 89
GMR, 13, 20, 102, 121 Ion beam milling, 353
devices, 20, 122 Ionic liquid, 93
oscillatory, 158 Iron, 46, 96
superparamagnetism, 164 carbide, 49, 50
Glassy solid, 65 nanostructures, 48
380 Subject Index
nanowires, 8 magnetic circular dichromism, 234
oxide, 50 Magnetic dipole Interaction, 269
Irradiation, 19 Magnetic force microscopy, 25, 32, 33,
reactive ion etching, 24 317, 328, 344
Isomer Shift, 270 Magnetic nanoparticles, 63, 315
characterization, 64
Kinetics, 91 encapsulated, 63
iron-oxide, 316
Landau, 123 monodispersive, 64
Landau-Lifschitz Gilbert (LLG) single domain, 315
equation, 183 synthesis, 65
Landau-Lifschitz-Gilbert-Slonczewski Magnetic random access memory
(LLGS) equation, 183 (MRAM), 20, 122, 174, 310, 351
Length scales, 1, 2 Magnetic relaxation, 70
Critical, 3 magnetic resonance imaging, 316
Lifshitz, 123 Magnetic semiconductors, 123
Ligand, 46 Magnetic transmission x-ray
Liquid crystal, 112 microscopy, 357
Lithography, 174 Magnetic tunnel junctions, 20, 122,
electron beam, 345 173, 251, 310
Local density approximation, 200 Magnetic viscosity, 70
Logic processing devices, 343 Magnetization
curve, 6
Magnetic, 19 coherent rotation, 8
device, 173 curling, 8
interface, 22 dynamics, 183
metals, 89 local, 173
microstructure, 23 mesoscopic scale, 6
moments, 6 macroscopic scale, 7
nanoparticles, 63, 112 reversal, 5, 6
nanostructures, 45, 89 Magnetization reversal,
ordering, 3 nanowire, 8
recording media, 8, 46, 122 thin film, 7
switching, 19, 20, 45 Nanodot, 11
Magnetic anisotropy, 19, 21 Ultrathin films, 97
energy, 281 Magnetoelectric, 121, 123
interfacial, 28 Composite, 129
local, 20 linear, 124
perpendicular, 19, 21, 26 Magnetoresistance, 121, 153
out-of-plane, 21 angular dependence, 19, 21, 22
Magnetic beads, 315, 318 extrinsic, 23, 44
Magnetic bubbles, 8 longitudinal, 154
Subject Index 381
Lorentz, 23 Nanowires, 10
ordinary, 154 Ni-Cu/Cu, 164
spin-orbit coupling, 23 Multi-segment, 89
transverse, 154 nanowires, 89
tunnelling, 20, 193, 194, 252
Magnetorestrictive, 129 Nanobeam, 57
micro-biosensor, 317 Nanocluster, 66, 228
Magnetostriction, 344 Nanocrystalline, 89
Magnetotransport, 102 Nanocrystalline magnetic
Medicine, 315 deposits, 99
Mesoscopic, 6 Nanodots, 7
Metal, 89 Nanofabrication, 46
Metal chloride, 63 three dimensional, 46
Metal hydroxide, 65 Nanolithography, 348
Metal oxide, 65 Nanomagnetism, 2
Metastable phase, 235 Nanomagnet, 353
Metastable precursor alloys, 101 Nanoparticles, 7
Metastability, 101 Nanorod, 48, 50
Microarrays, 316 Self-standing, 50
micro-Hall effect, 317 Nanoscale, 1, 21
Micromagnetic, 346 science and technology, 1
Microparticles, 316 Nanotechnology, 63, 89
Mg Ferrite Nanoparticles, 78 Nanowires, 7, 8, 349
MOKE, 349 multilayered, 8
Mono-domain, 344 multi-segment, 89
Mono-layer, 7, 66 Nernst equation, 90, 143
monolayer Néel wall, 7
nanocluster, 66 Néel relaxation, 316
Moore’s law, 345 Néel temperature, 67, 211
Morphology, 65 Nickel, 96
Mössbauer spectroscopy, 75, 99, 269 nanowires, 8
Conversion, 276 Ni ferrite, 74
Transmission, 276 bulk crystal, 74
Mott, 12 Nickel hydroxide, 66
Multiferroic, 121, 126 crystal, 66
Multilayers, 22, 89, 102, 139 Ni-Zn Ferrite, 73
Co/Cu, 162 Nuclear magnetic resonance, 316
Co-Cu/Cu, 164 Nuclear probe technique, 267
Co-Ni-Cu/Cu, 164, 166 Nuclear Resonance Scattering, 297
Co-Ni/Cu, 164
Co/Pt, 22 Oligonucleotide, 315, 319, 321, 323,
Ferromagnetic, 22 324, 326
382 Subject Index
Organic, 46 Radioactive, 269
precursor, 46 Radio-frequency, 316
metallic, 41 Raman spectroscopy, 232
Oxygenation, 231 Rare earth, 123, 193
Oxoanion, 96 Rare earth Nitrides, 193
preparation, 196
Particle, 63 electronic structure, 199
magnetic, 112 magnetic, 211
metal Hydroxide, 66 Read heads, 1, 45
metal Oxide, 66 Recrystallization, 102
Paramagnetic, 63 Redox, 91
Pauli paramagnet, 4 Reflection, 66
Permeability, 6 X-ray, 66
Perovskite, 127 Reflectivity, 206
Phase, 71 Relaxation, 318
Phase segregation, 101 Remanence, 8
Photon, 300 Resistivity, 153
Physics Nobel Prize, 20, 121 in metals, 153
Piezoelectric, 129 Rhenium, 55
Plating, 95 nanorod, 55
Fe-Co-Ni, 95 Rhodamine, 80
suspension, 112 RKKY, 179, 236
Pluralistic ferrite particles, 73
Roughness, 93
Polarization, 123, 218
Rutherford backscattering
electric, 123, 127
spectroscopy, 199
X-ray, 298
Polarization curve, 91, 92
Scanning electron microscope, 48
Polarizer, 189
Schottky barrier, 152
Planar, 189
Self-assembly, 110
Perpendicular, 190
Semiconductor, 13, 20, 46
Potentiostatic, 91, 147
dilute magnetic, 196
Precursor, 71, 89
transistor, 13
Printed circuit boards, 89
substrate, 151
Protein, 315
single bath method, 103
Pseudo-torque, 176
Single domain, 8
Pulsation, 298
Silica, 63
Pulsed laser deposition, 231
cage, 63
Quantum dot, 46 Silanization, 80
Quantum computers, 194 Size confinement, 7
Quantum magnetic tunneling, 64, 70 Spectroscopy, 65
Quantum size effects, 64 Fourier transform infrared, 65
Subject Index 383
Spin, 2 Suspension plating, 112
scattering, 19, 20 Synchrotron radiation, 297
transport, 194 coherent, 299
Spin angular momentum, 175 incoherent, 299
Spin-charge coupling, 194, 195
Spin-dependent, 22, 42 Template, 10, 89
spin diffusion length, 343 AAO, 10
Spin field effect transistor, 194
polymeric, 10
Spin-logic devices, 194
method, 89
Spin polarisation, 248, 274
Tetrahedral, 74, 78
Spin-polarised, 19, 173, 223, 248, 275
Thermal properties, 64
Spin polarized tunneling, 278
Thermodynamic, 91
Spin transfer torque, 20, 173
Thin films, 7, 277
dynamics, 186
magnetic, 89
microscopic, 176
Tissue, 316
transverse, 182
Transistor, 14, 343
Spin torque effect, 13, 359
spin transistor action, 14
Spin valve, 13
spin field effect, 194
Spinel, 233
Transition metals, 1
Spinel structure, 72
Transition-metal chlorides, 64
Spintronics, 12, 121, 139, 193
Transition temperature, 68
devices, 19, 173, 194, 278, 343
Transmission electron microscope, 47,
SQUID magnetometer, 65, 325
Standard hydrogen electrode, 90 65, 228
Stoner-Wohlfarth, 346 high resolution, 229
Structure, 66 image, 103
two-dimensional, 66 Transport, 8
three-dimensional, 66 Tri-layer, 19
Surface roughness, 28 Pt/Co/Pt, 23
Surface reconstruction, 98 Tunnelling, 20, 193, 194
Superconductor tip, 249
Superlattice, 21, 139, 277 Ultrathin magnetic films, 96
Fe/Co, 278
Fe/Cr, 280 Vortex, 8
Fe/V, 284
structure, 148
Warren’s formula, 66
Superparamagnetism, 11, 164
Wave function, 2
Superparamagnetic, 63, 73, 104, 165,
orbital, 1
226, 234, 318, 344
Supporting electrolyte, 93
Susceptibility, 67 X-ray absorption near-edge
AC, 67 fine-structure, 66
384 Subject Index
X-ray diffraction, 66, 225 X-ray absorption fine-structure, 65
diffractogram, 101 X-ray absorption spectroscopy (XAS),
high angle, 149 209
low angle, 150 X-ray emission spectroscopy (XES),
pattern, 66 209
X-ray magnetic circular dichromism, X-ray sources, 298
234
X-ray powder diffraction, 65 Zero-field-cooled (ZFC), 67, 72
Sptitroncs man cuates mowta* mgaeDc moments
to ritegratetogicfunctions and ncr-wtalie irfonrabon
sl>age on lh* s * w ptittom As is cflai the casern
artfensed ( T U » s o m e , a t o x c s arc m a * thrcugft
tre sjWhess ol now! materials a r t high quaity new
physics rate-tits. Gant * agnetonoslance <rxl dilute
magnetic senwxmfrttofs are two sutfi eaamtfes.
H&frtvEf. ttie femanVatle ptfentUi of spiiromcs for
quantum amputation faces mapy ehalenges when
It can© to control*^ sfcnjnaneousiy sever* obits
encoded in magnefc moments.
A t e a t f W rtrodutfon looacepfc in funomagnw*
and sptrt'onics. the teit rewews recent techniques
and their achievements in the synthesis and
fatrcab^iclmsgnrtcninostrxft/es. Tbenwrods
presented here emctasae Cotton i p cf top down
approaches to nancrtofe nanowrcs and t h h ttms-
They intfude tocuswl >on beam irradolin electron
beam-iniced tfwn«tf vapour deposition, chemical.
and electrochemical methods The later pan of
me boofc reviews majnetoetectfic materials, the
giant magnetee&sUnce in magnetic supertattces.
Cynamics effects In spn \mt# torgjjs os:i Italors.
diWe magnetic ondei rare earth n i n e s with meter
resonaxe scattering, and Mossfcaaer spoftroscopy in
sprttoncs. Fnaly, the last part of this t a k ctaoGses
apptcaWrs to magne:>; sloragc sr>3 ho magictem NANOMAGNETISM
Nanoroagnetisna and Sgiolronics wilt be useful :o
SPINTTONICS
gradate students andresearchersand engineers in toca^+VaKn^Chafacterea(on
tne M C (■ nawsciffKe. ardA»lt3fcns
World Scientific
www. wo' I d s c i e n t i 11 c. com
??athc 9%3TlV , j738W