Vous êtes sur la page 1sur 429

HANDBOOK of MAGNETIC

MATERIALS

VOLUME 14

EDITED BY

K.H.J. BUSCHOW
Van der Waals-Zeeman Institute
University of Amsterdam
Amsterdam
The Netherlands

I
2002
ELSEVIER
Amsterdam - Boston - London - New York - Oxford - Paris
San Diego - San Francisco - Singapore - Sydney - Tokyo
ELSEVIER SCIENCE B.Y.

P.O. Box 211,1000 AE Amsterdam, The Netherlands

© 2002 Elsevier Science B.Y. All rights reserved.

This work is protected under copyright by Elsevier Science. and the following terms and conditions apply to its use:

Photocopying
Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the
Publisher and payment of a fee is required for all other photocopying. including multiple or systematic copying, copying for
advenising or promotional purposes. resale. and all forms of document delivery. Special rates are available for educational
institutions that wish to make photocopies for non-profit educational classroom use.

Permissions may be sought directly from Elsevier Science via their home page (http://www.e1sevier.com) by selecting 'Customer
Support' and then 'Permissions'. Alternatively you can send an e-mail to: permissions@elsevier.com. or fax to (+44) 1865
853333.

In the USA. users may clear permissions and make payments through the Copyright Clearance Center. Inc.. 222 Rosewood Drive.
Danvers, MA 01923, USA; phone: (+1) (978) 7508400. fax: (+1) (978) 7504744. and in the UK through the Copyright Licensing
Agency Rapid Clearance Service (CLARCS). 90 Tottenham Coun Road. London WIP OLP, UK; phone: (+44) 207 631 5555;
fax: (+44) 207 631 5500. Other countries may have a local reprographic rights agency for payments.

Derivative Works
Tables of contents may be reproduced for internal circulation. but permission of Elsevier Science is required for external resale
or distribution of such material.
Permission of the Publisher is required for all other derivative works. including compilations and translations.

Electronic Storage or Usage


Permission of the Publisher is required to store or use electronically any material contained in this work. including any chapter or
pan of a chapter.

Except as outlined above. no pan of this work may be reproduced. stored in a retrieval system or transmitted in any form or by
any means. electronic. mechanical. photocopying. recording or otherwise. without prior written permission of the Publisher.
Address permissions requests to: Elsevier Science Global Rights Department, at the fax and e-mail addresses noted above.

Notice
No responsibility is assumed by the Publisher for any injury and/or damage to persons or propeny as a mailer of products liability.
negligence or otherwise, or from any use or operation of any methods. products. instructions or ideas contained in the material
herein. Because of rapid advances in the medical sciences. in particular, independent verification of diagnoses and drug dosages
should be made.

First edition 2002


Library of Congress Cataloging in Publication Data
A catalog record from the Library of Congress has been applied for.
British Library Cataloguing in Publication Data
Handbook of magnetic materials
Vol. 14 edited by K.H.J. Buschow
1. Magnetic materials 2. Magnetism
I. Buschow, K. H. J.
538.4'4

ISBN: 044451144X(VoI.14)
ISBN: 0444 85313 8 (Series)

Working together to grow


libraries in developing countries
www.elsevier.com I www.bookaid.org I www.sabre.org

Transferred to digital printing 2008


PREFACE TO VOLUME 14

The Handbook series Magnetic Materials is a continuation of the Handbook series


Ferromagnetic Materials. When Peter Wohlfarth started the latter series, his original
aim was to combine new developments in magnetism with the achievements of earlier
compilations of monographs, producing a worthy successor to Bozorth's classical and
monumental book Ferromagnetism. This is the main reason that Ferromagnetic Materials
was initially chosen as title for the Handbook series, although the latter aimed at giving a
more complete cross-section of magnetism than Bozorth's book.
In the last few decades magnetism has seen an enormous expansion into a variety of
different areas of research, comprising the magnetism of several classes of novel materials
that share with truly ferromagnetic materials only the presence of magnetic moments. For
this reason the Editor and Publisher of this Handbook series have carefully reconsidered
the title of the Handbook series and changed it into Magnetic Materials. It is with much
pleasure that I can introduce to you now Volume 14 of this Handbook series
Magnetoelectronics is a novel and rapidly developing field, where new functionalities
are created by combining and utilizing simultaneously two degrees of freedom, the
charge and the spin of the carriers. This new field is frequently referred to as spin-
electronics or spintronics. It includes spin-utilizing devices that need neither a magnetic
field nor magnetic materials. In semiconductor devices, the spin of the carriers has only
played a very modest role so far because well established semiconductor devices are
non-magnetic and show only negligible effects of spin. However, interesting enhanced
spin-related phenomena were observed in magnetic semiconductors and diluted magnetic
semiconductors already many years ago. A review of diluted magnetic semiconductors
has been presented by Kossut and Dobrowolski in Volume 7. In a way compatible with
the present-day electronic materials, diluted magnetic semiconductors can be prepared
by introducing high concentrations of magnetic ions into the parent non-magnetic
semiconductors. Semiconductors based on III-V compound are widely used for high-
speed electronic devices as well as for optoelectronic devices. Moreover, heterostructures
based on the GaAs/(Al,Ga)As systems have proven to be a convenient testing ground for
novel physical concepts and devices. The introduction of magnetism into 111-V compounds
therefore, opens up the possibility of using a variety of magnetic and/or spin-dependent
phenomena. not present in the conventional non-magnetic parent compounds. Preparation
and properties of ferromagnetic III-V based semiconductors are reviewed in Chapter 1,
including heterostructures.
Nanoscale thin films and multilayers, nanocrystalline magnetic materials, granular
films, and amorphous alloys have attracted much attention in the last few decades, in
the field of basic research as well as in the broader field of materials science. Such

v
vi PREFACE TO VOLUME 14

heterogeneous materials display uncommon magnetic properties that virtually do not occur
in bulk materials. This is true, in particular with respect to surface (interface) magnetic
anisotropy and surface (interface) magnetostrictive strains and giant magnetoresistance.
The local atomic arrangement at the interface differs strongly from that in the bulk.
The local symmetry is lowered, so that some interactions are changed or are missing
altogether. The interface atoms may be envisaged as forming a new phase and some
properties characteristic of this phase may become predominant for the entire system.
This becomes particularly evident in the case of interfacial magnetostriction which can
lead to a decrease (almost to zero) or to an increase (over the bulk value) of the resulting
magnetostriction of the nanoscale system. In Chapter 2 of the present Volume, the
magnetoelasticity of heterogeneous materials is treated in much details. Generally, the
dimensions of a magnetostrictive material change when the material is subjected to a
change in magnetic field. Hence, magnetostrictive materials can be applied in transducers,
which directly convert electrical energy into mechanical energy. They are useful in the
manufacture of sensors, actuators, controllers, force and displacement as well as other
electro-acoustic devices. For these applications, transducer materials in the form of thin
films are of special interest because cost-effective mass production is possible, compatible
to microsystem processing technologies. In addition, magnetostrictive thin films are
particularly promising as microactuator elements like cantilevers or membranes, since
they combine high-energy output, high-frequency and remote-control operation. Due to
this potential, interest in such giant magnetostrictive thin films has rapidly grown over
the past few years and results are reviewed in this Chapter 2. This chapter is a logical
extension of previous wok on magneto-elastic effects published in this handbook series
over the years. Bulk giant magnetostrictive materials based on rare-earth compounds
were reviewed by Clark (Volume I), quadrupolar interactions and magneto-elastic effects
in rare-earth intermetallics were treated by Morin and Schmitt (Volume 5) and thermal
expansion anomalies and spontaneous magnetostriction of these compounds were reviewed
by Andreev (Volume 8).
There are various forms of the interplay of magnetism and superconductivity, which
can be divided into competition and coexistence phenomena. For instance, a strong
competition is found in high- T e cuprates. In these materials, depending on the doping rate,
either Neel-type antiferromagnetism or superconductivity may occur, both based on the
copper d-electrons. A coexistence of localized magnetic moments (e.g. from 4f-elements)
with superconductivity is known to occur in systems where the concentration of these
moments is sufficiently small or where they are antiferromagnetically ordered and only
weakly coupled to the conduction electrons. A review on the interplay of magnetism
and superconductivity in various types of intermetallic compounds has been presented
by Fischer in Volume 5 of the Handbook. An extensive review on the normal state
magnetic properties of cuprate high-temperature superconductors and related materials
has been presented by Johnston in Volume 10. A striking feature distinguishing the
superconducting RT2B2C compounds from other superconductors is the following: For
certain combinations of the Rand T elements superconductivity and antiferromagnetic
order have been found to coexist and more importantly, the values of the magnetic ordering
temperature TN are comparable in magnitude with the values of the superconducting
transition temperatures T e . This means that the magnetic energy is comparable with the
PREFACE TO VOLUME 14 vii

superconducting condensation energy. Therefore the investigation of these compounds


is expected to result in new insights into the interplay between superconductivity and
magnetism. The high values of T~ demand that in the quaternary borocarbides, different
from the situation in high- T c cuprates and the classical magnetic superconductors, the
exchange coupling between the rare-earth magnetic moments is the dominant magnetic
interaction rather than magnetostatic interaction. Obviously the exchange coupling is
mediated by the conduction electrons, Consequently also the interaction between the
magnetic moments and the conduction electrons must be relatively strong in the quaternary
borocarbides. A comprehensive review on the current status of research of the quaternary
borocarbide superconductors, starting from their discovery, is presented in Chapter 3 of
this Volume. For the reasons mentioned, the magnetic and as well as the superconducting
properties of this interesting class of materials is discussed together.
During the years, intermetallic gadolinium compounds have adopted a special position
in the study of 4f electron magnetism. The reason for this is the fact that the gadolinium
moment consists only of a pure spin moment, orbital contributions to the moment being
absent. As a consequence, gadolinium compounds have been regarded as ideal test
benches for studying exchange interactions, free from complications due to crystal field
effects. Large spontaneous magnetoelastic effects are frequently associated with rare earth
compounds in which crystal fields are operative and in which the rare earth moments also
have an orbital contribution. Surprisingly, equally large spontaneous magnetoelastic effects
have been observed in some Gd compounds, showing that the contribution of the exchange
interaction to spontaneous magnetoelastic effects can become of equal importance as the
crystal field contribution. In several of the Gd compounds so-called magnetostructural
transitions occur where giant spontaneous as well as forced magnetoelastic effects can
be correlated with structural transitions. In Chapter 4 a review is given of experimental
studies of spontaneous magnetoelastic effects in Gd compounds, offering the possibility
to estimate the relative contribution of exchange striction to the total spontaneous
magnetoelastic effects in materials where also crystal field related contributions are
present.
Volume 14 of the Handbook on the Properties of Magnetic Materials, as the preceding
volumes, has a dual purpose. As a textbook it is intended to be of assistance to those
who wish to be introduced to a given topic in the field of magnetism without the need
to read the vast amount of literature published. As a work of reference it is intended for
scientists active in magnetism research. To this dual purpose, Volume 14 of the Handbook
is composed of topical review articles written by leading authorities. In each of these
articles an extensive description is given in graphical as well as in tabular form, much
emphasis being placed on the discussion of the experimental material in the framework of
physics, chemistry and material science.
The task to provide the readership with novel trends and achievements in magnetism
would have been extremely difficult without the professionalism of the North Holland
Physics Division of Elsevier Science B.Y., and I wish to thank Paul Penman for his great
help and expertise.

K.H.J. Buschow
Van der Waals-Zeeman Institute
University of Amsterdam
CONTENTS

Preface to Volume 14 v

Contents ix

Contents of Volumes 1-13 xi

List of Contributors xv

1. III-V Ferromagnetic Semiconductors


F. MATSUKURA, H. OHNO and T. DIETL 1
2. Magnetoelasticity in Nanoscale Heterogeneous Magnetic Materials
N.H. DUC and P.E. BROMMER 89
3. Magnetic and Superconducting Properties of Rare Earth Borocarbides of the
Type RNi 2 B 2 C
K.-H. MULLER, G. FUCHS, S.-L. DRECHSLER and V.N. NAROZHNYI . . . 199
4. Spontaneous Magnetoelastic Effects in Gadolinium Compounds
A. LINDBAUM and M. ROTTER 307

Author Index 363

Subject Index 405

Materials Index 413

ix
CONTENTS OF VOLUMES 1-13

Volume 1
1. Iron, Cobalt and Nickel, by E. P. Wohlfarth 1
2. Dilute Transition Metal Alloys: Spin Glasses, by J. A. Mydosh and G.J. Nieuwenhuys 71
3. Rare Earth Metals and Alloys, by S. Legvold 183
4. Rare Earth Compounds, by K. H. J. Buscfum 297
5. Actinide Elements and Compounds, by W. Trzebiatowski 415
6. Amorphous Ferromagnets. by F E. Luborsky 451
7. Magnetostrictive Rare Earth-Fe^ Compounds, by A. E. Clark 531

Volume 2
1. Ferromagnetic Insulators: Garnets, by M. A. Giileo I
2. Soft Magnetic Metallic Materials. byG.Y.Chin and J. H. Wernick 55
3. Ferrites for Non-Microwave Applications, by P. I. Slick 189
4. Microwave Ferrites, by J. Nicolas 243
5. Crystalline Films for Bubbles, by A.M. Eschenfelder 297
6. Amorphous Films for Bubbles, by A. H. Eschenfelder 345
7. Recording Materials, by G. Bate 381
8. Ferromagnetic Liquids, by S. W. Charles and J. Popplewell . 509

Volume 3
1. Magnetism and Magnetic Materials: Historical Developments and Present Role in Industry
and Technology, by U, Enz I
2. Permanent Magnets; Theory, by H. Zijlstra 37
3. The Structure and Properties of Alnico Permanent Magnet Alloys, by R. A. McCurrie 107
4. Oxide Spinels, by S. Krupitka and P. Novak 189
5. Fundamental Properties of Hexagonal Ferrites with Magnetoplumbite Structure.
byH.Kojima 305
6. Properties of Ferroxplana-Type Hexagonal Ferrites, by M. Sugimota 393
7. Hard Ferrites and Plastoferrites. by H. Stablein 441
8. Sulphospinels. by R. P. van Stapelt 603
9. Transport Properties of Ferromagnets. by I. A. Campbell and A. Fert 747

Volume 4
1. Permanent Magnet Materials Based on 3d-rich Ternary Compounds, by K. H. J. Buschaw . . . . 1
2. Rare Earth-Cobalt Permanent Magnets, by A J. Srrnai . 131
3. Ferromagnetic Transition Metal Intermeiallic Compounds, by J. G. Booth 211

XI
xii CONTENTS O F VOLUMES 1-13

4. Intermetallic Compounds of Actmides. by V. Sechmsky and L Havela 309


5. Magneto-Optical Properties of Alloys and Intermetallic Compounds, by K. H. J. Buschow . . . 493

Volume 5
1. Quadrupolar Interactions and Magneto-Elastic Effects in Rare-Earth Intermetallic Compounds.
by P Morin and D. Schmitt I
2. Magneto-Optical Spectroscopy of f-Electron Systems, by W. Reim and J. Schitenes 133
3. INVAR: Moment-Volume Instabilities in Transition Metals and Alloys, by E.F. Wassermtm . . . 237
4. Strongly Enhanced Itinerant Intermelallics and Alloys, by P. E. Bmmmerand J.J. M. Franse . . . 323
5. First-Order Magnetic Processes, by C Asti 397
6. Magnetic Superconductors, by 0. Fischer 465

Volume 6
1. Magnetic Properties of Ternary Rare-Earth TransUion-Metal Compounds, by H.-S. Li and
J.M.D.Coey 1
2. Magnetic Properties of Ternary Intermetallic Rare-Earth Compounds, by A. Szylula 85
3. Compounds of Transition Elements with Nonmetals. by O. Bechnan and L. Lundgren 181
4. Magnetic Amorphous Alloys, by P. Hansen 289
5. Magnetism and Ouasicrystals. byR. C. O'Hundley. R.A. Dunlap andM. E McHenry 453
6. Magnetism of Hydrides, by G. Wiesinger and G. Hilscher 511

Volume 7
1. Magnetism in Ultrathin Transition Metal Films, by II. Gradmann 1
2. Energy Band Theory of Metallic Magnetism in the Elements, by V.L Moruzzi and
P.M.Man-us 97
3. Density Functional Theory of the Ground State Magnetic Properties of Rare Earths and Actinides.
by M. S. S. BrtH>ks and B. Johansson 139
4. Diluted Magnetic Semiconductors, by J. Kossut and W. Dobrmvolski 231
5. Magnetic Properties of Binary Rare-Earth 3d-Transition-Metal Intermetallic Compounds,
by J.J. M. Franse and R.J'. Radv.an.ski 307
6. Neutron Scattering on Heavy Fermion and Valence Fluctuation 4f-systems.
by M. Ldoewenhaupl and K. H. Fischer 503

Volume 8
1. Magnetism in Artificial Metallic Superlattices of Rare Earth Metals, by J.J. Rhyne and
R. W. En* in 1
2. Thermal Expansion Anomalies and Spontaneous Magnetostriction in Rare-Earth Intermetallics
with Cobalt and Iron, by A. V. Andreev 59
3. Progress in Spinel Ferrite Research, hy V. A. M. Brabers 189
4. Anisotropy in Iron-Based Soft Magnetic Materials, by M. Soinski and A. J. Moses 325
5. Magnetic Properties of Rare Earth-Cui Compounds, hy Nguyen Hoang Luang and
J.J.M. Franse 415

Volume 9
1. Heavy Fermions and Related Compounds, hy G.J. Nieuwenliuys 1
2. Magnetic Materials Studied by Muon Spin Rotation Spectroscopy, hy A. Schenck and F.N. Gygax 57
CONTENTS OF VOL UMES 1-43 xiii

3. Interstitially Modified Intermetallics of Rare Earth and 3d Elements, by H. Fujii and H. Sun . . . .¾).¾
4. Field Induced Phase Transitions in Ferrimagnels. by A. K. Zvezdin 4()5
5. Photon Beam Studies of Magnetic Materials, by S. W. Lmesey 545

Volume 10
1. Normal-State Magnetic Properties of Single-Layer Cuprate High-Temperature Superconductors and
Related Materials, by D.C. Johnston 1
2. Magnetism of Compounds of Rare Earths with Non-Magnetic Metals, by D. Gignoux and D. Schmitt 239
3. Nanocrystalline Soft Magnetic Alloys, by G. Herzer 415
4. Magnetism and Processing of Permanent Magnet Materials, by K.H.J. Buschow 463

Volume 11
1. Magnetism of Ternary Intermetallic Compounds of Uranium, by V. Sechovsky and L. Have la . . 1
2. Magnetic Recording Hard Disk Thin Film Media. byJ.C. Ladder 291
3. Magnetism of Permanent Magnet Materials and Related Compounds as Studied by NMR.
by Cz. Kapu.ua. PC. Riedi and G.J. Tomka 407
4. Crystal Field Effects in Intermetallic Compounds Studied by Inelastic Neutron Scattering, by O. Maze 493

Volume 12
1. Giant Magnetoresistance in Magnetic Multilayers, by A. Barthe'lemy, A. Fertand F. Petrvff . . . I
2. NMR of Thin Magneiic Films and Superlattices. by P.C. Riedi. T. Thomson and G.J. Tomka . . . 97
3. Formation of 3d-Moinents and Spin Fluctuations in Some Rare-Earth-Cobalt Compounds.
by N.H. Due and P.E. Brtmmer 259
4. Magnetocaloric Effect in the Vicinity of Phase Transitions, by A.M. Ttshin 395

Volume 13
1. Interlayer Exchange Coupling in Layered Magnetic Structures, by D.E. Biirgler.
P. Grtinberg, S.O. Demokritpx-andM.T. Johnson 1
2. Density Functional Theory Applied to 4f and 5f Elements and Metallic Compounds, by M. Richter 87
3. Magneto-Optical Kerr Spectra, by P.M. Oppeneer 229
4. Geometrical Frustration, by A. P. Ramirez 423

Material chroniony prawem autorskim


LIST OF CONTRIBUTORS

P.E. Brommer, Van der Waals-Zeeman Instituut, Universiteit van Amsterdam, Valckenier-
straat 65, 1018 XE Amsterdam, The Netherlands
T. Dietl, Research Institute of Electrical Communication, Tohoku University, Sendai, Japan
S.-L. Drechsler, Leibniz-Institut filr Festkorper- und Werkstofforschung Dresden, P.O. Box
270116, D-O1171, Dresden, Germany
N.H. Due. Cryogenic Laboratory, Faculty of Physics. National University of Hanoi,
334 Nguyen Trai, Thanh xuan, Hanoi, Vietnam
G. Fuchs. Leibniz-Institut fur Festkorper- und Werkstofforschung Dresden, P.O. Box
270116, D-O1171, Dresden, Germany
A. Lindbaum, Institut fur Festkorperphysik, Technische Universitat Wien, Wiedner
Hauptstrasse 8-10/138, A-I040 Wien, Austria
F. Matsukura, Laboratory for Electronic Intelligent Systems, Research Institute of
Electrical Communication, Tohoku University, Sendai, Japan
K.-H. Muller, Leibniz-Institut fur Festkorper- und Werkstofforschung Dresden, P.O. Box
270116, D-O1171, Dresden, Germany
V.N. Narozhnyi, Leibniz-Institut fUr Festkorper- und Werkstofforschung Dresden, P.O. Box
270116, D-O1171, Dresden, Germany
H. Ohno, Laboratory for Electronic Intelligent Systems, Research Institute of Electrical
Communication, Tohoku University, Sendai, Japan
M. Rotter, Institut fur Angewandte Physik, Technische Universitlit Dresden, D-01062,
Dresden, Germany
chapter 1

III-V FERROMAGNETIC SEMICONDUCTORS

F. MATSUKURA
Laboratory for Electronic Intelligent Systems
Research Institute of Electrical Communication, Tohoku University
Sendai
Japan

H.OHNO
Laboratory for Electronic Intelligent Systems
Research Institute of Electrical Communication, Tohoku University
Sendai
Japan

T. DIETL

Institute of Physics, Polish Academy of Sciences


Warszawa
Poland

Handbook of Magnetic Materials, Vol. 14


Edited by K.H.J. Buschow
© 2002 Elsevier Science B.V. All rights reserved
CONTENTS

I. Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2. Preparation of III-V magnetic semiconductors by molecular beam epitaxy 6
2.1. (Ga,Mn)As . 6
2.2. (In,Mn)As grown on GaAs 10
2.3. (In,Mn)As grown on (AI,Ga)Sb II
2.4. Other Ill-V magnetic semiconductors . 12
3. Lattice properties . 15
3.1. Lattice constants . . . . . . . . . . . 15
3.2. Local lattice configuration (EXAFS) 17
3.3. Atomic-scale observations by scanning tunneling microscopy (STM) 18
4. Spin and charge states of Mn in III-V magnetic semiconductors. 18
4.1. Electron spin resonance (ESR) . 19
4.2. Optical spectroscopy . . . . . . . . 19
4.3. X-ray magnetic circular dichroism (XMCD) 20
4.4. Photoemission 21
5. Magnetic properties . 21
5.1. Magnetization. 21
5.2. Magnetic anisotropy 24
5.3. Magnetic domains . 26
5.4. Cantilever magnetometry 26
6. Magnetotransport properties 27
6.1. (Ga,Mn)As 27
6.2. (In,Mn)As . 34
6.3. Infrared and far infrared optical conductivity . 37
6.4. Cyclotron resonance 37
7. Magneto-optical properties 38
7.1. Faraday rotation .. 39
7.2. Magnetic circular dichroism (MCD) 40
8. Origin of ferromagnetism . 41
8.1. First-principles studies . 41
8.2. Parameterized Hamiltonians . . 45
8.3. Hole states and hole mediated exchange interactions 47
8.4. Mean-field Zener model and its application to (Ga,Mn)As 50
8.5. Comparison of theoretical and experimental results 55

2
III-V FERROMAGNETIC SEMICONDUCTORS 3

8.6. Limitations and refinements of the mean-field Zener model .. 59


9. Heterostructures . . . . . . . . . . . . . . 61
9.1. Basic properties of heterostructures 61
9.2. Spin-dependent scattering. interlayer coupling. and tunnel magnetoresistance in trilayer structures 64
9.3. Resonant tunneling diodes (RTDs) .. . . . . . . . . . . . . . . . 67
9.4. Spin-injection in ferromagnetic semiconductor heterostructures 70
9.5. Photo-induced ferromagnetism in (In.Mn)AslGaSb . . . . . 72
9.6. Electric-field control of ferromagnetism in gated structures . 73
9.7. Ferromagnetic imprinting .. . . . . . . . . . 75
10. Ferromagnetic semiconductors at room temperature 75
10.1. Theoretical suggestions 75
10.2. Cautionary remarks. 77
10.3. Experimental results 77
11. Summary and outlook . 79
Acknowledgements . 79
References 79
1. Introduction

Modem information technology utilizes the charge degree of freedom of electrons in semi-
conductors to process the information and the spin degree of freedom in magnetic mate-
rials to store the information. Magnetoelectronics is a new fastly developing field, where
the two degrees of freedom, the charge and the spin of the carriers, are utilized simul-
taneously to create new functionalities. In more general terms, this new field is referred
to as spin-electronics or spintronics to include those spin-utilizing devices that need nei-
ther the magnetic field nor magnetic materials. The magnetoresistance (MR) sensors made
of multilayers containing metal ferromagnets, showing giant magnetoresistance (GMR)
or tunneling magnetoresistance (TMR), are today's best known successful magnetoeloec-
tronics devices based on the interplay between the two degrees of freedom (Prinz 1998;
De Boeck and Borghs 1999; Wolf 2000; Ball 2000; Ziese and Thornton 200 I; Wolf et al.
2001).
In semiconductor devices, the spin of carriers has played a minor role so far because
the most-well established semiconductor devices based on Si and GaAs are non-magnetic
and show only negligible effects of spin. On the other hand, from the physical points
of view, the enhanced spin-related phenomena due to the coexistence of the magnetism
and semiconductor properties have been recognized in magnetic semiconductors and
diluted magnetic semiconductors (DMS) (or semimagnetic semiconductors; SMSC) since
the 60s. The family of magnetic semiconductors encompasses europium and chromium
chalcogenides (rock-salt type: EuSe, EuO and spinels: CdCr2S4, CdCr2Se4), which show
ferromagnetic order at low temperatures with the Curie temperature Tc :::; 100 K. They
have been extensively studied, because of their peculiar properties resulting from the
exchange interaction between itinerant electrons and localized magnetic spins (s-f and s-d
exchange interactions) (Kasuya and Yanase 1968; Methfessel and Mattis 1968; Mauger and
Gotard 1986). Owing to these interactions, magnetic semiconductors exhibit a rich variety
of striking optical and transport phenomena, which are strongly affected by the magnetic
field and the magnetic order, particularly near the metal-to-insulator transition (MIT).
However, difficulties in material preparation and in fabrication of heterostructures make
this family of compounds less attractive from the application point of view. Manganites
(perovskite: (La,Sr)Mn03 and related materials), which show colossal magnetoresistance
(CMR), are magnetic semiconductors, whose studies have been particularly active over the
recent years. Their ferromagnetic order, beginning at "" 350 K, originates from the double-
exchange interaction. Properties of manganites and their epitaxial heterostructures are
currently studied aggressively (Coey et al. 1999; Tokura and Tomioka 1999; Tokura 2000).
Their compatibility to the well-established electronic devices is an open issue because of
the differences in both crystal structure and constituting elements.

4
III-V FERROMAGNETIC SEMICONDUCTORS 5

DMS are based on non-magnetic semiconductors, and are obtained by alloying them
with a sizable amount (a few percents or more) of magnetic elements, such as Mn. The
studies of DMS and their heterostructures have offered a wide variety of materials and
structures, making it possible to explore further the effect of the exchange interaction in
semiconductors. Most of the work had been centered around II-VI based materials such as
(Cd,Mn)Te, (Zn,Co)S, (Hg,Fe)Se, where the valence of group II cations is identical to that
of most magnetic transition metals (Furdyna and Kossut 1988; Kossut and Dobrowolski
1993; Dietl 1994). Although this made them relatively easy to prepare, difficulties in
doping of II-VI-based DMS to either p- or n-type as well as relatively weak bonds made
these materials less attractive for applications. The magnetic properties of II-VI DMS
are dominated by the antiferromagnetic super-exchange interactions among the localized
spins, which result in paramagnetic, spinglass or antiferromagnetic behavior depending on
the concentration of the magnetic ions and temperature. Recent progress in doping of II-VI
materials is gradually changing this situation (Shibata et at. 1988; Baron et at. 1994), for
example, hole mediated ferromagnetism was found in p-type II-VI DMS with Tc < 10 K
(Haury et at. 1997; Ferrand et at. 2001; Hansen et a1. 200 1). Understanding of the carrier-
mediated ferromagnetism in semiconductors was put forward by a study of ferromagnetism
in IV-VI DMS such as (Pb.Sn.Mnj'Ie (Story et al. 1986). However, IV-VI DMS and their
heterostructures are again rather difficult to prepare.
An approach compatible with the present-day electronic materials is to make non-
magnetic semiconductors magnetic, and even ferromagnetic, by introducing a high
concentration of magnetic ions. III-V compound semiconductors are widely used for high-
speed electronic devices as well as for optoelectronic devices. Moreover, heterostructures
based on the GaAs/(Al,Ga)As systems have proven to be a convenient test bench for new
physics and device concepts. Introduction of magnetic III-V compounds opens, therefore,
up the possibility of using a variety of magnetic and/or spin-dependent phenomena, not
present in the conventional non-magnetic III-Vs, in the optical and electrical devices
already established (fig. 1). The proposal of III-V based magnetic semiconductors with
various sets of host materials and transition metals was put forward in 1970s (Galazka
1977), and some experimental studies were then initiated. At that time, however, III-V
materials with a sizable concentration of uniformly distributed magnetic elements were
not obtained due to the low solubility of transition metals in III-V semiconductors (Aliyev
et a1. 1980). The application of non-equilibrium methods to grow III-V-based DMS was
rewarded by successful molecular beam epitaxy (MBE) of uniform (In,Mn)As films
on GaAs substrates (Munekata et at. 1989). Subsequent discovery of the hole-induced
ferromagnetic order in p-type (In,Mn)As (Ohno et a1. 1992) encouraged researchers to
investigate GaAs-based system (De Boeck et at. 1996) and led to the successful growth of
ferromagnetic (Ga,Mn)As (Ohno et a1. 1996a). Currently, a number of groups is working on
the MBE growth of (Ga,Mn)As and related heterostructures to advance the understanding
of this new class of materials (Ohno et a1. 1996a; Ohno 1999; Hayashi et a1. 1997a;
Nishikawa et a1. 1997; Van Esch et a1. 1997; Sadowski et at. 1998; Kawakami et at. 2000;
Potashnik et a1. 200 1; Schott et a1. 200 1; Liu et a1. 200 1).
This chapter reviews the properties of ferromagnetic III-Vs, and is organized in the
following way. Section 2 describes the preparation of ferromagnetic III-Vs, and is followed
by the presentation of lattice properties in section 3. In section 4, spin and charge states of
6 F. MATSUKURA et aI.

Fig. I. Concept of spin-electronics (spintronics). In semiconductor spin-electronics spin properties as well as


electronic and optical properties are utilized at the same time.

magnetic ions in III-V DMS are discussed. The magnetic and magnetotransport properties
are presented in sections 5 and 6. respectively. Section 7 summarizes optical and magneto-
optical properties. Possible origin of ferromagnetism in III-V-based DMS is discussed in
section 8. Properties of heterostructures are presented in section 9 and recent developments
of room-temperature ferromagnetic semiconductors are introduced in section 10. Summary
and outlook is given in section II.

2. Preparation of 111-V magnetic semiconductors by molecular beam epitaxy

In order to observe magnetic cooperative phenomena in diluted magnetic systems. one


needs to introduce a sizable amount of magnetic elements (a few percents or more).
which is beyond their solubility limit in III-V semiconductors. Although non-equilibrium
epitaxial growth methods such as molecular beam epitaxy (MBE) could offer doping
in excess of the thermodynamic solubility limit. segregation of impurities during MBE
growth was an obstacle in obtaining high concentrations of magnetic elements (DeSimone
1982). It was demonstrated that low temperature MBE (LT-MBE. growth temperature
Ts < 300°C) can lead to successful epitaxy of (In.Mn)As with a few percents of Mn,
in which the solubility limit is overcome as well as the segregation phenomena. and the
formation of unwanted second phases is suppressed (Munekata et al. 1989). InAs was
chosen as a host compound because it was thought to be a suitable material for low
temperature growth due to its relatively small bond energy.

2.1. (Ga,Mn)As
Typical MBE growth of (Ga,Mn)As is carried out by using solid source MBE with
elemental sources Ga. Mn, In. Al and As, usually without intentional doping. Mn provides
both localized spins and holes due to its acceptor nature. Epitaxial films of (Ga.Mn)As
are grown on semi-insulating GaAs (00 I) substrates at a typical growth rate of 0.6-
0.8 JLmlhour under As-stabilized conditions. Normally. either a GaAs buffer layer or an
III-V FERROMAGNETIC SEMICONDUcrORS 7

(Al,Ga)As buffer layer is grown before epitaxy of (Ga,Mn)As. To control strain in the
film, strain-relaxed thick (In,Ga)As ('" 1 jlm) with the lattice constant Q o greater than the
subsequent (Ga,Mn)As layer can be employed. The Mn composition x in the Ga I-x Mn, As
films can be determined from measurements of Q o by x-ray diffraction (XRD), once the
dependence Qo(x)is calibrated by other means, such as electron probe micro-analysis
(EPMA) or secondary ion mass spectroscopy (SIMS).
The growth of (Ga,Mn)As can be initiated by simply commencing the Mn flow during
the LT-GaAs growth and keeping the substrate temperature Ts constant at 250°C. No
special precautions are needed at the start of (Ga,Mn)As growth. However, properties of
(Ga,Mn)As films do depend on the growth parameters such as As overpressure and Ts
(Matsukura et al. 1998a; Shimizu et al. 1999; Schott et al. 200 1). This may stem from the
influence of these parameters on the degree of compensation of the Mn acceptors by deep
donors, primarily As antisites which are known to be present with a high concentration in
LT-GaAs (Look 1991; Luysberg et al. 1998). However, as long as the established growth
procedure is followed, the properties of (Ga,Mn)As films are reproducible; for example,
for a given Mn composition x, the ferromagnetic transition temperature Tc can always be
maintained in the range of 2000x ± 10 K (Matsukura et al. 1998b).
Reflection high-energy electron diffraction (RHEED) patterns are used to monitor
the surface reconstruction during the growth. RHEED of GaAs [or (AI,Ga)As] buffer
layer grown at Ts '" 570°C shows the well-known (2 x 4) pattern, which changes to the
c(4x4) pattern when Ts is lowered to 48D-520°C, and remains c(4x4) below. Whereas
the RHEED pattern of GaAs layers grown at Ts = 250°C shows a (I x I) pattern (no
reconstruction) (fig. 2a), that of (Ga,Mn)As is (I x2) during and after the growth (fig. 2b).
The origin of this difference is not yet clear. When the Mn flux and/or Ts are too high, the
RHEED pattern indicates the appearance of a second phase on the surface, which is MnAs
in the hexagonal NiAs-structure (fig. 2d). When Ts is too low, spotty RHEED pattern
appears showing that the growth mode changes from the two-dimensional (2D) to three-
dimensional (3D), resulting in a polycrystalline material (fig. 2c). The maximum value
of x obtained so far is about 0.07 at 250°C, and increases up to '" 0.10 at Ts = 200°C
(Takamura et al. 200 1). Attempts to increase x even further have so far been unsuccessful
because of the surface segregation that occurs even at low growth temperatures. At a fixed
value of x = 0.035, epitaxial (Ga,Mn)As films can be grown at Ts varying from 160 to
320°C (Shen et al. 1997a).
Clear RHEED oscillations are observed at the initial growth stage (also without
Mn) under the conditions appropriate for (Ga,Mn)As epitaxy. Thus, the growth is two-
dimensional, and its rate can be determined from the oscillation period (fig. 3) (Shen et al.
1997a, 1997b). Monte Carlo simulations of RHEED oscillations have been carried out
for LT-GaAs employing the cubic solid-on-solid model (Yasuda and Ohno 1999). The
results show that the RHEED oscillations are related to an enhancement in migration of Ga
adatoms caused by a surfactant effect of the excess As overlayer. The surfactant effect of
As is demonstrated also by the surface visualization (Guo et al. 2000; Tazima et al. 2001a).
Scanning tunneling microscopy (STM) study of LT-GaAs shows that the step density on
the surface increases monotonically with the decrease of Ts, indicating the monotonous
decrease of Ga migration length. This result suggests that the origin of RHEED oscillation
may also relate to the decrease of the Schwoebel barrier height due to the formation of
8 F. MATSUKURA et a1.

Fig. 2. Reflection high energy electron diffraction (RHEED) patterns taken from [110] azimuth. (a)
Low-temperature grown GaAs at 250o e. (b) (Ga,Mn)As at 250 o e. (c) 170o e. and (d) 320 0 e (Shen et a1. I997a).

tshu er openingGaAs
T=600°C

t
~ GaAs
~ ~500C
W
I-
Z (Ga,Mn)As
T=250°C
s

2s
......
TIME-
Fig. 3. Temporal evolution of RHEED specular beam intensity (along [110] azimuth) for GaAs at 6OOo e. GaAs
at 250o e. and (Ga.Mn)As at 250 0 e from top to bottom (Shen et at. I997a).
III-V FERROMAGNETIC SEMICONDUCTORS 9

.... .
..... .....
.
Growth inhibited,
.
..... ..... formation of MnAs
300 ~\ ..... ..... •
\ .....
.....

--
p
\
\ Metallic (Ga,Mn)As
,-
.....
..... ~

. ...CIl
\
, /
I

200 ~ , /Insulating·
......... _ .... ", (Ga,Mn)As

Roughening

Polycrystal
100 "'--_......_-"""'--_......._ _... . .
0.00 0.02 0.04 0.06 0.08
Mn composition x in Ga 1_xMnxAs

Fig. 4. Schematic diagram of properties of {Ga,Mn)As films in relation to the growth parameters. Lines provide
a rough guide (Ohno 1998; Shen et aI. 1999).

small growth islands (Tazima et al. 200 Ib). The enhancement of the RHEED oscillation
of (Ga,Mn)As is observed, when the lattice-relaxed (In,Ga)As is used as a buffer layer
(Matsukura 1997) or (Ga,Mn)As contains a few percents of In (Katsumoto 1999). In these
cases, it is considered that In atoms act as additional surfactant.
A schematic phase diagram of MBE growth is depicted in fig. 4 (Ohno 1998; Shen et al.
1999). Recently it was shown that metallic (Ga,Mn)As with x = 0.1 can be obtained by
the use of a modified MBE growth technique at Ts = 150°C, migration-enhanced epitaxy
(MEE), where the beam fluxes of source materials are precisely controlled (Sadowski et al.
zoois.zoouo.
The surface morphology investigated by atomic-force microscopy (AFM) shows that
(Ga,Mn)As with 2D growth mode has a flat surface being comparable with a GaAs surface
[root-mean-square (RMS) of the roughness < 0.5 nm]. On the other hand, (Ga,Mn)As with
segregated RHEED pattern has a rather rough surface, characterized by about 10 times
larger RMS, which is probably due to small MnAs clusters (Yang et al. 2000).
The homogeneity of as-grown (Ga,Mn)As and the precipitation of MnAs clusters after
annealing at around 600°C are confirmed by transmission electron microscopy (TEM)
(De Boeck et al. 1996). The disappearance of the (Ga,Mn)As phase by annealing at
temperatures above 400°C is also confirmed by the disappearance of x-ray diffraction
peaks of (Ga,Mn)As (Shen et a1. 1997a). Low-temperature annealing at '" 300°C changes
lattice constant, magnetic, and electrical properties of (Ga,Mn)As, which is thought to
result from the evaporation of excess As atoms that form complexes with Mn acceptors
(Hayashi et a1. 200 I; Potashnik et al, 200 I). By additional doping of high concentration of
Sn, n-type (Ga,Mn)As can be also grown (Satoh et a1. 1997,2001).
10 F. MATSUKURA er aI.

It has been shown that digital alloys. multilayer structures consisting of GaAs and
less-than 1 monolayer (ML)-thick MnAs. which exhibit the ferromagnetic order at low
temperatures. can be grown by atomic-layer epitaxy (ALE) (Chen et al. 2000; Kawakami
et al, 2(00). The growth of (Ga.Mn)As with nominal Mn composition up to 0.04 by
metal-organic vapor-phase epitaxy (MOVPE) has also been carried out. MOVPE-grown
(Ga.Mn)As includes MnAs microclusters due to the high-growth temperature of 400-
600°C. which is necessary for the decomposing of precursor compounds (Hartmann
et a1. 2000; Heimbrodt et a1. 2001). MBE growth of (Ga.Mnjas on GaAs (41l)A and
Si (00l) substrates and (In.Ga.MnjAs on InP (001) substrates and the observation of their
ferromagnetism also have been carried out (Omiya et al. 200 1; Zhao et al. 2002; Ohya et a1.
200 I; Siupinski et al. 2002a).

2.2. (In.Mn)As grown on GaAs


The overall trend in the properties of (In.Mn)As grown by MBE can be summarized in
terms of x and Ts (Munekata et al. 1989. 1990. 1991; Ohno et a1. 1991). When (In.Mn)As
layers are grown directly on the GaAs substrate. either Il'- or p-type conduction is observed
depending on x and Ts. At Ts of 200°C. thick (In.Mn)As layers (> I JLm) grown on
GaAs (00l) substrates are n-type. The anomalous Hall effect. which is proportional to the
perpendicular component of the magnetization of the film. has not been observed in n-type
films. Donor-type defect formation in the InAs host lattice at growth temperature is most
probably responsible for the n-type conduction in such (In.Mn)As layers. Defect formation
may also be enhanced by the high density of mismatch dislocations in the (In.Mn)As
layers; the lattice mismatch in question is about 7%. When Ts < 200°C. polycrystalline
growth occurs.
At higher Ts of 400°C. the layers are p-type at low x « 0.03). MnAs ferromagnetic
clusters start to form above x = 0.03. and the films become eventually n-type at high
x values of 0.18. The hole concentration is identical to the Mn concentration up to
x = 0.004 (p = 7 X 10 19 cm'). at which the hole concentration peaks and then rolls off
with the increase of x. Thus. in the low x region of thick p-type layers grown at 300°C. Mn
behaves as a shallow acceptor. At x = 0.026 the hole concentration is 3 x 10 18 cm >'. No
epitaxy is observed at Ts > 400°C. The phase diagram showing the relation between film
properties and growth parameters is shown in fig. 5 (Ohno et al. 1991). Low-temperature
annealing of (In.Mn)As results in an increase of p and Te. similar to (Ga.Mnj.As (Endo
et al, 2001; Hashimoto et al. 2002).
The homogeneity of (In.Mn)As films grown at '" 200°C and the existence of MnAs
clusters in the samples grown at '" 300°C are confirmed by TEM observation. The
annealing at 450°C modifies the size of the clusters (Guha and Munekata 1993; Van Esch
et a1. 1995).
InAs quantum dots (QDs) fabricated by the self-organized growth mode (Stranki-
Krastanov mode) have attracted a great deal of interest. The interplay between dimensional
confinement and magnetism is certainly an interesting topic to pursue. The formation of
self-organized (In.Mnjas dots was demonstrated by growing (In.Mn)As on (00l), (211)B.
(311)B GaAs substrates by MBE at Ts = 350°C (Guo et al. 1998, 1999). In order to obtain
uniform structures. the growth temperature should be kept as high as possible. close to the
point of the phase separation. Photoluminescence from (In.MmAs QDs has been observed
(Sadowski et al. 200lc).
III-V FERROMAGNETIC SEMICONDUcrORS 11

( (

400 Growth Inhibited

p-lnMnAs p-lnMnAs+MnAs
6300- .
'?..-
CIl
t- x-O.03

200
n-lnMnAs (x s 0.24) .

Polycrystal
l\ I
100 ' - - - - - - - - - - - - , 'r--~
0.00 IJ 0.25
Mn composition x in In,."Mn"As

Fig. 5. Schematic diagram of properties of (In,Mn)As films grown directly on GaAs (001) substrate in relation
to the growth parameters (Ohno et al. 1991).

2.3. (In,Mn)As grown on (Al,Ga)Sb


Contrary to the results of thick layers directly grown on GaAs substrate, thin (In,Mn)As
layers « 30 nm) grown at 200°C pseudomorphically on thick (Al,Ga)Sb buffer layers
(200-500 nm grown on (001) GaAs substrates) are p-type (Ohno et al. 1992; Munekata
et al. 1992). A schematic phase diagram of thin (In,Mn)As layers on (Al,Ga)Sb buffer
layer is shown in fig. 6 (Munekata 1995). The p-type conduction results, most probably,
from the acceptor nature of Mn and low dislocation density in the pseudomorphic thin
layers. Although the lattice mismatch between the (In,Mn)As layers and the (AI,Ga)Sb
buffer layers is small, depending on the growth procedure different initial growth modes
have been observed (Shen et al. 1997c). Careful minimization of As flux is required to
obtain 2D growth at the initial stage of (In,Mn)As epitaxy (x = 0.02-0.04) on (Al,Ga)Sb,
otherwise 3D growth takes place, as witnessed by RHEED patterns. It is interesting to note
that despite streaky RHEED patterns that appear after 10-20 nm growth of (In,Mn)As,
the magnetic properties, probed by the anomalous Hall effect, show a correlation with
the initial growth mode. In particular, the 3D nucleation results in superparamagnetism,
whereas the 2D nucleation leads to ferromagnetism with square hysteresis (Shen et al.
1997c). The tensile strain associated with the lattice mismatch between (In,Mn)As and
(Al,Ga)Sb (0.6-1.3%) makes the easy axis to be perpendicular to the film plane (Munekata
et al. 1993). The increase of the thickness beyond the critical value (about 55 nm
for (In,Mn)As/AlSb) causes lattice relaxation, which affects magnetic anisotropy. The
thickness of the (In,Al)As spacer layer (AI composition r - 0.15) between (In,Mn)As layer
and AISb buffer also influences the magnetic properties of (In,Mn)As (Munekata et al.
1992).
12 F. MATSUKURA et al.

.
300 ,
x= x .
I

I
C

p-lnMnAs + MnAs
\
.. - - - - - - - -T=TC

p-lnMnAs on (AI,Ga)Sb I GaAs(001

200 ~

. .
0.0 0.1 0.2
Mn composition x in In 1•xMn xAs
Fig. 6. Schematic diagram of properties of (In,Mn)As films grown on GaAs substrate with thick (AI,Ga)Sb buffer
layer between the two in relation to the growth parameters (Munekata 1995).

2.4. Other III-V magnetic semiconductors


With the advances of the non-equilibrium crystal growth techniques, other III-V magnetic
semiconductors than (Ga,Mn)As and (In,Mn)As with different host semiconductors and
different transition metals have appeared and the investigation of properties of these new
materials are underway.
• (Ga,Fe)As: The growth of (Ga,Fe)As with Fe compositions up to ~ 0.1 was carried out
by LT-MBE at Ts = 260-350°C on GaAs substrate (Haneda et al. 2000a). The lattice
constant of (Ga,Fe)As becomes smaller with the increase of Fe composition, reflecting
the shorter bond length of Fe-As than that of Ga-As. The elevation of Ts leads to
the precipitation of Fe-related clusters. The extended x-ray absorption fine structure
(EXAFS) analysis revealed that Fe atoms in (Ga,Fe)As grown at low Ts substitute
into Ga sites in the zinc-blende structure. whereas that Fe atoms in high- Ts grown
samples form Fe clusters and/or Fe-As complexes (Soo et al. 200la). Van-Vleck type
paramagnetism occurs in (Ga.FefAs, as the temperature dependence of magnetization is
rather weak at low temperatures « 10 K). The conduction type is n-type and a photo-
induced MR effect is observed below 100 K. In GaAs-Fe composite structures (GaAs
including Fe clusters), photo-induced magnetization as well as MR effect is observed
even at room temperature (Haneda et al. 2000b, 200 I), The properties of (Ga.Fe.Mnj.As
with transition metal compositions up to 0.04 have been investigated. The results show
that (Ga,Fe,Mn)As becomes insulating with the increase of Fe content, and that there are
contributions of ferromagnetic and paramagnetic terms to the film magnetization. The
ratio of paramagnetic part to ferromagnetic part increases with the increase of Fe content
(Moriya et al. 2002).
III-V FERROMAGNETIC SEMICONDUcrORS 13

• (Ga,Cr)As: (Ga,Cr)As layers with Cr compositions up to 0.1 were grown by LT-MBE at


Ts = 250°C on GaAs substrates (Saito et al. 2001). The magnetization measurements
show that (Ga.CrjAs is superparamagnetic and the Curie-Weiss plot has a positive
Curie temperature, indicating that the dominant interactions between the Cr ions are
ferromagnetic. The value of the effective magnetic moment of the Cr ions is consistent
with that of the divalence state of Cr, suggesting that the Cr atoms act as acceptors.
Whereas there is a report of p-type conduction of (Ga,Cr)As (Okazawa et al. 1999,
2001), the determination of the conduction type by Hall measurements is rather difficult
due to a small Hall voltage, suggesting that the conduction is dominated by hopping.
Ferromagnetism below r - 45 K of (Ga,Cr)As with higher Cr composition (0.14) is also
observed. From photoemission spectroscopy, (Ga,Cr)As with Cr compositions larger
than 0.25 is metallic (Yamada et al. 2(01). Ferromagnetic properties of (Ga,Cr,Mn)As
grown by MBE are reported (Akinaga et al. 2000a). However, Tc is lower than that
of (Ga,Mn)As with the same Mn composition, which may be due to the compensation
effect.
• (Ga,Mn)N: bulk crystals with x up to 0.02 were grown by a resublimation method
(Gebicki et al. 2(00), and microcrystals with x up to 0.005 were obtained by an
ammonthermal method (Zajac et al. 200 Ia). Both compounds show extra Raman peaks
induced by disorder stemming from Mn incorporation. Magnetization measurements
reveal that these compounds are paramagnetic and from its temperature dependence
antiferromagnetic nearest-neighbor interaction coupling v- -2 K) is obtained (Zajac
et al. 200lb).
Structural properties of MBE-grown (Ga,Mn)N with x up to 0.09 on AIN buffer onto
Ah03 substrate grown at Ts = 650-750°C by MBE using RF-nitrogen plasma source
have been investigated (Kuwabara et al. 200 Ia, 200 Ib). Both cross-sectional TEM and
plane-view scanning electron microscopy (SEM) images show no visible second phase.
Extended x-ray absorption fine structure (EXAFS) analysis indicates that the Mn atoms
are incorporated in the Ga sites. Magnetization measurement revealed that there exist
contributions from both ferromagnetic (even at room temperature) and paramagnetic
phases. Kuwabara et al. suppose that the ferromagnetic part may originate from the
presence of ferromagnetic or ferrimagnetic second phases of Ga-Mn and/or Mn-N, since
the samples are highly resistive and thus no carrier-induced ferromagnetism is expected.
According to the Curie-Weiss analysis of the paramagnetic part, the paramagnetic
Curie-Weiss temperature 9p is negative in the very dilute Mn composition regime
(9p '" -8 K with S'" 2.5 for [Mn] = 7 x 10 19 cm- 3), and changes into positive when
the Mn composition is increased (9p '" 20 K with S '" 2.5 for [Mn] = 8 x 1020 cm >').
The EXAFS and near-edge x-ray absorption fine structure (NRXAFS) analyses on the
samples prepared by the same authors indicate that the most part of Mn substitutes
into the Ga sites, the valency of Mn is 2+, and that there is a possible formation of
Mn clusters (Soo et al. 2001b).
An n-type (Ga,Mn)N film with x = 0.07 on Al203 substrate was grown by MBE at
865°C using a nitrogen plasma source (Overberg et al. 2(01). The magnetization at 10 K
is nonlinear as a function of the magnetic field and small hysteresis are visible, indicating
that the film is ferromagnetic. Negative magnetoresistance and nonlinear dependence of
the Hall resistance are assigned to spin effects, though the negative magnetoresistance is
reminiscent of a weak localization effect.
14 F. MATSUKURA et at.

MBE-grown (Ga,Mn)N films with x = 0.06 and 0.09 were prepared using NH3 as
nitrogen source (Sonoda et al. 2002). The results of magnetization measurements are
similar to those of Kuwabara et al. However, Sonoda et al. claim that their (Ga,Mn)N
films show the ferromagnetic behavior even at room temperature. Indeed, the estimation
of Tc from temperature dependence of magnetization is 940 K, and no Mn-Ga and Mn-N
compounds with such high Tc have previously been found. (One should note that rather
high Tc ('" 750 K) has been observed in Mn-Ga alloys (Bither and Cloud 1965). Very
recently, the room temperature ferromagnetism in (Ga,Cr)N (Hashimoto et al. 2(02) and
(Ga,Mn)P:C (Theodoropoulou et al. 2(02) has also been reported.)
The magnetic properties of p-GaN implanted with high doses (3-5%) of Mn (annealed
at 700-1000°C after doses) have been investigated. The result shows that the sample is
ferromagnetic with Tc '" 250 K (Theodoropoulou et al. 200 1a).
(Ga,Mn)N film prepared by post growth Mn doping using solid state diffusion shows
ferromagnetic behavior at room temperature, which is confirmed by the observation of
an anomalous Hall effect (Reed et al. 2(01).
It seems that a considerable amount of work is needed to clarify the structural and
magnetic properties of (Ga,Mn)N.
• (Ga,Fe)N: GaN films doped with Fe, with concentrations up to '" 3 x 10 19 cm- 3
were grown by MBE at several Ts from 380 to 520°C directly on sapphire (0001)
substrates. Ferromagnetic behavior with Tc '" 100 K is observed only in the sample
grown at '" 400°C, in spite of quite a low concentration of Fe (Akinaga et al.
2000b). GaN:Fe films (Fe up to 6 X 102 1 cm >') grown by MBE at Ts = 500-8OO°C
show a superparamagnetic behavior (Kuwabara et al. 200 Ia, 200 1b) together with
superparamagnetic contributions of possible Ga-Fe and/or Fe-N inclusions. The EXAFS
analysis suggests that the decrease of Ts leads to a structural transition from wurtzite to
zinc-blende structure, and this transition may be related to the origin of ferromagnetism
in the GaN film with Fe (Ofuchi et al. 2oola). The emission channeling result on
annealed Fe implanted GaN (Fe concentration 10 17_10 18 cmr') shows that the majority
of Fe (80%) occupies substitutional Ga sites (Wahl et al. 200 I). p-GaN implanted with
a high dose of Fe (3-5%) shows ferromagnetic behavior, Tc '" 250 K (Theodoropoulou
et al. 200lb).
• (Ga,Mn)Sb and (Ga.CrjSb: The growth of bulk GaSb crystals heavily doped by Mn was
reported earlier, but whether an alloy between GaSb and Mn is formed remains unclear
(Aliyev et al. 1980; Adhikari and Basu 1984). GaSb films with a few percent of Mn or
Cr were also grown by MBE at Ts = 250-560°C (Abe et al. 2000,2001). The surface
morphology of MBE-grown samples observed by atomic force microscopy (AFM)
shows that these films contain clusters, which may be transition-metal-antimonide
compounds. The size of the clusters becomes smaller with the decrease of Ts.
For GaSb with Mn, magnetization measurements show a ferromagnetic behavior
even at room temperature, indicating the existence of ferromagnetic Mn-Sb clusters.
The larger coercive force for higher Ts may reflect the larger size of these clusters.
The increase of the magnetization at low-temperatures suggests the formation of
ferromagnetic zinc-blende (Ga,Mn)Sb. According to the saturation value of the
magnetization, only about 10% of the nominal Mn concentration contributes to the
formation of (Ga,Mn)Sb grown at Ts = 560°C, but 30% of Mn contributes to (Ga,Mn)Sb
m-v FERROMAGNETIC SEMICONDUCTORS 15

if Ts = 250°C. Magnetotransport measurements determine the properties of (Ga,Mn)Sb


without the influence of Mn-Sb clusters, and the data show a ferromagnetic behavior
with Tc ,.... 10 K and r - 30 K for Ts = 560°C and 250°C, respectively (Abe et a1. 2(00).
In the case of GaSb with Cr, both antiferromagnetic and ferromagnetic contributions
are detected. Due to an antiferromagnetic nature of Cr-Sb compounds, the magnitude of
the magnetization decreases with the increase of Cr content, whereas ferromagnetic hys-
teresis persist even at room temperature. The result of magnetotransport measurements
shows that (Ga,Cr)Sb grown at Ts = 550°C is antiferromagnetic at 1.5 K. (Ga,Cr)Sb
grown at 250°C is highly resistive at low temperatures « 100 K), which may be due
to the compensation of intrinsic holes in epitaxial GaSb films by the Cr ions (Abe et a1.
2001).
• Mn doped InSb: The properties ofInSb:Mn single crystals with [Mn] < 3.5 x 10 17 cm- 3
grown by the Czochralski method have been investigated extensively (Obukhov and
Pepic 1989; Obukhov 1993, 1996; Henriques et a1. 1999). Mn in InSb behaves as
a shallow acceptor (activation energy E a = 7 meV), and a metal-insulator transition
(MIT) occurs at low Mn concentrations (2x 10 17 cm- 3) due to a relatively large Bohr
radius of the bound holes. The interactions between the Mn spins are predominately
antiferromagnetic. A strong spin-dependent coupling between the Mn spins and holes
gives rise to large magnetoresistance effects, leading to the field-induced insulator-to-
metal transition in InSb:Mn at low temperatures (Obukhov 1996; Henriques et a1. 1999).
An anomalous Hall effect, whose coefficient is larger for lower Mn concentrations
(2 x 10 16 < [Mn] < 3 x 10 17 cm- 3) , has been observed (Obukhov and Pepic 1989).

3. Lattice properties

3.1. Lattice constants


X-ray diffraction (XRD) measurements show that (Ga,Mn)As has the zinc-blende structure
without detectable second phase. The results for the (004) reflection using Cu- Ko radiation
show that the lattice constant a of (Ga,Mn)As increases with the increase of x as shown in
fig. 7a (Ohno et a1. 1996a). Asymmetric double-crystal XRD on (224) or (115) reflection
demonstrates that the (Ga,Mn)As films are fully strained at least up to 2 JLm (fig. 7c)
(Shen et a1. 1999). The reciprocal space mapping on the (004) plane confirms also that
the (Ga,Mn)As layers are fully strained (Sadowski et a1. 2000). This rather high critical
thickness is probably due to the low growth temperature, which prevents dislocations
from nucleating (Price 1991). The direction of the strain can be controlled by using a
thick lattice-relaxed (In,Ga)As buffer layer, which has a larger lattice constant than that of
(Ga,Mn)As as shown in fig. 7b (Ohno et a1. 1996b ). The peak corresponding to (Ga,Mn)As
on (In,Ga)As is located at the higher angle side of GaAs, indicating that the film is now
under a tensile strain. In order to calculate the relaxed lattice constant Qo, it is assumed that
elastic constants of (Ga,Mn)As are the same as in GaAs (Poisson ratio: v = 0.311). On the
other hand, since thick (In,Mn)As films on GaAs are fully relaxed, a o of (In,Mn)As can be
directly determined from the positions of the diffraction peaks.
The dependencies ao(x) for (Ga,Mn)As and (In,Mn)As, as determined by XRD, are
summarized in fig. 8. In both materials, ao depends linearly on x following Vegard's
16 F. MATSUKURA et aI.

(a) (Ga,Mn)As
x=0.015
x=0.035
=0.07~ \,
.' I \
...

•I
f

(b)

~
'fi) 65 66
c:
~ 2(} (deg)
c:
- (c) (224) reflection
" - high incident angle
, \ - - low incident angle
, l-(Ga,Mn)As_
'I 2IJ.m
I (AI,Ga)As , GaAs
, I
-1500 -1000 -500 o 500
to (arcsec)

Fig. 7. X-ray diffraction curves for (Ga,Mn)As films obtained with Cu Ka radiation. (a) Mn concentration
dependence of peak positions [(004) reflection] of 150-nm thick (Ga,Mn)As grown on GaAs with compressive
strain (Ohno et aI. 1996a). (b) (Ga.Mn)As grown on (In,Ga)As buffer layer with tensile strain. (c) Double-crystal
x-ray diffraction curves for a 2 /lm-thick (Ga.Mn)As showing the asymmetric (224) reflection with high- and
low-incident angle (Shen et aI. 1999).

law, which for (Ga,Mn)As assumes the form a o = 0.566(1 - x) + 0.598x (nm) (Ohno
et at. 1996a). The lattice constant is known to depend on the growth conditions such as
As pressure and/or growth temperature due to the corresponding excess of As (Shimizu
et at. 1999; Haneda et at. 2000a). A growth of (Ga,Mn)As under other conditions gives
a o = 0.5654(1 - x) + 0.5901x (nm) (Sadowski et at. 2001a). It is shown that the lattice
constant of (Ga,Mn)As depends on the growth condition, probably due to the excess As
incorporation and the formation of a Mn-As complex (Schott et at. 2(01). The extrapolated
values of ao(x) for x ~ 0 are in good agreement with the actual GaAs and InAs lattice
constants, respectively. The extrapolated lattice constants for hypothetical zinc-blende
III-V FERROMAGNETIC SEMICONDUCTORS 17

0.61 r--.....- ......--,........ ..,.-.-oIIIIIIr--.,


..-..
E 0.60 (In,Mn}As
c:
'-'

C 0.59

~c:
o
o 0.58
Q)
0567
o E
Eto 0.57
s • 0566
-.J
0 5 6 5 ' - - - - -....
0.566 nm 000 0~4 008
0.56 .........._ .....- -......_ ....._ -...._ .....
0.0 0.2 0.4 0.6 0.8 1.0
Mn Composition x
Fig. 8. Cubic lattice constant ao versus Mn composition x in Gal_xMnxAs and Inl_xMnxAs films. Inset shows
the magnified view of the results for (Ga,Mn)As films (Munekata et al. 1989; Ohno et aI. I996a).

MnAs determined from (Ga,Mn)As and (In,Mn)As data show a good correspondence. This
suggests that virtually all Mn atoms occupy the substitutional sites. The lattice constant of
hypothetical zincblende MnAs in the ferromagnetic state has been predicted to be 0.59 nm
by first-principle calculations (Shirai et al. 1998; Ogawa et al. 1999).

3.2. Local lattice configuration (EXAFS)


An extended x-ray absorption fine structure (EXAFS) study, carried out using the Mn
K-edge, of thick (~ 111m) (In,Mn)As reveals that Mn is incorporated substitutionally into
the In sites. This is especially true for low x (,..., 0.0 I) samples grown at low Ts (200°C).
Although the substitutional signal is still dominant, either increasing of x above 0.1 or
raising Ts to 300°C results in a modification of the local structure. In the case of greater
x or higher Ts, the local structure shows that Mn is incorporated in the form of hexagonal
MnAs (NtAs-structure) (Krol et al. 1993; Soo et al. 1996).
Fluorescence EXAFS studies of a (In,Mn)As thin layer (10 nm) grown on a GaSb buffer
layer and of (In,Mn)As quantum dots (QDs) on GaAs were also performed. The results
show that in the thin (In,Mn)As layer, the In-site substitutional Mn and the NiAs-type
MnAs coexist, whereas the majority of Mn atoms are substituted into the In-sites of InAs
in (In,Mn)As QDs. It is argued that the difference of the strain deformation between the thin
layer (with strain) and thick layer and QDs (strain relaxed) is responsible for the differences
in the local structure of the Mn atoms (Ofuchi et al. 200 Ib).
18 F. MATSUKURA et al.

EXAFS measurements of (Ga,Mn)As (x = 0.005 and 0.074) also indicate that Mn atoms
are substitutionally incorporated into the Ga sublattice (Shioda et al. 1998). The Mn-As
bond length is found to be 0.249-0.250 nm, longer than the host Ga-As bond length
(0.244 nm) and shorter than the expected bond length of Mn-As in the hypothetical zinc-
blende MnAs (0.259 nm).

3.3. Atomic-scale observations by scanning tunneling microscopy (STM)

Cross-sectional scanning tunneling microscopy (XSTM) measurements were carried out


for a cleaved (110) surface of GaAs doped with Mn. The samples were cleaved in an
ultrahigh vacuum (UHV) chamber in order to expose the atomically flat (110) surface. The
measurements have been done in the UHV chamber at room temperature.
The defects and impurities of Mn-doped GaAs (Mn composition < 0.001) layers grown
at 400°C are identified in the XSTM images. The results show that Mn impurities have
negative charge, indicating that Mn acts as an acceptor. The Mn acceptor concentration
deduced from the XSTM images is in a good agreement with the hole concentration
determined by Hall measurements at room temperature. As-vacancy defects are also
observed. Since the number of the vacancies increases with time, they are presumably
formed after the cleavage due to the desorption of As atoms from the surface. No As-
antisite related defects are detected in the samples grown at 400°C (Tsuruoka et al. 2000,
2002).
XSTM images for Mn-doped GaAs (Mn composition = 0.005) grown at 255°C show
that there are numerous As-antisites, the concentration of which is '" 1 x 1020 em -3,
similar to LT-GaAs. The Mn concentration determined from the images is 7 x 1019 em -3,
which is consistent with the nominal Mn concentration, '" 1 x 1020 em -3. The tunneling
spectrum reveals the presence of a state in the midgap region caused by the As-antisite
donors and a shift of the Fermi energy brought about by the incorporation of Mn because
of its acceptor nature (Grandidier et al. 2000).

4. Spin and charge states of Mn in 111-V magnetic semiconductors

Various properties of Mn impurity centers have been investigated by many methods, such
as magnetic resonance and magnetization measurements, for a long time. It is expected
that there are three possible electronic states of the Mn impurity substituting a trivalent
cation: AO(d4 ) and AO(d 5+h) for Mn 3+ , and A-(d5 ) for Mn 2+ . AO denotes the neutral
center, A - is the negatively charged center, and the notation in parentheses is the electronic
configuration of the d electrons. There have been no reports on the observation of AO(d4 )
neutral centers in GaAs. In contrast, the anisotropy of some of electron spin resonance
(ESR) lines in the illuminated n-type GaP:Mn can be explained in terms of AO(d4 ) centers
(Kreisel et al. 1996) that undergo a Jahn-Teller distortion, as observed for Cr(3d 4 ) in GaAs
(Krebs and Stauss 1977).
In the case of the AO(d4 ) center the hole resides in the 3d shell. However, strong Hund's
intra-site exchange interaction may favor a state having five d electrons and a loosely bound
hole. This is the case of the AO(d5+h) configuration, where the A 0(d 4 ) center traps tightly
an electron in the 3d shell forming the high spin, S = 5/2, 3d5 configuration, and this
III-V FERROMAGNETIC SEMICONDUCTORS 19

negatively charged Mn ion binds the hole in an effective mass state. Experimental results
discussed below indicate that the ground state of the Mn impurity in III-V compounds
corresponds to such AO(d 5+h) configuration.

4.1. Electron spin resonance (ESR)


The result of ESR measurements for bulk Mn doped GaAs (GaAs:Mn) with a Mn
concentration of ~~ I x 1017 cm- 3 is interpreted in terms of the AO(d 5+h) acceptor state.
Owing to the antiferromagnetic exchange coupling between the d-electrons and the p-like
hole, the total angular momentum is J = I (Schneider et aJ. 1987; Masterov et aJ. 1988).
The ESR spectra for bulk GaAs:Mn with a Mn concentration of 10 17 '" 10 18 cm- 3 show
resonance with unresolved hyperfine structure even at 4 K, which can be attributed to the
existence of A- (d 5 ) centers (Almelsh and Goldstein 1962). Results of low-temperature
magnetization measurements on GaAs:Mn with a hole concentration of 5 x 10 18 cm- 3 at
300 K are consistent with the coexistence of AO(d 5+h) and A-(d 5 ) centers (Mac et aJ.
1994).
Electron spin resonance (ESR) spectra for (Ga,Mn)As grown by LT-MBE show usually
only one resonance corresponding to g = 2.0 (Nojiri et aJ. 1998; Szczytko et aJ. 1999a).
This resonance can be attributed to A - (d5 ) centers. The ESR intensity becomes weak
around the Curie temperature; i.e., the observed signal is due to the ferromagnetic
resonance (FMR). The description of the observed signal in terms of the standard FMR
formula, in which the magnetic crystaJline anisotropy is neglected, indicates that the
magnitude of magnetization is smaller than that expected for the given x value. This
suggests that only a part of the Mn spins contributes to the ferromagnetic order in
(Ga,Mn)As. The conclusion is consistent with results of other studies that will be discussed
below. From the fine structure of the ESR spectrum of (Ga,Mn)As with x < 1.5 x 10-3 ,
the single-ion crystal-field anisotropy of the Mn spin energy was determined (Fedorych
et aJ. 200 I).
No signal of AO(d 5+h) centers is usually detected in (Ga,Mn)As grown by LT-MBE
(Nojiri et aJ. 1998; Szczytko et aJ. 1999a). The reason can be either the compensation by
the antisite donors in the low x limit or the high hole concentration for larger x values,
which leads to screening of the Coulomb potentials of the A- (d5 ) centers, resulting in
a low ionization energy of the holes and eventually to the insulator-to-metal transition.
A similar situation occurs in the case of (In.Mn)As layers (Szczytko et aJ. 2oola). A
negligible contribution of the centers other than A - (d5 ) suggests that the double-exchange
mechanism of the coupling between the Mn spins is ineffective. as this mechanism requires
the coexistence of the Mn ions with a different valence.

4.2. Optical spectroscopy


The results of infrared absorption measurements on GaAs:Mn prepared by the solid-
state diffusion method are also in good agreement with the AO(d 5+h) center model
(Linnarsson et aJ. 1997). According to infrared spectroscopy and photoluminescence (PL)
measurements for GaAs:Mn with a Mn concentration of '" 1018 cmr'. this acceptor level
is located 113 meV above the top of the valence band (Chapman and Hutchinson 1967;
Ilegems et aJ. 1975). Two photoluminescence (PL) lines observed by Liu et aJ. (1995) in
20 F. MATSUKURA et al,

magnetic fields up to 30 T were identified as radiative recombination of the hole bound to


Mn with the conduction band and donor electrons, respectively. From the field-induced PL
line splittings, an effective Lande factor of the neutral acceptor g = 2.47 was determined,
confirming qualitatively the ESR results discussed above. This work corroborated also an
earlier conclusion (Schairer and Schmidt 1974) about the absence of transitions involving
excitons bound to neutral Mn acceptors in GaAs:Mn. This surprising result was explained
by Bhattacharjee and Benoit a la Guillaume (2000) taking into account the presence of
a strong exchange interaction between the holes and Mn ions. In contrast, Sapega et al.
(200 I) assigned a weak line in their PL spectrum to such a transition. At the same time,
this transition energy corresponds to a resonant enhancement of spin-flip Raman scattering
studied in detail by Sapega et al. (200 I). However, to interpret their Raman spectra, a
coupling of bound excitons to more than one Mn ion had to be invoked by Sapega et al.
(2001). It appears probable, therefore, that the PL line and spin-flip Raman scattering in
question originate from complexes involving the hole interacting with a pair of nearest
neighbor Mn ions or with other types of Mn clusters. Isolated neutral acceptors, in tum,
being unable to bind any exciton, are invisible in spin-flip Raman spectroscopy. This would
explain why the p-d exchange energy determined by Sapega et al. (200 I) is much smaller
than those imply by other experiments.

4.3. X-ray magnetic circular dichroism (XMCD)


X-ray absorption spectroscopy (XAS) provides direct information of the 3d electronic
structure of Mn, since the 2p electron is ensured to be excited into 3d state because
of the dipole selection rule. The x-ray magnetic-circular dichroism (XMCD) spectrum
obtained for 150-nm thick (Ga,Mn)As with x = 0.02 in the photon energy region 630-
660 eV shows the two groups of rich peak structures associated with 2p3/2 and 2PI/2 final
state holes. The rich structures are caused by the spin-orbit interaction of the holes and
the Coulomb and exchange interactions between the 2p core and 3d shell (Ohldag et al.
2000). The temperature dependence of XMCD at 642.2 eV, at which the 2p3/2-related
signal has maximum intensity, can be well fit by a mean-field model with Tc of 37 K,
which is consistent with Tc: of (Ga,Mn)As with x = 0.02. Comparison of measured and
calculated XMCD spectra shows that a linear combination of Mn 3d5 (80%) and Mn 3d6
(20%) configurations gives the best agreement. This configuration appears to occur for
all Mn atoms whose spins are, therefore, highly localized and the magnetic moment is
about 4.5 /-LB. From the ratio of the observed MCD to the calculated MCD, only 13%
of Mn in (Ga,Mn)As contributes to the ferromagnetic order. This is consistent with the
magnetization measurement for (Ga,Mn)As with x = 0.023, where considerable amounts
of Mn spins behave as paramagnetic spins (Oiwa et al. I998a). For as-grown samples, XAS
shows two-component contributions of Mn to the spectrum. Since one of them disappears
after low-temperature heat-treatment « 300°C), it may be related to Mn-As complexes
(Katsumoto et al. 2001). The analysis of XAS and XMCD based on the cluster model
shows that the orbital momentum of Mn electrons is small and that the sign of the p-d
interaction is antiferromagnetic (Veda et al. 200 I). The result shows that the 3d electron
»,
count ofMn is '" 5 (Ao(d 5+h) or A -(d5 which is consistent with the lack of the AO(d4 )
centers in the ESR signals.
II1-V FERROMAGNETIC SEMICONDUCTORS 21

4.4. Photoemission
The core-level x-ray photoemission spectrum of the Mn 2p core level for (Ga,Mn)As
with x = 0.074 was measured and was analyzed by a configuration interaction (CI)
cluster-model assuming a Mn 2+ and Mn3+ ground state (Okabayashi et al. 1998). For
the d 5 configuration, the p-d exchange energy (which is conventionally referred to as NoP
for DMS) should be negative and NoP "" -1.2 eV is obtained for A-(d5 ) centers with an
optimized parameter set.
The resonant photoemission technique was used for several (Ga,Mn)As layers to
elucidate the nature of the Mn 3d partial density of states (DOS) (Okabayashi et al. 1999,
2001a, 200lb). The main structure of Mn partial DOS extends from the Fermi energy EF
down to -4.5 eV below it. The configuration interaction (CI) model calculation indicates
an enhanced Mn 3d electron count of 5.3. The largest contribution to the DOS at EF comes
from As 4p states. However, the Mn partial DOS extends up to the top of the valence
band, suggesting partial d character of the conducting holes. On the other hand, in-situ
measurement on an as-grown sample shows a suppression of the d character of the holes
at Fermi level, suggesting that the observed d-character of the conducting holes may be
related to the experimental error due to the surface condition (Okabayashi et al. 200lc). In
the region extending from the Fermi energy EF down to 0.5 eV below it, an increase in
emission intensity due to Mn-induced states is observed in Gal-xMnxAs with x = 0.035
and 0.069 (Okabayashi et al. 200la, 200lb). Since an insulator-to-metal transition occurs
in the vicinity of this Mn concentration (Oiwa et al. 1997; Matsukura et al. 1998b), it is
tempting to assign this DOS to Mn acceptor states merging with the valence band. No clear
Fermi edge is observed, which may be due to a relatively low hole concentration and poor
metallic conduction.

5. Magnetic properties

5. J. Magnetization
Magnetic properties of III-V DMSs can be measured by direct magnetization measure-
ments as well as magnetotransport measurements. In this chapter, we focus on the magnetic
properties of (Ga,Mn)As obtained by direct magnetization measurements. For (In,Mn)As,
since there are only limited magnetization measurements, we will describe magnetic and
magnetotransport properties together in the next chapter.
Direct measurements of the magnetization M of (Ga,Mn)As layers as a function of
magnetic field B and temperature T have been done using a commercially available
superconducting quantum interference device (SQUID) magnetometer. The temperature-
independent diamagnetic response of the thick GaAs substrate (which could occasionally
show a slight temperature dependence of unknown origin) can be determined from a
separate measurement of only the same GaAs substrate used for the epitaxial growth. It
can be also determined from the low-temperature high-filed magnetization measurements,
where the magnetization of the (Ga,Mn)As layer should saturate, or from high-temperature
measurements, where the magnetization of the (Ga,Mn)As of the epitaxial layer on the
substrate should be negligible. The diamagnetic component is then subtracted from the
total response to obtain the magnetization of the magnetic layer.
22 F. MATSUKURA et a1.

x =0.035, 150 nm
0.04 r-........-...-............- ................
B /I plane
003 r - - - - = ~ " '~", o o - o - <

~ooo
E
-003. >-0-<>0
-002
5K

LJ
1
i0

0.03r--------,

~ ~o 0
l:;.
~~ b
b
6
~
0000 100 200
T(K)
-0.04 ...........--t..............._ ................
-0.5 0.0 0.5
B (T)
Fig. 9. Magnetic field dependence of the magnetization at selected temperatures for a 150-nm thick
Gal_xMnxAs film with a Mn composition x = 0.035. The magnetic field is applied parallel to the sample sur-
face (direction of magnetic easy axis) except for the closed circles at 5 K taken in perpendicular geometry. The
solid line for 5 K shows the magnetization determined from transport measurements. The upper left inset shows
a magnified view of the magnetization in the parallel field at 5 K. The lower right inset shows the temperature
dependence of the remanent magnetization (Ohno et al. I996a).

Figure 9 shows magnetization curves at several temperatures of a ISO-nm thick


(Gao.96sMnO.03S)As layer grown on GaAs. where B is applied parallel and perpendicular
to the plane. As shown in the inset. when B is applied parallel to the plane. the M-B
curve shows a clear hysteresis as in the upper right inset. which indicates the presence
of ferromagnetic order. A paramagnetic response is often observed after closure of the
hysteresis in the magnetization curves. as seen in the 5 K curve of fig. 9 and shown
for (Ga.Mnjas with several Mn compositions in fig. 10. This paramagnetic response is
correlated with the transport properties of the film, the more metallic sample (in terms of
the metal-insulator transition (MIT» the less the portion of paramagnetic response (Oiwa
et aI. 1998a. 1999).
Shown in the lower left inset of fig. 9 is the temperature dependence of remanence of M
after switch off of the parallel B. which reveals that the Tc of the film is '" 60 K. Note also
that there is no indication of inclusion of MnAs with NiAs-structure (Tc '" 310 K) in the
trace. as only a negligible M is present above the Tc of (Ga.Mnjxs. This. however. is not
always the case. and a small non-zero M up to 300 K is occasionally observed. especially
in samples grown at higher Ts or having large x.
III-V FERROMAGNETIC SEMICONDUcrORS 23

0.10 _ - - _ - - . . - - - _ - - " ' "


(Ga,Mn)As I LT-GaAs x = 0.071
2K ••••<;)••••
<}o.'.~
.~
sa··...., ,."., -..-.~"-
••••A-

. ..
.·0 .. ~....
v
x = 0.053

~ 0.05 #j:.-..,_·' i}_·' i}_·~;;:' cM·


~ x=0.035
x= 0.022
t::.. _O-_O-J::>--~-
x· 0.015
0.00 fIIII:; _ _...._ _.a... ...
o 2 4
B (T)
Fig. 10. Magnetization for six samples of Ga I-x Mn, AslGaAs with Mn compositions x ranging from 0.015 to
0.071 at 2 K. The magnetic field is perpendicular to the sample surface for x = 0.035-0.071 and parallel to it
for x = 0.015 and 0.022. The dashed lines show fit to the mean-field Brillouin function and the solid line for
x = 0.015 (paramagnetic sample) a fit to the Brillouin function (Oiwa et al, 1998a).

The solid line in fig. 9 shows M determined by the transport measurements, where
the Hall resistance is almost proportional to the perpendicular component of M, as
described in the next section. The good agreement between M determined by SQUID
and transport measurements indicates that one can correctly determine M of (Ga,Mn)As
by magnetotransport measurements.
=
The saturation magnetization MSal (M at T 5 K and B > 5 T) of the samples shown in
figs 9-11 indicates that S = 2.0-2.5 when calculated using MSat = x Nog /LaS and nominal
x as well as neglecting the hole contribution (Dietl et al. 200lc). Although S is related
to the valence of Mn and thus to the mechanism of ferromagnetism, a more accurate
determination of S from magnetization measurements is not possible because of the error
involved in the determination of the x value (±IO%).
Figure Ila presents magnetization determined for a 200-nm thick film of
Gao.947MnO.053As. Here almost no hysteresis is observed because B is applied perpendic-
ular to the sample plane along the magnetic hard axis. The inset displays the temperature
dependence of the remanent magnetization for B II plane, showing that Tc is above lOOK.
Using Arrott plots to minimize the effect of magnetic anisotropy and domain rotation (Ar-
rott 1957), the spontaneous magnetization Ms at each temperature is determined from
the curves in fig. lla and plotted in fig. lib. The Tc of this sample is about 110 K. The
non-standard temperature dependence of Ms shows a rather steep increase at low T. The
temperature dependence of the inverse susceptibility 1/ X shown in the same figure gives
a paramagnetic transition temperature () consistent with the Tc determined from Ms. For
the extrapolation of 1/ x, data points above 150 K were used.
24 F. MATSUKURA et al.

5K
(a)
x= 0.053 25K
0.03 55K
B 1.. plane
100 K
E
::e 0.00 125 K
250K
B /I plan

-0.03

200
--<I
-0.06~===:::L;..--~--_.....:..I:",--"",,"-~
-5 5
J
(x10 ) 1.5

0
1.0
E .....
::e'" 0.0 ~
.5

0.000 100 200 3~0


T (K)
Fig. II. (a) Temperature dependence of the magnetization for 200-nm thick Gal_xMnxAs with x = 0.053.
The magnetic field is applied perpendicular to the sample surface (hard axis). The inset shows the temperature
dependence of the remanent magnetization (0 T) and the magnetization at I T in a field parallel to the film surface.
(b) Temperature dependence of the saturation magnetization MS determined from the data shown in (a) by using
Arroll plots (closed circles). Open circles show inverse magnetic susceptibility and the Curie-Weiss fit is depicted
by the solid straight line (Ohno and Matsukura 2(01).

The Curie constant C = g2J1.~xNoS(S + 1)/3kBT, determined from l/X-T curve


shown in fig. l lb gives a spin of S ~ 3, when the nominal value of the Mn concentration,
x = 0.053, is used. Here No is the cation density, g (= 2.0) the Lande factor of the Mn ions,
J1.B is the Bohr magneton, kB is the Boltzmann constant. Judging from this value of S, some
ferromagnetic clusters exist already above Tc ;
Magnetization measurements on semi-insulating (Ga,Mn)As:Sn, in which the holes are
compensated by Sn donors, point to a paramagnetic behavior. The Curie-Weiss (l/X-T)
plot gives TAF '" 2 K, indicating that the "intrinsic" mechanism of the interaction between
the Mn spins is the antiferromagnetic superexchange (Satoh et al. 1997,2001). This means
that the ferromagnetic coupling in (Ga,Mn)As is induced by the presence of the holes.

5.2. Magnetic anisotropy


As shown in fig. 9 for the magnetization at 5 K, the magnetic easy axis is in the plane of
the film, and shows a weak four-fold symmetry within the plane. The anisotropy energy
III-V FERROMAGNETIC SEMICONDUCTORS 25
0.15 ..... ......_ .......... ~

0.10

0.05 .,
0.00
r e
_ 00
0
-0.05 - -15

-
~ -0.10
••• 300
_. -600

--; 0.031====+=+~..-.oll-- ....+-.....-t


I 1.5 urn (Gao 9BSMnO o3s)As Irr~=ro=:=-==,=:: !II
a:: 0.02 LT-GaAs
T= 2.3 K
0.01

0.00
compressive slraln
-0.01

-0.02 ;I
-0.03 .........................l........................._ ......
-0.5 0.0 0.5
B (T)
Fig. 12. Hall resistance RHall of (a) (Ga,Mn)As/(In,Ga)As and (b) (Ga,Mn)As/GaAs as a function of the
magnetic field for various angles between the field and the sample surface normal. (Ga,Mn)As films in (a) and
(b) are under tensile and compressive strain. respectively. Clear hysteresis and angle independent heights of the
hysteresis in (a) show that magnetic easy axis is perpendicular to the sample surface. whereas the easy axis in (b)
is in-plane (Shen et al. 1997a).

K calculated from the difference between the two magnetization curves at 5 K with B.l..
plane and B II plane is K = 2.9 x 103 J/m3 • This anisotropy energy of (Ga,Mn)As is strain
dependent (Ohno et a1. 1996b; Shen et a1. 1997a) and is also expected to depend on the
hole concentration (Abolfath et a1. 2001; Dietl et a1. 2oo1c). The strain in the film shown
in fig. 9 is -0.24%. The magnetic easy axis can be made perpendicular to the plane by the
reversing the sign of strain in the film.
Figure 12 shows clearly the different directions of the magnetic easy axis for the different
buffer layers. Figure 12a shows the direction of the B dependence of the Hall resistance
RHall for (Ga,Mn)As with tensile strain on a thick lattice-relaxed (Ino.16Gao.84)As buffer
layer and fig. 12b for (Ga,Mn)As with compressive strain on a GaAs buffer layer. Only
the (Ga,Mn)As with tensile strain shows a clear hysteresis which suggests a change of the
direction of the magnetic easy axis. The height of the hysteresis is almost independent
26 F. MATSUKURA et al.

of the direction of the magnetic field, which shows that the direction of the easy axis is
perpendicular to the plane. On the other hand, (Ga.MnjAs with compressive strain shows
only a small hysteresis and no perpendicular component of the remanent magnetization.
The result confirms that the direction of the magnetic easy axis of (Ga,Mn)As with
compressive strain is in-plane (Ohno et al. I996b).
The significant increase of coercive force and Ti; of (Ga.Mn)As with spinglass
(Zn,Mn)Se overlayer on it are observed. The observed effects are technological important
to control the hardness of ferromagnetism. Although the origin of which is not clear yet,
it is most probably related to the exchange coupling between (Ga,Mn)As and (Zn,Mn)Se
(Liu et al. 2(01).

5.3. Magnetic domains


Using a scanning Hall microscope, a stripe-shaped domain structure has been observed in
a (Ga,Mn)As sample with tensile strain and perpendicular easy axis (Shono et al. 2000;
Fukumura et al. 2(01). The Baukhausen noise due to the scattering from the domain
wall movement has also been observed in magnetotransport measurements (Hayashi et al.
2(00).
Ac-susceptibility measurements on (Ga,Mn)As with x = 0.042 have been performed
in alternating B from 0.1-4 mT. In the temperature dependence of susceptibility, there is
a sharp peak at about 48 K at 0.1 mT, which suggests a ferromagnetic phase transition.
The temperature and magnetic field dependence is rather complicated and an increase of
B involves additional peaks (the number of which up to 4), which may be due to domain
formation and domain wall movement (Sadowski et al. 2(00). There is also a report about
ac-susceptibility measurements on (Ga,Mn)As with x = 0.07, which shows that there is
no difference between the field-cooled and zero-field cooled ac-susceptibilities measured
with B = 10 mT (Van Esch et al. 1997).

5.4. Cantilever magnetometry


Torsional magnetometry using submicron GaAs micromechanical cantilevers is useful for
measuring the magnetic properties of small samples over a wide range of applied magnetic
fields and temperatures. Since (Ga,Mn)As can be grown epitaxially on GaAs, it can be
incorporated in a GaAs cantilever using photolithography. l5-lLm-radius disk-shape mesas
of (Gao.962MnO.03S)As samples with 6 nm and 20 nm of thickness have been fabricated at
the end of the cantilevers and their magnetic properties have been investigated (Harris
et al. 1999). The cantilevers are 50 ILm wide and 100 nm thick and the longest lever
is '" 400 ILm long. The cantilever with (Ga,Mn)As shows a twist of several degrees in
the last 50 ILm due to the lattice mismatch between GaAs and (Ga,Mn)As. From the
measurement, it is found that the torsional magnetometers have quite a high sensitivity
of 3 x 106 ILB at 0.1 T. The obtained magnetization curves for 20-nm thick (Ga,Mn)As
shows ferromagnetic behavior with an in-plane easy axis, being consistent with the results
obtained with a SQUID magnetometer for much larger samples. From the temperature
dependence of the remanent moment, Te is determined as 39 K. Illumination of the sample
with a blue light-emitting diode (LED) has no measurable effect on the magnetization of
(Ga,Mn)As either above or below Te. Measurement on 6-nm thick (Ga,Mn)As shows no
ferromagnetism down to 350 mK, which is consistent with SQUID results (Tanaka 1998).
III-V FERROMAGNETIC SEMICONDUCTORS 27

6. Magnetotransport properties

Due to the presence of the anomalous Hall effect (known also as the extraordinary or
spin Hall effect), magnetotransport measurements provide valuable information on the
magnetism of thin films. The Hall resistance RHall is empirically known to be a sum of
ordinary and anomalous Hall terms,

(1)

Here, Ro and Rs are the ordinary and anomalous Hall coefficients, respectively; d
is the thickness of the conducting channel; Ml. is the component of magnetization
perpendicular to the sample surface (Chien and Westgate 1980), and Rs is proportional
to R~heel with temperature-independent proportionality constant, where Rsheel is the sheet
resistance. Usually, y is either 1 or 2 depending on the origin of the effect; the skew-
scattering mechanism results in y = I, whereas for the side-jump mechanism y = 2. The
proportionality constant, which determines the overall magnitude of the anomalous Hall
effect, scales with the strength of the spin-orbit coupling for the carriers at the Fermi level
as well as with the exchange energy describing the ratio of carrier spin polarization to
magnetization M«. Accordingly, at given Ml.' the effect is expected to be much stronger
for the holes than for the electrons in the tetrahedrally coordinated semiconductors.
Furthermore, since its magnitude depends on the degree of spin polarization of the carrier
liquid, it ceases to be proportional to the magnetization when the carrier spin-splitting
becomes comparable to the Fermi energy.
Because of the high sensitivity, the determination of the magnetization by magneto-
transport studies is the important technique for thin films of diluted magnets, in which the
magnitude of the total magnetic moment is rather small. Accordingly, recent years have
witnessed a renewed interest in the theory of the anomalous Hall effect (Hirsch 1999; Ye
et al. 1999; Zhang 2000; Ferrand et al. 2000; Crepieux and Bruno 2001; Jungwirth et al.
2002). Theoretical calculations for p-type OMS, based on the side-jump mechanism, sug-
gests an increase of the Hall conductivity with the decrease of the carrier concentration
(Jungwirth et al. 2002). Experimental results discussed below: (i) demonstrate the criti-
cal importance of the Hall effect in the assessment of the magnetic properties of III-V
ferromagnetic semiconductors; (ii) suggest that both side-jump and skew-scattering mech-
anisms operate (iii) point to various effects that can lead to differences between magne-
tizations determined by standard and Hall magnetometry. Furthermore, the accumulated
magnetoresistance data indicate a significance of the spin-disorder scattering as well as re-
veal various effects associated with the interplay between spin and localization phenomena,
specific to doped OMS in the vicinity of the metal-insulator transition (Dietl 1994).

6.1. (Ga,Mn)As
6.1.1. The Hall effect
Figure 13a presents the Hall resistance RHall at various temperatures plotted as a
function of the magnetic field for the same sample, for which magnetization data were
collected in fig. I I (200-nm thick Gao.947Mno.o53As). The inset shows the temperature
28 F. MATSUKURA et al.

0.03 ....
;- :..:.:..'~:..:.;...:.:;,,;.~

(a) x =0.053 2K
/ ............. 1b~~~~ ..~.
-c B ..L plane 25 K
55 K
/
1 , ~
~ .~ .. ~. 125 K

---
~

( ij 0.00 . ,..........................'1'-: . '05'" •• , •••.•:: .••••••••••


300 K

o::I ..........
",0"

t
_

g
~

0.4 .•.•.•._'~ . . . .
-r.;.... ~ ~ O O K
~2.5.~5~ ~.~ •• ""
..~.. / I
.,~'
..•.•
. ~ •. ~ •. ~

: -: -. -: . -: . ~
_/ ,J ./

• . . • . .
ll:.
····2K,

03
....25K ...•
..::""", _

-5 8?n
30CK
5
-0.03
-5 o 5
B (T)
0.08

...... III

-..,
Qj
Q)

o::lIls: 0.04
l1)
............ 0-
I
(ij

--
C
0:: 0.02 ::J
;::;:

0.0
---en
200 300
T (K)
120
(c)

----
~
80 ~

40 ~

8.00 0.04 0.08


X
Fig. 13. (a) Temperature dependence of the Hall resistance RHaJl for the same sample as in fig. II. The inset
shows the temperature dependence of the sheet resistance Rsheet. (b) Temperature dependence of the saturation
magnetization [RHaJlI RsheedS obtained using Arrott plots (closed circles) and inverse susceptibility I/XHaJl
(open circles), both from the transport data shown in (a). Solid lines depict [RHaJl 1Rsheells and II XHaJl calculated
assuming the mean-field Brillouin behavior for the Mn spin S = 5/2 and the Curie-Weiss law, respectively. (c) Mn
composition dependence of the magnetic transition temperature Te. as determined from transport data (Ohno and
Matsukura 2001).
III-V FERROMAGNETIC SEMICONDUcrORS 29

dependence of Rshcct- A close similarity between the results of figs 13a and lla indicates
that the contribution from the ordinary Hall term is rather small in the displayed field
and temperature range. Thus, if skew-scattering is assumed, RHall :=:= c RSheetM, where
c is a temperature-independent constant. Actually, a comparison of magnetization and
magnetotransport data suggests the value of the exponent y to be between 1 and 2
at low temperatures, and that the assumption of either mechanism leads to virtually
identical conclusions about the magnetic properties. In particular, since RHall/ Rshcct ex M,
Arrott's plots can be employed to determine the temperature dependence of spontaneous
magnetization Ms.
The outcome of transport measurements is summarized in figs 13b and 13c. The
values of Tc are in good agreement with those determined from the direct magnetization
measurements. As shown in fig. 13c, the value of the Curie temperature is almost
proportional to x up to x of about 5% according to Tc :=:= 2000x ± 10 K. However, a further
increase of x decreases Te. The origin of this behavior is unclear; it may result from the
compensation by interstitial Mn donors (Yamamoto and Katayama-Yoshida 1999; Masek
and Maka 200 1) and/or from changes in the local spin configurations (van Schifgaarde and
Mryasov 2(01).
The temperature dependence of Ms determined by the magnetotransport measurements
can be fitted rather well by the mean-field Brillouin function, as shown by the solid line
in fig. 13b (Matsukura et a1. 1998b). Owing to a moderate temperature and magnetic
field dependence of Rsheeh this conclusion remains valid assuming RHall :=:= c'R;heetM.
However, a rather different temperature dependence stems from the direct magnetization
measurements (fig. l lb). To conjecture about the origin of this difference one should
note that the anomalous Hall effect scales with the spin polarization of the carrier liquid.
This polarization is proportional to the magnetization only if (i) the spin-splitting is
much smaller than the Fermi energy and (ii) the contribution of the carriers to the
total magnetization can be disregarded. Furthermore, the anomalous Hall effect does
not provide information about the magnetization of the whole samples but only about
its value in regions visited by the carriers. Just in these regions the carrier-mediated
ferromagnetic interactions are strong. Thus, since near the metal-insulator boundary
the carrier distribution is highly non-uniform, magnetotransport and direct magnetic
measurements may provide different magnetization values (Dietl et al. 2000).
Parenthetically, no clear indication of the presence of MnAs clusters has been observed
in the transport results, even in the cases where direct magnetization measurements detect
their presence. One of possibilities is that the Schottky barrier formation around the MnAs
clusters prevents their interaction with the carriers.

6.1.2. Temperature and magnetic field dependence of resistivity


The temperature dependencies of the resistivity p in 200-nm thick layers of Ga I-x Mn, As
with different Mn content x are displayed in fig. 14. In terms of the metal-insulator tran-
sition (MIT), these films can be cost into two categories. Low- and high-Mn composition
samples (x < 0.03, x > 0.06) are on the insulator side of the MIT, whereas the layers con-
taining intermediate Mn concentrations (0.03 ~ x ~ 0.06) are metallic. All samples exhibit
a negative magnetoresistance (MR) at low temperatures. Quite generally, the MIT occurs
30 F. MATSUKURA et aI.

1 200-nm (Ga,Mn)As / (AI,Ga)As


10 2
1.0 x 10'

- E
o
c
......... 10.1 x
~ 0.015
0.022
~:-::--..r-L,~0.071

0.035
0.043
~ ..........__...a...._ _.....................__..z0.053
o 100 200 300
T(K)
Fig. 14. Temperature dependence of the resistivity p at zero magnetic field for Gal_xMnxAs films with
x = 0.015~.071. Samples with x = 0.035~.053 exhibit metallic behavior. The inset shows an expanded view
for the sample with x = 0.053 together with the dependence on the magnetic field (Matsukura et aI. 1999b).

if the kinetic energy of holes at the Fermi level is high enough to prevent localization by
disorder associated with the presence of Mn acceptors and compensating donors.
The temperature dependence of p in the metallic (Ga.Mn)As samples show a maximum
around Tc (the inset to fig. 14). where the negative magnetoresistance also peaks; p
decreases by 20% in B = 7 T (at 100 K) (see the insets to figs 13a and 14). Such
behavior suggests the presence of spin disorder scattering by thermodynamic fluctuations
of the magnetization. In particular. the peak around Tc can be interpreted in terms of a
critical scattering by packets of ferromagnetic ally coupled spins. whose correlation length
is comparable to the wavelength of the carriers at the Fermi level (Matsukura et al. 1998b;
Omiya et al. 2000). Such spin-disorder scattering may also perturb the position of the
mobility edge, affecting the temperature dependence of the conductivity (Van Esch et al.
1997). The negative MR occurs because the field-induced spin-alignment reduces the spin-
disorder scattering.
The effect of critical scattering on the resistance was evaluated assuming that the holes
reside in a simple parabolic band (Omiya et al. 2000). The corresponding contribution to
the resistivity has to take into account the presence of correlation between neighboring
spins. (SiSj) #- (S;)Oij (Dietl 1994).

2 kF m 2 fJ2 kBT
PS = 2iT - 2 -3-2"2[2x.d T • B) + XII(T. B)]. (2)
pe h g /-LB
III-V FERROMAGNETIC SEMICONDUcrORS 31

. . •
1.00 ~ T
I • P
xT
iT
(J

- 0.90 ~ -IT
E
0 -.T-'T1'
-
c:
~ 0.80 ~~£T .
~~2
~~TTT
~.IIl.T
~
0.70 ~

. . .
150 200 250 300
T(K)
Fig. 15. Temperature dependence of the resistivity for a 200-nm thick film of Gal_xMnxAs with x = 0.053 in
the high-temperature paramagnetic region. Solid squares and open circles show experimental data and the fitting
using eq. (2), respectively (Omiya et aI. 2000).

Here kF is the Fermi wave vector determined from the value of the hole concentration p
assuming a spherical Fermi surface, m is the hole effective mass taken as 0.5mo (mo is
the free electron mass), fJ is the exchange integral between the holes and the Mn spins,
and h is Planck's constant. The transverse and longitudinal magnetic susceptibilities are
determined from the magnetotransport data according to Xl. = aM / aB and X" = M/ B.
As shown in fig. IS, eq. (2) reproduces the high-temperature part of the temperature
dependence of p (Omiya et al. 2000); the data can be fitted by XT, the temperature
dependence characteristic for critical scattering. On approaching Tc; however, the XT
fit deviates from the experimental points. This is, presumably, because no dependence
of the magnetic susceptibility X on the wave vector q is taken into account. Since X (q)
is a decreasing function of q in ferromagnetic materials, eq. (2) overestimates the critical
scattering, particularly near Tc, where X (q = 0) diverges. As shown in fig. 16, the negative
MR can be explained by a reduction of spin-dependent scattering in the magnetic field,
according to eq. (2).
From the fit of eq. (2) to the temperature and field dependencies of p, the exchange
energy INofJ I was evaluated by Omiya et al. (2000). Both T and B fits yield INofJ I of
1.5 ± 0.2 eY. Though this value of the exchange energy compares favorably with that
determined by photoemission experiments on (Ga,Mn)As, 1.0-1.2 eV (Okabayashi et al.
1998), it should be stressed that it has been obtained disregarding effects of localization.
Another possible source of ambiguity is the complex valence band structure, not accounted
for in eq. (2).
A rather large negative MR, with a substantial anisotropy, has been observed in
'reentrant' insulating samples with high x (Oiwa et al. 1998b; Katsumoto et at. 1998).
A destruction of bound magnetic polarons, as invoked in an earlier work on (In,Mn)As
32 F. MATSUKURA et al.

x10-3 d=200nm
8.0 r"-P~f-r-T"""" __T""'I"""T'"""""""""'....-r-.......__'P-o

x = 0.053 'J exp


B -l plane --fit

-c:
E
o 200K

-c:..
7.5

.-. ~- ., ... '''; ....."".,..


250 K

7.0 .......
-10 -5 o 5 10
B (T)
Fig. 16. Magnetic field dependence of the resistivity for a 200-nm thick film of Ga I _ .r Mn x As with x = 0.053 in
the high-temperature paramagnetic region. The solid lines show the fitting using eq. (2) (Omiya et al. 2000).

(Ohno et al. 1992), and an increase in the hole kinetic energy due to giant spin-splitting,
are suggested as mechanisms leading to the enhanced conductivity in the magnetic field.
The latter can be particularly strong in p-type materials, in which the exchange splitting
mixes heavy and light hole subbands (Wojtowicz et al. 1986).
The theoretical model describes transport properties of (Ga.MnjAs (T and B depen-
dence) by using Zubarev's double-time Green's function rather well (Kuivalainen 2001).
The results show that the spin scattering (spin disorder scattering and critical scattering) as
well as ionized impurity scattering have significant contributions to the total carrier mobil-
ity. For better understanding, an improvement of the theory is necessary, which includes
the effect of disorder, the formation of polarons, etc.
Anisotropic magnetoresistance has also been observed in metallic (Ga,Mn)As, which
depends on the relative direction between the magnetic field and the current. For
(Ga,Mn)As with compressive strain grown on a GaAs buffer layer, the resistance is
higher when the magnetic field is perpendicular to the measuring current than when the
two are mutually parallel (Baxter et al. 2(02). On the other hand, for (Ga,Mn)As with
tensile strain grown on an (In,Ga)As buffer layer, the resistance is lower when a magnetic
field perpendicular to the current (Sugawara et al. 1997). The origin of the anisotropic
magnetoresistance is not clear yet.
Because of the dominance of the anomalous Hall term in wide temperature and
field ranges, it is not straightforward to determine the carrier type and concentration
in ferromagnetic semiconductors. Only at low temperatures and in very high fields, the
anomalous Hall term saturates, so that the ordinary Hall coefficient can be determined
from the remaining linear change of the Hall resistance in the magnetic field. Note that
although M saturates in relatively low magnetic fields, the negative MR usually persists,
and generates the field dependence of the anomalous Hall coefficient.
III-V FERROMAGNETIC SEMICONDUCTORS 33

30
30.5

-.a..... 20 -9.
'ii
30.0
50mK

:I: exp
z"iii 0:: --linear fit
rr. 10
29.5
22 24 26 28

0
(a) an
0.0

-e
l-0.1

-0.2 L L . _...._ ...... _ ~ _ ................- - '


o 10 20 30
B (T)
Fig. 17. Magnetotransport properties of a 2oo-nm thick film of Ga I-x Mnx As with x = 0.053 at 50 m K in high
magnetic fields. (a) Hall resistance. which is a linear function of the magnetic field in the high-field region (inset).
(b) Sheet resistance; negative magnetoresistance tends to saturate in the high-field region (Omiya et aI. 2000).

Measurement of RHall at 50 mK in the field range of 22-27 T on the sample with


x = 0.053 revealed that the conduction is of p-type, consistent with the acceptor character
of Mn. The determined hole concentration was p = 3.5 x 1020 em -3, about 30% of the
nominal concentration of Mn (fig. 17) (Omiya et al. 2000). A similar value of the hole
concentration. which is almost independent of x, has been obtained from the Seebeck
coefficient assuming a simple model of the valence band (Osinniy et al. 2001).
If all Mn centers are acting as acceptors in the metallic sample described above, 70%
of them must have been compensated by donors. The most natural candidates for these
donors are As antisite defects, which act as deep donors in GaAs. Accordingly, (Ga,Mn)As
should become insulating at room temperature when the density of As anti sites exceeds the
density of shallow acceptors. Because the magnitudes of these densities are comparable and
moreover fluctuate from run to run depending on subtleties of the growth conditions, we
expect the overcompensation to occur occasionally. However, no such 'overcompensated'
sample has been obtained so far. This seems to call for mechanisms controlling the
upper limit of the excess As concentration (Luysberg et al. 1998) and/or leading to
autocompensation of Mn but not to overcompensation (Walukiewicz 1988). One candidate
for the latter might be the Mn interstitial, which is a donor according to first principles
calculation (Yamamoto and Katayama-Yoshida 1999; Masek and Maka 2001). Another
34 F. MATSUKURA et at

candidate is the formation of disordered sixfold-coordinated centers with As, which would
act as a donor (Van Esch et al. 1997).
Since the As antisite is anyway one of the most important defects acting as the
compensating donor, the excess As influences substantially the magnetic and transport
properties of (Ga,Mn)As. The increase of substrate temperature and the decrease of the As
pressure reduces the density of excess As, which result in a decrease of the lattice constant
and an increase in both the hole concentration and conductivity. Importantly, this generates
a raise of Tc (Shimizu et al. 1999), confirming the critical role of band holes in the
ferromagnetism of (Ga,Mn)As. The annealing of (Ga,Mn)As at relatively low temperatures
(~ 300 K) leads to similar results due to the evaporation of excess As (Hayashi et al. 200 1;
Potashnik et al. 200 I).

6.2. (/n,Mn)As
6.2./. n-type (/n,Mn)As
Magnetization measurements reveal that n-type (In,Mn)As layers with x = 0.046-0.18
grown at 200°C are paramagnetic. The analysis in terms of the Curie-Weiss law shows
that the interaction between the Mn ions is antiferromagnetic with the nearest-neighbor
exchange inn = -1.9 K (von Molnar et al. 1991). The effective magnetic moment ofMn is
determined as Peff (= gJS(S + 1)) = 5.49 for x = 0.046, indicating that the Mn electrons
are mostly in the high spin d 5 configuration.
Magnetotransport measurements on n-type samples show no evidence for carrier-Mn
spin interactions (Ohno et al. 1991). In particular, no magnetization-dependent anomalous
Hall effect is observed within the experimental error.

6.2.2. p-type (/n,Mn)As


p-type (In,Mn)As shows ferromagnetism at low temperatures. Magnetotransport mea-
surements of thick p-type (In,Mn)As films grown on (00 I) semi-insulating GaAs reveals
that the temperature dependence of the Hall coefficient, RH, can be expressed, as is the case
of (Ga,Mn)As, by eq. (1) (Ohno et al. 1992). Since M = XH, B = ILoH, X = C/ (T - ())
in the paramagnetic temperature region, and assuming Rs = cp one obtains the Hall coef-
ficient as,

(3)

where the Hall resistivity PH all = dRHall. Equation (3) can describe the temperature
dependence of RH of a 1.3-Jlm thick (In,Mn)As sample (x = 0.013) grown directly on
a GaAs substrate at 275°C, especially from 20 to 200 K, as shown in fig. 18. The fit carried
out by adopting the calculated value of the Curie constant (using x = 0.013 and assuming
Mn spin S = 5/2) yields Ro, c and (), which are p = 2.2 x 10 19 cm- 3 (determined from
Ro), c = 5.6, and () = 3.8 K. The sign of () demonstrates that the interaction between the
Mn spins is ferromagnetic. A ferromagnetic Mn-Mn interaction has only been observed
in p-type samples so far, proving that this interaction is induced by the presence of the
holes. Equation (1) is shown to be valid at low temperatures also, where hysteresis appears
in the field dependence of PHall, as shown in the inset to fig. 19. This points to the
presence of ferromagnetic order. By comparing the remanent magnetization determined
III-V FERROMAGNETIC SEMICONDUCTORS 35

10.2 10.7

......... \1 V P
E ...... -:...\1 \1 \1 \1 \1 \1\1

-c 10.4 ...
--- .!--
10.9
:-..:

--
~ .........
......... I
t: 3
E
-
c
Q:l:
10-6
~ a _

RO
•••••••• _ .....
...
. . ......

/-----
.... . . . .
10. 11

CpxllJ o
10.6 10. 13
10 100 300
T (K)

Fig. 18. Temperature dependence of the Hall coefficient RH and resistivity p of a 1.3-JLrn thick film of
Inl_xMnxAs with x = 0.013. RH can be modeled over a wide range of temperatures as Ro + cPX/JLo, which
is shown by the solid line assuming C = 5.6. The susceptibility X (depicted by the dashed line) is calculated
assuming the Curie-Weiss law with x = 0.013, Mn spin S = 5/2, and the Curie-Weiss temperature 8 = 3.8 K
(Ohno et al, 1992).

0.015.... -...--,....""1"""".....- . . . - - , , - -_ _... 6

2.8 K

:l·
E
~
3.5 K
-
'tl
3
:0

0.005
0 -100
0 20 40 60 80
2 - 3

0
0 0 0 BdmT)
o 0

0.000 .....
o
----1...--.. . ---'1...--..
2 4 6 8 10
0

B (T)
Fig. 19. Magnetic field dependence of the diagonal resistivity P (open circles) and magnetization MHall (close
circles) determined from the ratio of the Hall and diagonal resistivities. MHall = PHali/CP, where C = 6.3, for
a 1.3-llm thick film ofInl_xMnxAs with x = 0.013. The solid line is a fit by the modified Brillouin function
BS(y). where S = 5/2 and y = SgIlBB/(T + To) with To = 1.5 K. The inset shows the hysteresis observed in
the Hall resistivity at 3.5 K (Ohno et al, 1992).
36 F. MATSUKURA et aI.

from the SQUID measurements at low temperatures with the temperature dependence of
the remanent part of PHalliP (PHall ~ RsM = cpM) as shown in fig. 19, one can again
evaluate the proportionality constant c as 6.3, in good agreement with c = 5.6 obtained
from the paramagnetic temperature region. This indicates that the mechanism responsible
for the anomalous Hall effect does not change between the two temperature ranges.
The magnetic field dependence of MHall (= PHalli Rs) (i.e., magnetization measured by
the anomalous Hall effect) can be expressed as a sum of ferromagnetic and paramagnetic
terms. The latter can be described by a modified Brillouin function (Gaj et a1. 1979), as
shown in fig. 19. The magnitude of saturation magnetization (the sum of the two responses
at B = 0.015 Tat T = 3.5 K) is 0.014 T,just the value expected for x = 0.013 and S = 5/2.

6.2.3. Pseudomorphic p-type (In,Mn)As


The (In,Mn)As layers grown on quasi-lattice matched buffer layer (GaSb or (AI,Ga)Sb)
films are either superparamagnetic or ferromagnetic at low temperatures depending on
the initial growth mode (Shen et a1. 1997c). They have rather high Tc (= 50 K) with
19
x ~ 0.07 and relatively low p (~2-3 X 10 cm >', determined from the Hall coefficient
at room temperature) (Slupinski et a1. 2002b). The magnetotransport properties show no
clear evidence for partial ferromagnetism, that is for the co-existence of ferromagnetism
and paramagnetism in the same films. A comparison of transport and magnetization
measurements revealed that the anomalous Hall coefficient Rs is rather proportional to p 2
than to P in the low-temperature regime (Oiwa et a1. 1999). Surprisingly, in contrast to the
case of (Ga,Mn)As and (In,Mn)As with high hole concentrations, the sign of the anomalous
Hall effect was found to be negative in (In,Mn)As films with low hole concentrations
(Munekata et a1. 1997). Such a sign reversal appears to be possible if skew-scattering by
compensating impurities dominates (Leroux-Hugon and Ghazali 1972; Chazalviel 1975;
Dorleijn 1976).
The magnetic anisotropy in (In,Mn)As has been investigated for various strains in the
film, by means of varying the lattice constant of the buffer layer (Munekata et a1. 1993). To
increase the tensile strain, (AI,Ga)Sb with high Al composition was employed, whereas to
decrease its magnitude, AI(Sb,As) was used as a buffer layer. The corresponding reduction
of the tensile strain diminishes the perpendicular magnetic anisotropy, which demonstrates
that strain is indeed responsible for the magnetic anisotropy. At the same time, no changes
in Tc are observed upon changing the strain (Fumagalli and Munekata 1996). The thickness
of the (In,Mn)As film affects also the magnitude of the magnetic anisotropy. For thick
samples there is no magnetic anisotropy due to the relaxation of the lattice mismatch,
whereas for very thin samples there is no ferromagnetic order due to the depletion of
the holes. The thick (In,Mn)As films without anisotropy exhibit long-term magnetization
relaxation at low temperatures, which suggests that a cluster-glass state is formed (Oiwa
et a1. 1999).
The magnetic properties of structures consisting of an (In,Mn)As layer grown onto an
(In,AI)As with an composition of Al '" 0.15 and an AISb buffer layer were examined by
means of the anomalous Hall effect. The weakening of hysteresis with the increase of the
intermediary (In,AI)As layer thickness was observed. This may result from a decrease of
overlap between the Mn spins in the (In,Mn)As layer and the holes accumulated at the
(In,AI)AsIAISb heterointerface (Munekata et a1. 1992).
111-V FERROMAGNETIC SEMICONDUCTORS 37

6.3. Infrared and far infrared optical conductivity


Infrared (IR) and far infrared (FIR) transmission spectra were collected between 10 and
12000 em -I for two (Ga,Mn)As samples with metallic de conductivity (Nagai et al. 200 I).
The thickness and Mn concentration of these two films were 2 ILm, x = 0.034 and 0.4 ILm,
x = 0.050, respectively. In the whole spectral region studied, the absorption coefficient
is larger for the sample with the higher Mn composition. In the absorption spectrum of
the sample with x = 0.034, a broad peak is observed around 1600 em"! (200 meV).
Most probably, this peak reflects both photo-ionization transitions between Mn impurity
states and the valence band as well as inter-valence band transitions, which have been
also observed in the IR absorption spectrum of bulk GaAs:Mn with a Mn concentration
of 4 x 10 17 cm- 3 (Chapman and Hutchinson 1967). There exists also quite a different
interpretation, which assigns the peak to the formation of polarons (Katsumoto et al. 200 I).
This peak is less visible in the sample with x = 0.050, presumably because of a greater
degree of overlap among the Mn impurity states, which leads to hole delocalization. In the
FIR region, the absorption increases monotonously with the photon energy, and no clear
Drude conductivity is observed down to 10 em -I. This behavior may reflect the influence
of quantum localization effects on the Drude conductivity as well as the coexistence of the
hopping and Drude terms in the vicinity of the MIT (Liu et al. 1993; Gaymann et al. 1996).
The dependence of the absorption coefficient on temperature and the magnetic field
in the FIR region is consistent with the corresponding behavior of the d.c. conductivity.
In particular, the optical density shows both a minimum around Te and an increase in
the magnetic field. When the absorption spectrum at Te is subtracted from the spectra
below Tc, a Drude-like absorption is obtained below 400 meV, whose magnitude increases
as temperature decreases. The Drude-like absorption and localization-induced effects are
observed also in (In,Mn)As (Hirakawa et al. 2001). A detected shift of the spectral density
towards higher energies was taken as an evidence for the double exchange spin-spin
interaction mechanisms (Hirakawa et al. 200 I). This conclusion awaits a verification by
considering the behavior of the sum rule in the case of the narrow-gap and complex valence
band structure, specific to p-type (In,Mn)As.
Theoretical simulations for the optical spectra were also been performed (Sakai et al.
200 I; Sakai and Suzuki 200 I).

6.4. Cyclotron resonance


Magneto-transmission spectra of GaAs:Mn with Mn a concentration of '" 9 x 10 17 cm- 3
grown by MBE were taken at several temperatures in magnetic fields up to 110 T by
the use of a FIR laser line of 119 ILm (10.4 meV) and 70.5 ILm (17.6 meV) (Matsuda
et al. 1998). It was found that the carrier mobility is too low above 270 K to observe
the cyclotron resonance (CR) and that carrier freeze-out occurs below 140 K. In the
intermediate temperature region a CR peak was observed in '" 40 T at 119 ILm. The
mobility determined from the CR linewidth at 220 K is 780 cm 2/Vs. The cyclotron mass
is 0.42mo and r - 0.47mo at 119ILm and 70.5 ILm, respectively.
'"v

On the other hand, 150-nm thick (Ga,Mn)As with x = 0.053 (Te = 88 K) shows no CR
up to 150 T at 119 ILm, which may result from too a low mobility value. In the investigated
temperature range from 20 to 150 K, the transmission decreases with the magnetic field,
38 F. MATSUKURA et al,

this behavior being consistent with a negative magnetoresistance observed in such samples.
The spin origin of the effect is corroborated by the fact that the magnetic field at which the
transmission ceases to change corresponds with to the saturation field of the magnetization
(Matsuda et al. 1998).
The CR of n-type (In,Mn)As layers with x = 0, 0.08, and 0.12 was observed at 5.5 J.l.m
(224 meV) and 10.6 J.l.m (117 meV). The cyclotron mass at 224 meV decreases from
0.054mo to 0.048mo for x increasing from 0 to 0.12. The conduction-band effective
mass depends also on temperature. For example, for the x = 0.08 sample, it decreases
from 0.051mo at room temperature to 0.045mo at 18 K. For l-um thick (In,Mn)As with
p""" 1019 cm'. the CR of the light-hole was detected (Matsuda et al. 2001).

7. Magneto-optical properties

Magneto-optical properties of (In,Mn)As and (Ga,Mn)As have been studied in order to


elucidate the origin of ferromagnetism as well as to explore the possibility of using
these materials as Faraday isolators, suitable for monolithic integration with the existing
semiconductor lasers.
The absorption edge of (Ga,Mn)As is not sharp, as shown in fig. 20 (Kuroiwa et al.
1998; Szczytko et al. 1999b). This is probably due to impurity band formation caused by
the high concentration of ionized Mn and compensating donors (Kuroiwa et al. 1998).
Even below the fundamental absorption edge, the absorption coefficient is rather large
due to free-carrier (Casey et al. 1975) and intra-Mn absorption. There is no report on the
observation of exciton states or photoluminescence, which is probably due to non-radiative
recombination, carrier screening, and the formation of an impurity band (Ando et al. 1999).

- - (Ga,Mn)As, x = 0.043
- - - LT-GaAs

<Ll
o
c
ro
~
E
(/)
c
.-~
1.3 1.4 1.5 1.6 1.7
Energy (eV)
Fig. 20. Optical transmission spectra of 2-llm thick films of Ga J -x Mnx As with x = 0.043 and low-temperature
grown GaAs at - 10 K (Kuroiwa et al, 1998).
III-V FERROMAGNETIC SEMICONDUCTORS 39

7.1. Faraday rotation


Figure 21a shows Faraday rotation spectra of (Ga,Mn)As (2 J-Lm, x = 0.043) at 10 and
300 K in the vicinity of the band gap energy (Kuroiwa et al. 1998). Compared to GaAs,
the magnitude of the Faraday rotation in (Ga,Mn)As is very large and proportional to
magnetization, as shown in fig. 21b. The observed value of the Faraday rotation is 6 x
104 0 fcm in 0.1 Tat 10 K, and the room temperature Verdet constant is 9 x 10- 2 0 fG cm.
Oscillations seen in the low-energy region at 10 K are caused by interference originating
from internal multiple reflections.
The Faraday rotation angle eF can be expressed as,

Ed an
eF~ --!:lE (4)
2hc aE '
where E is the photon energy, c is the light velocity, n is the refractive index, and !:lE is
the energy difference of the transitions for the two circular light polarizations. The value

15 (a) (Ga,Mn)As
=
x 0.043
.- 10 6T
~
"C
-~ 5
300 K
o
1.4 1.6 1.8
Energy (eV)
15 . ~:::A:=&=t5=~
r--..,..-...... {I'J~~~
10 ~ (b)
(Ga,Mn)As 300K
.- 5 ~ x= 0.043 1.49 eV
C>
I
- o e;..,:,..,:,..,:,..,;,,"-,-._
Q)
"C __-tJ-e-~-t~!t"_.-D:-.H~_B~
~
-5 ~
-10 K _ _ (.J-magnetization
-10 ~ 1.55 eV _ ~. (from transport)
,; .
-3 0 3 6
B (T)
Fig. 21. Faraday rotation for a 2-JLm thick film of Gal_xMnxAs with x = 0.043 measured as a function of
the photon energy in a magnetic field of 6 T at - 10 and 300 K (a) and as a function of the magnetic field at
- 10K, 1.55 eV and at 300 K. 1.49 eV (b). Solid lines show the magnetization determined from magnetotransport
measurements at the given temperatures (scaled to match the open symbols) (Kuroiwa et al. 1998).
40 F. MATSUKURA et aI.

of !:i.E inferred from eq. (4), using experimentally determined anjaE, is about 35 meV,
fairly independent of E between 1.25 and 1.6 eV. If one uses !:i.E = (Nof3 - Noa)x(Sz),
which is valid only at the band edge, then (Nof3 - Noa) ~ 1 eV is obtained. Here Noa
is the exchange integral for the conduction band, and (Sz) the thermal average of the Mn
spins in the direction of B, determined by an independent magnetization measurement. The
positive value of (Nof3 - Noa) reflects the positive sense of the Faraday rotation, which is
opposite to that of II-VI DMSs like undoped (Cd,Mn)Te. This surprising result is explained
if a large Burstein-Moss shift due to the high hole concentration specific to III-V DMS is
taken into consideration (Szczytko et al. 1999b; Dietl et al. 200 lc).
The magnetic field dependence of the Kerr-rotation in (Ga,Mn)As at low-temperature
shows clear hysteresis reflecting the ferromagnetic order (Sadowski et al. 200Id). A study
on reflectivity and polar Kerr-rotation of thin (In,Mn)As films grown on (Al,Ga)Sb buffer
layers showed that the squareness (the ratio of the Kerr-rotation angle at remanence to that
at saturation) can be as high as 100% (Fumagalli and Munekata 1996).

7.2. Magnetic circular dichroism (MeD)


Magnetic circular dichroism (MCD) is caused by the difference in the magnitude of photon
absorption for left-polarized (0"-) and right-polarized (0"+) light,

(5)

where a+ (T+) and a- (T-) are the absorption (transmission) coefficients for right and
left circularly polarized light, respectively. The incorporation of Mn enhances the MCD
signal at optical critical points as shown in fig. 22 (Ando et al. 1998). According to MCD

Eo GaAs (x10) x= 0
(a)

E, E , +/l 1

(Ga 1• xMn)As
x = 0.005
0
0
~

(c)

5K
1T

0 2 4
E (eV)
Fig. 22. Magnetic circular dichroism (MeD) spectra of (a) undoped semi-insulating GaAs substrate and (b), (c)
of epitaxial Gal_xMnxAs films at T = 55 K and B = I T. The spectrum ofGaAs is magnified ten times because
the signal is weaker than that of Ga I-x Mn, As (Ando et al. 1998).
III-V FERROMAGNETIC SEMICONDUCTORS 41

studies of (Ga,Mn)As in the reflectivity mode near the Eo critical point, the splitting of the
valence band points to a negative Nof3 value, similarly to the case of II-VI DMS. On the
other hand, a positive value of MCD deduced from absorption measurements appears to
suggest that Nof3 is positive (Szczytko et at. 1996; Shimizu et aI. 1998; Beschoten et at.
1999). As already mentioned, a larger absorption coefficient for the left light polarization
can be reconciled with the negative value of Nof3 by taking into account the Moss-Burstein
shift (Szczytko et at. 1999b; Dietl et at. 200lc). This mechanism operates also on the
insulator side of the MIT, that is in the case of samples with small Mn concentrations,
provided that the intra-center correlation energy U of Mn acceptors is smaller than the
Fermi energy in the impurity band.
In addition to the positive MCD, a negative contribution of magnitude that increases
with decreasing temperature, was observed in ferromagnetic (Ga,Mn)As (Beschoten et at.
1999). This peculiar temperature dependence can be explained in the framework of the
standard model of interband transitions in doped DMS. It results from a non-linear
dependence of MCD on magnetization at the Moss-Burstein edge (Dietl et at. 2001c). In
order to describe the experimental data more quantitatively, a model taking into account the
effect of disorder on the transition probability was also proposed (Szczytko et a1. 200lb).

8. Origin of ferromagnetism

Despite a considerable effort aiming at elucidating the nature of ferromagnetism in


III-V magnetic semiconductors, the form of the relevant minimal Hamiltonian and its
universality for all III-V compounds are still under an active debate. Such a situation
reflects the multifaceted environment, in which the ferromagnetism appears. Indeed,
conceptual and technical difficulties inherent to theory of strongly correlated and
disordered transition metal compounds are combined - in III-V magnetic semiconductors
- with the intricate physics of Anderson-Mott localization that is specific to heavily doped
semiconductors. Moreover, low-temperature epitaxy, by which most of the materials is
obtained, results in a large concentration of native defects such as antisites, which act
as compensating donors. Another possible source of local bond reconstructions is the
mechanism of self-compensation, occurring in heavily doped semiconductors once the
Fermi level reaches the energy triggering defect reactions. Structural faults may form
with neighbor transition metal impurities defect complexes exhibiting hitherto unexplored
magnetic characteristics. At the same time, strong compensation by donor-like defects
enhances the electrostatic disorder substantially, leading to deep and long-range potential
fluctuations that result in significant band tailing.
In this section, we first discuss predictions of ab initio computations in comparison
to photoemission and x-ray magnetic-circular dichroism data, discussed already in
the previous sections. Theoretical models employing parameterized Hamiltonians are
presented next emphasizing their successes and shortcomings in the description of
magnetic phenomena.

8. J. First-principles studies
Ab initio computations of the band structure in zinc-blende ferromagnetic semiconductors
were initiated by Shirai et at. (1998), who employed the full potential augmented plane
42 F. MATSUKURA et al.

wave (FLAPW) method within the local-spin density approximation (LSDA) for MnAs
and for a supercell structure of Gal-xMnxAs with x = 1/2. The total energy calculation
demonstrated that the ferromagnetic ground state has indeed lower energy comparing
to competing magnetic states and led to the correct value of the lattice constant. This
points to the usefulness of LDSA for the determination of the ground state properties
of the compounds in question. Later. the same group (Ogawa et al. 1999) adopted the
linear muffin-tin orbital (LMTO) atomic sphere approximation (ASA) and essentially
reproduced the results for x = I as well as extended the previous computations to ordered
alloys with x = 1/8 and 1/4. It has been noted that the band structure is half-metallic
for the ferromagnetic state, i.e., the Fermi level lies in-between the densities of states
corresponding to the majority spin-down and minority spin-up states. Furthermore, owing
to the p-d hybridization there appears a magnetic moment on the As sites. The As and Mn
moments are antiparallel. which confirms that the coupling between the carriers residing
on the anion orbitals and the Mn spins is antiferromagnetic, Nof3 < O. At the same time.
the magnetic moment within the Mn muffin-tin radius is reduced to about 4/LB. This may
suggest that the Mn ion is in d4 state, i.e., that the holes have largely d-like character. This
is not the case. however. according to the population analysis carried out by Sanvito et al.
(2000. who employed the linear combination of atomic orbitals (LeAO) within LSDA for
ab initio studies ofGat-xMnxAs. The results of Sanvito et al. (2001) imply that the charge
accumulated on the Mn d-shell is about 5.5 in the electron charge units. This demonstrates
that a transfer of electrons from the 3d Mn shell to sp bonds is actually outweighed by a
transfer in the opposite sense. resulting in a partial occupation of the 3d6-like state and a
corresponding reduction of the Mn magnetic moment.
The supercell sizes adopted by Sanvito et al. (200 I) corresponded to a wide range of
Mn concentrations in Gal-xMnxAs. for which Nof3 was evaluated. A surprising result is
that the apparent value of the p-d exchange energy Nof3* , determined from the computed
spin splitting of the valence band at the r point of the Brillouin zone, increases from
-5.5 eV for x = 0.06 to -8.2 eV for x = 0.02. This dependence Sanvito et al. (2001)
interpret in terms of the model put forward by Benoit a la Guillaume et al. (1992). who
considered the case of strong and attractive "chemical" and exchange potentials introduced
by the Mn atoms in II-VI DMS. It has been found that since in such a case the hole
wave function tends to concentrate around the Mn ions. the virtual-crystal and molecular-
field approximations cease to be valid. As a result, a bowing of the energy gap and
an enhancement of the spin splitting is expected, particularly in the range of small Mn
concentrations x. In an extreme case, the Mn can localize a hole, like Cu ions in under-
doped cuprate superconductors (Zhang and Rice 1988). By fitting the computed valence-
band splitting to the model of Benoit a la Guillaume et al. (1992). Sanvito et al. (200 I).
evaluated the bare value of the exchange energy Nof3 = -4.5 ± 0.3 eV and the "chemical"
(spin-independent) part of the total Mn potential to be close to zero. For these parameters,
the holes are not yet localized in the Zhang-Rice states but certainly the local potential
will significantly contribute to the binding energy of the Mn acceptor state. As discussed
in section 4. the experimental values of INof31 are smaller, Nof3 ~ -1 e V.
More recently. Zhao et al. (2001a) employed the FLAPW method within the generalized
gradient approximation (GGA) to investigate further electronic and magnetic properties
of ordered Gal-xMnxAs alloys with low and high Mn concentrations, 0.031 ~ x ~ 0.5.
III-V FERROMAGNETIC SEMICONDUCTORS 43

The determined Mn moment in the muffin tin (MT) sphere at given x was found to be
in a good agreement with the results of Ogawa et al. (1999). This indicates that LSDA
and GGA agree with each other very well in the description of magnetic properties for a
given Gal-xMnxAs structure. At the same time, the occupancy analysis indicates that the
charge accumulated on the d-shell in the MT sphere is 5.2, again greater than 5, although
slightly smaller than that obtained by Sanvito et al. (2001), as quoted above. Furthermore,
the difference in total energies of antiferromagnetic and ferromagnetic ground states,
normalized to one Mn atom, was found to increase from 0.12 eV for x = 0.0625 to 0.24 eV
for x = 0.5. This confirms that enlargement of Mn content in III-V compounds might
constitute a road towards higher Curie temperatures Tc:
The question about the performance of other transition metals in synthesizing functional
ferromagnetic III-V semiconductors was addressed by Shirai (200 1), who extended
FLAPW computation within LSDA to zinc blende arsenides, MAs, where M = Ti, V,
Cr, Mn, Fe, Co and Ni. It was found that the ferromagnetic order is more stable than the
antiferromagnetic state for VAs, CrAs, and MnAs, the greatest energy gain being predicted
for CrAs. Another interesting question concerns ground state properties of hosts other than
GaAs. Kato and Katayama-Yoshida (1999) carried out first principles studies of ordered
Gao.75MnO.25N, Gao.75MnO.25NO.7500.25, and Gao.75Feo.25N alloys within the LAPW-
LSDA scheme. The hole-mediated ferromagnetic ordering is favored in (Ga,Mn)N, as
confirmed in the later work by Sato and Katayama-Yoshida (2001c), but for the remaining
two alloys, in which there are no carriers, an antiferromagnetic ground state is predicted.
Ordered alloys of Gal-xMnxN with several Mn concentrations were studied by Kulatov
et al. (2001) employing the tight binding LMTO-ASA-LSDA approach. The Mn 3d band
was found to reside inside the band gap and to hybridize rather weakly with the band
states. Nevertheless, the ground state appears to be ferromagnetic, and its stability -
in contradiction to results of Kato and Katayama-Yoshida (1999) mentioned above -
increases with n-type doping but decreases with p-type doping.
The first principles studies referred to above have stimulated interesting discussions
about the accuracy of the adopted approximations. In particular, a question arises to what
extend supercell models as well as the LSD and scalar-relativistic approximations are
valid in the case of magnetic alloys in question. Akai (1998), by using the Korringa-
Kohn-Rostoker (KKR) within LSDA approximation and describing chemical disorder in
the cation sublattice by the coherent-potential approximation (CPA), obtained information
about the ground state energy ofthe random Inl-xMnxAs alloy with x = 0.06. Again, the
ferromagnetic phase was found to have lower energy than the local-moment disordered
state. Similarly to the results of the supercell LSDA calculations discussed above, the
ferromagnetic state is half metallic, and the states near the Fermi energy contain a
d-like component. On this ground, Akai (1998) assigns the ferromagnetism to the double
exchange mechanism (Zener 1951b). Another important aspect of Akai's work is the direct
demonstration that ferromagnetism is unstable once holes introduced by the Mn acceptors
become compensated by electrons originating from As antisites or Sn donors. Here, as in
most magnetic materials, antiferromagnetic superexchange dominates in the absence of
the carriers. Surprisingly, however, the Mn magnetic moment increases only slightly as a
function of compensation, from 4.2 to 4.3J.LB. This appears to constitute an additional hint
that the holes reside rather on p than on d shells.
44 F. MATSUKURA et al.

A destructive influence of antisites on the ferromagnetic phase in ordered GaJ-xMnxAs


alloys has been demonstrated also by Sanvito and Hill (200 I). In another work devoted
to effects of disorder, Schulthess and Butler (2001) compare results for a GaJ-xMnxAs
supercell, x = 0.0625, computed employing the Vienna ab initio simulation package
(VASP), with the outcome of CPA computations for an equivalent random alloy with
x = 0.05. According to the VASPdata, the difference in total energies of antiferromagnetic
and ferromagnetic ground states, normalized to one Mn atom, is !:i.E = 0.076 eV. For
unspecified reason, this stability energy is appreciably smaller than the values of 0.15 and
0.12 eV determined for the same Mn content by Sanvito and Hill (2001) and Zhao et al.
(2oo1a), respectively. According to the CPA results, the disorder does not qualitatively
affect the density of states but reduces !:i.E to 0.045 eV (Schulthess and Butler 2001).
Exchange energies of various molecular clusters of Cr, Mn, and Fe embedded in zinc
blende GaAs, GaN, and AIN have been computed within the LSDA by van Schlifgaarde
and Mryasov (2001). According to this work, magnetic energy leads to a strong short-range
attractive force between the magnetic dopants but whether the corresponding clustering is
kinetically allowed at growth temperatures has not been assessed.
In all works referred to above LSDA or GGA have been employed. These approxima-
tions are known (Kotani 2000) to underestimate the on-site correlation effects for 3d elec-
trons in non-metallic solids. This means, according to the Anderson model of magnetic
impurities, that magnitudes of p-d hybridization and related interactions will be overes-
timated. For instance, the values of Neel temperatures in zinc blende Mn chalcogenides,
evaluated within LSDA (Wei and Zunger 1993), are about two times larger than those de-
termined experimentally. This discrepancy is removed in a computation scheme, known as
LSDA+U, in which an external parametric potential is added, whose role is to displace the
d band, hence change the p-d coupling (Wei and Zunger 1993). Figure 23 compares the
partial density-of-states (DOS) for Mn d-electrons in an ordered Gal_xMnxAs alloy with
x = 0.0625 computed by Park et al. (2000) within LMTO-LSDA and LMTO-LSDA+U,
respectively. In agreement with the Parmenter (1973) model, the parametric potential is
characterized by two adjustable parameters, the correlation energy U and Hund's intra-site
exchange energy J. As expected, DOS from the LSDA+U model is sharper and located
further off the Fermi energy than the LSDA d bands. Another interesting aspect of the
theoretical results (Park et aI. 2000) is the observation that the coupling of two Mn spins
changes from ferromagnetic to antiferromagnetic when the distance between the pair in-
creases, a behavior reminiscent of the Ruderman-Kittel-Kasuya-Yosida oscillations. It
would be interesting to check the stability of the ferromagnetic phase and the magnitude
ofNof3 within the LSDA+U model.
It is tempting to compare the computed DOS with photoemission spectra. Certainly,
DOS obtained within the LSDA+U model agrees better with photoemission results
(Okabayashi et al. 1998,1999, 2001a, 2001b), though a direct comparison is hampered by
the influence of electron-hole correlation and multiplet effects on the form of excitation
spectra (Mizokawa and Fujimori 1993). According to the LSDA+U model, the Mn
occupancy increases to 5.4, and the participation of d states in the wave function of holes
near the Fermi energy is reduced compared to the LSDA results (Park et al. 2000). The
occupation number in excess of 5 is directly corroborated by x-ray magnetic circular
dichroism (XMCD) studies (Ohldag et al. 2000; Ueda et aI. 2001), and also, in a model
III- V FERROMAGNETIC SEMICONDUCTORS 45

---
>Q) 4 - - -LSDA
en --LSDA+U
Q)
+-'
ro
--
1i> 2 ,
\

", ,~ l1
en
Q)
+-'
ro J
J "
, I ",", I ,
"" ," .
0

-
1i>
0
>- -2
+-'
'00
c
Q)
0 -4

-8 -4 o 4 8
Energy (eV)
Fig. 23. The spin polarized Mn 3d partial density of states of Gao.93SMno.063As from the LSDA and the
LSDA+U (Park et al. 2(00).

dependent way, by photoemission experiments of Okabayashi et al. (1998, 1999), which


lead to the value 5.3 ± 0.1, as discussed in section 4. At the same time, the local magnetic
moment, which amounts 4.1J1.B in LSDA, increases to 4.4J1.B in LSDA+U (Park et al.
2(00). The latter compares favorably with the value 4.6J1.B determined by fitting XMCD
spectra (Ohldag et al. 2(00). According to results presented in section 5, the saturation
values of the magnetization point to a magnitude of the magnetic moment between 4 and
5J1.B, its more accurate determination being precluded by difficulties in the evaluation of
the Mn content x.

8.2. Parameterized Hamiltonians


Experimental and theoretical results presented in the previous sections make it possible to
sketch a picture of interplay between electronic and magnetic properties in (Ga,Mn)As.
This magnetic semiconductor can be classified as a charge transfer insulator, in which the
high spin state, S = 5/2, is stabilized by positive values of energies required to transfer an
electron either from the Mn 3d shell to the Fermi level or vice versa. Since Mn atoms
are divalent in the S = 5/2 configuration, they act as effective mass acceptors when
they substitute for trivalent Ga. The corresponding energy level should not be confused
with the Mn 3d shells. At the same time, it is important to realize that the quasi-atomic
Woodbury-Ludwig description of magnetic d orbitals is by no means valid in the case
of doping by 3d transition metals. Actually, the position and origin of both band and
3d-like local states are determined - to a large extend - by their strong hybridization
(Zunger 1986). In particular, the hybridization accounts for a large splitting between tZg
and eg components. Moreover, it leads to an admixture of the d 6 configuration to the
46 F. MATSUKURA et al.

wave function of the occupied states visible in XMCD and x-ray photoemission spectra
(XPS), as discussed above. Furthermore, the p-d interaction tends to renormalize the
energy of anion p-type orbitals adjacent to Mn atoms. This will contribute to the "chemical
shift" of Mn acceptors ionization energy in GaAs:Mn, and to the valence band off-set in
(Ga.Mn)As. Importantly. a part of the p-d interaction is spin-dependent. which according
to the Schrieffer-Wolf transformation, leads to a Kondo-like coupling between effective
mass carriers and localized spins (Kacman 200 1).
In view of the above discussion. the one-carrier effective mass Hamiltonian describing
the interaction with a Mn atom located at R, assumes the form,

(6)

where Ho(p) is the k . p Hamiltonian for the crystal structure in question, Vc and v:~
describe the Coulomb and short-range "chemical" part of the potential introduced by the
Mn ion, respectively. and I is the short-range exchange operator. It appears that as long as
magnetic ion occupies the tetrahedral position (so that there is no. e.g. static Jahn-Teller
distortion), the Kondo form of the spin dependent interaction is valid even for magnetic
ions with non-zero orbital momentum (Kacman 200 1). However. the orbital momentum.
and the associated spin-orbit interaction will affect the relation between magnetization M
and expectation value of spin operator S. At the same time. in order to describe correctly
the influence of the exchange interaction on the effective mass states, it is essential to take
the spin-orbit interaction into account in the k . p Hamiltonian Ho(p).
It is convenient to replace. in the spirit of the Kohn-Luttinger effective mass theory
for the valence subbands, short range potentials by appropriate matrix elements Wand fJ
according to W = (XlVsIX) and fJ = (XllIX). where X denotes a py-like component
of the Bloch wave functions for the rg point of the Brillouin zone. The parameter
fJ is the familiar exchange integral. whereas W describes the valence band off set
disregarding long-range Coulomb effects. Such a parameterized Hamiltonian (6) and its
variants constitute a starting point for a number of ferromagnetism models, which will be
outlined in this section. First. however, available information on the values of Wand fJ is
summarized.
As already mentioned. theoretical results of Sanvito et aI. (200 I) lead to NofJ ~ -4.5 eV
and W close to zero. Experimental sources of information on fJ and W are parameters
serving to describe photoemission and XMCD results. Both XPS (Okabayashi et al. 1998).
and resonant photoemission spectra (Okabayashi et al. 1999) can be described with the
same set of charge transfer energies in the cluster model (Mizokawa and Fujimori 1993).
which result in NofJ = -1.2 ±0.2 eV for Gal-xMnxAs with x = 0.074 (Okabayashi et al.
1998). At the same time. the values of the transfer energies inserted into the formula of Hass
(1991) imply that the contribution of the p-d interaction to INoW I is below leV. However.
the application of the model of Mizokawa and Fujimori (1993) to XMCD data leads to
somewhat different values of the charge transfer energies. from which NofJ ~ -0.34 eV
and Wpd > 0 in Gal-xMnxAs with x = 0.025 (Ueda et al. 2001).
Bhattacharjee and Benoit a la Guillaume (2000) adopted the Hamiltonian (6) in order
to describe experimental values of spin-flip and ionization energies of holes localized on
Mn acceptors in the case of weak doping. x < 0.1 %. Their results, obtained by employing
III-V FERROMAGNETIC SEMICONDUCTORS 47

the Baldereschi-Lipari effective mass theory, lead to NoP ~ -0.9 eY. However, for the
assumed shape of the acceptor wave function, the magnitude of INoPI would, presumably,
be about two times smaller when taking a more recent value of the spin flip energy, as
determined by Sapega et aI. (2000, 2001). According to Bhattacharjee and Benoit a la
Guillaume (2000), a positive value of No W having a magnitude larger than INoP I is
necessary in order to explain the "chemical shift" of the Mn acceptor. The positive sign
of W is consistent with the existence of a barrier for the hole injection from (Ga,Mn)As
to GaAs in p-i-n diodes (Arata et aI. 2001). It has been concluded, by comparing
results for three kinds of p-i heterojunctions: (Ga,Mn)AslGaAs, GaAs:Be/GaAs, and
GaAs:Be/(Al,Ga)As that the barrier height in (Ga,Mn)AslGaAs is too large to be entirely
assigned to band gap narrowing by many body effects within the carrier liquid. A barrier
height of 100 meV for x = 0.05, would imply NoW = 2 eV in the virtual crystal
approximation. Such a band off-set is consistent with a reduction of the energy gap in
(Ga,Mn)As comparing to GaAs, noted by Dietl et al. (2oo1c) and Szczytko et al. (2oo1b)
when analyzing optical transmission data of Szczytko et al. (1999b) and Beschoten et al.
(1999). Importantly, assuming a value of NoP = -1.2 eV, a consistent account of the
MCD spectra of Beschoten et al. (1999) is possible within the k . P model of the valence
band (Dietl et aI. 2oo1c). More recently, Szczytko et aI. (2oo1b) by taking the effect of
disorder on selection rules into account obtained an accurate description of their magneto-
transmission results (Szczytko et al. 1999b) with NoP = -1.0 eV. Finally, we recall
that a description of the temperature and field dependence of the resistance in terms of
spin-disorder scattering yielded INoPI of 1.5 ± 0.2 eV (Omiya et al. 2(00), assuming a
simple structure of the valence band and neglecting the q-dependence of the magnetic
susceptibility.
The body of findings presented above demonstrates that owing to the p-d hybridization
there exists an antiferromagnetic coupling between the hole and Mn spins in (Ga,Mn)As.
If described in terms of the Kondo Hamiltonian, the magnitude of the corresponding
exchange energy is, presumably, of the order of -I eV, NoP ~ -1.0 eV. Furthermore, in
the GaAs/(Ga,Mn)As heterojunction, the valence band edge of (Ga,Mn)As resides higher
than in GaAs, No W > 0, so that the potential Vs in eq. (6) is attractive for the holes.
However, to what extent this band offset is controlled by the p-d hybridization and what is
its exact magnitude is unknown at present.

8.3. Hole states and hole mediated exchange interactions


It is now well established that in the absence of free carriers the dominant exchange
mechanism is the superexchange in zinc blende magnetic semiconductors. This mechanism
leads to antiferromagnetic interactions, except perhaps for some Cr-based compounds,
for which a ferromagnetic coupling is theoretically predicted (Blinowski et al. 1996).
Remarkably, owing to the large exchange energy INoPI and the high density of states,
the hole-mediated ferromagnetic exchange interaction can overcome antiferromagnetic
superexchange (Dietl et ai. 1997). Indeed, as already emphasized, the presence of holes
is essential for the existence of the ferromagnetic order in Mn-based semiconductors.
The case of III-V magnetic semiconductors is particularly fortunate as Mn atoms act as
acceptors. It should be recalled at this point that electronic states in doped semiconductors
undergo dramatic changes as a function of the impurity concentration (Belitz and
48 F. MATSUKURA et aI.

Kirkpatrick 1994; Edwards and Rao 1995). Hence, the hole states, and possibly hole-
mediated exchange mechanisms, maya priori undergo dramatic changes as function of
the Mn content x and the concentration of compensating donors, ND in III-V magnetic
semiconductors.
The evolution of electronic states in doped semiconductors is governed by the ratio
of the average distance between the carriers 'c to the effective impurity Bohr radius aB,
determined by both Coulomb and short-range potentials of eq. (6). In the case of the holes
in (Ga,Mn)As, r c = (3j4rrp) 1/3, p = x N« - ND, and aB ~ 0.78 nm (Bhatt and Berciu
2(01). In the range of small impurity concentrations, 'c »aB, the holes are tightly bound
to acceptors. Hence, the conductivity vanishes in the limit of zero temperature. At non-zero
temperatures, the charge transport proceeds either via phonon-assisted hopping between
occupied and empty acceptors or by means of thermal activation from the acceptor levels
to the valence band. In a pioneering work Pashitskii and Ryabchenko (1979) evaluated
the strength of exchange interactions between localized spins mediated by band carriers
thermally activated from impurity levels. More recently, Wolff et al. (1996) considered
carriers localized on impurities and forming bound magnetic polarons (BMP). It was
found that there exists a range of parameters, in which the coupling between the BMP
is ferromagnetic. This idea was further explored by Bhatt and Wan (1999), who examined
by Monte Carlo simulations properties of a ferromagnetic phase transition driven by the
interactions between BMP.
Two other groups noted that a long-range exchange interaction between Mn spins can
be mediated by holes undergoing quantum hoping from the Mn-derived impurity states to
the extended valence band states. Inoue et al. (2000) adopted the Slater-Koster approach,
well known in the physics of resonant states, for the case of two magnetic impurities. It has
been found, by a model calculation, that the pairs of Mn spins coupled to the valence band
states have a lower energy in the ferromagnetic than in the antiferromagnetic configuration.
Litvinov and Dugajev (2001) suggested than the ferromagnetic spin-spin interaction
can originate from virtual excitations between the acceptor-like impurity level and the
valence band, a variant of the Bloembergen-Rowland indirect exchange mechanism. They
evaluated Curie temperatures by using a formula, derived originally for excitations between
valence and conduction bands, without proving its correctness for the case in question.
With the increase of the net acceptor concentration, the impurity band merges with the
valence band. For r; «aB, the holes reside in the band, and their quasi-free propagation is
only occasionally perturbed by scattering of Mn (eq. (6» and other defect potentials, whose
long-range Coulomb part is screened by the carrier liquid. Here, the celebrated Ruderman-
Kittel-Kasuya-Yosida (RKKY) mechanism, driven by intraband virtual excitations, is
expected to dominate. In the context of III-V magnetic semiconductors, this mechanism
was discussed by Gummich and da Cunha Lima (1990) and Matsukura et al. (I 998b). At
the same time, Dietl et al. (1997) demonstrated the equivalence of the RKKY and Zener
(l951a, 1951c) models, at least on the level of the mean-field and continuous medium
approximations. However, with no doubts, beyond those approximations such equivalence
can be questioned (Semenov and Stepanovich 200 I).
Within the Zener approach, and its nuclear spin variant (Frohlich and Nabarro 1940),
the degree of spin ordering, Mq , at given temperature T is found by minimizing the total
free energy of the spin and carrier subsystems, F[Mq ]. Here M q denotes the Fourier
III-V FERROMAGNETIC SEMICONDUCTORS 49

components of localized spin magnetization M(r), so that the minimum of F[Mq ] for
M q=O # 0 implies the ferromagnetic order. In general, however, other ground states,
such as non-collinear structures or spin-density waves, described by M q",O have to be
considered (Dietl et al. 1999). This is a rather versatile scheme, to which carrier correlation
and confinement (Dietl et al. 1997,1999; Haury et al. 1997; Jungwirth et al. 1999; Lee et al.
2000; Fernandez-Rossler and Sham 2001), k- P and spin-orbit couplings (Dietl et al. 2000,
2oo1c; Abolfath et al. 2001; Femandez-Rossier and Sham 2001) as well as disorder and
antiferromagnetic interactions (Dietl et al. 1997; Kossacki et al. 2000) can be introduced
in a controlled way, and within which a quantitative comparison of experimental and
theoretical results is possible (Dietl et al. 2oo1a, 2oo1c; Ferrand et al. 2001).
In view of the above discussion the question arises whether the hole-mediated
ferromagnetism appears in the insulator or in the metallic phase. It is well established
that the metal-insulator transition (MIT) occurs at re ~ 2.4aB in doped non-magnetic
semiconductors (Edwards and Sienko 1978). According to this criterion one gets the
critical hole concentration Pc = 4 X 10 19 cm- 3 for aB = 0.78 nm. Experimentally, the
MIT occurs at about 3.5% of Mn in (Ga,Mn)As, i.e., for Nox = 7 x 1020 cm- 3 (Oiwa
et al. 1997; Matsukura et al. 1998b; Katsumoto et al. 1998). A large difference between
these two values is presumably caused by the compensation (discussed above) as well
as by the enhancement of localization by the sp-d exchange scattering (Dietl 1994). The
latter is documented in (Ga,Mn)As by the presence of negative magnetoresistance and
associated insulator-to-metal transition driven by the magnetic field (Katsumoto et al.
1998). In addition to the MIT at x ~ 0.035, the reentrant insulator phase is observed
for x > 0.06 (Matsukura et al. 1998b), as discussed in section 6. Presumably, a self-
compensation mechanism is involved but no microscopic model has been proposed so
far.
Perhaps, the most intriguing property of the materials in question is that the ferromag-
netism is observed on the both sides of MIT (Oiwa et al. 1997; Matsukura et al. 1998b).
It is, therefore, interesting to contemplate the nature of electronic states in the vicinity of
the MIT in doped semiconductors. Obviously, the random spatial distribution of acceptor
and donor centers gives rise to strong spatial fluctuations in the carrier density and states
characteristics. According to the phenomenological two-fluid model there exist two kinds
of relevant states (Paalanen and Bhatt 1991). The first are strongly localized and thus singly
occupied states associated with the attractive potential of a single majority impurity. The
strongly localized carriers barely contribute to the conduction process. However, they pro-
duce a Curie-like component in the magnetic susceptibility and give rise to the presence
of BMP in magnetic semiconductors. Obviously, the impurity-like states dominate deeply
in the insulating phase but their presence is noticeable also in the metallic phase (Paalanen
and Bhatt 1991; Glod et al. 1994). The second pool of states determines the conductivity, so
that properties of these states are described by the scaling theory of MIT. Accordingly, the
corresponding localization radius ~ is rather controlled by interference of multi-scattering
processes than by the attractive potential of a single impurity. Thus, ~ of these weakly lo-
calized states is significantly larger than aB, and diverges on approaching the MIT from
the insulator side. It is worth noting that such a two-fluid model is consistent with a.c. con-
ductivity studies (Nagai et al. 2001), which show the coexistence of weakly and strongly
localized states near the MIT in (Ga,Mn)As. Furthermore, the merging of impurity and
50 F. MATSUKURA et aI.

band states in this range is substantiated by angle-resolved photoemission spectra in the


same system (Okabayashi et al. 2001a, 2oolb).
In order to tell the dominant mechanism accounting for the existence of long-range
spin order in ferromagnetic semiconductors it is instructive to trace the evolution of their
magnetic properties on crossing the MIT. Remarkably, in contrast to rather strong changes
of resistivity, the evolution of magnetic properties is gradual. This substantiates the notion
that thermodynamic properties do not exhibit any critical behavior at MIT as they are
insensitive to large-scale characteristics of the wave functions. Importantly, the values
of the Curie temperature are found to grow with the degree of the material metallicity
(Matsukura et al. 1998b; Katsumoto et al. 200 1; Potashnik et al. 200 1). Moreover,
the examination of the magnetization as a function of temperature and magnetic field
indicates that virtually all Mn spins contribute to ferromagnetic order in the most metallic
samples (Oiwa et al. 1997; Matsukura et a1. 1998b; Potashnik et al. 2(01). However,
on crossing the MIT (by lowering x), the relative concentration of ferromagnetic ally
coupled spins decreases substantially. According to XMCD results of Ohldag et al.
(2000), about 10% of the Mn spins is involved in ferromagnetism of Gal-xMnxAs
with x = 2%. Also ferromagnetic resonance studies (Szczytko et al. 1999b) and direct
magnetization measurements demonstrate that only a part of the spins contribute to
spontaneous magnetization, while the alignment process of the remaining moments occurs
according to a Brillouin function for a weakly interacting spin system (Oiwa et al. 1997).
Remarkably, the anomalous Hall effect reveals clearly the presence of the first component
but hardly points to the existence of any loose spins (Matsukura et al. 1998b).
The above findings indicate that Mn spins in the regions visited by itinerant holes are
coupled ferromagnetically. These holes set long-range ferromagnetic correlations between
the Mn spins, including those contributing to BMP that are formed around singly occupied
local states. Obviously, the ferromagnetic portion of the material, and thus the magnitude of
the spontaneous magnetization, grows with the dopant concentration, attaining 100% in the
metallic phase. Such a trend is confirmed by the available data, as discussed above. Thus,
the delocalized or weakly localized holes are responsible for ferromagnetic correlation
in (Ga,Mn)As (Dietl et al. 2(00). At the same time, mechanisms that involve strongly
localized states, such as excitations from impurity levels or a direct coupling between BMP,
appear to be of lesser importance.

8.4. Mean-field Zener model and its application to (Ga,Mn)As


In this section, theoretical foundations and application of the mean-field Zener model to
III-V magnetic semiconductors are discussed in some detail. The capabilities of the model
to describe various magnetic properties of (Ga,Mn)As are presented, too. In the final part,
limitations of the model and its numerous refinements put recently forward are discussed.
As already mentioned, it is convenient to apply the Zener model (Zener 1951a, 1951c)
by introducing the Ginzburg-Landau functional F[M(r») of the free energy density (Ma
1976), where M (r) denotes local magnetization of the Mn spins. The choice of M (r) as
the order parameter means that the spins are treated as classical vectors, and that spatial
disorder inherent to magnetic alloys is neglected. In the case of magnetic semiconductors
F[M(r») consists of two terms, F[M(r») = Fs[M(r») + FclM(r»), which describe, for
a given magnetization profile M(r), the free energy densities of the Mn spins in the
III-V FERROMAGNETIC SEMICONDUCTORS 51

absence of any carriers and of the carriers in the presence of the Mn spins, respectively
(Leroux-Hugon 1973; Dietl 1994). A visible asymmetry in the treatment of the carries
and of the spins corresponds to an adiabatic approximation, the dynamics of the spins in
the absence of the carriers being assumed to be much slower than that of the carriers.
Furthermore, in the spirit of the virtual-crystal model molecular-field approximations,
the classical continuous field M(r) controls the effect of the spins upon the carriers.
Now, the system thermodynamics is described by the partition function Z, which can be
obtained by a functional integral of the Boltzmann factor exp( - f dr F[ M (r)]/ kBT) over
all magnetization profiles. In the mean-field approximation (MFA), a term corresponding
to the minimum of F[M(r)] is assumed to determine Z with a sufficient accuracy.
If energetics is dominated by a spatially uniform magnetization M, the spin part of the
free energy density in the magnetic field H can be written in the form

(7)

Here, h(Mo) denotes the inverse function to Mo(h), where M o is the macroscopic
magnetization of the spins in the absence of carriers at a field h and temperature T. It
is usually possible to parameterize Mo(h) by the Brillouin function Bs according to

(8)

where two empirical parameters, the effective spin concentration XeffNO < xNo and the
temperature Teff > T, take the presence of the short-range superexchange antiferromag-
netic interactions into account (Gaj et al. 1979). The dependencies Xeff(X) and TAF(X) are
known for II-VI DMS compounds. However, as argued by Dietl et al. (2000, 2001c), the
antiferromagnetic short-range interaction is overcompensated by the ferromagnetic double
exchange coupling in the case of III-V DMS. According to the two fluids model intro-
duced in section 8.3, a part of the carriers is trapped on strongly localized impurity states,
and thus forms BMP. To gain the Coulomb energy, the singly occupied local states are pref-
erentially formed around close pairs of ionized acceptors. In the case of III-V materials,
one hole localized at two Mn ions generates, via Zener's (Zener 1951b) double exchange,
a strong ferromagnetic coupling that overcompensates the intrinsic antiferromagnetic in-
teraction (Blinowski et al. 1997). Accordingly, Xeff ~ x and Teff ~ T. By contrast, in II-VI
compounds in which acceptor cores do not carry any spin, and the degree of compensa-
tion is low, BMP are not preferentially formed around Mn pairs, so that the close pairs
remain antiferromagnetically aligned. The presence of a competition between the ferro-
magnetic and antiferromagnetic interactions in p-type II-VI DMS, and its absence in Mn-
based III-V materials, constitutes the important difference between those two families of
magnetic semiconductors.
It is clear from eqs (7) and (8) that Fs[M] monotonously increases with IMI, so that-
as expected - the minimum of Fs[M] corresponds to M = 0, for which the spin entropy
attains the highest value. It is convenient to introduce the spin susceptibility is related to
52 F. MATSUKURA et a1.

INoIi1 = 1.2 eV
0.0 LX~-_-~0.
r
;.;.0;.;;5_~"-~~_~ _ _--1

:>
Q)

>-
~ -0.2
Q)
c
W

-0.4

-2 -1 0 2
- - k (10 7 cm') _ _
k-LM kllM

Fig. 24. The computed valence band dispersion E(k) computed from the 6 x 6 Luttinger model for the wave
vector parallel and perpendicular to the Mn spin magnetization in (Ga,Mn)As. assuming that the spin splitting of
the heavy-hole band at the r point is 0.15 e V.

the magnetic susceptibility XO of the spins according to is = (gf..LB)2 XO. In the limit, where
Mo(T, h) = Xo(T)h,
(9)

which shows that the increase of Fs with M slows down with lowering temperature. In
contrast to Fs[Ml, owing to Zeeman splitting of the bands imposed by the sp-d exchange
interaction, the energy of the carriers, and thus Fc£Ml, decreases with IMI. Accordingly, a
minimum of F[Ml at non-zero M may develop in H = 0 at sufficiently low temperatures.
In order to take into account the complex structure of the valence band, Dietl et al.
(2000, 2oo1c) and Abolfath et al. (2001) have computed hole energies by diagonalizing
the 6 x 6 k . p Luttinger matrix together with the p-d exchange contribution taken in the
virtual crystal and molecular field approximation,

Hpd = {3sM/gf..LB. (10)

This tenn leads to spin splittings of the valence subbands, whose magnitudes - owing
to the spin-orbit coupling - depends on the hole wave vectors in a complex way
even for spatially uniform magnetization, as shown in fig. 24. It would be technically
difficult to incorporate such effects into the RKKY model, as the spin-orbit coupling
leads to non-scalar terms in the spin-spin Hamiltonian. At the same time, the indirect
exchange associated with the virtual spin excitations between the valence subbands, the
Bloembergen-Rowland mechanism (Dietl 1994; Kacman 200 1), is automatically included.
The model allows for biaxial strain (Dietl et al. 2000, 2oo1c; Abolfath et al. 2001),
III-V FERROMAGNETIC SEMICONDUCTORS 53

confinement (Fernandez-Rossier and Sham 2001), and is developed for both zinc blende
and wurtzite materials (Dietl et al. 2000, 2oolc). Furthermore, Dietl et aI. (2oo1c) take into
consideration the direct influence of the magnetic field on the hole spectrum. The carrier-
carrier spin correlation is described by introducing a Fenni-liquid-like parameter AF,
which enlarges the Pauli susceptibility of the hole liquid (Altshuler and Aronov 1985; Dietl
et al. 1997). No disorder effects are taken into account on the grounds that their influence
on thermodynamic properties is relatively weak.
Having the hole energies, the Helmholtz free energy density FC£M] can be evaluated
according to the standard procedure for the Fermi gas. By minimizing F[M] = Fs[M] +
Fc[M] with respect to M at given T, H, and hole concentration p, one obtains M(T, H)
as a solution of the mean-field equation,

where peculiarities of the valence band structure, such as the presence of various hole
subbands, anisotropy, and spin-orbit coupling, are hidden in FC£M]. Near the Curie
temperature Tc and at H = 0, where M is small, we expect FC£M] - FC£O] ,..., M 2 • It is
convenient to parameterize this dependence by a generalized carrier spin susceptibility
Xc, which is related to the magnetic susceptibility of the carrier liquid according to
X = AF(g* ILB)2XC. In terms of Xc,

(12)

By expanding Bs(M) one arrives at the well-known form for the mean-field value of Tc
(Dietl et al. 1997; Jungwirth et al. 1999)

(13)

For a strongly degenerate carrier liquid lSFI/ kBT » 1, as well as neglecting the spin-
orbit interaction, Xc = p/4, where p is the total density-of-states for intra-band charge
excitations, which in the 3D case is given by p = mOoskF/rr2h2. In general, however, Xc
has to be determined numerically by computing FdM). Large magnitudes of both density
of states and exchange integral specific to the valence band make Tc much higher in p-type
than in n-type materials with a comparable carrier concentration.
The above reasoning can easily be generalized to the case of a phase transition to
a spatially modulated ground state, characterized by non-zero magnetization M q • The
corresponding mean-field value of the ordering temperature Tc(q) is given by the solution
of the equation (Dietl et al. 1999)

(14)

where the carrier spin susceptibility can be determined from the standard linear-response
expression,

(15)
54 F. MATSUKURA et aI.

Here uitl
), fin), and Ei tl
) are the periodic part of the Bloch function, energy and

Fermi-Dirac distribution functions for the n-th carrier spin subband. In the case of cubic
symmetry, the susceptibility tensor is isotropic, X~ij) = Xe8;j. It has been checked within
the 4 x 4 Luttinger model that the values of Tc, determined from eqs (13) and (12), which
do not involve explicitly -t: and from eqs (14) and (15) in the limit q ~ 0, are identical
(Ferrand et al. 200 1). Such a comparison demonstrates that almost 30% of the contribution
to Te originates from interband polarization, i.e. from virtual transitions between heavy
and light hole subbands.
It is possible to extend the above approach to the case of low dimensional structures
(Dietl et al. 1997, 1999; Haury et al. 1997; Lee et al. 2000; Femandez-Rossier and
Sham 200 1). If the carriers occupy one electric subband, the mean-field value of ordering
temperature Te(q) is given by the formula that generalizes eq. (13) (Dietl et al. 1999)

(16)

where qJo(~) is the envelope function of the occupied subband in the confining potential, ~
and q are vectors in the 3 - d and d dimensional space, respectively.
We now tum to the ferromagnetic phase, T < Te. Here, in addition to M(T, H),
the evaluation of FdM] makes it possible to determine the energy density of magnetic
anisotropy. K (Dietl et al. 2000. 2oo1c; Abolfath et al. 2001) as well as the hole spin
polarization P = 2gJ1.B(aFdM)/aM)/({3p) and magnetic moment Me = -aFdM]/aH
(Dietl et al. 2oolc). Another important characteristic of any ferromagnetic system is the
magnetic stiffness A. which describes the energy penalty associated with a local twisting
of the direction of magnetization. Actually, in the experimentally important case of a
uniaxial ferromagnet. the energy functional is entirely described by K u , A, and the value
of magnetization M according to,

(17)

where ii (i) is the unit vector that specifies the local Mn spin orientation. and 9 is its angle
with respect to the easy axis. The latter is controlled by biaxial strain in epilayers (Ohno
et al. 1996b; Shen et al. 1997a. 1997b, 1997c. 1997d) and by confinement in quantum wells
(Dietl et al. 1997; Haury et al. 1997). Konig et al. (2000, 2001) have developed a theory
of magnetic stiffness A in III-V ferromagnetic semiconductors. Remarkably, A determines
the magnitude and character of thermodynamic fluctuations of magnetization. the width
and energy of domain walls as well as the spectrum of spin excitations. In particular, the
quantized energies of long-wavelength spin waves are given by

(18)

It is clear from eq. (17) that A describes how the Ginzburg-Landau functional F[8Mq ]
varies with q. Here, 8Mq are the Fourier components of the difference between local and
macroscopic magnetization M(T, H). In the long-wave limit. in which eq. (17) is valid,
III-V FERROMAGNETIC SEMICONDUCTORS 55

the magnitude of A is expected to be primarily determined by the magnetic stiffness of the


carrier subsystem, that is by Fd8Mq] and thus by X~ii)(q). Indeed, the distance between
the spins is smaller that that between the carriers, and the intrinsic spin-spin interactions
are short-range. It is seen, by comparing eqs (12) and (17), that X~ii) (q, T) for the direction
i perpendicular to M(T, H) is relevant, and its numerical evaluation from eq. (15) for a
given M(T, H) will provide A(T, H). By parametrizing X~(q) = X~(q = 0) - Cl.q2 one
gets

(19)

As long as the valence band splitting t::. = AFfJM/gJLB is much smaller than the Fermi
energy one expects the spin susceptibility, and thus C to be independent of M, and isotropic
for the cubic symmetry Cl. ~ clI' An important observation of Konig et al. (2001) is that
the magnetic stiffness computed within the 6 x 6 Luttinger model is almost by a factor of
10 greater than that expected for a simple doubly degenerate band with the heave-hole
band-edge mass. This enhancement, which stabilizes strongly the ferromagnetic order,
stems presumably from p-like symmetry of the valence band wave functions as well as
from interband q-dependent polarization.

8.5. Comparison oftheoretical and experimental results


In this section, the mean-field Zener model discussed above is employed to describe
experimental values of Curie temperature, spontaneous magnetization, anisotropy field,
and domain stripe width. Standard values of band-structure parameters, elastic constants,
and share deformation potentials of GaAs are adopted (Abolfath et aI. 2001; Dietl et al.
2oolc). The Mn spins are assumed to be in the d 5 configuration, so that S = 5/2 and
the Mn Lande factor g = 2.0. For the pod exchange energy NofJ = -1.2 eV is taken
(Okayabashi et al. 1998), which for the cation concentration of GaAs, No = 2.21 x
1022 cm- 3 , corresponds to Jpd == -fJ = 0.054 eVnm 3. The Fermi liquid parameter AF =
1.2 (Jungwirth et al. 1999) enters the enhancement of Te and of the valence band pod
exchange splitting t::. = AFfJM/(gJLB) at magnetization M of the Mn spins (Dietl et al.
200Ic).
The most interesting property of Gal-xMnxAs epilayers is the large magnitude of Te,
up to 110 K for the Mn concentration x = 5.3% (Ohno et al. 1996b; Matsukura et al.
1998b). Because of this high Te, the spin-dependent extraordinary contribution to the
Hall resistance RH persists up to 300 K, making an accurate determination of the hole
density difficult (Oiwa et al. 1997; Van Esch et aI. 1997; Matsukura et aI. 1998b; Shimizu
et al. 1999). However, the recent measurement (Omiya et al. 2000) of RH up to 27 T
and at 50 mK yielded an unambiguous value of p = 3.5 x 1020 cm- 3 for the metallic
Gao.947MnO.053As sample, in which Tc =110 K is observed (Matsukura et aI. 1998b).
The above value of p is about three times smaller than x No, confirming the importance of
compensation in Gal_xMnxAs.
As shown in fig. 25, the numerical results lead to Tc = 120 K for x = 0.05, and thus,
Tc: = 128 K for x = 0.053 and p = 3.5 x 1020 cm- 3. It seems therefore that the mean-field
Zener model, with no adjustable parameters, can explain the high values of Tc found in
S6 F. MATSUKURA et al.

Ga'_xMnxAs
x =0.05
-
Q'100
l,aNa l = 1.2 eV
AF =1.2 mOOS

.. '
..............................
10

1 .....-"'----....................."---.......-.--.......~ .......- .......................


0.01 0.1 1
20
Hole Concentration (10 ern")
Fig.2S. Curie temperature as a function of the hole concentration for Gao.9SMnO.osAs computed from the 6 x 6
Luttinger model (solid line). Straight dashed lines represent results obtained assuming large and small values
of the spin-orbit splitting !J.o , respectively. The dotted line is calculated neglecting the effect of the spin-orbit
interaction on the hole spin susceptibility (Dietl et al. 200lc).

Gal-xMnxAs. Furthermore, the scaling theory of electronic states near the MIT, discussed
in the previous sections, makes it possible to explain the presence of the ferromagnetism
on the both sides of the MIT, and a non-critical evolution of Te across the critical point
(Matsukura et a1. 1998b). A comparison between theoretical and experimental data in a
wider range of Mn and hole concentrations requires reliable information on the hole density
in particular samples, which is not presently available. In appears, however, that in the case
of both Gal-xMnxAs and Inl_xMnxAs on the insulator side of the MIT, the experimental
values of Te are systematically higher than those expected from the Zener model.
Turning to the temperature dependence of the spontaneous magnetization in the
ferromagnetic phase one should note that the total magnetic moment consists of spin and
hole contributions. The hole part was found (Dietl et a1. 2oo1c) to be negative and to attain
only a few percent of the total magnetization values, as the hole polarization is incomplete,
the hole concentration is smaller than that of the spins, and because of a partial cancellation
between the Pauli and Landau terms. If the hole liquid is only partially spin polarized (i.e.
161 < IEFD, which is usually the case in Gal_xMnxAs, M(T) is expected to grow at T ~ 0
according to the Brillouin function (Dietl et a1. 2oo1c), in agreement with the experimental
results (Matsukura et a1. 1998b), shown in fig. 13. For lower hole concentrations or higher
Mn content, M(T) will tend to its saturation value M; somewhat slower (Dietl et a1. 2oo1c).
This, together with a lowering of M (T) by spin wave excitations, may account for the
dependence M(T)/ M s = 1- AT 3/ 2 , detected experimentally for T ~ 0 (Potashnik et a1.
2001).
Both hydrostatic and axial strain affect the valence band, and thus alter the magnitude of
the density of states and Tc : Quantitatively, however, the effect is evaluated to be small
III-V FERROMAGNETIC SEMICONDUCTORS 57

(Dietl et al. 2001c). There exists another mechanism by which strain may affect Tc·
It is presently well known that the upper limit of the achievable carrier concentration
is controlled by pinning of the Fermi level by impurity or defect states in virtually all
compound semiconductors. Since the energies of such states in respect to bands vary
strongly with the bond length, the hole concentration and thus Tc will depend on strain.
Apart from Tc and Ms , it is interesting to consider means making it possible to tailor
magnetic anisotropy, and thus the direction of the spontaneous magnetization, the coercive
force, the switching field, the domain structure. Already early studies of the ferromagnetic
phase in Int-xMnxAs (Munekata et al. 1993) and Gat-xMnxAs (Ohno et al. 1996b;
Shen et al. I997a) demonstrated the existence of a sizable magnetic anisotropy. Magnetic
anisotropy is usually associated with the interaction between spin and orbital degrees of
freedom of the d-electrons, According to the model advocated here, these electrons are
in the d 5 configuration. For such a case the orbital momentum L = 0, so that effects
stemming from the spin-orbit coupling are expected to be rather weak. It has, however,
been noted that the interaction between the localized spins is mediated by the holes that
have a non-zero orbital momentum (Dietl et al. 2(00). An important aspect of the Zener
model is that it does take into account the anisotropy of the carrier-mediated exchange
interaction associated with the spin-orbit coupling in the host material (Dietl et al. 2000,
2001c; Abolfath et al. 2(01), an effect difficult to include within the standard approach to
the RKKY interaction.
A detail numerical analysis of anisotropy energies has been carried out for a number
of experimentally important cases (Dietl et al. 2000, 2001c; Abolfath et al. 2(01). In
particular, the cubic anisotropy as well as uniaxial anisotropy under biaxial strain have
been studied as a function of the hole concentration p. The computation indicates that for
the parameters of Gal-xMnxAs films grown along the [001] direction, the spontaneous
magnetization M lies in the (001) plane, and the easy axis is directed along the [100]
or along the [110] (or equivalent) crystal axis depending on the degree of the occupation
the hole subbands as well as on their mixing by the k . p interactions. As a result, the
easy axis fluctuates between [100] and [110] as a function of p, the preferred direction
for typical hole concentrations being [110]. The magnitude of the external magnetic field
Heu that aligns M along the hard direction in the (001) plane is evaluated to be up to
0.2 T (Dietl et al. 200 Ic). However, the orientation of the easy axis changes rapidly with p
and M. Therefore disorder, which leads to broadening of hole subbands, will presumably
diminish the actual magnitude of magnetic anisotropy. The field /-Lo H eu determines also the
magnitude of the switching field, which could be observed in microstructures containing
only a single domain. In macroscopic films, however, smaller values of the coercive field
/-LoHe are expected as actually observed: typically /-LoHe = 4 mT for the magnetic field
along the easy axis in the (001) plane in Gat-xMnxAs (Shen et al. 1997a).
It can be expected that strain engineering can efficiently control magnetic properties
resulting from the hole-mediated exchange. Indeed, sizable lattice-mismatch driven by
biaxial strain is known to exist in semiconductor layers. In some cases, particularly if
epitaxy occurs at appropriately low temperatures, such strain can persist even beyond the
critical thickness due to relatively high barriers for the formation of misfit dislocations. It
has been found that the biaxial strain leads to uniaxial anisotropy, whose magnitude can
be much greater than that resulting from either cubic anisotropy or stray fields. As shown
58 F. MATSUKURA et al.

1.2
x =0.05
.......
-
I-
c:
::x:::>
1.0

0.8
IfJNol =1.2 eV
AF =1.2
e
"U 20
(1) 0.6 3.5x10 em"
u:: /
>.
a. 0.4
......
0
0
0 0.2
U)
'2
-c
0.0
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
Biaxial Strain exx (%)
Fig. 26. Computed minimum value of the magnetic field Hun necessary to align the saturation value of
magnetization M s along the hard axis as a function of biaxial strain component €x. for two values of the hole
concentrations in Gao.95Mno.05As. The symbol [100]-> [0011 means that the easy axis is along [100]. so that
Hun is applied along [(01) (Dietl et al. 200lc).

in fig. 26 for the experimentally relevant values of p and M. the easy axis is predicted
to be oriented along the [00 I] direction for tensile strain, whereas it should reside in the
(001) plane for the case of unstrained or compressively strained films (Dietl et al. 2000,
200lc; Abolfath et al. 2(01). This is corroborated by the experimental study (Ohno et al.
I996b), in which either (In,Ga)As or GaAs substrate was employed to impose tensile or
compressive strain in Gai-xMnxAs, respectively. In particular, for the Gao.965Mno.035As
film on GaAs, for which t:x x = -0.24%, the anisotropy field /-LoHun = 0.4 ± 0.1 T is
observed (Ohno et al. 1996b; Shen et al. I997a, I997b, 1997c, 1997d), in quantitative
agreement with the theoretical results of fig. 26. This field is about two orders of magnitude
greater than that evaluated from the extrapolation of ESR data on single-ion anisotropy
at low x (Fedorych et al. 200 I), a result confirming the dominant contribution of the
holes to the magnitude of Hun. Though no theoretical computations have been performed
for Inl_xMnxAs, a qualitatively similar effect of biaxial strain is expected, in agreement
with results of Munekata et al. (1993), who tailored the easy axis direction by employing
(AI,Ga)Sb or AI(Sb,As) substrates with various Ga contents or As contents.
Recently, the structure of magnetic domains in Gal_xMnxAs under tensile strain has
been determined by micro-Hall probe imaging (Shono et al. 20(0). The regions with
magnetization oriented along the [00 I] and [00 i) easy axis form alternating stripes
extending in the [110] direction. This indicates, for either Bloch or Neel domain walls,
that the in-plane easy axis is rather along [110] than along [100] directions, a conclusion
consistent with the theoretical expectation for in-plane (cubic) magnetic anisotropy
presented above. As shown in fig. 27, the experimentally determined stripe width is
III-V FERROMAGNETIC SEMICONDUCTORS 59

.
10 r-----r--..--"""T"-.......
• --,----T""""'--.,
• .

-E - - - ---"
·
--
::1.
.c
:2
6 •
_........ - _ --
....
-- --
"
"
,-

......•.
-- ----
~
c:
'm
E
4~
• ·
o ..........................................rfJ
o 2 ·
."
0'--_ _......._ ....._ . & -......_ .
......._ .
....._ . & -...... -'

0.0 0.2 0.4 0.6 0.8 1.0


Relative Temperature Tffc
Fig. 27. Temperature dependence of the width of domain stripes as measured by SOOno et al. (2000) for the
Gao.957Mll().043As film with the easy axis along the growth direction (fun squares). Computed domain width is
shown by the solid line. The dashed line is computed assuming that the parameter Ac (eq. (20» is by a factor of
1.8 greater (Dietl et al. 200la).

W = 1.5 JLm at 5 K for a 0.2 JLm film of Gao.957MnO.043As on Gao.SdnO.16As, for which
a tensile strain of cxx = 0.9% is expected. According to the micromagnetic theory, W is
determined by the dimensionless parameter Ac, which is given by the ratio of the domain
wall and stray field energies,

(20)

where d is the film thickness.


Figure 27 presents values of W(T) calculated by Dietl et aI. (200la) in comparison
to the experimental data of Shono et al. (2000). Furthermore, in order to establish the
sensitivity of the theoretical results to the parameter values, the results calculated for a
value of Ac that is 1.8 times larger are included as well. The computed value for low
temperatures, W = I. I JLm, compares favorably with the experimental finding, W =
1.5 JLm. However, the model predicts a much weaker temperature dependence of W than
observed experimentally, which Dietl et al. (200la) link to critical fluctuations, disregarded
in the mean-field approach.

8.6. Limitations and refinements ofthe mean-field Zener model


It is obvious that the role of randomness in the distribution of Mn acceptors and extrinsic
defects, which is neglected in the model in question, will grow with approaching the
localized regime. According to the two fluid model discussed previously, there is a
60 F. MATSUKURA et aI.

coexistence of weakly localized carriers and bound magnetic polarons in this regime. In
this context particularly interesting are results of Bhatt and Berciu (200 I), whose starting
point is the impurity band of holes localized by the Mn acceptors. Interestingly, according
to their numerical mean-field treatment, positional disorder enhances Te. A further study
along the same lines (Kennett et al. 200 1) supports the two fluid picture, demonstrating
the presence of "strongly" and "weakly" coupled spins. It worth noting that the itinerant
carriers may set long-range ferromagnetic correlation between magnetic polarons. Since
Te is proportional to the square of the relevant spin vectors, the weight of the BMP
contribution may greatly exceed their relative concentration. The latter, together with the
disorder enhancement of Te mentioned above, may account for higher values of Tc on the
insulator side of the MIT than those expected from the Zener model.
Another important issue requiring further studies is the role of carrier-carrier correlation.
It is known that the effect of disorder on carrier-carrier interactions controls the
localization and enhances the spin susceptibility (Altshuler and Aronov 1985), and thus
the tendency towards ferromagnetism. However, spin-disorder scattering may limit the
efficiency of this process (Altshuler and Aronov 1985). If this is the case, LSDA (Jungwirth
et al. 1999; Lee et al. 2(00) can provide a reasonable evaluation of the relevant Fermi-liquid
parameter.
There are three other experimentally important situations, for which the mean-field
Zener model, as introduced above, ceases to be valid. The first one corresponds to the case
when an average time of carrier tunneling between typical Mn pairs (Vx~~3)-1 becomes
1
significantly longer than the inverse exchange energy INoPI- • Here V is the width of the
carrier band, and its magnitude, not the Fermi energy as sometimes suggested, constitutes
the relevant energy scale. For long tunneling times, the virtual-crystal approximations
break down, an effect detected in Cdl-xMnxS (Benoit a la Guillaume et al. 1992). A
modified double-exchange model will constitute an appropriate description of the carrier-
mediated exchange interaction in the strong coupling limit V < INoPI, where the holes are
bound in Zhang-Rice (Zhang and Rice 1988) states, and only occasionally hop between
the Mn ions. Here, a strong sensitivity of Te to the concentration of compensating donors
is expected. Dynamic mean-field theory, as developed for III-V magnetic semiconductors
by Chattopadhyay et al. (2001), can constitute an appropriate approach in this regime.
Another regime, in which the mean-field Zener model may cease to be valid, is that
of large carrier concentrations n > XeffNO. In the limit when the continuous-medium
approximation is obeyed, n « x No, the mean-field value of the ordering temperature
T(q) deduced from the RKKY model are identical, independently of microscopic spin
arrangement. If, however, n > XeffNO, important changes in the carrier response function
occur at the length scale of a mean distance between the localized spins. Accordingly,
the description of spin magnetization by the continuous-medium approximation, which
constitutes the basis of the Zener model, ceases to be valid. In contrast, the RKKY model
is a good starting point in this regime (Ferrand et al. 200 1), as it provides the dependence
of the interaction energy of particular spin pairs as a function of their distance. This makes
it possible to evaluate the system energy for a given distribution of the localized spins.
Here, randomness associated with the competition of ferromagnetic and antiferromagnetic
interactions can drive the system towards a spin-glass phase (Eggenkamp et al. 1995). In
the extreme case, n »XeffNO, the Kondo effect that is dynamic screening of the localized
II1-V FERROMAGNETIC SEMICONDUcrORS 61

spins by the sea of the carriers may preclude both ferromagnetic and spin-glass magnetic
ordering.
Finally, the accuracy ofthe mean field approximation (MFA) ought to be addressed. It is
well known that the MFA results are exact if the range of spin-spin interactions is infinite
(Fisher et al. 1972). The decay of the strength of the carrier-mediated exchange interaction
with the distance r between two Mn spins is described by the RKKY function. At small r,
the interaction is ferromagnetic, and then changes sign at r = 1.2rc , where rc is an average
distance between the carriers that mediate spin-spin coupling. This means that the MFA is
valid quantitatively at n «xeffNo, a conclusion consistent with the estimate of Tc taking
the spin wave excitations into account (Konig et al. 2000). Actually, however, the range
of validity of the MFA is significantly larger than that initially found (Konig et al. 2(00),
as the magnitudes of spin stiffness evaluated within the 6 x 6 Luttinger model are much
greater (Konig et al. 2001) than those obtained for a simple parabolic band (Konig et a1.
2000).
Recently, Monte-Carlo studies of carrier-mediated ferromagnetism in semiconductors
have been initiated (Sakai et al. 2001; Sakai and Suzuki 2001; Bosselli et al. 2000;
Schliemann et al. 200la, 200lb). Such an approach has a potential to test the accuracy
of the approximations mentioned above and to determine the actual spin configuration
corresponding to the ground state. Preliminary results appear to confirm the validity of the
MFA (Sakai et al. 200 I; Bosselli et al. 2(00), and indicate a possibility of the existence
of non-collinear magnetic structures in low-dimensional systems (Bosselli et al. 2(00).
More recent comprehensive simulations of Schliemann et al. (200 Ib) carried out within
the hybrid Monte Carlo scheme, identify the parameter space, in which the mean-field
Zener model may break down. It would be interesting to check separately the regions of
validity of particular approximations involved: MFA, virtual crystal approximation (VCA),
and molecular-field approximation, as well as to elucidate the role of electrostatic disorder.

9. Heterostructures

New physics such as the fractional quantum Hall effect has emerged from non-magnetic
semiconductor heterostructures. These systems have also been a test bench for a number
of new device concepts, among which are quantum well lasers and high electron mobility
transistors. Ferromagnetic III-Vs can add a new dimension to the III-V heterostructure
systems because they can introduce magnetic cooperative phenomena that were not present
in the conventional Ill-V materials.

9.1. Basic properties of heterostructures


9.1.1. Structural properties ofmultilayer structures
Figure 28 shows a typical double-crystal x-ray diffraction pattern of the (004) reflection
of a GaAsI(Ga,Mn)As superlattice grown on GaAs (001) substrate, taken by employing
CuKa( radiation (Shen et al. 1997d). The sample consists of 20 periods of nominally
11.4-nm thick GaAs and 12.l-nm thick (Ga,Mn)As with x = 0.054. Satellite peaks
witnessing the periodicity of the structure are clearly visible. Theoretical simulations of
the rocking curve were carried out by adding background noise of 10 cps. The values of
62 F. MATSUKURA et al.

d~I=11.79 nm - - experimental
d(Ga.Mn~1=11. 74 n - - - simulated
xMn=0.056

10'

10° ......WooI. ...._ .....................WooI. ...._ ..........- . .


-1500 -1000 -500 0 500 1000 1500
Relative Angle (arcsec)
Fig. 28. X-ray diffraction rocking curve of (Ga,Mn)AslGaAs superlattices with 20 periods (solid line). The
nominal thicknesses of the GaAs and (Ga,Mn)As layers and the Mn composition x are 11.4 nrn, 12.1 nm,
and 0.054, respectively. The dashed line shows the simulated rocking curve. The fit to the experimental curve
is obtained with GaAs and (Ga,Mn)As thicknesses and x of 11.14 nm, 11.79 nm, and 0.056, respectively (Shen
et al. 1997d).

the elastic constants and the Debye-Waller factor determined for GaAs were adopted for
zinc-blende MnAs. The best fit shown in fig. 28 reveals that x = 0.056 as well as that the
thicknesses of the GaAs and (Ga,Mn)As layers are 11.4 nm and 11.70 nrn, respectively,
in good agreement with the nominal values. The fact that almost the same line width is
obtained for the experimental and the simulated satellite peaks suggests a high quality of
the films and the interfaces. The high perfection of the structure is also confirmed by the
observation of thickness fringes due to perfect crystal diffraction from thin films.

9.1.2. Magnetic properties ofsuperlattices, thin films. and quantum wells


Transport measurements on multilayer structures demonstrate that the ferromagnetism
can be retained down to a (Ga,Mn)As layer thickness of 5 nm, below which the structure
becomes paramagnetic. On the other hand, there was a report on the persistence of
ferromagnetic properties in short-period superlattices consisting of m (Ga,Mn)As and n
GaAs monolayers, with 8:::; m :::; 12 and 4 :::; n :::; 8 (Sadowski et al. 200le). The reason
for these diverging conclusions is not clearly understood; it may be related to the Mn
segregation at the initial stage of growth, which leads to a depletion of Mn and/or the
distribution of the spin-polarized carrier (Louriero da Silva et al. 2001; Vurgaftman and
Meyer 2(01). Multiple (Ga,Mn)As quantum wells (QWs) were also fabricated and studied
by means of magnetization, magneto-optical and magneto-transport phenomena (Hayashi
et a1. 1997b, 1998). No ferromagnetic order in (Ga,Mn)As QWs with thickness less than
5 nm is observed. MCD results show clear evidence of the quantum confinement and the
formation of subband.
III-V FERROMAGNETIC SEMICONDUCTORS 63

(Gs o965MnO 035)AslL T-GaAs 10 K


1 J.lm 180 nm
~
r
_ .~ ..~ .., :-_-;.IO~~---'I
•• -

,. : , 10 nm
• . I
I .
. ' I
..
I' I

....u
--- --
0
0 500
n
0.0 0.5 1.0
B (T)
Fig. 29. Thickness dependence of the ratio of Hall resistance and sheet resistance RHall/ Rsheet, which is
proportional to magnetization perpendicular to the film plane, as a function of the magnetic field at 10 K. The
inset shows the thickness dependence of TC (Matsukura et aI. 1998a).

Figure 29 shows the thickness dependence of magnetization in (Ga,Mn)As films, as


determined by the Hall effect (note that RHaU/ Rsheet oc M) (Matsukura et al. 1998a). The
inset presents the values of Tc in these films. Apart from an increase in anisotropy and
a gradual growth of Tc, no significant changes are observed when the film thickness
is reduced from I /Lm to 10 nm. Below 5 nm, however, the (Ga,Mn)As films become
insulating. At the same time, Tc drops significantly, often below the lowest studied
temperature of 2 K. Similarly, the ferromagnetism disappears in (Ga,Mn)As QWs, if
their width is below 5 nm (Hayashi et al. 1997b). No systematic differences in the
behavior of superlattices, thin films or quantum wells have been detected. In contrast, the
ferromagnetism was observed in GaAs samples to which a submonolayer of MnAs was
inserted (Kawakami et al. 2000). Further studies are apparently necessary to understand
the thickness dependence of magnetic properties in these materials.

9.1.3. Band offset between (Ga,Mn)As and GaAs


A heterojunction is primarily characterized by the band alignment. This most fundamen-
tal property of any heterojunction is not well established in the case of Mn-doped III-V's.
The difficulty in determining the band offset arises from a number of reasons, such as a
high doping level in Mn-doped I1I-V's and a relatively small values of Mn concentrations,
which results in a tiny magnitude of the band offset. The close examination of current-
voltage (I-V) characteristics of p-i-n and p-p diodes made of (Ga,Mn)As and GaAs shows
that thermoionic emission at temperatures higher than Tc is the dominant mechanism of
the current transport (Ohno et al. 2000; Ohno et al. 2001; Arata et al. 2001). By ana-
lyzing the temperature dependence of the I-V characteristics, one can deduce the barrier
64 F. MATSUKURA et aI.

200 2

>

.
<l>
E
• E

- -•
"::l
.I::.
0)
'0) 100
i-GaAs v

1 ~


.I::.
....
<l>
.;::
);'

....
CO
.0 0
0
00 -
0

0 0
0.00 0.04 0.08
X

Fig. 30. Barrier height [). measured by current-voltage (I-V) characteristics of (Ga,Mn)As/GaAs diodes. [).
shown by closed circles is the barrier height between the Fermi energy of (Ga.Mn)As and the valence band top
of GaAs as shown in the inset. Open circles depict the effective Richardson constants. (Ohno et aI. 2001).

height between (Ga,Mn)As and GaAs measured from the Fermi level of (Ga,Mn)As. Re-
sults of such an analysis are depicted in fig. 30 (Ohno et al. 200 I), which presents the mea-
sured barrier height t:.. as a function of x , together with the associated effective Richardson
constant (A* / A). Inset shows the band structure of the measured samples. The findings
demonstrate that the holes flowing from the (Ga,Mn)As side have to overcome a barrier
of about 100 meV, the value being virtually independent of x. However, the interpreta-
tion of this number and the determination of the bare offset t:.. E v between (Ga.Mn )As and
GaAs valence band edges is by no means straightforward. In particular, the Fermi energy
of (Ga,Mn)As (usually of the order of 100 meV) has to be taken into account, which will
result in t:..Ev :::::: 200 meV. At the same time, however, band gap renormalization caused
by the hole-hole interactions, the hole coupling to the system of ionized impurities, and
the impurity band formation will shift the (Ga,Mn)As valence band edge towards higher
energies, reducing t:..Ev .

9.2. Spin-dependent scattering, interlayer coupling. and tunnel magnetoresistance in


trilayer structures
Ferromagnetlnonmagnetlferromagnet trilayer structures constitute the most fundamental
building block of modem magnetic sensors and storage elements, and are useful for
the examination of various magnetotransport processes. Especially important is spin-
dependent scattering in such structures, as this is the basic process that gives rise to the
effect of giant magnetoresistance (GMR). Equally important is the effect of tunneling
magnetoresistance (TMR) in structures containing a thin insulator as the nonmagnetic
layer. In order to investigate the nature of spin-dependent transport in systems made
of semiconductors only, various (Ga,Mn)As/(AI,Ga)As/(Ga,Mn)As structures have been
III-V FERROMAGNETIC SEMICONDUcrORS 65

0.3 r - -....T""',.......-r---....,......_.., 2.50


..
-- -- .
0.2

!
ct 0.1

00 1-A...._I.0001....._I.0001..................... ~ 2.48
~ -4 o 4 8
B (mT)
Fig. 31. Hall resistance RHall (circles) and sheet resistance Rsheel (triangles) versus magnetic field B at 25 K for
a (Gao.9sMno.os)As/(Alo.14Gao.86)As/(Gao.97Mno.03)As trilayer structure. Closed and open symbols show the
major and minor loops, respectively. Dashed arrows indicate sweep directions of the magnetic field. The minor
loop of RHall is skewed by the presence of a ferromagnetic coupling between the two (Ga,Mn)As layers (Chiba
et al. 2000).

prepared and studied (Akiba et al. 1998a, 1998b, 2000a; Hayashi et al. 1999; Chiba et al.
2000; Higo et al. 2001a, 2001b).
Figure 31 presents the field dependence of the Hall resistance and the magnetoresistance
at 25 K for a Hall bar pattern of a trilayer structure (Chiba et al. 2(00). The device consists
of two 30 nm Gal-xMnxAs layers separated by a 2.8 nm nonmagnetic Alo.14Gao.86As
layer, acting as a barrier for holes, whose height depends on the AI concentration. The
layers are grown onto a 50 nm Alo.30Gao.70As buffer and a I ILm relaxed InO.15Gao.85As
film. The Mn content x = 0.05 and 0.03 ofthe two (Ga,Mn)As layers results in a difference
of their coercive forces. The (Ino./SGao.8S)As film introduces a tensile strain, which makes
the magnetic easy axis to be perpendicular to the structure plane. With this easy axis
direction, the Hall effect can be used to monitor magnetization.
Parallel transport in trilayer structures is characterized by the MR ratio (Rsheet - Ro)/ Ro,
where Ro is the layer resistance in the absence of the external magnetic field, Rsheet(B = 0),
and for parallel magnetizations M of the two (Ga,Mn)As layers. A plateau seen in the Hall
resistance data collected in fig. 31, indicates that magnetizations of the two (Ga,Mn)As
layers are anti-parallel (note that RHall is proportional to a weighted average of M of the
two (Ga,Mn)As layers). A clear increase of the MR ratio is observed in the plateau region
between 3 and 6 mT. This demonstrates the presence of spin-dependent scattering in the
trilayer structures made of semiconductors only. The MR ratio decreases with the increase
of AI composition in the barrier, which reduces the number of holes traveling across it.
In the minor loop measurements, when the direction of magnetization of one of
the two layers is fixed, the hysteresis loops of RHall and the MR ratio show a good
correspondence (open symbols in fig. 31). This confirms that the observed MR effect
66 F. MATSUKURA et 31.

60
(a)
20K
30
--E
l-

-
~
0

-30 -,
,,
-...
6

--
-
0~

0:::0 3
(b)

~
:\
+
.%
#:
• !
0
0
T (K)

- 5
44

43

-
~ I
+
t
42
--6'
::tl

-
0::: 0

41
-50 0 50 100
B (mT)
Fig. 32. (a) Magnetization M and (b) tunneling magnetoresistance of 30 nm (Gao.95MnO.os)As/2.8 nm
AIAs/30 nm (Gao.97MnO.03)As tunnel junction at 20 K. Inset shows the temperature dependence of tunneling
magnetoresistance ratio (Chiba et 31. 2000).

is indeed due to spin-dependent scattering. The minor loop is shifted away from B =
0, which indicates that there is a ferromagnetic interlayer coupling between the two
(Ga,Mn)As layers. The magnitude of this coupling estimated from this shift is quite
small « I l.d/m 2 ) . The coupling is always ferromagnetic. although theory predicts an
antiferromagnetic interaction under certain sets of parameters (Jungwirth et al. 1999).
The magnitude of the coupling increases with the lowering of the Al composition in
the (AI.Ga)As barrier. This suggests that the interlayer coupling is mediated by holes.
Elastic neutron scattering and polarized neutron reflectivity measurements on short-period
(Ga.Mn)AslGaAs superlattices show also the presence of a ferromagnetic interlayer
coupling (Szuszkiewicz et al. 200 I; Kepa et al. 200 I). An interlayer coupling between
two (In,Mn)As layers separated by InAs has also been observed (Yanagi et al. 2(02).
Vertical transport measurements of trilayer structures reveal the presence of TMR.
Figure 32a shows findings obtained at 20 K for a device containing a 2.8 nm AlAs barrier
(Chiba et al. 2000). The structure was grown on a (In,Ga)As film to fix the magnetization
direction perpendicular to the plane. For TMR measurements. the electrodes were formed
on the top and bottom (Ga.Mn)As layers. The temperature dependence of the MR ratio
is shown in the inset to fig. 32. The difference in the coercive forces produces a plateau
visible in fig. 32a, in the region, where magnetizations of the two ferromagnetic layers
IIl-V FERROMAGNETIC SEMICONDUcrORS 67

-
N
E
0
25 (a)
8K -
minor loop
~
II . 60
-a 4> Ir ~
'I
I,
40
-l
3:
"
-~
I ::0
0 20 II
c II
JJ
'ii)
II
II
20
4>
0::: ~ , I
I I .o
15
-
-20 -10 0 10 20
B 1/ [100] (mT)
100 I I I

(b)

/---\
81/[100] 8K

-
~
0

0
:0::
50
~
0:::
~
~

-
/-----,.,
81/[110]"

--I
0 .. ~-.-
1.4 1.6 1.8 2.0 2.2
Barrier thickness (nm)
Fig. 33. (a) TMRcurves at 8 Kof a Gaj __ xMnxAs(x =4.0%.50 nm)/AiAs (1.6 nm)/Gal_xMnxAs (x = 3.3%.
50 nm) tunnel junction 200 J./-m in diameter. Bold solid and dashed curves were obtained by sweeping the
magnetic field from positive to negative and negative to positive. respectively. A minor loop is shown by a thin
solid curve. The magnetic field was applied along the (100) axis in the plane. (b) Barrier thickness dependence of
TMR values with the magnetic field applied along the [100] and [110] axes at 8 K (Tanaka and Higo 2001).

are antiparallel. A resistance increase is observed between 8 mT and 16 mT, in the field
region corresponding to an antiparallel configuration of magnetizations. The MR ratio
is about 5.5% at 20 K. This is TMR, because the barrier produced by AlAs is high
(0.55 eV), so that the hole transport across the AlAs layer proceeds by tunneling. The
TMR ratio decreases with temperature most probably due to a corresponding drop of
spontaneous magnetization in (Ga,Mn)As layers. A TMR ratio over 70% was observed in
a (Ga,Mn)AsIAIAs/(Ga,Mn)As structure with an AlAs thickness of 5 monolayers (1.6 nm)
as shown in fig. 33. Such a high value indicates that spin polarization of carriers in
(Ga,Mn)As is quite high (Higo et al. 2oo1a, 2oo1b; Tanaka and Higo 2001). A theoretical
calculation ofTMR with ferromagnetic DMS electrodes shows how the TMR ratio depends
on the Mn and hole concentrations (Lyu and Moon 2001).

9.3. Resonant tunneling diodes (RTDs)


Spontaneous magnetization in ferromagnetic semiconductors gives rise to spin splitting of
the conduction and valence bands due to the presence of exchange interaction. This spin
68 F. MATSUKURA et aI.

•••• HH

---- LH2
•••• HH3

GaAs
•... HH
(Ga.Mn)As GaAs .... LH1
I, GaAs:Be
•••• HH11.-......... _

AlAs AlAs
I GaAs 5 nm

o v
Fig. 34. Schematic valence band diagram of resonant tunneling diode structures. simplified diagram of energy
versus wave vector parallel to the interface. and resulting 1-V curve by spin-splitting of the valence band of
(Ga,Mn)As emitter.

splitting can be observed in current-voltage (l- V) characteristics of resonant-tunneling


diodes (RTDs) having a ferromagnetic emitter as shown in fig. 34. Nonmagnetic double
barrier AIAslGaAslAIAs RIDs with a (Ga,Mn)As emitter reveal spontaneous splitting
of resonant peaks below Tc of (Ga,Mn)As, even in the absence of an external magnetic
field (Ohno et al. 1998; Akiba et al. 1998a, 1998b). Figure 35a shows the temperature
dependence of dl /d V versus V of an RTD having a (Ga,Mn)As emitter. Clear spontaneous
splitting of resonant peaks labeled HH2 and LHI is observed below Tc of 60 K. The
structure in question was grown onto a p+ GaAs substrate and consists of several layers
(from top): 150 nm (Gao.965MnO.035)As/15 nm undoped GaAs spacer/5 nm undoped
AlAs barrier/5 nm undoped GaAs QW/5 nm undoped AlAs barrier/5 nm undoped GaAs
spacer/150 nm Be doped GaAs (p = 5 x 10 17 cm- 3)/150 nm Be doped GaAs (p =
5 x 10 18 cm"). Each label in fig. 35 indicates the resonant state in the GaAs well, where
in total six states occur.
Akiba et al. (2000b) performed a calculation of the corresponding dl /dV-V character-
istics, taking into account a strong k-dependent mixing and the presence of the exchange
III-V FERROMAGNETIC SEMICONDUCTORS 69

(a) T (I<)
- - - - - - -.......---~ .....-...-......... 300
- ........--...------":j250

2 3 4 5
M 1/(110), x = 0.05,
(b) p = 1x1Uo em",compressive strain 1 %

H1 /\ HH3 lI.t2 .•.... NrfJ(eV)


..·....lH1 / ,'·.~H2.~.~
.. .,",.\..
..··P.....'"_../_\.
, '"~\ '".HH4 -. 0
..... , ,' ~\.".:. - " -0.2
.... , .'\ ... ~ ~_

.... -,. .,', \' .-.. ., .• -.---


- , ..--
-0.5
...... ,r=.\;11- "-. ' •• _..... 7
-._ . •' " ....._._.-._. ._.-0.
.
.. . . . .
IIIIlIIooo
..
oo. -1-.
.. \
__ .... _ ~ .... ~ \. " _ _ _ _ _
. , ..... - -1.0
-1.2
~ ""---
," ,.".,----""'---..f -1 .5

0.0 0.2 0.4 0.6 0.8 1.0 1.2


V(V)
Fig. 35. Temperature dependence of the differential conductance dJjdV versus bias voltage V of a resonant
tunneling diode with a (Ga,Mn)As emitter. No magnetic field is applied (Ohno et al, 1998). (b) Calculated
resonant tunneling spectra as a function of the exchange energy Nof3 (Akiba et aI. 2000b).

interaction. Previous theoretical approaches to p-type RTD's (Chao and Chuang 1991; Ro-
drigues Bittemcourt et at. 1998) were supplemented by incorporating the p-d exchange
interaction into the 6 x 6 Luttinger-Kohn Hamiltonian. The resonant states in the GaAs
well were calculated by using the 4 x 4 Hamiltonian. The current density Jz for a given
hole concentration and voltage was then calculated as,

4
Jz=~
(2Jr) .
L'/_1
ff'(J E(EF
d 3k.(aE).T*Tij'[J(E)-f(E-eV)],ciI.
ak z
I,J, -
(21)
70 F. MATSUKURA et al.

Here. T* Tij is the transmission coefficient from the heavy or light hole state in the
emitter to the heavy and light hole states in the collector. f(E) the Fermi-Dirac
distribution function. Cit is the probability amplitude of the jth hole state in the lth band.
Figure 35b shows the calculated dl/dV-V characteristics with JNoPI as a parameter.
which corresponds to various values of magnetization in the present case. A I nm AlAs
barrier is assumed to avoid numerical instability. A 5% compressive strain is included in
the (Ga.Mn)As (x = 0.05) layer containing I x 1020 holes per em". The magnetization
is assumed to be in plane along the [110] direction. As can be seen in fig. 35b. the HH2
resonant peaks shows a clear splitting as NoP (magnetization) increases. whereas other
peaks (except probably LHl) do not show such pronounced splitting. the conclusions being
in accord with the findings. The splitting of the HH2 peak is so pronounced because the
dispersion of the HH2 resonance is similar to that of the dominant valence band state in
the emitter. Other resonant states either show a very different dispersion and/or lie high
in energy. where transmission maxima become broad. Note that no clear cut-off of the
current is observed neither in the experiment nor in theory because of a rather high hole
concentration. The present results indicate that because of a complex interplay between the
k . p and exchange interactions. it is necessary to go through a rather elaborate calculation to
understand the origin of the peaks and their splitting. Similar calculations were performed
for different structures with a DMS QW and adopting Keldysh formalism (Petukhov et al.
2000; Inoue et al. 2000; Nonoyama and Inoue 2001; Kuivalainen and Hovinen 2002).

9.4. Spin-injection inferromagnetic semiconductor heterostructures


Spontaneous magnetization in ferromagnetic semiconductors introduces an imbalance in
the spin population of the carries. Thus. these materials can be used as a source of
spin polarized carriers. which can be injected into non-magnetic structures. In spite of
the common belief that very fast hole spin relaxation precludes the effect. hole spin
injection from (Ga.Mn)As into GaAs was demonstrated (Ohno et al. 1999). Figure 36a
shows a schematic diagram of a light emitting diode employed to demonstrate the spin
injection. Partly spin polarized holes are injected from a p-type (Ga.MnjAs layer through
an intrinsic GaAs spacer into an (In.GajAs QW. where they recombine with unpolarized
electrons injected from nonmagnetic n-type GaAs substrate. The spin polarization of
the recombining holes and hence spin injection is demonstrated by the observation of
electroluminescence (EL) polarization. Figure 36b shows the relative polarization change
!l P as a function of the magnetic field at various temperatures. Two polarization states
are clearly observed at B = 0 at low temperatures. A series of experiments was done to
exclude effects of both fringing fields and magnetic circular dichroism from the adjacent
(Ga.Mn)As. Because the easy direction is in-plane for (Ga.Mn)As grown on GaAs. EL
was collected along the direction parallel to the QW plane. Since the selection rule at the
very bottom of the band in QW does not allow emission of circularly polarized light in this
direction. the effect of band filling (Le. nonzero k) needs to be taken into account to fully
understand the results.
Injection of electron spins is preferable from the application point of view as electrons
usually exhibit longer spin lifetime. Spin-injection experiments using n-type II-VI DMS
in the magnetic field were performed by Fiederling et al. (1999) and Jonker et al. (2000).
An appealing scheme for electrical electron spin injection from a ferromagnetic material
III-V FERROMAGNETIC SEMICONDUcrORS 71

(a)

GaAs spacer (il

(In,Ga)As (i)
GaAs (i)
GaAs spacer (i)I--------i
GaAs substrate (n)b:::::::::!!j!Zl:::::::::::::J

(b) --
"#. E = 1.34eV
d= 140 nrn , r- I.
1

..
~

.~
c:
0
cu
1= 2.8 rnA
--+-sweepup
-It-sweepdown
- . - 6K
-e- 16 K
"

~oof.

CU
0Q. -£-31 K
0 ... 5 2 K "
.5
\I)
C)
c:
CU
~ily' "./
;.",:,..___",d-"

--
'i1-
-
~

~

..•.
c:
-fi .;: :~.,. ~ 0..
-
~

\I) :~ . <3 if
.....:::cu -1 :. - :: ' -t~~\
,=-,_.rr .
I
0
100
~
-15 -10 -5 0 5 10 15
B (mT)
Fig. 36. (a) The structure of a light emitting diode for detection of electrical spin-injection. (b) Relative change
in light polarization t>.P as a function of the magnetic field at four different temperatures. Inset compares the
remanent magnetization as determined by SQUID magnetomery with the remanent t>.P (Ohno et al. 1999).

is the use of a broken gap heterojunction system such as InAsiGaSb, in which the valence
band edge of GaSb is energetically higher than the conduction band edge of InAs. The
successful growth of ferromagnetic (Ga,Mn)Sb should make it possible to inject spin
polarized electrons into nonmagnetic InAs (Abe et al. 2(00). Another way is to use an Esaki
tunnel diode, a method successfully developed by Koda et al. (2001) and by Johnston-
Halperin et al. (2002), who employed p-(Ga,Mn)Asln-GaAs diodes.
By solving the drift-diffusion equation for a GaAs-based pn junction, it has been shown
theoretically that the spin-polarization in the p region can be transferred into the n region
with high efficiency through the depletion layer (Zutic et al. 200 I). In this structure the
spin polarization in the n region can be controlled by the external applied electronic bias;
i.e., an increase of forward bias results in a decrease of the spin polarization.
72 F. MATSUKURA et aI.

9.5. Photo-inducedferromagnetism in (In,Mn)As/GaSb


The relationship between Tc and the hole concentration indicates the possibility of control
of magnetic properties isothermally by light irradiation, electric field, carrier injection, and
all other means that change the carrier concentration in semiconductors. The concept of
such devices was proposed already in 1960s in the context of work on rare-earth magnetic
semiconductors (Methfessel 1965; Methfessel and Holtzberg 1966).
(In,Mn)As (12 nm, x = 0.06)/GaSb heterostructure was found to exhibit photo-induced
ferromagnetism as shown in fig. 37 (Koshihara et al. 1997). The illumination by photons
with the energy larger than the bandgap of GaSb (""' 0.8 eV) turns the paramagnetic
sample without remanent magnetization to a ferromagnetic state with a clear hysteresis
at 5 K. which is documented by both magnetization and magnetotransport measurements.

3
X = 0.06
.... 2
'"'
0.2 T
Eci
1 0°,"
EF 0
~ °00
Ev 00 10 20 30 40 50
hole
(In,Mn)As GaSb T(K)
8

....-
(c)
4 5K
N
b 0
-
~
X

~ -4
-8

-....'i 1

- ';;
0

et -1 - - •before irradiation
- - after irradiation
-2 L - _......__......---JL....- '---.....J

-0.3 0.0 0.3


Fig. 37. Band edge profile of a (In,Mn)AslGaSb heterostructure. EC. Ev , and EF denote band edges of
conduction band. valence band, and Fermi level, respectively. (b) Temperature dependence of the magnetization
observed during cooldown in the dark (open circles) and warmup (solid circles) under a fixed magnetic field
of 0.02 T. The effect of light irradiation at 5 K is also shown by an arrow. (c) Magnetization curves at 5 K
observed before (open circles) and after (solid circles) light irradiation. Solid line shows a theoretical curve.
(d) Hall resistivity PH.II observed at 5 K before (dashed line) and after (solid line) light irradiation (Koshihara
et aI. 1997).
III-V FERROMAGNETIC SEMICONDUcrORS 73

The ferromagnetic state persists even after switching the light off and, at the same time,
a persistent conductivity is observed. If the sample is heated up to '" 45 K, the initial
state is recovered. It appears that the interface electric field separates photo-holes and
photo-electrons, the former being accumulated in the (In,Mn)As layer, which triggers
the ferromagnetism. According to the Hall measurements up to 15 T, a critical hole-
concentration that generates the ferromagnetic order is '" 3.8 x 1019 cm- 3 , whereas
the change of the hole concentration after illumination is as low as 1.4 x 1018 cm- 3.
Further studies are necessary in order to clarify the reason why there is a threshold
hole concentration and why only a slight increase of the hole concentration leads to
such a dramatic effect. A similar persistent light-enhanced magnetization is observed
in (lno.9sMno.os)(Aso.sSbo.2)lInSb heterostructure (Kanamura et al. 2(02). Reflecting
the small band-gap of InSb, a lower photon energy is more efficient than that for
the (In,Mn)As case. On the other hand, it is difficult to observe irradiation effects in
(lno.9sMno.os)(Aso.2Sbo.s)lInSb, which may be related to the small built-in electric field in
the structure.
Ferromagnetic (In,Mn)AslGaSb heterostructures with rectangular hysteresis show also
peculiar light-irradiation effects. In particular, the coercive force is drastically reduced by
the illumination, which suggests a reduction of the domain wall pining energy (Oiwa et aI.
2(01).
A magnetization enhancement of (Ga,Mn)As by circularly polarized light illumination
has been also observed (Oiwa et aI. 2(02).
These effects open up the possibility of novel magneto-optical memory devices.

9.6. Electric-field control offerromagnetism in gated structures


The modification of ferromagnetism was demonstrated also in a structure consisting of an
insulating-gate field-effect transistor (FET) with an (In,Mn)As channel (Ohno et aI. 2(00).
The 5-nm thick channel layer (x = 0.03) was grown on a 10 nm InAsl500 nm (Al,Ga)Sb
buffer and a GaAs substrate. A 0.8 /Lm gate insulator and a metal electrode completed the
device. The hole concentration in the channel was estimated to be 5-8 x 1013 cm- 2 from
the resistance changes with the gate voltage and Hall effect at room temperature. Figure 38
shows magnetization deduced from the Hall resistance at 22.5 K for three different values
of the gate voltage VG, +125, 0, and -125 V. The bias of VG = 125 V changes the
sheet-hole concentration by '" 3 x 1012 cm- 2. At zero gate bias, the channel is weakly
ferromagnetic as can be seen from the presence of a small hysteresis. The application of a
positive gate voltage depletes the channel and, thus, reduces the ferromagnetic interaction
mediated by the holes, which results in a paramagnetic behavior of the magnetization
without hysteresis. When the holes are accumulated by applying a negative gate voltage,
a clear hysteresis appears. The magnetization curve resumes its original values as the gate
voltage returns to 0 V. The 125 V swing give rise to ±6% change in the hole concentration
and results in a Tc change of ±4% (±l K).
The value of Tc calculated form the mean-field theory for three-dimensional (3D) case
(Dietl et aI. 2000, 2001c) is '" 18 K for (In,Mn)As with x = 0.03 and the hole density
1 x 1020 cm- 3 , which corresponds to p = 5 x 1013 cm- 3 in a 5 nm-thick film. For
this calculation the Luttinger-Kohn parameters given in by Wiley (1975), the energy of
the exchange interaction between the holes and Mn spins NofJ = -0.98 eV (Dietl et aI.
74 F. MATSUKURA et al.

0.04 ...:- ....


r 1.5K

( ~(
1
5K
(

.~o, .. .:::t'.... ,
~
." ~.Q~
. 20
0.02 ~
• ...-t""':'

-c
~
-1
~_L

fJ ~ ... ,....
-'.~ .
- ""'---.... -
~

-
~

iii 0.00
-0.5 0.0
B (T)

... J"-' I.,.'


O.
.~ .......
a::
J: ,,~
", / '
,
6"':.,..
. ,-,,:..-
. ....
22.5 K
VG
-0.02 .~- OV .
-- +125 V
) --125V

-0.04
-1.0
~--
-0.5 0.0
_.- OV

0.5 1.0
B (mT)
Fig. 38. Hall resistance RHali of an insulated gate (In,Mn)As field-effect transistor at 22.5 K as a function of
the magnetic field for three different gate voltages. RHali is proportional to the magnetization of the (In,Mn)As
channel. Upper right inset shows the temperature dependence of RHali' Left inset shows schematically the gate
voltage control of the hole concentration and the corresponding change of the magnetic phase (Ohno et al. 2(00).

200lc), as well as the enhancement factor AF = 1.2 describing carrier-carrier interactions


(Jungwirth et a1. 1999) are adopted. The calculation also shows that a modulation of
I::!..p = 3 X 1012 cm- 2 gives a change of Tc of r - I K. These evaluations show. therefore.
a rather good correspondence with the experimental findings. On the other hand, the
thickness of the (In,Mn)As channel in the examined FET structures may suggest that the
system is two-dimensional (2D). For 2D systems. the mean-field theory predicts no carrier
concentration dependence of Tt; as long as the 2D density-of-states (DOS) that determines
Tc is energy-independent (Dietl et a1. 1997). For the hole concentrations in question,
several 2D subbands are occupied and the valence band DOS is a complex function of
energy, so that we expect a modulation of Tc through a modulation of p. Moreover, since
the mean free path is comparable to the quantum well width. the disorder-induced mixing
of electric subbands turns DOS towards the 3D value (Kossacki et a1. 2000).
An additional contribution to the modulation of Tc may come from the particular
design of the FET device: a nonmagnetic InAs is placed beneath the magnetic (In,Mn)As
layer. Application of positive (negative) bias displaces the hole wavefunctions away from
(towards) the magnetic layer, resulting in a reduction (increase) of the interaction between
the holes and magnetic spins and hence in a reduction (increase) of Tc (Lee et a1. 2000).
Control of ferromagnetism in gated structure has also been observed in a group VI
ferromagnetic semiconductor Mn.Ge I-x (Park et a1. 2(02).
III-V FERROMAGNETIC SEMICONDUCTORS 75

9.7. Ferromagnetic imprinting


The manipulation of spin coherence of electron spins and nuclear spins in semiconductor
materials is one of the most important issues to pursue in spintronics for spin-based qubits
in future quantum computing. A proximity effect on spin coherence in nonmagnetic semi-
conductors in ferromagnet/semiconductor heterostructures has been observed (Kawakami
et a1. 200 1). The spin dynamics in GaAs on which a ferromagnetic layer, (Ga,Mn)As or
MnAs, is grown, is investigated using time resolved Faraday rotation (TRFR). The results
at 5 K show that the presence of a ferromagnetic layer strongly modifies the spin coherence
of electrons. The dependence of TRFR on applied magnetic fields shows hysteresis behav-
ior, and the effective field extracted from the data suggests that there is a nuclei-mediated
effective field. The results derived from all-optical nuclear magnetic resonance (Kikkawa
and Awshalom 2(00) on the samples also support the existence of the "imprinting" on
nuclear spins by the neighboring ferromagnetic layer.

10. Ferromagnetic semiconductors at room temperature

For the application of magnetic semiconductors, a Tc above room temperature is required.


Room-temperature ferromagnetic sulphospinels have been found in 1960's (Goodenough
1969; Van Stapele 1982), but the application of these ferromagnetic semiconductors
has been hampered by difficulties in their fabrication. Recent experimental as well
as theoretical progress in material science allows one to design and synthesize new
ferromagnetic semiconductors with Tc above room temperature.

10.J. Theoretical suggestions

The mean-field Zener model described in section 8 predicts Tc to lie above room
temperature for DMS containing large amounts of magnetic ions and carriers. Thus, the
elaboration of methods enabling a simultaneous increase of the magnetic ion and carrier
concentrations in DMS is one of the ways to be pursued (Dietl et a1. 2000, 2001c).
Furthermore, a chemical trend was theoretically established, suggesting that the highest
values of Tc can be achieved in materials containing light anions (Dietl et ai. 2000,
2001c). The tendency for higher Tc values in the case of lighter elements stems from
the corresponding increase in the p-d hybridization and the reduction of the spin-orbit
coupling. It can be expected that this tendency is not altered by the uncertainties in the
values of the parameters employed for the computation. In particular, if one could introduce
5% ofMn and 3.5 x 1020 cm- 3 of holes into wide-gap semiconductors, such as GaN, ZnO,
and C, these materials should be ferromagnetic at room temperature. The first-principle
calculation also predicts a rather stable ferromagnetism for these materials. The results
show that V, Cr, Fe, Co, or Ni doped ZnO is a half-metallic double-exchange ferromagnet,
Mn doped ZnO is an antiferromagnetic insulator which changes to a ferromagnet by
additional doping of holes, whereas Ti or Cu doped ZnO remains paramagnetic (Sato and
Katayama- Yoshida 2000). It is also predicted from the first-principle calculation that V-,
Cr-, or Mn-doped GaN is ferromagnetic without additional doping (Sato and Katayama-
Yoshida 200lc).
76 F. MATSUKURA et aI.

In order to solve the problems of the low solubility of magnetic ions in III-V semicon-
ductors and the difficulty of the carrier control in II-VI semiconductors, modulation-doped
III-VIII-VI superlattice structures have been proposed (Kamatani and Akai 200 Ia). Here, a
II-VI layer serves as the magnetic layer, which can contain a large concentration of Mn, and
III-V barrier layers. where impurities are incorporated. provide carriers to II-VI magnetic
layers. The results of the first principle calculation for a AIAs:C/(Cd.Mn)Te superlattice
shows that the ferromagnetic state is indeed stable for high concentrations of C.
The highest spin and the associated large magnitude of the on-site correlation energy U
account for the divalent character of the Mn atoms in a large variety of environments. This
results. in particular. in a large solubility of Mn in II-VI materials and its acceptor character
in a number of III-V compounds. A question arises about ferromagnetic properties of
semiconductors containing other magnetic components. One should recall in this context
the existence of. e.g., ferromagnetic europium chalcogenides and chromium spinels. In
those compounds, the ferromagnetism is not driven by free carriers. With no doubt, the
availability of intrinsic and n-type tetrahedrally-coordinated ferromagnetic compounds
would enlarge considerably the impact of semiconductor electronics.
Actually, a theoretical suggestion was made that superexchange in Cr-based II-VI
compounds can lead to a ferromagnetic order (Blinowski et al. 1996). High composition
(I(}-25%) of V or Cr doped ZnS, ZnSe. and ZnTe have been predicted by first-principle
calculations to exhibit the ferromagnetism at room temperature even without p- or
n-type doping (Sato and Katayama-Yoshida 2(01). Desirable material properties. such as
divergent magnetic susceptibility and spontaneous magnetization, can also be achieved in
the case of a strong antiferromagnetic super-exchange interaction. The idea here (Dietl
1994) is to synthesize a ferrimagnetic system that would consist of antiferromagnetically
coupled alternating layers containing different magnetic cations, e.g., Mn and Co.
In general terms. the transition metals (TM) other than Mn are no longer always
divalent, they lead to the presence of magnetic levels in the gap, and they are characterized
by a non-zero orbital momentum. These will considerably modify transport and optical
properties as well as enhance the sensitivity to co-doping and illumination. Furthermore,
an increase in magnetic anisotropy as well as an enlargement of the coupling to phonons
and thus a shortening of spin-lattice relaxation time are expected, too. Recent ab initio
calculations of Sato and Katayama-Yoshida (2000) suggest that V. Cr. Fe, Co, and Ni in
ZnO should provide carriers, which owing to the double exchange mechanism generate the
ferromagnetic order. We note that in the case of wide gap II-VI compounds studied so far,
neither of these transition metals (TM) introduces free carriers. However. for sufficiently
high TM concentrations a Mott-Hubbard transition is expected, leading to a transport
through the gap d-states. A co-doping with either shallow acceptors or donors could
make such transport. and the associated exchange interaction. more efficient. Since the
TMs act as deep donors and acceptors, the co-doping of such compounds with shallow
impurities (e.g.• by 0 for Mn in III-V compounds) constitutes a natural way to control
the ferromagnetic couplings (Katayama-Yoshida et al. 2001). Indeed, according to Sato
and Katayama-Yoshida (200 Ib). for Fe. Co, or Ni doped ZnO. the ferromagnetic state is
stabilized by the doping of shallow donors.
III-V FERROMAGNETIC SEMICONDUcrORS 77

10.2. Cautionary remarks


Before discussing experimental results, it is appropriate to enlist difficulties encountered
in assessing properties of new ferromagnetic semiconductors. The well-known difficulty
is the multi-phase character of materials grown by non-equilibrium techniques, such as
MBE. In particular, some phases may consist of ferromagnetic and/or ferromagnetic
precipitations, such as metallic and ferromagnetic MnAs. These precipitations, even if
too small to be detected by x-ray diffraction, can give the dominant contribution to the
total magnetic moment of the sample, particularly at high temperatures. Importantly, the
Curie temperature of the precipitations may not be identical to the tabulated values, and
may depend on their size and the host material. Another source of undesirable magnetic
signals originates often from magnetic impurities residing outside the studied layer, for
instance in the substrate. Since ferromagnetic semiconductors are characterized by colossal
magnetotransport and magnetooptical phenomena, a correlation between semiconductor
and magnetic properties allows one usually to rule out parasitic effects. However, stray
magnetic fields produced by ferromagnetic inclusions or their contribution to charge
transport may constitute a source of ambiguity. Additionally, multi-layer and/or multi-
carrier transport of semiconductor structures, together with localization, surface, and
interface phenomena, may generate strong magnetoresistance effects, usually hard to
separate from spin-related phenomena in the examined material.

10.3. Experimental results


10.3.1. (Ga,Mn)As
As mentioned above, the mean-field Zener model suggests that Tc values above 300 K
could be achieved in, e.g., Gao.9MnO.1 As, if such a large value of x would be accompanied
by a corresponding increase of the hole concentration. The elaboration of an annealing
procedure that increases Tc (Katsumoto et al. 1999; Poshtanik et al. 2001) as well as
a successful implementation of low-temperature atomic layer epitaxy to increase the
Mn composition x (Sadowski et al. 2oo1a) constitute examples of recent encouraging
developments in this direction.

10.3.2. (Ga,Mn)N
As discussed in section 2.4, indications of high temperature ferromagnetism in
(Ga,Mn)N have been reported by Sonoda et al. (2002) and Reed et al. (2001), whose layers
grown by ammonia-MBE or prepared by solid state diffusion show ferromagnetism well
above room temperature. Work is under way to rule out the influence of precipitations
as well as to establish how Tc depends on the Mn and carrier concentration. Possible
mechanisms accounting for the experimental observations have been put forward (Dietl
et al. 2oo1b).

10.3.3. (Cd,Mn)GeP2
Il-VI-V2 chalcopyrite OMS, (Cdl_xMnx)GeP2 was prepared by the solid phase
chemical reaction. Mn vacuum deposition (30 nm) on a single crystal of CdGeP2 and
the reacting process (500°C, 30 min) was carried out in an MBE chamber (Medvedkin
et al. 2000). The Mn/Cd composition ratio decreases rapidly with the depth. The average
78 F. MATSUKURA et al.

MnlCd ratio was determined as 20% for an effective thickness 0.5 JLm by energy
dispersive x-ray analysis. The value Tc: ,...., 320 K of (Cdl-xMn x )GePz was determined by
magnetization measurements. Clear hysteresis in the field dependence of magnetization
and the stripe magnetic domain pattern were observed by magnetic force microscopy
(MFM) even at room temperature. A large Faraday rotation of 5.7 x 104 deg/cm at an
energy gap of CdGePz (1.83 eV) was estimated from the magneto-optical Kerr effect
at room temperature. The energy gap of (Cd,Mn)GePz is two times larger than that of
CdGePz. Photoluminescence indicates also that the introduction of Mn enlarges the energy
gap.
Results of a first principles calculation shows that the antiferromagnetic state is more
stable than the ferromagnetic state, and that the energy gap decreases with the Mn
composition (Zhao, Y.-J.et al. 200 Ib). The reason for the discrepancies between theoretical
expectations and experimental results is not clear; it may stem from the substitution of
Ge for Mn in surface-doped samples. More recent plane-wave pseudopotential and KKR-
CPA calculations show that the intrinsic defects are responsible for the stabilization of the
ferromagnetic state (Mahadevan and Zunger 2002; Kamatani and Akai 200 Ib).

10.3.4. Co doped no,


It has been shown by means of magnetization measurements that anatase and rutile (two
forms of TiOz) doped with several percents of Co are ferromagnetic at room temperature
(Matsumoto et al. 200 I, 200 I). Co doped TiOz has been synthesized by laser ablation
MBE employing a combinatorial method, in which a series of thin films with different
compositions can be grown on a single substrate while keeping other growth parameters
virtually unchanged (Ohno 2001). No ferromagnetic order has been found for other form
of TiOz (blookite) and for other transition metals.

10.3.5. Co doped zoo


The magnetic properties of n-type transition-metal doped ZnO have been investigated
(Veda, K. et al. 200 I). The material is synthesized by the pulsed laser deposition, and 1%
of Al is added to produce n-type conduction. Co, Ni, Cr, or Mn serves as a source of the
magnetic spins. ZnO with 5% of Mn, Ni, or Co shows an antiferromagnetic behavior with
Neel temperatures of 310, 350, and 270 K, respectively. On the other hand ZnO with 5-
25% of Co exhibits ferromagnetic (or weak-ferromagnetic) behavior, which is confirmed
by magnetization measurements. The magnetic properties depend on carrier concentration
and mobility. Samples with higher carrier concentrations and mobilities show a stable
ferromagnetism, otherwise spin-glass behavior is observed.
V doped ZnO (V composition 0.05-0.15) with higher conductivity shows ferromagnetic
behavior at room temperature, whereas that with lower conductivity is nonmagnetic (Saeki
et al. 200 1).

10.3.6. Zinc-blende CrAs and CrSb


Thin epitaxial films (less than 3 nm) of CrAs and CrSb with zinc-blende structure can be
grown on GaAs substrates by MBE. Their Tc exceeds 400 K (Akinaga et al. 2000c; Zhao
et al. 200 1b). A zinc-blende structure is confirmed by in-situ RHEED collected during
the growth and ex-situ cross-sectional transmission electron microscopy (TEM). The
III-V FERROMAGNETIC SEMICONDUCTORS 79

preparation and properties of zinc-blende MnAs have also been investigated (Okabayashi
et al. 2001d; Hazama et al. 2(01). According to first-principle calculations (Shirai 2(01)
such materials possess a half-metallic electronic structure. Hence. being compatible with
the existing semiconductor electronics. these systems appear to be promising spin injectors.

11. Summary and outlook

Ferromagnetic semiconductors based on III-V compounds can be incorporated into III-


V based epitaxial structures allowing one to explore spin-dependent phenomena. not
available in structures made of nonmagnetic semiconductors alone. There are two major
directions for the exploration of spin-dependent phenomena in magnetic semiconductors.
One concerns with new functionalities and materials for classical devices. such as
optical isolators and modulators, magnetic sensors and memories. This direction requires
systems with Tc above room temperature. The other direction is quantum related:
new developments in magnetic III-V heterostructures combined with recent progress in
coherent manipulation of carriers (Kikkawa and Awschalom 1999; Malajovich et al. 2000.
200 1; Salis et al. 200 Ib) and nuclear spins (Kikkawa and Awschalom 2000; Salis et al.
2001a) pave the way for future quantum information technologies that will utilize spins in
semiconductors (Kane 1998; Loss and DiVincenzo 1998; Das Sarma et al. 2000; Vrijen
et al. 2(00). Ferromagnetic III-V heterostructures are excellent media to explore this new
field of semiconductor physics and technology. where both charge and spins play critical
roles. With no doubt. however. there is plenty of room for new ideas how to explore
outstanding properties of magnetic semiconductors. which have not yet been exploited.

Acknowledgements

The authors thank many collaborators for fruitful discussion. especially colleagues at
Tohoku University. The work at Tohoku University was supported partly by the Japan
Society for the Promotion and Ministry of Education. Culture. Sports. Science and
Technology. Japan. and the work at Polish Academy of Sciences was supported partly
by Foundation for Polish Science.

References

Abe, E.. F. Matsukura, H. Yasuda, Y. Ohno and Akiba, N., F. Matsukura, A. Shen, Y. Ohno, H. Ohno,
H. Ohno, 2000, Physica E 7, 981. A. Oiwa, S. Katsumoto and Y. lye, 199830 Appl.
Abe, E., K. Sato, F. Matsukura, Y. Ohno and H. Ohno, Phys. Lett. 73, 2122.
2001, Extend Abstracts (The 48th Spring Meeting, Akiba, N., F. Matsukura, Y. Ohno, A. Shen, K. Ohtani,
The Japan Society of Applied Physics) p. 1322. T. Sakon, M. Motokawa and H. Ohno, 1998b, Phys-
ica B 2S~2S8, 561.
Abolfath, M., T. Jungwirth, J. Brum and A.H. Mac-
Akiba, N., D. Chiba, K. Nakata, F. Matsukura, Y. Ohno
Donald, 2001, Phys. Rev. B 63, 054418.
and H. Ohno, 200030 1. Appl. Phys. 87, 6436.
Adhikari, T., and S. Basu, 1984, Jpn. J Appl. Phys., Akiba, N.• F. Matsukura, T. Dietl. K. Ohtani, Y.Ohno
Part I, 33, 4581. and H. Ohno, 2000b. presented at the Int. Conf. on
Akai, H., 1998, Phys, Rev. Lett. 81, 3002. Physics and Application of Spin-Related Phenom-
80 F. MATSUKURA et aI.

ena in Semiconductors. Sendai, Japan. Sep. 13-15. eds G. Landwehr and W. Ossau (World Scientific.
2000. Singapore) p. 861.
Akinaga, H.. S. Basu, M. Mizuguchi, T. Manago Bosselli, M.A .• A. Ghazali and l.C, da Cunha Lima.
and M. Shirai. 2000a, Extend Abstracts (The 61st 2000. Phys. Rev. B 62, 8895.
Autumn Meeting. The Japan Society of Applied Casey. H.C .• Jr.• D.D. Sell and K.W. Wecht. 1975. J.
Physics) p. 1138. Appl. Phys, 46. 250.
Akinaga, H.• S. Nemeth. J. De Boeck. H. Bender. Chao. C.Y.-P.. and S.L. Chuang. 1991. Phys. Rev. 43.
G. Borghs, H. Ofuchi and M. Oshima. 2000b. Appl. 7027.
Phys. Lett. 77. 4377. Chapman. R.A .• and W.G. Hutchinson. 1967. Phys,
Akinaga, H.• T. Manago and M. Shirai. 2000c. Jpn. J. Rev. Lett. 18. 443.
Appl. Phys. Part 2. 39. Ll118. Chattopadhyay, A.•S. Das Sarma and AJ. Millis. 2001.
Aliyev, M.I..I.S. Dadashev and G.I. Safaraliyev, 1980. Phys. Rev. Lett. 87.227202.
Phys. Met. MetaIlogr. 49. 166. Chazalviel, J.-N .• 1975. Phys. Rev. B 11.3918.
Alrnesh, N.• and B. Goldstein. 1962. Phys. Rev. 128. Chen. X.. K.P. Mooney, G. Comenescu, L. Guo.
1568. B.D. McCombe. Y. Sasaki. X. Liu, SJ. Poteshnik,
Al'tshuler, B.L.. and A.G. Aronov, 1985. in: Electron- P. Sciffer and J.K. Furdyna, 2000. APS March 2000
electron interaction in disordered conductors Meeting. 20-24 Mar. 2000. Minneapolis. USA.
(North-Holland. Netherlands) pp. 1-153. Chiba, D.• N. Akiba, F. Matsukura, Y. Ohno and
Ando, K., T. Hayashi. M. Tanaka and A. Twardowski.
H. Ohno, 2000. Appl. Phys. Lett. 77. 1873.
1998. 1. Appl. Phys. 83. 6548. Chien. C.L.. and C.w. Westgate. 1980. The Hall effect
and Its Applications (Plenum Press. New York).
Ando, M.• Y. Kanemitsu, K. Takamura, F. Matsukura
and H. Ohno, 1999. Fifth Symposium on Physics
Coey, J.M.D .• M. Viret and S. von Molnar. 1999. Adv.
Phys.48. 167.
and Application of Spin-Related Phenomena in
Crepieux, A.• and P. Bruno, 2001. cond-matlOI01376.
Semiconductors. 16-17 Dec. 1999. Sendai, Japan.
Das Sarma. S.• J. Fabian. X. Hu and I. Zutic, 2000. Su-
p.I90.
perlattice and Microst, 27. 285.
Arata. I., Y. Ohno, F. Matsukura and H. Ohno, 2001.
De Boeck. J.• R. Oesterholt, A. Van Esch, H. Bender.
Physica E 10. 288.
e. Bruynseraede, C. Van Hoof and G. Borghs, 1996.
Arrott, A.• 1957. Phys, Rev. 108. 1394.
Appl. Phys. Lett. 68. 2744.
Ball. P.• 2000. Nature 404, 918.
De Boeck. J.• and G. Borghs, 1999. Phys. World. No 4.
Baron. T.• S. Tatarenko, K. Sarninadayer, N. Magnea
p.27.
and J. Fontenille, 1994. Appl. Phys. Lett. 65. 1284.
DeSirnone. D., C.E.e. Wood and C.A. Evans Jr.• 1982.
Baxter. D.V.• D. Ruzmetov, 1. Scherschligt, Y. Sasaki. J. Appl. Phys. 53. 4938.
X. ue, J.K. Furdyna and c.a. Milke. 2002. Phys. Dietl. T.. 1994. (Diluted) Magnetic Semiconductors.
Rev. B. 65. 212407. in: Handbook of Semiconductors. ed. T.S. Moss.
Belitz. D.• and T.R. Kirkpatrick. 1994, Rev. Mod. Phys. Vol. 3B (North-Holland, Amsterdam).
57.287. Dietl. T.• A. Haury and Y. Merle d'Aubigne, 1997.
Benoit 11 la Guillaume. C; D. Scalbert and T. Dietl. Phys. Rev. B 55. R3347.
1992. Phys, Rev. B 46. 9853. Dietl. T., 1. Cibert, D. Ferrand and Y. Merle d' Aubigne,
Beschoten, B.• P.A. Crowell. I. Malajovich, D.D. Aw- 1999. Mater. Sci. Eng. B 63. 103.
shalom. F. Mastukura, A. Shen and H. Ohno, 1999. Dietl. T.• H. Ohno, F. Matsukura, J. Cibert and D. Fer-
Phys. Rev. Lett. 83. 2073. rand. 2000. Science 287.1019.
Bhatt. R.N .. and X. Wan. 1999. Intern. 1. Modem. Phys, Dietl. T.. J. KOnig and A.H. Maclsonald, 200la. Phys.
C 10.1459. Rev. B 64. 241201.
Bhatt. R.N .• and M. Berciu, 2001. Phys. Rev. Lett. 87. Dietl. T.• F. Matsukura and H. Ohno, 200lb. cond-
10723. matlOI09145. Phys, Rev. B. to be published.
Bhattacharjee, A.K.. and e. Benoit Ii la Guillaume. Dietl. T.. H. Ohno and F. Matsukura, 200lc. Phys. Rev.
2000, Solid State Commun. 113. 17. 63.195205.
Bither. T.A .• and W.H. Cloud, 1965. J. Appl. Phys. 36. Dorlrijn, 1.w.F.. 1976. Philips Res. Repts. 31. 287.
1501. Endo, T.. T. Slupinski, S. Yanagi, A. Oiwa and
Blinowski, J.• P. Kacman and J.A. Majewski. 1996. H. Munekata, 2001. Seventh Symposium on the
Phys. Rev. B 53. 9524. Physics and Application of Spin-Related Phenom-
Blinowski, 1.. P. Kacman and J.A. Majewski. 1997. in: ena in Semiconductors. 17-18 Dec. 2001, Yoko-
High Magnetic Fields in Semiconductor Physics II. hama. Japan. p. 33.
III-V FERROMAGNETIC SEMICONDUcrORS 81

Edwards. P.P.• and M.I. Sienko. 1978. Phys. Rev. B 17. Guo. S.P.. H. Ohno, A. Shen, F. Matsukura and
2575. Y.OOOo. 1998, Appl. Surf. Sci. 130-132,797.
Edwards. P.P.• and C.N.R. Rao (eds), 1995. Metal- Guo, S.P.. A. Shen, F. Matsukura, Y. 0000 and
Insulator Transitions Revisited (Taylor and Francis. H. 0000. 1999. 1. Cryst. Growth, 2011202. 684-
London). 688.
Eggenkamp, P.T.I.. H.I.M. Swagten, T. Story. Guo, S.P., A. Shen, H. Yasuda, Y. 0000. F. Matsukura
V.I. Lhvinov, C.H.W. Swuste and W.J.M. de Jonge, and H. 0000. 2000. J. Cryst. Growth 208. 799.
1995. Phys. Rev. B 51. 15250. Haneda, S.. M. Yamamura, Y. Takatani, K. Hara.
Fedorych, O.M., E.M. Hankiewicz, Z. Wilamowski S. Harigae aod H. Munekata, 2oooa, Jpn. 1. Appl.
and J. Sadowski. 2001. cond-matlOl06227 (submit- Phys. 39. L9.
ted to Phys. Rev. B). Haneda, S.• H. Munekata, Y. Takatani aod S. Koshi-
Fernandez-Rossier, J.• and L.I. Sham. 2001, Phys. Rev. hara, 2000b. 1. Appl. Phys. 87.6445.
B 64. 235323. Haneda, S., S. Kodhihara and H. Munekata, 2001,
Ferrand. D.. 1. Cibert, A. Wasiela, C. Bour- Physica E 10.437.
gognon, S. Tatarenko, G. Fishman. S. Kolesnik, Hansen. L., D. Ferrand. G. Richter. M. Thicrley.
1. Jaroszyiiski, T. Dietl. B. Barabara and D. Dufeu, V. Hock. N. Schwarz. G. Reuscher, G. Scmidt,
2000.1. Appl. Phys. 87, 6451. A. Waag and L.W. Molenkarnp. 2001. cond-
Ferrand. D., J. Cibert, A. Wasiela, C. Bpur- matlOl07619; submitted to Appl. Phys. Lett.
gognon, S. Tatarenko, G. Fishman. T. Andrearczyk, Harris. J.G.E.. D.D. Awshalom. F. Matsukura,
J. Jaroszyriski, S. Kolesnik, T. Dietl. B. Barbara and H. 0000, K.D. Maranowski and A.C. Gossard,
D. Dufeu, 2001. Phys. Rev. B 63. 085201. 1999, Appl. Phys. Lett. 75. 1140.
Fiederling, R.. M. Keirn. G. Reuscher, W. Ossau, Hartmann. Th .• M. Lampalzer, W. Stolz. K. Megges,
G. Scmidt, A. Waag and L.W. Molenkamp, 1999. L. Lorberth, P.I. Klar and W. Heimbrodt, 2000. Thin
Nature 402. 787. Solid Films 364. 209.
Fisher. M.E .• S.-K. Ma and B.G. Nickel. 1972, Phys. Hashimoto. M.. Y.-K. Zhou. M. Kanamura and
Rev. Lett. 29.917. H. Asahi, 2002. to be published in Solid State Com-
Frolich, H.• and ER.N. Nabarro, 1940. Proc. R. Soc. mun.
London Ser. A 175. 382. Hashimoto. Y., T. Hayashi. S. Katsumoto and Y. lye.
Fukumura, T.. T. Shono, K. Inaba, T. Hasegawa. 2002. J. Cryst. Growth 237-239.1334.
H. Koinuma, E Matsukura and H. Ohno, 2001, Hass, K.C.. 1991. in: Semimagnetic Semicondic-
Physica E 10. 135. tors and Diluted Magnetic Semiconductors. eds
Fumagalli, P., and H. Munekata, 1996. Phys. Rev. B 53. M. Averous and M. Balkanski (Plenum. London.
15045. 1991).
Furdyna, J.• and J. Kossut (eds), 1988. Semiconductor Haury. A.• A. Wasiela, A. Arnoult, J. Cibert,
and Semi metals. Vol. 25 (Academic. New York). S. Tatarenko, T. Dietl and Y. Merle d'Aubigne,
Gaj. 1.A.• R. Planel and G. Fishman. 1979. Solid State 1997. Phys. Rev. Lett. 79, 511.
Commun. 29. 435. Hayashi, T.• M. Tanaka. T. Nishinaga, H. Shimada.
Galazka, R.R.. 1977. Postepy Fyzyki 28. 601. H. Tsuchiya and Y. Ohtsuka, 1997a. J. Cryst.
Gayrnann, A.• H.P. Geserich and H.V. Lohneysen, Growth 175/176. 1063.
1995. Phys. Rev. B 52. 16486. Hayashi, T.• M. Tanaka, K. Seto, T. Nishinaga and
Gebicki, W.. J. Strzeszewski, G. Kalmer, T. Szyszko K. Ando, 1997b. Appl. Phys. Lett. 71. 1825.
and S. Podsiadlo, 2000, Appl. Phys. Lett. 76. 3870. Hayashi. T.• M. Tanaka, K. Seto, T. Nishinaga, H. Shi-
GI6d. P.•T. Dietl. M. Sawicki and I. Miotkowski, 1994. mada and K. Ando, 1998. J. Appl. Phys. 83. 6551.
Physics B 194-196.995. Hayashi. T.• H. Shimada, H. Shimizu aod M. Tanaka,
Goodenough. J.B .• 1969. 1. Phys, Chern. Solids 30, 1999. J. Crystal Growth 2011202. 689.
261. Hayashi. T.• S. Katsumoto, Y. Hashimoto. A. Endo,
Grandidier, B.. 1.P. Nys, C. Delerue, D. Stievenard, M. Kawamura, M. Zalalutdinov and Y. lye. 2000.
Y. Higo and M. Tanaka. 2000. Appl. Phys. Lett. 77. Physica B 284-288. 1175.
4001. Hayashi. T.• Y. Hashimoto. S. Katsumoto and Y. lye,
Guha, S.• and H. Munekata, 1993, J. Appl. Phys. 74. 2001. Appl. Phys. Lett. 78,1691.
2975. Hazama, Y.. K. Yamamoto, A. Nagashima and
Gummich, U.• and I.C. da Cunha Lima. 1990. Solid J. Yoshino. 2001. Seventh Symposium on Physics
State Commun. 76. 831. and Application of Spin-Related Phenomena in
82 F. MATSUKURA et al.

Semiconductors. 17-18 Dec. 2001. Sendai, Yoko- Y. Harada. S. Shin and K. Hirakawa. 2001. Mater.
hama. p. 160. Sci. Eng. B 84. 88.
Heimbrodt, W.• Th. Hartmann. PJ. Klar, M. Lam- Kawakami. R.K .. E. Hohnston-Halperin, L.F. Chen.
palzer, W. Stolz. K. Volz. A. Schaper. W. Treut- M. Hanson, N. Guebels, J.S. Speck. A.C. Gossard
mann. H.-A. Krug von Nidda. A. Loidl. T. Ruf and and D.O. Awshalom, 2000. Appl. Phys. Lett. 77.
V.F. Sapega, 2001, Physica E 10. 175. 2379.
Henriques. A.B .• SA Obukhov, N.F.Yr. Oliveira and Kawakami. R.K.. Y. Kato, M. Hanson. I. Malajovich.
V.A. Sanina, 1999, Pis'rna v ZhETF 69.358. J.M. Stephens. E. Johnston-Halperin, G. Salis.
Higo, Y.• H. Shimizu and M. Tanaka. 200la. Physica E A.C. Gossard and D.O. Awshalorn, 2001. Science
10.292. 294. 131.
Higo, Y.. H. Shimizu and M. Tanaka. 200lb. 1. Appl. Kenett, M.P.. M. Berciu and R.N. Bhatt. 2001. cond-
Phys. 89. 6745. matlO102315.
Hirakawa. K.• A. Oiwa and H. Munekata, 2001. Phys- Kepa, H.. 1. Kutner-Pielaszek, A. Twardowski.
ica E 10. 215. c.F. Majkrzak, 1. Sadowski. T. Story and
Hirsch. J.E .• 1999. Phys. Rev. Lett. 83. 1834. T.M. Giebultwicz, 2001. Phys. Rev. B 64. 121302.
Ilegems, M.. R. Digle and L.w. Rupp Jr.• 1975. J. Appl. Kikkawa, 1.M.• and D.O. Awschalom, 1999. Nature
Phys. 46. 3059. 397.139.
Inoue. J.• S. Nonoyama and H. Itoh, 2000. Phys. Rev. Kikkawa, J.M .• and D.O. Awschalom, 2000. Science
Lett. 85.4610. 287.473.
Johnston-Halperin, E .• D. Lofgreen, R.K. Kawakami. Kohda, M.. Y. Ohno, K. Takamura, F. Matsukura and
D.K. Young. I. Coldren. A.C. Gossard and D.O. Aw- H. Ohno, 2001. Jpn. J. Appl. Phys. 40. L1274.
shalom. 2002. Phys. Rev. B 65. 041306. Konig. J.. H.-H. Lin and A.H. MacDonald. 2000. Phys.
Jonker. J.T.. Y.D. Park. B.R. Bennett. H.D. Cheong. Rev. Lett. 84. 5628.
G. Kioseoglow and A. Petrou, 2000. Phys. Rev. B Konig. J.• H.-H. Lin and A.H. MacDonald. 2001. Phys-
62.8180. ica 10. 139.
Jungwirth. T.• W.A. Atkinson. B.H. Lee and A.H. Mac- Koshihara, S.• A. Oiwa, M. Hirasawa, S. Katsumoto,
Dolnald. 1999. Phys. Rev. B 59. 9818. Y.lye. C. Urano, H. Takagi and H. Munekata, 1997.
Jungwirth. T.. Q. Niu and A.H. MacDonald. 2002. Phys. Rev. Lett. 78.4617.
Phys. Rev. Lett. 88. 207208. Kossacki, P.. D. Ferrand. A. Amoult, J. Cib-
Kacman, P.• 2001. Semicond. Sci. Technol. 16. R25. ert, S. Tatarenko, A. Wasiela, Y. Merle
Kamatani, T.• and H. Akai, 2oola. Physica E 10. 157. d'Aubigne, K. Swiatek, M. Sawicki. J. Wro-
Kamatani, T.• and H. Akai, 2oolb. Seventh Sympo- bel. W. Bardyszewski and T. Dietl. 2000. Physica E
sium on Physics and Application of Spin-Related 6.709.
Phenomena in Semiconductors. 17-18 Dec. 2001. Kossut, J.. and W. Dobrowolski. 1993. in: Handbook
Sendai, Japan. p. 124. of Magnetic Materials. K.HJ. Buschow (ed.), Vol. 7
Kanamura, M.• Y.K. Zhou, S. Okumura. K. Asami, (North-Holland. Amsterdam).
M. Nakajima. H. Harima and H. Asahi, 2002. Jpn. Kotani, T.. 2000. J. Phys. Condo Matter 12. 2413.
1. Appl. Phys. 42. 1019. Krebs. J.J .• and G.H. Stauss. 1977. Phys. Rev. B 16.
Kane. B.E .• 1998. Nature 393. 133. 971.
Kasuya, T.. and A. Yanase. 1968. Rev. Mod. Phys. 40. Kreisel. 1.. W. Ulrici. M. EI-Metoui. A.-M. Vasson,
684. A. Vasson and A. Gavaix, 1996. Phys. Rev. B 54.
Katayama-Yoshida. H.• R. Kato and T. Yamamoto. 10508.
2001. J. Cryst. Growth. 231. 428. Krol, A.• Y.L. Soo, S. Huang. Z.H. Ming, Y.H. Kao,
Kato, R.. and H. Katayama-Yoshida. 1999. Fifth Sym- H. Munekata and LL Chang. 1993. Phys. Rev. B
posium on Physics and Application of Spin-Related 47.7187.
Phenomena in Semiconductors. 16-17 Dec. 1997. Kuivalainen, P.• 2001. Phys. Stat. Sol. (b) 227. 449.
Sendai. Japan. p. 233. Kuivalainen, P.• and A. Hovinen, 2002. J. Phys. D 35.
Katsumoto, S.• 1999. unpublished. 48.
Katsumoto, S.• A. Oiwa, Y. lye. H. Ohno, F. Matsukura, Kulatov, E.. H. Nakayama. H. Mariette. H. Ohta and
A. Shen and Y. Sugawara, 1998. Phys. Status Solidi Yu.A. Uspenskii, 2001. unpublished.
B 205.115. Kuroiwa, T.. T. Yasuda. F. Matsukura, A. Shen,
Katsumoto, S.. T. Hayashi. Y. Hashimoto. Y. lye. Y. Ohno, Y. Segawa and H. Ohno, 1998. Electron.
Y. Ishiwata, M. Watanabe. R. Eguchi, T. Takeuchi. Lett. 34. 190.
III-V FERROMAGNETIC SEMICONDUCTORS 83

Kuwabara, S., K. Ishii, S. Haneda, T. Kondo and Matsuda, Y.H.• T. Ikaida, N. Miura, M.A. Zudov,
H. Munekata. 200la. Physica E 10. 233. J. Kono and H. Munekata, 2001. Physica E 10. 219;
Kuwabara, S.• T. Kondo. T. Chikyow, P. Ahmet and Matsuda, Y.H.• T. Ikaida, N. Miura, Y. Hashimoto.
H. Munekata, 200lb. Jpn. J. Appl. Phys. 40. L727. S. Katsumoto, J. Kono, M.A. Zudov, G.D. Sanders,
Lee. B.• T. Jungwirth and A.H. MacDonald. 2000, CJ. Stanton and H. Munekata. Seventh Symposium
Phys. Rev. B 61. 15606. on the Physics and Application of Spin-Related
Leroux-Hugon, P.• 1973, in: New Developments in Phenomena in Semiconductors, 17-18 Dec., 2001.
Semiconductors. eds P.R. Walace. R. Harris and Yokohama, Japan. p. 41.
MJ. Zuckennann (Noordhoff, Leyden). Matsukura, F.• 1997. unpublished.
Leroux-Hugon, P.• and A. Ghazali, 1972. J. Phys. C 5. Matsukura, F., A. Shen, Y. Sugawara, T. Omiya,
1072. Y. Ohno and H. Ohno, 1998a, Presented at the
Linnarson, M.• E. Janzen. B. Monemar, M. KJevennan 25th Int. Symp. Compound Semiconductors, 12-16
and M. Thilderkvist, 1997. Phys. Rev. B 55. 6938. October 1998. Nara, Japan; Proc. 25th Int. Symp.
Litvinov, V.1. and Y.K. Dugaev, 2001, Phys. Rev. Lett. Compound Semiconductors. Institute of Physics
86.5593. Conference Series. No. 162 (lOP Publishing Ltd.
Liu, S.• K. Karrai, F. Dunmore. H.D. Drew. R. Wilson Bistrol, 1999) p. 547.
and G.A. Thomas. 1993, Phys. Rev. B 48. 11394. Matsukura, F.. H. Ohno, A. Shen and Y. Sugawara,
Liu, X.• Y. Sasaki and 1.K. Furdyna, 2001. Appl. Phys. 1998b. Phys. Rev. B 57. R2037.
Lett. 79. 2414. Matsumoto. Y., M. Murakami. T. Shono, T. Hasegawa.
Liu, X.C .• D. Heiman. 1. Hao and K.C. Hsieh. 1995. in: T. Fukumura, M. Kawasaki, P. Ahmet, T. Chikyow,
Proceedings International Conference on Physics S. Koshihara and H. Koinuma, 2001. Science 291.
of Semiconductors in High Magnetic Fields. ed. 854.
D. Heiman (World Scientific, Singapore) p. 658. Matsumoto. Y.. R. Takahashi. M. Murakami. T. Koida,
Look.D.e.. 1991, J. Appl. Phys. 70. 3148. X.-J. Fan, T. Hasegawa, T. Fukumura, M. Kawasaki.
Loss. D.• and D.P. Divincenzo. 1998. Phys. Rev. A 57. S. Koshihara and H. Koinuma, 2001, Jpn, J. Appl.
120. Phys. 40. Ll204.
Louriero da Silva. M.A. Boselli, LC, da Cunha Lima, Mauger. A., and e. Gotard, 1986, Phys. Rep. 141.51.
X.F. Wang and A. Ghazali, 2001, Appl. Phys. Lett. Medvedkin, G.A .• T. Ishibashi, T. Nishi. K. Hayata,
79.3305. Y. Hasegawa and K. Sato, 2000. Jpn. J. Appl. Phys.
Luysberg, M., H. Sohn, A. Prasad. P. Specht. 39. L949.
Z. Liliental-Weber, E.R. Weber. 1. Gebauer and Methfessel, S.• 1965. IEEE Trans. Mag. I. 144.
R. Krause-Rehberg, 1998,1. Appl. Phys. 83. 561. Methfessel, SJ., and F. Holtzberg, 1966. US Patent
Lyu, P., and K. Moon, 2001, Phys. Rev. B 64. 035201. 3.271.709.
Ma, S.-K .• 1976, Modem Theory of Critical Phenom- Methfessel, S., and D.C. Mattis. 1968, in: Encyclope-
ena (W.A. Benjamin, Reading). dia of Physics. Vol. XVIII/I. Magnetism. Magnetic
Mac. W.• A. Twardowski and A.M. Hennel, 1994, in: Semiconductors. ed. H.PJ. Wijn (Springer-Verlag.
Proceedings of the 22nd International Conference Berlin) pp. 389-562.
on Physics of Semiconductors. ed. D.J. Lockwood Mizokawa, T.• and A. Fujimori, 1993. Phys. Rev. B 48,
(World Scientific. Singapore. 1995) p. 1569. 14150.
Mahadevan, P.• and A. Zunger, 2002. Phys. Rev. Lett. Moriya, R.• H. Munekata, T. Kondo and A. Oiwa, 2002.
88, 047205 (2002). J. Cryst. Growth 237-239. 1344.
Malajovich, I., J.M. Kikkawa, D.O. Awschalom. Munekata, H.• 1995. Mater. Sci. Eng. B 31,151.
U. Berry and N. Samarth, 2000. Phys. Rev. Lett. Munekata, H., H. Ohno, S. von Molnar. A. Segmilller,
84.1014. LL Chang and L. Esaki, 1989. Phys, Rev. Lett. 56.
Malajovich, I.. U. Berry, N. Samarth and 777.
D.O. Awschalom, 2001. Nature 411. 770. Munekata, H.• H. Ohno, S. von Molnar. A. Harwit,
Mldek. J.• and F. Maka, 2001, Acta Phys. Polon. A 100. A. Segmilller and L.L. Chang. 1990. J. Vac. Sci.
319. Technol. B 8. 176.
Masterov, V.F., K.F. Shtelmakh and M.N. Barbashov, Munekata H.• H. Ohno, R.R. Ruf, R.I. Gambino and
1988, Fiz. Tech. Poluprovodn. 22.654 [Sov. Phys. L.L. Chang. 1991. J. Cryst. Growth 111, 1011.
Semicond. 22. 408). Munekata H.• T. Penny and L.L. Chang. 1992, Surf.
Matsuda. Y.H., H. Arimoto, N. Miura. A. Twardowski. Sci. 267. 342.
H. Ohno, A. Shen and F. Matsukura, 1998. Physica Munekata, H.. A. Zaslavsky, P. Fumagalli and
B~258.565. R.I. Gambino. 1993. Appl. Phys. Lett. 63. 2929.
84 F. MATSUKURA et al.

Munekata, H., T. Abe, S. Koshihara, A. Oiwa, M. Hira- Ohno, H.. F. Matsukura and Y. Ohno, 2001. Solid Stale
sawa, S. Katsumoto, y. lye, C. Urano and H. Takagi, Commun. 119.281.
1997,1. Appl. Phys. 81,4862. Ohno, Y., D.K. Young, B. Beshoten, F. Matsukura,
Nagai, Y.. T. Kunimoto, K. Nagasaka, H. Nojiri, H. Ohno and D.D. Awschalom, 1999. Nature 402.
M. Motokawa, F. Matsukura, T. Dietl and H. Ohno, 790.
2001, Jpn. 1. Appl. Phys. 40, 6231. Ohno, Y., I. Arata. F. Matsukura, K. Ohtani, S. Wang
Nishikawa, Y., Y. Satoh and 1. Yoshino, 1997, Sec- and H. Ohno, 2000, Appl. Surf. Sci. 159/160, 308.
ond Symposium on Physics and Application of Spin Ohno, Y.• I. Arata, F. Matsukura and H. Ohno, 2001,
Related Phenomena in Semiconductors, 27-28 Jan. Physica E, to be published.
1997, Sendai, Japan, p. 122. Ohya, E., H. Shimizu, Y. Higo, 1. Sun and M. Tanaka,
Nojiri, H., M. Motokawa, S. Takeyama, F. Matsukura 2001, Jpn. J. Appl. Phys. 41, L24.
and H. Ohno, 1998, Physica B 256-258, 569. Oiwa, A., S. Katsumoto, A. Endo, M. Hirasawa, Y. lye.
Nonoyama, S., and 1. Inoue, 2001, Physica E 10, 283. H. Ohno, F. Matsukura, A. Shen and Y. Sugawara,
Obukhov, SA, 1993, Solid State Commun. 88. 255. 1997, Solid Slate Commun. 103,209.
Obukhov, SA, 1996, Proceedings of 23rd Interna- Oiwa, A., S. Katsumoto, A. Endo, M. Hirasawa, Y. lye,
tional Conference on the Physics of Semiconductors F. Matsukura, A. Shen, Y. Sugawara and H. Ohno,
(World Scientific, Singapore) pp. 99-102. 1998a, Physica B 249-251, 775.
Obukhov, SA, and N.J. Pepic, 1989, Solid State Com- Oiwa, A., S. Katsumoto, A. Endo, M. Hirasawa, Y. lye,
mun. 70, 103. H. Ohno, F. Matsukura, A. Shen and Y. Sugawara,
Ofuchi, H., M. Oshima, M. Tabuchi, Y. Takeda, H. Ak- 1998b, Phys. Status Solidi b 205, 115.
Oiwa, A., A. Endo, S. Katsumoto, Y. lye, H. Munekata
inaga, S. Nemeth. 1. De Boeck and G. Borghs,
and H. Ohno, 1999, Phys. Rev. B 59, 5826.
200la, Appl. Phys. Lett. 78, 2470.
Oiwa, A., T. Siupinski and H. Munekata, 2001, Appl.
Ofuchi, H., T. Kubo, M. Tabuchi, Y. Takeda, F. Mat-
Phys. Lett. 78. 518.
sukura, S.P. Guo, A. Shen and H. Ohno, 200lb, J.
Oiwa, A., Y. Mitsumori, R. Moriya, T. Siupinski and
Appl. Phys. 89, 66.
H. Munekata, 2002, Phys. Rev. Lett. 88, 137202.
Ogawa, T., M. Shirai, N. Suzuki and I. Kitagawa, 1999,
Okabayashi, J.• A. Kimura, O. Rader, T. Mizokawa,
J. Magn. Magn. Mater. 196-197,428.
A. Fujimori, T. Hayashi and M. Tanaka. 1998, Phys.
Ohldag, H., V. Solinus, F.U. Hillebrecht, J.B. Goed-
Rev. B 58, R421 I.
koop, M. Finazzi, F. Matsukura and H. Ohno, 2000,
Okabayashi, J., A. Kimura, T. Mizokawa, A. Fujimori,
Appl. Phys. Lett. 76, 2928.
T. Hayashi and M. Tanaka, 1999, Phys. Rev. B 59,
Ohno, H., 1998, Science 281, 951.
R2486.
Ohno, H., 1999, J. Magn. Magn. Mater. 200, 110. Okabayashi, J., A. Kimura, O. Rader, T. Mizokawa,
Ohno, H., 2001, Science, 291,840. A. Fujimori, T. Hayashi and M. Tanaka, 2001a,
Ohno, H., H. Munekata, S. von Molnar and Physica E 10, 192.
L.L. Chang, 1991, J. Appl. Phys. 69, 6103. Okabayashi, J.• A. Kimura, O. Rader, T. Mizokawa,
Ohno, H.• H. Munekata, T. Penny, S. von Molnar and A. Fujimori, T. Hayshi and M. Tanaka, 200lb, Phys.
L.L. Chang, 1992, Phys. Rev. Lett. 68, 2664. Rev. B 64, 125304.
Ohno, H.. A. Shen, F. Matsukura, A. Oiwa, A. Endo, Okabayashi, J., K. Ono, T. Mano, M. Mizuguchi,
S. Katsumoto and Y. lye, 1996a, Appl. Phys. Lett. H. Horiba, S. Nakazoe, T. Kihara, K. Nakamura,
69,363. H. Kiwata, A. Fujimori and M. Oshima, 200lc, Ex-
Ohno, H., F. Matsukura, A. Shen, Y. Sugawara, tend Abstracts (The 48th Spring Meeting, The Japan
A. Oiwa, A. Endo, S. Katsumoto and Y. lye, 19%b. Society of Applied Physics) p. 1325.
in: Proceedings of the 23rd International Confer- Okabayashi, 1., K. Ono, M. Mizuguchi, M. Yamada,
ence on Physics of Semiconductors, eds M. Schef- A. Fujirnori, M. Oshima and H. Akinaga, 200ld,
fler and R. Zimmermann (World Scientific, Singa- Seventh Symposium on Physics and Application of
pore) pp. 405-408. Spin-Related Phenomena in Semiconductors, 17-18
Ohno, H., N. Akiba, F. Matsukura, A. Shen, K. Ohtani Dec. 2001, Yokohama, Japan, p. 147.
and Y. Ohno, 1998, Appl. Phys. Lett. 73, 363. Okazawa, D., K. Yamamoto and J. Yoshino, 1997,
Ohno, H., D. Chiba, F. Matsukura, T. Omiya, E. Abe, Third Symposium on Physics and Application of
T. Dietl, Y. Ohno and K. Ohtani, 2000, Nature 408, Spin-Related Phenomena in Semiconductors, 16-17
944. Dec. 1997, Sendai, Japan, p. 212.
Ohno, H.• and F. Matsukura, 2001. Solid State Com- Okazawa, D., K. Yamamoto, A. Nagashima and
mun. 117, 179. 1. Yoshino, 2001, Physica E 10, 229.
III-V FERROMAGNETIC SEMICONDUCTORS 85

Omiya, T.• F. Matsukura, T. Dietl. Y. Ohno, T. Sakon, Sadowski. I .• et aI.• 200ld. presented at Drugi Kongres
M. Motokawa and H. Ohno, 2000. Physica E 7. 976. Polskiego Towarzystwa Prozniowego, May 13-17,
Omiya, T.. F. Matsukura, A. Shen, Y. Ohno and 2001. Warszawa, Poland.
H. Ohno, 2001. Physica E 10. 206. Sadowski. 1.. K. Swil4tek. P. Svedlindh, R. Mathieu.
Osinniy, V.• A. Jedrzejczak, M. Arciszewska, W. Do- M. Karlsteen, 1. Kanski and L. Ilver, 2001e. sub-
browolski. T. Story and I. Sadowski. 2001. Acta mitted to Phys. Rev. B.
Phys. Polon. A 100. 327. Saeki. H.• H. Tabata and T. Kawai. 2001. Solid State
Overberg, M.E.. c.R. Abernathy. S.1. Pearton, Commun. 120,439.
N.A. Theordropoulou, K.T. MacCarthy and Saito. H.. W. Zaets, R. Akimoto, K. Ando, Y. Mishima
F. Hebard. 2001. Appl. Phys. Len. 79. 1312. and M. Tanaka, 2001, I. Appl. Phys. 89. 7392.
Paalanen. M.A .• and R.N. Bhatt. 1991. Physica B 169. Sakai. 0 .. S. Suzuki and K. Nishizawa, 2001. J. Phys.
223. Soc. Ipn. 70. 1105.
Park. 1.H.• S.K. Kwon and BJ. Min. 2000. Physica B Sakai. 0 .. and S. Suzuki. 2001. Physica E 10. 148.
281 & 282. 703. Salis. G.• D.T. Fuchs. J.M. Kikkawa, D.O. Awshalom,
Park. Y.D.• A.T. Habbicki, S.c. Erwin. C.S. Hellberg. Y. Ohno and H. Ohno, 2001a, Phys. Rev. Len. 86.
I.M. Sullivan. I.E. Manson. T.M. Ambrose. A. Will- 2677.
son. G. Spanos and B.T. Jonker, 2002. Science 295, Salis. G.• Y. Kato, K. Ensslin, D. Driscol, A.C. Gossard
651. and D.O. Awschalom, 2001b. Nature 414, 619.
Parmenter. R.H.• 1973. Phys. Rev. B 8.1273. Sanvito, S.• P. Ordej6n and N.A. Hill. 2001. Phys. Rev.
Pashitskii, E.A .• and S.M. Ryabchenko, 1979. Fizika B 63.165206.
Tverdogo Tela 21. 545. Sanvito, S.• and N.A. Hill. 2001. Appl. Phys. Len. 78,
Petukhov, A.G., D.O. Demchenko and A.N. Chantis, 3493.
2000, I. Vac. Sci. Technol. B 18.2109. Sapega, V.F..T. Ruf and M. Cardona, 2000. Solid State
Potashnik, S.1.• K.C. Ku, S.H. Chun, J.J. Berry. Commun. 114. 573.
N. Samarth and P. Schiffer. 2001. Appl. Phys. Len. Sapega, V.F.• T. Ruf and M. Cardona. 200 I, Phys. Stat.
79.1495. Solidi b 226. 339.
Reed. M.L.. N.A. El-Masry, H.H. Stadelmaier, Sato, K.. andH. Katayama-Yoshida, 2000. Ipn. J. Appl.
M.K. Ritums, M.1. Reed, e.A. Parker, I.e. Roberts Phys. 39. L555.
and S.M. Bedair, 2001. Appl. Phys. Len. 79. 3473. Sato, K.. and H. Katayama-Yoshida, 2001a, Ipn. J.
Price. G.L.. 1991, Phys. Rev. Lett 66. 469. Appl. Phys. 40. L651.
Prinz. G.A .• 1998. Science 282.1660. Sato, K.• and H. Katayama-Yoshida, 200lb. Ipn. 1.
Rodrigues Bitterncourt, A.C.. A.M. Cohen and Appl. Phys. 40. L334.
G.E. Marques. 1998. Phys. Rev. B 67.4525. Sato, K., and H. Katayama-Yoshida. 200lc, Jpn. I.
Sadowski, 1.. 1. Domagala. 1. Bak-Misiuk, K. Sqiatek, Appl. Phys. 40. L485.
I. Kanski, L. liver and H. Oscarsson, 1998. Acta Satoh, Y.• N. Inoue, N. Nishikawa and I. Yoshino.
Physica Polonica A 94.509. 1997. Third Symposium on Physics and Applica-
Sadowski, 1.. 1.Z. Domagala. 1. Bllk-Misiuk. tion of Spin-Related Phenomena in Semiconduc-
S. Koleswicki, K. Siatek, 1. Kanski, L. liver tors. 17-18 Nov. 1997. Sendai, Japan. p. 23.
and I. Strom. 2000. I. Vac. Sci. Technol. B 18. Satoh, Y.• D. Okazawa, A. Nagashima and I. Yoshino.
1697. 2001, Physics E 10,196.
Sadowski, I .• P. Mathieu. P. Svedlindh, 1.Z. Domagala, Schairer. W.• and M. Schmidt. 1974. Phys. Rev. B 10.
1. Bak-Misiuk, K. Swiatek. M. Karlsteen, I. Kanski, 2501.
L.Uver, H. Asklund and U. Sodervall, 2001a, Appl. Schliemarm,I., I. Konig. H.-H. Lin and A.H. MacDon-
Phys. Len. 78. 3271. ald. 200la. Appl. Phys. Len. 78. 1550.
Sadowski. I., P. Mathieu. P. Svedlindh, J.M. Karlsteen, Schliemann, 1.•1. Konig and A.H. MacDonald, 200lb.
I. Kanski, L.Uver. H. Asklund. K. Swil4tek,J.Z. D0- Phys. Rev. B 64. 165201.
magala, I. Bak-Misiuk and D. Maude. 2001b. Phys- Schon. G.M .• W. Faschinger and L.W. Molenkamp.
ica E 10, 181. 2001. Appl. Phys. Len. 79. 1807.
Sadowski. J.• E. Vanelle, D.S. Yee, D. Hrabovski, Schneider. 1., U. Kaufmann. W. Wilkening.
1. Kanski and L.Uver. 200k. presented at XXX in- M. Baeumer and F. Kohl. 1987. Phys. Rev.
ternational School on the Physics of Semiconduct- Len. 59, 240.
ing Compounds "Jaszowiec 2001". June 1-8.2001, Schulthess, T.C.• and W.H. Butler, 2001.1. Appl. Phys.
Ustrori-Jaszowiec, Poland. 89.7021.
86 F. MATSVKVRA et aI.

Semenov, Y.G., and V.A. Stephanovich, 2001, cond- Szczytko, J., W. Mac, A. Stachow, A. Twardowski,
matlO106406. P. Becla and J. Tworzydlo, 19%, Solid State Co-
Shen, A.• H. Ohno, F. Matsukura, Y. Sugawara, N. Ak- munn. 99, 927.
iba, T. Kuroiwa, A. Oiwa, A. Endo, S. Katsumoto Szczytko, J., A. Twardowski, K. Swiatek, M. PaI-
and Y. lye. 1997a, J. Cryst. Growth 175/176. 1069. czewska, M. Tanaka, T. Hayashi and K. Ando,
Shen, A., Y. Horikoshi, H. Ohno and S.P. Guo, 1997b. 1999a. Phys. Rev. B 60, 8304.
Appl. Phys. Lett. 71, 1540. Szczytko, J., W. Mac, A. Twardowski, F. Matsukura
Shen, A.. F. Matsukura, Y. Sugawara, T. Kuroiwa, and H. Ohno, 1999b, Phys. Rev. B 59.12935.
H. Ohno, A. Oiwa, A. Endo, S. Katsumoto and Szczytko, J.. A. Twardowski. M. Palczewska,
Y. lye. 1997c, Appl. Surf. Sci. 113/114. 183. R. Jablonski. 1. Furdyna and H. Munekata, 2001a,
Shen, A.. H. Ohno, F. Matsukura, Y. Sugawara, Phys. Rev. B 63. 085315.
Y. Ohno, N. Akiba and T. Kuroiwa, 1997d. Jpn. J. Szczytko, J., W. Bardyszewski and A. Twardowski.
Appl. Phys. 36. L73. 200lb. Phys. Rev. B 64.075306.
Shen, A.. F. Matsukura, S.P. Guo. Y. Sugawara, Szuszkiewicz, W.• E. Dynowska, B. Hennion, F. Ott.
H. Ohno, M. Tani, H. Abe and H.C. Liu, 1999, J. M. Jouanne, J.-F. Morhange, M. Karlsteen and
Cryst. Growth 201/202. 673. J. Sadowski, 2001. Acta Phys. Polon A. 100. 335.
Shibata. N., A. Ohki and A. Katsui, 1988. J. Cryst. Takamura, K.. F. Matsukura, Y. Ohno and H. Ohno,
Growth 93. 703. 2001. J. Appl. Phys. 89. 7024.
Shioda, R. K. Ando, T. Hayashi and M. Tanaka. 1998,
Tanaka, M., 1998. Physica E 2.372.
Phys. Rev. B 58.1100.
Tanaka, M., and Y. Higo, 2001, Phys. Rev. Lett. 87.
Shirai. M., 2001. Physica 10. 143.
026602.
Shirai. M.• T. Ogawa, I. Kitagawa and N. Suzuki. 1998.
Tazima, M., K. Yamamoto, D. Okazawa, A. Nagashima
J. Magn. Magn. Mater. 177-181, 1383.
and 1. Yoshino, 2001a. Physica E 10. 186.
Shimizu. H., T. Hayashi. M. Takenaka, Y. Nakano,
Tazima, M., A. Nagashima, K. Yamamoto. D. Okazawa
M. Tanaka and K. Ando, 1998. Fourth Symposium
and J. Yoshino, 2001 b. Extend Abstracts (The 48th
on Physics and Application of Spin Related Phe-
Spring Meeting. The Japan Society of Applied
nomena in Semiconductors. 3-4 Dec. 1998, Sendai,
Physics) p. 1322.
Japan. p. 53.
Theodoropolpu, N., A.F. Hebard. M.E. Overberg,
Shimizu. H.• T. Hayashi, T. Nishinaga and M. Tanaka,
C.R. Abernathy, SJ. Pearton, S.N.G. Chu and
1999. Appl. Phys. Lett. 74. 398.
RG. Wilson. 2001a. Appl, Phys. Lett. 78. 3475.
Shono, T., T. Hasegawa, T. Fukumura, F. Matsukura
Theodoropolpu, N.. A.F. Hebard. S.N.G. Chu,
and H. Ohno, 2000, Appl. Phys. Lett. 77, 1363.
Slupinski, T.• H. Munekata and A. Oiwa, 2002a, Appl. M.E. Overberg, C.R. Abernathy. SJ. Pearton,
R.G. Wilson and J.M. Zavada. 200lb. Appl. Phys.
Phys. Lett. SO. 1592.
Slupinski, T., A. Oiwa, S. Yanagi and H. Munekata, Lett. 79. 3452.
2002b, J. Cryst. Growth 237-239, 1326. Theodoropolpu, N.• A.F. Hebard, M.E. Overberg,
Sonoda, S., S. Shimizu. T. Sasaki. Y. Yamamoto and C.R. Abernathy, SJ. Pearton, S.N.G. Chu and
H. Hori, 2002. J. Cryst. Growth 237-239, 1358. RG. Wilson, 2002, cond-matl0201492.
Soo, Y.L., S.W. Huang, Z.H. Ming, Y.H. Kao, Tokura, Y.. 2000. in: Clossal Magnetoresistive Oxides,
H. Munekata and L.L. Chung. 1996, Phys. Rev. B ed. Y. Tokura (Gordon & Breach Science Publish-
58.4905. ers. London).
Soo, Y.L., G. Kioseoglou, S. Huang, S. Kim. Y.H. Kao. Tokura, Y.• and Y. Tomioka, 1999, J. Magn. Magn.
Y. Takatani, S. Haneda and H. Munekara, 2001 a, Matter. 200, 1.
Phys. Rev. B 63. 195209. Tsuruoka, T., R. Tanimoto, N. Tachikawa, S. Ushioda,
Soo, Y.L.. G. Kioseoglou, S. Kim. S. Huang. Y.H. Kao. F. Matsukura and H. Ohno, 2000, in: Proceedings
S. Kuwabara, S. Owa, T. Kondo and H. Munekara, 25th International Conference on Physics of Semi-
200lb. Appl. Phys, Lett. 79. 3926. conductors. Sep. 17-22. Osaka. Japan. eds N. Miura
Story. T., R.R Galazka, R.B. Frankel and P.A. Wolf, and T. Ando (Springer-Verlag. Berlin. Heidelberg.
1986. Phys. Rev. Lett. 56,777. 2001) p. 245.
Sugawara, Y.. N. Akiba, T. Kuroiwa, F. Matsukura, Tsuruoka, T., R. Tanimoto. N. Tachikawa, S. Ushioda,
A. Shen and H. Ohno, 1997, presented at The 300 F. Matsukura and H. Ohno, 2002, Solid State Com-
Symposium on the Physics and Application of Spin- mun. 121, 79.
Related Phenomena in Semiconductors. Sendai, Veda, K., H. Tabata and T. Kawai. 2001, Appl. Phys.
Japan. Nov. 17-18. 1997. pp. 30-33. Lett. 79, 988.
III-V FERROMAGNETIC SEMICONDUCTORS 87

Ueda, S.• S. Imada, T. Muto, Y. Saitoh, S. Suga, F. Mat- Physics and Application of Spin Related Phenom-
sukura and H. Ohno, 2001, Physica E 10. 210. ena in Semiconductors. 17-18 Dec. 2001, Yoko-
Van Esch, A.• W Van Roy. J. De Boeck. I. Franeois, hama, Japan. p. 60.
H. Bender. G. Borghs, L. Van Bockstal, R. Bogaerts Yamamoto, T.• and H. Katayama-Yoshida. 1999. Fifth
and F. Herlach, 1995. Mater. Sci. Forum 182-184. Symposium on Physics and Application of Spin Re-
623. lated Phenomena in Semiconductors. 16-17 Dec.
Van Esch, A.• L. Van Bockstal, J. De Boeck. G. Ver- 1999. Sendai, Japan. p. 170.
banck, A.S. Van Sternbergen, P.J. Wellman. B. Gri- Yanagi, S.• H. Munekata, Y. Kitamoto, A. Oiwa and
etens, R. Begaerts, F. Herlach and G. Borghs, 1997. T. Slupinski, 2002. J. Appl. Phys. 91. 7902.
Phys. Rev. B 56. 13103. Yang. J.. H. Yasuda, S. Wang, F. Matsukura, Y. Ohno
Van Schlifgaarde. M.• and O.N. Mryasov, 2001, Phys. and H. Ohno, 2000. Appl. Surf. Sci. 166. 242.
Rev. B 63. 233205. Yasuda. H.• and H. Ohno, 1999, Appl. Phys. Lett. 74,
Van Stapele, R.P.. 1982. in: Handbook on Mag- 3275.
netic Materials. Vol. 3. ed. K.H.J. Bushow (North- Yeo J., Y.B. Kim. A.J. Millis, B.I. Shraiman, P. Ma-
Holland. Amsterdam) p. 603-745. jumdar and Z. Te§anovic. 1999. Phys. Rev. Lett. 83.
Von Molnar. S., H. Munekata, H. Ohno and L.L. 3737.
Chang, 1991. J. Magn. Magn. Mater. 93. 356. Zajac, M.• R. Dorandziriski, J. Gosk, J. Szczytko,
M. Lefeld-Sosnowska, M. Kaminska, A. Twar-
Vnjen, R., E. Yablonovitch, K. Wang. H.W. Jiang.
dowski. M. Palczewska, E. Grzanka and
A. Balandin, V. Roychowdhury, T. Mor and D. Di-
W. G~bicki, 2001a, Appl. Phys. Lett. 78,1276.
Vincenzo, 2000. Phys. Rev. A 62. 012306.
Zajac, M .• J. Gosk, M. Kaminska, A. Twardowski.
Vurgaftman. I.• andJ.R. Meyer. 2001. Phys. Rev. B 64.
T. Szyszko and S. Podsiadlo, 200lb. Appl. Phys.
245207.
Lett. 79. 2432.
Wahl. U.• A. Vantomme, G. Langousche, J.G. Correia,
L. Peralta and The ISOLDE Collaboration. 2001.
Zener, c.. 1951a. Phys. Rev. 81.440.
Zener, C.• 1951b. Phys. Rev. 82,403.
Appl. Phys. Lett. 78. 3217.
Zener. c, 1951c. Phys. Rev. 83. 299.
Walukiewicz. W.• 1988. Phys. Rev. B 37. 4760.
Zhang. F.C.• and T.M. Rice. 1988. Phys. Rev. B 37.
Wei, S.-H .• and A. Zunger, 1993. Phys. Rev. B 48.
3759.
6111.
Zhang, S.• 2000. Phys. Rev. Lett. 85. 393.
Wiley. J.D .• 1975. in: Semiconductors and Sernimetals, Zhao. J.H .• F. Matsukura, K. Takamura, E. Abe,
Vol. 10. eds R.K. Willardson and A.C. Beer (Acad- D. Chiba and H. Ohno, 2001. Appl. Phys. Lett. 79.
emic Press. New York) p. 173. 2276.
Wojtowicz. T.• T. Dietl. M. Sawicki, W. Plesiewicz and Zhao. J.H .• F. Matsukura, E. Abe. D. Chiba, Y. Obno,
J. Jaroszyriski, 1986. Phys. Rev. Lett. 56. 2419. K. Takamura and H. Ohno, 2002. J. Cryst. Growth
Wolf. S.A., 2000. J. Supeconductivity 13. 195. and 237-239. 1349.
other papers in the same issue. Zhao. Y.-J.. WT. Geng, K.T. Park and A.J. Freeman.
Wolf. S.. D.D. Awshalom, R.A. Buhrman. 200la. Phys. Rev. B 64. 035207.
J.M. Daughton. S. von Molnar. M.L. Roukes, Zhao. Y.-J.• WT. Geng, A.J. Freeman and T. Oguchi,
A.Y. Chtchelkanova and D.M. Treger, 2001. 200lb. Phys. Rev. B 63. 201202.
Science 294.1488. Ziese, M.• and M.J. Thornton (eds), 2001. Spin Elec-
Wolff. P.A.. R.N. Bhatt and A.C. Durst. 1996. J. Appl. tronics (Springer. 200 I).
Phys. 79. 5196. Zunger, A.• 1986. in: Solid State Physics. Vol. 39. eds
Yamada. M.• K. Ono, M. Mizuguchi, J. Okabayashi, H. Ehrenreich and D. Turnbull (Academic Press.
T. Mano, M. Oshima and H. Akinaga, 2001. Ex- New York) p. 275.
tend Abstracts (Fall Meeting, The Japan Society of Zutic. I.. J. Fabian and S. Das Sarma, 2001. Phys. Rev.
Applied Physics). p. lOll; Seventh Symposium on B 64. 121201.
chapter 2

MAGNETOELASTICITY IN NANOSCALE
HETEROGENEOUS MAGNETIC MATERIALS

NGUYEN HUU Due


Cryogenic Labotatory, Faculty of Physics, National University of Hanoi
334 Nguyen Ttai, Thanh xuan, Hanoi
Vietnam

p.E. BROMMER
Van der Waals-Zeeman Instituut, Universiteit van Amsterdam
Valckeniersttaat 65, 1018 XE Amsterdam
The Netherlands

Handbook of Magnetic Materials, Vol. 14


Edited by K.H.J. Buschow
e 2002 Elsevier Science B.V. All rights reserved

89
CONTENTS

Abbreviations . . 91
List of symbols . 91
I. Introduction 93
2. Magnetoelastic effects. 95
2.1. Physical background of magnetoelasticity 95
2.2. Symmetry considerations .. 97
2.3. Surface and interface effects 105
3. Determination of magnetostriction of thin films. 106
3.1. The magnetoelastic cantilever method 106
3.2. The strain induced anisotropy method 108
3.3. Magnetostriction in spin valves . . . . 110
3.4. The strain modulated ferromagnetic resonance (SMFMR) method . III
3.5. The secondary electron spin-polarisation spectroscopy (SESPS) . 112
3.6. The strain-induced anisotropy due to the spontaneous strains .. 113
4. Giant magnetostriction in rare-earth-transition metal thin films . . . . 114
4.1. General consideration of magnetism in rare-earth-transition metal alloys . 114
4.2. Magnetostriction of amorphous rare earth based thin films .. 116
4.3. Magnetostriction of nanocrystalline rare earth based thin films. 132
5. Magnetostrictive multilayers . . . . . . . . . . . . . . . . . . . . 139
5.1. Nanocrystalline TbDyFe + ZrlNb multilayers . 139
5.2. Magnetostrictive spring magnet type multilayers (MSMM) 140
5.3. Interface magnetostriction of multilayers 151
6. Magnetoelasticity of rare-earth superlattices . 158
7. Magnetostriction of R-T sandwich films . . . . 163
8. Magnetostriction in nanocrystalline and granular magnetic materials 168
9. Huge magnetostriction in perovskites . 174
10. Potential applications of magnetostrictive materials 185
11. Summary and concluding remarks 190
Acknowledgements 192
References . . . . . 192

90
Abbreviations

a- amorphous
AF anti ferromagnetic
AFM anti ferromagnetic metal
AFI antiferromagnetic insulator
c crystalline
CEF crystalline electric-field
CEMS conversion electron Mossbauer spectra
CG cluster glass (state)
CL charge localisation
CMR colossal magnetoresistance
CO charge-ordered (state)
DE double exchange
EDW extended domain wall
EMD easy magnetisation direction
FM ferromagnetic
FMM ferromagnetic metal
GMR giant magnetoresistance
GMS giant magnetostriction
IT Jahn-Teller
MEMS rnicroelectromechanical system
MOKE magneto-optic Kerr effect
MSMM magnetostrictive spring magnet type multilayer
n nanocrystalline
p polycrysta1line
PMI paramagnetic insulator
R rare-earth
HR heavy rare-earth
LR light rare-earth
SAMR small-angle magnetisation rotation
SESPS secondary electron spin-polarisation spectroscopy
SMFMR strain modulated ferromagnetic resonance
T transition metal

List of symbols

X magnetic susceptibility
A magnetostriction

91
92 N.H. DUe and P.E.BROMMER

v Poisson's ratio
w spontaneous volume magnetostriction
a stress
X).. magnetostrictive susceptibility
Aam magnetostriction of amorphous matrix
Xb magnetoelastic susceptibility
Acr magnetostriction of nanocrystalline grains
Aeff effective magnetostriction
Pi (i=x,y,z) direction cosines of the measured magnetostriction
aj (i=x,y,z) direction cosines of the magnetisation
a] Stevens factor
AS saturation magnetostriction
AS surface/interface magnetostriction
At anisotropic (Joule) magnetostriction
AV volume magnetostriction
A divalent alkaline ions
Aij spin-spin coupling parameter
b elastic coefficient in thin films
B external magnetic field (/-LoH)
B bu1k , By,2 magnetoelastic coefficient in bulk materials
Be~ch exchange field
Bhf hyperfine field
bsurf surface/interface magnetoelastic coefficient in thin films
bv volume magnetoelastic coefficient in thin films
d grain diameter
e volume dilatation
E Young's modulus
gR Lande factor
J quantum number of the total angular momentum
1R the total angular momentum for the 4f ions
K anisotropy constant
MY magnetoelastic tensor
Mj sublattice magnetisation
MR magnetisation of the R sub lattice
MT magnetisation of the T sublattice
Ms spontaneous magnetisation
P volume fraction
t film thickness
TA annealing temperature
Tc Curie temperature
Ts substrate temperature
Zij number of nearest neighbours
r exchange integral
6 sperimagnetic cone angle
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 93

1. Introduction

The origin of magnetoelasticity is the interplay of the elastic properties of a material


and its magnetic state. The magnetoelastic coupling terms in the free energy give rise
to (extra) strains in a magnetic substance, i.e. the volume magnetostriction and the
Joule magnetostriction (anisotropic changes in linear dimensions). The concurring change
of the magnetic state can be induced either by a temperature variation (spontaneous
magnetostriction) or by application of a magnetic field (forced magnetostriction). The
magnetostriction varies from nearly 1% in rare-earth based intermetallic compounds to
almost zero for iron based amorphous and nanocrystalline alloys. Also the elastic moduli,
can be affected, e.g. the !l E effect. Conversely, an imposed strain may cause extra magnetic
anisotropy. Apart from these magnetoelastic coupling effects, a magnetised body may gain
magnetostatic energy by macroscopic, rather small (~I 0- 6 ) , deformations, i.e. the form
effect.
Nanoscale thin films and multilayers, nanocrystalline magnetic materials, granular films,
and amorphous materials have been and still are most attractive topics in the field of basic
and applied magnetic research. The local properties of these nanoscale heterogeneous
magnetic systems do vary in the scale of nanometers, for amorphous materials down to an
atomic scale. One is able to produce nanocrystals with a narrow size distribution, embedded
in an amorphous matrix at approximately constant distance between them. Granular solids
composed of magnetic particles diluted in a non-magnetic matrix are also heterogeneous
systems. Going down to an atomic scale, amorphous magnets can be considered as (on
nanoscale uniform) heterogeneous magnetic systems. The local anisotropy is assumed to
vary in strength and orientation with a uniform correlation length of only a few angstroms.
The exchange coupling between the adjacent magnetic moments shows an analogous
variation.
Heterogeneous materials exhibit unusual magnetic properties which do not occur, or
are negligible, in bulk materials: e.g. surface (interface) magnetic anisotropy and surface
(interface) magnetostrictive strains, giant magnetoresistance and so on. An important key
to understanding such effects is the knowledge of the magnetic behaviour at the interfaces,
where questions concerning the magnetic moments, the type of magnetic exchange
coupling and the spin orientation may be answered. The local atomic arrangement at the
interface differs strongly from that in the bulk. The local symmetry is lowered, some
interactions are changed or are missing. Indeed, the interface atoms may be considered
to form a new phase and any property characteristic of this phase may become a dominant
one for the whole system. It becomes particularly evident in the case of interfacial
magnetostriction which can lead to a decrease (almost to zero) or to an increase (over
the bulk value) of the resulting magnetostriction of the nanoscale system. In this handbook
series, magnetism in ultrathin transition metal films and muItilayers (Gradmann 1993) and
magnetism in artificial metallic superlattices (Rhyne and Erwin 1995) was presented in
the volumes 7 and 8, respectively. Herzer (1997) reviewed nanocrystalline soft magnetic
alloys in volume 10. Giant magnetoresistance in multilayers was presented in volume 12
by Barthelemy et al. (1999), and NMR in multilayers by Riedi et al. (1999).
The present chapter deals with the magnetoelasticity of heterogeneous materials.
Generally, the dimensions of a magnetostrictive material change when the material is
94 N.H. Due and P.E. BROMMER

subjected to a change in magnetic field. Hence. magnetostrictive materials can be applied


in transducers (as well as piezoelectric and shape memory ones). which directly convert
electrical energy into mechanical energy. They are useful in the manufacture of sensors.
actuators. controllers. force and displacement as well as other electro-acoustic devices.
For these applications. transducer materials in the form of thin films are of special
interest because cost-effective mass production is possible. compatible to microsystem
processing technologies. In addition. magnetostrictive thin films are particularly promising
as microactuator elements like cantilevers or membranes, since they combine high-energy
output. high-frequency and remote-control operation. Due to this potential, interest in such
giant magnetostrictive thin films has rapidly grown over the past few years. Owing to
the specifications related with applications in microelectromechanical systems (MEMS).
materials research has been focused on thin-film materials showing giant magnetostriction
(GMS) in combination with soft magnetic properties. Here, the rare-earth metals and
rare-earth-iron intermetallic compounds have been shown to generate very interesting
systems. In 1971. A.E. Clark at the Naval Ordnance Laboratory (NOL). now Naval Surface
Warfare Center discovered that bulk TbFe2 (Terfenol: Ter for Tb, fe for Fe. nol for
NOL as above) has the highest room-temperature magnetostriction, and his Terfenol-D
(D for Dy: TbxDYJ-xFe2. where x ~ 0.3) is still the best known, exploiting the huge
magnetostriction in combination with reduced magnetocrystalline anisotropy (Clark and
Belson 1972a, 1972b. 1972c). Wang et al. (2000) studied the magnetostriction and the
magnetisation process in a TbO.27DYO.73Fe2 single crystal. Indeed. as a tradition, research
on giant magnetostrictive thin films was focused on iron based rare-earth alloys. Later.
however, important progress has been made with the development of rare-earth-cobalt
alloys at the 'Laboratoire Louis Neel' in Grenoble (Betz et al. 1999). Due et al. (2001b)
proposed to name the amorphous Tb-Co compounds (with composition near to TbC02) as
'a-TerCoNeeI', by an obvious analogy to the name TerFeNol above. Still better results were
obtained by combining Fe and Co. Due et al. (1996) achieved a record magnetostriction
of 1020 x 10-6 on 'a-Terfeconeel', i.e. an amorphous Tb-(Fe,Co h thin film. whereas Due
et al. (2oooa) extended the study to 'a-Terfeconeel-D', i.e. amorphous (Tb,Dy)(Fe,Coh
films. In Vietnam's capital Hanoi. as a variant, the amorphous compound with composition
Tb(FeO.55C00.45) 1.5. to be named 'a-Terfecohan', was developed (Danh et al. 2000; Due et
al. 2ooob). Composite materials are known to exhibit sometimes outstanding properties,
which cannot be predicted simply from the behaviour of the constituents. Actually, a
very high magnetostrictive susceptibility has been observed in TbCo/FeCo and TbFe/Fe
multilayers (Quandt et al. 1997a, 1997b).
The tlE effect (the change in the value of Young's modulus E) is perhaps the best known
example of a change of the elastic moduli caused by magnetoelasticity (Du Tremolet de
Lacheisserie 1993). The tl E effect plays an important role in various practical applications
(Hernando et al. 1988; Gibbs 1992; Gibbs et al. 1996, 1997) including such spectacular
MEMS as micromachines and actuators using thin film and multilayer magnetostrictive
materials (Gibbs et al. 1996, 1997). For more insight on this effect and its applications,
however, we refer to the above-mentioned publications.
In this handbook series, previous work on giant magnetostriction (which was mainly
focused on rare-earth compounds) is excellently reviewed by Clark (1980) on "Magne-
tostrictive Rare-Earth-Fer Compounds", by Morin and Schmitt (1990) on "Quadrupolar
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 95

Interactions and Magneto-Elastic Effects in Rare-Earth Intennetallic Compounds" and by


Andreev on "Thermal Expansion Anomalies and Spontaneous Magnetostriction in Rare-
Earth Intennetallics with Cobalt and Iron" (Andreev 1996). Noteworthy are also the more
recent review papers by Szymczak (1997) on "Mechanisms responsible for magnetostric-
tion in heterogeneous magnetic systems" and on "From almost zero magnetostriction to
giant magnetostrictive effects: recent results" (Szymczak 1999). We mention also the re-
view report by Sander (1999) on 'The correlation between mechanical stress and magnetic
anisotropy in ultrathin films". Szymczak (2000) focuses attention on "Giant magnetostric-
tive effects in magnetic oxides". Much detailed practical information can be found in the
"Handbook of Giant Magnetostrictive Materials" by Engdahl (1999). The review of Due
(2001) "Giant magnetostriction in lanthanide-transition metal thin films" deals with amor-
phous thin films. Therefore, in the present chapter, this subject is treated concisely, fo-
cussing attention on more recent results.
The present chapter is organised as follows. After this introduction, section 2 is devoted
to the description of magnetoelastic phenomena. Problems in the determination of the mag-
netostrictive coefficients of thin films are presented in section 3. In secttion 4, magnetism
and magnetostriction in amorphous rare earth-transition metal alloys are summarised, and
the possibilities to develop giant magnetostriction in nanocrystalline thin films are dis-
cussed. Sections from 5 to 8 deal with an overview of the research of magnetostriction in
artificially structured materials, such as multilayers, rare-earth superlattices, R-T sandwich
films, nanocrystalline soft magnetic materials and granular solids. Doped manganese ox-
ides RI-xAxMn03, (RI-xAxhMn207, and cobaltates RI-xAxC003 have been shown to
exhibit colossal magnetoresistance. Their magnetostriction is presented in section 9. Some
actual and potential applications of magnetostrictive films in MEMS are briefly discussed
in section 10. Finally, a summary and concluding remarks are presented in section 11.

2. Magnetoelastic effects

2.1. Physical background of magnetoelasticity


Usually, a discussion of magnetoelastic effects is based on the minimisation of a free en-
ergy expression, which is the sum of the magnetostatic energy and the 'internal' free en-
ergy. The magnetostatic energy originates from the (long-range) dipolar interactions, giv-
ing rise to demagnetising fields, shape anisotropy and 'magnetic forces'. The minimisation
of this energy leads to deformations, which depend on the geometry of the sample (hence
the name form effect). These deformations can be inhomogeneous even when the mag-
netisation is uniform, and can contribute volume changes as well as anisotropic strains.
This effect exists in all magnetic materials even in the case of vanishing magnetoelastic
coupling. Its magnitude, however, is always small ( ~ 10-6 ).
As shown in the next subsection, the 'internal' magnetoelastic effects are described by
expanding the free energy as a sum of the elastic free energy, the magnetic free energy
and the magnetoelastic coupling terms. Symmetry considerations restrict the number of
possible terms (see e.g. Becker and Doring 1939). It is necessary, however, to consider the
underlying physical mechanisms too. The isotropic Heisenberg exchange interaction for
localised spins, for example, can be written as E j »i (-2AijSiSj), where the interaction
96 N.H. Due and P.E. BROMMER

parameters A;j may be assumed to be functions of the interatomic distance vector rij.
Taking into account the variation of r;j with a certain strain component e, one finds
magnetoelastic coupling terms like

(1)

Such a contribution can be expected not only for the volume magnetostriction but also
for other strain components (see e.g. the magnetostriction contribution "-'A~'O in eq. (4)
below). The temperature (and field) dependence of the magnetoelastic coupling parameter
b is mainly determined by the correlation function (S;Sj). Analogously, the anisotropic
Joule magnetostriction may originate from crystal field effects (also responsible for the
magnetic anisotropy, if the symmetry is low enough; see introduction) and from the
pseudo-dipolar exchange interactions, which are anisotropic and vary rapidly with the
interatomic distance (see e.g. Du Tremolet de Lacheisserie (1999) for explicit expressions).
The angular dependence is determined by the corresponding correlation function, and so
is the temperature dependence of the magnetostriction and that of the magnetic anisotropy.
Notice, however, that, by cancellation of terms due to symmetry, the (local or total)
magnetic anisotropy may vanish, where the magnetostriction does not. For itinerant-
electron systems, in addition, the band splitting may cause large deformations too. For
instance, in Re02 based compounds, the (volume) magnetostriction can reach values up to
"-'10- 2 , see Due and Goto (1999), Due and Brommer (1999).
An important physical origin of the magnetostriction in rare-earth-transition metal
based materials is indicated in fig. I. Here, the deformations are thought to be due to the
anisotropy in the orbital part. For 3d-atoms, the large extension of the 3d-wave function
invokes a strong interaction with the crystalline electric field, leading to (partly) quenching
of the orbital moment. The spin-orbit coupling is relatively weak ("-'0.015 eV per atom),
and, helped by the stronger exchange interactions ("-'0.1 eV), the spin moments can easily
be rotated into the applied-field direction, leaving the orbitals almost unaffected (fig. la).
The anisotropy and then the magnetostriction is small. For rare-earth ions, the spin-orbit
coupling LS is strong and a rotation of the total moment L + S = J forces simultaneously
a rotation of the orbitals. This results in not only a large magnetocrystalline anisotropy but
also in a deformation of the crystal lattice. This magnetostriction happens to be negative
when the distribution of the 4f-charge distribution is prolate, whereas the magnetostriction
is positive when the charge distribution is oblate. As illustrated in fig. l(b, c), one may
imagine that the surrounding positively charged neighbours are drawn to the negative
charge cloud (see also section 4, and fig. 8).
In the actual thin-film materials, large local strains and large variations in local
atomic arrangement may occur, possibly leading to large magnetic anisotropy's and large
magnetoelastic effects in the as-deposited materials. Nevertheless, even in amorphous thin
films, the local atomic arrangements appear to resemble those of the crystalline materials,
the more so after suitable annealing. Some relaxation may occur, however. Moreover,
one should take into account the possibility that anisotropic pair ordering is induced or
originates from the fabrication (e.g. sputtering) process. Models were worked out, for
example, by Mergel et al. (1993) and Huang et al. (1995). Anyhow, such considerations
do support the idea to look upon the whole range of materials, from crystalline down to
amorphous, as being governed by basically the same physical interactions.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 97

ED
ED 1,1tfi\ ED (a)
weak
~, spin-orbit coupling

ED

(b)
negative Joule
magnetostriction

(c:)
positive Joule
magnetostriction

Fig. I. Schematic representation of the phenomena of magnetostriction. The surrounding atoms, schematisc:d as
positive charges. are displaced from their initial symmetrical position (open circles) to their final strained positions
(black circles) due to the electrostatic interactions with the:aspherical electron distribution.

2.2. Symmetry considerations


With respect to a (fictitious) non-magnetic, unstrained state, the free energy is written as the
sum of an elastic contribution, a magnetic contribution and the magnetoelastic interaction.
The elastic free energy is minimal for vanishing strains, and here, for simplicity, its
expansion is restricted to terms of second order in the (six) different elements of the
(symmetrized) strain tensor: exx. eyy. ezz, exy. eyz. exz. The magnetic contribution
is described by an expansion in terms of a; (i = 1,2.3, corresponding to x, y, Z,
respectively), i.e. the direction cosines of the magnetisation direction. For simplicity, we
restrict the detailed discussion to the (first-order) magnetoelastic coupling terms, which
are linear in the strains and are of up to second order in {a;}. Taking into account
some underlying quantummechanical mechanisms, Callen and Callen (1963, 1965) have
presented an instructive group-theoretical analysis, thereby establishing a notation, which
has been used by many authors (many of them adding their own flavour, mostly adapting
the normalisation; see e.g. Du Trernolet de Lacheisserie (1993, 1995». A brief outline of
98 N.H. Due and P.E. BROMMER

the reasoning may establish and clarify the notation. Let us start by recalling that, for given
strains, the linear expansion, measured in the direction fJ = (f3I, f32, f33) is given by

!:i.f./f. = Exxf3t + 2Exyf31 f32 + eye!. (2)

This expression must be independent of the choice of the co-ordinate system. In line
with the general usage to look upon the 6 independent elements of the symmet-
ric strain (stress) tensor as components of a vector, we may elucidate this indepen-
dence of co-ordinate system by writing eq. (2) as the scalar product of the vectors
[E xx, Eyy, Ezz, exy.j2, eyz.j2, ezx.j2] and [f3t, f3i, f3j, f31f32.j2, f32f33.j2, f33f31.j2]. The
components of these vectors transform like [x 2, y2, z2, xy.j2, yz.j2, xz.j2], and, hence,
the scalar products transform as:

evidently invariant. This property remains intact, when one applies an orthogonal change
of basis vectors (in the 6-dimensional space). So, introducing the 'volume dilatation'

e = !:i. V/ V = (exx + eyl' + e:J,


we can replace the 'diagonal elements' {E xx , E yy , e:zl by

{(exx + eyy + ed/.j3 = e] .j3, (2ez: - exx - eyy)/.j6 = (e:: - e/3).j(3/2),


(exx - Eyy)/ .j2},

and analogously e.g. {f3r, f3i, f3jl by {I/ .j3, (f3j - 1/3).j(3/2), (f3? - f3i)/ .j21, and so
on. Now, again forming the scalar product yields

!:i.f./f. = e/3 + (3/2)(e:: - e/3)(f3j - 1/3) + O/2)(E xx - Eyy)(f3t - f3i)


+ [2exyf3If32 + eye!.], (3a)

which also can be written as

!:i.f./f. = e /3 + [(e:: - e!3)(f3] - 1/3) + 2exyf31 f32 + cycl.]. (3b)

For a material strained in one direction, one may check that !:i.f./f. equals ez:, the only non-
zero strain component, when the z-direction and the measuring direction are chosen along
the strain direction.
Next, we consider the symmetry operations of the system. The free energy is expanded as
a function of the strains (as defined above) and the corresponding 'harmonic polynomials'
h(uj). The resulting expression must be invariant under the symmetry transformations.
If the symmetry is low enough, one can reduce further the vector space(s) introduced
above, by choosing a suitable basis. The resulting irreducible subspaces are indicated
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 99

by the superscript a when one-dimensional, and by superscripts y, ... , when more-


dimensional. A one-dimensional subspace yields an invariant strain and a corresponding
harmonic polynomial (possibly isotropic), directly. Any product of these invariants
(possibly originating from different irreducible subspaces) forms also an allowed invariant.
For a more-dimensional subspace, e.g. spanned by functions {hi}, one knows from a
fundamental group-theoretical theorem, that the only (second-order) invariant is I;j(hj)2.
The only invariant (bilinear) magnetoelastic coupling terms are I;jeihi ' where lei}
j}
(the strains) and {h must transform according to the same irreducible representation y.
Then, the magnetostrictive strains follow from the minimisation of the free energy
part Y2CYI;j(ei)2 + BYI;jeihj, i.e. ei = -(BY /CY)hi, governed by the (first-order)
'coefficient of magnetostriction' .AX = -(BY / CY), for the 'y combination' Hei}, {hi}].
The free-energy contribution can be written as -Y2(By2/CY)I;jhi2, showing explicitly
the well-known magnetoelastic contribution to the magnetic anisotropy.
As we have seen above, there is always at least one one-dimensional irreducible
subspace, i.e. the invariant trace of the strain tensor, e = exx + e xy + ezz (analogously
x 2 + y2 + Z2 = r 2 and ar
+ a~ + a~ = 1).
For an isotropic material, the space spanned by the five remaining strain components
cannot be reduced further. Consequently, there are only two magnetostriction modes.
The energy density can be written down directly, in principle for either irreducible
representation separately:

F = Y2C ae 2/3 + B a,oe/3 +Y2CY[(e zz -e/3)2 +2e;y +cycl.]

+ By,2[(e zz - e/3)(a~ - 1/3) + 2e xya\a2 + cycl.], (4)

For the magnetoelastic coupling parameters (Ba,o, BY,2), the first superscript indicates the
irreducible representation, the second one the degree of the harmonic polynomial in {ai}.
Notice that the bracketed expressions in eq. (4) can be rewritten in a form analogous to that
in eq. (3a):

(e z: - e/3)2 + cycl. = (3/2)(e zz - e/3)2 + (l/2)(e xx - e yy)2, (Sa)

(e zz - e/3)(a~ - 1/3) + cycl. = (3/2)(e zz - e/3)(a~ - 1/3)

+ (l/2)(e xx - eyy)(ar - aD. (5b)

Moreover, still another fashionable way to write such expressions follows from e.g.:

(3/2)(e zz - e/3)2 = (2/3){e zz - (exx + e yy)/2}2. (5c)

The coupling gives rise to 'magnetic stresses', i.e. the magnetic pressure pmag = Ba,o /3,
the diagonal traceless stress tensor elements (Tx~ag = -BY,2(ar - 1/3) (cycl.) and the
off-diagonal stresses (Txymag = -By·2 a 1a 2 (cycl.). In response to these stresses, the
equilibrium strains can be found by minimising the free energy contributions, resulting
100 N.H. Due and P.E.BROMMER

A,cx,O
z

/i 1.,>0

~-
Ms=O H=O

Fig. 2. The two principle modes of observable magnetostriction for an isotropic magnetic substance.

in e = Aa,O = _Ba,o [c" and {(e u - e/3) = ).Y.2(a~ - 1/3), e xy = Ay·2 a 1a 2 (cycl.j}, with
)'y,2 = _BY'o IcY . Plugging in these results in eq. (3b), we find

Lli/i =A f1
• /3 + )"Y.2[{ (a~
O
- 1/3)(pj - 1/3) + cycl.] + {2a,a2PI P2 + cycl.I],
(6)

Starting from a perfectly demagnetised state (with (af) = (a~) = (a~) = 1/3, (ala2) =
0, and so on), the ideal 'saturated' relative change of length measured along the
field direction (common to the magnetisation direction: a3 = P3 = 1) would be ).,11 =
(Lli/i)"at - Aa,o/3 = 2Ay,2/3 = AS (by definition; subscript S indicates 'saturation').
The relative change in the plane perpendicular to the field would be (Lli/fh =
)".1 = - ~).,s. As the materials always present an anisotropic demagnetised state, it is
necessary to apply the field in two perpendicular directions and determine ).,y,2 =
(All - A.1), and As = 2/3(AII - A.1), independent of the demagnetised state of the mag-
netic materials.
The two principle modes of the magnetostriction ().,a.O and X, or ).,Y,2) introduced above
are illustrated in fig. 2. With respect to the non-magnetic fictitious state, a spherical,
isotropic sample exhibits a relative volume change LlV / V = Aa.O, when it becomes
magnetic. In addition, when one forces the moments to be directed along an applied
magnetic field B, an anisotropic deformation is induced, which transfers the sphere into an
ellipsoid with the same volume.
For a lattice with cubic symmetry, the 6 strain components are grouped into the
one-dimensional representation a: basis e/.,/3, a two-dimensional representation y:
basis {(e zz - e /3).,/(3/2), (e xx - eyy)/ .,/2}, and a three-dimensional one s: basis
{exy.,/2. eyz.,/2, ezx.,/2}. The modes of deformation can be described as follows. The
isotropic mode e/.,/3 '" (exx + eyy + ezz) gives a volume change and does not reduce
the symmetry. The anisotropic y-modes reduce the cubic symmetry by varying the lattice
parameters without modifying either the volume or the angles. Finally, the s-modes reduce
the symmetry by shearing without changing either the volume or the lattice parameters (see
fig. 3). Notice that, in this notation, the co-ordinate system should coincide now with the
cubic axes.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 101

---
TJ LT'
_..1
I I I I
I I
I I I I
I I
,< --- I I

Ala 'A' At
A24

Fig. 3. The normal modes of deformation and the corresponding magnetostriction modes for cubic and uniaxial
symmetry.

The different contributions to the free energy can be written down immediately,
by analogy to the procedure followed above. Here, we give the contribution of the
magnetoelastic coupling to the energy density in the notation of Du Tremolet de
Lacheisserie (J 993, 1995):

FmetlV = ( B;'O) (Exx + f yy + Ell)

+ By,z[23( Ezz - Exx + EYY ) (z


2 a3 -
ar +2 a~)
+ ~(ExX - Eyy)(ar - aD]

+ 2B E'Z(E
YZaz a3 + Ezxa3a) + Exya)az), (7a)

Notice the "splitting" of the (isotropic) BY'z in the two parts BY'z and Bd. A similar
splitting occurs for the elastic energy density, of course:
102 N.H. Due and P.E. BROMMER

1 <ll 2
Fel/V= -c -(cxx+cvv+c:z) + ...
2 3 --

I
+ 2c
y[2(
3" czz -
cxx+cyy)2 1
2 + 2(c X X - cyy)
2]
(7b)

The group-theoretical stiffness parameters can be expressed in the conventional (cubic)


elastic stiffness coefficients:

The material is (elastically) isotropic when c" = cY •


The magnetostriction in the cubic symmetry can be described as

A = ~Aa.o + A .2
y
G(a~ - a
f;
a~)(~~ - ~r; ~i) + ~(af - a~)(~r - ~i) ]
+ 2At:.2(a2a3~2~3 +a3al~3~1 +ala2~1~2) +.... (8)

Here {~;} represent the direction cosines of the measurement direction (of A = ~l/l) with
respect to the crystal axes.
The first-order coefficients ofmagnetostriction of cubic materials are then

Now, the anisotropic Joule magnetostriction is essentially described by the two coefficients
p.2 and A£,2. The familiar magnetostriction constants AIOO and Alii represent the relative
change in length measured in the [100] and [Ill] direction, respectively, when the
magnetisation is directed along the named measuring direction. They are related to the
coefficients given above by:

AIOO = (2/3)A y •2 and Alii = (2/3)A£,2 (see also fig. 3).

For uniaxial (hexagonal) symmetry the 6 strain components are subdivided in two
(invariant) one-dimensional subsets (indicated by the superscript a, and subscripts 1 and
2 for the volume dilatation and the axial deformation, respectively), and two different
two-dimensional subsets, indicated by y for deformations in the (hexagonal) plane, and
by e for 'skew' deformations. These modes are also depicted in fig. 3. In this case, the
magnetostriction can be expressed as

A= ~A<l
3 I
·o +A<l.O(R2
2"'3
_~)3 + [~A<l3.2 I
+ ~A<l
2 2
.2(R2 "'3
_ ~)](a2
3 3
_~) 3

+ ).Y'2[~(~r - ~i)(ar - aD + 2ala2~1 ~2 ]


+ 2)..t:.2(a2a3~2~3 + a3 al ~3~d. (9)
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 103

Notice here the uniaxial deformation, A;'O, independent of the direction of the magnetisa-
tion, and the contribution to the volume magnetostriction, A~,2, which does depend on the
magnetisation direction.
Experimentally, one often tries to measure the magnetoelastic coupling parameters by
measuring the anisotropy induced by a tensile stress, say, applied to an isotropic 'ribbon'
(Villari effect). Choosing the x-axis along the direction of the applied stress a , we have
only one non-zero component of the stress tensor: U x x = o . In the notation established
above, we write the extra energy terms as -UxxE xx = -UExx = -ue/3 - (uj3){2(Exx -
ej3) - (Ezz - ej3) - (Eyy - ej3)}, whence the extra strains e =ujc" ,and (exx - ej3) =
2u j3c Y, (Eyy - ej3) = (ezz - ej3) = -u j3c Y, giving rise to the magnetoelastic coupling
contributions [B U,Oej3 = ]B,,·ouj3c U = -AU,Ouj3, and [BY,2(e x x - ej3)(ar - Ij3) +
cycl. = (3j2)By,2(E xx - ej3)(ar - Ij3) = ] By,2(u jcY)(af - Ij3) = -Ay,2u (a r -
Ij3) = -(3A suj2)(af - Ij3). In other words, the tensile stress induces a uniaxial
anisotropy

(10)

In performing such experiments on isotropic materials, one is accustomed to express


the elastic stiffness parameters in the experimentally more readily accessible 'technical'
parameters E (Young's modulus) and v (Poisson ratio). The relative change in length, in
the direction of the tensile stress a is, by definition, given by (6.1jl)1I = a j E, whereas
v = I(1:11jlhj(1:11jl) I I· For several magnetostrictive films and substrates, E and v values
are listed in table I. Some useful relations are:

E = (CII - cl2)(1+ v) = (1 + v)c Y = 3c"c Yj(2c" + c Y) = c"(1 - 2v),


v = cl2j(CII + Cl2) = (c" - c Y)j(2c" + c Y ) ; I - v = (c" + 2c Y)j(2c" + c"),

Since thin magnetic films are deposited on a non-magnetic substrate, the actual strains,
and thus the induced anisotropy, depend on the magnetoelastic coupling coefficients of

TABLE I
Young's modulus (E) and Poisson's ratio (v) for several magnetostrietive films and different substrates

Material E (GPa) v References

a-R-Fe 40 0.4 Wada et aI. (l997e)


a-R-Fe 50 0.3 Quandt (997)
n-R-Fe 80 0.3 Quandt (997)
a-R-Co 80 0.31 Due et aI. (1996)
TbFeJFe 80-260 Quandt and Ludwig (1999)
W 345 0.28 Wada et aI. (l997e)
Ta 186 0.3 Wada et aI. (l997e)
Other metals ~200 0.31 Betz (1997)
Glass 72 0.21 Due et al, (1996)
Si [1(0) 130.19 0.278 Betz (1997)
Si (110) 169.16 0.037 Betz (1997)
\04 N.H. Due and P.E. BROMMER

the film, and on the elastic properties of the substrate as well as on those of the film.
Choosing the z-axis to be normal to the planar surface of the film (and substrate), the strains
exx, eyy , and exy in the magnetic thin film are imposed by the corresponding strains in the
substrate (more precisely: the equality of the strains is imposed as boundary condition).
This situation is treated for instance by Du Tremolet de Lacheisserie (1995). Notice also
the treatment of the case of hexagonal symmetry by Gutjahr-Loser et al. (2000).
In the experiment discussed above, for instance, the tensile stress a is applied (along
the x-direction) to the substrate. For an amorphous (not extremely thin) film neglecting
surface effects, one finds then two contributions to the induced anisotropy:

3/2(e xx +eyy)(a~ - 1/3){-By·2c" J(c" +2cY ) }


+ By,2{Y2(e xx - eyy)(a? - aD + 2exyala2}. (lla)

Neglecting the influence of the film, we find for the strains in the substrate (and thus also
in the film): exy = 0, and

(exx + eyy) = a(l - vs ) / E s, (exx - e yy) = a(l + vs ) / E s


(subscript s means 'substrate').

Recalling )"y,2 = -By,2/c Y , 3c Y c" [tc" + 2c Y ) = Er/(l - vr) and c Y = ErI(I + vr)
(subscript f means 'film'), one finds

I
Find(a) = 2aAy·2(Er/Es){(a5 - 1/3)(1- 1.1,)/0- vr)

- (a? - aDO + vs)/O + vr)}. (lIb)

For E r = E sand vr = v.. eq. (II b) is reduced to eq. (0). In case the magnetisation vector
is confined to the film plane we have

and in case the magnetisation vector is confined to the xz-plane (a2 = 0)

Find(a) = Kind(a)(a5 - const)


=aA y,2(Er/ Es){[(l - vsvr)/(I - v?)]a5 - const]. (lId)

The elastic properties of the substrate can be determined more accurately than those of the
film. Hence, in principle, the magnetoelastic coupling parameter By,2 can be determined
more accurately than the magnetostriction coefficient Ay •2 • An analogous conclusion can
be drawn with respect to 'bending' experiments (see e.g. section 3.1: the magnetoelastic
cantilever method).
One may note that the relation between the actual surface layer (a few atomic layers, see
next subsection) and the 'film' is quite analogous to the relation between the 'film' and the
'substrate'. Thus, the above derivations can be applied in that case too.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 105

Finally, we remark that for the higher-order coupling terms, i.e. non-linear in the
strains, analogous symmetry considerations lead to the correct 'group-theoretical' invariant
magnetoelastic energy contributions (see e.g. Du Tremolet de Lacheisserie (1993». In fact,
it has been demonstrated (see e.g. Sander (1999» that, in accordance with the suggestion
of O'Handley and Sun (1992) for epitaxially grown films, second-order magnetoelastic
coupling parameters must be taken into account because of the large 'epitaxial' strains
due to the mismatch between substrate and film. Fiihnle and Komelj (2000; see also
references therein) proposed a way to determine these second-order parameters separately
by measuring magnetostrictive stress and magnetic anisotropy in combination.

2.3. Surface and interface effects


The discussion up to now is satisfactory for bulk magnetic materials, even when ribbon-
shaped, having a thickness of some micrometers, still very substantial as compared with
the interatomic distances. For nanoscale thin films and multilayers, however, surface
and interface effects are important. Neel (1953, 1954) recognised that at the surface,
the breaking of the symmetry (leading to lower co-ordination numbers and 'missing'
pair interactions) would give rise to surface magnetic anisotropy. Within localised-spin
models, the surface magnetic anisotropy is associated with effects occurring only at
the surface planes. Within itinerant-electron models, the changes in magnetic anisotropy
due to the surface are related to electronic states extending into the interior of the
magnetic materials. In an analogous way, the (magneto)elastic coefficients can be affected.
Moreover, as pointed out by Du Tremolet de Lacheisserie (1995), the lower symmetry at
the surfaces may necessitate extra (magneto)elastic coupling coefficients. Up to now, most
theoretical models for surface magnetostriction are still rather phenomenological, although
Freeman's group appears to have developed a promising simplification such that 'ab initio'
calculations of the surface or interface magnetostriction become feasible (see references in
Szymczak 1997; Shick et al. 1998; Wu et al. 1998; Wu and Freeman 1999 and Freeman
et al. 1999). Otherwise, the problem of surface magnetostriction can be developed in the
spirit of the Neel model of the surface anisotropy (Zuberek et al. 1994). Detailed theoretical
calculations of the surface contribution to the magnetoelastic energy have been performed
in the framework of the (pseudo- )dipolar pair model (Szumiata et al. 1993, 1999). Du
Tremolet de Lacheisserie and McGrath (1995) have shown that single-ion terms can also
be incorporated in such an approach.
The relative influence of the surface (or interface) effects, of course, must decrease
with increasing thickness t of the layer(s). Since the surface effects contribute 'per unit
surface area', one defines, for the layer, effective parameters such that Beff , or beff, equals
Bbulk + 2b surf / t, Here, the factor 2 is put in, because a layer has two surfaces. In practice,
this simple 1/ t dependence works satisfactorily. For nanocrystallites, both the volume
fraction and the 'volume to surface ratio' of the crystallites (i.e. their radius) must be taken
into account (see also section 8). In connection to these effects, (non-linear) contributions
to the magnetoelastic coefficients due to surface strains and surface roughness are expected
to be considerable.
106 N.H. Due and P.E.BROMMER

3. Determination of magnetostriction of thin films

For bulk ferromagnetic materials, the magnetostriction can be determined directly by


measuring the relative deformation in one direction as a function of the magnetic field:
"A(H) = !!..l/l. Thin magnetic films, however, are deposited on the planar surface of a
non-magnetic substrate. As discussed in the preceding section, in such bimorphs, the
observed strains depend on the magnetoelastic coupling coefficients of the film, as well
as on the elastic properties of the substrate and those of the film. Moreover, in case
surface or interface contributions become important, a larger number of coefficients is
needed to describe the magnetoelastic effects because of the symmetry lowering (see
preceding section). The experimental methods treated below do not permit to determine all
of the magnetoelastic-coupling coefficients. Nevertheless, quantities relevant for technical
applications can be determined experimentally, of course.
Various experimental methods have been developed for investigating the magnetoelastic
properties of thin films and nanoscale magnetic systems. In the following subsections,
we discuss the most important ones: (i) the magnetoelastic cantilever, (ii) strain induced
anisotropy, (iii) magnetostriction in spin valves, (iv) strain modulated ferromagnetic
resonance, (v) secondary-electron spin-polarisation, and (vi) strain-induced anisotropy due
to the spontaneous strains.
For technical applications (sensors), Hristoforou et al. (1998) have developed an
interesting method to determine the field-dependence of the magnetostrictive strain, based
on measuring the delay due to Bloch-wall motion.

3.1. The magnetoelastic cantilever method


In this method a magnetic film deposited on one side of a cantilevered substrate is
magnetised. The resulting 'magnetic stresses' will, for example, tend to lengthen the
magnetic film. Hence, the substrate will be bent (fig. 4b), with anticlastic deformation (i.e,
in saddle form, see fig. 4a, where the field is taken to be directed along the x-direction).
Obviously, the bending will depend on the elastic properties of both the film and the
substrate, as pointed out by Du Tremolet de Lacheisserie and Peuzin (1994, 1996), Van
de Riet (1994), and Weber et al. (1994). Notice that some errors in the older literature
were amended in these works. See also the debate in Klokholm and Jahnes (1996). The
theory of bending of plates was applied to obtain precise results, e.g. for non-isotropic
substrates (Ciria et al. 1995), see also the review of Sander (1999). Szymczak (1999)
(see also references therein) stresses the importance of the boundary conditions due to
the clamping of the cantilever, and concludes that often 'finite-element' calculations are
necessary. Also Iannotti and Lanotte (1999) treat the effect of clamping.
The cantilever bending-technique requires a sensitive displacement detection such as a
capacitance probe (Klokholm 1976, 1977), optical interferometry (Sontag and Tam 1986),
a tunnelling tip (Wandass et al. 1988) or angular detection (e.g. laser beam deflection, Son-
tag and Tam 1986; Trippel 1977; Tam and Schroeder 1988; Betz 1997; Sanderet al. 1998).
Using the laser beam deflection method, one determines the bending angle "', and
subsequently the displacement D(~ L . ."/2) at the end of the film (x = L; fig. 4b). Then,
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 107

z L

x
Fllm",ltb thickness tf
Substrate with
~ thickness t•
.Holder

(a)
Fig. 4. (a) Anticlastic deformation of a bimorph in a magnetic field (saddle fonn). (b) Fixation of a bimorph
on its sample holder and measurement principle of the defonnation of any magnetostrictive bimorph using the
deflection of a laser beam.

an (effective) magnetoelastic coupling coefficient, or rather an 'effective magnetic stress'


b(H) is calculated as

b(H)
Es
= 3(1 + vs)
D(H)
L2
t; e, ts L\(I/R )
r; = 6(1 + v ) ts
2

L .
(12)
s

Here, t is the thickness, E is the Young modulus, v is the Poisson ratio; the subscripts f
and s stand for film and substrate, respectively (see also preceding section). In the second
equality (applied, for instance, by Sander et aI. 1999), RL is the (longitudinal) radius of
curvature (signs should be chosen in such a way that positive b corresponds to negative
magnetostriction, since e.g. ).,y.2 = -By·2/c Y). Assuming that the magnetic film is thin
in comparison to the substrate, but that surface effects can be neglected, and ignoring an
isotropic contribution, Du Trernolet de Lacheisserie (1995, 1999) has shown that, for an
isotropic film magnetised to saturation in the direction a, the magnetoelastic parameter
b(a) measures the bulk value By·2 as

(13)

In practice, one determines bn(H) with the field directed along the film direction (x
axis), and b1. (H) with the applied field along the y-direction (still inside the film plane).
=
At saturation, one has at = I and az I, respectively. When applied to cases where
surface (interface) effects are important, it turns out that, with these field orientations, only
108 N.H. Due and P.E. BROMMER

the effective parameter b y · 2 (corresponding to By·2 in bulk material) can be determined


(Du Tremolet de Lacheisserie and Peuzin 1994; Du Tremolet de Lacheisserie 1995).
The difference between the saturated b sat values just yields the (effective) magnetoelastic
coupling coefficient b y · 2 = bll sat - b.l sat [= By·2 for relatively thick films]. In order to
facilitate comparison with other data, it is common practice to express these results as
'coefficients of magnetostriction' All, A.l and Ay·2 by:

(14a)

(l4b)

In general, it is difficult to determine correctly the elastic parameters E f and Vf of thin


films. Hence, this translation (eq. (l4a), (l4b» may introduce a large uncertainty in the A
values.

3.2. The strain induced anisotropy method


This method is based on the Villari effect: applying a uniaxial stress to a ferromagnetic
substance induces a magnetoelastic anisotropy which may modify all the parameters of its
magnetisation curve, e.g. magnetic susceptibility, coercive force, and so on. Some experi-
mental techniques to measure the strain-induced anisotropy are discussed shortly below.
An original technique was developed by Konishi et al. (1969) and extended later on
by Narita et al. (1980). This method is known as the small-angle magnetisation rotation
(SAMR) method: a static bias field HII and a tensile stress (a) are applied in the direction of
the film; a small-amplitude ac driven field H.l = H.l max sin(wt) is applied perpendicular
to HII' It is this ac magnetic field that induces a magnetisation rotation, which can be
detected as an induced voltage in a sensor coil wound around the film axis. This response
is measured as a function of the applied stress. i.e. of the strain-induced anisotropy. An
experimental SAMR set-up is illustrated in fig. 5. The sensitivity of this method was
2 x 10-7 (Narita et al. 1980) and even much higher. namely 10-9 (Hernando et al. 1983).

Glassy
Ribbon

Fig. 5. Experimental SAMR set-up. After Du Tremolet de Lacheisserie (1993).


MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 109

O'Handley et al. (1993) presented a method which allows magnetoelastic coefficients to


be determined from arbitrarily shaped M -H curves taken at different strains, by calculating
the area between them over a conveniently chosen magnetisation range. This reduces
error in the magnetoelastic coefficients by avoiding hysteretic effects at low fields and
ambiguity about saturation at high fields. These authors illustrated how the magnetic and
magnetoelastic coefficients can be determined from the M-H loops using the magneto-
optic Kerr effect (MOKE).
For thin films with a perpendicular anisotropy (easy axis normal to the film plane),
Huang et a1. (1995) determined magnetostriction (As) and anisotropy (K) values, by using
the so-called 'initial-susceptibility method'. An ac magnetic field was applied in the film
direction, and the induced magnetisation component in the field direction was measured.
These authors deduced K and AS from the X -I vs. a plots, using:

110 M 2
X= s (15)
2K - 3Asa

Here, M s is the saturation magnetisation. Notice, that apparently, the stress transfer from
the kapton substrate to the deposited film was ignored (compare eq. (15) with eq. (11d».
A consistent result was obtained by determining the anisotropy from the hysteresis
loops measured in the hard and easy magnetisation directions, using a vibrating sample
magnetometer.
Experiments using a standard electromagnet can be performed by measuring the strain
modulated biased transverse susceptibility (Kraus 1989). In this measurement, a static bias
field is applied in the film plane, now perpendicular to the film axis, while the ac driving
field is parallel to that axis. The susceptibility (transverse to the bias field) is detected by
the pick-up coils and its reciprocal value is then a linear function of both the de bias field
(H de) and the magnetoelastic field (Hu = 3Asa/ Ms; see also eq. (l lcj):

I I
X;; = -(Hde - H a - Hs). (16)
Ms

Here, Hs is the anisotropy field (Ny - Nx)M s •


An increasingly important tool to determine the strain-induced anisotropy is MOKE
(magneto-optical Kerr effect). In section 2 we mentioned already the calculations by
Freeman et a1. (1999). Experimentally, e.g. Ali and Watts (1999); (see also references
therein) apply a bending device to induce strains in a controlled way, and determine
the (local) curvature and the strains by optical interferometry or by direct measurement
(stylus). The properties of the substrate are incorporated in a finite-element modelling
calculation, thus allowing an absolute determination of the film properties. Compare also
Stobiecki et al. (2000), who studied the strain induced anisotropy in FeB/CulFeB trilayers,
using Kerr magnetometry (MOKE).
Other methods based on the Villari effect can be found in the paper by Lachovicz and
Szymczak (1984).
110 N.H. Due and P.E. BROMMER

3.3. Magnetostriction in spin valves


An alternative to the cantilever beam method, called the bending method, was introduced
by Baril et al. (1999), who measured the strain-induced magnetoelastic energy for spin
valve structures. A spin valve consists of two thin conducting ferromagnetic layers,
separated by a non-magnetic layer, which is thick enough to reduce the magnetic coupling
between the two magnetic layers. The upper layer is coupled to the top antiferromagnetic
layer. The lower one is deposited on the substrate (see Dieny et al. 1991b). The samples are
strained by means of the bending device shown in fig. 6. The strain in the film is tensile,
and as the bending is small, it is considered to remain in the film plane. The M-H loop is
directly determined from the deduced resistance (R) versus magnetic field curve

M= 2M s (
R(a ) - R(O) I)
- - . (17)
~RGMR 2

Here, a is the angle between the magnetisation directions of the two ferromagnetic layers
which form the spin-valve.
Assuming that the magnetic moment of the upper layer is rigidly fixed, and taking
into account the (weak) coupling of the two ferromagnetic layers (coupling field Hi), the
magnetostriction can be calculated from the area between the curves obtained without and
with strain (subscripts u and s, respectively):

In this measurement, the hysteresis problem was solved by applying a constant


transverse field. In eq. (18), this transverse field does not occur due to cancelling of terms.
The same holds for the (in-plane) anisotropy and the shape anisotropy. Baril et al. (loc. cit.)
claim that this is a reliable method, giving results consistent with the cantilever method and
with a sensitivity better than 1 x 10- 7 . It is perhaps worthwhile, however, to consider all

~ x
substrate
~Vp it
1

Fig. 6. Bending tool for the measurements of magnetostriction in spin valves. After Baril et al. (1999).
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS III

assumptions critically, taking into account the weak coupling of the ferromagnetic layers
and also the elastic properties of the substrate.

3.4. The strain modulatedferromagnetic resonance (SMFMR) method


Magnetic resonance has been recognised as a powerful technique to study the magnetocrys-
talline anisotropy energy of ferro magnets (ferromagnetic resonance FMR) and crystalline
electric fields in paramagnets (electron paramagnetic resonance, also named electron spin
resonance ESR). The technique of ferromagnetic resonance under stress was firstly devel-
oped by Smith and Jones (1963) (see also Smith 1968) in order to derive magnetostriction
coefficients from the shift in the resonant field caused by the application of a known sta-
tic stress to a magnetostrictive thin film. The shift arises due to the Villari effect. Using a
dynamic stress instead of the static one, Henning and den Boef (1978) have improved the
performance of the strain modulatedferromagnetic resonance (SMFMR). This technique
has been applied to measure the magnetostriction of amorphous wire (Kraus and Schnei-
dner 1977) and multilayer films (Zuberek et al. 1987, 1988, 1995,2000). Apparently, the
SMFMR method has some specific advantages, the most conspicuous one being the high
sensitivity (magnetostrictive strains as small as 10-9 can be detected) and a possibility to
measure several magnetoelastic tensor components.
The resonance equation for a thin film, which is subject to an uniaxial stress, and which
exhibits an intrinsic uniaxial perpendicular anisotropy field, may be expressed as (Bushnell
and Vittoria 1993)

(~) 2= [Hr + Ms - Hal- + 3UAs/M] X [Hr + 3UAs/Ms]' (19)

where H r is the resonant field, M s is the saturation magnetisation, Hal- is the anisotropy
field (= 2Kl-/ M«; Ks. is the perpendicular uniaxial magnetic anisotropy constant), y is the
gyromagnetic ratio and the equilibrium position of M 5 is assumed to be collinear with Hr.
Since the microwave frequency and gyro magnetic ratio are constant, changes in the
applied stress will result in a shift in the resonant magnetic field. Indeed, the 'shift' equation
is

flHr -3A5
--=-- (20)
flu Ms

Clearly, this equation allows to determine AS through the values of the stress a , the
resonant-field shift flH r and the saturation magnetisation Ms. Details of the experimental
set-up may be found in the review papers by Bushnell et al. (1992), Le Gall etal. (1989) and
in Vukadinovic's thesis (Vukadinovic 1988). Unfortunately, up to now the measurements
are limited to room temperature.
In traditional discussions based on eq. (20), the resonance conditions have been
expressed in terms of the magnetostriction coefficients. However, Du Tremolet de
Lacheisserie (1995) has found that the elastic coefficients of the substrate (SIll and SI2 S )
have entered his formulas of the resonance frequency. He showed that in the SMFMR
method only the b y •2 and ba •2 coefficients can be determined.
112 N.H. Due and P.E. BROMMER

3.5. The secondary electron spin-polarisation spectroscopy (SESPS)

The surface of a bulk solid is generally subject to relaxation strains (mainly normal to the
surface) that can be as great as 10%. These arise from the lower symmetry and different
electronic charge distribution there relative to the bulk. The surface of a cubic solid is
generally tetragonally distorted relative to the bulk. The surface of a thin film is subject to
the same relaxation effect as is the surface of a bulk sample. The surface of a material has
an electronic structure that reflects lower symmetry and reduced co-ordination numbers
relative to the interior, say, deeper than a few atomic layers. The secondary electrons,
whose spin polarisation is measured as a probe of the magnetisation, come just from
these outermost few layers. Thus, the spin-polarisation of secondary electrons, detected
from the asymmetric spin-orbit scattering, can be used to monitor selectively and directly
the surface magnetisation. Subsequently, by straining the sample (surface), the surface
magnetoelastic coupling can be determined by measuring the strain dependence of the
surface magnetisation.
An experiment to measure the asymmetry as a function of applied magnetic field at
different values of strain was set up by Sun and O'Handley (1991). It is shown in fig. 7.
From the measured asymmetric spin-orbit scattering, A, the surface magnetisation, M surf
is determined as

(21)

Here, N is the number of magnetic atoms per unit volume, n is the total number of valence
electrons (n = n+ +n-), nB is the valence-band magneton number (nB = n+ -n-), and S

MANIPULATOR

FIELD WINDINGS

Fig. 7. Experimental SESPS set-up. After Sun and O'Handley (1991).


MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 113

the asymmetric scattering efficiency of the detector. Then, from the strain dependence of
the magnetisation curves, the surface magnetoelastic coefficient B surf is derived from

(22)

=
with e t:x x , and where m 2(H) = (af(H») in our notation. The value of the Poisson
ratio of the bulk (ribbon) material, v, was taken equal to 1/3. Notice, that here B surf
is considered as energy density, so analogous to the bulk quantity By·2, evidently now
for the 'surface layers' only. The authors derived this expression by assuming isotropic
magnetoelastic behaviour. One may, however, look upon the relation between the 'surface
layers' and the 'bulk ribbon', as equivalent to the relation between a thin film and the
substrate. Eq. (22) should then be compared with (the exx-derivative of) eq. (IIa). Taking
(a~) = (a~) = (1 - {af)/2, one may deduce that, actually, the authors did apply the
(almost equivalent) approximation [1 - vv surf + (v - vsurf )j 2]f[(1 - vsurf )] = 4/3, instead
of (1 + v) = 4/3: indeed. again the elastic properties of the 'carrier' (here the ribbon) and
the 'thin layer' (here the surface layers) do enter into the expressions.
In this experiment, B surf was found to be (6.0 ± 0.2) x 105 J/m3 for the amorphous
C076Cr4B20 alloy, approximately 3 times the bulk value [+ 1.8 x lOS J/m3 ], and to be
-(1.6 ± 0.2) x lOS J/m3 for the amorphous Fe77Cr6B17 alloy, about half the bulk value
[-3.0 x 105 J/m3 ]. The 'surface magnetisation', was found to be 92% and 63% of the bulk
values for the Co-based and the Fe-based amorphous alloy, respectively.

3.6. The strain-induced anisotropy due to the spontaneous strains


Bochi et aI. (1994) proposed a phenomenological equation that describes the film thickness
dependence of the total effective magnetic anisotropy energy density of an epitaxial
sandwich fcc (111 )Co/Cu. They showed that by fitting the measured effective anisotropy
energy density with their equation, one can extract the surface magnetic anisotropy as well
as the surface magnetoelastic coupling coefficients. A surprising result was obtained for fcc
(l1I)Co/Cu superlattices (grown, on GaAs (110) substrates, by Lee et aI. 1990): the Neel
surface magnetic anisotropy energy is KS sin 2(9), with KS = 0.47 erglcm 2, whereas the
surface magnetoelastic coefficient BS is found to be BS = -25.5 erg/em", In this method,
instead of artificially created strains (as in the preceding subsection 3.3) the strain induced
anisotropy arises in the process of sample preparation, due to the Co-Cu misfit m = 1.9%.
In fact, the strain e was taken to follow e = mtcu/(tco + tcu). The actual strain e was
determined by X-ray scattering.
Of course, the different methods are often combined. (Clamped) cantilevers (section 3.1)
and bending devices (section 3.2) have also been used to determine the 'spontaneous'
strains during the depositing process, or those caused by different thermal-expansion
coefficients. As an example, we refer to Sander et aI. (1999), who determined the
magnetoelastic coupling parameters as well as the epitaxial misfit stress in ultrathin
Fe(100)-films on W(100) substrates. The change of sign of the magnetostriction constant,
positive for Fe films thicker than 20 nm, and negative for thinner films, is ascribed to the
strain dependence of the magnetoelastic coupling (higher order terms, see remarks in the
114 N.H. Due and P.E. BROMMER

final part of section 2.2), in combination with the fact that the strains are much larger in
the thinner films. Gutjahr-Loser et al. (2000) performed a similar study on (ultra)thin Co-
layers, grown on W(OO 1). These authors suggest that a description with higher-order terms
only may be too simple, because of the structural changes in the (ultra)thin layer forced
by the epitaxial strains. Notice also that for such ultrathin layers, magnetic-anisotropy
oscillations and quantum well states were observed (Weber et al. 1995, 1996a, 1996b).

4. Giant magnetostriction in rare-earth-transition metal thin films

4.1. General consideration ofmagnetism in rare-earth-transition metal alloys


The rare earth-transition metal compounds are considered to be formed by the association
of the relatively narrow 3d band with a wider 5d band with higher energy. The
electronegativity difference between the constituents gives rise to a transfer of 5d electrons
towards the unfilled 3d band. Since the screening of the nuclear potentials by the electrons
is modified, the two bands draw together, leading to 3d-5d hybridisation states at the top of
the 3d band and at the bottom of the 5d one. This strong 3d-5d hybridisation does not only
result in the formation of the rare-earth intermetallics, but also determines the physical
properties of the existing compounds. In the compounds based on rare-earth and transition
metals, it is found, as a general rule that, due to the hybridisation between the 3d and
5d states, the 4f-3d spin-spin coupling is always antiferromagnetic (Campbell 1972; Due
1997). Taking into account the coupling between the spin and orbital moments of the 4f
electrons, one can explain the parallel and antiparallel alignments of the 3d (Fe.Co.Ni)- and
4f-moments in the light- (J = IL - 51) and heavy- (J = L + 5) rare-earth compounds.
The rare-earth elements can be separated into two groups, as already mentioned, of the
light elements consisting of Ce, Pr, Nd, Pm, Srn, Eu and the heavy ones consisting of
Gd, Tb, Dy, Ho, Er, Tm and Yb. Each of these groups can be subdivided into two sub-
groups according to the sign of the Stevens factor CX). This division takes into account
the angular distribution of the charge density of the 4f electrons, which is in the oblate
(CX) < 0) or prolate (CX) > 0) form (fig. 8). As mentioned in section 2 (fig. I), with the
oblate distribution (CX) < 0), the magnetic moment is perpendicular to the plane as in Nd
and Tb, for example. In this case, the magnetostriction is positive. For the case where
CX) > 0, i.e. the electronic density distribution in the prolate form, the magnetic moment is
aligned along the axis of the "rugby" and the magnetostriction is negative.
Huge magnetostriction is expected to occur in those compounds, which combine a high
rare-earth concentration with a high ordering temperature. This is the case for the Laves
phase RFe2 compounds. The corresponding crystalline RC02 compounds, however, show
a much lower Curie temperature (Tc) due to the metamagnetic behaviour of the 3d(Co)
electrons (Due and Goto 1999; Due and Brommer 1999). They exhibit huge volume as well
as Joule magnetostriction, but at low temperatures only. For this reason, these compounds
are not particularly interesting for applications.
Amorphous alloys are characterised by a structural disorder where each atom constitutes
a structural unit. In this state, the low mass density and the loss of the periodicity
enhance the localisation of the 3d electrons in the rare earth-transition metal alloys. In
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 115

Light-lanthanide elements
J= IL-si

<XJ<O <XJ>O

~
Ce
e: Pr
roNd
~Pm (D
8m Eo

Heavy-lanthanide elements
J=L+S

<XJ<O <XJ>O

([)
Tb
ro
Dy
(l)
Do
([)
Er
~Tm (l)
Yb

Fig. 8. Angular distribution of the 4f charge density of lanthanide atoms for Jr = J (effective moment parallel
to the z-axis). After Thole in Coehoom (1990). In Ceo Pr, Nd, Tb, Dy, Ho the charge density is oblate (aJ < 0).
in Pm, Srn, Br, Tm, Yb it is prolate (aJ > 0). In Gd, Lu (L = 0). the charge density has spherical symmetry.

amorphous alloys. at a certain concentration, the 3d magnetic moment is higher, but the 4f-
3d exchange interactions are somewhat weaker in comparison with the crystalline alloys.
These parameters, however, follow the same tendencies with varying the concentration
(Due and Givord 1996). These results confirm again the systematic variation of the 3d-
5d hybridisation and its influence on the magnetism of the rare earth-transition metal
compounds. The decrease of the transfer and of the associated hybridisation allow to
maintain the strongly ferromagnetic character over a large range of concentrations, i.e.
there exists a large value for the critical rare-earth concentration where the magnetism
disappears in the amorphous alloys. These effects lead to an enhancement of the Curie
temperature when going from the crystalline state to the amorphous state: for the R-Co
alloys, e.g. the ordering temperature of c-TbC02 equals 230 K (Due and Brommer 1999),
whereas Tc of a-TbC02 is comparable with that of a-TbFe2 (Tc > 400 K) (Hansen et
al. 1989; Hansen 1991). For the a-RFe compounds. however, the magnetic properties are
not improved by the amorphisation, but just the opposite. Due to the amorphous structure,
116 N.H. DUC and P.E. BROMMER

Tb

f (a)
Co ~c.
(b)

Tb

~ (c)
Fe
~
(d)
Fe

Fig. 9. Sperimagnetic structures in amorphous Sm-Co (a), Tb-Co (b), Srn-Fe (c) and Tb-Fe (d).

the Fe-Fe interatomic distance is distributed and the Fe-Fe exchange interactions can be
sometimes positive or negative. This leads to frustration and freezing of the Fe moments,
and to the occurrence of asperomagnetism (M s > 0) or speromagnetism (M s = 0). The
latter is observed for Y l-xFex (Coey 1978; Chappert et al. 1981).
Sperimagnetism occurs in two-subsystem structures like rare-earth-transition metal
alloys of composition RI-x T x where the 4f-3d exchange interactions are not modified (Due
and Givord 1996; Danh et al. 1998), but the large spin-orbit coupling of the non-S-state
rare-earth gives rise to large local anisotropy (local easy magnetisation axis). Taking into
account the sign of the RT-coupling for the light rare earth and the heavy ones. we can again
divide the amorphous R-T alloys into four groups of sperimagnetic structures as illustrated
in fig. 9:
a) asperomagnetic LR. with collinear T (like Sm-Co),
b) asperomagnetic HR. with collinear T (like Tb-Co),
c) asperomagnetic LR, with non-collinearT (like Sm-Fe),
d) asperomagnetic HR, with non-coIlinearT (like Tb-Fe),
In these sperimagnetic structures, the subsystem magnetisations are reduced with respect
to the collinear magnetic structure in the crystalline compounds. As a consequence, the
saturation magnetostriction As is decreased.

4.2. Magnetostriction ofamorphous rare earth based thin films


The common approach to amorphous magnetostrictive thin films is to reduce the
macroscopic anisotropy to achieve saturation of the magnetostriction in low magnetic
fields. As a tradition. the R-Fe based alloys were thought to be the best candidates.
giving a giant magnetostriction in bulk as well as in film materials. The magnetic and
magnetostrictive properties and magnetic anisotropy of the amorphous binary TbI-xFex
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 117

400

300

-e 200

100

o
o 0.2 0.4 0.6
lJoH (T)
Fig. 10. Room-temperature magnetostriction for TbI_xFex films. After Miyazaki et a1. (1997).

films as a function of the Th concentration have been the subject of many investigations
(Forester et al. 1978; Hansen et al. 1989; Hayashi et al. 1993; Quandt et al. 1994a; Grundy
et al. 1994; Huang et al. 1995; Hernando et al. 1996 and Miyazaki et al. 1997). Jerems and
Greenough (1999) studied amorphous melt-spun (ThFe2)t-xBx alloys.
Magnetostriction curves measured in magnetic fields up to 0.7 T applied parallel to
the film plane for several a-Tht-xFex thin films are presented in fig. 10 (after Miyazaki
et al. 1997). For all samples, except for the x = 0.944 sample, which has the bee-Fe
structure, the field dependence of the magnetostriction shows a relatively large random
anisotropy. The values of the magnetostriction in the as deposited a-ThI-xFex thin films
collected from different sources are summarised in fig. 11. Although there is some
scatter they show a compositional variation of the magnetostriction, which is very similar
to that observed for polycrystalline Tb-Fe compounds. A magnetostriction maximum
occurs around 0.55 ~ x ~ 0.67. In fact, at JLoH = 0.7 T, the largest magnetostriction of
about 480 x 10-6 was found at x = 0.67 which corresponds to an alloy of the ThFe2
composition (Miyazaki et al. 1997, see fig. 11). Note that, for these systems, Tc shows
its maximum at the same composition range. Thus, the giant magnetostriction could
be an optimal combination of the lanthanide concentration and the magnetic ordering
temperature. The magnetostriction maximum shifts to higher Th contents with decreasing
applied magnetic field, and a magnetostriction maximum of 220 x 10-6 was obtained
at x = 0.58 in JLoH = 0.1 T. Indeed, Danh et al. (2000) and Due et al. (2ooob) have
succeeded to enhance the magnetostriction even at lower fields PI.U = 340 x 10-6 at
118 N.H. Due and P.E. BROMMER

Tbtel7 Tbte:u TbFe3 TbF~.

Tb1_.Fe.
300 • .o.eTp.dlll84)
• .o.44Tputdyol.0I.1t194)
• .1.5T".....0I. .. 1I7III
. . . 1.8T....".ot.ot.ll1l13)
200

100


0.1 0.2 0.3 0.4 0.5
I-x

Tb1.•Fe.
o 0.7 T
400 <l. 0.3T
o 0.1 T
300

200

100 til
o

o 0.2 0.4 0.6


I-x
Fig. 11. Magnetostriction as a function of Tb concentration for Tbl_xFex films. After Quandt (1997) (upper
panel) and Miyazaki et aI. (1997) (lower panel).

l-toH = 20 mT in TbI_AFe,Coh with x = 0.6 (fig. 12). The saturation magnetostriction


depends strongly upon the preparation condition. Films with a perpendicular magnetic
anisotropy show a higher magneto striction, however, one needs higher applied fields to
rotate the magnetisation into the film plane and then to attain saturation (see fig. 12) (Schatz
et aI. 1994; Due et al. 2000b). This perpendicular anisotropy is usually thought to be
induced during the deposition process. Heat treatments at temperatures up to T A = 450°C
are often performed to relieve any stress. Heat treatments do cause a number of clear
differences in the magnetic properties, such as the magnetic anisotropy change from
perpendicular to parallel (Riedi et al. 1998; Danh et aI. 2000; Due et al. 2000b), the
reduction of the magnetic coercive fields and the development of magnetostriction at low
fields (see e.g. fig. 12). In order to obtain a soft magnetostrictive behaviour for Tb-Fe
alloys, other attempts to reduce the magnetic anisotropy have been undertaken. Altering
the deposition conditions, e.g. by applying a r.f. bias voltage (Quandt 1994), the tensile
stress obtained can lead to an in-plane magnetic easy axis and to dramatically improved
magnetostriction at low fields. Miyazaki et al. (1997) and Wada et al. (I997a, 1997b)
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 119

600

500

400
-r
.'-'..
= 300
~
«
200

100

0
-0.8 -0.4 0 0.4 0.8
l10H (T)

Fig. 12. Parallel magnetostrietive hysteresis loops for Tb(Feo.55COO.45) 1.5 films: (I) as-deposited film and
(2) after annealing at 350°C and (3) at 450°C. After Due et aI. (2000b).

have also succeeded to enhance the low-field magnetostriction, however, by varying the
substrate temperature and the heat treatments. Magnetic-field annealing (Due et al. 1996;
Betz 1997) created a well-defined in-plane uniaxial anisotropy. Pasquale et aI. (2000)
studied stress-induced anisotropy in ThFe thin films. At the composition Th54Fe46 these
authors found optimal properties for sensor applications requiring magnetic softness and
stress sensitivity. It is interesting to compare these results with those obtained by Jiang et
al. (2000) on amorphous TbDyFe sputtered films. Notice also that for a-DYI-xFex films,
the magnetostriction has a maximum at x = 0.7 (Williams et aI. 1994). Speliotis et aI.
(1998) added Nb and Zr to ThDyFe-films in order to influence the (nano)crystallisation.
For nanocrystalline (Tho.3DYo.7)4o(Feo.95Nbo.OS)60 films, deposited at 470°C, a room-
temperature magnetostriction of 500 ppm at 4 kOe and a coercivity lower than 30 Oe
were observed.
A negative magnetostriction was actually observed in amorphous Sm-Fe thin films in
accordance with the positive sign of the Stevens factor (see section 4.1) (Hayashi et al.
1993; Honda et al. 1993, 1994). For these films, the room-temperature magnetostriction
increases rapidly in low fields due to the in-plane anisotropy. The maximum absolute 1-
values of about 250-300 x 10-6 at 0.1 T and 300-400 x 10-6 at 1.6 T were obtained on
films with 30-40 at% Sm. Honda et al. (1994) have used these magnetostrictive films for
the fabrication of trimorph ThFelpolyimide/SmFe cantilevers (see fig. 86a, in section 12).
Applying a bias voltage, it was again possible to alter the stress state to tensile stress,
which resulted in a perpendicular anisotropy due to the negative but increased saturation
magnetostriction (Quandt and Ludwig 1997).
In contrast to Tb-Fe films, crystallisation of Sm-Fe films does not result in a higher
saturation magnetostriction. The hysteresis, however, was found to be significantly
increased. Boron added to the SmFe2 alloy improves the possibility to form the amorphous
120 N.H. DUC and P.E. BROMMER

20 320
Tbt_xCo x

-
~
~
\Ill
15 240

-
~
'-' c
10 160 C
.c'" .s
0.62
5 80
0.53
0 0
0 0.5 1.0 1.5 2.0
~H(T)

Fig. 13. Room temperature magnetostriclion for several a-Tbj c, Cox alloys. After Givord et al. (1995).

state, reduces the local magnetic anisotropy energy (Polk 1972) and, thus, can enhance
the low-field magnetostriction. This was examined by Kim (1993) on the de-sputtered
(SmFez)lOo-xBx system with a thickness of 0.3 mm. In these alloys, the saturation
magnetisation decreases with increasing x, whereas the saturation magnetostriction
significantly increases. The highest saturation magnetostriction, of -670 x 10-6 at 1.0 T
with an effective magnetostriction of -490 x 10-6 at 0.03 T has been obtained for a-
(SmFez)99.Z6BO.74. A similar result was reported for amorphous bulk alloys (Fujimori
et a1. 1993; Shima et a1. 1997). Batt et a1. (2000) observed decreasing (absolute)
magnetostriction values upon substituting Co for Fe in melt-spun Sm(Fe,Co)z ribbons and
hot-pressed compacts.
The magnetostriction of a-Thu.j Co, (0.78 ~ x ~ 0.38) thin films was studied
intensively by Betz et a1. (1999) (see also Givord et a1. 1995; Betz 1997). Their reasons
for expecting a larger magnetostriction in a-Tb-Co than in a-Tb-Fe is that effectively
(see fig. 9(b and d)), the Tb-Co exchange interaction in the Tb-Co films are expected
to be higher, and consequently, (i) the Tb-sperimagnetic cone-angle will be reduced, and
(ii) the ordering temperature in a-Tb-Co will be raised. Indeed, it was shown that these
alloys are ferrimagnets above room temperature for x ~ 0.62. For x ~ 0.67, T C reaches
approximately 500 K, which is already higher than that of a-TbFez. Room-temperature
magnetostriction is shown in fig. 13 for several a-Tb l-xCOx films. The magnetostriction is
always positive. The compositional variation of magnetostriction is shown in fig. 14. It is
clearly seen that the room temperature magnetostriction increases rapidly with increasing
Co content as soon as the films become magnetic at that temperature, and reaches a
maximum around x = 0.71 (bll = -20 MPa, b Y ' z = -24 MPa under JLoH = 1.9 T which
corresponds to All = 320 x 10- 6 , AY' Z = 400 x 10-6 assuming Er = 80 GPa and Vr = 0.3).
For comparison, the magnetostriction results for a-TbFe measured in a field of 1.6 T by
Hayashi et a1. (1993) are also included in the same fig. 14. These earlier published results
were corrected by multiplying with a correction-term (1 - vs)/(l + vs) for the use of
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 121

25
Tb
1'1
FeI
~/
...•••• -,.
"
400

350
20 (Hayashi et al.) ... ,
, 300
.-.
~
~
15
. 250 »;c
::g , N
'-' 200 .-.
....
... ;.
~
10

~
~ 150 J

100
5
I 50

0 0
0.4 0.5 0.6 0,7 0.8 0.9 1
X

Fig. 14. Magnetostriction as a function of Tb concentration for a-TbI_xCOx thin films (after Betz 1997) and for
a-TbI_xFex ones (after Hayashi et al, 1993).

0 0

\
-5 \ -5
.-.
~ \\ +
~
,\...... ...r:r
U
0
~
'-'
-10 ...~'!.a-:
~::t"r~~ ~
-10 g
~
... i
~
I
-IS -15
-:P
·20 -20
0 0.5 1 I.S 2
~O H (T)

Fig. 15. Magnetostriction of SmO.37CoO.63 and TbO.36COO.64. After Betz (1997).

the former Klokholm's formula (Betz 1997; Betz et al. 1999; see also the discussion in
section 2.2).
It is also meaningful to consider the magnetostrictive susceptibility obloH (and/or
0).,10 H), which is of significance for applications. For the films under consideration, at low
applied magnetic field, the magnetostrictive susceptibility is maximal for x = 0.65. Betz
et al. (1999), thus, have succeeded to show the similarity of the compositional variation of
the magnetostriction in both a-ThFe and a-ThCo alloys and opened up a promising series
of magnetostrictive alloys for potential applications.
A comparison of the magnetostriction of Sm-Co and Tb-Co systems is shown in
fig. 15 for SmO.37CoO.63 and Tb0.36COO.64. The Sm-Co alloy shows a behaviour, which
122 N.H. Due and P.E. BROMMER

is comparable to that of the Tb-Co one, but with the opposite sign and a somewhat smaller
magnetostriction in absolute value. In addition, the Sm-Co alloy is magnetically harder
than Tb-Co: its magnetostriction saturates around 0.3 T and the coercive field is about
0.03 T. The high-field magnetostrictive susceptibility, however, is smaller in the case of
Sm-Co. Since crystal field effects are supposed to be proportional to a)(O~}«r4r/ao)2)
(see e.g. the review by Franse and Radwariski 1993, table 2.4), it is particularly interesting
to compare these factors and the observed magnetostriction for these two alloys. The values
of a) (O~}«r4f/ao)2) are +40.2 x 10- 2 and -54.8 x 10-2 and the values of by·2 at 300 K
are + 11.0 and -18.3 MPa for the corresponding Sm-Co and Tb-Co films, respectively.
One finds that the ratio between the operators is comparable with that between the two
magnetoelastic coefficients:

la)(O~}«r4r/ao)2)TbCo /aJ(O~}«r4r/ao)2)SmCol = 1.37 and


\bIlTbCo/bnSmCol = 1.67.

A few amorphous Sm-Co alloys have also studied been by Quandt et al. (1994).
A (negative) magnetostriction has been confirmed, somewhat smaller (by 30%-50%) than
that of Tb-based alloys.
The maximum magnetostriction found in the amorphous state for both Tb-Fe and Tb-Co
alloys is much lower than in the crystalline state. There are three reasons to be considered
for this difference. (i) The structure is not the same in the crystalline and the amorphous
state. Nevertheless, it is often argued that the local environment in the amorphous state
is reminiscent of that found in the crystalline one (see e.g. Hernando et al. 1997). As a
starting point for the discussion, differences in the local environment can be neglected. The
measured differences in magnetostriction between the crystalline and amorphous states
might then be attributed to the facts that (ii) some compositions have a lower ordering
temperature in the amorphous state which means that the magnetocrystalline anisotropy
and so the magnetoelastic coefficient is lower and (iii) the sperimagnetic arrangement of
the Tb-moment in the amorphous case gives rise to a distribution of the Tb-moments which
lower the projected magnetisation and thus the magnetostriction.ln order to verify the latter
argument, Betz (1997) has compared the mean Tb-magnetic moment in a-TbI-xCox and
c-Tb I _ x Co x alloys and determined the variation of the low-temperature Tb-sperimagnetic
cone angle 0. The results show that 0 decreases from 75 degrees to 40 degrees when x
increases from 0.5 to 0.75. This result is comparable with that published in the literature
(Cochrane et al. 1978; Danh et al. 1998).
By assuming that the local environment in the amorphous state is similar to the
crystalline one, a further approach to lower the remaining anisotropy is by eliminating
the fourth-order anisotropy by substitution of Dy for Tb (Williams et al. 1994; Wada et al.
1996, 1997a, 1997b, 1997c, 1997d; Miyazaki et al. 1997). The well-known composition of
the bulk Terfenol-D alloy is TbO.3DYO.7Fe2. For a-(Tb,DY)0.42FeO.58 thin films, Miyazaki
et al. (1997) found that compensation of the magnetic anisotropy also occurs at the
Tb:Dy ratio 3:7. Room temperature magnetostriction curves for a-(TbI-xDYx)0.42FeO.58
thin films are presented in fig. 16. Note that with increasing x the value of A decreases
and tends to saturate with smaller magnetic field. In addition, the magnetostriction data
MAGNETOELASTlCITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 123

400 x=O.O

~-= 300

-
..-.I

-e
200
0.5

100 0.89
1.0

0
0 0.2 0.4 0.6 0.8
J.1oH (T)
Fig. 16. Room-temperature magnetostriction for (Tbl-xDYx)0.42Feo.58 films. After Miyazaki et aI. (1997).

measured at J-LoH = 0.008, 0.05 and 0.07 T are plotted as a function of x in fig. 17a. At
J-LoH = 0.05 T and J-LoH = 0.7 T, A does not decrease monotonically with x, but exhibits a
broad peak around x = 0.7. This concentration dependence of A is rather similar to that of
bulk polycrystalline (TbI-xDYx)Fe2 compounds (see fig. 17b). It indicates the near-zero
magnetic anisotropy in a film of the Terfenol-D composition. The atomic short-range order
of sputtered amorphous films, thus, can be considered as quite similar to that of crystalline
bulk samples, and the same origin of the magnetic anisotropy as well as magnetostriction
can be expected for both film and bulk alloys. We will come back to this aspect in the
discussion for the a-(Tb,Dy)(Fe,Co) films below.
As mentioned above, one of the ways to obtain a larger magnetostriction at room
temperature, is to enhance the lanthanide magnetisation by increasing the ordering
temperature and by diminishing the sperimagnetic cone angle. Due et al. (1996) have
successfully substituted Co for Fe in a-R-(Fe,Co) alloys. The T-T interactions tend to
be stronger in (Fe.Co)- than in either Fe- or Co-based alloys (Gavigan et al. 1988).
This results in an increase of T c for a given R:T ratio. In general, however, R-Fe
exchange interactions are larger than the equivalent R-Co interaction energies (Liu et al.
1994; Due 1997). This arises from the fact that the Fe moment is significantly larger
than the Co one, while the R-T intersublattice exchange constant is approximately the
=
same for T Fe and Co. Moreover, at the optimum concentration, the stronger R-FeCo
exchange energies should then lead to a closing of the sperimagnetic cone angle and thus
124 N.H. Due and P.E.BROMMER

400
a-(Tb 1-.Dy.)O.41Feo .S8
300
(a)
~ -
-
~
-e
200

100

0
0 0.2 0.4 0.6 0.8 1.0
X

2000
(Tb 1•• Dy.)O.41 Feo .S8
(bulk)

~--
=
(b)

-
~
1000

500
0.1 T

0
0 0.2 0.4 0.6 0.8 1.0
X

Fig. 17. Magnetostriction as a function of Dy content for a-(TbI-xDYx)O.42Feo.58 films (a) and bulk
(TbI_xDyx )0.42FeO.58 alloys (b). After Miyazaki et al. (1997) and refs therein.

to an enhancement of the magnetostriction. Due et a1. (2oooa, 2000c) have studied the
(TbO.27DYo.73)(Fel-xCoxh system, in which the Tb:Dy ratio of I : 2.7 was fixed as the
same as that of Terfenol-D. At low temperature, all these as-deposited compounds of the
composition (Tbo.27Dyo.73)(Fel-xCoxh are magnetically rather hard. The coercive field
starts, at x = 0, with the highest value of 3.4 T, and then decreases rapidly with increasing
Co concentration down to about 0.5 T for 0.67 ~ x ~ 1.0. At 4.2 K, the saturation
magnetisation, M». exhibits a maximum at x = 0.47 (see fig. 18). This compositional
variation of M s is in contrast to the behaviour observed for the corresponding crystalline
alloys where M s always shows a minimum in the middle of the composition range due to
the enhancement of the (opposite) 3d(Fe,Co) magnetic moment. In the amorphous case,
however, an increase in M 3d will close the sperimagnetic cone of the rare-earth moments.
The maximum observed at x = 0.47 reflects that, at low temperatures, the enhancement of
M3d is smaller than the associated increase in (MTb.Dy)' At room temperature, the films
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 125

1000

800 ....- - - -

-1 400 ..
..·..·r······......•...........•............•..-...-.-300- K .

200

oL......<~--l.~--'--'--'-...J-..L-..L-J.--JL-.l--l.--'--'--'--'-...&.....J

o 0.2 0.4 0.6 0.8


x
Fig. 18. Variation of spontaneous magnensation as a function of x at 4.2 K and 300 K for
ThO.27DYO.73(CoxFel-x h thin films.After Duc et al. (2000a, 2000c).

30 500

25
400

.-. 20
ell
300
-
~ ~
6 15 __-:::IIV .-.
o
-::...... ~ ~: ....~: ~ ....~ ~: ..· ~.'· o .. 200 J
10 A" •••••••.•:&., .,.9.. •... .2 •

5 {:~·:~·:H:: : : : :·e'· ·I'1· · \ -,). ... 100

0- •..••.•... 0- .
o o
0.0 0.2 0,4 0.6 0.8 1.0
x
Fig. 19. Magnetostriction measured at 0.06 T for ThO.27DYO.73(CoxFel-x h thin films: (I & I') as-deposited;
(2 & 2') annealed at 150°C; (3 & 3') annealed at 250°C. Solid lines represent by •20. y · 2 ). dashed lines: bll ()'II)'
The measuringdirectionalong the sample length is indicated. After Duc et al. (2QOOa, 2000c).

become magnetically rather soft. The strongest coercive field (of 15 mT only) is found
at x = 0.63. The room-temperature spontaneous magnetisation M s. however, is almost
independent of the Co-concentration (see fig. 18).
Different field-annealing effects on the magnetostriction are evident across the whole
composition range, as summarised in fig. 19. For the Co-rich alloys, b U increases
126 N.H. DUC and P.E. BROMMER

5.0 2.0

4.8 1.8
--.
....
-•
.:::,fI:l A.6 1.6 ...":~
--.
1\
';i'
V
4.4 1.4
.J=
-....
~

'-"

4.2 1.2
a·(Tb0.21Dy0.13)(Fe1-"Co)
II 2

4.0 1.0
0 0.2 0.4 0.6 0.8 1.0
X

Fig. 20. Variation of (MR) and M3d(Fe,Co) calculated from magnetostriction data as a function of x for
ThO.27DYo.73(CoxFel-x)z thin films. After Due et aI. (2000a. 2000c).

significantly after annealing (dashed lines), whereas b y ,2 remains virtually unchanged.


For the Fe-rich alloys, we see the opposite effect in that b y ,2 increases significantly after
annealing while bll remains virtually unchanged. The increase in b y · 2 may be associated to
the decrease of the saturation field after annealing. In fact, the largest magnetostriction of
Ay ,2 = 480 X 10-6 and All = 250 x 10-6 is found in the middle of the composition range
(at x = 0.47) and it can be obtained in the rather low applied magnetic field of 0.06 T. Note
that, due to the substitution of Co for Fe, the ordering temperature in c-Tb(Fe l-xCOx h is
increased. but without enhancement of the magnetostriction (Dwight and Kimball 1974;
Belov et aI. 1975). At present, the magnetostriction increase is considered to be caused
by the closing of the sperimagnetic cone angle due to the enhancement in M 3d in the
substituted a-R(Fe,Coh alloys. A detailed analysis was performed by Due et al. (2000a),
yielding values between 48° and 53° for the R-sperimagnetic cone-angle 0, in accordance
with some literature values (Coey et aI. 1981; Hansen 1991; Danh et al. 1998). The
variation in e also implies a variation in the average (Tb.Dy) moment as a function of
x (see fig. 20). From the measured magnetisation data (fig. 18), M3d can be determined
as a function of x (see also fig. 20). Clearly, a similar composition dependence of M 3d
as observed in the crystalline R-(Fe,Co) alloys is found and a maximum is reached for
x = 0.47 where there is sufficient Co to ensure good ferromagnetic T-T coupling as well as
sufficient Fe giving the larger magnetic moment. This 3d magnetic moment enhancement
caused by Co substitution was confirmed by Mossbauer studies (Danh et aI. 2000; Due et
aI. 2000b): for a-Tb(Feo.55C00.45)1.5, a hyperfine-field value Bhf = 24.5 T was reported,
whereas it equals only 21 T for a-Tbfe-.
It is well known that the substitution of Dy for Tb gives rise to an increase of the
magnetostriction at low magnetic fields, through the reduction of the saturation field. It
is, however, also accompanied by a reduction of the saturation magnetostriction. The Co
substitution in the R-(Fe,Co) alloys, in combination with the effects of field annealing,
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 127

60
3 :
50
40
: ...
/;:::- 2 800

600
-.. --~:--- 1 'f--
I• ;•
~ 30 /I

~20 ..,.. f /
, ~
~
400
--
8
-e
,.Q

10
.,
; ~
200 J

o 3 2
.l.. o
-10
0.0 0.5 1.0 1.5 2.0
JloH (T)
Fig. 21. Room temperature magnetostrietion for Tb(Feo.4SC00.5Sh.l: as-deposited (l), annealed at 150°C (2)
and 250°C (3). After Due et aI. (1996).

results in an enhancement of both the low-field and the saturation magnetostriction. Thus,
it shows a possibility to enhance the magnetostriction in this type of alloys by increasing
the Tb concentration. Indeed, Duc et al. (1996) found a record giant magnetostriction
of AY,2 = 1020 x 10-6 at l-toH = 1.8 T, with All = 585 x 10-6 at l-toH = 0.1 T, in
a-Tb(Fe0.45CoO.55 h.l. The field dependence of the magnetostriction for these alloys is
presented in fig. 21. Here we see that a magnetostriction of Ay,2 (= All - A.d = 800 x 10-6
at 1.8 T is already observed for the as-deposited film. The comparison between bl! (AU)
and b1.(A.d indicates clearly the anisotropy of the sample. If the zero-field state is fully
isotropic, then bU = -2b1. and if it is isotropic in the plane, then bU = -b1.. For the
results presented in fig. 21, b1. = - ibll' indicating a certain initial anisotropy. After
field annealing, bll increases, and bs. is significantly reduced (in absolute magnitude)
in agreement with the fact that the easy axis becomes better defined. In addition, it is
important to note that b y ,2 actually increases after annealing at 250°C, leading to the
very large magnetoelastic coefficient of -60 MPa (i.e. AY.2 = 1020 x 10-6) at saturation.
This result is particularly interesting as the magnetostriction was almost completely
developed at only 0.3 T. In comparison, the magnetostriction of TbO.32(Feo.45C00.55)O.68
and TbO.32COO.68, prepared in the same condition, is presented in fig. 22. Clearly, the Fe
for Co substitution increases the magnetostriction by a factor of 2.
The temperature dependence of the magnetostriction is shown in fig. 23 for a
TbO.36(FeO.5C00.5)O.64 film. The magnetostriction decreases linearly with increasing
temperature up to the ordering temperature of about 423 K. Figure 24 illustrates five
normalised magnetostriction loops of an amorphous Tb40Fe60 film. This film shows a T C
of about 100°C (373 K). The measurement reveals not only the variation of the saturation
magnetoelastic coupling constant and the Curie temperature of the investigated film, but
also gives information on the change of the hysteresis curve with temperature.
128 N.H. DUC and P.E. BROMMER

40
Tb:M = 1:2 600

30 500

....... 400 ~
.......
If 20
300 ;,
:E
'-'
.Q 200 '-'
..
10
0.61 100
0 0

-100
-10
0 0.5 I 1.S 2
~O H (T)

Fig. 22. Room temperature magnetostriction for as-deposited Tb(Feo.45COO.5512.I and TbC02.1. After Betz
(1997).

30
-:.... Tb (Fe
I

Co)
480
0.36 0.5 0.5 0.64
25
..~ . ~o H =1.9 T - 400

. . 20 - 320 ~
~
:E ......... ...
....
N

~
-

I
IS

10
-,•....... 240 .......

160
"Cl
"Cl

!
5

o
f-
'........ 80

o 50 150

Fig. 23. Magnetostriction of Tb(Feo.5CoO.512. I as a function of temperature. After Betz (1997).

Figure 25 presents low-field magnetostriction data for a-(TbI-xDYx)(Feo.45CoO.55h.l


thin films with different TblDy ratios, after annealing at 250°C. It exhibits obviously a
decrease in the magnetostriction and saturation field with increasing Dy content, but with
a maximum in the initial magnetostrictive susceptibility for x = 0.73 as observed in other
Terfenol-D based alloys. Here, we see a partial anisotropy compensation for x = 0.73 as
found for crystalline (TbO.27Dyo.73)Fe2. As shown in fig. 26, the anisotropy constant K
deduced from the magnetisation measurements has also a minimum for this composition.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 129

1.O1-_~

Q.8

CUSr-_~~ ...

SJ= 0.4

Q.8

Fig. 24. Normalised magnetostriction as a function of the applied field, at different temperatures, for a Tb(},4Feo.6
thin film (d = 3 ILm). After LUdwig and Quandt (2000».

,..--.,..---r----r--.,.--.,..--...., 800

40 600

0.5

-
&1-
0.73
20
200
10 1.0

o -.&.._---''--_...I-_.....J,_ _-L-_--' 0
o 0.1 0.2 0.3
~ H (T)
o

Fig. 25. Room temperature parallel magnetostriction for (Tbl-x Dy, )(Feo.45COO.55 h.1 measured along the hard
axis after annealing at 250°C. After Due et al. (1996).

This indicates that even in amorphous samples. in which the uniaxial K2 terms are expected
to dominate due to the local distortions from cubic symmetry, there is stilI a significant
contribution from the K4 cubic anisotropy terms. It is interesting to contrast this with the
numerical simulations for a-TbFe2 which have shown that the fourth-order anisotropy
energy per atom was an order of magnitude smaller than the second-order anisotropy
energy per atom, albeit that the fourth-order term was of the same order of magnitude
130 N.H. DUC and P.E. BROMMER

16
14
...,......
-
...:.:...e
'-'
~
12
10
8

-.c ""'

.c":::
6
4
2
0
0 0.2 0.4 0.6 0.8 1
x

Fig. 26. Variation of the ratio bll/ b.1 and the anisotropy constant K as a function of x for
(TbI-xDYx)(Feo.4sCo0.5S)2. After Due et al. (\996).

as in the crystalline TbFe2 Laves phase (Kaneyoshi 1984). The present result implies that
K4 tends to be larger than expected and that the local environment is reminiscent of that in
the cubic Laves phase. In the same figure, the ratio Ibll/bl.l shows a minimum at the same
composition, implying that at this concentration the anisotropy is then less well defined by
field annealing.
The field dependence of the magnetostriction is usually associated with different types of
magnetisation processes. For a system of randomly oriented spins and random distribution
of domain walls, the magnetisation process takes place in two steps (Chikazumi 1964):
first, the motion of 1800 domain walls leads to a magnetisation of Mo without any
contribution to the magnetostriction. In the second step, the spins rotate into the
direction of the applied magnetic field, leading to the change of both magnetisation and
magnetostriction. For amorphous alloys of random three-dimensional spin orientation and
of a random distribution of domain walls, Mo is expected to equal M max/2. In this case,
the relation between magnetostriction and magnetisation is given as (Schatz et at. 1994):

A(H)/A max = [2M(H)/M max - 1]3/2. (23)

For the rotation of the magnetisation out of the easy axis into the field direction, the
magnetostriction is related to the magnetisation as (Chikazumi 1964):

A(H)/A max = [M(H)/Mmaxf. (24)

The normalised magnetostriction as a function of normalised magnetisation for films


with perpendicular and parallel anisotropy is plotted in fig. 27(a, b) for
(TbO.27DYO.73)O.42FeO.58 (Schatz et at. 1994). The films with in-plane anisotropy appear
to show high magnetostriction and magnetisation at low fields (the coercivity is less than
0.01 T), due to the easy rotation of the spins in the isotropic plane. The motion of 180 0
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 131

1 I I I I , 1
Perpendicular ,•
, -
0.8 anisotropy ., 0.8
f-
,
-
8 0.6
(a) ,, - 0.6
,< = ,,
<,
-< 0.4 -
" Theory - 0.4
, , .. '
0.2 f- ,, - 0.2
a
a
.-r-·
0.2 0.4
","
I

0.6
I

0.8
I

1
a
M/Mmax

1 1

,,--
I I I I

0.8 f- In-plane anisotropy 0.8
(b)

-<8=
0.6 f- 0.6
,"
- ,,. -
<,
-< 0.4 0.4
Theory,.
0.2 - ,, - 0.2
:~
a I I I
a
a 0.2 0.4 0.6 0.8 1
M/Mm a x
Fig. 27. Normalised magnetostriction as a function of normalised magnetisation for (ThO.27DYO.73)O.42Fe0.58
films with perpendicular (a) and parallel (b) anisotropies. Dashed curves are calculated with eqs (24) and (23).
respectively. After Schatz et aI. (1994).

domain walls does not contribute to the magnetostriction, and eq. (23) can be applied. For
the films with perpendicular anisotropy, the magnetisation is governed by rotations out of
the EMD into the plane, requiring larger external fields. Here, eq. (24) holds. An anal-
ogous presentation is made for Tbo.4(Feo.ssCoo.4S)O.6 (Terfecohan; Due et aI. 2000b). In
this case, the sample with perpendicular anisotropy shows again a similar variation as de-
scribed in eq. (24). For the sample with parallel anisotropy, however, the onset of 'AJ'A max
occurs at a M value which is even higher than M max/2(see fig. 28). It is possible that
the spins are pulled into the plane by the shape anisotropy and only the random in-plane
(two-dimensional) oriented spin structure is formed. This would lead to a magnetisation
remanence, which is appreciably higher than in the case of the random three-dimensional
spin orientation.
132 N.H. DUe and P.E. BROMMER

1.0

0.8

-e 0.4

0.2

0.0
0.0 0.5 1.0

Fig. 28. Experimental and theoretical relations between normalised magnetostriction and magnetisation for
amorphous Terfecohan [Tb(Fe0.55eOO.45) 1.51 films. (1) and (2): theoretical curves described by eqs (23) and
(24), respectively. (_) as-deposited, (e) T A = 250oe, (A) 350 0 e and (0) 450 oe. After Due et al. (2000b).

Finally, in view of possible applications, the interesting values of the magnetoelasticity


and its magnetostrictive susceptibility (Xb = db/d(lLoH) (and/or XJ.. = dA/d(lLoH» are
summarised in table 2. Although the results are not so good as for the multilayers to be
discussed in the next section, the best results obtained for the single-layer films are good
enough for rnicroactuator applications. We mention here, for example, for Th(Fe,Co)1.5
films, the suitable values Ay · 2 = 1020 x 10-6, 1L0H e = 6 mT and XJ..II = 1.8 x 10- 2 T- 1•
As will be presented later in this chapter (section 10), many magnetostrictive devices have
been designed and manufactured, using rare earth based amorphous thin films like those
discussed here.

4.3. Magnetostriction ofnanocrystalline rare earth based thin films


A great disadvantage of the amorphous R-Fe thin films, from the point of view of
application in MEMS, is the comparatively low Curie temperature (T c ~ 400 K for
amorphous Terfenol-D films). Polycrystalline Terfenol-D films exhibit a room temperature
magnetostriction of As ~ 1500 X 10-6 and T c = 650 K. These films, however, are not
suitable for micro-mechanical applications due to their comparatively high coercive fields.
Material designs require always a synthesis between the advantage of the crystalline
films (giant magnetostriction and high T c) and good magnetic softness of the amorphous
films (low coercivity). In this context, the nanocrystalline structure is expected to be
a reasonable compromise. Most of the work on nanocrystalline ThDyFe alloys was
started with rapidly quenched ribbons (Kikuchi et al. 1993; Tanaka et al. 1995; Lim
et al. 1994; Ooike et al. 1993). Addition of 5 to 8 at% B yields a fine grain structure
with average diameter d ~ 10 nm. The best results {coercive field 1L0He 20 mT, =
magnetostriction A ~ 400 X 10-6, in an applied field lLoH =
1.0 T} have been obtained
for (ThO.3DYO.7 )0.3 Feo.62 BO.08 ribbons (Tanaka et al. 1995). Using different preparation
conditions, a nanocrystalline structure with d =
10 to 20 nm can be achieved directly
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 133

TABLE 2
Comparison of the magnetoelastie data for magnetostrietive bulk and thin-film materiaIs

y•2 A,y.2 ilbli/ilB Refs


MateriaIs b
(MPa) (10- 6) (Mpaff)

Bulk crystalline
Terfenol-D 11>o.27DYO.73Fe2 -101 2400 568 [IJ
Single layer films
a-Th Fe2 -19.4 321 20 [2J
a-11>0.27Dyo.73Fe2 -17.2 300 50 [3]
n-Tho.27Dyo.73Fe2 -49.0 800 [4J
a- ThC o2 -24.5 400 155 [5J
a-ThO.27 Dyo.73CO2 -15.1 260 190 [3J
a-Th(Feo.45 Coo.55 h -63.5 1040 300 [6,7J
a-Th(Feo.55 Coo.45) 1.5 -65.9 1080 1100 [8]
a-Tho.27DYO.73 (Feo.45Coo.55h -20.15 330 430 [6,7J
a-SmFel.6 25.9 -380 [9J
a-( SmFe2>w.26 BO.74 45.6 -670 [I 0]
a-SmCo2 11.0 -161 40 [7J
a-Sm(Feo.5S Coo.42) 1.54 27.4 -320 76 [IIJ

Multi/ayers
Tho.4 Feo.6/Fe -20 300 [I2J
Tho.27DYO.73Fe2 /Fe -12 650 [13]
Tho.27Dyo.73Fe2/Finemel 300 [l3J
Th(Feo.55 Coo.45) 1.5/ Fe -39 3040 [14J
Tho.4 Feo.6/Feo.5Coo.5 -28 410 1000 [12]
Tho.37 Feo.63/ Feo.65Coo.35 -31.1 600 4000 [12J
Tho..27Feo.73/ Feo.75Coo.25 -27 348 4800 [12J
Tho.. I sFeo.S2/Feo. 75COO. 25 -44.5 890 [I2J
1b(Feo.55C oo.45)1.5/ Feo.85Coo.15 -32 530 7850 [IS]

Sandwich system
Ndo.25Coo. 75/Tho.28 Coo. 72/Ndo.25 Coo.75 -15.2 248 560 [8]
Tho.28Coo.n/Ndo.25Coo.75/Tho.28Coo.n -16.5 270 117 £7,8,I6J

[IJ Clark (1980), [2J Hayashi et aI. (1993), [3J Due et aI. (2000a), [4] Riedi et aI. (1998), [5J Betz et aI.
(1999), [6J Due et aI. (1996), [7J Betz (1997), [8J Due et aI. (2000b), [9J Honda et aI. (1994), [IOJ Kim (1993),
[I I] S. David, unpublished data, [12] Quandt and Ludwig (1997), [I3] Farber and KronmUller (2000b), [l4J Due
et aI. (2oola), [15] Due et aI. (2oo1b), [I6J Givord et aI. (1996).

by melt spinning of DYO.3FeO.56BO.14 ribbons, without further annealing treatment (Lim


et aI. 1994). The influence of several additive elements on the magnetic and crystallisation
behaviour of these materials was investigated with Mo (Kikuchi et a1. 1993) and Mn (Ooike
et a1. 1993).
Schatz et a1. (1993, 1994), Williams et aI. (1994), Miyazaki et al. (1997) and Wada et
a1. (1997c, 1997d) reported investigations on the magnetic and magnetostrictive properties
of amorphous and crystalline Tb-Dy-Fe thin films. Effects of the substrate temperature,
of the annealing treatment on the microstructure, on the magnetic domains and then
on the magnetic and magnetostrictive properties of the 'hot grown' Terfenol-D films
have been investigated by Wada et aI. (1997c, 1997d). The changes in the in-plane
134 N.H. Due and P.E. BROMMER

1200
700
r s S.OxlO-# I PG s "m·1 2.5
b. ... M/I,_"_cc"'
0
• .l,,_',JIIfI 1000

,
9""
600
o • HelkOe
06<>: IttllUl-eryrbllJlud 2.0
..... : cry6llllJlud
u 500
u 800
~
r> S,ld,-4 tr.. _"'
E o :-M,"S11Oo',.." cc·,
-a
Q)

.IC
400
~o,,"-Uu
E
Q,
e,
600 ....
a
1.50
~
CD

.....II
It)
9"" ~/ .IC

...
111 2
~
:::: 300 .' :::: ...
111

1.0~
,
I 400 -e
I

.,.
I
200 I
I

I
[J I

I
I
200 0,5
I
100 [J I
I

.------------~-----_cr-----~
[J

0 0 0.0
400 450 500 550 600 650

Substrate temperature I K
Fig. 29. The changes of the in-plane magnetisation, coercivity and magnetostriction as a function of substrate
temperature Ts in TbO.27DYO.73Fe2 films. After Wada et al. (1997c).

magnetisation, coercivity and magnetostriction as a function of substrate temperature (T s)


are summarised in fig. 29. The sharp increases in the magnetisation and magnetostriction
at T s from 405 K to 410 K were associated with the formation of very fine nanocrystalline
grains (smaller than 5 nm). The small change of coercivity confirmed the non-drastic
structure change in this temperature range. At temperatures T s between 410 K and 600 K,
the magnetisation, coercivity and magnetostriction remain almost constant. In accordance
with microstructure studies, these results indicate a release of the internal stresses induced
during the depositions. At T s over 600 K, the development of grain sizes from 10-20 nm
to 40-100 nm was observed. This recrystallisation causes the second step of the sharp
increase of the magnetisation, coercivity and magnetostriction as illustrated in fig. 29. For
these films, it is worth to mention that surface oxidation with a thickness of 50 nm was
observed.
For films formed at T s below 405 (i.e. with grains below 5 nrn) the measured hysteresis
loops and the observations of the magnetic domains in applied magnetic fields seem to
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 135

0.8

)C
III
e O.6
-
~:::

~:: :'0.4

0.2

0.2 0.4 0.6 0.8 1.0


M"I M"max
Fig. 30. Normalised magnetostriction as a function of normalised magnetisation for the nanocrystalline (n) and
polycrystalline (p) (ThO.27DYO.73)Fe2 films. After Wada et al. (I997c).

indicate a preferential perpendicular magnetisation by spin rotations with low anisotropy


in these nanocrystalline films (denoted as n-films). The magnetisation of the polycrystalline
films formed at Ts between 610 K to 625 K (p-films), however. was suggested to be
governed by the motion of domain walls at low magnetic fields. The change in the
normalised in-plane magnetostriction AU/As as a function of the normalised magnetisation
of these films is plotted in fig. 30. At low magnetic fields. the n-film shows a much
higher magnetostrictive response to the magnetisation than the grown p-film. The parabolic
dependence of the in-plane magnetostriction on in-plane magnetisation, observed for the n-
film, is in good agreement with the arguments mentioned above. and with the experimental
results for the motion of 90° domain walls, i.e. the rotation of spins with perpendicular
anisotropy into the plane (see also eq. (24) and fig. 27a). For the p-film, almost no
magnetostriction takes place up to M/ M max = 0.2. This behaviour seems to be appropriate
for random distributions of spins. as described by Schatz et a1. (1994) (see also eq. (23».
Additional annealing of these 'hot grown' nanocrystalline Terfenol-D films at 630 K for
10 h increased the nanograins of below 5 nm to around 50 nm. An enhancement of the
magnetostriction (up to 1500 x 10-6). however. occurs only for annealing during 2 h.
Further annealing causes the magnetostriction to decrease due to the oxidation (Wada et
a1. 1997d). No appreciable increase in grain size was observed in a similar annealing for
a film with d = 10-20 nm. In addition, Riedi et a1. (1998) showed that annealing 'hot
grown' films is not advantageous because of the formation of iron-rich phases like RFe3
and R6Fe23 which exhibit even less magnetostriction and higher magnetic coercivity. These
authors. however. reported their results of annealing the 'cold grown' (Tho.3Dyo.7) I-x Fex
(with x ~ 0.3) films as in fig. 31. The best magnetic properties were obtained for the sample
136 N.H. DUC and P.E. BROMMER

800 2°
i' 03
0 600
'<,

=
t.
-e 400 t.
t. t. 0
200 t. I 0
4
0
0
0
0.4 t. amorphous -
E 0.3 o intermediate 40
J: 0
o crystalline
:::t 0.2
20
0.1 0 0
I 3
0.0 • •
3
20 0
600
0 0
SZ 4
~

....U 500

400 t. t. 0I
Ii t.
300
as-spat. 400 600

Fig. 31. The saturation values of the in-plane magnetostriction All, the in-plane coercivity ItOHq and the Curie
temperature Tc as a function of the annealing temperature T A for (TbO.3DYO.7)O.3Feo.7: (0) amorphous, (0)
intermediate and (t» crystalline states. (I. 2 and 3. 4 serve to distinguish different samples.) After Riedi et aI.
(1998).

annealed at T A = 600°C for 10 minutes with a grain size around 10 nm, a saturation
magnetostriction A = 860 x 10-6 but /-LO He = 120 mT.
The appearance of a perpendicular anisotropy in the mixed state of both amorphous and
crystalline structure was reported by Miyazaki et al. (1997) for (TbO.3DYO.7 )O.33FeO.67 films
fabricated above 673 K (4{)()°C) (see fig. 32). The thermal stability and the reproducibility
of those films were studied also. For films prepared with substrate temperatures above
673 K (400°C), the magnetostriction changes remarkably after 3 months. This is due to
ageing effects, related to the formation of the Laves phase (Tb,Dy)Fe2.
According to the random anisotropy model (Herzer 1990; Hofmann et al. 1992), a
further reduction of the crystal size will reduce the coercive field, if the exchange length
is larger than the average grain size. Hence, technologies which, on the one hand. can
enhance the nucleation of grains, but, on the other hand, may limit the grain growth, must
be applied to achieve a fine nanocrystalline structure. For this purpose Zr and Mo have been
chosen as additives. In the binary system Fe-Zr the most stable intermetallic compound is
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 137

Amorphous ~
in plane out of plane in plane

200

,-..
150
...
'i'
Q
'-'
O.7T
O.lT
-e 100 .OST

SO t r - - - - - - - - - ,

o '-=:=i:==::::C==~---.:t:...--.J
100 200 300 400 SOO 600
Substrate temperature (0C)

Fig. 3Z. Magnetostriction as a function of substrate temperature for (TbO.3DYO.7 )Fe2' The anisotropy type and
the structural morphology is indicated too. After Miyazaki et al. (1997).

- (Tbo.1l DY•.") •., ,Fe ..


- - «(Tb01lDYol,)."Fe )Zr•."

100

-0.8 -0.4 0.0 0.4 0.8


J.LoH'J:I [T]
Fig. 33. Room-temperature parallel magnetostncuon loops for (Tbo.27DYO.73)O.32Feo.68 and
[(TbO.27DYO.73)O.32Feo.681Zro.03 films after annealing at 973 K. After Winzek et al. (1999).

the cubic Laves phase ZrFez. So the solubility of Zr in Fe is negligible, and it can be
assumed that Zr is substituted for (Tb,Dy) in (Tb,DY)I-xFeZZrx alloys. In comparison
to (Tb,Dy)Fez, ZrFez exhibits a more negative enthalpy of formation, and therefore the
addition of Zr may enhance the nucleation of grains. Mo, however, has a much higher
melting point, and intermetallic compounds containing Fe show a much lower stability.
The existence of the Laves phase MoFez is still questionable. Moreover, the large Mo
atoms reduce the diffusion, and therefore the addition of Mo was thought to limit the grain
138 N.H. Due and P.E. BROMMER

800 o
- / 0 - (Tb•.71DYo.;n)Q.lZFe•.61
700 - 6 - «Tho.nDyo.Z7).,lFeo.61)Zr. Ol 6 t:.
~ «Tho.."DYo.Z7).,!eo.61)Mo... o
600
500 •
-..
~,. . .,

0
...... 400
~
300
200
100
~6A..-..
o ~ ~-t:>.

(a)
0
t... [minI 10 60 360 10 60 360 10 60 360 10 10
Tit [K] 773 803 833 873 973

300 - 4 - (Th•.nOY•.27)•.11Fe•.61
- 6 - «Th•.nOy.1I).nFe•.61)Zr001
250 - 0 - «Th•.71DY027)•.lZFe•.61)MoDJ)<
crystallized 0
E=' 200
E
...... 0-0
150
::x::"
::t" 100
X-ray amorphous
SO -4~
!_6-t:.
(b)
0
t... [min] 10 60360 10 60 360 10 10
TA [K] 773 833 873 973
Fig. 34. Magnetostriction (Asl, measured in I T (a), and coercivity (IJ,OHc) (b) of (TbO.27 DYO.73 lO.32 FeO.68,
[(Tho.27DYO.73lo.32Feo.68IZro.03 and [(TbO.27DYO.73lo.32Feo.68IMoo.04 films, after different heat treatments at
temperatures TA and annealing times IA. After Winzek et al. (1999).

growth. Practically, the influence of the additives Zr and Mo on the crystallisation and on
the magnetic properties was studied by Winzek et al. (1999). Typical magnetostriction
loops are presented in fig. 33 for two films crystallised at 973 K for 10 minutes, one
without additive and the other with 3 at% Zr. It is clearly seen that the Zr-substituted
= =
film shows All 430 x 10-6 at /-LOH 1.0 T and /-LoHe = 120 mT, a strong improvement
with respect to the starting alloys: All = 230 x 10- 6 and /-LoHe = 300 mT. The saturation
magnetostriction and the corresponding coercivity of all annealed films, without and with
additives, are summarised in fig. 34(a, b). These additives were considered as causing an
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 139

enhancement of the growth of RFe2 grains and to hinder the formation of RFe3 ones. It is
assumed that the latter was responsible for the high coercivity values above 150 mT. Farber
and Kronmiiller (2000a) reported also a reduction of the coercivity by 30% by the addition
of 2 at% of Zr. In general. it should be noted that a reduction of the average grain size of
the cubic Laves phases below 10 nm (and consequently coercivity values below 100 mT)
could not be achieved in single-layer films. The grain growth, however. can be controlled
in nanometer-scaled multilayers with interlayers of Nb. This will be presented in the next
section.

S. Magnetostrictive multilayers

5.1. Nanocrystalline TbDyFe + Zr/Nb multi/ayers


Nanocrystalline structures combine the advantages of a crystalline film and the soft
magnetic properties of an amorphous structure. These properties. however. can be achieved
when the grain sizes (d) are smaller than the Bloch wall width of the crystalline material.
For R-Fe alloys. the critical grain size has been estimated to be de ,...., 10-15 nm and
a grain diameter below 5 nm is desirable to reduce coercivity significantly. The results
of magnetostrictive single-layer films presented in the preceding section show that grain
growth can be affected by several factors. but that it is rather difficult to reduce the average
grain size below 5 nm. Fischer et al. (1999) and Winzek et al. (1999) have applied a method
to inhibit grain growth by constructing multilayers in a special way. They fabricated a
multilayer system containing TbDyFe + Zr layers with thickness of 5 nm, separated by
Nb-layers with an average thickness of 0.25 nm. Their microstructure investigations have
shown that this multilayer structure shifts the crystallisation temperature (for 10 minute
annealing) from between 773 and 873 K (for single layers) up to about 923 K. At this
annealing temperature, nanocrystallites of less than 5-10 nm were formed. After such
a heat treatment, the corresponding single-layer film exhibits a polycrystalline structure
with average grain sizes much larger than 20 nm (see section 4.3). This remarkable
influence of the Nb spacer layers is thought to be due to reduced dimensions and an
increase of interfacial surface area. Magnetic and magnetostriction data of the as-deposited
and annealed TbDyFe + ZrlNb multilayers are summarised in fig. 35. After 10 min
annealing at temperatures from 873 K to 973 K, the transition temperature increased
from T c = 333 K to 592 K, accompanied by an increase of the parallel magnetostriction
from 265 x 10-6 to 520 x 10-6 • while the coercive fields (increasing from #LoBe = 5 to
75 mT) stay distinctively below 100 mT. The variation of the magnetostriction, the ordering
temperatures and the coercive fields are strongly correlated to the formation of the highly
magnetostrictive RFe2 Laves phases. As mentioned above, the Nb spacer layers do inhibit
the grain growth. and also increase the crystallisation temperature. Because of its strong
affinity to iron. the additive Zr enhances local formation of the highly stable ZrFe2 Laves
phases, which then act as nuclei for the similar, but less stable. RFe2 Laves phase. The
difference of 34 K in the transition temperatures of single-layer TbDyFe + Zr (T c = 626 K,
see also section 4.3) and multilayer films may result from differing average grain sizes
and/or chemical composition. In the as-deposited multilayer the chemical composition is
periodically varying spatially. In the annealed multilayer, however, due to the interdiffusion
140 N.H. DUC and P.E. BROMMER

600 , - - - - - - - - - - - - - - : - . . 100
multilayer

... - ..... .. .: - ,.
,.,' .-~.
[500 - -.~. 75
.:: a'
~-

E- f /
F
oc
- Tc [K]-..... / ... ~_c· :::::
~400
as ... , 50 3'
..<
~300
.. ,4' .:j

.....
25
\-.u
---"- ----' _. -A
._._.-._-- -'
200 u - ""__OJ
o
~

873 898 923 948 973


T.~[KI

Fig. 35. The in-plane magnetostriction All_ coercivity J.l.oHc and Curie temperature TC as a function of the
annealing temperature T A for ThDyFe + ZrlNb multilayers. Open square symbols indicate the phase transition
temperature of an amorphous rest phase. After Fischer et al. (1999).

of atoms between layers, modification of the layers must be taken into account. Due et aI.
(2000d) have observed that an annealing at 350°C for 1 hour can extend the interfacial
spacing with I nm. Farber and Kronmtiller (2000b) observed that the multilayer structure
was destroyed at T A = 800 K (see also section 5.3 and fig. 56).

5.2. Magnetostrictive spring magnet type multi/ayers (MSMM)


Originally, the research on giant magnetostriction was based on homogeneous R-T-
alloys. As described in section 4, attempts to reduce the driving fields required for
giant magnetostriction are concentrated around techniques for reducing the macroscopic
anisotropy, e.g., to control the Tb:Dy ratio in order to achieve compensation of fourth-
order anisotropy, to use amorphous or nanocrystalline materials to reduce anisotropy, etc.
Furthermore, the saturation field H s actually is determined by the anisotropy field K 12M s-
and thus can be reduced by increasing the saturation magnetisation M s, instead of by
decreasing the anisotropy constant K. For a given Rvconcentration, which is optimised with
respect to giant magnetostriction, e.g. at the 1:2 R:T ratio, an increase of the 'I'-sublattice
magnetisation by substitution, for instance, will increase the total magnetisation in the R-
T alloys with R = light lanthanide, but will reduce the total magnetisation in the alloys
with a heavy lanthanide (due to the ferrimagnetic nature, when the R-moment exceeds the
T-moment). An increase in the R concentration, then, can increase M s- but also results in
a lowering of the ordering temperature with the opposite effect. Thus, it is difficult to see
how M s can be notably increased using homogeneous R- T alloys. Nevertheless, one has
found this possibility by combining two different magnetic materials using an approach
similar to those developed for the permanent "spring magnets". For the spring magnets,
one matches a material which has a high magnetisation with another one which possesses
a strong coercive field. These two materials are coupled magnetically. Here, multilayers are
fabricated by combining also two different materials, one having a large room-temperature
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 141

Fig. 36. Schematic view of a magnetostrictive "spring magnet" multilayer (MSMM).

magnetostriction (like, e.g., a-Tb-tfe.Co) alloys) and the other being magnetically soft
and having a high magnetisation (like, for example, (Fe.Co) alloys). The structure of this
spring-magnet type multilayer is illustrated in fig. 36. The thickness of these layers must be
enough for magnetic coupling but they must be thinner than the magnetic exchange length,
for which domain walls cannot be formed at the interfaces, i.e., the thickness should range
between I nm to 20 nm (Givord et al. 1993, 1996; Wtichner et al. 1995, see also section 7).
In this state, the 3d-3d exchange interactions ensure that parallel coupling of the transition-
metal magnetic moments persists throughout all multilayers. Without creation of domain
walls at the interfaces, the multilayer behaves as one piece of material. Then, magnetisation
processes result from the average of the magnetic characteristics of each individual layer.
Assuming that the 3d-3d exchange interactions are infinite and the interfacial magnetic
anisotropy is negligible, one can calculate the magnetisation, the magnetic anisotropy and
the magnetostriction of the multi layers from the corresponding values of the simple alloys
and the (average) thickness tTb and tT of the individual Tb-(Fe,Co) layers and the (Fe.Co)
layer, respectively (Betz 1997):

_ MTtT - MTbtTb
( M-
) , (25a)
tT tTb +
(K) = KTtT + KTbtTb, (25b)
tT + tTb

(b y •2htT + (b y ,2)Tb tTb


(b y ,2) = - - - - - - - (25c)
tT +1Tb

In the as-deposited composite multilayers, the Tb-based layers were amorphous, while
the transition metal layers were found to be nanocrystalline with a mean grain size being
equal to the layer thickness (Quandt and Ludwig 1997). In spite of the compressive
stress, these multilayers exhibit an in-plane EMD. The magnetisation of a TbFelFe and
TbFeIFeCo multilayer series is presented as a function of the transition metal layer-
thickness in fig. 37 (Quandt and Ludwig 1997; Quandtet al. 1997a, 1997b). Comparison of
the experimental data with theoretical calculations based on either parallel or antiparallel
coupling between the Tb-Fe and the T layers strongly supports the antiparallel coupling,
i.e. the (Fe.Co)- and Fe-magnetic moments are coupled ferromagnetically throughout
the entire multilayer and are coupled antiferromagnetically to the magnetisation of the
142 N.H. DUC and P.E. BROMMER

1.5 .-----...--,...--.--y---.--,...-....,......--.",_--.~_.....,

1.0

....,..,.
•/ • lbFe(4.5 nm)/Fe(x nm) expo
0.5 ••••• -lbFe/Fe parallel coupling
•••• • - lbl'e/Fe ontiparallel coupling
....
...... • lbFe(7 nm)/FeCo(x nm) expo
-1bFe/FeCo parallel coupling
- - lbl'e/FeCo antiparallel coupling
0.0 '--........L-.--'-_"""'-_"'----'_--'-_........_ " ' - -........._ ..............
o 2 4 6 8 10
X (om)

Fig. 37. Saturation magnetisation of the TbFelFe and TbFelFeCo multilayers as a function of the transition-metal
sublayer-thickness, in comparison to a simple model for exchange coupled layers considering either parallel or
anti parallel coupling of the TbFe and the transition metal layers. After Quandt and Ludwig (1997).

ferrimagnetic TbFe layers due to the dominating Tb moments in these layers. This
behaviour is confirmed for TbFelFeCoBSi multilayers (Quandt and Ludwig 1999).
The in-plane magnetisation loop of the as-deposited Tb(Feo.ssCo0.4s)1.5(7.5 nm)!
Fe(5 nm) multilayer is plotted in fig. 38a together with the corresponding loop of a
giant-magnetostrictive Terfecohan Tb(Feo.SSCoO.4S) 1.5 single-layer film (Due et al. 2000d.
200la). It reveals the reduced (but non-negligible) hysteresis and saturation field for
the multilayer. This behaviour is confirmed in the magnetostriction loops (fig. 38b).
Figures 39a and 39b show the magnetostrictive hysteresis loops of a TbFeIFe and a
TbFelFeCo multilayer. respectively. For both multilayer systems. high saturated values
were obtained for the magnetoelastic coefficient b y · 2 = (b ll -b1.): -28 MPa (TbFelFe) and
-41 MPa (TbFelFeCo) in a field as low as 20 mT (corresponding to the Ay ,2 = (All - A1.)
values 580 ppm and 850 ppm. respectively). These values. however. are still lower than
those of the best TbFeCo single-layer films and also the uniaxial easy axis does not seem
to be well established in these systems. The magnetostriction improvement in TbFeIFeCo
compared to TbFelFe multilayers is due to the (extra) magnetostrictive contribution of
the FeCo layers. which exhibit a saturation magnetostriction exceeding 100 x 10-6
(Quandt and Ludwig 1997; Betz 1997). In fig. 40a, the saturation magnetoelastic
coupling coefficient b; = (3/2)b y ,2 of an annealed (Tb4oFe6o)/(FesoCoso) multilayer is
compared with that of single films of Tb40Fe60 (with sputtering conditions optimised
for attaining the large saturated b s value of about 41 MPa. see Quandt 1997) and
of (Sm.Fe.B) (optimised for low-field performance. at the cost of the saturated value.
which for the (Sm.Fe.B) system can be doubled, for a Sm content of 36.8%. see
Lim et al. 1998). In fig. 40b a comparison is presented between the magnetoelastic
coupling coefficient bll of the same multilayer and values for some other multilayers:
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 143

1.0

0.5

~ 0.0

-0.5
(a)

-1.0
-0.8 -0.4 0.0 0.4 0.8
floR (T)
700
600
SOO
.-..

-
"=
~

400
300
200
A,
100 (b)

0
-0.8 -0.4 0 0.4 0.8
l10B (1)

Fig. 38. In-plane magnetisation (a) and magnetostriction (b) of a Terfecohan Tb(Feo.55COO.45>t.5 single layer
and a TerfecohanJFe multilayer film. Definitions of the magnetostriction remanence (A.r) and the coercive field
(AHe> are illustrated for the Terfecohan film. After Due et a1. (200la).

Sm(TbI SFeS2)/(Fe75C025), (TbDyFe)/(FeSiCuNbB) and (Tb33Fe67)/(FesoB20). Ludwig


and Quandt (2000) reported the possibility to control the orientation of the magnetic easy
axis by magnetic annealing and, thus, to enhance the magnetostriction in the considered
direction (see figs 41(a, bj). A uniaxially anisotropic multilayer TbFelFe can also be
created by depositing under a magnetic field H def (Le Gall et al. 2(00). Typical field
dependence of the magnetoelastic coefficient bY' of an isotropic TbFelFe multilayer
deposited without a bias field (H dep = 0) is presented in fig. 42a. The symmetry of b"
and as. with almost the same saturated values (""-"2 MPa) demonstrates the quasi-isotropic
character of the magnetoelastic properties of the multilayer (see also eq. (13) in section 3).
When deposited under a bias field in the x-direction (H dx = H dep I- 0), the multilayer
presents the original magnetic (fig. 43) and magnetoelastic (fig. 42b) behaviour associated
with strong uniaxial anisotropy. After saturation along the easy axis, the moments remain
aligned along that axis, in a single domain, when the applied field H" is lowered down to
zero (MriMs = 1). In reverse fields, the magnetisation is switched in a short field range.
144 N.H. DUC and P.E. BROMMER

200
(8)

-
=-
IIIl
-
-
~
.Q
0

-
'T'
c
-e
\ -Tbo.JeG.6(4.5 om/Fe(6.5 om)

0
10 -200
\ -Tbo.J~.6
...
20
-0.05
~
0.00
-
0.05 0.10
-400
0.15

JJoH (T)
600

(b) 400

_
ClIl
=-
-10
II TbFe(7nm)/FeCo(8nm)
200

b --
-
~
.Q
0

l-
0
-
-e
10 -200

20 -400
-0.05 0.00 0.05 0.10 0.15
JJeH (T)
Fig. 39. Magnetostrictive hysteresis loops of a TbFelFe (a), compared with a TbFe single film, and of a
TbFelFeCo (b) multilayer. After Quandt and Ludwig (1997).

Apparently, 'in field' depositing reduces also slightly the coercivity (from 6 mT to 4 mT).
High-amplitude flexural and torsional-oscillation modes were observed for these films.
In the absence of a long-range anisotropy in amorphous TbFeCo layers, along with
negligible magneto-crystalline anisotropy in FeCo layers, the coercivity (between 4.5 mT
and 7 mT) of MSMM is usually determined by stress-induced anisotropy. Chopra et a1.
(2000) showed that the magnetoelastic constraints at the TbFeIFeCo interfaces (due to
different values of the magnetostriction in adjacent layers) lead to biaxial stresses. They
developed a model, which expresses the magnetic coercivity as follows

(26)

Here, E is Young's modulus of TbFeCo.


MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 145

30
(7nn~/'''~
25 I

III
Q.
~

,Q-

10

5
(a)
0
• ~ .4) 20
mT
4D 8D 8D

IJaH.
45

:. -15
:IE
,Q= -10

o 20 40 60
IJoH- mT

Fig. 40. A comparison of the magnetoelastic coupling coefficient of an annealed (Th40Fe60)/(FesOCoso)


multilayer with the optimised Th40Fe60. SmFeB (a) and (ThISFeS2)/(Fe7SCOZS), (ThDyFe)/(FeSiCuNbB),
(Th33Fe67)/(FesoB20) (b) ones. After Ludwig and Quandt (2000).

This model succeeds in predicting the correct order of magnitude of the experimental
coercivity values in these MSMM.
In attempts to improve the soft-magnetic properties of the highly magnetostrictive
nanocrystalline layers (see above, and see section 5.l) by preparing MSMM's with
soft magnetic interlayers, Farber and Kronmuller (2000b) have studied ThDyFelFinemet
multilayers (Finemet is a nanocrystalline FeSiBNbCu soft magnetic alloy; see Herzer
1997). After production, the deposited ThDyFe as well as the Finemet are amorphous.
Subsequent annealing leads to relaxation of the amorphous phases and to crystallisation in
the (soft magnetic) Finemet layers. The properties of the multilayer can be described by
eqs (25a-25c), i.e. by the mixture rule of two perfectly coupled components. Applying
eq. (25c), the contribution of the individual layers to the magnetoelastic coupling
146 N.H. DUC and P.E. BROMMER

0.12....--.....---,----.--,.--.........--,----....------,

II

0.06 ..

0.00 ~-------~

.L

41 0.1 Q.2

ClO6 r-----.---..-----.---~-__.-:--_,._---.,....---...,

aea II

.L (b)

-4.1 0.1 Q.2

Fig. 41. Magnetostriction loops (parallel and perpendicular measurement direction) of an as-deposited (open
symbols) and a magnetic-field annealed TbO.4FeO.6(7 nm)lFeo.5COO.5(9 om) multilayer (closed symbols):
(a) annealing field aligned perpendicular to the long axis of the cantilever (parallel measurement direction),
(b) annealing field aligned parallel to the long axis of the cantilever. Magnetostriction in arbitrary units, field
in tesla. After Ludwig and Quandt (2000).

was deduced from their magnetoelastic data presented in fig. 44. Deviations from a
linear behaviour are ascribed to the compensation of the magnetic moments. For these
multilayers, the TbDyFe (amorphous Terfenol-D) contribution (b y ,2 = -15 to -18 MPa)
and the Finemet contribution (b y •2 = 15 MPa) have opposite signs. whereas the Fe
contribution (b y ,2 = 2 MPa) is small. with the same sign. For the bulk materials.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 147

(a)
Htl
o
IV
a..
~ .1
.....
':a
-2

-3 y&
~x Ht,
(IbFC'JFe)x40
~=O'
as-sput.
-4

1 tJ;;-
• a..
X fL, 8"

.
-2 ~.'
...~:r
.'.Ii
'

•••
•"0•i .
iI
",
~
~
.
.
"

••
~.
. .- (IbFeJFe)x40 ~ ~

-3

.'
..• ""
'

.. 0 '
.
~
.:
:
*
Hd• p 0
as-sput
". "
,,~~

",
-4 L---''''''-.......I_--'-_--L_........ _-L- ---J

·100 -so o so 100

J.1oH (mT)
Fig. 42. Magnetoelastic coefficient versus magnetic field applied parallel and perpendicular to the long axis of
the cantilever (parallel measurement direction) of an as-sputtered TbFelFe multilayer deposited (a) without a de
field (Hdep = 0) and (b) with Hdep #- O.After Le Gall et aI. (2000).

Farber and Kronmuller (2ooob) quote the b y •2 values -22 MPa. 6 MPa and -6 MPa for
a-Terfenol-D, Fe and Finemet, respectively.
As regards the temperature dependence of the magnetoelastic-coupling coefficient.
multilayers show a great advantage with respect to single films. Measurements by Ludwig
and Quandt (2000) are presented in fig. 45 for a (Tb40Fe60)/(FesoCoso) multilayer. Note
that. upon increasing the temperature. the slope of the magnetoelastic loop as well as the
coercive force remain relatively constant. whereas the saturation field and the saturation
value of the magnetoelastic-coupling coefficient are reduced. At 225°C. the saturation
magnetostriction is still 20% of that at room temperature.
148 N.H. Due and P.E. BROMMER

1.0 (TbFelFe)x40
Hdep :# 0
.
~: ~
.
0.5 as-spur,

f 0.0
IL,-+;':
Y fiJ.
~.~ ~~
-1.0P==::::::.._--'"··::..··-,.; L.::::+. a, I

-100 -50 o 50 100


J1JI (mT)
Fig. 43. Easy-axis and hard-axis hysteresis loops of an anisotropic. as-sputtered ThFelFe multilayer. deposited
under a dc field (Hdep # 0). After Le Gall et al. (2000).

.-.. 20 0
=:I
=..
:;
'-'
::::: 15
.Q
.-..
~
,!::
J
\; 10 ........
J .............

-
+
'-'
I
5

0
0.0 0.2 0.4 0.6 0.8 1.0
tsoj/tnrlc
Fig. 44. Plot of the weighted magnetostriction. -(I + tsof!ltstric)bll' of (D) Terfenol-DlFe and ('il) Ter-
fenol-DlFinemet multilayers, as a function of the thickness ratio. t",ftltstric. of the soft magnetic layers (Fe and
Finemet, respectively) and the magnetostrictive ones (Terfenol-D), After Farber and Kronmiiller (2oooa. 2ooob).

The large magnetostrictive susceptibility in Tb-FeCo/FeCo multilayers and its technical


character were reported and discussed in detail by Due et al. (2001a) in a study of
Terfecohan/Fe and Terfecohan/(Feo.sCoo.s) multilayers. Like in magnetic hysteresis loops,
there is a so-called magnetostriction remanence (Ar) and a coercive field J...Hc, where A = 0
in the magnetostrictive hysteresis loop (see definitions in fig. 38). Note that, experimentally,
J... He is observed to be equal to M He. Due to this magnetostriction remanence, the

possible change of the magnetostriction in (re)magnetising fields (i.e. in H > 0), 6A =


AS - Ar. is smaller than that in demagnetising (reverse) fields (i.e. in H < 0), where
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 149

25 50 75

Fig.45. Normalised magnetoelastic coupling coefficient of a Tb0.4Feo.6(7 nm)/Feo.5CoO.5(9 nm) multilayer. as


a function of external field at different temperatures. After Ludwig and Quandt (2000).

/),.).., = )..,S - )..,(>.. H c) = )..,s. The low-field dependence of the magnetostrictive susceptibility
is presented in figs 46(a, b) for the TerfecohanlFe and Terfecohan/(Feo.sCoo.s) multilayers,
respectively. As can be seen from these figures, the field direction dependence mentioned
above may not be so clear for the case of the as-deposited films (see curves I in
figs 46(a, bj), but certainly becomes rather important for the annealed films (see curves 2
in figs 46(a, bj), For the annealed TerfecohanlFe multilayer, X>"II shows a maximum
(of 1.3 x 10- 2 T-I) in low magnetising fields. A sharp and larger maximum (of
3.5 x 10- 2 T- 1) , however, is exhibited in demagnetising fields just above >..Hc. The
as-deposited Terfecohan/(Feo.sCOO.s) multilayer shows even a rather small and broad
magnetostrictive susceptibility maximum (X>"II.max = 0.7 x 10- 2 T- 1 only) in magnetising
fields. In demagnetising fields, X>"II.max reaches a value of 5.0 x 10- 2 T- 1, which strongly
exceeds that of 2.0 x 10-2 T- 1 required for the application in magnetostrictive devices
(Du Tremolet de Lacheisserie 1993). This magnetostrictive susceptibility maximum in
demagnetising fields could be considered as a good working point for the magnetostrictive
films in MEMS. For the samples under investigation, the working point is at JLoH dc ~
6 mT. By applying a bias magnetic field at this magnetic field, a magnetostriction change
/),.).., ~ 10- 4 can be obtained in an a.c. field of the magnitude of 2 mT.
Attempts to reduce JLoHc also imply the possibility to shift the working point to lower
fields. For this purpose, Quandt and Ludwig (1999) have prepared TbFeJFeCoBSi multi-
layers. It was shown that the FeCoBSi layers have improved the magnetic softness of the
multilayer. Due (2002) and co-workers have succeeded to prepare Terfecohan/(Yo.zfeo.s)
multilayers with JLoH c = 0.5 mT (see fig. 47b). Initially, this multilayer consists of amor-
phous TbFeCo layers and not-well crystallised (Y, Fe) layers. In this state, the multi-
layer exhibits already a soft magnetic and magnetostrictive character with a coercivity
JLoH c = 3 mT and a parallel magnetostrictivesusceptibility X>"II.max = 3.8 x 10-2 T- 1 (see
150 N.H. DUC and P.E.BROMMER

0.04
(8) 9 9

0.03

1.-
0.02
~

0.01

0
0
-0.1 -0.05 0 0.05 0.1
0.05
": '":
I I
(b) j ~
0.04
.'" "" 2
.' -,
"
" "
"
0.03 ':" :'"
1.- "
" "
" "
" ""
~ 0.02 ~ ::
q q;
~~4
o.<;fl',0 t

--
0.01

0.00
~
-:e% ~
Q:.

-0.1 -0.05 0 0.05 0.1


Il DU (I)

Fig, 46. Low-field dependence of the absolute value of the parallel magnetostrictive susceptibility for
TerfecohanlFe (a) and Terfecohan/(Feo.5COO.5) (b) multilayers: curves (I) as-deposited, curves (2) annealed
multilayers. After Due et al. (200la).

figs 47 and 48(a. bj), This magnetostrictive softness has been strongly improved by heat
treatments: ll-oHc = 0.3 mT and X}"II = 13 x to- 2 T- 1 in a field of 1.8 mT (fig. 48(a. bj).
These novel properties are associated with the development of a nanostructure in the (Y.
Fe) layers. As indicated by the XRD results (fig. 49), the as-deposited 'Ierfecohan/(Y, Fe)
multilayer shows no clear Bragg peaks. but only a rather broad contribution located around
20 = 45° (i.e. the (110) reflection of bee-Fe). Along with the results of the Mossbauer stud-
ies (see fig, 50). this could imply that the (Y, Fe) layers are amorphous, The (Ito) Bragg
reflection of the bee-Fe structure is enhanced for the samples annealed at T A = 250°C.
Finally, they are visible in the sample annealed at 350°C (fig. 49), This transformation
from the initially not-well crystallised phase into the bee-phase in the (Y. Fe) layers is con-
firmed by the characteristic value of the hyperfine field (Bhf = 33 T) observed from the
Mossbauer spectrum (see fig. SOc).
Aside from the hE effect which is an interesting feature for designing resonant
mechanical devices. MSMM show a GMR (hR) effect (Quandt and Ludwig 1999.2(00).
Thus, as already introduced in section 3.3. the bending of a magnetostrictive bimorph
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS lSI

Fig. 47. Magnetic hysteresis loops for Terfecohan/(YO.2FeO.g)multilayers: (a) as-deposited. (b) after annealing
at 350°C. After Due (2002).

can be monitored by measuring the magnetoresistance. A comparison of the normalised


magnetoresistance and the magnetostriction effect for an annealed (Tb4oFe60)/(FesoC050)
multilayer is shown in fig. 51. The figure exhibits that both loops are nearly identical. This
allows combining actuator and sensor functions at the same time.

5.3. Interface magnetostriction ofmultilayers


The magnetostriction of multilayers is found to change as the thickness t of the
magnetic layer is changed (Zuberek et al. 1987. 1988; Awano et al. 1988; Dime and
Denisson 1989; Nagi et al. 1988). These changes in the magnetoelastic properties have
been attributed to magnetostrictive strains. which are localised at the interface. Such
magnetostrictive effects are called "surface magnetostriction" (Szymczak et aI. 1988).
Similar effects are observed for the surface magnetic anisotropy (Gradmann 1993). Surface
and interface effects were already discussed in section 2.3. It is often claimed that the
changes in elastic and magnetoelastic properties of multilayers are due to the presence
152 N.H. Due and P.E. BROMMER

200

ISO
,,~

-=
'-"
'"
.<
100

SO
(a)
0
-IS -10 -5 0 5 10 IS
IlJl (mT)

0.15
(b)

--
-'E-
:::::
..:
0.10 2

~
0.05

0.00
-15 -10 -5 o 5 10 15
J.LoH (ml)

Fig. 48. Low field dependence of the parallel magnetostriction (a) and the magnetostrictive susceptibility (b) for
Terfecohan/(Yo.zFeo.s) multilayers: (I) as-deposited, (2) annealed at 350°C. After Due (2002).

of interdiffusion layers. which are formed at the interfaces. Nowadays, however, one can
fabricate multilayers, which have negligible interface diffusion. Nevertheless, their surface
anisotropy and magnetostriction are considerable. Independent experimental evidence for
the intrinsic character of the surface magnetoelastic coupling has been given by Sun
and O'Handley (1991), who used the spin-polarisation of secondary electrons, detected
from the asymmetric spin-orbit scattering, to monitor selectively and directly the surface
magnetisation (see section 3.5). These measurements were performed using amorphous
ribbons and therefore interdiffusion processes had no effect.
Szymczak (1997, 1999) has stressed the possibility to distinguish between pure
interface effects, i.e. surface magnetostriction, and the effects of an interface diffusion
layer. Since magnetic anisotropy and magnetostriction have the same origin, the surface
magnetostriction is expected to have an intrinsic character. In Szymczak's notation (Voigt
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 153

~10)(Fe,CO)

--"-.a T A =350°C

-
:I

of T A =250 °C

'-'

As-dePCIsited

35 40 45 50 55
2 theta (degrees)

Fig. 49. X-ray diffraction patterns of a Terfecohan/(Feo.85Coo.15) multilayer. After Due (2002).

Velocity (mm/s)
o

"--1 :il;zsJ
+12

0 10 20 30 40
1.00
=
e
1.02 c.

.-
::1::2\1
';1 (b)
Gf)
"":::l.
e
.&I
S
~
~

1.00 0 10 20 30 40
1.02
20
15 (c)

10
5
0
0 10

Fig. 50. Mossbauer spectra (CEMS) and hyperfine-field distributions of a Terfecohan/(YO.2Feo.8) multilayer:
(a) the as-deposited film, (b) after annealing at 250°C and (c) 350°C. After Due (2002).
154 N.H. DUC and P.E. BROMMER

1.0

II
o.s -0- "IIII'*" " iCI!

-
.<
~1i'Y"'idiai

is u
45
l.

-1..0
-1GD -75 -25 0 2i 50 75 100
PaH.. mT
Fig. 5 I. Comparison between the normalised magnetostriction and the normalised magnetoresistance of a
ThOAFeo.6(8 nm)lFeO.5CoO.5(9 nm) multilayer annealed at 250°C. After Ludwig and Quandt (2000).

notation), it means that the magnetoelastic tensor [M] in artificially structured materials
should consist of two parts:

(27)

where [M]bulk describes the bulk magnetoelastic coupling and where [M]surf describes
surface magnetoelastic coupling and/or the contribution to [M] due to the interdiffusion
layers. This relation should be compared with the relation in our notation, discussed
in section 2: BelT = B bu1k + 2b surfjt, where the factor 2 is put in, because a layer has
two surfaces. The components Mil and MI2 are related to V· 2 = -By·2jc Y and A£,2 =
- Bd j c", respectively, in such a way that, for (bulk) isotropic material, one has MIl = As
and the 'isotropy condition' Mil = -2MI2. The intrinsic surface magnetoelastic tensor
is expected to be strongly anisotropic, while magnetoelastic interactions in interdiffusion
layers lead usually to an isotropic form of [M]surf. Information about the symmetry of
[M]surf, therefore, is of great importance to determine the intrinsic nature of magnetic
interactions in the interfaces. As an example, we give here some results obtained by
Szymczak and co-workers (see e.g. Szymczak 1997) applying the SMFMR technique (see
section 3.4). Experimental data obtained for several Ni-based multilayers are presented
in fig. 52(a, b). The fitting parameters for the magnetoelastic tensor components Mij
are listed in table 3. In isotropic approximation, (Mil )bulk should be equal to the
magnetostriction constant for polycrystalline material As. The difference between As
(= -35 X JO-6) of polycrystalline nickel and (Mil )bulk was attributed, for Ni/Ti, with
(Mldbulk = -45 x 10- 6 , to the texture of the magnetic sublayers, and for Ni/Pb, with
(Mil )bulk = -19.5 X JO- 6, to the strong reduction of the magnetisation (Zuberek et
aJ. 1991). The strongest surface-tensor anisotropy, i.e. the ratio (MI dsurf/(MI2)surf, was
observed for Nil Ag multilayers. This is taken as clear evidence for the existence of
intrinsic surface magnetostriction. The SMFMR technique has also applied to study
the magnetostriction in Co-based multilayers (see Szymczak 1997). In these cases,
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 155

20-r--------------,
....
* -
I~
.-
-..
~
' -'

i
0

-- • --'&

~ -20- - ; MIJ

.--.
Jt--

!'P--- (a)

0.0 0.1 0.2 0.3 0.4 0.5

l/tNl (am")

15

.,=
-..
'-'

i -1
0

*" "
-." "
.
"

.- "
...."
.-"
""
.-
.-"
i "" .-.-
-3
, "" ,At

.- .-"
.- "
"...... Mil
(b)
.- .-
-4
0.0 0.1 0.2 0.3
I
l/tNl (nm- )

Fig. 52. Dependence of the magnetoelastic tensor components Mil and MI2 on the inverse Ni layer thickness
for several Ni-based multilayers. (a) (.&) NiIPb and (*) NilAg. (b) (.&) NilTi and (*) NilC multilayers. After
Szymczak (1997) and refs therein.

however, no indication of surface magnetoelastic coupling was observed (e.g. in Co/Ag


multilayers).
GdlFe multilayers fabricated by rf-sputterring exhibit clearly a magnetostriction con-
tribution due to interdiffusion layers. Samples deposited by de-sputtering, however, have
sharp interfaces. The variation of their effective (saturation) magnetostriction O'eff) is de-
scribed as a function of the inverse of the iron sublayer thickness, tFe -I, (see fig. 53). In
this case, the magnitude of the bulk (Abulk) and surface magnetostriction (Asurf) can also
be deduced assuming the following relation:

Aeff = Abulk + 2tie' Asurf. (28)


156 N.H. DUC and P.E. BROMMER

TABLE 3
Values of M II and M 12 tensor components and their ratio MIl / M 12 for several multi layers

Samples Bulk Surface


Mil MI2 MII/MI2 Mil MI2 MII/MI2
(10-6 ) (10- 6) (10- 6 ) (10- 6 )

Niffi -45 21 -2.14 67 -31 -2.16


NilC -32 17 -1.88 SI -45 -I.S0
NilAg -36 17 -2.12 II -7 -1.57
NilPb -19.5 8 -2.44 14 -6 -2.33

After Szymczak (1997).

-1,---------------,
,•
GdlFe multilayers "
-2
, ,
, ,,
,
,,
,,
, ,"
,,
-5--
,,
*
0.0 0.1 0.2 0.3 0.4 0.5

lItF e (nm")

Fig. 53. Magnetostriction as a function of the inverse Fe-layer thickness for GdlFe multilayers. After Szymczak
(1997) and refs therein.

It turns out that Asurf (= 10.2 x 10- 6 ) exceeds the observed bulk magnetostriction Abulk
(= 6.5 x 10-6 )
(Szumiata et al. 1993; Zuberek et al. 1995).
An excellent agreement with relation (28) was obtained for the magnetostriction
data of FeCo/Au multilayers (fig. 54) (Zuberek et al. 2000), from which the values
6 6
Abulk = 57.8 x 10- and Asurf = -18.2 x 10- were deduced. Note that the obtained
value of Abulk corresponds to that of polycrystalline FesoCoso alloys (Bozorth 1951).
Lafford et al. (1995) reported a similar analysis for the magnetostriction of C032Pd6S!Ag
multilayers (see fig. 55). Unfortunately, since only one magnetoelastic tensor component
(or saturation magnetostriction) is determined, definite conclusions cannot be drawn from
such experiments.
Investigations on SmFeBffbFeB multilayers have shown that strain and stress are trans-
ferred effectively at the interface (Shima et al. 1997). In these SmFeBffbFeB multilayers,
the thickness of the layers was varied, and it was found that the magnetostriction is sensi-
tively affected by Young's modulus, the Poisson ratio and the thickness of the constituent
layers.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 157

FeCo/Au multilayers


35+-~-r----.....,.---r-~-,--""'---r-..---i
0.0 0.1 0.2 0.3 0.4 0.5 0.8

Fig. 54. Magnetostriction as a function of the inverse Fe-layer thickness for FeCoiAu multilayers. After Zuberek
et al. (2000).

C0 32Pd68!'Ag multilayers
......
'f

-
=
.... .&JO
!

-
)
a~ .1000

·1500

·2000
0 2 4 , 10 12 14

teoN (om)
Fig. 55. Aeff· tCoPd as a function of tCoPd for CoPdlAg multilayers. After Szymczak (1999) and refs therein.

The changes in magnetic and magnetostrictive properties of multilayers which are


due to the presence of interdiffusion layers formed at the interfaces, were studied by
Quandt et al. (I 997a, 1997b) and Due et a1. (2000<1), for TbFeCoIFe multilayers. In
these multilayers, the amorphous state was formed in the giant magnetostrictive TbFeCo-
layers, while the Fe-layers were crystallised. This is confirmed by XRD investigations.
Annealing at temperatures T A from 150°C to 350°C made the interface spacers to expand
and the core of the individual layers to shrink. Assuming the same value of magnetisation
for the TbFeCo-layers and the TbFeCoIFe interfaces, Due et al. (2000d) have estimated
the thickness of the extended interface spacer due to the interdiffusion (t x ) by applying
expression (25a) in the following form:

MFe(tFe - tx) - MThFeCo(tThFeCo + tx)


(M TbFeCo/Fe ) =
tFe
+ tTbFeCo . (29)
158 N.H. Due and P.E. BROMMER

rI' 1..........
..
..-;~~
... - .....
"' :.~

.
"''''
...

..
. .te':,....
;... •

~
..., ...
-

.\
>

I
I
300 I
I
I

-.
I

200 T crysl (single: film)


E
:r:"
~100 1:
I
I
I

700 750 800 850


T.[K]

Fig. 56. The coercive field of different Terfenol-DlFe multi layers with Terfenollayer thickness Istric = 10 nm,
and Fe layer thickness: Isoft = 2 nm (e), 6 nm (TSB) and 8 nm (.). The dashed line shows the crystallisation
temperature of a 2-/Lm thick Terfenol-D film. After Farber and Kronmiiller (2000a, 2000b).

At T A = 350°C, Ix was found to be about 1 nm. Auger electron spectroscopy (AES) depth
profiling showed that the interdiffusion was not found after annealing at 280°C, but that
the layer structure was almost destroyed after annealing at 480°C (Quandt et al. 1997b).
Annealing at temperatures above 820 K destroys the multilayer structure and then the
film loses its soft-magnetic character. This is clearly seen from the microstructure graph
illustrated in fig. 56.
Annealing at medium temperature leads to an increase of the magnetostrictive suscep-
tibility and a decrease of the saturation field (as well as coercive field), while annealing
at higher temperatures usually reduces dramatically saturation magnetisation and then the
saturation magnetostriction due to the expansion of the interface spacers. This is due to the
formation of the so-called 'magnetostriction dead layers' (Farber and Kronmiiller 2000a).
In case, however, a perpendicular magnetic anisotropy is created, associated with the mod-
ification of the structure and/or composition in the interface, the magnetostriction can be
enhanced slightly. This phenomenon is observed in TbFeCoIFe multilayers annealed at
350°C (Duc et al. 2000d).

6. Magnetoelasticity of rare-earth superlattices

Magnetic rare-earth superlattices RIM (M = Y, Lu) behave in a variety of ways. Two


remarkable features in these artificial structures are: (i) helical magnetic order is found to
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 159

propagate through non-magnetic layers and (ii) different magnetic phases are identified in
the superlattices when comparing to bulk elements. The first feature has been explained in
the framework of the RKKY interactions through a spin density wave and the discreteness
of the interleaving material, which produces an increase of the magnetic period. The
strain induced in the crystalline structure by the mismatch between layers of different
elements was thought to be responsible for the second magnetic feature. The strain can
couple to the magnetisation either by modifying the indirect exchange as was suggested
to explain the suppression of the conical c-axis ferromagnetic phase transition in ErfY
(Borchers et al. 1991), or by altering the energy balance between the exchange and
magnetoelastic contributions. The latter mechanism accounts for magnetic changes in
Dy-based supersublattices. Bulk Dy has a ferro-helix first-order transition at 85 K. It is
suppressed or enhanced up to 100% depending on whether the Dy lattice was expanded (in
DylY) (Erwin et al. 1987) or compressed (in DylLu) (Beach et al. 1993), respectively.
O'Donovan et al. (1998) performed low-temperature X-ray measurements on a DylLu
superlattice and found that the spontaneous ferromagnetic transition is accomplished by
(continuous) formation of orthorhombic domains which preserve the area of the unit cell
in the basal plane. The importance of epitaxial clamping in this system was emphasized.
Similarly, the c-axis cone phase was observed to be suppressed in HolY (Jehan et al.
1993), while ferromagnetic order existed below 30 K in HolLu superlattices with less than
20 Ho-atomic planes (Swaddling et al. 1992).
As regards the stress due to the misfit of lattice parameters, magnetoelastic studies of
superlattices were carried out by Ciria et al. (1995), Del Moral et al. (1996, 1998) and
Amaudas et al. (1996), applying a low-temperature cantilever technique. Their results have
elucidated the role of epitaxial strains. The magnetoelastic coefficient By·2 (corresponding
to the distortion of the cylindrical symmetry of the superlattice basal plane; see section 2,
eq. (9» was determined from the magnetoelastic stress measurements (fig. 57). The thermal
variation of BY measured in a field of 12 T is plotted in fig. 58 for the {H06fY6}100 and the
{H0311Lu 19 }50superlattices. (The subscripts denote the number of atomic planes per layer,
and the number oflayers in the superlattice, respectively. Alternatively: [H031 /Lu 19] x 50.)
The BY values for the {H06/Y61100 sample are much larger than those for bulk Ho.
Moreover, its thermal variation deviates from Callen and Callen's law of the cubic power
variation of the reduced magnetisation (m 3 ) at low temperature for bulk rare earths. For
the HolLu superlattices with thick Ho and thin Lu layers (e.g. {H03t1LuI9}) the BY(T)
variation approaches that of bulk Ho at high temperatures, but below 50 K it shows a
deviation. This was attributed to an opposite-sign surface magnetostriction which scales as
m4 at low temperatures and as m 2 at high temperatures. At low temperatures, BY (T) is
simply expressed as

(30)

with BVol Y (0) = + 1.43 GPa and B~urf(O) = -1.38 GPa. The surface magnetostriction is
masked in {H06/Y6} I()() due to the large epitaxial effect.
Magnetoelastic stress isotherms (T a and (Tb of the {H040/Lu 15 }50 superlattice, clamped
along the a and b axes of the hcp structure, respectively, and with the applied magnetic
field along the easy b axis are presented in fig. 59 (Del Moral et al. 1998). The sudden
160 N.H. Due and P.E. BROMMER

l
9-
1O
--4.05
,
4
e
~.I ·z
o
o 2 4 6 • 10 12
Appliedfield (T)

Fig. 57. Magnetoelastic stress isotherms for a (H06fY61100 superlattice: 0"0 and O"b correspond to clamping
along Q- and b-superlaltice axes, respectively. Inset: magnetic phase diagram «0) from magnetoelastic stress and
(e) from magnetisation measurements; FM- ferromagnetic. F - fan and H - helical phases). After Del Moral
et al. (1996).

0.5
• • •
0.4 •• • • •
'2 0.3 •
tS........
~ 0.2
...........
I:Q ~o\ ;~.~ ~ ~.
•• •• b. {HoJ-lI6l!lO •
0.1 ..
., ~.

0
0 50 100 150
Temperature (K)
Fig. 58. Thermal variation of the magnetoelastic coupling parameter BY for the superIattices (H06fY61100 (e)
and (H031 fLu 19150 (6), and for bulk Ho (dashed - - -). The dash-dotted and continuous lines are calculated (see
main text). After Amaudas et al. (1996).

onset of stress and saturation at low temperatures is interpreted as a direct transition from
a helical to a ferromagnetic state. As the temperature is raised, the strain increases less
abruptly, indicating the transition to a fan structure. The variation of the magnetoelastic
coupling coefficient BY (T), or MY (= 2(0' a - O'b» in the notation of Del Moral et al.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 161

o
[Ho.JLuu1x.SO

~I
\ -0.1-

I"---
"~-.:~=:::====~1.cac:
:: 110< fJ
.e
IlOK
---------==
~
r,-------------1OK -0.2
0

-eel
Q..
0....-
-0.04
m~~==========o4OK ~

II:)
.. &oK

-0.08

o 246 8 W ~
Applied field (T)

Fig. 59. Magnetoelastic stress isotherms for a IH040/Lu IS150 superlattice: Ua and Ub correspond to clamping
along the Q- and b-superlattice axis, respectively. After Del Moral et al, (1998).

(1998), as a function of the number of Ho-atomic planes at T = 10K is presented in fig. 60


for {HonlLu15}50 superlattices. MY was expressed as:

(31)

Here, the first term describes the volume (bulk) value of MY, the second term is the
interface contribution and the third term is determined by the lattice misfit. It turned
out from this analysis that M~ol = 0.275 GPa. M:Urf/(c/2) =
-7.0 GPa and D~ol =
-116 GPa, where c is the c-axis lattice parameter for Ho. This leads to the following
conclusions: (i) M~ol equals the bulk value for Ho, (ii) the interface magnetoelastic stress
is very strong compared with the volume one. up to 6.4 times larger for nHo = 8 and of
opposite sign and (iii) the misfit stress is of the same order of magnitude as the interface
magnetoelastic stress. In the spirit of the description of eq. (30) and assuming 'single-ion'
CEF contributions for the rare-earth ion, the expression for M~ol can be generalised as

(32)

Here, i 5/2 is the reduced Bessel function, and its argument L -I (m) is the inverse
Langevin function (see Callen and Callen 1963, 1965). Using the values given above for
162 N.H. Due and P.E. BROMMER

---. 103 103


~
c,
0
'-' [Hon I LUIS] X 50
---. 1()2 > 1()2 3:
::s ~-. .c!
...J
C ::s
+0 ::t
0
::t 10'
c:: t 10 ' ---.
'-' 0
ci. "tl
~
~:l '-'
::E 10° 10°
10' 102 103 104
nHo

Fig. 60. The variation of the basal plane cylindrical symmetry breaking magnetoelastic stress, M[xp. at 10K and
at an applied magnetic field of 12 T, multiplied by (nHo + nLu). as a function of nHo (where nHo and nLu are
the number of atomic planes in the Ho and Lu blocks. respectively) for IHo,.f[..u I S I so superlattices. The line
indicates the fitting by the theoretical model. After Del Moral et al. (1998).

0.7
X HO,.ILu,~
0.6
{;.
HolOlLuu
---.0.5
~
{;. • Ho",ILu,~
p..
o 0.4
Ho./'-u ,S

'-'
0.3

::E 0.2

0.1
O'--_.l-_-'--_-'--_...J-_~_ ......- - L . . - - - J
o 20 40 60 80 100 120 140
Temperature (K)
Fig. 61. The variation with temperature of the magnetoelastic stress, MY (IZ T) multiplied by (nHo + nLu)/nHo
for IHolIlLuls lso superlattices. The lines are obtained by adapting reduced-magnetisation power laws. see Del
Moral et al. (1998).

M~ol' M:urr/ (c / 2) and D~ol' and taking ex = 4 below a certain temperature and ex = 2 above
that temperature, the fitting obtained for {HolIlLuI5}s0 superlattices with n = nHo = 14,
30,40 and 45 (fig. 61) is reasonably satisfactory. This, once more, confirms the relevance of
the interfacial stress contribution to the magnetoelastic stress. In addition, it does suggest
strongly that both volume and interface magnetoelastic coupling originate from single-ion
crystal-field terms.
Using X-ray spectrometry, De la Fuente et aL (1999) measured the thermal dependence
of the a and c lattice parameters in a {Er3ZlLu IO}40 superlattice. Again, strong single-ion
CEF contributions, originating from the ErlLu interfaces, were observed in the volume
and tetragonal distortions. Their analysis reveals also important contributions caused by
epitaxial misfit.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 163

7. Magnetostriction of R·T sandwich films

Sandwich films of the type RTIR'TIRT made by stacking coupled layers with typical
thicknesses of 100 nm have been intensively studied in the last decade (Dieny et al.
1990, 1991a, 1991b; Givord et al. 1993, 1996; Wiichner et al. 1995). Because properties
such as magnetisation or anisotropy differ from one layer to the next, the reversal of the
magnetisation occurs at a different value of the coercive field for each layer. When the
reversal takes place in a given layer but not in the adjacent one, a domain wall will be
formed at the interface between the layers, in order to minimise the exchange energy.
Such a domain wall has the particular feature of extending over the whole film surface
and therefore is referred to as an extended domain wall (EDW). In these systems, the
magnetostriction, which is fundamentally different from that observed in the multilayers
discussed in the preceding section, is associated with the creation of this domain wall.
Magnetostriction associated with domain wall formation has been known for a long time
as being due to the progressive rotation of magnetic moments making up the domain wall.
Normally, this effect is small since the volume occupied by the domain wall is always quite
small. In the systems with EDWs, however, the domain wall can occupy an extremely large
fraction of the total volume of the sample.
The effects ofEDW formation were investigated on sandwiches consisting of Nd-CofIb-
ColNd-Co (system 1) and Tb-Co/Nd-Co/Tb-Co (system 2), in which a well defined easy
axis was created by annealing at 150aC in a field of 2 T. The Tb magnetic moment
is dominant in the Tb-Co layers at room temperature (Givord et al. 1996; Betz 1997).
In order to better understand the magnetisation process, the coupling between one pair
of layers was suppressed by a thin oxide layer at the interface. The configuration of
magnetisation and of the magnetic moments in zero-magnetic field is illustrated in fig. 62a
for the sandwich system 2 (layer thicknesses 57/67/57 nm). The magnetisation loop
at room temperature is shown in fig. 63. Starting from the high-field state, where the
magnetisation of the system is well saturated in the applied-field direction, we see that
the Co moments between layers are antiparallelly coupled and an EDW is formed at the
coupled interface (fig. 62b). As the field decreases, a (positive) critical field (of 8.5 mT)
is reached, where the EDW at the Nd-CofIb-Co coupled interface is suppressed by the
reversal of the moment in the Nd-Co layer (fig. 62c). Note that magnetisation reversal
in the (coupled) Tb-CO layer is prohibited by the large coercivity (associated with the
sperimagnetic Tb moments). The reversal of the uncoupled TbCo layer occurs at -46 mT,
and, finally, for the coupled TbCo at a higher field of -96 mT due to the interlayer Co-Co
exchange coupling and the (re)creation of the EDW (fig. 62(d, ej). The magnetostriction is
shown in fig. 64. While the high-field magnetostriction exhibits a behaviour similar to that
observed in single layer Tb-Co films, the low-field magnetostriction measured along the
easy axis shows a rather complex field dependence with magnetostriction anomalies and,
in particular, extremely large magnetostrictive susceptibilities. In the Nd-CofIb-ColNd-
Co sandwich system, a magnetostrictive susceptibility (ob/o#LoR) of 556 MPa/f was
observed even at #LoR = 2 mT (Betz 1997). This field dependence can be understood on
the basis of the magnetisation process depicted above. At saturation, the magnetostriction
(which can be either positive or negative) is maximum. As the field is reduced, the
magnetostriction decreases as the EDW becomes larger (and the Tb sperimagnetic cone
164 N.H. DUC and P.E. BROMMER

(a) magnetic moment magnetisation


zero-field
un-coupled Tb •
, '..•.
........ Co '.n ..·.·.:. _ " ,-'.t.
.- Tb-Co

• ........ Co
> ••. ,,,,..:.,-- .•_."'"
interface ~ ~.', ''''''''~ .~.

Nd
• Nd-Co
coupled
interface
~
Tb • ........ Co .- Tb-Co

(b)
positive Tb - -•• +-- Co - . Tb-Co
high-field
Nd • ........ Co • Nd-Co
EDW-.j{~1~UF.~;~~
Tb • +-- Co - . Tb-Co

(c) Tb
..,
• +--
..
;..-"
Co -. Tb-Co
positive
~,

low-field Nd • +-- Co
• Nd-Co

Tb • +-- Co -. Tb-Co

(d)
negative
Tb • ........ Co .-
• " " ',.... -<~ ',~," .,
Tb-Co

low-field Nd • +-- Co
• Nd-Co

Tb • +-- Co -. Tb-Co

Tb ••- - ........ Co . - Tb-Co


(e)
negative Nd. +-- Co • Nd-Co
high-field ,I.'r.
~
~tl{~.~);~
.,;~u..:.'f'1.:
~:1:"""-~)J~t·:~,,c;,··~
• • _" _~!';llr:<61'", ••;lI;'!!V":(R~.".~
Tb· ........ Co . - Tb-Co

Fig. 62. Configuration of magnetisation and magnetic moment in the sandwich TbCoINdCofIbCo film in zero
magnetic field (a) and during the (de)magnetisation processes (b-e).
MAGNETOELASTIClTY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 165

0.015
Tb-Co/Nd-Co/Tb·Co
0.010 300K

N
...... 0.005
eu
':3 0.000
e
II)
- -0.005
:E
-0.010

-0.0 15
- 0.2 ·0.11 o 0.1 0.2
~H (T)

Fig. 63. Magnetisation versus applied field measured on the sandwich Tb-CoINd-CoITb-Co system at 300 K.
After Givord et aI. (1995).

0.0 IS

Tb·Co/Nd·Cofl·b-Co
300K hord
8xl.

...... -0.5
"'
0..

-
:::E

m -1.0

-1.5
·0.3 - 0.15 0 0.15 0.3
lloH (T)
Fig. 64. Magnetostriction versus applied field measured on the sandwich lb-CoINd-CoITb-Co system at 300 K.
After Givord et aI. (1995).

opens up a little bit). At around 8.5 mT, the EDW is destroyed. All Tb moments contribute
now to the magnetostriction, and thus a large change is observed. When the external
field decreases and changes its direction, the cone of Tb moments opens up further,
leading to the observed decrease of the magnetostriction. At the coercive field for reversal
of the magnetisation in the uncoupled TbCo layer (-46 mT), a discontinuity in the
magnetostriction takes place, since these Tb moments become more aligned along the
applied field. Further variation of the field opens the cone in the coupled TbCo layer
still more (evidently exceeding the closing of the cone in the uncoupled layer). Finally,
at -95 mT the EDW is re-created with a sudden increase of the disorder in the magnetic
166 N.H. Due and P.E. BROMMER

0.4 (a)

0.2

(25nm TbFei 8nm FeCoBSI)


0.0 t---------i+t--------~

-0.2

-0.4

-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6

·6
(b)
-4

-2

O~-----__1f_-----,.tt_-~+_----__r":I

81....--..._.L.-......._.L.--..._......~_.L____'__....L..___'___'
-0.6 -0.4 -0.2 0.2 0.4 0.6

Fig. 65. Magnetisation and magnetostriction versus applied field measured on a TbFe(25 nrn)/FeCoBSi(8 nm)
multilayer. After Quandt and Ludwig (1999).

configuration giving rise to a large transition in the magnetostriction. Still further increase
of the field does narrow the EDW and closes the Tb cone gradually, in both Tb-Co
layers.
Quandt and Ludwig (1999) studied the influence of domain wall formation on the
magnetic and magnetostrictive behaviour of multilayers TbFelFeCoBSi(8 nm) with various
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 167

700
600
500
~ 400
- 300
=
~
«:
200
100
o
-0.8 -0.4 o 0.4 0.8
Ii.H (1)
Fig. 66. Magnetostriction hysteresis loops of a ThFeCo single-layer film and a FetrbFeCoIFe sandwich films.
After Due et al, (200lc).

thickness of the TbFe-layer (see section 5.2). For a TbFe thickness of 25 nm, domain walls
(EDW) can be formed at the interfaces. and the magnetisation process is comparable to
that in the sandwich systems under consideration. In this case. the transition (rotation
of the TbFe moments into the field direction) occurred in a rather wide field range
(say. from 0.05 T to 0.3 T; fig. 65a). because of the high number of layers. which
are not absolutely identical. Like above. the field dependence of the magnetostriction
(fig. 65b) was explained in terms of the local orientation of the magnetisation vectors.
For positive magnetostriction materials (b < 0). a magnetisation orientation parallel or
antiparallel to the measurement direction results in a positive strain, while a perpendicular
orientation leads to a negative strain. In low fields. the layers are still exchange
coupled and thus the magnetostriction is similar to those of thinner layer thicknesses.
When the (TbFe) magnetic moments are rotated into the field direction. the domain
wall is formed, and some TbFe moments rotate out of the external field direction.
The negative magnetostrictive susceptibility is associated with this process. At higher
(parallel) fields, the domain wall width is reduced. and the (positive) magnetostriction is
recovered.
The sandwich system discussed above is a composite material. which gives evidence
for the existence of the domain wall and of its influence on the magnetostriction. These
systems are not optimised with respect to magnetostriction. Nevertheless. their high mag-
netostrictive susceptibility is interesting. For applications. it may be worthwhile to mention
here (fig. 66) the magnetostriction data of a Fe(25 nm)!fbFeCo(5000 nm)/Fe(25 nm) sand-
wich (Due et al. 200lc). In comparison with the corresponding magnetostrictive TbFeCo
single-layer film. the magnetic softness was strongly improved for this sandwich right after
depositing. The (not so high: All = 180 x 10-6) saturation magnetostriction is already well
developed at /LoR ~ 20 mT.
168 N.H. Due and P.E. BROMMER

8. Magnetostriction in nanocrystalline and granular magnetic materials

Nanocrystalline materials obtained by controlled crystallisation of Fe-rich amorphous


alloys are composed of crystalline grains embedded in a residual amorphous matrix. These
materials belong to a wide class of heterogeneous structures, including organo-metallic
complexes or metallic clusters deposited on graphite or build in polymers on a molecular
scale (see e.g. Gubin et al. 2000; Koksharov et al. 2000), and composites consisting of
nanoscale up to microscale grains embedded in either a metallic binder (see e.g. Herbst
et al. 1997; Pinkerton et al. 1998) or in some resin or polymer (see e.g. Duenas and
Carman 2000). Granular solids consisting of magnetic particles embedded in an immiscible
medium have been intensively studied in the last decade, because of the interest of their
physical properties as well as of their technological applications. Indeed, these materials
exhibit novel phenomena such as superparamagnetism, giant magnetoresistance and giant
magnetic coercivity (Chien 1991; Hernando et al. 1997; Due et al. 2001d and references
therein).
In this subsection we start by continuing, in a way, the discussion in section 4.3 on the
attempts to control the nanocrystallisation process. There, attention was focused on the
possibility to manipulate the magnetic anisotropy. In section 5.1, it was shown that the
crystal growth can also be limited by the layer thickness in multilayer structures In the
present section, we discuss in particular the magnetostrictive properties. Originally, the
excellent soft magnetic properties of these materials have been associated with the absence
of magnetostriction in order to minimise magneto-elastic energies. In the amorphous state,
however, the material reveals a (high) positive magnetostriction of As ~ +23 X 10-6 . In
the nanocrystalline composite material, this positive magnetostriction is compensated by
the negative magnetostriction of the crystalline material. The effective magnetostriction
of nanocrystalline alloys at different stages of crystallisation is illustrated in figs 67(a,
b) for (Fe,Cu,Nb,Si,B) alloys (see Herzer 1997 and references therein). This effective
magnetostriction Aeff is usually interpreted as the simple volume average of the positive
contribution from the amorphous matrix (Aam) and the negative one from the grains (Acr)
(Herzer 1991):

(33)

where p is the crystalline volume fraction.


As an example, fig. 68 shows the decrease of Aeff with increasing crystalline volume
fraction. Clearly, in order to attain near zero magnetostriction in nanocrystalline Fe-based
alloys it is necessary to have a large crystalline volume fraction with negative magnetostric-
tion. It was reported that Acr ~ -6 x 10-6 for a-Fe8oSi20 grains in the (Fe,Cu,Nb,Si,B)
system and that Acr ~ -4 x 10-6 for a-Fe grains in the (Fe,Zr,B) system (Suzuki et al.
1991). Tejedor et al. (1998) report a change of sign of the magnetostriction upon anneal-
ing Fe 73.5 Cu I Nb3Si 16.5B6 ribbons. With respect to nanocrystalline Fe 73.5 Cu I Nb3Si 13.5B9
rapidly solidified ribbons, Chiriac et al. (1999) stressed the importance of the 'inheritance'
of the short range order in the melt. Holzer et al. (1999) found that the low-temperature
temperature dependence of the magnetostriction in such ribbons follows a Bloch T 3/ 2 law.
See also the low-temperature studies of Skorvanek et al. (2000) on the crystalline-fraction
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 169

-
<D
I
30
(a)
,-:. :a--_I._,
Fe73.5Cu1N~SixB22.S.x
annealed 1h at T.

-.e
0
... ··········'··0
\ .
T"""
20 \ :
as cast \ :
\ :
C \ :
0 \ :
\ :
15
....III
'C
10 e :
Si = 16.5\ 0.13.5 at%
.,<>-

-0
Q)
c:
C) 0

• l>..
'. e-....o'
-·- ___ e
O.
.'
0·'

m
:2 -5 I.

0 400 500 600 700


Annealing Temperature, T« (DC)
15
to (b) Fe-Cu1 Nb3SixBz.x
I annealed1h 540'C

-
0
T""" o ./"'0-' o; SI+B(.\~)
10 I ,e, o :18.5 0 20.5
f/l
0 I • 22.5 0 23.5
-e I
,
I "'0...
C
0 5 6' bn.
15 ~=t-Fe·Cu,SixNb5.7B13

-....
'C
III
0
Q)
0
I
~Fe84Nb789
):::Fe-CUO_1Zr6-782.0
8'8-
b,-
'Q--
c:
C) '<>
m
::! -5
0 5 10 15 20
Si-Content, X (at%)
Fig. 67. The saturation magnetostriction, AS. of Fe-Cu-Nb-Si-B alloys: (a) influence of the annealing temperature
and (b) influence of the Si content in the nanocrystalline state. The figure includes the data for Fe-Nb-B (A.) and
Fe-(Cu)-Zr-B (6) alloys. After Herzer (1997) and refs therein.

dependence of the magnetostriction of Nanoperm-like (Fe,Nb,B) alloys. Upon increasing


the annealing temperature, and thus increasing the (nano)crystalline fraction, Chiriac et
aI. (2000a, 2000b) observed a change of sign of the magnetostriction in nanocrystalline
Fe90Hf7B3 ribbons. In a study of annealed nanocrystalline (Fe,Zr,Nb,B) melt-spun rib-
bons, Makino et aI. (2000) found minimal core losses in a Fes5.5Zr2Nb4BS.5 alloy, which
exhibited zero nagnetostriction.
In many cases, however, the above simple model did not fit the experimental results even
though the changes in ).am with the evolution of crystallisation were considered (Slawska-
Waniewska et aI. 1994; Herzer 1993). In this case, an additional surface contribution
was introduced. Indeed, nanocrystals usually have a diameter of 10-15 nm and hence a
170 N.H. DUC and P.E. BROMMER

<0

6 20

C 10
.Q
U

I
'1:

0+----------=__._----1
c
Cl
CIl
::i: -10 +-.....--.--.--.--..---,-~__r_~---l
0.0 0.2 0.4 0.6 0.8 1.0
Crystalline Fraction

Fig. 68. Dependence of the saturation magnetostriction, As, on the fraction of the precipitated bee Fe-Si grains
for samples annealed at different temperatures and times. After 1'warowski et al, (1995).

TABLE 4
The fitted parameters (see text) of the saturation magnetostriction for several nanocrystalline alloys and
multilayers

Samples Acr Aam AS(S/V) AS k Refs


(10- 6 ) (10- 6 ) (10- 6 ) (10- 6 nm) (10- 6 )

FeSSZr7B6Cu2 alloys -4 2.2 6.55 10.9 12.1 [1,2]


Fe73.SCul Nb3Sil3.sB9 alloys -6 25 3.37 5.57 -11.5 [2,3,4]
FelC multilayers 45.7 [5]
FeCoiAu 58.6 -18.2 [6]
FelGd -6.5 I [2]

[I] Slawska-Waniewska et al. (1996), [2] Szymczak (1999), [3] Slawska-Waniewska et al, (1997), [4] Gutierrez
et al, (1995), [5] Zuberek et al, (1994), [6] Zuberek et al. (2000).

large fraction of their atoms are surface atoms. Moreover, it was found from Mossbauer
measurements that the Fes5Zr7B6Cu2 nanocrystalline sample, annealed at 540°C, consists
of three components: crystalline, amorphous and interfacial phases with the volume
fractions 52.5%, 31.3% and 16.2%, respectively (Slawska-Waniewska et a1. 1997). The
hyperfine field Bhf is 33 T for the crystalline phase, distributes below 25 T for the
amorphous phase, and ranges from 20 to 35 T for the interfacial phase. Assuming that
the grains are spherical and 10 nm in diameter, the thickness of the interfacial layer was
estimated to be about 0.6 nm. Taking into account the different contributions, the effective
magnetostriction of the nanocrystalline materials has been approximated by

Aeff = pAcr + (1 - p)(A am + kp) + p(3A surf )1R. (34)

Here, the parameter k expresses the changes of the magnetostriction due to the change
in composition of the residual amorphous matrix with the evolution of crystallisation. In
fact, a linear approximation Aam(p) = Aam(0) + kp is applied. The last term describes
the surface effects in which R is the effective radius of the grains, i.e, 3/ R is the
surface to volume ratio for the spherical grains (SI V in table 4). Figure 69a shows the
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 171

'9 25 •
0
.....
..!'
20 I " ...
...
(a)

15 • experimental
l:
0
10
ts
'i::
'1ii 5
~l: 0
Ol p,"
III -5
~
(b)
'9 4
0
.....
..<."
l:
0
ts -2
-
'i::

Gi
Ul
0 -4
• experimental
l:
Ol -6
III 0 20 40 60
~
Crystalline Fraction p [ % ]

Fig.69. Saturation magnetostriction, AS, versus crystalline fraction, p. in (a) FeS9Zr7B4 and (b) FeSSZr7B6Cu2
nanocrystalline alloys. After Slawska-Waniewska et al. (1996).

magnetostriction for Fes9Zr7B4, observed by SMFMR (see section 3.4). In a first fitting
attempt, Slawska-Waniewska et al. (1996) did apply eq. (34), with Aam(o) + kp replaced
again by Aam (p), because the linear approximation mentioned above could not be applied.
Actual Aam(p) values were estimated from magnetostriction data obtained on a series of
amorphous Fe93-xZr7Bx alloys. Further deviations were ascribed to the 'snowflake-like'
crystallites (with high SN values). Figure 69b shows the same attempt for FeS5Zr7B6CU2,
where almost spherical grains with a diameter of 10 nm were observed. pm(p) values
were now deduced from magnetostriction data obtained on amorphous Fe91-xZr7BxCu2
alloys. The indicated linear contribution of the 'surface term' was deduced then. Since
for this series, a linear approximation for Aam(p) appears to work well, in later studies
eq. (34) was applied (at the same time replacing the value Acr = -7 x 10-6 by the
more common value for a-Fe grains Acr = -4 x 10-6 ) . Figure 70 shows the observed
saturation magnetostriction of Fe73.5Cu,Nb3SiI3.5B9 nanocrystalline alloys as a function
of the volume fraction of the crystalline phase p. Here, eq. (34) has been applied too.
The fitted parameters used to describe the experimental data are listed in table 4. The
positive value of k in the case of Fes5Zr7B6CU2 indicates that the magnetostriction of
the amorphous phase increases with a decrease of Fe content. This is in agreement
with the experimental results of Slawska-Waniewska et al. (1996) mentioned above. In
the case of Fe73.5CU I Nb3Si I3.5B9, the negative value of k reflects the decrease of the
magnetostriction of the residual, Band Nb enriched, amorphous matrix with increasing
crystalline fraction. The important feature of the results discussed is the difference of
172 N.H. DUC and P.E. BROMMER

15
'1/
0
-,-,
10
.....
~

~ 5 ..................~"." =:~, a

0
_·_~ •••••O·· _<;4·· · ~.O •.... Q •••• 1
-5
0.0 0.2 0.4 0.6 0.8 1.0
crystalline fraction

Fig. 70. Saturation magnetostriction, AS. versus nanocrystalline fraction in Fe73.SCul Nb3Sil3B9 alloys. The
numbers 1.2.3 denote the different contributions according to eq. (34). After Szymczak (\999) and refs therein.

the signs for the bulk and surface magnetostriction. It seems to be a general rule as
similar observations were found in multilayers and superlattices, see also sections 5
and 7. Notice, however, that Areas et a1. (1998) found that, at applying transverse fields
to nanocrystalline Fe73.sSilS.5B7Nb3CUJ samples, for low fields the total elongation
is negative whereas it shifts to positive values for increasing transverse applied fields.
This behaviour was explained as a consequence of spatial fluctuations of the average
magnetostriction, correlated in length to the domain width. Obviously, the problem of
surface/interface magnetoelasticity requires further studies, in particular, magnetostriction
measurements above the Curie temperature of the amorphous phase, where only the
ferromagnetic grains should give a magnetic contribution, simplifying separation between
bulk and surface effects.
Alves et a1. (2000) have studied stress-induced anisotropy in Finemet- and Nanoperm-
type nanocrystalline materials, like Fe74.3Nb2.7CuISils.5B6.5 and Fes4Nb3.5Zr3.5BgCul,
applying flash stress-annealing. In Finemet a strong transverse anisotropy was induced,
for example 2600 J/m3 after annealing at 610°C for 30 sunder 400 MPa, higher than by
field-annealing (about 10 J/m3 ) . These effects are assumed to be located within the grains
in magneto-elastic form due to tensile back stress exerted by the amorphous phase. In
Nanoperm type nanocrystalline alloys, the authors observed a low longitudinal anisotropy
(around 20 J/m3 ) , in spite of a negative magnetostriction of the Fe crystallites (around
-4.1 ppm). This observation was also interpreted by the correlation between the sign of
the induced anisotropy and grain size due to the contribution of interface properties.
The existence of the surface contribution to the effective magnetostriction of nanocrys-
talline alloys has been confirmed theoretically in terms of the dipolar model
(Szumiata et al. 1999). These authors showed that, due to the limited radius of the nanopar-
tides, additional magnetostrictive stresses are localised at the interfaces. The evaluation of
the influence of the dipolar interaction on the magnetostriction in crystalline grains of per-
fect spherical shape surrounded by a magnetic environment of about 0.5 nm with either
crystalline or amorphous structure has been calculated. A similar method was previously
used to obtain the surface and volume anisotropy (Draaisma and de Jonge 1988) and to
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 173

deduce the surface magnetostriction and the volume magnetostriction in ultrathin films
(Szumiata et a1. 1993), and nanoparticles (Zuberek et a1. 1998).
The effective magnetostriction of a system consisting of highly magnetostrictive
spherical grains embedded in a non- (or low-)magnetostrictive matrix has been studied
by Herbst et a1. (1997). In this case the magnetostriction of the system can be described as

Aeff= Acr p (1.l5-1.26v-0.2 Ip 2/ 3) , (35)


1 - 1.25v

where v is the Poisson ratio for the low-magnetostriction component. This model gives
a satisfactory description of the fill factor dependence of the effective magnetostriction
observed on composites consisting of SmFe2 particles (obtained by ball milling) in an Al
or Fe matrix. See also Pinkerton et a1. (1998), who reported on the trade-off between high
magnetostriction and desirable mechanical properties (hardness, density). For hot-pressed
SmFe2lFe, in an applied field of 1.9 T, the saturation magnetostriction varies from 100 ppm
for mechanically robust high-density material, up to 900 ppm for low-density material. In
an attempt to reduce the anisotropy, Pinkerton et a1. (1998) substituted Nd for Sm in melt-
spun Smt-xNdxFe2 ribbons, which are transformed into Sml-xNdxFe2lFe composites by
(optimal) annealing. The intrinsic coercivity H ci of the melt-spun ribbons decreases from
2.0 kOe for SmFe2 to 0.5 kOe for Srno.5Ndo.5Fe2. As x increases from 0 to 0.5, A of the
composites decreases roughly linearly with x from 430 to 80 ppm.
Duenas and Carman (2000) have optimised the magnetostrictive properties of compos-
ites containing up to 50% (microscale) Terfenol-D grains in a non-metallic binder (epoxy).
During preparation, the elongated grains are oriented in a magnetic field, and are thought
to form a fiber-like structure. Such composites are durable and are easily machined or
otherwise manufactured into complex shapes, but usually their magnetostrictive response
is comparatively low. By choosing an optimal Terfenol-D volume fraction, related to the
selected non-metallic matrix, a composite can be obtained of which the magnetostrictive
response is comparable to that of Terfenol-D itself.
Nan and Weng (1999) have developed a Green function method to determine self-
consistently the effective magnetostrictive properties of composites. The authors claim
that their method, in principle, can be used to study the effects of material constants and
microstructure, such as anisotropy, particle shape and orientation relative to the applied
magnetic field.
Armstrong (2000a, 2000b) calculated the nonlinear multiaxial magnetostriction versus
applied magnetic-field curves of oriented Terfenol-D magnetostrictive particulate compos-
ites for very general conditions of magnetic field orientation, multiaxial applied stress,
particulate volume fraction, and particulate aspect ratio. The applied curing field does
preferentially align individual magnetic particles along easy magnetisation directions and
promotes the formation of particle chains parallel to the field.
Applying an optical cantilever method, Ohnuma et al. (1999) performed magnetostric-
tion studies on the (Col-xFex)-AI-O granular system. They found a composition depen-
dence of the magnetostriction which was almost the same as that of bulk Co-Fe alloys
(Bozorth 1951), although somewhat smaller (see fig. 71). In fact, zero magnetostriction
was found for as-deposited Fe-AI-O and (Coo.94Feo.06)-AI-O films, and, after annealing
174 N.H. DUC and P.E. BROMMER

100

80 Co-Fe
(",,(from Bozorth)
..r : ....... _.... . .~ '"
I ".. -,

.=..
'-'
60
.
I
I
I
",

" ,,
'
-e 40
! 0----
20 ""'"

·20
0.0 0.2 0.4 0.6 0.8 1.0

Fig. 71. Composition dependence of the magnetostriction, A. of (Col_xFex )-AI-O soft magnetic films. After
Ohnuma et al. (1999).

at 300°C. also for (CoO.92Feo.og)-AI-O. The latter composition exhibits optimum proper-
ties for use as core material of an inductor (see section 12). because it combines the zero
magnetostriction with high resistivity (as the other films. about 200 /LQcm) and acceptable
saturation magnetisation (induction about 1.2 T. somewhat lower than the other composi-
tions) with a large anisotropy field (which vanishes for Fe-AI-O).
The random anisotropy model describes the magnetostriction in amorphous alloys
(Szymczak and Zuberek 1982. 1983). and explains also the suppression of the local
magnetocrystalline anisotropy in nanocrystalline alloys (Herzer 1997). This model has also
been adapted to the case of granular systems (Szymczak et al. 1999). Finally. the interaction
between small metallic spheres is of the RKKY type. Its anisotropic part is thought to be
responsible for the oscillations of the magnetic anisotropy in nanoscale magnetic films
(Back et al. 1997). According to Szymczak et al. (1999). this anisotropic part should also
contribute to the magnetostriction in heterogeneous magnetic systems.

9. Huge magnetostriction in perovskites

In the preceding sections. magnetostriction has been associated mostly with the properties
of rare-earth ions and 3d itinerant-electrons. In this section. we discuss the huge
magnetostriction occurring in another kind of material. i.e. in perovskites and some related
manganese oxides exhibiting colossal magnetoresistance. Although much experimental
work has been performed on bulk samples. thin films of these ceramics can be produced by
(pulsed) laser ablation deposition. Consequently. these materials are regarded as possible
alternatives for the more conventional epitaxially grown thin films.
Among the perovskite oxides. only manganites RI-xAxMn03 and cobaltates
Rl-xAxC003 (R: trivalent rare-earth; A: divalent alkaline ion) show colossal magne-
toresistance. The most attractive feature of their magnetic behaviour is the co-existence
of metallic conductivity and ferromagnetism. Hole-doping (substituting A 2+ for R3+ )
in the parent compounds LaMn03 and LaCo03 creates mixed valence in Mn and Co
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 175


-- t·s
-
b
)oC
'0
U
0

.....
Ol lOll
U


....,
S
<I

100 ISO 200 250 300 350 400 450 500


TEMPERATURE (K)

Fig. 72. Linear thermal expansion of Lao.6YO.07Cao.33Mn03' Inset: anomalous thermal expansion contribution
(61/ l)p. After Ibarra et al. (1995).

ions (Mn3+, Co 3+ and Mn 4 + , Co4 + ) and produces a ferromagnetic metallic state for
x ~ 0.25. Notice also that Mathieu et al. (2000) produced doping by adding monova-
lent K in Lao.96-yNdyKo.04Mn03. Doped manganites and cobaltates show not only colos-
sal magnetoresistance but also various magnetic-field induced phenomena such as mag-
netic structural, ferromagnetic metal (FMM)-paramagnetic insulator (PMI) phase transi-
tions, collapse of charge-ordered states under fields and many others. Extensive studies
have revealed that the observed features can be attributed to double-exchange (DE) in-
teractions (Zener 1951; De Gennes 1960), in competition with other mechanisms such
as antiferromagnetic (AFM) superexchange, Jahn-Teller (JT) effect, charge-orbital order-
ing, etc. A detailed (spin-orbit coupling) model was presented by Feiner and Oles (2000).
The magnetic properties of these doped perovskite manganites are reviewed by Ramirez
(1997), Ibarra and de Teresa (1998), Coey et al. (1999) and Szymczak (2000). Alejandro
et al. (2000) presented ESR-studies of the relation between the magnetic properties and
IT-distortions. Calculations performed by Nakano et al. (2000) revealed that IT distortions
work co-operatively to enhance the optical-response incoherence caused by orbital (eg )
fluctuations.
First observations of large magnetostrictive effects in manganites have been re-
ported in 1995 by Ibarra et al. (1995). The measurement was performed on doped
LaO.60Y0.lJ7Cao.33Mn03 perovskite. Colossal magnetoresistance has been observed in
this polycrystalline sample, where t::.RjR ~ 10000% just below T c (i.e. at 140 K, at
B = 6 T; Jin et al. 1995). The thermal-expansion (fig. 72) shows an extra contribu-
tion (with respect to the phonon contribution) for temperatures above Tc (= 160 K,
the temperature of the FMM-PMI transition) up to a temperature T» (~ 320 K). In this
temperature range, T C < T < T p, a large negative volume magnetostriction (of order
of 1.0 x 10- 3) was observed (see fig. 73), ascribed to the change from a charge lo-
calised PMI high-volume state to a FMM low-volume one, upon application of a mag-
netic field (or when the long-range magnetic order sets in). Similar effects were observed
176 N.H. DUC and P.E. BROMMER

a 3
J. lot
:i ..... 'y•. I7C... H MDOl
§ 0 ,...,........... ~ IS K

348.5 K
e -3

' ...<,-,"
---...;;;......
130K

~-6
Cl ~'-.
......--..~
"'
~-
~ ·9
200K

~'12 ................. _-- ....- ....---.,.


~

CS 170K
> -IS
0 5 10 IS
APPUED MAGNETIC FIElD H, (Teala)

Fig. 73. Volume magnetostriction isotherms of Lao.6YO.07Cao.33Mn03 at selected temperatures. After Ibarra et

.
aI. (1995) .

,--..,.--,.--,..-..,.--,.--.,
~
,......, ~
l"l
~
-
b :it
!
s .
.
.
. '.
<1 0 SO 100
T(K)
'

.-,.. , ....'
.'
----_ .. -. '
o 100 200 300 400 500 600
T(K)
Fig. 74. Linear !hennal expansion of Gdo.5Sro.sMn03 in zero field and under applied fields of 6 T and 12 T.
The dashed curve represents a fit of the high-temperature linear thermal expansion using a Griineisen law and a
Debye temperature eD = 500 K. After Garcia-Landa et aI. (1998a. 1998b).

in Lao.62Tbo.o5Ca0.35Mn03 (De Teresa et al. 1996). Wang et al. (1998) ascribe the shift
of the transition temperature upon applying a magnetic field, observed in La2/3Call3Mn03
and (Ndo.6ThO.4hj3SrII3Mn03, to magnetostrictive effects. Garcfa-Landa et al. (1998a,
1998b) studied the oxygen-isotope effect on the field-induced I-M transition in
Pr2l3Ca1l3Mn03,and showed that the heavier isotope favours the insulating state. Hayashi
et al. (200 1) observed the metamagnetic transition to the FMM state in pulsed fields up
to 45 T in Ndo.45Sm0.55Mn03. Two different mechanism responsible for magnetostric-
tion can be clarified: (i) in the PMI phase at T > Tc, the magnetostriction is isotropic
0"11 = A.d and gives rise to a volume distortion. This behaviour is related to the quenching
of the charge localisation under the application of a magnetic field. (ii) In the FMM phase,
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 177

20
0
-20

---
.s-
-40
'f
<:
-60
' -'
-80
-100
-120
-140
107
1()6

lOS
---eu 10"
c
'-' UP
Q.
(b)
IOZ
----T=80K
10 1 --T=60K
-T=10K
10°
0 2 4 6 8 10 12
J.1oH (T)
160
10K
140

.!!---
(C)
120
60K
E 100
s SO SOX
~ 60 100
40
200K
20

2 4 6 8 10 12 14
J.l.H (T)

Fig. 7S. MagnetosJriction(a), magnetoresistance (b) and magnetisation (c) isotherms for G<1o.5 SrO.5 Mn03. After
Garda-Landa et aI. (1998a. 1998b).

at T < Tc. the magnetostriction is anisotropic (AI = All - 1..1. '" 10-4) and typical in value
and shape to that of a ferromagnetic 3d metal; its magnetovolume effect is negligible.
The existence of Tv is also observed in Gdo.sSro.sMn03 (Garda-Landa et aI. 1998a.
1998b). see fig. 74. For this manganite, however. no FMM-PMI transition is observed.
Instead. a gradual enhancement of the charge localisation (CL) with lattice distortion
has been proposed as the mechanism responsible for the increase in the resistivity with
decreasing temperature. Irreversibility's and sharp anomalies in the magnetostriction,
magnetisation and magnetoresistance isotherms take place at low temperatures (T <
90 K) (see fig. 75(a---<:». which have been attributed to the crossover from the CL to an
inhomogeneous charge-ordered (CO) state. For temperatures lower than 42 K. the CO state
was found to coexist with a cluster glass (CG) state.
178 N.H. DUC and P.E. BROMMER

-
cg
b
~~\1'1"-~-1Iliiii~~:::::::~=240 220 K
K
200 K

-.,. -100
)(

.... ·150 180 K

~ ·200
~--160K

.250~d-~b).~:;;:;;;;~iiiiiiiiiiiiiiiii
o 2 3 4 5
Magnetic field (T)

Fig. 76. Magnetostriction isotherms for (Nd 1_ y Srn , )O.5Sr0.5 Mn03 (Y = 0.938). After Kuwahara et aI. (1996) .

•18 r-- - - - - --,


-18

-14
.:1)

s>
a
-e-'
-20 «
---
-I

·10
~

Smcontent
Fig. 77. Composition dependence of magnetostriction and magnetoresistance for (Lal-xSmx)zI3SrI/3Mn03'
After Cao et al. (1999).

Magnetostriction measurements have also been interpreted as proof for the existence of
magnetic polarons above T c in La213Srl/3Mn03 perovskites (De Teresa et al. 1997). See
below. however. for an alternate interpretation. as discussed by Demin et al. (1997) and
others.
In a (Ndl-ySmy)o.sSro.sMn03 (y = 0.938) single crystal. magnetostriction shows a
clear hysteresis and abrupt change in applied magnetic fields (fig. 76). These phenomena
were described as a field-induced first-order insulator to metal phase transition. which
accompanies a metamagnetic (from AFI to FMM) transition (Kuwahara et al. 1996).
A large linear (Joule) magnetostriction with a value up to -180 X 10-6 was observed for
bulk polycrystalline (Lal-xSmxhj3SrII3Mn03 (x = 0.33) at T = 77 K (Cao et al. 1999).
Joule magnetostriction and magnetoresistance show a similar concentration dependence
(see fig. 77).
A somewhat different picture is presented by Demin et al. (1997. 1998) and Koroleva and
Demin (1999). on the basis of their work on Lal-xSrxMn03 and also by Abramovichet al.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 179

30
Lao.oSr .1MnO
'J'II1 .1[
/"
20

j
V
./
III[

II.
o~
<: r-,
·10
-,
/
~ lOO~
I iIO I[

·20
o 5 20 25

Fig. 78. Magnetostriction isotherms for Lao. 9SrO.1 Mn03' After Popov et aI. (1998).

(2000), on the basis oftheir work on Smo.ssSro.4sMn03' According to these authors these
systems should be treated as 'magnetic two-phase states', i.e., in this case, ferromagnetic
'clusters' embedded in a antiferromagnetic semiconductor. The structural phase transition
from the orthorhombic semiconducting phase to the metallic rhombohedral phase is then
caused by (or accompanied by) a transition to a conducting mainly ferromagnetic two-
phase state (either AF clusters in a conducting FM matrix, or percolation and tunnelling
between the increased FM clusters, still in the AF matrix). Maignan et aI. (2001) ascribe the
possibility to form a similar two-phase structure in CaMn l-xRux03 perovskites to specific
relationships between Ru and Mn ions. Respaud et aI. (200 I) determined B vs T magnetic
phase diagrams of electron-doped Sml-xCaxMn03 in fields up to 50 T. Introducing Sm
leads to increasing FM correlations. For x < 0.8 abrupt metamagnetic transitions were
observed, whereas around 0.85 phase separation was found.
Popov et aI. (1998) studied magnetostriction and field-induced structural phase transi-
tions in a series of Lal-xSrxMn03 (x = 0,0.05,0.1,0.175) perovskite single crystals
(twins), in pulsed magnetic fields up to 23 T. Anomalies in the All (8) curves were ob-
served at 20 T for LaMn03. These were associated with a spin-reorientation AyFI: .... AI:Fy
when the magnetic field B is applied along the b-axis (AyFI: indicates an antiferromagnetic
structure with moments along the b-direction in combination with a small ferromagnetic
canted moment in the c-direction; the proposed transition can - almost - be regarded
as an ordinary spin-flop transition). For Lao.9Sro.IMn03 (Tc ~ 170 K), the longitudinal
magnetostriction has no anomalies at low temperatures (T < 130 K), while near Tc: the
magnetostriction increases rapidly up to one order of magnitude and even changes in sign
(see fig. 78). This complicated behaviour was suggested to be associated with a structural
phase transition related to a polaron ordering process. A strong hysteresis of the magne-
tostriction was observed for x = 0.175 (fig. 79). Because of these phenomena accompa-
nying the structural transition, the initial state of the crystal is not re-established after the
magnetic-field pulse.
180 N.H. Due and P.E. BROMMER

1~,B25S O,175M% 0 3
20
I'-- ~
-- -
t>
...1 41(
./
->
10 - / 111 L
/

5
/ / /
/ V
1511 I(
(
o
o 5 10 15

Fig. 79. Magnetostriction isotherms for Lao.825SrO.175 Mn03. After Popov et aI. (1998).

1.5

"
5r=0.11
...
-
(a)
= 1.0
H=O
S
0.5

0.0
a 50 100 150 200 250 300
Temperature (K)

1.5 (b)
Sr = 0.11 {

~ 1.0 f
s 0.5
~o-"
0
I
: p'o-o-"-o-O~
0.0 ~ ,-
0.0 0.1 0.2 0.3 0.4 0.5
M 2 (Am 2/kg)2

Fig. 80. Plot of w(T) (a) and w(M2) (b) for Lao.89SrO.11 Mn03. After Dabrowski et aI. (2000).

Kadomtseva et al. (2000) measured thermal expansion and longitudinal and transverse
magnetostriction in pulsed magnetic fields up to 25 Tin Lal-xSrxMn03 single crystals
(x = 0.1, 0.125 and 0.15). The results were ascribed to a suppression of the O' phase and
field-induced transitions to a new orbital-ordered ferromagnetic state.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 181

(b) La2-zxSr'+2xM~ x=

0.36

0.4

0.45

o 100 200 o 1 200


Temperature (K) Temperature (K)

Fig. 81. Temperature dependence of the striction along (a) the ab-plane and (b) the c-axis in the absence of
magnetic fields for La2_2xSf(+2xMn207 (0.3 ~ x ~ 0.45) crystals. Arrows indicate the magnetic transition
temperature. After Kimura et al. (1998).

Mukhin et al. (2000) found that increasing Sr doping in Lal-xSrxMn03 does


suppress the (quasi- )ferromagnetic resonance frequency, whereas the (quasi- )AF resonance
frequency is only slightly decreased (20%). This behaviour corresponds to a canted
magnetic structure and its evolution with increasing x.
Dabrowski et at. (2000) measured the magnetostriction (and thermal expansion in fields),
the magnetisation and the resistivity of Lal- xSrxMn03 perovskite samples (x = O.ll,
0.13 and 0.165), which were also used in neutron diffraction studies (Dabrowski et al.
1999). Large striction effects were observed at the ferromagnetic transition. This indicates
coupling of the local spin moments and charge carriers to the crystal lattice. Magnetic
fields suppress coherent Jahn-Teller (IT) orbital ordering near x = 0.13 and remove
incoherent IT orbital ordering near x = 0.165. The observed strong spontaneous volume
magnetostriction was associated with the square of the magnetisation w(T) '" M 2(T) (see
fig. 80). The authors claim to have derived, in this way, a correct description of the intricate
phase transitions and spin reorientation in the framework of the localised spin model.
Eto et al. (200 I) studied the effect of hydrostatic pressure on the magnetostriction and
the magnetisation of a polycrystalline Euo.58Sro.42Mn03 sample. At 6 K and ambient
pressure, the estimated magnetostriction of about 400 ppm falls down in increasing fields
182 N.H. Due and P.E. BROMMER

s: x-o.3
---
<=
' -'
<=
...
'-'
X-O.4

.J
;;:...
:c
~
J
<I

"-s
<=
' -'

--J.
;;:...
:c
J<I

----
<=
i
Q.
;;:...
:I:
~

o 246 o 246 8
~oH (T) JIoH (T)

Fig. 82. Field dependence of the in-plane magnetostriction (6Lab/Lab(0)1 (upper panel), the inter-plane (or
c-axis) magnetostriction (6LcI Lc<O)] (middle panel), and the in-plane magnetoresistance (Pab/Pab(O)] (lower
panel), at several temperatures for La2-2., Sri +2>:Mn207, (aHc): x = 0.3 and (dHf): x = 0.4. After Kimura et
al. (1998).

from I T to 2.4 T, i.e. in the 'antiferromagnetic insulator' (AFI) region. The field of 2.4 T
corresponds to the AFI-FMM (insulator-metal) transition as signalled by the observed
CMR. With increasing pressure, the magnetostriction becomes smaller and the metal-
insulator transition occurs at lower fields. In decreasing fields (from 5 T down to zero
field), no transitions were observed: the (partial) FMM state apparently remains. In fact,
at intermediate pressures, the magnetostriction measured in increasing fields becomes
negative with respect to the values observed in decreasing fields. At pressures above
1.15 GPa, the magnetostriction almost disappears, indicating that then EUO.58 Sr0.42Mn03
behaves as a ferromagnetic metal (FMM state).
The 3D perovskites can be regarded as the limit for n ~ 00 of the (doped) layered
manganese oxides with formula (RI-yAv)n+IMnn03n+1 [R: rare earth, A: divalent
cation]. Colossal magnetoresistance has been found in some highly anisotropic, almost
two-dimensional, bilayered (n = 2) members, based on (La, Sr), (La, Ca), (Nd, Sr) or (Sm,
Sr), see e.g. references in Szymczak (2000) or Garcia-Landa et al. (2001). These results
triggered further research on magnetostrictive effects. Giant magnetostriction (GMS, up
to A ....., 0.05%) has been reported for bilayered (n = 2) La2-zxSrl+zxMnz07 crystals,
which exhibit huge magnetostrictive effects (Kimura et aI. 1998). This system shows a
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 183

(a)

+
Mn06 octahedron

f=l=3z2-,2~
x2-f-<Cl ~3Z2_~
eg x2-f~ EF

~
EF _
=
=
e
,band
,= --
xy, yz , zx e"band
XV, yz, zx
~11 2g

(b) doping level


.

..
• 'maonetic
moments
x-0.4 x-O.4S

So [AJ 3.860(2) 3.876(2) 3.878(1)


eo [AI 20.380(2) 20.140(2) 20.053(2)
Mn-0(1) [AI 1.98 1.95 1.94
Mn-O(2) [AI 2.06 2.00 1.97
Mn-O(3) [AI 1.93 1.94 1.94

Fig. 83. (a) Schematic electronic structure of a Mn3+ ion in a Mn06 octahedron with IT distortion. The in-plane
eg band in the layered manganite shows a different band dispersion and bandwidth depending on the respective
orbital states. (b) Doping-level dependence of lattice distortion at room temperature in La2-2x Srl+2x Mn207.
Thick arrows on the right hand of respective crystal structures indicate the spin structures within a bilayer unit at
low temperatures. After Kimura et aI. (1998).

large spontaneous distortion at T c (fig. 81) as well as a highly anisotropic field-induced


striction with respect to the crystal axes (fig. 82). The magnetoelastic behaviour does vary
systematically with the doping level (x) and may be attributed to the field-induced change
in the orbital-state occupancy in the two-dimensional band. The most striking result is
the difference in sign between the magnetostriction of the x = 0.3 sample and that of the
x = 0.4 sample (see fig. 82). By applying magnetic fields in the ab-plane, the ab-plane
=
expands but the c-axis shrinks for x 0.3. Just the opposite is observed for x = 0.4. To
understand the origin of these magnetotransport and magnetoelastic properties, the orbital
degrees of freedom of the eg-like conduction electrons of Mn 2+ were taken into account.
The electronic structure of Mn3+ ions in Mn06 octahedron with IT distortion is sketched in
fig. 83a. It is clearly seen from this figure that an expansion of the c-axis stabilises the 3z2 -
r 2 orbital state. The (doped) e g electrons, however, prefer to occupy the x 2 _y 2 orbital state
when the c-pararneter is diminished. Doping-level dependence of lattice distortion at room
184 N.H. DUC and P.E. BROMMER

800x10" r----------------,

-400 '--''--'-...L---'---'--L-~_'_..L_.L_L_l'__'___L~~

o 2 4 6 8 10 12 14 16

lloH (T)

Fig. 84. Paral1el and perpendicular magnetostriction isotherms of Lao.5SrO.5Co03 at selected temperatures.
After Ibarra et al. (\998).

temperature in La2-2xSrl+2xMn207 crystals is presented in fig. 83b. The lattice parameter


of the a (b) axis slightly increases with increasing .r, whereas that of the c axis decreases
more rapidly. Since the orbital character is strongly correlated with lattice deformations,
the observed structural distortions, induced by changing temperature, magnetic field and
doping level, reflect sensitively changes in the orbital state. In the case of the low-doped
samples (x ~ 0.36), the cooling- or field-induced distortion appears to stabilise the x 2 _
y2 orbital state. By contrast, the lattice distortion toward the FM state (or by applying
magnetic field) has been understood in terms of the stabilisation of the 3z 2_r2 orbital state.
In a study of a series of perovskites R(Co,Mn)03 and R(Mn,Ni)03 (R = Eu, Gd, Tb, Dy,
Y), Troyanchuk et al. (1997) found large negative magnetostrictive effects of the order of
10- 4 in Eu(Coo.sMnO.S)03 at 4.2 K. The magnetostriction was not saturated in the highest
available field of 12 T. For Th(Coo.sMno.s)03, the magnetostrictive effects were somewhat
smaller. Moreover, at 5 T a change of sign was observed, ascribed to the metamagnetic
transition.
The magnetostrictive properties of Lal-xSrxCo03 cobaltates were studied by Ibarra et
al. (1998). It was found that the magnetostriction in the compounds with x = 0.0 and
x = 0.08 is negligible. Large magnetostrictive effects were observed for x = 0.3 and
x = 0.5. Their behaviour, however, is different from that of manganites. The parallel
and perpendicular magnetostriction isotherms are presented in fig. 84 for Lao.sSrO.SCo03.
Since All and AJ.. are of opposite sign and display large values, a huge anisotropic
magnetostriction is obtained: At = All - Al., is as large as 1 x 10-3 (fig. 85). The maximum
volume magnetostriction (w = All + Al.) is reached near Tc ::::: (250 K), with a value of
about -140 x 10-6 at B = 14.2 T, which is one order of magnitude lower than found in
the manganites. Such a huge anisotropic magnetostriction cannot be explained on the basis
of the usual spin-orbit coupling mechanism. Moreover, this finding suggests a different
nature of the magnetovolume effect in cobaltates compared to manganites. Indeed, it was
proposed that the magnetostriction of Lal-xSrxCo03 has its origin in orbital instability of
C0 3+ ions under the influence of a magnetic field, giving rise to a transition from a non-
degenerate low-spin (t2g 6 , S = 0) state to an orbital degenerated intermediate-spin (lzg se g,
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 185

2500
<a) lSK
2000 La•.,Sr•. l CoO]

1500
....
«-
Cl 1000

SOO

0
1200
<b) lSK
1000 La•. 5 Sr•. 5 CoOl

800 lOOK

....
«- 600
Cl
400

200

0
0 S 10 1S
J10H (T)

Fig. 85. Anisotropic magnetostriction (At) isotherms for Lao.7SrO.3Co03 (a) and Lao.5SrO.5Co03 (b) at selected
temperatures. After Ibarra et aI. <1998).

S = 1) state. In this configuration. the two-fold degenerate eg level of intermediate Co3+


splits into two singlets (L = 0). and the triplet tZg into a singlet and a doublet (L = 1). Due
to the degeneracy of the doublet level with nonzero angular momentum, a strong intra-
atomic spin-orbit coupling is created which couples to the lattice strain to give rise to a
large anisotropic magnetostriction.

10. Potential applications of magnetostrictive materials

Technical aspects and potential applications of magnetostrictive materials have been


presented by Du Tremolet de Lacheisserie (1993). Different kinds of magnetostrictive
devices based on Terfenol-D were reviewed by Houqing et al. (1997), Claeyssen et al.
(1997) and Ludwig and Quandt (2000). Marin and Hernando (2000) present a number of
applications using. on the one hand. highly magnetostrictive amorphous materials (e.g.
FeSiBCuNb wires), and, on the other hand, zero-magnetostrictive amorphous, highly
inductive materials, such as (Feo.06Coo.94hz.sSil2.sBls. According to Osaka (2000), zero
magnetostriction is also a prerequisite for electrodeposited soft magnetic films to be used
for magnetic recording heads 'in the next century'.
It is also interesting to point to the so-called 'gyromagnetic' effect, observed in Fe-Si-B
amorphous wires with a particular magnetic-domain structure (Chiriac et al. 2000b).
A large growing interest is manifest today for physical micro-systems of reduced
dimensions. typically between 10 and 103 JLm. Such systems may satisfy the need to
186 N.H. Due and P.E. BROMMER

perform functions which are not fulfilled by existing electronic circuits, i.e. sensing
functions (micro-sensors) and functions of interacting with the environment (micro-
actuators). In the cases where a simple motion is to be obtained, magnetostrictive
amorphous R-T thin films, which combine several specific properties, are very promising.
The working principle of the magnetostrictive thin-film devices is based on the bending
transducer which consists at least of a film-substrate compound, where the substrate is
non-magnetostrictive. Upon magnetisation the magnetostriction in the film causes the
compound to bend, similar to the bending of a bimetallic strip. The advantages of
magnetostrictive devices with respect to their piezo-electric competitors are the larger
deformations, higher forces and energy densities, lower sound velocity and Young's
modulus, and low operating voltage. Moreover, direct electric contacts can be avoided. The
disadvantage is the coil, which is difficult to make small because of the field requirements.
Developing films with giant magnetostriction at low fields is necessary to solve these
problems. In addition, as magnetostrictive devices using thin films are very small, the
price of material is not a problem. So, such MEMS could find large scale applications, for
instance in electronics, optics, medicine, automobile industry, geophysical explorations,
and ocean environmental protection as well. At present, some devices are already used for
specific applications. We present here some typical examples, such as a micromechanical-
switch, a micro-motor and a micropump, demonstrating the above-mentioned advantages
of magnetostriction, especially for the fact that moving parts are wireless.
The simplest motion which can be thought of is realised by a bimorph bending (as
schematised in fig. 4) as a mechanical switch. In this case, although Sm-T films exhibit less
pronounced magnetostriction compared to Tb-T ones, the combination of a negative- with a
positive-magnetostriction film allows the fabrication of magnetostrictive bimorphs which
enhance the total deflection, and reduce the initial curvature of cantilevers (fig. 86a, see
Honda et al. 1994). A 3-mm long cantilever actuator is found to exhibit a large deflection
of above 100 JLm in a magnetic field as low as 0.03 T. With this cantilever, a deflection
of more than 500 JLm at resonant frequency in an alternating field of 0.03 T has been
reached.
One of the main drawbacks of the magnetostrictive actuators is their thermal drift. For
bulk Terfenol-D actuators, the thermal expansion (10- 5 K- 1) develops strains comparable
to the magnetostrictive ones for t1T = 150 K. For a simple rectangular bimorph cantilever
as illustrated in fig. 86a, the strains due to thermal expansion and to the magnetoelastic
coupling are comparable for only t1T = 75 K (Du Tremolet de Lacheisserie et al. 1998).
A torsion based, thermal-drift free microactuator was invented by Betz (1996, 1997). It
is basically a unimorph structure composed of one magnetostrictive film deposited on
a passive substrate. The special feature is a square shape maintained by hinges at three
comers (figs 87a, b). The useful displacement due to magnetostriction is obtained at the
fourth free comer, without thermal displacement. The different deformed shapes are due to
the anisotropy of magnetostriction strains and the isotropy of thermal strains.
Various types of the so-called magnetostrictive "inchworm" type of motor have been
proposed and built. A two-leg travelling machine using a magnetostrictive bimorph
actuator with 7.5 JLm-thick polyimide is shown in fig. 86b (Honda et al. 1994; Arai and
Honda 1996). In an alternating magnetic field, it can travel in one direction. The maximum
speed of approximately 5 mmls was obtained around the mechanical resonant frequency
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 187

(a)

Tb-Fe(1 ,urn) Polyirnide


(7.5,urn)

(b)

Sm-Fe (1 ,urn)

Fig. 86. (a) A side view of the cantilever actuator fabricated by combining TbFe and SmFe films. After Honda et
al, (1994). (b) Two-leg linear magnetostrictive micromotor. After Honda et al, (1994).

(a)

(b)

Fig. 87. Drift free microactuator: (a) thermal deformation and (b) magnetostriction deformation. After Betz
(1997).

of 200 Hz. Similarly, a many-leg linear-motor was also fabricated using a micromachined
Si(llO) substrate (10 /Lm) and TbFe films, see fig. 88 (Halstrup et al. 1996; Claeyssen
et al. 1997). The 4 /Lm-thick TbFe films were deposited on both sides of the substrate.
188 N.H. DUC and P.E. BROMMER

Silicon plate

?"~f,s1 {

• Leg Magnetostrictive thin film


}J2 }J2

Fig. 88. Many-leg linear magnetostrictive micromotor. After Claeyssen et al. (1997).

Applying a magnetic excitation field of 10 mT at a frequency of approximately 775 Hz and


a magnetic bias field of 10 m'T, this motor can be operated at a speed of 3 mm/s. A rotating
magnetostrictive ultrasonic micromotor, however. consists of a free Ti rotor (20 mm
diameter. thickness of 100 J-Lm. three 300 J-Lm-thick teeth) on which a magnetostrictive
film of 4.6 J-Lm-thick in three sectors is deposited. An ac magnetic field of 20 mT and a
bias magnetic field of 40 mT are applied transversally to the film. Operating at a frequency
of 2.86 klfz, the propulsion mechanism of vibrating teeth on a friction layer leads to a
rotational speed of 0.5 rls (Claeyssen et al. 1997).
Gibbs et al. (1997) describe a MEMS pressure sensor based on the magnetoelastic
properties of an amorphous Fe-Si-B-C coating (derived from METGLASS2605SC) on
GaAs substrate.
The world-wide first magnetostrictive thin film micropump prototype was made
of four laser micro-machined Si(lOO) wafers in combination with a bimorphous Tb-
DyFe(l5 J-Lm)/Si(l (0)(50 J-Lm)/SmFe(l5 J-Lm) membrane and cantilever-type passive
valves (fig. 89) (Quandt and Seemann 1996; Quandt and Ludwig 1997; Quandt et al.
1997b). The radial magnetic field circuit is directly placed on the membrane to permit
a good penetration by the magnetic field. The micropump is operated using an oscillating
rectangle pulse for the membrane actuation. At the frequency of 2 Hz a maximum yield of
approximately 290 J-LVmin of methanol or an outlet pressure of 4.9 mbar can be reached.
Later. Quandt and Ludwig (2000) proposed a more simple construction based on Si mem-
branes coated with giant magnetostrictive TbFeIFeCo multilayers.
Optical microactuators like a micromirror for scanning applications have been proposed
by Orsier et al. (1996). A two-dimensional bimorph (Si + magnetostrictive film) is driven
remotely by two differently oriented magnetic fields working at different frequencies
in order to obtain the bending and torsional vibrations due to the magnetostrictive
strain.
Berger et al. (2000) proposed to use magnetostrictive thin films in scanning probe
microscopy. This idea allows overcoming misinterpretation of MFM images due to the
tip magnetisation by using non-magnetic tips.
(Co.Fe)-AI-O granular films were used in preparing the core (dimension of 2 mm x
2 J-Lm x 300 J-Lm) of a high-frequency spiral shaped inductor. The best properties were
obtained with (Coo.92Feo.o8)-AI-OISi02 multilayers, which have magnetostriction almost
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 189

TbOyFe film (20 11m)


51 (RI1 0 nvn x 50 11m)
SmFe 111m (20 11m)

Si (100) (180. 240 11m).


by laser micropallerning

Fig. 89. Schematic cross-section of the magnetosttictive membrane type micropump. After Quandt (1997).

zero after annealing as the core material. A maximum quality factor Qrnax = 25 at a
frequency of 300 MHz was observed (Ohnuma et aL 1999). A coil using a granular film
with >.. = 10 x 10-6 had Qrnax = 8 at a frequency of 120 MHz.
Combination of magnetostrictive and piezoelectric materials have also been suggested
(Arai et at. 1995). Using a piezoelectric substrate (or thin film) on which a magnetostrictive
film is deposited, one can operate in two different modes: either the magnetostrictive
film can be used as an actuator leading to a sensor signal in the piezoelectric substrate,
or the function of both layers could be exchanged using the inverse effects in both
materials.
A stress-operated memory device consisting of an ellipsoidal magnetic particle array
and an electrostrictive grid is proposed by Novosad et al, (2000). In this device, the
magnetic state of the particle can be controlled by the magnetostriction effect. The design
is sketched in fig. 90a. In the writing process, the driving voltages are simultaneously
applied to two pairs of the selected contacts (e.g. ±UI and ±U2 in the illustration)
resulting in an electric field with a vector sum of two electric fields. Varying the intensity
and the polarity of the applied voltage (fig. 90b) can effectively rotate the direction and
the amplitude of the electric field. Consequently, the rotatable stress can be generated
at the selected intersection in the piezoelectric grid, which initiates the magnetisation
reversal of the magnetic particle (e.g. the magnetisation of particle B is switched, whereas
the final magnetic states of the particles A and C are not affected). For the design of
this kind of stress-operated memory device, ferroelectic materials such as lead zirconat
titanate (PZT), barium strontium titanate (BST) and lead lanthanum zirconate titanate
(PLZT) would be suitable. The magnetic particles should be magnetically soft with large
magnetostriction. In a (microscale) prototype Ni particles were applied, but Pd-Co and
the spring type multilayers (MSMM, discussed in section 5.2) were also mentioned as
promising materials.
190 N.H. Dueand P.E. BROMMER

(a)

(b)

Fig. 90. (a) Schematic design of a stress-operated memory device. (b) The magnetic states of particles A, B and
e (see (a» as a function of a driving electric field. The magnetisation of the particle B is switched, whereas the
final magnetic states of the particles A and e are not affected. After Novosad et al. (2000).

11. Summary and concluding remarks

Recent results for nanoscale heterogeneous magnetic systems have been reviewed. For
applications, both zero-magnetostriction (soft-magnetic) materials and materials exhibiting
giant magnetostrictive effects (for actuators) are of interest. Reliable magnetostrictive
devices (MEMS) have been designed on the basis of amorphous Terfenol and Terfenol-D
(TbDyFe2) alloys, although the magnetostriction of these amorphous alloys has been found
to be one order of magnitude lower than that of their traditional, well-known crystalline
counterparts. Magnetic investigations have shown that the GMS-properties of amorphous
R-Fe alloys suffer from the sperimagnetic character of the Fe and R subsystems.
Fortunately, an alternative solution has been found by preparing nanocrystalline R-Fe
alloys in which the magnetic anisotropy is reduced, while the exchange energy (ordering
temperatures) and the GMS remain satisfactorily high. By optimising the annealing
temperature and time, i.e. the nanocrystallisation, the magnetostriction value can be
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 191

doubled. In particular, a giant parallel magnetostriction (An) of 800 x 10-6 has been
achieved in a (Tbo.7DYO.3)O.3FeO.7 film with grain sizes around 10 nm. This size, however,
is rather close to the exchange length in R-T alloys, so the coercive field is increased
(less suitable for MEMS, where fields of a few mT are preferred). In addition, annealing
promotes the segregation of other Fe-rich phases, which - on the one hand - have
low magnetostriction, but - on the other hand - may improve the magnetostrictive
softness. Good results are obtained for nanocrystaIIine multi layers in which the crystallite
growth is delimited by the layer thickness. In amorphous thin films, however, it is strongly
preferable to replace iron by cobalt. Near the RC02 composition, the amorphous alloys
present higher ordering temperatures and higher magnetostriction than the equivalent iron-
based alloys. In fact, the magnetostriction has been optimised in a series of thin films
of the type a-(Tb,Dy)(Fe,Co)n. These materials showed a high, record, magnetoelasticity
of p.2 = -63.5 MPa and AY. 2 = 1020 x 10-6 , which is almost fully developed at low
fields. GMS obtained in this series of alloys has been explained in terms of an increase in
the ferromagnetic coupling strength within the (Fe.Co) subsystem, and the effect of field
annealing in inducing a well-defined uniaxial anisotropy. StilI better performances were
obtained on spring type Rtf MSMM, where the saturation field of the magnetostrictive,
amorphous Tb-FeCo is lowered by means of exchange coupling with the soft-magnetic
Fe-Co layers. In addition, the increase of the coupling exchange field between the layers
does close the cone angle in the sperimagnetic Tb-FeCo structure, thus increasing its
magnetostriction. In this case, layers must be sufficiently thin, in order that a noticeable
volume of the R-T layer may be submitted to the large molecular field. Deposition under
an applied field and field annealing are rather effective in creating a well-defined uniaxial
anisotropy and, then, in increasing the magnetostriction in the direction of interest. In view
of applications, some interesting values of the magnetostriction and of the magnetostrictive
susceptibility are summarised in table 2.
Apart from technical considerations for applications, we stress that studies on magnetoe-
lastic effects in thin films also are of fundamental interest. At the surface of any alloy, the
symmetry is broken, for instance because neighbouring atoms are missing. Thus, surface
(and interface) contributions to the magnetoelastic coupling must be taken into account.
This contribution is usually negligible in bulk materials, but this is no longer true for thin
films and multilayers where surface and interface effects on the magnetic and magnetostric-
tive properties have been experimentally evidenced. For a better understanding of the very
complex properties of these systems, however, very complex further experimental as well
as theoretical studies are needed.
Significant improvements in the field of giant magnetostriction have stimulated the
design and manufacturing of microactuators and motors (sometimes prototypes only),
taking advantage of wireless magnetic excitation, in first instance at room temperature
(or higher temperatures). For cryogenic applications, magnetostrictive actuators require
low-temperature magnetostrictive materials. In practice, single crystals of TbO.6DY0.4Zn I
(Terzinol) were used (Teter et aI. 2000). The major advantage of this material is that a
50 mT apllied field suffices to achieve a large magnetostriction of 5 x 10-3. In this context,
perovskites and rare earth superlattices are also possible candidates.
192 N.H. DUC and P.E. BROMMER

Acknowledgements

The collaboration of N.H. Due with Dr. D. Givord, Dr. K. Mackay, Dr. J. Betz and Dr. E.
Du Tremolet de Lacheisserie at the Laboratoire Louis Neel, CNRS, Grenoble (France) has
been very useful for him, not only to familiarise him with this field of research, but also
to go further into it. N.H. Due is also particularly grateful to Prof. Dr. T.D. Hien, Prof. Dr.
N.H. Luong, Dr. N.H. Chau and co-workers, who have started together with him to build
equipment for measuring thermal expansion and magnetostriction using the capacitance
method, and to develop research on magnetovolume effects and magnetostriction of the
lanthanide-transition metal intermetallics at the Cryogenic Laboratory in Hanoi since
1981. The support of the Vietnam National University, Hanoi within the project QG.99.08
has been useful to complete this work. The work was also partially supported by the
Vietnamese programme for Fundamental Research under project 420.301.

References

Abramovich, A.I., L.I. Koroleva, A.V. Michurin, Barthelemy, A., A. Fert and F. Petroff, 1999, in:
O.Y.Gorbenko and A.R. Kaul, 2000, Physica B 293, Handbook of Magnetic Materials, vol. 12, ed.
38. K.HJ. Buschow (Elsevier Science, North-Holland,
Alejandro, G., M. Tovar, A. Butera, A. Caneiro, Amsterdam) p. I.
M.T. Causa, F. Prado and R. Sanchez, 2000, Physica Batt, AJ., R.A. Buckley and H.A. Davies, 2000, Sen-
B 284-288, 1408. sors Actuators A 81, 170.
Ali, M., and R. Watts, 1999,1. Magn. Magn. Mater. Beach, R.S., 1.A. Borchers, A. Matheny, R.W Erwing,
202,85. M.B. Salamon, B. Everitt, K. Pettit, J.l. Rhyne and
Alves, F., 1.B. Desmoulins, D. Herisson, 1.F. Rialland c.a Plynn, 1993, Phys. Rev. Lett. 70, 3502.
and F. Costa, 2000, 1. Magn. Magn. Mater. 215-218, Becker, R., and W.D. Doring, 1939, Ferromagnetismus
387. (Springer, Berlin).
Andreev, A.V., 1996, in: Handbook of Magnetic Mate- Belov, K.P., O.P. Elyutin, G.I. Kataev, D. Kim,
rials, vol. 8, ed. K.HJ. Buschow (Elsevier Science, S.A. Nikitin, G.V. Pshechenkova, L.I. Solntseva,
North-Holland. Amsterdam) p. 59. G.N. Surovaya and V.P.Taratynov, 1975, Phys. Met.
Metall. 39, 50.
Arai, K.I., and T. Honda, 1996, Robotica 14,477.
Berger, R., F. Krause, A. Dietzel, l.W. Seo, J. Fom-
Arai, K.I., T. Honda, W. Sugawara and M. Takezawa,
peyrine and 1.P. Locquet, 2000, Appl. Phys. Lett.
1995, Technical Digest of the 13th Sensor Sympo-
76,616.
sium (unpublished), 477.
Betz, 1., 1996, Proc. Actuators 96, Axon, Bremen (Ger-
Areas, 1., A. Hernando, 1.M. Garcfa-Beneytez and
many), p. 283.
M. Vazquez, 1998, J. Magn. Magn. Mater. 186,283.
Betz, J., 1997, Thesis, Universite de Grenoble.
Armstrong, WD., 2000a, 1. Appl. Phys. 87, 3027. Betz, 1., K. Mackay and D. Givord, 1999, 1. Magn.
Armstrong, W.D., 2000b, Mater. Sci. Eng. A285, 13. Magn. Mater. 207,180.
Amaudas, J.I., A. del Moral, M. Ciria, GJ. Tomka, Bochi, G., O.-S. Song and R.c. 0' Handley, 1994,
C. de la Fuente, P.AJ. de Groot, R.C.C. Ward and Phys. Rev. B SO, 2043.
M.R. Wells, 1996, J. Magn. Magn. Mater. 156,421. Borchers, J.A., M.B. Salamon, R.W. Erwin, U. Rhyne,
Awano, H., O. Taniguchi, T. Katayama, F. Inoue, R.R. Du and C.P. Flynn, 1991, Phys. Rev. B 43,
A. Itoh and K. Kawanishi, 1988,1. Appl. Phys. 64, 3123.
6107. Bozorth, R.M., 1951, Ferromagnetism (Van Nostrand,
Back, C.H., W. Weber, Ch. Wursch, A. Bishof, D. Pes- New Jersey).
cia and R. Allenspach, 1997, 1. Appl. Phys. 81, Bushnell, S.E., and C. Vittoria, 1993, IEEE Trans.
5054. Magn. 29, 3379.
Baril, L., B. Gurney, D. Wilhoit and V. Speriosu, 1999, Bushnell, S.E., WB. Nowak, SA Oliver, C. Vittoria,
1. Appl. Phys. 85, 5139. 1992, Rev. Sci. Instrum. 63, 2021.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 193

Callen. E.R .• and H.B. Callen. 1963. Phys. Rev. 129, Danh, T.M.• N.H. Duc and N.P. Thuy, 1998. J. Magn.
578. Magn. Maler. 185. 105.
Callen, E.R .• and H.B. Callen. 1965. Phys. Rev. 139. Danh, T.M.• N.H. Duc. H.N. Thanh and J. Teillet, 2000,
A455. J. Appl. Phys. 87. 7208.
Campbell. I.A., 1972. J. Phys. F 2. L47. De Gennes, 1960. Phys. Rev. 118. 141.
Cao, Q.Q .• J. Wu. K.M. Gu, S.Y. Zhang and Y.W. Du. De la Fuente. C.. A. del Moral. J.I. Arnaudas,
1999, J. Appl. Phys. 85, 4494. J. Go. R. Sarthour, R. Cowley. M.R. Wells and
Chappert, J., J.M.D. Coey, A. Lienard and J.P. Rebouil- R.C.C. Ward, 1999, J. Magn. Magn. Mater. 196,
lat. 1981, J. Phys. F 11, 2727. 140.
Chien, C.L.. 1991. J. Appl. Phys. 69, 5267. De Teresa, J.M .• J. Blasco. M.R. Ibarra, J. Garcia,
Chikazumi. S., 1964, Physics of Magnetism (Wiley. e. Marquina. P. Algarabel and A. del Moral. 1996.
New York). J. Appl. Phys. 79. 5175.
Chiriaco H.• M. Tomuta and M. Neagu, 1999. Nano-
De Teresa, J.M., M.R. Ibarra, P.A. Algarabel, C. Ritter.
struct, Mater. 12, 851.
C. Marquina, J. Blasco. J. Garcia, A. del Moral and
Chiriac, H.• M. Lozovan, M. Neagu and e. Hison,
Z. Arnold. 1997. Nature 368. 256.
2000a. J. Magn. Magn. Mater. 215. 378.
Del Moral. A., J.1. Arnaudas, M. Ciria, M.R. Wells
Chiriaco H., C.S. Marinescu and T.-A. Ovari, 2OOOb,
and R.C.C. Ward. 1996, J. Magn. Magn. Mater.
Sensors and Actuators 181. 126.
157/158.539.
Chopra, H.D .• OX Yang and P. Wilson. 2000. J. Appl.
Del Moral, A., M. Ciria, J.1. Amaudas, M.R. Wells.
Phys. 87, 5780.
Ciria, M., J.1. Arnaudas, A. del Moral, GJ. Tomka, R.C.C. Ward and e. de la Fuente. 1998. J. Phys.:
e. de la Fuente, P.A. de Groot, M.R. Wells and Condens. Matter. 10, L139.
R.C.C. Ward, 1995, Phys. Rev. Lett. 75,1635. Demin, R.V., L.1. Koroleva and A.M. Balbashov, 1997,
Ciria, M., J.1. Arnaudas, C. Dufour, V. Odemo, Phys. Lett A 231, 279.
K. Dumesnil and A. del Moral, 1998, Thin Solid Demin, R.V., L.1. Koroleva and A.M. Balbashov, 1998,
Films 317. 303. J. Magn. Magn. Mater. 177-181, 871.
Claeyssen, E, N. Lhermet, R. Le Letty and P. Bouch- Dieny, B., D. Givord, J.M. Ndjaka and J.M. Alameda,
iIIoux, 1997, J. Alloys Compo 258, 61. 1990. J. Appl. Phys. 67, 5677.
Clark, A.E., 1980, in: Handbook of Ferromagnetic Ma- Dieny, B., D. Givord and J.M. Ndjaka, 1991a, J. Magn.
terials, vol. I. ed. E.P. Wohlfarth (Elsevier Science, Magn. Maler. 93, 503.
North-Holland. Amsterdam) p. 539 Dieny, B.• V.S. Speriosu, S.S.P. Parkin. B.A. Gurney.
Clark, A.E., and H. Belson. I 972a, AlP Conf. Proc. D.R. Wilhoit and D. Mauri, 1991b, Phys. Rev. B 43,
No.5, p. 1498. 1297.
Clark. A.E.• and H. Belson, 1972b, Phys. Rev. B 5, Dime, EW.A.• and CJ.M. Denisson, 1989. J. Magn.
3642. Magn. Mater. 78.122.
Clark. A.E., and H. Belson. 1972c. IEEE Trans. Magn. Draaisma, HJ.G., and WJ.M. de Jonge, 1988, J. Appl.
MAG-8.477. Phys. 64, 3610.
Cochrane, R.W.• R. Harris and MJ. Zuckermann, 1978.
Du Tremolet de Lacheissene, E., 1993, in: Magne-
Phys. Rep. 48. I.
tostriction: Theory and Applications of Magneto-
Coehoom, R., 1990. in: Proceedings of the NATO-
elasticity (CRC Press. Boca Raton, USA).
ASI "Supermagnets, Hard Magnetic Materials", eds
Du Tremolet de Lacheisserie, E., 1995. Phys. Rev. B
GJ. Long and E Grandjean (Kluwer, Dordrecht).
51.15925.
Coey, J.M.D.• 1978, J. Appl. Phys. 49, 1646.
Coey, J.M.D.• D. Givord, A. Lienard and J.P. Rebouil- Du Tremolet de Lacheisserie, E., 1999, Magnetism
(Presses Universitaires de Grenoble).
lat. 1981, J. Phys. F 11. 2707.
Coey, J.M.D.• M. Viret and S. von Molnar. 1999, Adv. Du Tremolet de Lacheisserie, E.• and J.C. Peuzin,
Phys.48, 167. 1994. J. Magn. Magn. Maler. 136. 189.
Dabrowski. B.• X. Xiong. Z. Bukowski. R. Dybzin- Du Tremolet de Lacheisserie, E., and O. EK. McGrath.
ski. P.W. Klamut, J.E. Siewenie, C.W. Kimball, 1995, J. Magn. Magn. Mater. 147. 160.
O. Chmaissem, J.D. Jorgensen and S. Short, 1999. Du Tremolet de Lacheisserie, E., and J.C. Peuzin,
Phys. Rev. B 60, 7006. 1996. J. Magn. Magn. Maler. 152, 231 (see
Dabrowski. B.• L. Gladczuc, A. Wisniewski. A. Szew- K1okholm and Jahnes, 1996).
czyk. M. Gutowska, S. Kolesnik. C.W. Kimball and Du Tremolet de Lacheisserie. E., K. Mackay, J. Betz
H. Szymczak. 2000. J. Appl. Phys. 87, 301 I. and J.e. Peuzin, 1998. J. Alloys Compo 275, 685.
194 N.H. DUC and P.E. BROMMER

Due, N.H., 1997, in: Handbook of Physics and Chem- Eto, T., G. Oomi, E.V. Sampathkumaran, A. Sundare-
istry of the Rare Earths, vol. 24, eds K.A. Gschnei- san. M. Kosaka and Y. Uwatoko, 2001. Physica B
dner, Jr. and L. Eyring (Elsevier Science, North- 294-295, 111.
Holland. Amsterdam) p. 339. Fahnle, M., and M. Komelj, 2000, J. Magn. Magn.
Due, N.H., 2001, in: Handbook of Physics and Chem- Mater. 220. LB.
istry of the Rare Earths. vol. 33. eds K.A. Gschnei- Farber. P.• and H. Kronmiiller, 2000a, J. Magn. Magn.
doer. Jr. and L. Eyring (Elsevier Science, North- Mater. 214, 159.
Holland, Amsterdam). Farber. P.. and H. Kronmiiller, 2000b, 1. Appl. Phys.
Due, N.H., 2002, Proc. JEMS200J, Grenoble, to ap- 88,2781.
pear in 1. Magn. Magn. Mater. Feiner, L.F., and A.M. Oles, 2000, Physica B 259-261,
Due, N.H .• and D. Givord, 1996, J. Magn. Magn. 796.
Mater. 157-158, 169. Fischer, S.F.. M. Kelsch and H. Kronmiiller, 1999, J.
Due, N.H., and P.E. Brommer. 1999, in: Hand- Magn. Magn. Mater. 195, 545.
book of Magnetic Materials, vol. 12. ed. K.HJ. Forester, D.W.. C. Vittoria, J. Schelleng and P. Lubitz,
Buschow (Elsevier Science, North-Holland. Ams- 1978, J. Appl. Phys. 49, 1966.
terdam) p. 259. Franse, lJ.M., and RJ. Radwariski, 1993, in:
Due, N.H., and T. Goto, 1999, in: Handbook of Physics Handbook of Magnetic Materials, vol. 7, ed.
and Chemistry of the Rare Earths. vol. 26. eds K.HJ. Buschow (Elsevier Science. North-Holland,
K.A. Gschneidner, Jr. and L. Eyring (Elsevier Sci- Amsterdam) p. 307.
ence, North-Holland. Amsterdam) p. 171. Freeman, AJ., R. Wu, M. Kim and V.I. Gavrilenko,
Duc, N.H., K. Mackay, J. Betz and D. Givord, 1996, J. 1999,1. Magn. Magn. Mater. 203. I.
Appl. Phys. 79, 973. Fujimori, H.• 1.Y. Kim, S. Suzuki, H. Morita and
N. Kataoka, 1993,1. Magn. Magn. Mater. 124, 115.
Due, N.H .• K. Mackay, 1. Betz and D. Givord, 2000a,
Garda-Landa. B., L.M. De Teresa, M.R. Ibarra. C. Rit-
J. AppJ. Phys. 87, 834.
ter, R. Drost and M.R. Lees, 1998a, J. Appl. Phys.
Due N.H., T.M. Danh, H.N. Thanh, J. Teillet and
83,7664.
A. Lienard, 2000b, J. Phys.: Condens. Matter. 12,
Garda-Landa, B., M.R. Ibarra, L.M. De Teresa,
8283.
G. Zhao, K. Conder and H. Keller, I998b. Solid
Due, N.H., K. Mackay. J. Betz, Z. Sarkozi and
State Comm. 105,567.
D. Givord, 2000c, J. Phys.: Condens. Matter. 12,
Garda-Landa. B., C. Marquina, M. Hilbers,
7957.
M.R. Ibarra, P.A. Algarabel, A. del Moral,
Due, N.H., N.A. Tuan, D.T.N. Anh, T.M. Danh and
G. Balakrishnan, M.R. Lees and D. McK Paul,
N.P. Thuy, 2000d, Proceedings of the 8th Asian
2001 Physica B 294-295, 107.
Pacific Physics Conference, Taipei. August 7-10,
Gavigan. J.P.• D. Givord, H.S. Li and J. Voiron, 1988.
2000 (World Scientific, Singapore).
Physica B. 149, 345.
Due, N.H., T.M. Danh, N.A. Tuan and J. Teillet, 200la, Gibbs, M.R.J., 1992, Physica Scripta T45, 115.
Appl. Phys. Lett. 78, 3648. Gibbs, M.RJ., C. Sherwood, J.L. Dancaster,
Due, N.H .. F. Richomme, N.A. Tuan, D.T. Huong Gi- P.E.M. Frere and AJ. Jacobs-Cook. 1996, IEEE
ang. T. Verdier and J. Teillet, 200lb. Proc. JEMS Trans. Magn. 32,4950
200 I, to be published. Gibbs. M.RJ., R. Watts, W. Karl, A.L. Power and
Duc, N.H., T.M. Danh, N.A. Than and 1. Teillet, 200lc, R.B. Yates, 1997, Sensors Actuators A59, 229.
unpublished. Givord, D.• A.D. Santos. Y. Souche, J. Voiron and
Due, N.H., L.Q. Than, A. Fnidiki. 1. Ben Youssef, S. Wllchner. 1993. J. Magn. Magn. Mater. 121.216.
H. Le Gall and 1. Teillet, 200ld. submitted to J. Givord, D., 1. Betz, N.H. Due, K. Mackay and E.
Appl. Phys. du Tremolet de Lacheisserie, 1995, Proc. 2nd In-
Duenas, T.A., and G.P. Carman. 2000. 1. Appl. Phys. ternational Workshop on Materials Science. Hanoi,
87,4696. eds F.F. Becker, N.D. Chien. lJ.M. Franse and
Dwight, A.• and C. Kimball, 1974. Acta Crys. B 30, T.D. Hien, p. 287.
2791. Givord, 0 .. 1. Betz, K. Mackay, J.e. Toussaint, 1. Vo-
Engdahl, G., 1999, Handbook of Giant Magnetostric- iron and S. Wiichner. 1996,1. Magn. Magn. Mater.
tive Materials (Academic Press. New York). 159,71.
Erwin. R.W., J.J. Rhyne. M.B. Salamon, J.A. Borchers, Gradmann, U., 1993, in: Handbook of Magnetic Mate-
S. Sinha. R. Du, J.E. Cunningham and C.P. Flynn, rials, vol. 7. ed. K.HJ. Buschow (Elsevier Science.
1987, Phys. Rev. B 35, 6808. North-Holland, Amsterdam) p. I.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 195

Grundy, p.1., D.G. Lord and P.1. Williams, 1994, 1. Houqing, Z., L. Jianguo, W. Xiurong, X. Yanhong and
Appl. Phys. 76, 7003. Z. Hongping, 1997, J. Alloys Compd. 258. 49.
Gubin, S.P., V.V. Kislov, v.v. Kolesov, E.S. Soldatov Hristoforou, E., H. Chiriac and M. Neagu, 1998. Sen-
and A.S. Trifonov, 2000, Nanostruct. Mater. 12, sors and Actuators A 67,49
1131. Huang. J., C. Prados, J.E. Evetts and A. Hernando,
Gutjahr-Loser, Th., D. Sander and J. Kirschner, 2000, 1995, Phys. Rev. B 51, 297.
J. Magn. Magn. Mater. 220, L1. Iannotti, V., and L. Lanotte, 1999. J. Magn. Magn.
Gutierrez, J., P. Gorria, 1.M. Barandiaran and Mater. 202. 191.
A. Garcia-Arribas, 1995, in: Nanostructured and Ibarra. M.R.•and J.M. de Teresa, 1998, J. Magn. Magn.
Nanocryst. Mater., eds M. Vazquez and A. Her- Mater. 177-181, 846.
nando (World Sci. Publ., Singapore) p. 500. Ibarra, M.R., P.A. Algarabel, C. Marquina, J. Blasco
Halstrup, B., J. Betz, K. Mackay, J.C. Peuzin and and 1. Garda, 1995, Phys. Rev. Letters 75. 3541.
N. Lhermet, 19%, in: Micro Systems Technologies Ibarra. M.R., R. Mahendiran, C. Marquina, B. Garcia-
%. eds. H. Reichl and A. Heuberger (VDE-Verlag. Landa and J. Blasco. 1998. Phys. Rev. 57. R3217.
Berlin) p. 457. Jehan, D.A.. D.F. McMorrow. R.A. Cowley,
Hansen, P., 1991, in: Handbook of Magnetic Materi- R.C.C. Ward, M.R. Wells. N. Hagmann and
als. vol. 6, ed. K.H.J. Buschow (Elsevier Science, K.N. Clausen, 1993. Phys. Rev. B 48,5594.
North-Holland, Amsterdam) p. 289. Jerems, F.. and R.D. Greenough. 1999. J. Appl. Phys.
Hansen, P.• C. Clausen, G. Much, M. Rosenkranz and 85,4500.
K. Witer. 1989, J. Appl. Phys. 66, 756. Jiang. X.• H. Xu. Ch. Jiang and Sh. Gong. 2000. 1. Al-
Hayashi. Y., T. Honda, K.1. Arai, K. Ishiyama and loys Compo311, 86.
M. Yamaguchi, 1993, IEEE Trans. Magn. 29. 3129. Jin, S., H.M. O'Bryan, T.H. Tiefel, M. McCormack and
Hayashi. T., N. Miura, K. Noda and H. Kuwahara,
W.W. Rhades, 1995. Appl. Phys. Lett. 66,382.
2001. Physica B 294-295. 115.
Kadomtseva, A.M., Yu.F. Popov. K.I. Kamilov,
Henning. J.C.M.• and J.H. den Boef, 1978, J. Appl.
G.P. Vorobev, A.A. Mukhin, V.Yu. Ivanov and
Phys. 16, 353.
A.M. Balbashov, 2000, Physica B 284-288.1410.
Herbst, J.C.M., T.W. Capehart and F.E. Pinkerton,
Kaneyoshi, 1984. Amorphous Magnetism (CRC. Boca
1997, Appl. Phys. Lett. 70, 3041.
Raton. FL).
Hernando. A.• C. Prados and C. Prieto, 1983, J. Magn.
Kikuchi. S.. T. Tanaka, S. Sugimoto, M. Okada,
Magn. Mater. 37, 169.
M. Homma and K.l. Arai, 1993, J. Magn. Soc. Jpn.
Hernando, A.• M. Vazques, J. M. Barandiaran, 1988. J.
17.267.
Phys. E 21. 1121.
Kim. J.Y., 1993, J. Appl. Phys. 74. 2701.
Hernando. A.• C. Prados and C. Prieto, 1996. 1. Magn.
Magn. Mater. 157-158,501 Kimura, T.• Y. Tomioka, A. Asarnitsu and Y. Tokura,
Hernando, A., C. Gomez-Polo, M. EI Ghannarni and 1998, Phys. Rev. Letters. 81. 5920.
A. Garcia Escorial, 1997, J. Magn. Magn. Mater. Klokholm, E., 1976. IEEE Trans. Magn. 12.819.
173,275. Klokholm, E.• 1977. IBM Tech. Disci. Bull. 19.4030.
Herzer, G.• 1990, IEEE Trans. Mag. 26. 1397. Klokholm, E.• and C.V. Jahnes, 19%, J. Magn. Magn.
Herzer. Goo 1991, in: Proc. Int. Symp. on 3d Transition- Mater. 152.227.
Demi Metal Thin films, Magnetism and Processing Koksharov, Yu.A., S.P. Gubin, l.D. Kosobudsky,
(Japan Soc. for the Promotion of Science, Sendai, M. Beltran. Y. Khodorkovsky and A.M. Tishin,
Japan) p. 307. 2000,1. Appl. Phys. 88, 1587.
Herzer, G., 1993, Phys. Scr. T 49.307. Konishi, S.• S. Sugatani and Y. Sakurai, 1%9, IEEE
Herzer, G., 1997, in: Handbook of Magnetic Materi- Trans.Magn.~(;·5. 14.
als. vol. 10, ed. K.H.J. Buschow (Elsevier Science, Koroleva, L.I., and R.V. Demin, 1999, Physica B 259,
North-Holland. Amsterdam) p. 415. 816.
Hofmann. B., T. Reninger and H. Kronmuller, 1992, Kraus, L.. 1989.1. Phys.: Sci. Instrum, F 22.943.
Phys. Sat. Sol. (a) 134,247. Kraus, L., and J. Schneidner, 1977. Phys. Stat. Sol. 39&,
Holzer. D., I. Perez de Albeniz, R. Grossinger and K161.
H. Sassik, 1999, J. Magn. Magn. Mater. 203, 82. Kuwahara, H., Y. Tomioka, Y. Morotomo,
Honda, T., Y. Hayashi. K.I. Arai, K. Ishiyama and A. Asamitsu, M. Kasal, R. Kumai and Y. Tokura,
M. Yamaguchi. 1993. IEEE Trans. Mag. 29. 3126. 19%. Science 272, 80.
Honda, T., K.l. Arai and M. Yamaguchi, 1994, J. Appl. Lachovicz, H.K., and H. Szymczak. 1984. J. Magn.
Phys, 76, 6994. Magn. Mater. 41. 327.
196 N.H. DUC and P.E. BROMMER

Lafford, A., R. Zuberek and M.RJ. Gibbs, 1995. J. Ohnuma, S., N. Kobayashi, T. Masurnoto, S. Mitani
Magn. Magn. Mater. 140-144,577. and H. Fujirnori, 1999, J. Appl. Phys. 85, 4574.
Lee, C.H., H. He, FJ. Larnelas, W. Vavra, e. Uher and O'Handley, R.C., and S.W. Sun, 1992,1. Magn. Magn.
R. Clarke, 1990, Phys. Rev. B 42,1066. Mater. 104-107, 1717.
Le Gall, H., N. Vukadinovic, O. Navaro and J. Ben O'Handley, R.C., O.-S. Song and e.A. Ballentine,
Youssef, 1989, Proc. Joumee S.E.E. (SEE Ed. Paris) 1993, 1. App!. Phys. 74, 6302.
p.145. Ooike, T., S. Ishio and T. Miyazaki, 1993, J. Mag. Soc.
Le Gall, H., 1. Ben Youssef, F. Socha, N. Tiercelin, Jpn. 17,271.
V. Preobrazhensky and Pernod, 2000, J. Appl. Phys. Orsier, E., T. Hiramoto, J. Betz, K. Mackay,
87,5783 J.e. Peuzin, D. Givord and A. Garnier, 1996, Meca-
Lim, S.H., T.H. Noh, LK. Kang, S.R. Kim and tronics '96, Besancon (unpublished), p. 537.
S.R. Lee, 1994, J. Appl. Phys. 76, 7021. Osaka, T., 2000, Electrochimica Acta 45, 3311.
Lim, S.H., Y.S. Choi, S.H. Han, HJ. Kim, T. Shima Pasquale, M., A. Infortuna, L. Martino, e. Sasso,
and H. Fujimori, 1998,1. Magn. Magn. Mater. 189, e. Beatrice and S.H. Lim, 2000, J. Magn. Magn.
I Mater. 215-216, 769.
Liu, J.P., F.R. de Boer, P.F. de Chatel, R. Coehoom and Pinkerton, F.E., T.W. Capehart, J.F. Herbst,
K.HJ. Buschow, 1994,1. Magn. Magn. Mater. 134, E.G. Brewer and e.B. Murphy, 1998, J. Appl.
159. Phys. 83, 7252.
Ludwig, A., and E. Quandt, 2000, 1. Appl. Phys. 87, Polk, D.E., 1972, Acta Metal!. 20,485.
4691. Popov, Yu. F., A.M. Kadomtseva, G.P. Vonrob'ev,
Maignan, A., C. Martin, M. Hervieu and B. Raveau, V.Yu. Ivanov, A.A. Mukin, A.K. Zvezdin and
2001, Solid State Comm. 117,377. A.M. Balbashov, 1998, J. App!. Phys. 83, 7160.
Makino, A., T. Bitoh, A. Kojima, A. Inoue and T. Ma- Quandt, E., 1994,1. Appl. Phys. 75, 5653.
sumoto, 2000, J. Magn. Magn. Mater. 215-216, Quandt, E., 1997,1. Alloys Compd. 268,128.
288. Quandt, E., and K. Seemann, 1996, in: Micro Systems
Marin, P., and A. Hernando, 2000, J. Magn. Magn. Technologies 96, eds. H. Reichl and A. Heuberger
Mater. 215-216, 729. (VDE- Verlag, Berlin) p. 451.
Mathieu, R., Y. Guo, P. Svedlindh, P. Nordblad and Quandt, E., and A. Ludwig, 1997, Mat. Res. Soc.
R. Wlippling, 2000, Physica B 284-288, 1432. Symp. Proc., vol. 459 (Materials Research Society)
Mergel, D., H. Heitzmann and P. Hansen, 1993, Phys. p.565.
Rev. B 47, 882. Quandt, E., and A. Ludwig, 1999, J. Appl. Phys. 85,
Miyazaki, T., T. Saito and Y. Fujino, 1997, J. Magn. 6232.
Magn. Mater. 171, 320. Quandt, E., and A. Ludwig, 2000, Sensors Actuators
Morin, P., and D. Schmitt, 1990, in: Handbook of Mag- 81,275.
netic Materials, vol. 5, ed. K.H.J. Buschow (Elsevier Quandt, E., B. Gerlach and K. Seemann, 1994, J. Appl.
Science, North-Holland, Amsterdam) p. I. Phys. 76, 7000.
Mullin, A.A., v.yu. Ivanov, V.D. Travkin, S.P. Lebe- Quandt, E., A. Ludwig, J. Mencik and E. Nold, 1997a,
dev, A. Pimenov, A. Loidl and A.M. Balbashov, J. Alloys Compd. 258, 133.
2000, Physica B 284-288, 1414. Quandt, E., A. Ludwig, J. Betz, K. Mackay and
Nagi, Y., M. Senda and T. Toshima, 1988, J. Appl. D. Givord, 1997b, J. Appl. Phys. 81, 5420.
Phys. 63, 2356. Ramirez, A.P., 1997, J. Phys. Condens. Matter 9, 8171.
Nakano, H., Y. Motome and M. Imada, 2000, Physica Respaud, M., H. Rakoto, 1.M. Broto, 1. Vanacken,
B 284-288, 1406. C. Martin, A. Maignan, B. Raveau and S. Askenazy,
Nan, C.-W., and GJ. Weng, 1999, Phys. Rev. B 60, 2001, Physica B 294-295, 119.
6723. Rhyne, Ll., and R.w. Erwin, 1995, in: Handbook of
Narita, K., J. Yamasaki, H. Fukunaga, 1980, IEEE Magnetic Materials, vol. 8, ed. K.H.J. Buschow (El-
Trans. Magn. ~(;-16,435. sevier Science, North-Holland, Amsterdam) p. I.
Neel, L., 1953, Comptes Rendus 237, 1468. Riedi, K., M. Schnell, F. Schatz, M. Hirscher, B. Lude-
Neel, L., 1954,1. Phys. Radium 15, 225. scher, W. Sigle and H. Kronmuller, 1998, Phys. Stat.
Novosad, V., Y. Otani, A. Ohsawa, S.G. Kim, Sol. (a) 167, 195.
K. Fukamichi, J. Koike, K. Maruyama, O. Kitakami Riedi, zc., T. Thomson and GJ. Tomka, 1999, in:
and Y. Shimada, 2000, J. Appl. Phys. 87, 6400. Handbook of Magnetic Materials, vol. 12, ed.
O'Donovan, K.V., R.S. Beach, M.B. Salamon and K.HJ. Buschow (Elsevier Science, North-Holland,
e.P. Flynn, 1998,1. Magn. Magn. Mater. 186, 161. Amsterdam) p. 97.
MAGNETOELASTICITY IN NANOSCALE HETEROGENEOUS MAGNETIC MATERIALS 197

Sander, D., 1999, Rep. Progr. Phys. 62,809. Szymczak. H.• R. Zuberek, R. Krishnan, M. Tessier.
Sander, D., R. Skomski, A. Enders, C. Schmidthals, K.B. Youn and C. Sella. 1988, ICMFS-12 Confer-
D. Reuter andJ. Kirschner. 1998.J. Phys. D31, 663. ence Le Creusot, p. 3.
Sander. D., A. Enders and J. Kirschner. 1999, J. Magn. Szymczak, H.• R. Zuberek and J. Gonzalez, 1999. J.
Magn. Mater. 198-199.519. Magn. Magn. Mater. 191, 199.
Schatz. F., M. Hirscher, G. Aik and H. Kronmuller, Tam. A.C.• and H. Schroeder. 1988,1. Appl. Phys. 64.
1993. Phys. Stat. Sol. (a) 137, 197. 5422.
Schatz. F., M. Hirscher, M. Schnell, G. Aik and H. Kro- Tanaka. T.. H. Arimura, S. Kikuchi, S. Sugimoto.
nmuller, 1994,1. Appl. Phys. 76. 5380. M. Okada, M. Homma and K.I. Arai, 1995. 1. lpn.
Shick. A.B.• D.L. Novikov and AJ. Freeman. 1998. I. Inst. Metals 59, 447.
Magn. Magn. Mater. 177-181, 1223. Tejedor, M.• B. Hernando, M.L. Sanchez, V.M. Prida,
Shima. T.• H. Yokoyama and H. Fujimori, 1997. I. Al- I.M. Garcia-Beneytez, M. Vazquez and G. Herzer.
loys Compd. 258.149. 1998, J. Magn. Magn. Mater. 185.61.
Skorvanek. I.. P. Duhaj and R. Grossinger, 2000. I. Teter, J.P., G.c. Homer and L. Bromberg. 2000. J.
Magn. Magn. Mater. 215-216. 431. Appl. Phys. 87, 6313.
Slawska-Waniewska, A.• P. Nowicki, H.K. Lachowicz, Trippel. G.• 1977. IBM Tech. Rep. No. TR28.094.
P. Gorria, 1.M. Barandiaran and Hernando. 1994, Troyanchuk, 1.0., N.V. Samsonenko, A. Nabialek and
Phys. Rev. B 50, 6465. H. Szymczak. 1997, I. Magn. Magn. Mater. 168,
Slawska-Waniewska, A.• R. Zuberek and P. Nowicki. 309.
1996,I. Magn. Magn. Mater. 157-158, 147. Troyanchuk, 1.0.. N.V. Kasper. D.D. Khalyavin,
Slawska-waniewska, A., A. Roig, E. Molins, H. Szymczak, R. Szymczak and M. Baran, 1998,
I.M. Greneche and R. Zuberek, 1997. 1. Appl. Phys. Rev. Lett. 80. 3380.
Phys. 81, 4652. Twarowski, K.. H.K. Lachowicz, M. Kumiski,
Smith, A.B.• 1968, Rev. Sci. Instrum. 39, 378. A. Slawska-Waniewska and G. Herrer, 1995.
Smith, A.B.• and R.V. Jones, 1963. J. Appl. Phys. 34. J. Magn. Magn. Mater. 150, 85.
1283. Van de Riet. E., 1994, J. Appl. Phys. 76, 584.
Sontag. H., and A.C. Tam, 1986. IEEE Trans. U1trason. Vukadinovic, N.• 1988. Thesis. University of Paris.
Ferroelectr. Freq. Cont. 33. 500 Wada, M.• H.H. Uchida, Y. Matsumura, H. Uchida and
Speliotis, A., O. Kalogirou, N. Vouroutzis and D. Niar- H. Kaneko, 1996, Thin Solid Films 281. 503.
chos, 1998.1. Magn. Magn. Mater. 187. 17. Wada. M.• H.H. Uchida and H. Uchida. 1997a, J. Al-
Stobiecki, T., 1. Wrona. M. Czapkiewicz, C. Pra- loys Compounds 258. 143.
dos, FJ. Castano, D. Garcia. M. Vazquez, Wada, M.• H. Uchida and H. Kaneko, 1997b. J. Alloys
M. Kopcewicz, A. Grabias and R. Zuberek, 2000, Compounds 258, 169.
1. Magn. Magn. Mater. 215-216, 566. Wada, M.• H.H. Uchida and H. Uchida, 1997c. J. Al-
Sun. S.W.• and R.C. O'Hand1ey, 1991. Phys. Rev. Lett. loys Compounds 258. 174.
66.2798. Wada, M., H. Uchida and H. Kaneko. 1997d. 1. Alloys
Suzuki. K., A. Makino. N. Kataoka, A. Inoue and Compounds 258, 42.
T. Masumoto, 1991, I. Appl. Phys. 70. 6232. Wandass. 1.H.• J.S. Murday and R.I. Colton. 1988,
Swaddling, P.P.. D.F. MacMorrow, I.A. Simpson. Preprint, Naval Research Laboratory.
R.C.C. Ward. M.R. Wells and K.N. Clausen. 1992, Wang. J.H.• H.Y. Chen, J.H. Wu. Z.X. Liu, T.Y. Chen
J. Phys.: Condens. Malter. 5. L487. and D.S. Dai, 1998. Solid State Comm. 108. 701.
Szumiata, T., H. Szymczak and R. Zuberek, 1993, Wang. B.W., S.C. Busbridge, Y.X. Li, G.H. Wu and
IEEE Trans. MAG·29. 3132 A.R. Piercy. 2000. J. Magn. Magn. Mater. 218.198.
Szumiata, T.• R. Zuberek, I. Gonzalez. A. Slawska- Weber. M.• R. Koch and K.H. Rieder. 1994. Phys. Rev.
Waniewska and H. Szymczak. 1999, 1. Magn. Lett. 73, 1166.
Magn. Mater. 203, 262. Weber. W.• C.H. Back, U. Ramsperger, A. Vaterlaus
Szymczak, H., 1997. J. Appl, Phys. 81. 5411. and R. Allenspach, 1995. Phys. Rev. B 52. 14450.
Szymczak, H., 1999. J. Magn. Magn. Mater. 200,425. Weber, W.• C.H. Back, A. Bischof, Ch. Wilrsch and
Szymczak, H., 2000, J. Magn. Magn. Mater. 211, 186. R. Allenspach, 1996a, Phys. Rev. Lett. 76.1940.
Szymczak, H., and R. Zuberek, 1982, J. Phys. F 12, Weber, W., A. Bischof and R. Allenspach, 1996b. Phys.
1841. Rev. Lett. 76.3424.
Szymczak, H., and R. Zuberek, 1983, 1. Magn. Magn. Williams, P.I.. D.G. Lord and P.I. Grundy, 1994, 1.
Mater. 35. 149. Appl. Phys. 75, 5257.
198 N.H. DUC and P.E. BROMMER

Winzek, 8., M. Hirscher and H. Kronmuller, 1999. J. Zuberek, R., D. Zymierska and H. Szymczak. 1994.
Alloys Compounds 283, 78. Acta Phys. Polonica A 85, 439.
Wu. R. and A.1. Freeman. 1999. J. Magn. Magn. Mater. Zuberek, R., K. Fronc, H. Szymczak, A. Nabiaiek,
200.498. T. Stobiecki and J. Sokulski, 1995. J. Magn. Magn.
WU,R.Q .• L.1. Chen. A. Shick and A.1. Freeman, 1998, Mater. 139, 157.
J. Magn. Magn. Mater. 177-181, 1216. Zuberek, R., N. Murillo, J. Gonzalez. A. Slawska-
Wiichner, S.• J, Voiron, J.C. Toussaint and J.1. Prejean. Waniewska, K. Fronc, T. Kulik and lA. Gonza-
1995. J. Magn. Magn. Mater. 148,264. lez, 1997a. Proc. Rapidly Quenched and Metastable
Zener, c., 1951. Phys, Rev. 82, 403. Materials, Bratislava, Slovakia. 25-30 Aug. 1996.
Zuberek, R.. H. Szymczak. R. Krishnan. K.B. Youn Supplement (Elsevier) p. 213.
and C. Sella, 1987,lEEE Trans. Magn. 23, 3699. Zuberek, R.• K. Fronc, E. Mosiniewicz-Szablanska and
Zuberek, R..H. Szymczak. R. Krishnan and M. Tessier. H. Szymczak. I997b. J. Phys., IV Proc. Soft Mag-
1988. J. Phys. 49. C8, 1761. netic Materials vol. 13. Grenoble. 24-26 Sep. 1997.
Zuberek, R., A. Wawro, H. Szymczak and M. Baran. Zuberek, R.• N. Murillo, J. Gonzalez, J.M. Blanco and
1991, in: Proc. of the 5th Int. Conf. On Phys. P. Garcia-Tello. 1998. Nanostruct. Mater. 8. 711.
Of Magnetic Materials. eds W. Gorzkowski, Zuberek, R., A. Wawro, H. Szymczak. A. Wisniewski,
M. Gutowski. H.K. Lachowicz and H. Szymczak W. Paszkowicz and M.R.1. Gibbs. 2000, r. Magn.
(World Scientific. Singapore) p. 425. Magn. Mater. 214. 155.
chapter 3

MAGNETIC AND SUPERCONDUCTING


PROPERTIES OF RARE EARTH
BOROCARBIDES OF THE TYPE RNi2B2C

K.-H. MOLLER, G. FUCHS, S.-L. DRECHSLER


Leibniz-Institut fUr Festkorper- und Werkstofforschung Dresden
POB270116, 0-01171 Dresden
Germany

V.N. NAROZHNYI

Institute for High Pressure Physics, Russian Academy of SCiences


Troitsk, Moscow Region, 142190
Russia

Handbook of Magnetic Materials. Vol. 14


Edited by K.H.J. Buschow
@ 2002 Elsevier Science B. V. All rights reserved

199
CONTENTS

I. Introduction . . . . . . . . . . . . . 202
I. I. Discovery of the RNi2B2C superconductors 202
1.2. Boron and carbon based superconductors . . . 202
1.3. On the interplay of superconductivity and magnetism .. 207
1.4. Specific features of the RNi2 B2C compounds . 215
2. Crystal structure and chemical composition . 218
2.1. The LuNi2B2C-type structure . . 218
2.2. Lattice distortions due to magnetoelastic effects . 221
2.3. Single-, double- and triple-layer borocarbides (nitrides) 223
2.4. Related R-T -B-C(N) phases . 224
3. Basic properties of YNi2B2C and LuNi2B2C . 226
3. I. Normal state electronic properties and the superconducting stale 226
3.2. The upper critical field. . . . . . . . . . . . . . . . . . . 230
3.3. Magnetotransport . 234
3.4. Characteristics of superconducting YNi2B2C and LuNi2B2C 240
4. Magnetic and superconducling properties of RNi2B2C . 241
4.1. Magnetic order and the crystalline electric field . 243
4.2. CeNi2B2C 246
4.3. PrNi2B2C 247
4.4. NdNi2B2C 252
4.5. SmNi2B2C . . . 253
4.6. GdNi2B2C .. 254
4.7. TbNi2B2C .. 256
4.8. DyNi2B2C 257
4.9. HoNi2B2C 259
4.10. ErNi2B2C 266
4.11. TmNi2B2C .. 268
4.12. YbNi2B2C .. 270
5. Vortex lattices in RNi2B2C superconductors .. 272
5.1. Vortex lattice in non-magnetic borocarbides 272
5.2. Vortex lattice and magnetic order in ErNi2B2C and TmNi2B2C . 276
5.3. Vortex pinning and magnetic order . 277
6. Superconductivity in R(Ni,T}2B2C and (R, R')Ni2B2C . 277
6.1. R(Ni,T}2B2C compounds (T = Co, Cu, Pd, PI etc.) 277

200
MAGNETIC AND SUPERCONDUCfING PROPERTIES OF RNi2B2C 201

6.2. Effects of disorder . . . . . . . . . . . . . . . . . . . . 279


6.3. Magnetic impurities in a nonmagnetic superconductor 286
6.4. Nonmagnetic impurities in an antiferromagnetic superconductor 288
7. Conclusions . . . 289
Acknowledgements 291
References . . . . . 291
1. Introduction

J.J. Discovery ofthe RNi2B2C superconductors

Superconductivity in quaternary rare-earth transition-metal borocarbides (RTBC ) has been


discovered when for seemingly single-phase polycrystaIline samples of the hexagonal
compound YNi4B, at about 12 K, a drop in resistivity as well as in susceptibility
has been observed (Mazumdar et al. 1993). However, the superconducting phase in all
investigated YNi4B samples was at least a minor fraction of the material (~ 2%). It
had been suggested that the superconductivity in the YNi4B samples may be due to
a phase stabilized by the presence of an element other than Y, Ni and B. This was
supported by the observation of bulk superconductivity in polycrystaIline material with
the nominal composition YNi4BCo.2 (Nagarajan et al. 1994). At the same time Cava et al.
(l994a, 1994b) reported results on superconductivity in mu1tiphase YPd5B3Co.35 with a
transition temperature Teas high as 23 K and single-phase materials of the composition
RNhB2C (R =Y, Lu, Tm, Er, Ho with r, ~ 15.5 K, 16.5 K, 11 K, 10.5 K, 8 K),
respectively. Obviously, the superconducting behaviour of the YNi4B and YNi4BCo.2
samples mentioned above is caused by YNi2B2C. Consequently this was the discovery of
the first superconducting quaternary intermetaIlic compound. In the case of the Y-Pd-B-C
system the classification of the phase being responsible for T c ~ 23 K was much more
complicated because, so far, only multiphase superconducting material can be prepared
for this system. Not all of the phases in superconducting Y-Pd-B-C materials could be
identified and evidence for at least two superconducting phases has been reported (Hossain
et al. 1994a). Only quite recently it has finally been shown by a microanalysis technique
that YPd2B2C is the 23 K superconducting phase (Dezaneti et al. 2(00). A typical
de susceptibility-versus-temperature transition curve for polycrystaIline LuNhB2C and
YNi2B2C with T c ~ 16.5 K and 15 K is shown in fig. 1. The growth of very high quality
single crystals of nickel borocarbide superconductors (see e.g. Xu et al. 1994) almost
immediately after their discovery has had a profound impact on the quality of the work
performed. Thus many of the pitfalls of the early research on other complex materials, such
as high T c superconductors, carried out on polycrystaIline samples of variable quality, have
essentially been avoided (Cava 2(01).

J.2. Boron and carbon based superconductors


According to the BCS theory of superconductivity the critical temperature

(I)

202
MAGNETIC AND SUPERCONDUCI1NG PROPERTIES OF RNi2B2C 203

RNi2B2C
o.5 ~&TI-1
~
::i 0

;-2
..
I
I

i
Cl
::J
i:! 14 16
T(KI
IS
E 0 .0
Gi
'1'
o
FC
i" -0.5
~ R=V Lu
ZFC
-1. 0
.J
5 10 15 20
TemperalUf. (K)

Fig. I. Temperature dependence of the de magnetic susceptibility of LuNi2B2C and YNi2B2C in a magnetic
field of 20 De. ZFC and FC means zero field cooling and field cooling, respectively (after Nagarajan et al, 1994).

is determined by the Debye temperature OD representing the phonon spectrum, the normal
state electron density of states N(EF) at the Fermi level and some measure V of the
electron-phonon interaction (Bardeen 1992). Although formula (1) had been derived for
simple systems with the superconductivity driven by electron-phonon interaction, under
the condition N(EF)V « I, it has been successfully applied to describe qualitatively
superconductivity in a very wide class of materials. The value of OD monotonically
increases with the inverse mass of the atoms participating in the lattice vibrations of the
considered material. Therefore low-mass elements and their compounds are considered
as candidates for superconductors with high critical temperature. Thus, monatomic or
diatomic forms of metallic hydrogen are expected to exhibit superconductivity at quite high
temperatures (Ashcroft 1968; Richardson and Ashcroft 1997). However, hydrogen based
superconductivity has not yet been found. The difficulty is to have, simultaneous with
the large Debye temperature, conduction electrons with a large N(EF) and a sufficiently
large V. Besides hydrogen, lithium and beryllium other light elements such a boron
and carbon should be beneficial for increasing T c . For carbon this prediction has been
excellently confirmed recently (see table 1). A critical temperature as high as 117 K
could be achieved upon hole doping of the pristine solid version of the fullerene C60
intercalated by CHBr3 (Schon et al. 2001, see fig. 2). Because C60 is a strong electron
acceptor and, consequently, chemical hole doping of this material is difficult it was done
by electrostatic means using a field-effect transistor device geometry. An advantage of this
method is that, contrary to chemical doping, the doping level can be varied without changes
and imperfections in the crystal lattice. The intercalation with electronically inert spacer
molecules such as CHCh or CHBr3 results in a lattice expansion and, consequently, the
electronic bands narrow and the density of states increases. Thus the doping level and the
value of N(EF) can be controlled independently. It should be noted that the frequency
spectrum of C60 intramolecular vibrations extends to high energies and several of these
modes couple significantly to the electronic states providing considerable intramolecular
~

TABLE 1
Some boron and carbon containing superconductors
Boron and borides Carbon and carbides Borocarbides
T, (K) Space gr. T, (K) Space gr. T, (K) Space gr.
Structure Structure Structure
B 11.2 [I] HP C60 52 [l2] ET LuB2C2 2.4 [26] P4/mbm
~/CHBr3 117 [13] ET YB2C2 3.6 (26) LaB2C2
YBI2 4.7 (2) Fm3m Rb3C60 28 (14) Fm3m
ZrB12 5.8 [2) UB I2 BiF3 M02BC 7.5 (27) Cmcm
BEDT-TTF- 12.8 [l5) HP organic MOzBC
YB6 7.1 [2) Pm3m based salt ~
LaB6 5.7 [2] CaB6 YC2 4.0 [16] I4Immm LuNi2B2C 16[28) 14/mmm ;I:
CaC2 ScNi2B2C 15 [29) LuNi2B2C 3:
MgB2 39 [3) P61mmm La2C3 11 [l7] 143d ThNi2B2C 8 [30) c:
C-
ReB2 6.3 [4) A1B2 (Y,Thh C3 17 [l8) PU2C3 YNi2B2C 15.5 [28] c-
m
YPd2B2C 23 [31,32,35) :oc
~
NbB 8.3 [5] Cmcm MOS6 C 44 13 [l91 Fm3m YPt2B2C 10 [331
NaCI ~
TaB 4.0[5) CrB NbCy 11.8 [20) YRu2B2C 9.7 [341

M02B 5.1 [61 14/mcm M02C 12.2 [51 orthorh.


CuA12
Re3B 4.7 [4) Cmcm LaNiC2 2.7 [21] Amm2
Re3B (La,Th)NiC2 7.9 [22) CeNiC2
LuRuB2 10.0 (7) Pnma
YRuB2 7.8 (7) LuRuB2 LaBrC 7.1 [23J C2Im
YIC 10.0 [24) Gd2C212
YOS3B2 6.0 (7) P6Immm Y(Br,I)C 11.6 [241
LuOs3B2 4.7 [8) CeC°3B2
3:
TABLE 1
(Continued) ~
('5
Boron and borides Carbon and carbides Borocarbides
Te(K) Space gr. Te(K) Space gr. T e (K) Space gr. ~
o
Structure Structure Structure CIl

I
M03Al2C 10 [l9J P4132
~B4 8 9 [9,IOJ P42/nmc tl-Mn
LuRh.i B4 8 12 [9,IOJ CeC04B4 [II] MgNi3C 8.5 [25] Pm3m
YR!4 B4 10 11 [9,1OJ SrTi03
HP- under high pressure, ET - electric field induced doping using field-effect transistor geometry,
[I] Eremets et aI. (2001), [2] Matthias et aI. (1968), [3] Nagamatsu et aI. (2001), [4] Strukova et aI. (2001), [5] Savitskii et aI. (1973), [6] Havinga et aI. (1972), [7]
Ku and Shelton (1980), [8] Lee et aI. (1987), [9J Fischer and Maple (1982), [10] Maple and Fischer (1982), [11] The alternative types LuRu4B4 (s.g. I41/acd) and
~
LuR!4B4 (s.g. Ceca) have also been reported for theRR!4B4 compounds (RogI1984), [12] Schon et aI. (2000), [13] Schon et aI. (2001), [14] Rosseinsky et aI. (1991),
[IS] WJ1liams et aI. (1991), [16] Gulden et aI. (1997), [l7J Giorgi et aI. (1969), [18] Krupka et aI. (1969), [19J Fink et aI. (1965), [20] Gusev et aI. (1996), [21J Lee et aI.
s~
.."

(1996), [22] Lee and Zeng (1997), [23] Simon et aI. (1991), [24] Henn et aI. (2000), [25] He et aI. (2001), [26] Sakai et aI. (1982), [27] Lejay et aI. (1981), [28] Cava ~
et aI. (l994b), [29] Ku et aI. (1994), [30] Lai et aI. (1995), [31] Tominez et aI. (1998), [32] Dezaneti et aI. (2000), [33] Cava et aI. (1994d), [34] Hsu et aI. (1998), [35] iii
CIl
Cava et aI. (1994a). o"rl
~
Z
':;'
..,
t:tI
o

~
206 K.-H. MULLER et aI.

A Drain
Source Gate I
2 Al20a1 i: :::tl

o L.......L-.o.....L...L..L..L-L......u......L...................L............... ...J
o 20.-0 80 80 100 120 1.-0 180 180 200
Temperature (K)

Fig. 2. Resistivity-vs.-temperature transition curves for some C60 based superconductors. (A) Variation of the
hole doping from 1.3 to 3.2 holes per C60 molecule. Inset: the field-effect transistor geometry used in the
experiment. (B) Comparison of optimum hole-doped C60. as grown and intercalated with CHCI3 and CHBr3)
respectively (Schon et al. 2(01).

electron-phonon interaction. Therefore these phonon modes and their coupling to the
electrons will not much be influenced by the doping and intercalation procedure. The
maximum value of T c achieved in C60 by hole doping without lattice expansion is 52 K
(see table I; Schon et al. 2000). Electron doping of C60, which can be done by chemical
as well as electrostatic means, results in lower values of T c because of the higher density
of states in the valence band than in the conduction band. As an example, the chemically
electronically doped bulk material Rb3C60 has a T c of 28 K (Rosseinsky et al. 1991; see
table I). Superconductors with remarkably high critical temperatures have also been found
among organic compounds and inorganic carbides (see table I).
At ambient pressure boron is an insulator consisting of 12-atom icosahedral units. As
reported by Eremets et al. (2001), under high pressure B becomes not only metallic,
as predicted by Mailhiot et al. (1990) but even superconducting and it has a positive
pressure derivative of the critical temperature d T c / dP . Pressure induced superconductivity
has also been found in organic compounds (e.g. BEDT-TTF in table I), spin-ladder
cuprates (Uehara et al. 1996) and many other materials. Obviously pressure can cause,
through various mechanisms, crystallographic and electronic structures that are favorable
for superconductivity. On the other hand the electronic bands of a metal will broaden if
the material is compacted which is consistent with the fact that negative dT c/dP has been
observed for many superconductors (Wijngarden and Griessen 1992). Therefore pressure
MAGNETICAND SUPERCONDUCTING PROPERTIES OF RNi2B2C 207

-8Co) 80

c:
~ 60
'-'
0 s::::PMg
:E
- 40

~
.;!l
</>
20
~
0
0 20 40 60 80 100

Temperature (K)

Fig. 3. Resistivity-vs.-temperature transition curve and crystal structure of MgB2 (after Nagamatsuet al. 2(01).

induced superconductivity, also in the case of boron, is expected to be characterized by


a nonmonotonic pressure dependence of T c with a maximum value of T c at a certain
pressure. Such a behaviour has been confirmed, e.g. for iron (Shimizu et a1. 200 1) and spin-
ladder cuprates (Dagotto 1999). Superconductivity is also known for many borides (see
table 1). The most surprising example is MgB2 a binary compound with a simple crystal
structure, which is well known for many years (Russel et a1. 1953). But, unbelievably, its
transport and magnetic properties had not been investigated until quite recently although
there was an intensive search, on a large international scale, for higher values of T c in the
family of binary compounds. The highest critical temperatures were achieved for A15-
type compounds with a maximum value of about 23 K which could not be improved
since the early seventies until the discovery of the high- T c cuprate superconductors in
1986 (Bednorz and MUller 1986; Wu et a1. 1987). In 2001 Nagamatsu et al, (2001)
found a critical temperature as high as 40 K for MgB2 (see fig. 3). Electronic structure
calculations show that MgB2, being essentially metallic boron held together by covalent
B-B and ionic B-Mg bonding, is electronically a typical sp metal (Kortus et al. 2(01).
The crystal structure of MgB2 may be regarded as that of completely intercalated graphite
with carbon replaced by boron. Thus the band structure of MgB2 is graphitelike, but with
n bands falling deeper than in graphite, and two-dimensionality features are assumed
to be important for the superconductivity in this compound as well as in the surface
of the above discussed solid C60 doped by electrostatic means (Ann and Pickett 2001;
Belashchenko et a1. 200 I). A strong influence of 2D effects on T c had been discussed by
Ginzburg (1964, 2000) but these aspects are yet to be understood. On the other hand, the
electronic structure of the quaternary borocarbides RNizB2C is clearly three-dimensional
(see section 3.1). The lattice structure of LuB2C2 and YB2C2 (see table 1) contains well-
separated BC layers, suggesting 2D behaviour. However, the electronic properties of these
low-Z', superconductors are not yet well investigated.

J.3. On the interplay ofsuperconductivity and magnetism


The discovery of the RTBC superconductors generated great excitement for two reasons.
First. T c ::::; 23 K in the Pd-system was, at that time. the highest known transition
208 K.-H. MULLER et aI.

temperature for bulk intermetallics. Such a high T c had been reported for thin Nb3Ge-
films, two decades before (Gavaler et al. 1974). Apart from the relatively high values
of their T c the RTBC have attracted a great deal of attention because they contain rare-
earth magnetic moments in high concentration, which are coupled by exchange interaction.
The interplay between the two collective phenomena magnetism and superconductivity has
been an active area of interest for many years (reviews: Fischer and Maple 1982; Maple and
Fischer 1982; Bulaevskii et al. 1985; Fischer 1990). In this section we will briefly review
this problem starting with compounds where superconductivity and magnetism completely
(to our present knowledge) exclude each other, then continuing with systems for which
some kind of coexistence of these two phenomena was observed and finishing with the
recently discovered coexistence of superconductivity and weak itinerant ferromagnetism.

1.3.1. Superconductivity and magnetic ordering as antagonistic phenomena


In the usual BCS theory of superconductivity electrons are paired with opposite spins
and, obviously, they cannot give rise to magnetically ordered states. Hence magnetic order
and superconductivity should be antagonistic. In high- T c cuprate materials, depending on
the doping rate, the Cu 3d electrons (or holes) contribute to a localized antiferromagnetic
(or spin glass) state or they participate in superconductivity, i.e. the two phenomena do not
coexist (Aharoni et al. 1988; Luke et al. 1990).
Intriguing forms of competition between superconductivity and ferromagnetism have
recently been reported for the elements carbon and iron, where the two cooperative
phenomena are related to different crystallographic structures. As discussed in section 1.1
pristine C60 consisting of dominantly van der Waals bounded C60 molecules becomes
superconducting if doped by electrons or holes. On the other hand, under sufficiently
high pressure and temperature, a layered rhombohedral structure of C60 forms where,
within the layers, the C60 molecules are covalently bound. This phase is metastable at
room temperature and ambient pressure. It shows a spontaneous magnetization, which is
assumed to be based on unpaired electrons created by structure defects (Makarova et al.
2001; Xu and Scuseria 1995).
It is well known that, at pressure above 10 GPa, Fe transforms from a ferromagnetic
cubic phase into a non-ferromagnetic hexagonal phase. Wohlfarth (1979) argued that the
hexagonal iron may be a low-temperature superconductor. This prediction has now been
confirmed by Shimizu et at. (2001) who found superconductivity in Fe below 2 K at
pressures P between 15 and 30 GPa. An interesting open question is whether high-pressure
iron is an unconventional superconductor with Cooper paring mediated by magnetic
fluctuations (as proposed by Fay and Appel in 1980) instead of phonons. At T = 0 the
superconductivity disappears at a quantum critical point (P ~ 30 GPa in fig. 4). This may
be due to reduced magnetic fluctuations or to a reduced density of states N (E F) caused by
electron-band broadening at higher densities.

1.3.2. Superconductors with magnetic impurities


The superconducting state can coexist with magnetic moments of localized electrons
(e.g. of 4f type). It was experimentally found by Matthias et at. (l958a) that for rare-
earth impurities substituted into a superconductor T c rapidly decreases with increasing
impurity concentration and that superconductivity is completely destroyed beyond a
MAGNETIC AND SUPERCONDUCfING PROPERTIES OF RNi2B2C 209

2000.----------.,.----,

hcp

o 20 40 60
Pressure (GPa)

Fig. 4. Proposed temperature-pressure phase diagram of iron (after Saxena and Littlewood 2(01) with the
ferromagnetic body-centered cubic (bee) phase. the paramagnetic face-centered cubic phase (fcc) and the
hexagonal close-packed phase (hcp).

critical concentration of the order of one percent. This has been well understood by
a theoretical approach of Abrikosov and Gor'kov (1961) who took into account that
scattering by magnetic impurities leads to pair breaking. However, many systems with
rare-earth magnetic moments show deviations from the behaviour predicted by Abrikosov
and Gor'kov (AG). As has been proven theoretically (Keller and Fulde 1971; Fulde and
Peschel 1972) and confirmed by many experiments, effects of crystalline electric fields
on the magnetic moments result in a weaker decrease of T c with increasing impurity
concentration compared to the AG prediction. On the other hand, it was demonstrated
theoretically by Muller-Hartman and Zittartz (1971) and experimentally by Riblet and
Winzer (1971) that effects of hybridization and strong correlation (Kondo effect) may
cause a considerably stronger reduction of T c than predicted by AG. Furthermore the
AG predictions will fail for higher concentrations of the rare-earth magnetic moments
which are usually coupled by certain types of indirect exchange interaction and show
cooperative magnetic phenomena. The first example where such deviation from the
AG behaviour has been realized is CeRu2 where over 30% of non-magnetic Ce can be
replaced by Gd (Matthias 1958b; Peter et al. 1971), Tb (Hillebrand and Wilhelm 1970;
Ferdinandez-Baca and Lynn 1981) or Ho (Lynn et al. 1980; Willis et al. 1980) before
superconductivity is suppressed. The measurements of susceptibility, specific heat and
Mossbauer effect as well as neutron scattering clearly indicated that the ordering of the
heavy-rare-earth magnetic moments in these materials is of spin-glass or, strictly speaking,
cluster-glass type with short range ferromagnetic order (Roth 1978; Davidov et al. 1977).
For the pseudobinary systems (Gd,La)Ru2 (Jones et al. 1978) and (Nd,Th)Ru2 (Huser et al.
1983) which are superconducting spin glasses or cluster glasses, similar as the (R,Ce)Ru2
systems mentioned above, even so called reentrant superconductivity occurs as shown
in fig. 5 for the compound Nd(l.3sThO.6SRu2. The competition between superconductivity
and ferromagnetic short-range order results in a complicated non-monotonic temperature
dependence of the susceptibility. In the temperature range around 1 Kelvin the material is
superconducting IxI < 0). Cooling below this temperature range would return the material
into the normal state (X I > 0). A reentrance of superconductivity is observed below 0.2 K.
210 K.-H. MULLER et aI.

o 0.5 1.0 1.5 2.0


Temperature T (K)

Fig. 5. Reentrant superconductivity in Nd(l.3SThO.6SRu2 (after HUser et aI. 1983). Upon cooling, first, the real
part of the ac susceptibility, X', becomes negative, indicating superconductivity. Then the material reenters the
normal state (X' > 0) and reentrance of superconductivity occurs at lower temperatures.

The reason why reentrant behaviour occurs in (Gd,La)Ru2 and (Nd,Th)Ru2 but not in the
(R,Ce)Ru2 system may be that in the former compounds the magnetic correlation length
of the spin-glass state is closer to the superconducting coherence length than in (R ,Ce)RU2
compounds. The reentrance of superconductivity at low temperatures (in fig. 5) is attributed
to the well known fact that, in diluted magnetic materials, larger ferromagnetic clusters
become unstable at low temperatures (HUser et al. 1983; Nieuwenhuys et al. 1979).

J.3.3. Superconductivity and local-magnetic-moment cooperative phenomena


To understand the interplay of superconductivity and magnetism in systems containing
localized magnetic moments in high concentration Gor'kov and Rusinov (1964) extended
the AG theory taking into account cooperative magnetic phenomena. They concluded
that ferromagnetism would destroy superconductivity because the conduction electrons
will be polarized by exchange interaction with the ordered magnetic moments. Ginzburg
(1956) had pointed out, already before, that superconductivity and ferromagnetism in
(type-I) superconductors can only coexist if the magnetic induction caused by the
magnetization M s is smaller than the critical field of the superconductor, i.e, the
spontaneous magnetization has to be sufficiently small. For type-Il superconductors this
conclusion has to be modified as only states must be excluded which, at the same
time, are homogeneously magnetized and homogeneously superconducting. This will be
achieved if the induction caused by M s is smaller than the lower critical field He I. An
alternative solution of the dilemma is the self-induced formation of vortex structures (see
e.g. Fulde and Keller 1982). A further possibility is that the electromagnetic coupling of
superconductivity and magnetism causes an oscillating magnetic (instead of homogeneous
ferromagnetic) order which coexists with a homogeneous superconducting state. The
wavelength of the oscillations is governed by the penetration depth A of the superconductor
(Blount and Varma 1979; Matsumoto et al. 1979). An alternative mechanism for oscillatory
magnetic order has been proposed by Anderson and Suhl (1959): the strength of
the exchange interaction between the rare-earth magnetic moments mediated by the
conduction electrons (RKKY interaction) is changed in the superconducting state because
the electron-spin susceptibility is reduced in the long-wavelength range. Consequently
MAGNETIC AND SUPERCONDUCTING PROPERTIES OF RNi2B2C 211

the effective exchange interaction in the superconducting state will have a maximum at
a finite wavelength, leading to an oscillatory magnetic state, even if the material would
be ferromagnetic in the absence of superconductivity. The wavelength of this state is
controlled by the coherence length ~ of the superconductor.
It was predicted by Baltensperger and Strassler (1963) that antiferromagnetic order may
coexist with superconductivity. The first examples of compounds where true long range
magnetic order coexisting with superconductivity has been observed are ternary Chevrel
phases RM06SS and RRh4B4 compounds (see Fischer and Maple 1982; Maple and Fischer
1982). In these materials there is a separate fully occupied rare-earth sublattice. It is
assumed that the magnetic moments and the superconducting electrons in these compounds
belong to different more or less "isolated" sublattices, which supports superconductivity
to exist despite the high concentration of localized magnetic moments (Lynn 200 I).
The magnetic ordering temperatures are low (~ I K) whereas T e is considerably larger.
Therefore it cannot be excluded that magnetostatic interaction dominates the energies in
the magnetic subsystem. It was found that in ErRh4B4 (Fertig et al. 1977) and HoM06Ss
(Ishikawa and Fischer 1977) superconductivity is in competition with ferromagnetic long-
range order, which results, in a reentrant behaviour and in coexistence of superconductivity
with oscillatory magnetic states (Thomlinson et al. 1982; Lynn et al, 1984). For most of
the superconducting RM06SS and RRh4B4 compounds the magnetic interactions favor
antiferromagnetic order with a magnetic unit cell on a length scale small compared to
~ and A which results in a relatively weak influence on the superconducting state i.e.
antiferromagnetic order and superconductivity do readily accommodate one another. The
antiferromagnetic transition in these materials has been confirmed by neutron scattering
(see Thomlinson et al. 1982). Initially this transition had been observed as an anomaly in
the upper critical field (Ishikawa et al. 1982). In particular, a near-reentrant behaviour has
been found for some of the antiferromagnetic ternary compounds i.e. reentrant behaviour
occurs if a sufficiently high magnetic field is applied, as shown in fig. 6 for GdM06SS.
To explain the near-reentrant behaviour it is usually argued (see Maple and Fischer 1982)
that, in the vicinity of the antiferromagnetic ordering temperature TN, the applied field
induces a remarkable degree of ferromagnetic order which has been confirmed for various
compounds.
In the case of GdM06Ss, additionally, large spin fluctuations below TN have been
assumed to enhance the near-reentrant behaviour (Ishikawa et al. 1982). Machida et al.
(1980b) extended the theory of antiferromagnetic superconductors (Baltensperger and
Strassler 1963), taking into account effects of the antiferromagnetic molecular field caused
by aligned local magnetic moments in addition to spin fluctuations. They found that the
anomalies of Ha in RM06SS (R = Gd, Tb, Dy) can be explained by the formation
of energy gaps of spin density waves on the Fermi surface. Morozov (1980) as well as
Zwicknagl and Fulde (1981) integrated the concept of Baltensperger and Strassler (1963)
into the Eliashberg theory and they found that the influence of the antiferromagnetic
staggered magnetization on the phonon-mediated quasiparticle attraction also results in
an anomaly, in particular a reduction below TN, of H eZ(T).
The cuprates RBazCu307-,s with the orthorhombic (nearly tetragonal) Rl23-type
structure exist for R = Y and all 4f elements with the exception of Ce and Tb, For
0<8 < 0.6 they are cuprate-mixed-valence high-Z', superconductors, with the exception
ZIZ K.-H. MULLER et al.

1.0
GdMo~8
0.8

as 0.6

IX 0.4

0.2

0.5 1.0 1.5


Temperature (K)

Fig. 6. Resistance-vs-temperature curves of a GdMo6Sg sample for different values of the applied magnetic
field. indicating near-reentrant superconductivity i.e. reentrant behaviour at finite field only (nominal composition
Gd1.2Mo6Sg; after Ishikawa et aI. 198Z).

of R = Pro The value of T c is about 90 K and it practically does not depend on the choice
of R. GdBazCu307 shows three-dimensional antiferromagnetic ordering with TN ~ 2.2 K
and a staggered magnetic moment of 7.4JlB which is close to the Hund's rule Gd 3+ free
ion value (Paul et al. 1988). Since TN does not much change if [, is increased from 0
to I and the material becomes a semiconductor with antiferromagnetic ordering of the
Cu 2+ magnetic moments (Dunlap et al. 1988) the Gd magnetic order cannot be dominantly
governed by indirect exchange via conduction electrons (RKKY interaction) and the two
antiferromagnetic structures on the R and the Cu sublattices are only weakly coupled to
each other. On the other hand the value of TN ~ 2.2 K is too high to be explained by
dipolar interactions only. Thus the type of magnetic coupling of the R magnetic moments
is not yet fully understood. For R = Nd, Srn, Dy, Er, Yb the single-R 3+-ion crystal field
splitting results in magnetic (doublet) ground states and the RBa2Cu307 compounds with
these R elements show antiferromagnetic ordering with TN ~ I K. For R = Dy and Er
the R magnetism (as well as the Cu magnetism) is two-dimensional (Lynn 1992). For
R = Ho the crystal field ground state in the R 123 structure is a singlet. Nevertheless,
antiferromagnetic ordering (TN = 0.17 K) has been observed also for this compound and
the Ho magnetic moments have assumed to be induced in the electronic singlet ground
state by nuclear hyperfine interaction (Dunlap et al. 1987). In these R 123 superconductors
the superconductivity persists below TN. Hence there is no measurable effect of the
ordered magnetic moments on superconductivity. This supports that exchange interaction
between the conduction electrons and the rare-earth magnetic moments is minor and pair-
breaking due to exchange scattering is weak. On the other hand, the relatively high value
of TN ~ 2.2 K (for Gd123) suggests that some small indirect exchange between the rare-
earth magnetic moments operates across the CU02 layers (Fischer 1990). The situation
is totally different for Prl23 where antiferromagnetic order of the Pr magnetic moments
develops at T N ~ 17 K and superconductivity does not occur. The superconductivity in
Prl23 which has been recently reported by Zou et al. (1998) is not yet understood or
even finally secured. It may be connected with a modified composition of the samples as
well as a modified occupation of the lattice sites by Pr and Ba ions (see Narozhnyi and
MAGNETIC AND SUPERCONDUCTING PROPERTIES OF RNi2B2C 213

Drechsler 1999). The anomalous behaviour of Prl23 has been attributed to hybridization
of Pr 4f states with 0 2p states, completely disrupting the quasiparticles which form the
Cooper pairs in the CU02 planes and dramatically increasing the exchange interactions
between the Pr magnetic moments (Fehrenbacher and Rice 1993; Lynn 1997; Skathakumar
et al. 1997). A further consequence of the hybridization of the Pr 4f electrons, besides the
enhanced value of TN and the absence of superconductivity, is a considerable interaction
of the Cu magnetic subsystem with the Pr subsystem: contrary to the behaviour of the
RI23 materials mentioned above Prl23 shows Cu antiferromagnetism in the whole range
of 0 = 0 ... 1 with an ordering temperature TN[CU] of about 300 K (instead of TN[CU]
~ 410 K for YI23 at 0 ~ 0.9). Furthermore, below TN ~ 17 ... 20 K an incommensurate
magnetic structure develops involving both the Pr and the Cu moments where the Cu
moments are found to be non-collinear (see Boothroyd 2000).
Coexistence of superconductivity and magnetic order has also been reported for
ruthenocuprates with typical composition RuSr2RCu20g or RuSr2(R,CehCu201O, with
R = Sm, Eu or Gd, where the magnetic-ordering temperature TN = 100 ... 180 K is
much higher than T; = 15 .. .40 K (Bauernfeind et al. 1995; Braun 2001). Neutron
diffraction experiments (Lynn et al. 2001) have shown that TN is related to basically
antiferromagnetically ordered magnetic moments in the Ru sublattices and, in the case
of RuSr2GdCu20g, the Gd moments order independently antiferromagnetically at 2.5 K.
The small ferromagnetic component reported for these materials for temperatures below
TN is attributed to spin canting resulting in weak ferromagnetism of Dzyaloshinsky-Moria
type and to explain the coexistence of this type of magnetism with superconductivity it has
been assumed that the magnetically ordered Ru sublattice is practically decoupled from the
superconducting CU02 planes (Bernhard et al. 1999; FeIner 1998, FeIner et al. 1999).
In the Heusler alloy ErPd2Sn superconductivity and antiferromagnetic order coexist
although there is no clear separation between the superconducting and the magnetic
sublattices and T c ~ 1.17 K is not much different from T N ~ I K (Shelton et al. 1986;
Stanley et al. 1987). However, the focus on this interesting compound was short lived
because of the discovery of the high T c cuprate superconductors (Lynn 200 I).
An interesting theoretical prediction is that, similar as in the p-wave superconductors
discussed in the next subsection, non-magnetic impurities in an antiferromagnetic
superconductor cause pair breaking (Morozov 1980; Zwicknagl and Fulde 1981) whereas
non-magnetic impurities in a non-magnetic superconductor are not expected to destroy
superconductivity (Anderson 1959).

1.3.4. Superconductivity and itinerant-electron magnetism


Fay and Appel (1980) predicted unconventional superconductivity (p-wave paring, i.e.
spin triplet pairing) mediated by longitudinal spin fluctuations to coexist with itinerant
ferromagnetism if the magnetization is small enough. These authors also declared ZrZn2
as a candidate for this phenomenon. For reasons of time-reversal symmetry, in p-wave
superconductors all impurities are pair breakers (Foulkes and Gyorffy 1997) and, therefore,
superconductivity will be observed only in very clean samples. This behaviour is different
from that of BCS (s-wave) superconductors where nonmagnetic impurities do not destroy
superconductivity (Anderson 1959). Matthias and Bozorth (1958) had found that ZrZn2 is
ferromagnetic although both elements, Zr and Zn, are non-ferromagnetic. These authors
214 K.-H. MULLER et aI.

60

r*_
~ ... TC UGe2
• ....
g 40 ~"-
"-
!! ,.
::J
~ Ferromagnetism ,.
8- \

.,,
E \.
{!! 20 Superconductivity
10 TSC 'I
\
I

00 1 2
Pressure (GPa)

Fig. 7. The temperature-pressure phase diagram of UGe2 (Saxena et al. 2000). TC is the Curie temperature and
Tee the superconducting transition temperature (normally denoted Tel.

also were the first who suggested that ZrZn2 could be a superconductor. Wohlfarth (1968)
showed that ZrZn2 is a weak itinerant d-electron ferromagnet. Now the superconductivity
in ZrZn2 has been confirmed for very pure samples (Pfleiderer et al. 200 1). At ambient
pressure the Curie temperature T c and the critical temperature Teare 28.5 K and 0.29 K,
respectively. Under hydrostatic pressure T c and T c disappear at the same pressure.
Therefore the ferromagnetic state is assumed to be a prerequisite for superconductivity
and the above mentioned p-wave pairing mechanism is assumed to be valid for this
material. Superconductivity coexisting with weak itinerant ferromagnetism has also been
reported for UGe2 (Saxena et al. 2000) and URhGe (Aoki et al. 2001) and it has been
assumed to be based on the same magnetic p-wave pairing mechanism as in ZrZn2 (Huxley
et al. 2001). In UGe2 the superconductivity is pressure induced and, as in the case of
ZrZn2, it disappears at the same pressure as the ferromagnetism (see fig. 7). However,
the magnetic moments are expected to be more localized in UGe2 than in ZrZn2 because
they are due to f -electrons rather than d -electrons. Therefore Suhl (200 I) and Abrikosov
(200 1) developed an alternative pairing model based on interaction of the conduction
electrons with ferromagnetically ordered localized spins which can only lead to an s-wave
order parameter. This concept is supported by experiments of Bauer et al. (2001) who
showed that, different from the case of ZrZn2. high-purity specimens with long mean free
paths are not necessary in UGe2 in order to observe superconductivity near the critical
pressure where the magnetic ordering temperature vanishes. Furthermore, for the s-wave
superconductivity not to be destroyed by magnetism the metal has to be a heavy-fermion
type which also is supported by the experimental results of Bauer et al. (200 I).
Superconductivity has also been found to coexist and compete with itinerant-electron
antiferromagnetism (spin density waves) which has extensively reviewed by Gabovich
et al. (2001).
In summary, there are various forms of the interplay of magnetism and superconductiv-
ity, which can be divided into competition and coexistence phenomena. In the elements C
MAGNETIC AND SUPERCONDUCfING PROPERTIES OF RNi2B2C 215

and Fe different types of crystal structure and bonding between the atoms, both varied
by preparation routes or thermodynamic parameters such as pressure, result in one of the
antagonistic cooperative phenomena ferromagnetism or superconductivity. Strong compe-
tition is found in high- T c cuprates where, depending on the doping rate, Neel-type antifer-
romagnetism (or spin glass) or superconductivity occur, both based on copper d-electrons.
Coexistence of localized magnetic moments (e.g. from 4 f -elements) with superconductiv-
ity is known for systems where the concentration of these moments is small enough or they
are antiferromagnetically ordered and only weakly coupled to the conduction electrons.
Even weak ferromagnetism of such localized moments can coexist with superconductiv-
ity.In RuSr2GdCu20g and (R,Ce)RuSr2Cu201O-8, probably, the Ru subsystem with weak
ferromagnetism of Dzyaloshinsky-Moriya type is weakly coupled to and coexists with su-
perconducting CU02 layers. Most surprising is the recently reported coexistence of weak
itinerant ferromagnetism with superconductivity based on d or f electrons in ZrZn2, UGe2
and URhGe.

1.4. Specific features ofthe RNhB2C compounds


A striking feature distinguishing the superconducting RTBC from other superconductors
known until 1994 is that for certain combinations of elements Rand T superconductivity
and antiferromagnetic order have been found to coexist in RT2B2C with the values of the
magnetic ordering temperature TN being comparable with the T c values (see fig. 8) i.e. the
magnetic energy is comparable with the superconducting condensation energy. Therefore
the investigation of these compounds is expected to result in new insights into the interplay
of superconductivity and magnetism. In addition to many specific studies in this field,
published so far, there are various reports and review articles summarizing experimental
and theoretical results on the superconducting and magnetic properties of these materials
and comparing them with other superconductors, as e.g. Maple (1995), Canfield et a1.
(1997b), Lynn (1997), Takagi et a1. (1997), Andreone et a1. (1998), Canfield et al. (1998),

LuTmEr Ho Dy Tb Gd
20 /
,.'.'
:: 15 ~/ "
,.,-
,/TN
~ .".
" 10 ,.,
c • ,.
IV ,
~5 .,~ •
.~.,/,.
0&;...--.......- -.....- -....
o 5 210 15
=
DG (g -1) J (J + 1)

Fig. g. Critical temperatures for superconductivity, T c. and for antiferromagnetic ordering, TN, for RNi2B2C
compounds with R = Lu, Trn, Er, Ho, Dy, Tb and Gd. DO is the de Gennes factor, g the Lan~ factor and
J the total angular momentum of the R3+ Hood's rule ground state. The straight lines represent rough linear
approximations.
216 K.-H. MULLER et al.

FeIner (1998), Gupta (1998), Nagarajan and Gupta (1998), Paranthaman and Chakoumakos
(1998), Schmidt and Braun (1998), Hilscher and Michor (1999), Naugle et al. (1999),
Drechsler et al. (1999), Schmiedeshoff et al. (2000), Tominez et al. (2000), Drechsler
et al. (2oo1b), Muller and Narozhnyi (2001a). Also articles in this field are collected in
the Proceedings of the NATO Workshop "Rare Earth Transition Metal Borocarbides",
held in Dresden, Germany in June 2000 (Muller and Narozhnyi 2oo1b). The high values
of TN demand that in quaternary borocarbides, different from the situation in high- T c
cuprates and the classical magnetic superconductors, exchange coupling between the rare-
earth magnetic moments is the dominant magnetic interaction rather than magnetostatic
interaction. Obviously the exchange is mediated by conduction electrons. Consequently
also the interaction between the magnetic moments and the conduction electrons must be
relatively strong. Figure 8 shows a linear scaling of TN and, roughly approximated, also of
T c with the de Gennes factor

(2)

of the R 3+ Hunds' rule ground state where g is the Lande factor and J the total angular
momentum (de Gennes 1958). Such so called de Gennes scaling, at the same time for
TN and T c- is known for various isostructural metallic R compounds, which is due to
the fact that both effects, antiferromagnetism and the suppression of superconductivity are
governed by exchange interaction of conduction electrons with R 4 f electrons. In some
approximation both, TN and the difference I:i. T c of the critical temperature compared to
that of a nonmagnetic (DG = 0) reference material can be written as

(3)

where I is the strength of the exchange interaction between 4 f electrons and the conduc-
tion electrons and N(EF) is the density of states at the Fermi level (Fischer 1990). From
fig. 8 it can be seen that both cases, TN < T« (R = Tm, Er, Ho) and TN > T c (R = Dy)
occur in the series RNhB2C. A similar phase diagram as that in fig. 8 had been predicted
by Machida et al. (1980a). As can be seen in fig. 9, de Gennes scaling does not work if a
larger class of materials is considered within the series RNi2B2C. (As will be discussed
in subsection 2.1, two lattice structures have been reported for ScNi2B2C. Therefore the
position of Sc in fig. 9 is questionable and, in reality, the dashed line may be monotonous.)
In particular there is a strong influence of the lattice parameters on T c which cannot be
explained by only taking into account the variation of N(EF) in the expression (3). Obvi-
ously, T c also very much depends on nonmagnetic effects (see subsection 3.1).
Useful means for the investigation of both superconducting and magnetic subsystems
are measurements under high hydrostatic pressure P. The first results on the influence of
P on T; of RNi2B2C (R = Y, Lu, Tm, Er and Ho) were reported by Schmidt and Braun
(1994). Other groups report on results of high pressure studies also on other borocarbides
(Gao et al. 1994; Murayama et al. 1994; Alleno et al. 1995b; Looney et al. 1995; Carter
et al. 1995; Uwatoko et al. 1996; Bud'ko et al. 1996; Weht et al. 1996; Meenakshi et al.
1996, 1998; Jaenicke-Roessler et al. 1999, Cappannini et al. 1998; Oomi et al. 1999,2001;
Murdoch et al. 1999; Matsuda et al. 2001; Dertinger 200 1; Falconi et al. 200 1). High-
pressure studies on YNjzB2C at room temperature do not indicate any structural transition
MAGNETIC AND SUPERCONDUCfING PROPERTIES OF RNi282C 217

Lu
~>---.",y
~ 15 (LuY)- " ,,
,,
,,
\
\

'.Th
\
ADy \
\
\
\
\
Sm Nd '-
Yb d Ce \ La
O'--_L.--.1JL.-oI..1IL.Z.lI~I1.....""""'~
QD Q~ Q~ Q~ QTI Q ~
Lattice parameter a (nm)

Fig. 9. Transition temperature and lattice parameter a for RNi282C compounds with LuNi282C type structure
(see subsection 2.1) for non-magnetic (Lai et aI. 1995) and magnetic R elements. The curves are guides to the
eye. The (Lu, Y)-value is taken from Freudenberger et aI. (1998a).

up to P = 16 GPa (Meenakshi et al. 1996, 1998). The bulk modulus is measured to be


200 GPa and estimated to be 270 GPa from calculations based on the TB-LMTO method
(Meenakshi et al. 1996). Values of the bulk modulus close to those mentioned above have
been obtained for LuNi2B2C (210 GPa) within the local-density approximation (Weht et al.
1996). The results on the influence of P on T c in these materials are contradictionary. For
example, Schmidt and Braun (1994) found that Tc(P) decreases linearly with pressure at
rates of -0.058 KlGPa. This is in agreement with the results of Murayama et al. (1994), but
is in contradiction with the data of Alleno et al. (1995b) who found a Tc(P) dependence
with positive initial slope of +0.03 KlGPa and with a peak centered at P ~ 0.52 GPa.
These observations indicate that even the sign of the pressure dependence of T c depends
on the microstructure of the samples (Alleno et al. 1995b). Contrary to the behaviour of T c-
the magnetic ordering temperature TN has found to increase with increasing pressure for
all RNhB2C compounds investigated so far, i.e. for R = Gd (Bud'ko et al. 1996), Ho
(Carter et al. 1995; Uwatoko et al. 1996; Dertinger 200 1) and Er (Matsuda et al. 200 1).
The effect of pressure on the interplay between superconductivity and magnetism in
borocarbides has been discussed for R = Tm (Oomi et al. 1999), Er (Matsuda et al. 2(01)
and Ho (Uwatoko et al. 1996; Carter et al. 1995; Oomi et al. 2001; Dertinger 2(01). It
was pointed out that the much stronger suppression of H c2 under high pressure observed
for ErNhB2C and TrnNi2B2C, compared to RNhB2C compounds with non-magnetic R,
may be connected with some instability of the superconducting state of these magnetic
superconductors. Most extensively the influence of pressure has been investigated for
HoNi2B2C (see also subsections 4.9.3 and 4.9.4).
In this article we will report on the current status of research on the quaternary
borocarbide superconductors starting from their discovery. We will concentrate on the
magnetic and superconducting properties of RNhB2C compounds. Section 2 is devoted
to the typical crystal structure of RNi2B2C and lattice distortions caused by magnetic
ordering but also to other compounds and lattice structures which are related to RNi2B2C.
Z18 K.-H. MULLER et al.

Section 3 briefly summarizes electronic and superconducting properties of RNi2BZC


compounds with nonmagnetic R elements. Special features are Fermi surface nesting
characterized by the nesting wave-vector (0.55,0,0) and phonon softening at the same
wave vector (subsection 3.1), and the positive curvature of the upper critical field as
a function of temperature, He2(T), discussed in subsection 3.2. RNjzB2C compounds
with 4 f elements R are considered in section 4. Among them, Ce and Yb are
interesting because, in RNi2BzC, they show intermediate 4f-valence and heavy fermion
behaviour, respectively. DyNjzB2C is outstanding because it is one of the exceptional
antiferromagnetic superconductors with TN > T e . In HoNjzB2C three different types of
magnetic order occur and the competition between superconductivity and magnetism is
most complex. An exciting feature of ErNjzB2C is the coexistence of superconductivity
with some kind of weak ferromagnetism. Results on flux line lattices in the borocarbides,
including the transformation from hexagonal to square lattices, are presented in section 5.
The investigation of pseudoquaternary compounds, reported in section 6, provides some
more insight into the pair-breaking mechanisms in the borocarbides. A short summary and
conclusions are presented in section 7.

2. Crystal structure and chemical composition

2.1. The LuN;zB2C-type structure

With the investigation of superconducting rare-earth transition-metal borocarbides the


new LuNi2B2C-type structure, space group I4/mmm, has been discovered which can be
considered as the ThCr2Si2-type, space group I4/mmm, interstitially modified by carbon
(Siegrist et al. 1994a, 1994b). Figure 10 shows the nonmodified and the modified structures
with Th -. Gd, Cr -. Co, Si -. Band Lu -. Gd, Ni -. Co, respectively. The family of
ternary rare-earth transition-metal metalloid compounds with the ThCr2Si2-type structure
is very large (Just and Paufler 1996) and a broad variety of magnetic and electronic
properties have been observed in it. For example in SmMn2Ge2 both Sm and Mn carry a
magnetic moment and two metamagnetic transitions occur connected with giant magneto-
resistance effects (Brabers et al. 1993). Different collective phenomena as heavy-fermion
behaviour, superconductivity and magnetic order have been found in the exotic compound
CeCu2Si2 (Steglich et al. 1995). The LuNi2B2C-type structure has three open parameters,
the two lattice constants a and c and the coordinate z of the boron atom. It has been
pointed out by Godart et al. (1997) that the values of a and c of RNjzB2C compounds
show a certain dispersion indicating a domain of existence which is in agreement with
the variety of physical properties observed in many individual cases. The structure of the
RNjzB2C compounds is highly anisotropic with a ratio cia of about 3. It has alternating
sheets of Ni2B2 tetrahedra and RC layers. In a good approximation, the parameters c and
z linearly decrease with increasing radius of R (where R is assumed to be in the trivalent
oxidation state) whereas a linearly increases with the radius of R, with the exception of Ce
(see subsection 4.2). Thus while going through the series of R elements from Lu to La, the
structure shows a contraction along the tetragonal c-axis but an expansion perpendicular
to it i.e. a decrease of the anisotropy characterized by cia and the boron shifts away from
the RC-Iayers more in the vicinity of the Ni layers. However, the radius variation of the
MAGNETIC AND SUPERCONDUCTING PROPERTIES OF RNi2B2C 219

(8) (b)

Fig. 10. (a) GdC02B2 has the ThCr2Si2-type structure, where Gd resides on the Th, Co on the Cr and B on
the Si sites, respectively. (b) GdC02B2C has the LuNi2B2C-type structure i.e, the ThCr2Si2-type interstitally
modified with C atoms. The lattice constants are a = 3.575 Aand c = 9.561 Afor GdC02B2 (Feiner et aI. 1984),
a = 3.548 Aand c = 10.271 Afor GdC02B2C (Mulder et aI. 1995). respectively.

rare earth does not much affect the B-C distance and the B-Ni distance. Consequently,
there is a remarkable reduction of the B-Ni-B tetrahedral angle from 108.8° for Lu to 102°
for La which is expected to influence the variation of the electronic structure within the
series of RNi2B2C compounds. The Ni-Ni distance in LuNhB2C (2.449 A) is smaller
than that in metallic Ni (2.492 A) and underlines the metallic character of the RNizB2C
compounds.
Table 2 shows the known RT2B2C compounds (R: Sc, Y, La, Th, or 4f or 5f el-
ements; T : 3d. 4d or 5d elements). Table 3 contains the superconducting compounds
from table 2 as well as their superconducting transition temperatures T c and, if existing,
magnetic ordering temperatures TN. Superconductivity in CeNizB2C has been reported
(El Massalarni et al. 1998a) and is exceptional in that this is the only superconducting
RNizB2C compound with a light rare-earth R and Ce is in a mixed-valence state (see
subsection 4.2).
Single crystals on which most of the physical investigation has been performed have
been grown by the flux method (Xu et al. 1994) and it has also been proven to be possible
to grow single crystals by floating zone (Takeya et al. 1996) and zone melting methods
(Behr et al. 1999a). Various melting techniques have been used to prepare polycrystalline
materials (see e.g. Torninez et al. 2000). c-axis aligned or even epitaxial RNizB2C thin
films have been successfully prepared by pulsed laser deposition (Cimberle et al. 1997;
Hase et al. 1997, 2(01) as well as magnetron sputtering technique (Arisawa et al. 1994;
Andreone et al. 1996). Non-equilibrium methods as rapid quenching (Strom et al. 1996;
Freudenberger 2(00) or mechanical alloying (GUmbel et al. 2(00) have been utilized to
search for metastable phases. In some systems metastable phases form also in normal
melting procedures. For example, ScNizB2C (Ku et al. 1994; Tomilo et al. 2001, 1999)
and ThNhB2C (Sarrao et al. 1994; Hossain et al. 1994b; Zandbergen et al. 1994b) have
been found to be metastable. In the case of ScNizB2C two different tetragonal lattice
=
structures have been reported with the lattice parameters a 0.332 nm, c 1.004 nm =
220 K.-H. MULLER et al.

TABLE 2
Known R-T -B-C compounds with the LuNi2B2C-type structure. Compounds printed in bold face are
superconductors

CeCo2B2C GdNi2B2C LuC02B2C SmRh2B2C YCo2B2C


CeNi2B2C GdRh2B2C LuNi2B2C TbNi2B2C YNi2B2C
CePt2B2C HoC02B2C NdNi2B2C TbRh2B2C YPd2B2C
CeRh2B2C HoNi2B2C NdPt2B2C TbNi2B2C YPt2B2C
DyNi2 B2C HoRh2B2C NdRh2B2C ThPd2B2C YRu2B2C
DyPt2 B2C LaIr2B2C PrNi282C TbPt2B2C YbNi282C
DyRh2B2C LaNi2B2C PrPt2B2C ThRh2B2C
ErNi2B2C LaPd2B2C PrRh2B2C TmNi2B2C
ErRh2B2C LaPt2B2C ScNi2B2C UNi2B2C
GdC02B2 C LaRh2B2C SmNi2B2C URh2B2C

TABLE 3
8orocarbide superconductors with LuNi2B2C-type structure and their superconducting transition temperature T c
and magnetic ordering temperature TN

Compound Tc<K) TN(K) Compound Tc<K) TN(K)

CeNi2B2C 0.1 [IJ YRu2B2C 9.7 [20J


DyNi2B2C 6.2 [2J, 6.4 [3J II [2,18J LaPd2B2C 1.8 [21)
HoNi2B2C 8 [4J, 7.5 [5J 5 ... 8 [8,9,IOJ ThPd2B2C 14.5 [15J
ErNi282C 10.5 [4,5J 6.8 [11,12J YPd2B2C 23 [4,16,19J
TmNi2B2C II [4,5J 1.5 [13,14J LaPt2B2C 10 ... 11 [l7,22J
LuNi2B2C 16.5 [4,5J PrPt2B2C 6 {17,22J
YNi2B2C 15.5 (4] YPt2B2C 10 .. _11 [l7,22J
ScNi2B2C 15 [6J ThPt2B2C 6.5 [I5J
ThNi2B2C 8 [7J
[I) EI Massalami et al. (1998a), [2J Cho et al. (1995a), [3J Tomy et al. (1995), [4J Cava et al. (1994b), [5J Eisaki
et al. (1994), [6J Ku et al. (1994), [7J Lai et al. (1995), [8] Grigereit et al. (1994), [9J Goldman et al. (1994),
[10] Canfield et al. (1994), [I I) Sinha et al. (1995), [12] Zarestki et al. (1995), [13] Cho et al. (1995b), [14] Lynn
et al. (1997), [15] Sarrao et al. (1994), [16] Tominez et al. (1998), [17] Cava et al. (l994d), [18] Dervenagas
et al. (1995a), {19] Dezaneti et al. (2000), {20] Hsu et al. (1998), {21J Jiang et al. (1995), {22] Buchgeister et al.
(1995).

and a = 0.354 nm, C = 1.055 nm where only the latter phase has been found to be
superconducting (Tomilo et al. 2(01). Therefore the dashed curve in fig. 9. in reality. may
be monotonic because the lattice constant a of superconducting ScNi2B2C should be close
to that of YNbB2C. Also it was found that some of the RNi2B2C compounds are rather
stable. Thus YNi2B2C starts oxidizing and decomposing only above 850°C (Buchgeister
and Pitschke 1996).
Neutron diffraction data seem to provide evidence that all the crystallographic sites
in RNbB2C are fully occupied and there is no site mixing (Tominez et al. 2000).
However. the conventional diffraction techniques may be not enough sensitive to determine
interchange or defects on the Band C sublattices in these compounds. Such effects may
be one of the reasons why superconducting and magnetic properties of the borocarbide
superconductors do very much depend on the metallurgical state of the samples (Lynn
et al. 200l).
MAGNETIC AND SUPERCONDUCfING PROPERTIES OF RNi2B2C 221

TABLE 4
Structural and magnetic properties of RC02B2 and RC02B2C phases with the ThCr2Si2 and LuNi