You are on page 1of 172

A study of exchange interaction,

magnetic anisotropies, and ion beam


induced effects in thin films of
Co2-based Heusler compounds

Oksana Gaier

Vom Fachbereich Physik der Technischen Universitat Kaiserslautern


zur Verleihung des akademischen Grades
Doktor der Naturwissenschaften genehmigte Dissertation

Betreuer: Prof. Dr. B. Hillebrands


Zweitgutachter: Prof. Dr. C. Felser

Datum der wissenschaftlichen Aussprache: 15.07.2009


D 386

Abstract
This thesis is devoted to the study of the exchange interaction and magnetic
anisotropies in thin films of Co2 -based Heusler compounds with the chemical
composition Co2 YZ. The strength of the exchange interaction plays an important role in the temperature dependence of the spin polarization of Heusler compounds, which is a crucial point for the performance of the devices incorporating
these materials. Magnetic anisotropies strongly influence the magnetic domain
configuration and the magnetization reversal processes, and additionally provide insight into the spin-orbit interaction. Spin-orbit coupling is an important
parameter for applications of Heusler compounds in devices based on the spin
degree of freedom.
The studies of exchange interaction and magnetic anisotropies in Heusler
compounds are carried out by Brillouin light scattering spectroscopy and magneto-optical Kerr effect magnetometry. Particular attention is given to the dependence of exchange and magnetic anisotropies on the chemical composition
and atomic ordering of the Heusler films. The influence of the chemical composition on exchange and anisotropies is studied for compounds wherein Y=Mn,Fe
and Z=Al,Si, as well as the quaternary compound Co2 Cr0.6 Fe0.4 Al. The influence of the structural order is investigated for the Co2 MnSi films exhibiting a
varying degree of the L21 order. The gradual variation of the L21 order in the
Co2 MnSi films is achieved by annealing the samples at different temperatures
after their deposition.
We observe that a larger ordering degree in the investigated Heusler compounds is related to a stronger exchange interaction, which is evidenced by
larger values of the exchange stiffness D found in the L21 -ordered films compared to those with a less ordered B2 structure. Moreover, we observe a linear
increase of the exchange stiffness D upon increasing number of valence electrons
in the unit cell evoked by the change of composition (e.g. replacement of Mn
by Fe). The highest D value is found for the Co2 FeSi Heusler compound. This
value is nearly as large as the highest D value reported up to this day which
is found for Fe1x Cox alloys at a composition of x 0.5. Furthermore, most
of the investigated films are found to exhibit a four-fold magneto-crystalline

anisotropy, which becomes weaker with increasing ordering degree in the films.
In L21 -ordered films, the magneto-crystalline anisotropy is even found to be negligible, pointing to a vanishing spin-orbit coupling in the more ordered Heusler
compounds. This is a great advantage for possible applications based on the
transport of spins since the information carried by spins cannot be lost by
coupling to the orbital motion.
Given the strong dependence of the exchange interaction and magnetic
anisotropies on the structural order, this thesis also explores the structural
modification of Heusler films using ion irradiation. The irradiation experiments
performed in this thesis utilize both light 30 and 130 keV He+ and heavy 30 keV
Ga+ ions. Experiments with He+ are carried out for Co2 MnSi films with an
initially predominant B2 order, with and without a simultaneous mild anneal.
Ga+ irradiation is performed for L21 -ordered Co2 FeSi films at room temperature only. For the study of the influence of ion irradiation on the structural,
magnetic and electronic properties of Heusler films various techniques are employed including x-ray diffraction, Brillouin light scattering spectroscopy and
magneto-optical Kerr effect magnetometry. In some cases additional investigations by means of superconducting quantum interference device magnetometry,
photoemsission at high energies, and x-ray absorption and circular magnetic
dichroism are also carried out.
Our results show that the crystallinity of the Co2 MnSi films remains largely
unaffected by the bombardment with He+ ions in the whole range of applied
fluences. Moreover, at particular fluences, we observe an improvement of magnetic and electronic properites of the Co2 MnSi layer towards the bulk material
after the room temperature irradiation. However, no evidence of the transition
to the L21 phase is found. For Ga+ irradiation experiments on Co2 FeSi films, we
find a strong destructive impact of the ion beam on the magnetic properties of
Co2 FeSi. For larger fluences, we even observe a breakdown of the ferromagnetic
order.

Kurzfassung
Die vorliegende Arbeit beschaftigt sich mit der Untersuchung der Austauschwechselwirkung und der magnetischen Anisotropien in d
unnen Schichten aus
Co2 -basierten Heusler-Verbindungen mit einer chemischen Zusammensetzung
Co2 YZ. Die Starke der Austauschwechselwirkung spielt eine wichtige Rolle f
ur
das Verstehen der Temperaturabhangigkeit der Spinpolarisation der HeuslerVerbindungen. Dieses Verstandnis ist ein entscheidender Punkt f
ur die Effizienz von Bauelementen, welche diese Materialien integrieren. Die magnetischen Anisotropien beeinflussen im groen Mae die magnetische Mikrostruktur und den Ummagnetisierungsprozess, und stellen auerdem eine direkte
Verbindung zu der Spin-Bahn-Wechselwirkung dar. Die Spin-Bahn-Kopplung
ist ein wichtiger Parameter f
ur technologische Anwendungen, welche auf dem
Spin-Freiheitsgrad basieren.
Die Untersuchungen der Austauschwechselwirkung und der magnetischen
Anisotropien in d
unnen Heusler-Schichten werden mittels Brillouin-Lichstreuspektroskopie sowie magneto-optischer Kerr-Effekt-Magnetometrie durchgef
uhrt. Ein besonderes Augenmerk wird dabei auf die Abhangigkeit dieser magnetischen Eigenschaften von der chemischen Zusammensetzung der HeuslerSchichten und von ihrer kristallographischen Ordnung gelegt. Der Einflu der
chemischen Zusammensetzung auf die Austauschkonstanten und auf die magnetischen Anisotropien wird an einer Serie von Verbindungen untersucht, in
welcher Y=Mn,Fe und Z=Al,Si sind, und welche auch die quaternare Verbindung Co2 Cr0.6 Fe0.4 Al einschliet. Der Einflu der kristallographischen Ordnung
wird an Co2 MnSi-Schichten untersucht, deren L21 -Ordnungsgrad graduell zunimmt. Die Variation des Ordnungsgrades wird dabei durch die Variation der
Temperatur erreicht, bei welcher die Schichten nach ihrer Deposition getempert
werden.
Es zeigt sich, dass ein hoherer Ordnungsgrad der Heusler-Verbindungen mit
einer starkeren Austauschwechselwirkung korreliert. Die experimentell ermittelte Spin-Steifigkeit D ist bei den L21 -geordneten Verbindungen groer als
bei solchen, die eine weniger geordnete B2-Struktur aufweisen. Auerdem,
wird eine lineare Zunahme von D mit zunehmender Anzahl der Valenzelek-

tronen in der Einheitszelle beobachtet, wobei die Zunahme der Valenzelektro


nenzahl durch die Anderung
der chemischen Zusammensetzung erzeugt wird
(z. B. Ersatz von Mangan durch Eisen). Die hochste Spin-Steifigkeit wird bei
Co2 FeSi beobachtet. Sie ist vergleichbar mit dem hochsten bis heute berichteten
D-Wert, welcher in Kobalt/Eisen-Legierungen beobachtet wurde. Dar
uber
hinaus weisen die meisten untersuchten Heusler-Schichten eine vierfache magnetokristalline Anisotropie auf, welche mit der sich verbessernden kristallographischen Ordnung schwacher wird. In den Schichten mit der L21 -Ordnung ist
sie sogar vernachlassigbar, was auf eine verschwindende Spin-Bahn-Kopplung
hindeutet. Dies ist von einem groen Vorteil f
ur mogliche Anwendungen, die auf
dem Transport von Spins basieren, da bei einer geringen Spin-Bahn-Kopplung
die Information, welche die Spins tragen, nicht verloren gehen kann.
In Anbetracht der starken Abhangigkeit der Austauschwechselwirkung und
der magnetischen Anisotropien von der kristallographischen Ordnung, wird
im Rahmen der vorliegenden Arbeit zusatzlich die Modifikation der HeuslerSchichten durch Ionenbestrahlung untersucht. Die Bestrahlungsexperimente
werden mit 30 und 130 keV He+ - sowie mit 30 keV Ga+ -Ionen durchgef
uhrt.
+
Dabei erfolgt die Bestrahlung mit He -Ionen an Co2 MnSi-Schichten, die nach
ihrer Herstellung grotenteils eine B2-Ordnung aufwiesen. Die Bestrahlung
wird sowohl bei Raumtemperatur als auch bei erhohten Temperaturen durchgef
uhrt. Die Bestrahlung mit Ga+ -Ionen erfolgt an L21 -geordneten Co2 FeSi Schichten. Der Einflu der Ionenbestrahlung auf die strukturellen, magnetischen und elektronischen Eigenschaften der Heusler-Verbindungen wird mittels Rontgenbeugung, Brillouin-Lichtstreuspektroskopie sowie magneto-optischer Kerr-Effekt-Magnetometrie untersucht. In einigen Fallen werden auch
Photoemission bei hohen Energien, SQUID-Magnetometrie, Rontgenabsorption
und magnetischer Zirkulardichroismus verwendet.
Die Untersuchungen der mit He+ -Ionen bestrahlten Co2 MnSi -Schichten
zeigen, dass ihr kristalliner Zustand im gesamten Bereich der gewahlten Ionenfluenzen nach der Bestrahlung grotenteils erhalten bleibt. Auerdem wird
bei bestimmten Fluenzen eine Verbesserung der magnetischen und elektronischen Eigenschaften beobachtet, sodass diese nach der Bestrahlung denen des
Volumenmaterials nahe kommen. Jedoch wird kein Phasen
ubergang zur L21 Struktur beobachtet. Die Bestrahlung von Co2 FeSi-Schichten mit Ga+ -Ionen
hat dagegen einen destruktiven Einfluss auf die magnetischen Eigenschaften
von Co2 FeSi. Bei groeren Fluenzen wird sogar eine Zerstorung der ferromagnetischen Ordnung beobachtet.

Contents
1 Introduction
2 Concepts in magnetism
2.1 Fundamentals . . . . . . . . . . . . . . . . .
2.1.1 Exchange interactions . . . . . . . .
2.1.2 Magnetic anisotropies . . . . . . . . .
2.1.3 Magnetic domains . . . . . . . . . . .
2.2 Magneto-optical Kerr effect . . . . . . . . .
2.2.1 Phenomenological description . . . .
2.2.2 Microscopic origin . . . . . . . . . . .
2.3 Spin wave dynamics . . . . . . . . . . . . . .
2.3.1 Landau-Lifshitz torque equation . . .
2.3.2 Spin waves in single magnetic layers .
2.3.3 Brillouin light scattering . . . . . . .

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

5
5
5
9
11
14
15
17
18
19
20
22

3 Half-metallic Heusler compounds


25
3.1 Structural properties of Co2 -based Heusler compounds . . . . . 26
3.2 Half-metallicity . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1 Origin of the band gap in Co2 -based Heusler compounds 30
3.2.2 Effects destroying half metallicity . . . . . . . . . . . . . 31
3.3 Magnetic properties of Co2 -based Heusler compounds . . . . . . 36
3.3.1 Magnetic moments and Slater-Pauling behavior . . . . . 36
3.3.2 Formation of local magnetic moments . . . . . . . . . . . 38
3.3.3 Exchange interaction between magnetic moments . . . . 39
3.3.4 Curie temperature . . . . . . . . . . . . . . . . . . . . . 40
3.3.5 Orbital magnetism . . . . . . . . . . . . . . . . . . . . . 41
4 Experimental methods
4.1 X-ray diffraction . . . . . . . . . . . . . . . . . . . . . . . . . .

43
43

vi

Contents

4.2

4.3

4.4

4.1.1 Generalities . . . . . . . . . . . . . . . . .
4.1.2 Application to Co2 YZ Heusler compounds
4.1.3 X-ray diffractometer . . . . . . . . . . . .
Kerr effect techniques . . . . . . . . . . . . . . . .
4.2.1 MOKE magnetometry . . . . . . . . . . .
4.2.2 MOKE microscopy . . . . . . . . . . . . .
Brillouin light scattering spectroscopy . . . . . . .
4.3.1 Measurement geometries . . . . . . . . . .
4.3.2 Brillouin light scattering spectrometer . .
Ion irradiation . . . . . . . . . . . . . . . . . . . .
4.4.1 Ion-solid interactions . . . . . . . . . . . .
4.4.2 Experimental setup . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

43
45
47
48
48
50
51
51
52
55
56
58

5 Experimental results I - Magnetic properties of Co2 -based Heusler


films
5.1 Film preparation and pre-characterization .
5.1.1 Overview of investigated samples . .
5.1.2 XRD investigations of Co2 MnSi films
5.2 Determination of exchange constants . . . .
5.2.1 Co2 MnSi . . . . . . . . . . . . . . . .
5.2.2 Co2 FeAl . . . . . . . . . . . . . . . .
5.2.3 Co2 Cr0.6 Fe0.4 Al . . . . . . . . . . . .
5.2.4 Co2 FeSi . . . . . . . . . . . . . . . .
5.2.5 Discussion . . . . . . . . . . . . . . .
5.3 Magnetic anisotropies . . . . . . . . . . . . .
5.3.1 Co2 MnSi . . . . . . . . . . . . . . . .
5.3.2 Co2 Cr0.6 Fe0.4 Al . . . . . . . . . . . .
5.3.3 Co2 FeAl . . . . . . . . . . . . . . . .
5.3.4 Co2 FeSi . . . . . . . . . . . . . . . .
5.3.5 Discussion . . . . . . . . . . . . . . .
5.4 Magnetization reversal . . . . . . . . . . . .
5.4.1 Magnetization reversal process . . . .
5.4.2 Influence on coercivity . . . . . . . .
5.5 Quadratic magneto-optical Kerr effect . . . .
5.5.1 Co2 FeSi . . . . . . . . . . . . . . . .
5.5.2 Co2 MnSi . . . . . . . . . . . . . . . .
5.5.3 Discussion . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

59
60
60
61
64
65
69
71
72
73
80
80
83
85
86
89
93
93
96
96
97
98
100

Contents

vii

6 Experimental results II - Modification of Heusler films by ion


irradiation
6.1 Preliminaries . . . . . . . . . . . . . . . . . .
6.1.1 Overview of investigated samples . . .
6.1.2 Simulations of the irradiation process .
6.2 He+ irradiation of Co2 MnSi . . . . . . . . . .
6.2.1 RT irradiation 30 keV He+ . . . . . . .
6.2.2 RT irradiation with 130 keV He+ . . .
6.2.3 Irradiation with 30 keV He+ above RT
6.2.4 Discussion . . . . . . . . . . . . . . . .
6.3 Ga+ irradiation of Co2 FeSi . . . . . . . . . . .
6.3.1 MOKE investigations . . . . . . . . . .
6.3.2 Discussion . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

103
103
104
106
111
112
122
124
128
129
129
131

7 Summary and outlook

133

A Calculation of BLS intensity

137

Chapter 1
Introduction
Heusler compounds, with the general composition X2 YZ , which have been
known since the beginning of the last century, are currently attracting great interest due to their possible use in the novel field of spin-dependent devices, also
known as spintronics or magnetoelectronics [1]. This interest in Heusler compounds stems from the half-metallic character of their spin-split band structure,
i.e. metallic behavior for one spin component (majority spins), and insulating
behavior for the other one (minority spins), as predicted by ab initio calculations for many compounds of this material class [2, 3]. As such, these materials
may exhibit a 100 % spin polarization at the Fermi level, which would make
them ideal candidates for spin polarizers or spin detectors, amongst other applications.
In addition to half-metallicity, Heusler compounds exhibit several other features which make them suitable candidates for technological applications. Their
electronic band structure, the width, and position of the insulating gap particular can be tuned with respect to the Fermi level by changing composition [4].
Consequently, the design of new materials with electronic and magnetic properties adequate for a certain application is possible with this material class.
Moreover, several Heusler compounds possess relatively high Curie temperatures [5, 6], which is a prerequisite for the stability of the performance of devices incorporating ferromagnetic materials. The highest Curie temperatures
are found for the so-called Co2 -based Heusler compounds, which have the chemical composition Co2 YZ (with Y being a transition metal and Z an element
from the III-V groups) and crystallize in the L21 structure. This explains the
particular attention given to the Co2 -based Heusler compounds.
An indispensable precondition for a successful implementation of Heusler
compounds into real devices, however, is a good understanding of their electronic and magnetic properties. In the last decade, a lot of theoretical and

Introduction
experimental work has been carried out to explain the appearance of the ferromagnetic order in these materials and to elucidate the phenomena which might
be detrimental to the insulating gap in the minority spin channel. In particular, it was found out that inelastic electron-magnon interactions, which are
irrelevant at low temperatures but gain in importance at room temperature,
can create states near the Fermi level in the minority gap, thus reducing the
idealized 100 % spin polarization [79]. As such, the investigation of thermal
magnons in Heusler compounds is an important issue. In this context, a key
parameter is the exchange stiffness D which describes the energy of a magnon,
and is related to the exchange constant A which expresses the energy of aligned
spins in a magnetic material. The knowledge of these parameters also plays an
important role in micromagnetic simulations and the study of dynamic phenomena. However, a systematic experimental investigation and comparison of the
exchange stiffness in Heusler compounds is still lacking till today. In addition to
the inelastic electron-magnon interactions, atomic disorder in the crystal lattice
is also found to have a detrimental effect on the spin polarization [10,11], since it
can create states in the insulating gap as well. The influence of various types of
disorder which might occur in the L21 Heusler structure on the electronic structure of Heusler compounds is a topic of numerous theoretical and experimental
works. However, the question of how the atomic disorder would influence the
inelastic electron-magnon interactions has not yet been investigated.
Another magnetic property which is of equal importance for spintronics applications as the exchange interaction is the spin-orbit coupling. Since this
interaction is responsible for the coupling of the spin to the orbital angular
momentum, it would have a strong impact on the depolarization of highly spin
polarized currents required in the spin-dependent devices. Therefore the investigation of spin-orbit coupling for potential spintronics materials such as
Heusler compounds is a crucial issue towards the optimization of the performance of devices based on these materials. A possible way to get insight into
the spin-orbit interaction inherent to these materials is a separate determination of spin and orbital magnetic moments, which can be done by means of
x-ray magnetic circular dichroism for example. An easier way is the investigation of the magneto-crystalline anisotropies by magneto-optical techniques,
an example of which is the magneto-optical Kerr effect magnetometry. The
anisotropy constants K determined by such magneto-optical techniques are a
direct measure for the strength of the spin-orbit interaction. There are few
reports dealing with the determination of K values for the technologically relevant material class of Heusler compounds. However, a systematic experimental
investigation and comparison of anisotropy constants is not present for these
materials.
The goal of this thesis is a systematic investigation of the exchange and

Introduction
anisotropy constants for Co2 -based Heusler compounds. For this purpose, studies of thin films consisting of Co2 -based Heusler compounds are carried out
by means of Brillouin light scattering spectroscopy and magneto-optical Kerr
effect magnetometry. The investigations are performed with the particular
aim of revealing the dependence of exchange and anisotropy constants on both
the chemical composition and atomic ordering of the Heusler films. Moreover,
this thesis also explores the structural modification of Heusler films using ion
irradiation. Usually, high-temperature annealing is applied to Heusler films to
obtain the highly ordered L21 phase. This technique, however, often suffers
from the drawback of interdiffusion between layers, and is not compatible with
semiconductors when envisaging devices consisting of Heusler-semiconductor
hybrid structures. Irradiation with light ions such as He+ might provide an
alternative to the high-temperature annealing since it was shown to induce a
transition from a disordered to a more ordered phase for the FePt(Pd) binary
alloys [12, 13].
In the following, we first briefly introduce magnetic phenomena generally
necessary for the understanding of the performed experiments in Chapt. 2.
Thereafter, a comprehensive overview of the crystallographic, electronic, and
magnetic properties of the Co2 -based Heusler compounds will be given in
Chapt. 3, and the employed experimental methods will be described in Chapt. 4.
Finally, studies of exchange and magnetic anisotropies of Co2 -based Heusler
compounds will be presented in Chapt. 5, followed by the results obtained in
ion-irradiation experiments performed with Heusler films in Chapt. 6.

Introduction

Chapter 2
Concepts in magnetism
This Chapter gives an overview of magnetic phenomena which are important
for the understanding of the present work. First of all, the exchange interaction and magnetic anisotropies are introduced, which form the core subject of
this thesis. Thereafter, the macroscopic and microscopic behavior of a ferromagnet in an external magnetic field is described. To lay out the fundamental
principles of the experimental techniques employed in this thesis introductory
theoretical descriptions of spin waves in thin layers and magneto-optical Kerr
effect are given. It should be noted that it is not our goal to give a complete
mathematical description of these phenomena, and only those aspects will be
addressed in more detail which are necessary for the general understanding of
the experiments performed within this work and for the discussion of the results
in Chapts. 5 and 6. For more detailed descriptions, the reader is referred to
textbooks such as Magnetism by J. Stohr and H. C. Siegmann [14].

2.1
2.1.1

Fundamentals
Exchange interactions

In solids, the electronic orbitals of neighboring atoms overlap, which leads to


the correlation of electrons. This results in the interatomic exchange interaction
that makes the total energy of the crystal depend on the relative orientation of
spins localized on neighboring atoms. The exchange interaction is the largest
magnetic interaction in solids (1 eV) and is responsible for the existence of
parallel, i.e. ferromagnetic, and antiparallel, i.e. antiferromagnetic, spin alignment. The exchange interaction might be mediated by different mechanisms

Concepts in magnetism
depending on the material system under consideration. The most important
mechanisms are described below.
Direct exchange
The direct exchange arises from a direct overlap of electronic wave functions of
the neighboring atoms and the Pauli exclusion principle, which requires different
symmetry properties from the spatial and spin parts of the electronic wave
function. In a two-spin system, the exchange energy is defined as the energy
difference between the parallel and antiparallel spin configuration. For a manyelectron system, the exchange energy is given by the expectation value of the
Heisenberg Hamiltonian
X
i S
j ,
Hex = 2
Jij S
(2.1)
i<j

where Jij is the exchange integral describing the coupling between two spins or
i and S
j 1 . Depending
magnetic moments represented by the spin operators S
on the interatomic distances (i.e. orbital overlap) the values of Jij might have

Jexch (a.u.)

r / rd

Figure 2.1: Direct exchange energy as a function of the interatomic distance r divided
by the radius of the d orbital rd (adapted from Ref. [15]). A negative value of the
exchange energy results in an antiferromagnetic ground state alignment of spins,
whereas a positive value leads to a ferromagnetic ground state coupling. This is
illustrated by red arrows. The curve is also known as the Bethe-Slater curve.
1

Here the capitalized symbols represent coupled spins forming the atomic magnetic mo-

ment. Orbital moment is assumed to be zero.

2.1 Fundamentals

a positive or negative sign, resulting in the parallel or antiparallel ground state


configuration of spins, respectively. The direct exchange is a short-range interaction. If the interatomic distance is too large (i.e. wave function overlap too
small) the direct exchange coupling is not strong enough to overcome thermal
excitations, giving rise to paramagnetism.
Itinerant exchange
In metallic 3d systems, like Co, Fe or Ni, both the atomic moments and the
exchange interaction are due to the delocalized, also called itinerant, electrons.
It is thus not possible to describe the magnetic properties of metals in terms of
atomic or ionic moments coupled by the intersite exchange interaction, and one
has to rely on the band model of electrons. As a consequence of the Coulomb
repulsion of electrons and their kinetic energy, the bands with opposite spin
orientation are exchange split (Fig. 2.2) giving rise to a non-zero total magnetic moment, and the appearance of ferromagnetism. This exchange splitting
is commonly described using the Stoner model where the exchange splitting
energy is given by Eex = IM . Here, I is the so-called Stoner exchange parameter and M is the averaged atomic magnetization. Within this model, the
stability of the spontaneous magnetic order is given by the Stoner criterion
I N (EF ) > 1 ,

(2.2)

where N (EF ) is the density of states at the Fermi level. A high density of states
at the Fermi level and strong exchange splitting thus favor metallic ferromagnetism. The total magnetic moment is not fixed by atomic rules. Instead it
4

Energy (eV)

s-electrons

EF 0

0
-2

DEex
d-electrons

-4

-2
-4

Ni

-6
-8

-6
-8

Density of states

Figure 2.2: Spin resolved density of states of Ni adapted from Ref. [16]. Eex denotes
the exchange splitting energy.

Concepts in magnetism
depends on the unequal populations of electrons with different spin orientations,
defined by the position of the Fermi level in the conduction band. Therefore,
the magnetic moments in 3d metals and their alloys exhibit quite a regular
dependence on the number of valence electrons, which is known as the SlaterPauling dependence [1719] and will be discussed in more detail in conjunction
with the Heusler compounds in the next Chapter.
RKKY exchange
RKKY indirect exchange interaction (named after Rudermann, Kittel, Kasuya
and Yoshida) can play an important role when there is no direct overlap of the
wave functions with unpaired electrons. In this case the interaction between
two magnetic moments at sites i and j is mediated by the polarization of
the conduction electrons. Characteristic for this coupling mechanism is an
oscillatory behavior of the exchange integral J which changes its sign as a
function of distance between the localized moments (Fig. 2.3). Amongst others,
this mechanism explains the coupling of 4f electrons in rare earth materials
and is responsible for the oscillatory interlayer exchange coupling in multilayer
GMR structures. In contrast to the direct exchange interaction, this type of
interaction is long-range.

JRKKY (a.u.)

4
2
0

-2
-4
-6

10

15

20

25

2kFr

Figure 2.3: RKKY exchange energy in dependence of the interatomic distance r


multiplied by the radius of the Fermi sphere kF (adapted from Ref. [15]).

Superexchange
Superexchange is another kind of indirect exchange interaction and is important in ionic solids such as the transition metal oxides and fluorides. The most

2.1 Fundamentals

Mn

Mn

Figure 2.4: Schematic illustration of the superexchange interaction between two Mn


atoms mediated by an oxygen atom.

prominent example is MnO. The interaction between the magnetic Mn atoms


is mediated by the diamagnetic oxygen though the overlap of the metals 3d
and oxygens 2p orbitals, and a partial delocalization of the involved electrons
(Fig. 2.4). In the case of a parallel orientation of the magnetic moments located
at the metal centers, no delocalization occurs, which makes the antiferromagnetic alignment energetically favorable. Generally, the size of the superexchange
depends on the magnitude of the magnetic moments on the metal atom, the
orbital overlap between the metal, and the non-metallic element and the bond
angle.
It should be noted that other exchange mechanisms exist apart from those
described above. Biquadratic exchange [20, 21] and Dzyaloshinskii-Moriya interaction [22, 23] are just two examples of them. However, these mechanisms
will not be discussed here.

2.1.2

Magnetic anisotropies

In the absence of an external magnetic field, the magnetization M of a magnetic


solid usually tends to lie along one or several axes. Energy is required to displace
the magnetization from these preferential directions. The magnetic anisotropy
is defined as the energy that is necessary to turn M into any direction different
from the preferred axes. Magnetic anisotropies might be caused by different
mechanisms, and are generally described as different contributions Fani to the
free energy density of a magnetic system. For this, Fani is advantageously
expanded into a series of components i of the unit vector pointing into the
direction of magnetization:
X
Fani =
Ki,j,k 1i 2j 3k .
(2.3)
i,j,k

The parameters Ki,j,k in Eq. (2.3) are the so-called anisotropy constants, which
are accessible experimentally. The knowledge of these parameters is generally

10

Concepts in magnetism
sufficient for the description of different anisotropy contributions.
In thin films investigated within the framework of this thesis, magnetocrystalline and shape anisotropy make a major contribution to the magnetic
anisotropy. In the following, these two contributions are described in more
detail. Depending on the magnetic system under consideration, various other
anisotropy contributions might become important. In thin films and multilayers
strain effects can give rise to the so-called magneto-elastic anisotropy. Interface anisotropy contributions might become relevant as well, depending on the
interface properties of the different layers. A comprehensive overview of these
contributions and the description of their origins can be found for example in
Ref. [15].
Magneto-crystalline anisotropy
The origin of the magneto-crystalline anisotropy (MCA) lies in the spin-orbit
interaction which, in a simple semiclassical picture, is due to coupling of the
orbital moment (resulting from the motion of the electron around the nucleus)
and the spin angular momentum. The orbital motion represents a current loop
that gives rise to a magnetic field in the center of the loop. This field interacts
with the spin angular momentum, coupling the spin and orbital moments. In
a quantum mechanical treatment, the spin-orbit interaction is described by the
Hamiltonian
e~2 1 d(r)
s l = nl (r) s l .
HSO =
(2.4)
2m2e c2 r dr
The spin and angular momentum (represented by the operators s and l in
Eq. (2.4), respectively) couple via the electrostatic potential of the nuclear
charges (r), which has the largest gradient d(r)/dr for small distances r from
the nucleus. The expectation value of nl (r) is called the spin-orbit coupling
constant, or spin-orbit parameter. Its value is of the order of 10-100 meV. The
spin-orbit interaction is thus considerably weaker than the exchange interaction
(1 eV).
In single crystalline materials, the bonding is anisotropic, i.e. the overlap
of the atomic wave functions depends on the crystallographic directions. This
gives rise to an anisotropy of the orbital magnetic moment, which results in
different values of the spin-orbit energy (Eq. (2.4)) associated to different crystallographic directions. The symmetry of the MCA is apparently that of the
crystal lattice. A quantitative treatment of the MCA by means of ab initio
calculations is still not satisfactorily possible. Therefore, a phenomenological description, in the form of a series expansion given in Eq. (2.3), is often
used with i,j,k being the direction cosines defined with respect to particular

2.1 Fundamentals

11

crystallographic axes. For the case of a cubic system, which is of a particular importance for the present work, the number of coefficients in Eq. (2.3) is
substantially reduced and the MCA contribution is given by
(cub)

FMCA = K0 + K1 (12 22 + 22 32 + 32 12 ) + K2 12 22 32 .

(2.5)

The parameters K1 and K2 are the so-called first and second anisotropy constants for a cubic system. Usually, the interplay between these two anisotropy
constants determines the direction of hard and easy axes [15]. In case of a
(001)-oriented film with an in-plane magnetization, a positive value of K1 results in an easy axes direction parallel to the h100i crystallographic directions.
The h110i directions become the easy axes when K1 is negative.
Shape anisotropy
Besides the magneto-crystalline anisotropy described above the shape of the
sample gives rise to magnetic anisotropy as well. Shape anisotropy is mediated
by the long-range dipolar interaction:

1 X 1
(rij mi )(rij mj )
Edipdip =
mi mj 3
.
(2.6)
3
2
20 i6=j rij
rij
The summation is over all atomic magnetic dipoles mi and mj at a distance
rij and every pair of dipoles is only counted once.
In a thin film with homogeneous magnetization, the dominant contribution to the dipolar energy arises from the demagnetizing field created by the
uncompensated magnetic moments at the film surface. It is given by
E=

1
0 MS2 sin2 ,
2

(2.7)

where is the angle between the surface normal and the magnetization vector
MS . The dipolar anisotropy energy is thus minimized for an angle of 90 , i.e.
for magnetic moments lying in the plane of the layer. The in-plane alignment
of MS keeps the magnetic flux lines in the film plane, hence reducing the stray
field energy.

2.1.3

Magnetic domains

Since the beginning of the last century, it is a well known fact that the magnetization is not uniform in a ferromagnetic material, but rather arranged into
domains with different orientation of magnetization. This domain configuration

12

Concepts in magnetism
(a)

(b)

Figure 2.5: 180 and 90 domain wall.

minimizes the stray field energy. Within each domain, the magnetic moments
are aligned parallel to each other and point into one of the preferential directions that are determined by the magnetic anisotropies in the absence of an
applied magnetic field. A large variety of domain patterns exists depending
on the specific properties of the ferromagnetic sample under investigation [24].
Single crystalline ferromagnetic films with two easy axes in the sample plane,
like those studied in this work, typically exhibit domain configuration, where adjacent domains are oriented at 90 or 180 with respect to one another (Fig. 2.5).
The change of the magnetization direction between the adjacent domains
does not occur abruptly but is characterized by a slight tilt of the microscopic
magnetic moments in the boundary regions. These boundary regions are several
tens of nanometers wide and called domain walls. Two types of domain walls are
distinguished, which are named after their discoverers Bloch and Neel. These
two types of domain walls are schematically illustrated in Fig. 2.6. While in
the former the rotation of the magnetization vector occurs in a plane which
is parallel to the plane of the domain wall, in the latter the rotation takes
place in a plane perpendicular to it. Bloch walls are more common in bulk-like
thick films, while Neel walls are often observed in thin films, where a surface
stray field is avoided by the rotation of the moments within the surface plane.
The width of a domain wall is determined by the exchange and the anisotropy
energy [24].
(b)

easy axis

(a)

Figure 2.6: Rotation of magnetization in a (a) Bloch and (b) Neel wall [25]. The
local preferential direction of magnetization changes by 180 in both cases.

2.1 Fundamentals

13

In an external magnetic field H0 , which is assumed to be higher than a


certain critical field, the system will try to align all the magnetic moments
parallel to the field direction in order to minimize the Zeeman energy
Z
E Z = 0
M H0 d3 r .
(2.8)
V

In the above equation, 0 is the vacuum permeability and M is the magnetization distribution in a given volume V . The alignment of the magnetic moments
can occur through either the coherent rotation of magnetization or the growth
of domains in which the magnetization lies in a favorable direction at the expense of unfavorable ones by a motion of domain walls.
The response of a magnetic material to the external magnetic field is represented by a magnetization or hysteresis curve, an example of which is shown
in Fig. 2.7. A hysteresis loop is characterized by the saturation magnetization
MS , the remanence Mr , the saturation field HS and the coercive field (or coercivity) HC . The HC strongly depends on the details of the reversal mechanism.
If a uniform rotation of the magnetization occurs, as is assumed in the StonerWohlfarth model [26], HC is equal to the anisotropy field. In most systems,
where the nucleation of domains plays an important role in the magnetization
reversal process (which is the case for the Heusler films studied in this thesis),
the coercivity is usually smaller than the anisotropy field. Its value is determined in addition to the anisotropy constants by the number of local defects.
For domain walls, defects act as pinning centers which increase the stability
of the domain walls against the externally applied fields. This results in higher

Magnetization, M

MS
MR

HS

HC

Applied field, H

Figure 2.7: Definition of the main features of the hysteresis curve: saturation magnetization MS , remanence Mr , saturation field HS and coercive field or coercivity
HC .

14

Concepts in magnetism
values of HC .
Experimentally, the magnetization reversal curves can be obtained by using
magneto-optical Kerr effect magnetometry. Since this technique was employed
in this thesis, the underlying magneto-optical Kerr effect will be described in
the following. The experimental setup will be described in Chapt. 4.

2.2

Magneto-optical Kerr effect

When linearly polarized light is reflected from a magnetic film, its polarization
becomes elliptic and the principal axis is rotated. This effect is known as the
magneto-optical Kerr effect (MOKE) named after its discoverer John Kerr [27].
Accordingly, the rotation angle and ellipticity are referred to as Kerr angle
and Kerr ellipticity . In general, these two quantities can be expressed
in terms of complex amplitudes of the electric field vector. The quantities
and are proportional to the magnetization of the film. Which component of
the magnetization is probed in the experiment depends on the measurement
geometry.
Figure 2.8 illustrates three different Kerr effects, which are the (a) polar, (b)
longitudinal and (c) transverse MOKE. For the polar effect, the magnetization
direction is perpendicular to the film surface. For the longitudinal and transverse effect, the magnetization is lying in the film plane and is oriented either in
a parallel or perpendicular fashion with respect to the plane of light incidence.
The polar MOKE is strongest at the perpendicular incidence. In addition to
the polar, linear and transverse MOKE which exhibit a linear dependence on
the respective components of magnetization (i.e. MP , ML and MT ), there are
also higher order effects, for which the Kerr angle and ellipticity depend on
product terms involving the polar, longitudinal and transverse magnetization
components.
In a phenomenological picture, the different kinds of MOKE are described

MT
MP

polar

ML

longitudinal

transverse

Figure 2.8: Different magneto-optical Kerr effects that exist depending on the relative
orientation of the magnetization (red arrow), sample surface and the plane of the
incidence of light.

2.2 Magneto-optical Kerr effect

15

using material tensors in combination with the Jones matrix formalism usually introduced in textbooks on optics, or with a more generalized Yeh matrix
approach [28]. The microscopic origin of MOKE lies in the splitting of electronic levels due to both exchange and spin-orbit interaction. In the following
two Subsections, both the phenomenological description and the microscopic
origin of MOKE are looked at in more detail.

2.2.1

Phenomenological description

Optical and magneto-optical properties of a magnetized crystal are described


by the permittivity tensor ij , which can be expanded into a series in the
components of the sample magnetization M :
(0)

ij = ij + Kijk Mk + Gijkl Mk Ml + ,

(2.9)

(0)

where Mi are the components of M . The terms ij , Kijk and Gijkl denote
the components of the dielectric tensor and the linear and quadratic magnetooptical tensors, respectively [29]. The number of independent components of
these tensors is reduced taking into account the crystal symmetry and Onsagers
relation:
ij (M ) = ji (M ) .

(2.10)

For cubic crystals, this results in only one free (complex) parameter in the
(0)
constant term ij , another one in the linear term Kijk and three additional
parameters in the quadratic term Gijkl [30]:
(0)

ij = d ij ,
Kijk = K ,
(2.11)

Giiii = G11 ,
Giijj = G12 ,

i 6= j ,

G1212 = G1313 = G2323 = G44 ,


where ij is the Kronecker delta-function. Therefore, in the case of an in-plane
magnetization, the off-diagonal elements of ij can be written as

16

Concepts in magnetism

xy = yx

G
= 2G44 +
(1 cos 4) ML MT
2

G
sin 4 (ML2 MT2 ) ,
4

(2.12)

xz = zx = K ML ,
yz = zy = K MT ,
where is the crystal orientation with respect to the plane of the light incidence.
The term G is the so-called magneto-optical anisotropy parameter, and G =
G11 G12 2G44 . Analogous treatment of other crystal symmetries can be found
for example in Ref. [31, 32].
In the literature, the Kerr rotation and the Kerr ellipticity are often
regarded as the real and the imaginary part of a complex Kerr amplitude ,
i.e. = i . can be given by the following expressions (up to the first
order in the off-diagonal components of the permittivity tensor) [3336]
rps
s =
= As yx + Bs zx ,
rss
(2.13)
rsp
p =
= Ap xy + Bp xz .
rpp
In the above equations, s and p stand for the polarization of the incident light
perpendicular and parallel to the plane of incidence, respectively. The terms
rps , rss , rsp and rpp denote material specific reflection coefficients, whereas the
terms with mixed indices give the fraction of p-polarized light converted to spolarized light after the reflection and vice versa. The weighting optical factors
As/p and Bs/p are even and odd functions of the angle of incidence, respectively,
and can be expressed analytically in the limiting cases of dFM /4 NFM or
dFM /4 NFM . Here, dFM and NFM is the thickness and the refractivity
index of the ferromagnetic layer.
Substituting ij from Eq. (2.12) into Eq. (2.13) results in the following
expression for the complex Kerr amplitude [30, 37]

G
s/p = As/p 2G44 +
(1 cos 4) ML MT
2
(2.14)
G
As/p
sin 4 (ML2 MT2 ) Bs/p K ML ,
4
where + (-) is related to the Kerr s (p) effect. Equation (2.14) is a final
expression of the Kerr effect in the case of an in-plane magnetized film with

2.2 Magneto-optical Kerr effect

17

cubic symmetry. In addition to the linear longitudinal Kerr effect described by


the last term in Eq. (2.14), there are two separate quadratic contributions to
the total Kerr effect proportional to ML MT and ML2 MT2 . In contrast to the
linear effect, which does not depend on the crystal orientation , the quadratic
effect exhibits anisotropy, i.e. it oscillates as a function of . The magnitude
of the anisotropy of the quadratic signal is determined by the magneto-optic
anisotropy parameter G.

2.2.2

Microscopic origin

As we have shown in the previous Subsection, the magneto-optical Kerr effect


is related to the off-diagonal permittivity tensor elements. The elements of
the linear magneto-optical tensor can be expressed in terms of the microscopic
electronic structure using the Kubo formalism [38, 39], according to which the
dissipative part of the off-diagonal component of the conductivity tensor2 has
the form (for > 0)
Im[xy ()] =

X
e2
f (Ei ) [1 f (Ef )]
4 m2e i,f

(2.15)

[|hi|
p |f i|2 |hi|
p+ |f i|2 ] (Ef Ei ~)) .
In the above equation is the light frequency, the total volume, f (E) the
= p
x p
y are the linear momentum operators
Fermi-Dirac function and p
x = i ~ /x). The right hand side of Eq. (2.15) describes the absorp(e.g. p
tion of a photon by an electron through an electric dipole transition from an
occupied initial state |ii to an unoccupied final state |f i. The matrix elements
hi|
p+ |f i and hi|
p |f i give the transition probabilities for left (+ ) and right
( ) circularly polarized light, respectively. The Dirac -function expresses the
condition of energy conservation.
A non-vanishing Kerr effect thus requires different absorption probabilities
for + and light. This is only possible if the electronic levels involved in
the dipole transitions are split by both the exchange and spin-orbit coupling as
schematically shown in Fig. 2.9 for the case of p d transitions. The selection
rules for these transitions are
s = 0 , l = 1 , m = 1 .
2

(2.16)

The corresponding dispersive component Re[xy ()] can be obtained using the Kramers-

Kronig relations [40]. The permittivity tensor is related to the conductivity tensor through
the expression ij = ij + (i/)ij .

18

Concepts in magnetism

spin-down e-

d9

(b)

|108,

absorption

Dm = 1

left circ.
pol. light
Dm =-1

right circ.
pol. light
Dm =1

D 0- D ex + D SO
D 0- D ex -D SO

d8

|218,
|2-18,

spin-up e-

2D SO

|219,

p8

D0

Dm = 1
|2-19,

p9

Dm = -1

2D SO

Fermi
level

Dm = -1

|109,

2D ex

(a)

D 0+ D ex + D SO
D 0+ D ex

-D

SO
-

hw

Figure 2.9: (a) Schematic representation of electric dipole transitions between magnetically perturbed p and d states. For clarity reasons only transitions between the
levels with (l = 2, ml = 1) and (l = 1, ml = 0) are shown. (b) Absorption spectra
for left and right circularly polarized light (adapted from Refs. [38, 41]).

The allowed dipole transitions are represented by vertical arrows in Fig. 2.9(a).
For clarity reasons only transitions between the levels with (l = 2, ml = 1) and
(l = 1, ml = 0) are shown. When both the exchange and spin-orbit coupling
are present, the absorption spectrum is different for left- and right-circularly
polarized light (Fig. 2.9(b)), which results in a non-zero Kerr effect. If either
spin-orbit coupling or exchange is absent, the absorption spectrum becomes
equal for both polarizations, leading to a vanishing Kerr effect.

2.3

Spin wave dynamics

The concept of spin waves, as dynamic eigen excitations of a magnetic system,


was proposed by Bloch [42] in order to explain the temperature dependence of
the magnetization of ferromagnetic materials [14,15]. A spin wave, or magnon,
consists of a collection of spins that coherently precess about the magnetization
direction (Fig. 2.10). In the case of small wavelengths, the spin-wave characteristics are mainly determined by the exchange interaction, whereas for spin
waves with large wavelengths the dipolar interaction dominates. Therefore two
types of spin-wave modes are generally distinguished, the so-called dipolar and
exchange modes. In the following an introduction to the properties of spinwaves in thin films is given, starting with the fundamental equation in the field
of spin dynamics, the Landau-Lifshitz torque equation.

2.3 Spin wave dynamics

19

Figure 2.10: Side view (top) and top view (bottom) of a spin wave in a onedimensional magnetic moment chain.

2.3.1

Landau-Lifshitz torque equation

The fundamental equation of motion in magnetism is the Landau-Lifshitz torque


equation [43]
1 dM

= M Heff ,
(2.17)
dt
which describes the precessional motion of the magnetization vector M in an
effective magnetic field Heff . The parameter = gB /~ is the gyromagnetic
ratio, where g is the spectroscopic splitting factor (also called Lande g-factor).
The effective field Heff is the sum over all the external and internal magnetic
fields acting on the magnetization and is usually composed of the following
contributions
Heff = H(t) + Hdem + Hex + Hani .
(2.18)
H(t) includes an externally applied static field and fluctuating fields, which
originate from the precessional motion of spins and might be additionally
caused by an external microwave field. The term Hdem is the demagnetizing
field created by the uncompensated magnetic moments at the surface, whereas
Hex stands for the exchange field due to exchange interaction. The last term
in Eq. (2.18) takes into account the internal fields associated with magnetic
anisotropies.
In order to obtain the dispersion relations of various spin-wave modes, the
Landau-Lifshitz equation must be solved together with the Maxwell equations
in the so-called magnetostatic approximation
H = 0,

(2.19)

(H + 4 M ) = 0 .

(2.20)

Since the fluctuations in M (t) and H(t) associated with spin waves are small
compared to the static values, the magnetization and the field vectors are usually split into time-independent static parts MS and H0 and dynamic parts
m(t) and h(t), and the above set of equations is solved for the dynamic components only, assuming appropriate boundary conditions for surfaces and interfaces [44, 45].

20

Concepts in magnetism

2.3.2

Spin waves in single magnetic layers

As already mentioned in the introductory part of this Section, different spinwave modes might appear in thin films. Their frequency can be largely dominated by either the exchange or the dipolar interaction.
Different types of dipolar spin wave modes are summarized in Fig. 2.11. If
the magnetization M and the wave vector q lie both in the film plane and are
perpendicular to each other a magnetostatic surface mode (MSSM) or DamonEshbach mode is excited. If both M and q are collinear the so-called magnetostatic backward volume modes (MSBVM) appears. The magnetostatic forward
volume modes (MSFVM) are observed when M is oriented perpendicular to
the film surface. In the present work, MSSM geometry was exclusively used.
Therefore, only this type of dipolar spin waves will be treated in more detail in
the following.
The Damon-Eshbach mode is characterized by the localization of the mode
energy (i.e. localization of the amplitude of the dynamic magnetization) close
to the film surface and an exponential decay of the precessional amplitude
along the film normal. Furthermore, it exhibits a nonreciprocal behavior, which
means that the propagation is possible for either positive or negative direction
of the wave vector but not for both. In the magnetostatic limit, i.e. weak
exchange contribution, and in the case of negligible anisotropies, the dispersion

Frequency (GHz)

9.0

8.5

magnetostatic surface
wave (MSSW)

MS

8.0
7.5

magnetostatic forward
volume wave (MSFVW)

7.0

MS

6.5
magnetostatic backward
volume wave (MSBVW)

6.0
q

5.5

MS

10

q|| d
Figure 2.11: Typology of the spin wave modes in a magnetic film as a function of the
directions of the magnetization MS and the in-plane wave vector qk (adapted from
Ref. [46]).

2.3 Spin wave dynamics

21

relation of the Damon-Eshbach mode is given by [46]:

(q)
= H0 (H0 + 4MS ) + (2MS )2 1 e2qk d ,

DE

(2.21)

where qk is the in-plane component of the wave vector q and H0 is the external
magnetic field. For fixed experimental conditions (i.e. H0 and qk are constant),
the frequency of the Damon-Eshbach mode only depends on the saturation
magnetization. Therefore, trends observed for DE provide valuable information
on the behavior of MS .
Apart from the surface localized Damon-Eshbach mode, the so-called perpendicular standing spin waves (PSSW) are accessible in the MSSW geometry
as well. These modes are of the exchange type and are formed by a superposition of two spin waves propagating in opposite directions perpendicular to the
film. The wave vectors of the PSSW modes are thus quantized, q = p/d, with
p being the quantization number and d the film thickness. Their frequencies
can be approximately described by [46]:
2

(q)
2A 2 2A p 2
= H+
q +

MS k MS d
(2.22)
p

"
#
!

2
4MS /H
2A
2A p
H+
+H
qk2 +
+ 4MS ,
MS
p/d
MS d
where A is the exchange constant. The detection of PSSW modes therefore
allows for the determination of the exchange interaction in a given magnetic
sample. In the literature, the so-called exchange stiffness constant or spin
stiffness D is frequently found along with the exchange constant A. Both
quantities are related through the expression [A5]:
A=

DMS
.
2gB

(2.23)

Equations (2.21) and (2.22) provide a proper description of the spin-wave


frequencies as long as the condition qk d 1 is fulfilled, and only in the frequency regions where no mode crossing occurs. For the spin waves investigated in the Heusler films in this thesis, qk is typically 1.7 105 cm1 , and
d = 30 80 nm. Consequently, qk d 1 is not longer valid. Therefore a numerical approach is necessary for the determination of the spin-wave frequencies
which is based on a model described in detail in Refs. [44, 45]. This numerical
approach also allows the determination of MS and A values, as will be shown
in Chapt. 5.

22

Concepts in magnetism

2.3.3

Brillouin light scattering

Photons can interact with magnons, i.e. the elementary quanta of spin waves,
via inelastic photon scattering which is known under the name of Brillouin
light scattering. Generally, two different scattering processes are distinguished,
in which a magnon is created (Stokes process) or annihilated (anti-Stokes process), resulting in an energy shift of the scattered photon (Fig. 2.12). Both
photon-magnon scattering processes satisfy the laws of energy and momentum
conservation:
~i = ~s ~ ,
~ki = ~ks ~k .

(2.24)
(2.25)

In these expressions, i,s and ki,s denote the frequency and the wave vector of
the incoming and the scattered photon, respectively. The parameters describing
the magnon are without any index. For light scattering from thin films the
perpendicular component of the wave is not conserved anymore due to the
break of translational symmetry. In this case, Eq. (2.25) is only valid for kk
which is the wave vector component parallel to the film plane.
Besides the one-magnon scattering processes schematically shown in
Fig. 2.12, higher order scattering processes exist, in which more than one
magnon are involved. Since the latter are not relevant for the present thesis,
they will not be considered further. In the literature, photon-magnon interactions are usually treated separately from the magneto-optical effects such

ws=wi-w
ks=ki-k

ws=wi+w
ks=ki+k
w, k

wi, ki

w, k

wi, ki

Stokes process

anti-Stokes process

Figure 2.12: Schematic representation of photon-magnon scattering processes. A


magnon can be created (Stokes process) or annihilated (anti-Stokes process) resulting
in a gain or loss in energy and momentum of the scattered photon. i and ki denote
frequency and wave vector of the incoming photon, s and ks describe the scattered
photon and and k are assigned to the magnon.

2.3 Spin wave dynamics

23

as Kerr or Faraday effect. However, they are closely related to each other.
In the quantum mechanical picture, the magnon scattering and the classical
magneto-optical effects can be described by a common Hamiltonian [47, 48].
But even without going into the details of the quantum mechanical treatment,
the analogy between these effects can be shown as will be demonstrated in the
following.
Macroscopic description
In a classical treatment, the light scattering from a solid occurs due to the
electric polarization introduced in the solid by the external optical electric field.
The relationship between the polarization P and the electric field E is given
by
X
Pi = 0
ij Ej ,
(2.26)
j

where ij is the susceptibility tensor. The above expression can be rewritten as


a function of the permittivity tensor introduced in Sect. 2.2.1 using the relation
ij = ij +ij . At this point, the common origin of the Brillouin light scattering
and the MOKE effect becomes evident. In the presence of spin waves, the offdiagonal permittivity tensor elements, i.e. the elements responsible for the light
scattering of magnons, have a similar form to Eq. (2.12), the only modification
being that the components of the static and dynamic magnetization vectors
Ms and m(t) have to be used instead of ML and MT .
For crystals with cubic symmetry probed in the Damon-Eshbach geometry
(Fig. 2.11), where the direction of the applied magnetic field and of the static
component of magnetization is defined as the x-axis and the in-plane component
of the spin wave vector qk lies along the y-axis, the microscopic spins precess
about the x-axis and the dynamic magnetization has only two components my
and mz . In this case the off-diagonal components of ij related to spin waves
can be expressed as [49]
xy = yx = Kmz + 2G44 Ms my
xz = zx = Kmy + 2G44 Ms mz ,

(2.27)

where K and G44 are the components defined in conjunction with MOKE in
Sect. 2.2.1. It should be noted that in deriving Eq. (2.27) Ms was assumed
to lie along one of the cubic crystalline axes and terms quadratic in mi were
neglected. The latter is justified since |m(t)| |Ms |.
Substituting Eq. (2.27) in Eq. (2.26) and assuming a linearly p-polarized

24

Concepts in magnetism
incident wave (i.e. Ex = 0; Ey ,Ez 6= 0) one obtains only one component of P :
Px = (2G44 Ms Ey KEz )my
+ (KEy + 2G44 Ms Ez )my .

(2.28)

For the s-polarized incident wave having only the component Ex , the substitution of Eq. (2.27) in Eq. (2.26) yields two components of P :
Py = 2G44 Ms Ex my KEx mz ,
Pz = KEx my + 2G44 Ms Ex mz

(2.29)

These relations imply the well-known fact that an incident s-wave changes to
a p-wave upon the interaction with a magnetic excitation and vice versa. It
is interesting to note that for the incoming photon the dynamic modulation
of polarization expressed in Eq. (2.28) and (2.29) represents a phase grating
propagating with the velocity of spin wave. Therefore, the photon-magnon
interaction can be understood in a classical picture as Bragg scattering from
a moving grating resulting in the Doppler shift of the frequency of the Braggscattered light:
s = i k v i .
(2.30)
Vectors k and v in this expression denote the wave vector and the velocity of
the spin wave.

Chapter 3
Half-metallic Heusler
compounds
The interest in the intermetallic compounds with the general formula X2 YZ,
where X and Y are transition metals and Z an element from the III-V groups,
emerged in 1903, when F. Heusler discovered Cu2 MnAl to be ferromagnetic even
though it consists only of non-ferromagnetic elements [50]. Since this discovery,
numerous investigations were made to clarify the chemical and magnetic ordering in these materials [5154] and different models were proposed to explain the
microscopic interactions leading to the appearance of ferromagnetism [5557].
It was also found that some of the ternary intermetallic compounds with the
chemical composition XYZ exhibit a ferromagnetic behavior as well. Today
this class of materials is known as half- or semi-Heusler compounds, whereas
the X2 YZ compounds are named full-Heusler compounds. Among the halfHeusler compounds, PtMnSb attracted particular attention due to its extremely
large magneto-optical Kerr rotation [58], which made this compound a suitable
material for the magneto-optical recording technology.
The discovery of the half-metallic property in NiMnSb and related compounds by de Groot et al. in 1983 [2] launched many numerical works devoted
to the prediction of other half-metallic materials. Several other Heusler compounds turned out to posses the half-metallic property, as suggested from ab
initio calculations. Among them are the Co2 -based Heusler compounds, which
gained a particular interest and became most widely studied in this field due
to their high Curie temperatures. However, despite the huge achievements that
have been obtained in the last decade in magnetic tunnel junctions (MTJs)
using Co2 YZ compounds as electrodes [5964], an experimental proof of the
half-metallicity in these materials is still missing. Different mechanisms that

26

Half-metallic Heusler compounds


are thought to explain the suppression of the half-metallic state will be presented in this Chapter. The main goal of this Chapter, however, is to provide
an overview of the crystallographic, electronic and magnetic properties of the
full-Heusler compounds where X=Co. A special attention will be given to the
materials studied within this work, which are Co2 MnSi, Co2 FeAl, Co2 FeSi and
Co2 Cr0.6 Fe0.4 Al (CCFA in the following).

3.1

Structural properties of Co2-based Heusler


compounds

Heusler compounds with the general formula Co2 YZ crystallize in the L21 structure which is shown in Fig. 3.1. The cubic unit cell consists of four interpenetrating fcc sublattices, two of which are occupied by Co atoms, and the other
two by the Y and Z atoms, respectively. The two Co sublattices are positioned
at (0,0,0) and ( 21 , 12 , 12 ) in the unit cell, while those of Y and Z are at positions
( 41 , 14 , 14 ) and ( 34 , 43 , 34 ), respectively. The L21 structure represents the most ordered phase of the Co2 -based Heusler compounds. However, partial disorder
can exist for a given chemical composition, in the form of interchange of atoms
between different sublattices. If the ( 14 , 14 , 14 ) and ( 34 , 34 , 43 ) sites are randomly
occupied by either of Y or Z with a 50 % probability, the structure is referred
to as the B2 structure. The interchange of atoms in Co and Y sublattices results in the DO3 type of disorder. A random occupation of all four sublattices
leads to a structure which is called the A2 structure [6567].
Experimentally, the information about the structural properties, such as lat-

Co
Y
Z

Figure 3.1: L21 structure of Co2 YZ Heusler alloys. The lattice consists of four
interpenetrating fcc sublattices, two of which are occupied by Co, the other two by
Y and Z atoms, respectively.

3.1 Structural properties of Co2 -based Heusler compounds

27

tice constants and the type of disorder present in the Heusler samples, can be
obtained by x-ray and neutron scattering. The bulk lattice parameters of Co2 based Heusler compounds, which are studied within this work or appear in the
discussions of experimental results in Chapts. 5 and 6, are listed in Table 3.1. It
is interesting to note that there is a systematically smaller lattice constant for
Si containing compounds. Structural information concerning atomic disorder
utilizing both x-ray and neutron scattering is provided by the analysis of relative intensities of superstructure reflections, as will be discussed in more detail
in Chapt. 4. Since the intensities of neutron diffraction peaks can be determined in general with a much greater accuracy than those of the corresponding
x-ray peaks [54], neutron scattering is a more favorable technique for the quantitative determination of the degree and type of disorder. However, a neutron
source is not easily accessible. Thus, neutron scattering cannot be routinely
used for sample characterization. X-ray diffraction, which is easily accessible,
suffers from the drawback of a very small difference in the atomic scattering
factors between Co and second transition metal Y, to which the superstructure
reflections are sensitive (Chapt. 4). Therefore, additional characterization by
other techniques is usually necessary to eliminate possible ambiguities. In this
context, local experimental techniques such as x-ray absorption spectroscopy
(XAS) in combination with circular magnetic dichroism (XMCD), Mossbauer
spectroscopy and nuclear magnetic resonance spectroscopy (NMR), which are
sensitive to the short-range order, are commonly used. In particular, NMR was

Compound
CCFA

NV (e /f.u.) Structure

aexp (
A) Ref.

27.8

B2

5.737

[68]

Co2 MnAl

28

B2

5.756

[5]

Co2 MnSi

29

L21

5.654

[5]

Co2 MnGe

29

L21

5.743

[5]

Co2 FeAl

29

B2

5.730

[68]

Co2 FeGa

29

L21

5.741

[69]

30

L21

5.640

[70]

Co2 FeSi

Table 3.1: Experimental bulk lattice parameters of selected Co2 YZ Heusler compounds. The listed compounds are sorted by the increasing number of valence electrons NV (see also Sect. 3.3.1). Compounds studied within this thesis are marked
with (*).

28

Half-metallic Heusler compounds


demonstrated to a be a very powerful method for the characterization of atomic
disorder [7173].
In order to obtain Heusler films with the well-ordered L21 structure, high
temperature annealing is usually required since the B2/L21 pase transition in
the bulk occurs at about 1000K in most of the Co2 -based Heusler compounds
[7476]. It is interesting to note, however, that the L21 phase is not always
the most stable one from the thermodynamic point of view. Regarding the
series of materials studied in this thesis, CCFA preferably crystallizes in the B2
structure [77] while L21 structure is more stable for Co2 MnSi and Co2 FeSi.

3.2

Half-metallicity

The concept of a half-metal was first introduced by de Groot et al. on the


basis of band structure calculations for the NiMnSb half-Heusler compound [2].
This term is employed to describe material systems with an asymmetry of the
spin-split band structure: while the majority spin band has a metallic behavior
(i.e. a non-vanishing density of states (DOS) at the Fermi level), the minority
electrons exhibit a semiconducting character (i.e. a band gap around the Fermi
level) (Fig. 3.2). The conduction electrons are thus 100% spin polarized.
Since the discovery of the half-metallicity in NiMnSb, many other materials have been theoretically predicted to be half-metals, among them the Co2 based Heusler compounds. First theoretical evidence of the half-metallicity
in Co2 MnSi was given by Ishida et al. [78] and was later confirmed by various
authors [7982]. K
ubler et al. reported for the first time the existence of a band
gap for the minority electrons in the related Co2 MnAl Heusler compound [83].
In this compound, however, the Fermi energy falls into the upper tail of the
valence band and a non-vanishing spin-down DOS is found at the Fermi level
(Fig. 3.2(b)). More recent calculations showed similar results for the electronic
structure of this material [80, 82, 84]. Nevertheless, Co2 MnAl remains an interesting candidate for applications because its expected spin polarization is
still comparatively high. The half-metallic behavior of the Co2 FeSi Heusler
compound has been reported by Wurmehl and collaborators [70] and Gercsi
et al. [11]. In case of the Al containing counterpart, different results of band
structure calculations can be found in the literature. While no half-metallic
state is found for Co2 FeAl in Refs. [77, 85], a clear band gap for minority spins
is shown in Refs. [11, 82].
The differences in the reported electronic band structures mainly result
from different computational schemes used in ab initio calculations by different authors. Most of the theoretical investigations are based on the density

3.2 Half-metallicity

29

DOS (states/eV)

Co2MnSi

0
-4
4

(a)
Co2MnAl

0
-4

(b)
-6

-3
0
Energy (eV)

Figure 3.2: Spin-resolved density of states (DOS) for (a) Co2 MnSi and (b) Co2 MnAl
[86]. The upper and lower panels in (a) and (b) show the DOS for the majority
(spin-up) and minority (spin-down) electrons, respectively, as indicated by the red
arrows.

functional theory (DFT), which expresses the ground state energy of an atomic
system as a function of its electron density [87]. The main problem of DFT
is that the exact potentials describing the exchange interaction of the electrons are not known and approximations are required. The basic assumption
underlying these approximations is a delocalized or itinerant behavior of the
valence electrons. However, as will be discussed in Sect. 3.3.1, the Co2 -based
Heusler compounds exhibit characteristics of systems with localized magnetic
moments. Therefore, additional corrections such as electron-electron correlation potentials must be introduced in the calculations, which take into account
a partial localization of the valence electrons [82, 88]. This very rough insight
into the band structure calculations of the Heusler compounds makes evident
that the prediction of the half-metallic state strongly depends on the details
of the computational approach. In particular, the width and the position of
the minority band gap with respect to the Fermi level might vary depending
on the calculation scheme. It should be noted as well, that in most cases the
half-metallic property is predicted for the ground state only, i.e. for T=0 K.
The electronic structure of Co2 YZ Heusler compounds and their physical
properties might be tuned by alloying with a fourth element, which opens a
new way to design materials with desired characteristics [4,11,75,89,90]. In the
quaternary alloys, the atoms in one of the sublattices are partially substituted
by the atoms of another element. CCFA, obtained from Co2 CrAl by doping

30

Half-metallic Heusler compounds


with Fe atoms, represents one of the most prominent examples in this class
of Heusler compounds. In this case, a partial substitution of Cr resulted in a
large magnetoresistance effect of about 30% which was not observed for the
initial ternary compound [91]. Miura et al. investigated in detail the complete
substitutional series Co2 Cr1x Fex Al from the theoretical point of view. Similar
to the case Co2 MnAl, a non-vanishing DOS is expected in the minority band
of Co2 Cr0.6 Fe0.4 Al according to these calculations, for both B2 and L21 ordered
phases.

3.2.1

Origin of the band gap in Co2 -based Heusler compounds

Even though Heusler compounds are known for nearly a century, the mechanisms responsible for their particular electronic structure remained unexplained
for a long time. In case of Co2 -based compounds, it was just in the year 2002
that Galanakis and collaborators proposed a mechanism leading to the appearance of the minority band gap which became generally accepted by the Heusler
community. As has been pointed out in their original work, the gap in the
minority DOS of Co2 -based Heusler compounds results from the hybridization
of d electrons taking place between the Co and X atoms as well as between the
Co atoms sitting on second-neighbor positions [80]. In the following this will
be discussed in more detail.
Figure 3.3 schematically shows the behavior of d electrons for X=Mn. In a
first step the five d orbitals of each Co atom couple according to their symmetry group representations and form bonding (eg , t2g ) and antibonding (eu , t1u )
orbitals. In a second step the interaction between the hybridized Co states and
the d states of the Mn is considered. The doubly-degenerate Co eg orbitals hybridize with the e (dz2 , dx2 y2 ) orbitals of the Mn, creating a doubly-degenerate
bonding and antibonding eg states, that are situated below and above the Fermi
level, respectively. The t2g Co orbitals couple with the t2 (dxy , dyx , dzx ) states
of the Mn atom, giving rise to a low-lying triplet t2g state with a bonding character and a triplet antibonding t2g state above the Fermi level. The remaining
antibonding Co orbitals (eu and t1u ) do not hybridize, since no states exist on
the Mn atom with the same symmetry group representation, and thus remain
non-bonding. The Fermi level falls between these two non-bonding Co orbitals.
Therefore, a real minority gap exists in the Co2 YZ Heusler compounds. The
size of the gap is largely determined by the Co-Co interaction. It should be
noted here, that in principle the same hybridization of electronic states may
occur for the majority electrons as well, opening a band gap in the spin-up
DOS. The exchange splitting, however, shifts the spin-up states of Mn to lower

3.2 Half-metallicity

31
t2g
eg

eu
dz ,d x -y
2

Co

dxy , dyx , dyz

t1u
eg
t2g

dz ,d x -y
2

Mn
dz ,d x -y
2

dxy , dyx , dyz

eu

Co

dxy , dyx , dyz

EF

t1u

Co-Co
eg
t2g
t2g
eg

Figure 3.3: Schematic illustration of the origin of the minority band gap in Co2 YZ
Heusler compounds proposed by Galanakis et al. [80].

energies where they form a common band with the Co states.


In the above discussion the Z atom has been completely disregarded. This is
justified by the fact that its low-lying s and p states do not directly contribute to
the formation of the minority band gap. However, since these states contribute
to the total number of occupied and empty states, the Z atom is important
for the position of the Fermi level within the minority band gap. In Fig. 3.2
for example, a clear shift of the Fermi level towards the conduction band is
visible when Al is substituted by Si in the Co2 MnZ compound. Moreover, the s
and p states have been shown to play an important role for the distribution of
electrons in the various symmetry distinguished states (t2g and eg ) at Co and
Y sites [82]. This is of particular importance in the discussion of the specific
magnetic moments at the Co and Y sites as will be shown in Sect. 3.3.2.

3.2.2

Effects destroying half metallicity

We already mentioned before, that ab initio predictions of the half-metallic


property for Co2 -based Heusler compounds are generally made for the ground
state. Moreover, a bulk material is usually assumed in the theoretical investigations, and coupling of the spin to the orbital moment is neglected. Halfmetallicity has yet to be proven in thin films of Co2 YZ Heusler compounds
even though high tunnel magnetoresistance (TMR) ratios corresponding to relatively high spin polarization of the conducting electrons have been achieved
at low temperatures in MTJs using these materials as electrodes [5964]. The
reason for the lack of the experimental evidence of the half-metallic state, i.e.

32

Half-metallic Heusler compounds


100% spin polarization, are believed to be additional electronic states appearing in real thin film samples in the minority DOS and closing the band gap.
These additional states within the band gap might be introduced by various
mechanisms, which will be briefly described in the following. The mechanisms
leading to the reduction of the theoretically predicted 100 % spin polarization
of Heusler compounds can partially be expected to play also an important role
for the exchange interactions and magnetic anisotropies in these materials.

Disorder effects
The effect of structural disorder, such as antisites (replacement of one kind of
atom by another) and swaps (interexchange of atoms in different sublattices)
on the half-metallic property of Co2 -based Heusler compounds has been studied
theoretically by various authors [10,11,77,84,90,92,93]. It has been found that
the electronic structure of these materials can be significantly modified by the
presence of structural disorder, including a complete destruction of the halfmetallic band gap. This detrimental effect arises from new minority electronic
states introduced at the Fermi level by the imperfections in the crystal lattice.
The appearance of the new minority states can be understood as follows. For
the hybridization of electronic d states at Co and Y atoms leading to a gap in the
minority DOS, the crystal field symmetry plays a crucial role (see Fig. 3.3 above
and related discussion). Swapping of the atoms or antisites induce changes in
the crystal field symmetry at a given Co or Y site, which results in additional
splitting of the d orbitals. Hence, new states can appear in the region of the
minority band gap.
The degree of the destructive impact of disorder varies substantially depending on the kind of disorder and the material system under consideration.
For example, for Co2 MnSi Picozzi and collaborators have shown that only in
the case of Co antisites (replacement of Mn by Co) is the half-metallic property
destroyed due to the formation of a peak in the minority DOS at the Fermi
level [10]. In the case of Mn antisites (replacement of Co by Mn) as well as
for Co-Mn and Mn-Si swaps, the half-metallicity is preserved. For the same
material system Galanakis and collaborators have demonstrated that a partial
substitution of Mn by Si atoms (and vice versa) up to 20% does not destroy
the half-metallic character of the Co2 MnSi Heusler compound [86]. Similar
results have been reported for Co2 FeSi by Gercsi et al. [11] and are shown in
Fig. 3.4. The half-metallicity is preserved when up to 25% of B2 disorder (due
to Fe-Si swaps) is introduced in the crystal structure, even though additional
electronic states start to appear at the lower edge of the minority conduction
band (Fig. 3.4(a)). The DO3 type of disorder broadens the gap, but 100% spin

3.2 Half-metallicity

33
40
B2 (25%)

L21

20
0

DOS (states/eV)

-20

(a)

-40

DO3 (12.5%)

20
0
-20

(b)

-40
A2 (12.5%)

20
0
-20

(c)

-40
-6

-5

-4

-3

-2

-1

E-EF (eV)

Figure 3.4: Spin-resolved density of states (DOS) of Co2 FeSi Heusler compound with
(a) B2, (b) DO3 and (c) A2 types of disorder taken from Ref. [11]. DOS of the L21
ordered phase is shown in grey.

polarization at the Fermi level is still conserved (Fig. 3.4(b)). In the presence
of the A2 type of disorder, however, new states appear at the Fermi level, thus
closing the half-metallic gap (Fig. 3.4(c)). In this context it is worth mentioning that the DO3 type of disorder significantly increases the total energy of the
system, and thus is thought to be unlikely to happen [10, 77].

Spin-orbit coupling
Even in an ideally prepared single crystal, electronic states in the half-metallic
gap of the minority band may be introduced due to spin-orbit coupling. The
spin-orbit interaction couples the majority and minority spin channels to one
another [94, 95]. This has the consequence that spin-flip scattering processes
can take place, producing additional states in the gap of the minority band.
Thus, 100% spin polarization cannot exist even in a hypothetically ideal crystal
in the ground state. It should be noted, however, that the effect of spin-orbit
interaction on the spin polarization of Co2 -based Heusler compounds is very
small due to comparatively small orbital magnetic moments (see Sect. 3.3.5)
and the spin polarization at the Fermi level can still be very high. Sargolzaei
and collaborators have demonstrated in their theoretical studies that in case
of Co2 MnSi, the reduction of the spin polarization due to this effect is of only
3% [96]. While the spin polarization is 100% without taking the spin-orbit
coupling into account, it amounts to 97% when the spin-orbit interaction is

34

Half-metallic Heusler compounds


included in the band structure calculations. In case of Co2 MnAl a decrease from
97% to 95% is reported. At the edges of the band gap the effect of spin-orbit
interaction can become more pronounced since there exist spin-down states and
therefore the probability for the spin-flip scattering of the majority electrons
is higher compared to the middle part of the band gap. The position of the
Fermi level within the half-metallic band gap will therefore largely determine
the impact of this effect on the spin polarization.
Temperature effects
At finite temperature there are mainly two effects giving rise to additional states
in the minority gap and thus destroying the half-metallic character. First,
thermal fluctuations of individual spins create a certain spin disorder and noncollinearity effects appear [97, 98]. In regions with short-range order, the local
spin quantization axis is not necessarily parallel to the average moment direction. Moreover, spin axis of each atom can vary with respect to that of its
neighbors if the short-range order is not perfect. The latter is more significant in
multicomponent systems like Heusler compounds. Due to the non-collinearity
of spins at T > 0, there always exists a partial projection of a spin-up wave
function of each atom onto the spin-down states of its neighbors generating
new states in the minority band gap. Note that in this case the non-vanishing
DOS in the gap of the minority spin channel is symmetric with respect to the
Fermi energy, i.e. new states are generated for both electrons and holes. The
second source of depolarization are the so-called nonquasiparticle (NQP) states
which originate from a superposition of spin-up electron excitations and virtual
magnons [7, 9, 99]. In contrast to the non-collinearity effects, the NQP states
result in an asymmetric band gap filling. At low temperatures these states are
introduced just above the Fermi energy. Note that if magnetic anisotropies are
taken into account, the cut-off of the NQP states is slightly above the Fermi
energy. At elevated temperatures the distribution of the NQP states is broadened and crosses the Fermi energy. This results in the loss of the half-metallic
property. Recently, experimental evidence of NQP states in the Co2 MnSi Heusler compound has been given by MTJ spectroscopy measurements [100].
Apart from non-collinearity effects and the appearance of NQP states, other
effects leading to the finite temperature degradation of half-metallicity are discussed in literature. For example, a change in hybridization strength between
the d states at elevated temperatures is proposed by Lezaic and collaborators [98], leading to a shift of the conduction and valence band edges in the
minority DOS. This has the consequence that the Fermi energy is no longer
located in the gap but falls into the minority conduction band and the halfmetallicity is lost. Moreover, the defect densities will increase at T > 0 leading

3.2 Half-metallicity
to changes in the electronic structure described in the beginning of this Subsection.

Surfaces and interfaces


At surfaces and interfaces the half-metallic property can be lost due to the
introduction of surface and interface states at the Fermi level in the minority spin channel. Amongst the Co2 -based Heusler compounds investigated in
this thesis, Co2 MnSi is the one whose surface and interface properties have
been most widely studied by ab initio calculations [78, 101105]. For example,
Galanakis has shown that the half-metallic character of this compound is lost
at both Co- and MnSi-terminated surfaces [101]. A lower coordination number of Co and Mn atoms at the surface reduces the hybridization between the
Co minority d states and the Mn ones. For Co-terminated surfaces, the Co
spin-down bands are shifted to higher energies with the effect that the Fermi
energy falls in the middle of the d band. For MnSi-terminated surfaces the unoccupied d-like Mn states shift to lower energies closing the minority band gap.
Hashemifar and collaborators, however, have shown that the half-metallicity
of the Co2 MnSi Heusler compound can be preserved if the surface has the
MnMn-termination [102].
The half-metallicity of the Co2 MnSi/MgO and Co2 MnSi/GaAs interfaces
have been theoretically studied as well [103105]. The former is of particular
importance for magnetic tunnel junctions, whereas the latter is considered for
spin injection into semiconductors. Similar to the discussion of surfaces, interface state appear in the half-metallic gap because of the deviating atomic
environment of Co and Mn atoms in the topmost layer. These new states at
the Fermi level can be filled by the electrons from the minority valence band
through inelastic scattering processes such as electron-phonon, electron-magnon
and electron-electron scattering [106] and thus make the spin-down electrons
contribute to the total conductivity of the interface. Therefore, the polarization
of the current passing through such an interface in a MTJ is reduced, and spin
injection into a semiconductor becomes less efficient.
Before switching over to the magnetic properties of Co2 -based Heusler compounds it should be pointed out that in real samples all of the above effects
will contribute to the appearance of additional states within the band gap.
Therefore special care must be taken when for example the effect of structural
disorder is discussed on the spin polarization obtained in MTJs with B2- and
L21 -ordered electrodes.

35

36

Half-metallic Heusler compounds

3.3

Magnetic properties of Co2-based Heusler


compounds

Due to their particular electronic structure, Co2 -based Heusler compounds exhibit several interesting magnetic properties. For example, most of these materials are strong ferromagnets with very high Curie temperatures. The origin
of the ferromagnetic behavior, however, is rather complicated in these material
systems and is still one of the most interesting problems in modern magnetism.
In this Section, we give an overview of the magnetic properties of the Co2 -based
Heusler compounds that are relevant for the discussion of the experiments performed within the frame of this work. We also report on the current status
in the understanding of the mechanisms behind these properties. The most
important parameters of materials studied in this thesis are summarized in
Table 3.2, and will be addressed in detail in the following.

3.3.1

Magnetic moments and Slater-Pauling behavior

One of the most striking characteristic of the half-metallic Heusler compounds


are their integer magnetic moments. To understand this behavior, we should go
back to our discussion of the origin of the half-metallic gap in Sect. 3.2.1. There
we have shown that Co2 YZ Heusler compounds have eight occupied minority
d states per unit cell, namely the doubly-degenerate eg very low in energy,
the triply-degenerate t2g state and the triply-degenerate t1u orbital situated
just below the Fermi level. Apart from these d states, there are one s and
NV (e /f.u.)

mcalc (B /f.u.)

mCo (B )

mY (B )

TC (K)

27.8

3.48

0.94

1.24/2.92

750

Co2 MnSi

29

5.00

0.98

3.29

985

Co2 FeAl

29

5.00

1.22

2.97

1000

Co2 FeSi

30

6.00

1.50

3.14

1100

Compound
CCFA

Table 3.2: Magnetic properties of Co2 -based Heusler compounds investigated in this
work. All values are adapted taken from Ref. [82]. Calculated total magnetic moments are given per formula unit (f.u.). TC are experimental Curie temperatures
determined for bulk samples.

3.3 Magnetic properties of Co2 -based Heusler compounds

37

Magnetic moment per atom m (B)

2.5
2.0
1.5
1.0
0.5
0.0
6

8
9
10
Valence electrons per atom

11

Figure 3.5: Slater-Pauling curve for selected Co2 -based Heusler compounds adapted
from Ref. [82]. Magnetic moments of 3d transition metals and their binary alloys are
given for comparison.

three p states, which are much lower in energy (5-6 eV below EF ) and do
not contribute to the formation of the band gap. Thus, in total the Co2 -based
Heusler compounds possess 12 occupied minority states per unit cell. The
total magnetic moment is given by the number of excess majority electrons
with respect to minority electrons. Introducing NV as the total number of
valence electrons in the unit cell, the magnetic moment m of the Co2 YZ Heusler
compounds obeys the rule
m = NV 24 .
(3.1)
This is known as the generalized Slater-Pauling rule [107], which is an analogue
to the Slater-Pauling behavior of the binary transition metal alloys [1719].
The number 24 in Eq. (3.1) is the sum of completely occupied minority and
majority states, i.e. twice the number of the occupied minority states. Since in
the Co2 YZ Heusler compounds NV is an integer number, an integer magnetic
moment follows from this rule. However, this is no longer true for the quaternary compounds such as CCFA, where the non-integer site occupancy results
in the non-integer values of the magnetic moments.
The Slater-Pauling behavior of Co2 YZ Heusler compounds is shown in
Fig. 3.5 in comparison with 3d transition metals and their binary alloys. Note
that here the mean magnetic moment per atom is displayed as a function of
the averaged number of valence electrons per atom. There are two clearly
distinguishable regions in the Slater-Paling curve, namely the one with a positive slope for NV . 8 e and the other one with a negative slope. Whereas

38

Half-metallic Heusler compounds


the former describes the systems with localized magnetic moments, the latter
is the range of itinerant magnetism [107, 108]. Thus, the Co2 -based Heusler
compounds fall into the region of systems with localized magnetic moments.

3.3.2

Formation of local magnetic moments

A microscopic mechanism resulting in the formation of localized magnetic moments in the Co2 -based Heusler compounds was first given by K
ubler and collaborators [83]. It should be noted, however, that in these early calculations,
where different X2 MnZ compounds where investigated, a confinement of magnetization to the Mn atoms was indicated. The localized nature of the magnetic
moments is explained by the large exchange splitting of the Mn spin-up and
spin-down d states, which results in only spin-up d states being supported by
the Mn atom. The spin-down electrons are almost completely excluded from
the Mn sites. This localized exclusion is equal to a formation of a localized
region of magnetization.
It was later recognized that both Co and Y atoms significantly contribute
to the total magnetic moment while the contribution of Z atom is negligible.
Nevertheless, the Z atom plays an important role for the formation of the
individual magnetic moments at Co and Mn sites. An overview of calculated
total and element-specific magnetic moments of ternary Heusler compounds
studied in this thesis is given in Table 3.2. The compounds are listed according
to their total number of valence electrons per unit cell.
Except for CCFA , the total magnetic moments of the listed Co2 -based
Heusler compounds follow the Slater-Pauling rule (Eq. (3.1)). The magnetic
moments localized at Al and Si sites (not shown in the table) are at least one
order in magnitude smaller than those of other constituent elements, and have
a negative sign [86]. The small moments found at sp sites are mainly due to
a polarization of these atoms by the surrounding, magnetically active Co and
Mn atoms. The coupling mechanism between the Co magnetic moments and
between the moments of Co and Mn sublattices will be addressed in more detail
in the next Subsection.
Table 3.2 clearly shows that there is a substantial magnetic moment located
on the Co site. This moment arises from the two unoccupied bands in the
minority conduction bands, i.e. the double-degenerate eu states in Fig. 3.3.
Replacement of Al by Si results in an increase of localized magnetic moments
on both Co and Mn or Fe sites, which is due to the filling of the majority
bands with an additional electron. The interesting point is, however, that Co
atoms contribute much more to the magnetic moment in the Si compounds
with respect to the Al compounds. The reason for this behavior is a stronger

3.3 Magnetic properties of Co2 -based Heusler compounds


bonding interaction between Co and Si as compared to the Co-Al case due to
the higher electro-negativity of Si [82].

3.3.3

Exchange interaction between magnetic moments

Recently, extensive theoretical studies have been reported by Kurtulus et al.


and Sasioglu et al. elucidating the coupling mechanism of the localized magnetic
moments in Co2 -based Heusler compounds [81, 109]. In the case of Co2 MnSi,
both groups found that the interaction between the nearest Mn and Co atoms
provides the leading interaction for the stabilization of the ferromagnetism in
this system. The magnetic moments localized at Mn and Co sites are coupled
ferromagnetically to each other via the direct exchange coupling mechanism.
This is in contrast to the mechanism proposed by K
ubler in his early theoretical
investigations where the Mn-Mn interaction was proposed to be responsible
for the appearance of ferromagnetism in Co2 MnSi [83]. In addition to the
leading Co-Mn interaction, the Mn-Mn interaction and the coupling of the
nearest Co magnetic moments located on different Co sublattices were found
to be important for the stabilization of ferromagnetism in Co2 MnSi. These
contributions, however, are about one order in magnitude smaller than the
leading Co-Mn interaction. Although the Mn atoms have a larger magnetic
moment, the ferromagnetic interaction between Co atoms is stronger, which is
explained by a smaller distance of the Co atoms as compared to that of the Mn
atoms. Due to a large Mn-Mn separation (> 4
A) there is no significant direct
exchange interaction. The coupling of Mn moments is believed to be mediated
by a RKKY interaction, since the characteristic oscillations of the exchange
parameter are found in dependence of the interatomic distance. Co atoms
positioned in the same sublattice exhibit a similar behavior and contribute as
well to the stabilization of ferromagnetism in the Co2 MnSi Heusler compound.
Results published recently by Thoene et al. are in agreement with the above
described coupling mechanism in Co2 MnSi [110].
For other Co2 -based Heusler compounds considered in this thesis, there are
no such detailed studies of the interatomic exchange interactions in the literature. However, since they exhibit similar structural and electronic properties,
the exchange mechanisms in these materials can reasonably be expected to be
similar to those described for the Co2 MnSi Heusler compound. This assumption is supported by the results of investigations carried out for Co2 CrAl [81],
which is not studied here. It was shown that the same coupling mechanisms are
responsible for the stabilization of the ferromagnetic state in this compound.
The strength of the exchange interaction is found to be dependent on the
lattice constant. In particular, the lattice contraction leads to an enhancement

39

40

Half-metallic Heusler compounds


of intra- and intersublattice exchange constants [81,109]. Therefore, an increase
of the exchange constants is expected in the investigated series of Co2 -based
Heusler compounds for those with a smaller a value, which we have indeed
observed (see Chapt. 5).

3.3.4

Curie temperature

Due to the robustness of the ferromagnetic order, which arises from the strong
inter- and intrasublattice exchange interactions, the Co2 -based Heusler compounds exhibit remarkably high Curie temperatures TC (Table 3.2). Their TC
values are comparable to those of pure 3d metals, even though a non-magnetic
element is present in the lattice. When considered as a function of the number of
valence electrons, the TC of Co2 -based Heusler compounds show a linear dependence similar to the Slater-Pauling behavior discussed in Sect. 3.3.1. The same
trend was reported for TC values as a function of the magnetic moment [108].
Figure 3.6 illustrates these two dependencies.
The origin of the linear dependence between TC and NV found in Co2 -based
Heusler compounds was carefully investigated by K
ubler et al. by means of ab
initio calculations [3]. The authors estimate the Curie temperature using the
expression
"
#1
2X 2 1 X 1
L
,
(3.2)
kB TC =
3 N qn jn (q)

1200

(a)

Co2FeSi
Co2FeAl

1000

Co2MnSi

800

Co2MnSn

Co2MnAl

600
Co2CrGa

400

Co2VGa
Co2CrAl

200
Co2VSn

Co2TiAl

25

26

27

28

1400

Curie temperature (K)

Curie temperature (K)

where L is the local magnetic moment at site and jn (q) are three exchange

Co

(b)

1200

Fe

1000
800

Ni

600

Co2YZ
X2YZ; X=Co

400

Fit Co2YZ

200

Elements

0
29

Valence electrons

30

0.0

0.5

1.0

1.5

2.0

2.5

Magnetic moment per atom (B)

Figure 3.6: (a) Measured Curie temperatures of different Co2 -based Heusler compounds in dependence of the number of valence electrons per unit cell [3]. (b) Measured Curie temperatures as a function of the magnetic moment per atom. TC values
of the 3d metals Fe, Co, and Ni are shown for comparison as well [108].

3.3 Magnetic properties of Co2 -based Heusler compounds


functions, which describe the inter- and intrasublattice exchange interactions
described in the previous Section. The exchange functions are calculated using
the frozen-magnon approach [111]. The expression in the square brackets in
Eq. (3.2) represents an exchange average. Hence TC is determined by both the
magnetic moment and the average exchange value. The linear dependence of TC
on the number of valence electrons is the result of the trends obtained for both
of these factors. The same interplay is responsible for the linear trend found
for TC and the magnetic moment. For the discussion of experimental results
in Chapt. 5 it is interesting to note that in K
ublers calculations the exchange
average is found to exhibit a nearly linear increase as a function of valence
electrons in the range of 28 to 30 electrons, i.e. the range of the compounds
studied in this thesis.

3.3.5

Orbital magnetism

Because of their cubic structure Heusler compounds have been considered as


materials with completely quenched orbital moments for a long time. Recently,
however, noticeable orbital moments were found in x-ray magnetic circular
dichroism (XMCD) studies for Co2 FeAl and CCFA [93, 112] as well as for
Co2 MnGe [113] and Co2 FeSi [70], leading to an increased interest in the orbital magnetism of Co2 -based Heusler compounds from the theoretical point
of view. Several computational studies were dedicated to the determination of
element-specific magnetic moments in Co2 YZ compounds [96, 114]. However,
none of them could give a satisfactory agreement with the experimental results:
the calculated orbital moments are usually 2-4 times smaller compared to the
experimental values. The discrepancy is thought to arise from difficulties in a
proper treatment of the orbital magnetism in ab initio calculations on the one
hand and systematic artifacts in the analysis of XMCD data on the other. It
is worthwhile to mention, however, that the orbital moments are found to be
larger in Fe containing compounds than in those with Y=Mn. Since the magnetocrystalline anisotropy is directly related to the orbital magnetic moments
(see Sect. 2.1.2), a larger anisotropy is therefore expected in the investigated
compounds with Y=Fe.

41

42

Half-metallic Heusler compounds

Chapter 4
Experimental methods
In this Chapter, the experimental techniques which were used for the analysis and modification of the structural and magnetic properties of investigated
Heusler films are described (namely x-ray diffraction, MOKE magnetometry
and microscopy, Brillouin light scattering spectroscopy, and ion irradiation).

4.1

X-ray diffraction

X-ray diffraction (XRD) is a commonly used method for the investigation of


the structural properties of thin films. There is a number of textbooks dealing
with the fundamentals of this experimental technique, and it is not our goal
to give a complete representation here. In the following, only aspects of x-ray
diffraction which are necessary for the understanding of experiments carried
out with Heusler films will be addressed. A comprehensive overview of x-ray
diffraction techniques can be found for example in Ref. [115].

4.1.1

Generalities

Let us consider the x-rays reflected from neighboring crystal planes in a crystalline sample, as is schematically shown in Fig. 4.1. The resulting reflections
interfere with each other. A constructive superposition of the beams occurs
whenever the path difference is an integer of the x-ray wavelength . For a
given angle between the incident beam and the crystal plane characterized
by Miller indices (hkl), the path difference of the reflected neighboring beams
is 2dhkl sin , dhkl being the spacing between the scattering planes. This leads

44

Experimental methods

d
}

dsinq
Figure 4.1: Diffraction from neighboring planes of the periodical crystal lattice.

us to the well known Bragg equation


2dhkl sin = n .

(4.1)

This equation predicts the peak positions in a XRD pattern, and can be used
to determine the lattice constant for a chosen lattice direction [hkl] from a
recorded Bragg scan.
Not every set of Miller indices for which Eq. (4.1) is satisfied provides a finite
intensity in the Bragg pattern. For some (hkl) combinations, annihilation of
the associated peak occurs due to the destructive interference caused by the
internal structure of the unit cell. The influence of the internal structure of
the unit cell on intensity of reflections is taken into account by introducing the
structure factor [115]
F =

N
X

fn exp[2i(hun + kvn + lwn )] ,

(4.2)

n=1

where the summation extends over the total number of atoms N in the unit
cell. The intensity of the beam diffracted by all the atoms of the unit cell is
proportional to F 2 . In Eq. (4.2) (un , vn , wn ) are the relative coordinates of the
n-th atom and fn denotes its atomic scattering factor (also called form factor ).
The latter is determined by the number and the distribution of electrons in
the atom, and depends on the wavelength and the scattering direction with
respect to the crystallographic axes.
Structure factors F are generally sensitive to the disorder within a unit cell.
The intensities of Bragg reflections can therefore be used to characterize the
type and the degree of disorder present in a sample. Quantitatively, the degree
of disorder is usually described by the long-range order parameter S, which in
case of binary alloys is defined as [115]
S=

rA FA
,
1 FA

(4.3)

4.1 X-ray diffraction

45

where rA is the fraction of A sites occupied by the A atoms and FA is the


fraction of A atoms in the alloy. S = 1 corresponds to a completely ordered
sample, whereas no long-range order is present for S = 0. In the latter case
however the sample is not necessarily completely disordered. The sample can
consist of domains wherein the atoms form a superlattice, but the occupation
of sublattices differs from one domain to another. This is known as shortrange order, and its detection requires local probing techniques such as x-ray
absorption spectroscopy [14, 116] or nuclear magnetic resonance [71, 72, 117].

4.1.2

Application to Co2 YZ Heusler compounds

The L21 structure of Co2 YZ Heusler compounds gives rise to non-zero Bragg
reflections only when the indices h, k and l of the reflecting planes are either all
even or all odd. The planes with even indices are additionally subdivided into
those with odd or even ratios of (h + k + l)/2 such that three types of reflecting
planes are generally distinguished with the following structure factors [54]:
F (111) = 4|fY fZ | ,
F (200) = 4|2fCo (fY + fZ )| ,

(4.4)

F (220) = 4|2fCo + fY + fZ | .
F (111) and F (200) contain difference terms and are thus sensitive to the disorder in the unit cell. The corresponding reflections are called superlattice or
superstructure reflections. In contrast, F (220) is the sum of all atomic scattering factors. Therefore, this type of reflection is independent of order and is
referred to as fundamental reflections [115].
As it has been pointed out in Sect. 3.1, various types of disorder might occur
in the Co2 -based Heusler compounds, affecting the two groups of superlattice
reflections in a different way. Thus, it is not possible to characterize the disorder
by a single parameter as was the case for binary systems. In general, a set of
order parameters must be introduced as discussed by Webster [54]. However,
if one type of disorder predominates, the state of disorder in Heusler samples
can be described in terms of a preferential disorder parameter together with
a random disorder parameter S [5]. In the case of B2 preferential disorder (i.e.
exchange of Y and Z atoms), S is defined as for the binary alloys (Eq. (4.3)) and
specifies the fraction of Y atoms occupying Z sites. The (200) superlattice
reflections are unaffected by this type of disorder since the form factors of Y and
Z atoms are summed in the corresponding structure factor F (200) (Eq. (4.4)).
Their intensities are only reduced by a factor S 2 . The (111) reflections are,
however, sensitive to the disorder between the Y and Z sublattices, since F (111)

46

Experimental methods
contains the difference of the corresponding form factors. Their intensities are
reduced by a factor (1 2)2 S 2 [5]. In case of a perfect B2 structure (i.e.
= 0.5 and S = 1) the (111) superlattice peaks vanish, while the intensity of
the (200) peaks remains unchanged when compared to a perfect L21 structure.
A complete disorder (i.e. S = 0) would lead to vanishing intensities of (200)
reflections and corresponds to the A2 structure.
Both disorder parameters and S can be determined by comparing the
experimental intensities of superlattice lines with the intensities theoretically
expected for a perfectly ordered L21 structure:
S 2 = (I200 /I220 )exp /(I200 /I220 )theory ,
(1 2)2 S 2 = (I111 /I220 )exp /(I111 /I220 )theory .

(4.5)

The normalization to the fundamental (220) reflection in the above expressions


is necessary to rule out the influences of the experimental setup. The use of
the (220) reflections for the normalization is based on the fact that they provide the strongest intensity, which helps to reduce the experimental error when
evaluating Eq. (4.5). Some authors, however, also use the (400) fundamental
reflection for the normalization of the (200) intensity. This has the advantage
that (400) reflections are observed in the normal -2 scan of a single-crystal
thin film. For the detection of (220) reflections a pole figure must be recorded,
which requires more effort and measurement time. It is worth mentioning that
(111) superlattice reflections with high intensities do not necessarily mean a
high degree of the L21 order as might be concluded from the previous discussion. An increased normalized intensity of the (111) reflections in comparison
to the corresponding theoretical value can also occur if the Co-Y preferential
disorder (i.e. DO3 type of disorder) is present in the sample. In this case, a
simultaneous decrease of the normalized (200) intensities must be observable
as well.
When applying the above method for the analysis of single crystalline Heusler
films several factors have to be kept in mind. First of all, a reference sample
with a perfect L21 structure is not available, so the experimental intensities
must be compared to simulated ones. Dedicated software, such as PowderCell [118] and CrystalDiffract [119] used in this work, allow for the calculation
of powder diffraction patterns only. Therefore, total integrated intensities of
the Bragg reflections have to be determined experimentally for an accurate
comparison with the powder pattern simulations. Moreover, the simulation
software does not take into account the fact that the intensities of the diffraction spots are influenced by geometrical effects in the sample. These mainly
result from the dependence of the sample volume which contributes to the scattering intensity on the incidence angle (a smaller value of results in a larger

4.1 X-ray diffraction

47

scattering volume). Therefore, additional correction factors of the form sin


must be introduced into the Eq. (4.5). Finally, it should be pointed out that
the above described formalism provides only a rough estimation of the degree
of order since it assumes only one type of preferential disorder in the sample.
This formalism, however, is widely-used in the literature and was therefore
partially applied in the present thesis. Generally, careful simulations including
various types of disorder are necessary for a quantitative analysis of the crystal
structure.

4.1.3

X-ray diffractometer

XRD structural characterization of the Co2 YZ Heusler thin films was carried
out at their places of fabrication, i.e. at the Johannes Gutenberg University
of Mainz or at the Tohoku University in Sendai. Figure 4.2 shows the x-ray
diffractometer available in the group of Prof. Dr. Y. Ando in Sendai. It is a
4-circle Bruker D8 Discover model equipped with a 2D x-ray detector. The
rotational degrees of freedom are indicated by the dashed lines. The radiation
source is a Cu anode providing Cu-K line with =1.54
A.
For the Co2 YZ Heusler films, (200) superlattice reflections are specular reflections, i.e. they are found in the plane formed by the incident beam and the
normal of the film surface. Therefore, they are easily accessible in the -2

sample
holder
j

detector

x-ray
source
c
w

q
Figure 4.2: Bruker D8 Discover x-ray diffractometer located at the Tohoku University
in Sendai, Japan, with indicated rotational degrees of freedom (photograph kindly
provided by S. J. Hermsdorfer). For the adjustment of the sample, -, - and -circles
are used. The source and detector are positioned on the -circle.

48

Experimental methods
scans. (111) as well as (220) refections are off-specular and pole figures are
required for the analysis of their intensities. To measure the pole figures, the
sample is first adjusted by rotating the sample holder around the -axis in order to bring the particular reflection into the specular plane. Thereafter, the
diffracted intensities are detected in a chosen range of for different values of
such that all the equivalent reflections are collected in this measuring procedure. For the operation, display and data processing Brukers GADDS software
was used. Similar measurements were performed in Mainz.

4.2

Kerr effect techniques

In this Section two experimental methods employed in this thesis are introduced
which are based on the magneto-optical Kerr effect discussed in Sect. 2.2. One
of them is static MOKE magnetometry, which was used to record magnetization reversal curves. The other is MOKE microscopy which we applied to the
elucidation of the microscopic details of the reversal mechanism in thin Heusler
films.

4.2.1

MOKE magnetometry

The employed MOKE magnetometer is a home-made apparatus operated using


a software called Aurora [120,121]. In the following, we give a basic description
of the setup, as well as the measurement principles for standard linear MOKE.
Thereafter, measurement procedures will be presented which allow the detection
of second order contributions to the MOKE signal.
Experimental setup
A schematic representation of the MOKE setup is given in Fig. 4.3(a). A diode
laser with a wavelength of =670 nm is used as a light source. The beam passes
through a polarizer where it becomes s-polarized, and is then focused onto the
sample. The incidence angle is =45 and the laser spot size is 300 m.
After the reflection on the sample, the beam is collimated and focused onto
the diodes of a photodetector. On its way to the detector, the reflected light
passes a Wollaston prism where it is divided into two orthogonally polarized
beams with intensities I1 and I2 . The normalized difference of these intensities
I/(I1 +I2 ), where I = I1 I2 , constitutes the detected signal. The Wollaston
prism and the photodiodes form a fixed detector unit, which is rotatable around

4.2 Kerr effect techniques

49
photodiodes
detector
unit

polarizer

H7
]

[100

Wollaston
prism

(a)

H5

[110]

H6

laser

H4

H3

(b)

H8

H1
H2

Figure 4.3: (a) Schematic representation of the Kerr magnetometer. (b) Sketch of the
8-directional method used for the determination of the QMOKE signal in saturation.
The arrows represent the directions of the externally applied magnetic field.

the optical axis of the setup. For I 0, the normalized differential signal is
proportional to the Kerr rotation.
Before the beginning of a measurement, an automatic adjustment of the
photodetector is carried out. In this adjustment procedure, the detector unit is
rotated until the difference signal I = 0. With the applied magnetic field the
rotation of the elliptical polarization of the reflected beam is changed by the
Kerr angle as discussed in Sect. 2.2. This change of polarization is registered as
a non-zero intensity difference I by the photodiodes. Magnetization reversal
curves can thus be measured by successively changing the values of the magnetic
field. Since the sample is mounted on a sample holder which can automatically
be rotated and horizontally moved, measurements of the magnetization reversal
can be carried out at different in-plane orientations and positions of the laser
beam on the sample.
Detection of second order contributions
Second order contributions to the MOKE signal (also called quadratic MOKE
or QMOKE) are accessible in two different ways. On the one hand, they can be
obtained from the Kerr rotation loops measured in the longitudinal geometry
described above by the antisymmetrization of the measured loops [A2,122]. On
the other hand, QMOKE loops can be directly measured at a nearly normal
angle of incidence. The former method utilizes the fact that second order
contributions lead to an asymmetric shape of the loops, i.e. different shape of
the loop branches for increasing and decreasing H. Since the first order MOKE
is symmetric in H, both these contributions can be separated using the relation
sym/asym = [inc (H) dec (H)]/2, where inc/dec denotes the loop branch for
increasing or decreasing H, respectively. The latter procedure exploits the
fact that the LMOKE vanishes at = 0 and the detected signal is only

50

Experimental methods
proportional to the second order effect in case of an in-plane magnetized sample.
In the measurements performed within this thesis the angle of incidence was
=0.5 .
For a quantitative determination of the strength of the second order contributions proportional to ML MT and ML2 MT2 , an 8-directional method introduced in Ref. [37] was employed. The basic principle of this method is that a
magnetic field which is high enough to saturate the sample is applied in eight
different directions as is sketched in Fig. 4.3(b) and the Kerr rotation is measured. The two distinct QMOKE contributions can be calculated from the
corresponding Kerr rotations (Hi ) using the expressions
sat,ML MT = [(H1 ) + (H5 ) (H3 ) (H7 )]/4 ,

(4.6)

sat,M2L M2T = [(H8 ) + (H4 ) (H2 ) (H6 )]/4 .

(4.7)

Note that this method also yields the LMOKE Kerr rotation in saturation
sat,ML = [(H8 ) (H4 )]/2. Application of the 8-directional method at different in-plane sample orientations (Fig. 4.3(b)) provides an insight into the
anisotropy of the second order effect.

4.2.2

MOKE microscopy

Kerr microscopy investigations of Heusler films were carried out at the LeibnitzInstitut f
ur Festkorper- und Werkstoffforschung (IFW) Dresden in the group
of Dr. R. Schafer. A schematic illustration of the microscope used is shown
in Fig. 4.4. The magnetic contrast primarily arises from different signs of
Kerr amplitudes resulting from domains with different directions of magnetization [24]. An optimum contrast is usually achieved for nearly crossed polarizer
and analyzer since the intensity of the regularly reflected light is minimized
in this case [24]. Sensitivity to the longitudinal component of magnetization
is obtained when the polarizer is set to s-polarization while the angle of the
analyzer is close to p-polarization. The transverse component of magnetization
gives rise to the magnetic contrast when polarizer and analyzer are set to 45 ,
respectively, relative to the plane of light incidence. However, since the Kerr
amplitudes are small and a certain background intensity is always present, the
magnetic contrast remains very low. This problem is effectively overcome when
digital image acquisition and processing are used. In a standard measurement
procedure, which was used in this thesis as well, first an image of a magnetically saturated state is recorded. This reference image is then subtracted from a
subsequently recorded image containing magnetic contrast. In this way structural and non-magnetic contributions to the contrast are largely eliminated.

4.3 Brillouin light scattering spectroscopy

Figure 4.4: Schematic picture of the employed Kerr microscope (adapted from [24]).

To visualize the distribution of both longitudinal and transverse components


of magnetization in a particular domain pattern, the microscope settings have
to be modified (from sp-configuration to 45 -configuration) while keeping
the magnetization state of the sample. For a correct background subtraction
a second reference image has to be recorded at the end of the image sequence,
which is related to the modified microscope setting.

4.3

Brillouin light scattering spectroscopy

Brillouin light scattering spectroscopy is based on the interaction of photons


with magnons, as introduced in Sect. 2.3.3. The magnon energy is basically
measured as a function of the frequency shift of the scattered photon. Depending on the properties of investigated materials two different scattering
geometries can be used in the experiment which will be briefly described in the
following. Thereafter, details of the BLS spectrometer employed in this thesis
will be addressed.

4.3.1

Measurement geometries

The two most frequently used BLS scattering geometries are the so-called forward scattering and backward scattering geometry illustrated in Fig. 4.5. In
the forward scattering geometry (Fig. 4.5(a)) the laser beam is focused onto a
transparent sample and is collected behind it after passing the medium. This
geometry allows only for the detection of excitations with kk smaller than the

51

52

Experimental methods
(b)

(a)

ki

j
ki

Dkmax=kisinj

Dkmax=2kisinj

Figure 4.5: Schematic representation of commonly used Brillouin light scattering


geometries. (a) shows the scattering in forward, (b) in backward direction.

wave vector of the incident photon. In case of opaque samples backward scattering geometry is used (Fig. 4.5(b)) which makes a larger range of wave vectors
accessible. For grazing incidence the maximum momentum ~kk transferred to
(or from) the magnon is twice the momentum of the incident photon. Knowing
the angle of incidence and the wavelength of the exciting laser the transferred
wave vector kk can be easily calculated using the expression:
kk = ki sin =

4
Laser

sin .

(4.8)

In all the BLS measurements performed in this work the backward scattering
geometry was employed.

4.3.2

Brillouin light scattering spectrometer

The frequency selective part of the BLS spectrometer used in this work is a
tandem Fabry-Perot interferometer (FPI) developed by Sandercock [123, 124].
Its operating mode is described in the second part of this Section. In the
beginning, the details of the experimental setup are explained.
Experimental setup
Figure 4.6 provides a schematic view of the BLS setup used in this work.
A diode pumped, frequency doubled Nd : YVO4 laser with a wavelength of
=532 nm is used as a light source. A beam splitter divides the laser beam into
two parts, one of which (2 %) serves as a reference beam for the stabilization
of the interferometer and the other one (98 %) is focused onto the sample with
an objective lens. The backscattered light (elastic and inelastic contributions)
is collected by the same objective lens and sent through the interferometer. On
its way from the sample to the interferometer the beam passes polarizer which

4.3 Brillouin light scattering spectroscopy

53

solid state laser


reference
beam

spatial filter

spectral filter

photodetector
shutter
system

Computer control
- stabilization
- data acquisition
and processing
- visualization

FPI 2

scan stage

FPI 1

spatial filter

polarization
analyzer

sample

Figure 4.6: Schematic view of the BLS setup used in this work.

is set perpendicular to the initial polarization direction of the laser beam and
suppresses the light stemming from the interaction with phonons. This is possible since the polarization of light remains unaffected by the scattering from
phonons, while light scattered by a magnon is rotated by 90 . Before entering
the interferometer the backscattered light passes through a spatial filter where
the background noise is suppressed. A computer-controlled double shutter system regulates the light flow entering the interferometer. It assures that in the
frequency range of the laser light a much weaker reference beam is sent through
the interferometer, thus protecting the photomultiplier from the high intensity
of the elastically scattered light. After frequency selection in the tandem FPI,
which will be discussed in the following paragraph, the beam passes through a
second spatial filter for additional background suppression, and is detected by
a photomultiplier. Data acquisition and visualization as well as stabilization
of the interferometer are carried out by using the TFPDAS3 programm [125].
Details of the interferometer alignment and stabilization procedure will not be
discussed here and can be found for example in Ref. [126].
Tandem Fabry-P
erot interferometer
Before addressing the operating mode of the tandem FPI, which consists of two
FPIs, we briefly summarize the physics behind a single FPI. The FPI consists

54

Experimental methods
of two plane and highly reflecting mirrors mounted accurately parallel to each
other. The intensity of light transmitted through the FPI at perpendicular
incidence is described by the transmission function [126, 127]:
T = T0

1
,
1 + (4F 2 / 2 ) sin2 (/2)

(4.9)

and is determined by the interference of light multiply reflected between the


two reflecting surfaces. In Eq. (4.9), T0 is the maximum possible transmission
determined by losses in the system and F is the finesse which mainly depends
on the reflectivity of mirrors. As is shown in Eq. (4.9), the FPI provides the
maximum transmission when the phase difference between two interfering
beams is an integer of 2. Since the path difference s of two interfering beams
equals the double optical distance d between the mirrors and
= 2

s
,

(4.10)

where is the wavelength of light, for a given mirror spacing d the interferometer will transmit only those wavelengths which satisfy the condition:

d=n ,
2

(4.11)

with n being an integer. Varying the spacing d allows thus for scanning the
intensity of light at different wavelengths. However, an unambiguous interpretation of the spectrum is impossible because of the periodicity of the transmission
function (Eq. (4.9)). One way to solve this problem is to increase the spacing
of two neighboring transmission maxima (also called the free spectral range
FSR) by decreasing d. This results in an increase of width of the transmission
maxima since both parameters are related via
F =

(4.12)

where F is the finesse of the interferometer. The consequence is a decrease in


resolution.
An unambiguous frequency determination without any resolution deterioration becomes possible with a tandem arrangement of two etalons developed by
Sandercock [123, 124]. In this arrangement, the light passes in series through
two etalons FPI1 and FPI2 (Fig. 4.6), which are tilted by the angle with respect to each other. The right mirrors of both FPIs are mounted on a common
translation stage which is translated, i.e. the mirror spacing is varied. Due to
the relative orientation of both etalons a change of optical distance in the first

4.4 Ion irradiation

55

Transmission

FPI 1

FPI 2

Tandem FPI

reference
signal
inelastic
signal

M ir r o r s p a c in g d
Figure 4.7: Transmission functions of both single and the tandem FPIs. The higher
order transmission is strongly suppressed. The inelastic signal is normally much
weaker compared to the reference signal.

FPI by d results in a change of mirror spacing of d sin in the second FPI.


The higher orders of the second etalon are thus transmitted later through FPI2
than this is the case for FPI1. This results in a strong suppression of higher
order transmission maxima as is shown in Fig. 4.7. However, the neighboring
maxima do not disappear completely since the transmission function (Eq. (4.9))
does not abruptly becomes zero.
The weak intensity of the inelastic signal requires a high contrast of the
interferometer. In the tandem FPI used in this work a high contrast is achieved
by multiple passes of the beam through both etalons realized using a system
of retroreflectors and mirrors (Fig. 4.6). As is shown in Fig. 4.6, the beam
goes six times though the FPIs before passing a spectral filter and entering the
detector. The performance of the tandem FPI is very sensitive to the stability
of the mirror positions. An active stabilization mechanism implemented into
the TFPDAS3 program guarantees the parallel alignment and correct spacing
between the mirrors of both etalons is maintained.

4.4

Ion irradiation

Ion irradiation, and irradiation with light He+ ions in particular, has been
demonstrated to be an excellent tool for tailoring of magnetic properties of

56

Experimental methods
technologically relevant thin films and multilayers. In this context, most experiments have focused on the perpendicularly magnetized Co/Pt systems [128131]
and exchange-biased systems [132138]. In the former case, a rotation of the
out-of-plane magnetization into an in-plane direction was observed due to ioninduced modification of the interface anisotropy, while in the latter case, both
the magnitude and direction of the exchange bias field could be tuned by He+
irradiation. In addition, light ion irradiation has been applied to in-plane magnetized ferromagnetic films [12, 139141], structures with interlayer exchange
coupling [142], spin valve systems [143,144] and magnetic tunnel junctions [145].
A comprehensive overview of applications of He+ irradiation as well as of irradiation with different ion species is given in Refs. [146, 147]. In the field of
heavy ion irradiation, applications involving Ga+ ions, which are a standard
ion source in the focused ion beam (FIB) systems, are particularly to mention. Most of the above applications have in common that the modification of
magnetic properties results from ion-induced disorder and interfacial mixing.
However, Ravelosona et al. have shown that an enhancement of chemical order
can also be induced by He+ irradiation [12, 140]. In both disordered and partially ordered FePt thin films, they observed an increase of the L10 long-range
order parameter as a result of bombardment with 130 keV He+ ions at elevated
temperatures [12]. An improvement of ordering after He+ irradiation was also
reported by the same group for similar FePd thin films [140].
Based on these results, the effect of He+ irradiation on Heusler films was
investigated in the present thesis. In particular, the feasibility of an ion-induced
enhancement of order in Heusler compounds was studied. In view of the potential micro-patterning application for Heusler films, the effect of Ga+ irradiation
on their properties was also studied using a FIB. The details concerning the
energy of ions, applied fluences etc. will be given in Chapt. 6. Here, we introduce the basics of the ion-solid interactions with a special attention to the
differences between the irradiation with light and heavy ions. Subsequently, the
experimental setups used for the ion irradiation of the Heusler films are briefly
described.

4.4.1

Ion-solid interactions

An energetic ion moving through a solid loses its kinetic energy due to the interaction with the target material, and finally implants itself into the irradiated
specimen. Depending on the energy and atomic number of the incoming ion
and the composition of the target material, different mechanisms might play a
dominant role for the energy loss. Basically, two different energy-loss mechanisms are distinguished: (1) elastic nuclear collisions with target atoms, in

4.4 Ion irradiation


which ions kinetic energy is transmitted into translatory motion of the target
atom according to the mass ratio of the collision partners, and (2) inelastic electronic collisions, in which the electrons of the target atoms are excited, possibly
to the point of ionization [148]. These two combined energy-loss channels define
the stopping power for the incoming ion in the solid. This stopping power is
defined as the total energy loss per unit length perpendicular to the surface of
the irradiated specimen



dE
dE
dE
=
+
,
(4.13)
dx tot
dx n
dx e
and which determines the penetration depth or range R of the ion given by
Z 0
dE
R=
.
(4.14)
E0 dE/dx
In the above equation, E0 denotes the initial kinetic energy of the impinging
ion.
Nuclear stopping predominates for ions with low energy and high atomic
numbers. This process can involve large energy transfers and strong deviations from the ions initial incident trajectory. If the energy transferred to the
target atoms is large enough, these atoms can be displaced from their lattice
sites. This results in vacancy and interstitial pairs lattice defects, also called
Frenkel defects. The displaced atoms can undergo further collisions on their way
through the material, giving rise to collision cascades involving the creation of
various other types of lattice defects. Collision cascades occurring within a limited volume are usually referred to as dislocation spikes and are characterized
by a strong lattice heating in the outer parts of the cascade, which can even
result in the local melting of the target material. In this context one therefore
also uses the term thermal spike. Electronic stopping becomes important for
ions with high energy and low atomic numbers. Due to smaller energy transfers per collision event, the lattice disorder is significantly smaller compared
to the case of predominant nuclear stopping. Furthermore, extended collision
cascades are absent.
The relative importance of these two energy-loss mechanisms as a function of
ion kinetic energy is shown in Fig. 4.8 for Ga+ and He+ impinging on Co2 MnSi.
The curves were calculated using the software SRIM (Stopping and Range of
Ions in Matter) [149, 150], which is a standard tool for simulations of ion-solid
interactions. For both cases, energy-dependent ion ranges were calculated and
are also included in Fig. 4.8. The ion range is defined as the depth at which the
ion is stopped in the irradiated specimen. One clearly sees that light ions, for
which the electronic energy loss predominates, travel a much larger distance in
the solid before they come to rest.

57

58

Experimental methods

30

dE/dx n

150

100

100
50
50
0

dE/dx e

10

20

30

40

0
50

Ion energy (keV)

Energy loss (eV / D)

150

Ion range (D)

Energy loss (eV / D)

200

25

2000

He in Co2MnSi

6
5
4
3

1500

2
1

20

15

dE/dx e

0.0 0.5 1.0 1.5 2.0 2.5 3.0

1000

10
500
5
0

(b)

dE/dx n

10

20

30

Ion range (D)

200

250 (a) Ga in Co2MnSi

40

0
50

Ion energy (keV)

Figure 4.8: Nuclear and electronic energy losses as well as ion ranges in Co2 MnSi for
(a) Ga+ with atomic number Z = 31 and (b) He+ with Z = 2 calculated with SRIM.
The inset in (b) is a magnification of the low-energetic range.

The energy deposited into the solid which is not used for the creation of
defects is dissipated as lattice vibrations, i.e. heat. This can lead to an enhanced diffusivity and recombination of Frenkel defects, resulting in a partial
recovery of defects. In alloys, a thermally activated phase transformation can
also occur. In this context it should be noted that the description of ion-solid
interactions presented so far is restricted to irradiation-induced effects inside
the solid. However, depending on the ion species and the target material surface
effects such as sputtering can become important as well. This particularly applies to irradiation with heavy ions and large ion fluences. A detailed overview
of ion irradiation effects on the properties of a solid is given, for example, in
Ref. [148].

4.4.2

Experimental setup

Irradiation of Heusler films with He+ ions was carried out by Dr. J. Fassbender
and collaborators at the Forschungszentrum Dresden-Rossendorf employing a
DANFYSIK low energy implanter of the Series 1090. Irradiation with Ga+
ions was performed by B. Reuscher using focused ion beam equipment of the
type FEI ALTURA 865 Dual Beam located at the Institut f
ur Oberflachenund Schichtanalytik (IFOS), Kaiserslautern.

Chapter 5

Experimental results I Magnetic properties of


Co2-based Heusler films

This Chapter is organized as follows. In the beginning, an overview of the investigated samples will be given, in which the sample structure is described including the information about the method of fabrication as well as film thicknesses
and crystallographic properties. Subsequently, the results of XRD characterization of Co2 MnSi films will be presented, which appears important in view of
extensive studies performed on these samples with a particular goal to reveal
the influence of a gradual change of the L21 ordering on their magnetic properties. Thereafter, the main results of the experimental work focused on the
determination of exchange constants and magnetic anisotropies of Co2 -based
Heusler films will be presented and discussed, which is one of the central issues
of the present thesis. Finally, the second-order (so-called quadratic) magnetooptical Kerr effect (QMOKE), appearing in the Co2 -based Heusler films will be
addressed.

60

Experimental results I - Magnetic properties of Co2 -based Heusler films

5.1

Film preparation and pre-characterization

5.1.1

Overview of investigated samples

Co2 MnSi
Thin Co2 MnSi (100) films, epitaxially grown on a MgO(100) substrate covered by a 40 nm thick Cr(100) buffer layer, were provided by the group of
Prof. Dr. Y. Ando in Sendai. Inductively coupled plasma-assisted magnetron
sputtering was employed for the deposition of the Co2 MnSi layers. The samples
were covered by a 1.3 nm Al capping layer to prevent oxidation of the films. For
our investigations, two such sample sets were fabricated. The films of the first
sample set have different thicknesses d = 20, 30, 40, 60 and 80 nm, respectively.
After deposition, all of these films were annealed at Ta = 500 C, resulting in a
predominant L21 structure, as was determined by XRD measurements. Films
of the second set are 30 nm thick. Their post-growth annealing temperatures
were varied in 25 steps between 350 and 500 C. This provided Co2 MnSi films
with a gradual variation of the of the L21 order, as will be demonstrated in
Sect. 5.1.2.
Co2 FeAl
The Co2 FeAl(100) sample investigated in this thesis was fabricated in Mainz
by Dr. M. Jourdan and collaborators. The sample consists of an 80 nm thick
Co2 FeAl layer, which was epitaxially grown on a single-crystalline MgO(001)
substrate covered with a 10 nm thick MgO buffer layer. For the deposition
of both the MgO buffer and the Co2 FeAl layer, magnetron sputtering was
employed. A post-growth anneal at 550 C provided a Co2 FeAl film with the
B2 structure, as confirmed by XRD measurements. A 3 nm thick AlOx capping
layer was deposited on top of the structure to prevent sample oxidation. The
sample exhibits a saturation magnetization of 4.66 B /f.u. measured at RT by
a superconducting quantum interference device (SQUID) magnetometer.
Co2 Cr0.6 Fe0.4 Al (CCFA)
The CCFA(100) structure under investigation is a MgO(100)/Cr(8 nm)/
CCFA(80 nm)/Al(2.5 nm) epitaxial structure provided by Dr. M. Jourdan and
collaborators. The Cr buffer layer was deposited by electron beam evaporation
onto a single-crystalline MgO substrate, while the epitaxial CCFA films were
subsequently deposited by dc magnetron sputtering [151]. XRD measurements

5.1 Film preparation and pre-characterization


revealed a B2 growth of the film. Furthermore, a saturation magnetization of
2.5 B /f.u. was determined by SQUID measurements at RT.
Co2 FeSi
Studies of the Co2 FeSi compound were carried out on two sample sets fabricated
in Mainz by Dr. G. Jakob and collaborators. The first set, which was mainly
used for the determination of exchange and anisotropy constants, consists of
MgO(100)/Cr(30 nm)/Co2 FeSi(d)/Al(4 nm) epitaxial structures with two different thicknesses d = 20 and 60 nm. For the growth of both the Cr buffer and
Co2 FeSi(100) layers pulsed laser deposition (PLD) was employed using a KrF
excimer laser ( = 248 nm) as a light source [152]. After deposition, the films
were annealed at 450 C, resulting in the L21 structure formation, which was
determined by XRD measurements.
The second sample set, which was in particular used for the investigation of
the QMOKE, consists of four Co2 FeSi(100) films with respective thicknesses of
11, 21, 42 and 98 nm. The films were deposited directly on MgO(100) substrates
by RF magnetron sputtering, and were covered by a 4 nm thick Al protective
layer. XRD measurements revealed L21 structure of the fabricated films. A
more detailed description of the sample preparation as well as of the structural
characterization can be found in Ref. [153].

5.1.2

XRD investigations of Co2 MnSi films

All XRD data that will be discussed in the following were acquired and analyzed
by S. J. Hermsdoerfer during his research stay in the group of Prof. Dr. Y. Ando
at the Tohoku University, Sendai.
The x-ray -2 scans obtained from the Co2 MnSi samples annealed at different temperatures Ta are summarized in Fig. 5.1. For all the studied films,
clear peaks corresponding to the (200) and (400) reflections of Co2 MnSi are
observed in the specular geometry indicating epitaxial (100) growth. Furthermore, pole figures of all Co2 MnSi samples show the presence of the (111) peak
that is characteristic of the L21 structure. An example of such a pole figure is
presented in Fig. 5.2(a) for the Co2 MnSi film annealed at 450 C. Figure 5.2(b)
shows the pole figure of the corresponding (220) fundamental reflections.
Before discussing further crystal structure of the fabricated Co2 MnSi films,
we would like to comment on MgO substrate reflection visible in the recorded
-2 scans. In Fig. 5.1, the MgO reflection appears as a double peak, whose
intensity is comparable with the intensities of the Co2 MnSi reflections. For this
kind of samples, however, the MgO reflection is usually observed as a single peak

61

62

Experimental results I - Magnetic properties of Co2 -based Heusler films

Figure 5.1: -2 scans of MgO/Cr(40 nm)/Co2 MnSi (30 nm)/Al(1.3 nm) films annealed at temperatures ranging from 350 C to 500 C.

with a much higher intensity compared to the Co2 MnSi reflections. The shape
of the MgO peak observed in our case is not related to the sample properties,
but is a saturation effect of the detector. If the scattered intensity is above
a certain limit, the detector is automatically blocked during the measurement
until the intensity is reduced again. This gives rise to the observed double peak
structure of the MgO reflection.
The intensities of the (111) superlattice reflections detected in the corresponding pole figures were used to estimate the L21 ordering degree in the
Co2 MnSi films annealed at different Ta . For this purpose, the long-range order

(a)

(b)

j-axis

111
202
111

100

220
100

111
220

c-axis

202

111

Figure 5.2: X-ray pole figures for (a) (111) and (b) (220) equivalent reflections of the
Co2 MnSi film annealed at Ta = 450 C.

5.1 Film preparation and pre-characterization

63

parameter SL21 was calculated using the following expression:


2
SL2
=
1

I(111)exp /I(220)exp
.
I(111)theory /I(220)theory

(5.1)

Here, Itheory denotes the peak intensities obtained from powder pattern simulations obtained using the CrystalDiffract software. Corrections for the multiplicity factor m and by the film thickness factor Gt were applied, according to
the following expression:
Itheory = Isim m1 Gt .

(5.2)

The film thickness factor was adapted from Ref. [154] and is of the form
Gt = 1 exp (2d/ sin ), where is the linear absorption coefficient of
Co2 MnSi and d is the film thickness. Isim is the calculated XRD intensity
for a powder sample.
The relationship between the long-range order parameter SL21 and Ta is
presented in Fig. 5.3. The values of SL21 , represented by full symbols, increase
from approximately 0.6 to about 0.9 with increasing Ta , except for the sample
annealed at Ta =475 C. This feature is probably related to a statistical fluctuation of the experimental parameters during the preparation procedure. For
this reason, the Co2 MnSi film with Ta = 475 C is considered as an outlier
and will not be considered further in the analysis of the experimental data. To
be sure that the increase of SL21 is due to an increasing fraction of L21 and

1.0
0.9

0.8
0.7
0.6
0.5
350

375

400

425

450

475

500

Annealing temperature (C)

Figure 5.3: Long-range order parameters SL21 (full symbols) and SB2 (open symbols) for Co2 MnSi films annealed at different temperatures after deposition. The
experimental error of S values corresponds to the size of the data points.

64

Experimental results I - Magnetic properties of Co2 -based Heusler films


not due an increase of the DO3 type of disorder (see discussion in Sect. 4.1.2),
we also estimated the degree of order in the Co sublattice. This was done by
calculating the B2 ordering parameter SB2 using the following expression:
2
SB2
=

I(200)exp /I(400)exp
.
I(200)theory /I(400)theory

(5.3)

The corresponding values are shown in Fig. 5.3 by open symbols and exhibit
a slight decrease from 0.99 to 0.93 at Ta = 400 C. For larger Ta , SB2 takes a
constant value of 0.95, which is slightly smaller than the initial value. From
these results it is concluded that, at least in the range of Ta > 400 C, the
observed increase of SL21 is indeed due to an increasing L21 fraction. The reason
for the drop of SB2 values for Ta 400 C with respect to Ta = 350 and 375 C
is not clear. It might be related to a interdiffusion with the Cr buffer layer in
this temperature range, which is a well-known problem in Heusler structures
with a Cr buffer layer [152, 155157].

5.2

Determination of exchange constants

To determine the values of the exchange constants of Heusler films, BLS spectra
were recorded at different values of the external magnetic field H and at different angles of incidence of the probing light beam corresponding to different
transferred wave vectors qk = (4/) sin . The former is necessary to confirm the magnonic origin of peaks observed in the BLS spectra, while the latter
is required for an unambiguous separation of the DE from the PSSW modes,
which is possible because of a much stronger dependence of the DE mode on
the amount of the in-plane component of the wave vector. The experimental
dependencies of the spin-wave frequencies on H and qk are subsequently compared with simulations performed using a theoretical model described in detail
in Ref. [44]. The exchange constant A, the Lande g-factor, the saturation magnetization MS and the magnetic anisotropies K are the free parameters in these
simulations. In Sect. 2.3.2 it has been shown that for the range of qk used in our
investigations the frequency of the DE mode is mainly determined by the values of MS and g, whereas the frequencies of the PSSW modes are particularly
determined by the A/MS ratio and g. Furthermore, the Lande g-factor can be
determined independently from the slope of the BLS frequency dependence on
the applied magnetic field, df /dH. Therefore, the parameters MS , A/MS and
g are found rather independently in the fitting procedure.

5.2 Determination of exchange constants

5.2.1

65

Co2 MnSi

Measurements on samples with different thicknesses

25 H=2.0 kOe
d=60nm

80
H=1.5 kOe

d=40nm

-40

-20

20

40

BLS frequency (GHz)

PSSW1

DE

DE

PSSW1

-60

60

DE

100 j=55
20
15

d=20nm

(c)
j=65

PSSW2

120

10
50

140

PSSW2
DE
PSSW1

d=40nm

j=45
PSSW2

(b)

40

5
0

H=1.0 kOe

-60

-40

-20

20

40

BLS frequency (GHz)

60

j=45

60

PSSW1

PSSW1

PSSW2

PSSW1
DE

30

DE

DE
PSSW2
DE
DE

PSSW1

PSSW2

100

d=80nm

PSSW2

BLS intensity (a.u.)

150

PSSW2

PSSW1

200

DE

PSSW1

(a)

PSSW1

Figure 5.4(a) shows BLS spectra for 20, 40, 60 and 80 nm thick Co2 MnSi films,
which were recorded at an external magnetic field H = 1.5 kOe applied along
the the [110] direction of Co2 MnSi and a transferred wave vector qk = 1.67
105 cm1 . As expected from the dispersion relations given in Sect. 2.3.2
(Eqs. (2.21) and (2.22)), the frequency of the DE mode in Fig. 5.4(a) increases
with increasing film thickness while the frequencies of the PSSW modes shift
to lower values.
The magnonic origin of the peaks present in Fig. 5.4(a) was confirmed by Hdependent measurements, the results of which are demonstrated in Fig. 5.4(b)
for the 80 nm thick Co2 MnSi film. In both the Stokes and anti-Stokes part of the
spectrum, the peak positions move towards higher frequencies with increasing
magnetic field, revealing their magnonic origin.
For a correct assignment of the observed peaks to different spin wave modes,
BLS spectra were recorded at different values of , i.e. different qk , and are
shown in Fig. 5.4(c) for the 80 nm thick Co2 MnSi film. While the spectral
positions of the PSSW modes are not significantly affected by the variation of
(and thus qk ), the frequency of the DE mode exhibits a clear shift to larger
frequencies with increasing . Therefore, the peak originating from the DE

j=30

d=80nm

20
0

j=20
-40 -30 -20 -10 10 20

30

40

BLS frequency (GHz)

Figure 5.4: BLS spectra of (a) Co2 MnSi films with different thicknesses d acquired
at an applied external field H = 1.5 kOe and a transferred wave vector qk = 1.67
105 cm1 , (b) 40 nm thick Co2 MnSi film measured at different values of H at qk =
1.67 105 cm1 and (c) 80 nm thick Co2 MnSi film recorded at H = 1.5 kOe and
different angle of incidence . Solid lines are guide to the eye, showing expected
peak positions as follows from the model.

66

Experimental results I - Magnetic properties of Co2 -based Heusler films


mode excitation is easily identified.
Figure 5.5 shows the results of numerical simulations (solid lines) along
with the experimentally determined BLS frequencies (symbols). An equally
good agreement between the simulations and the experimental data is achieved
for two different values of A, namely A = 2.35 0.1 erg cm1 (Fig. 5.5(a)(c)) and A = 0.6 0.1 erg cm1 (Fig. 5.5(d)-(f)). For both A values, the
saturation magnetization and the Lande g-factor are found to be MS = 970 emu
cm3 (4.72 B /f.u.) and g = 2.05, respectively. These MS and g values are in
agreement with those reported by other groups [158160]. In the first case,
where A = 2.35 0.1 erg cm1 (the corresponding exchange stiffness is D =
575 20 meV
A2 ), the calculations describe all observed PSSW modes. When
A = 0.6 0.1 erg cm1 (D = 575 20 meV
A2 ), only even PSSW modes seem
to be observed in the experiment.
Due to the ambiguity of the PSSW mode assignment, the correct value of
60

60

BLS frequency (GHz)

(a)
50

50

40

40

30

30

BLS frequency (GHz)

(d)

Stokes
anti-Stokes

50

(e)

40

30
20 j =45

20

40

d (nm)

80

10

14

PSSW2

18

DE
PSSW1

16

1000

1500

H (Oe)

PSSW1

2000

Stokes
anti-Stokes

(c)

d=80 nm, H=1500 Oe


26 A=0.6 merg/cm
24

Stokes
anti-Stokes

60

18

d=40 nm
j=45, q||=1.67A105 cm-1
A=0.6 merg/cm
PSSW3

20

H=1500 Oe
A=0.60 merg/cm

20

16

PSSW4

30

DE

22

12
28

50

40

10

d=40 nm
5
j=45, q||=1.67A10 cm-1
A=2.35 merg/cm
PSSW1

Stokes
anti-Stokes

10
60

d=80 nm, H=1500 Oe PSSW2


26 A=2.35 merg/cm
24

20
H=1500 Oe
A=2.35merg/cm

10
60

PSSW2

DE

j=45

20

28

(b)

Stokes
anti-Stokes

PSSW4

DE

22

PSSW3
20

14

PSSW2
PSSW1

Stokes
anti-Stokes

(f)

12
0.0

0.5

1.0

1.5
5

2.0

-1

q|| (H10 cm )

Figure 5.5: Comparison of calculated (solid lines) and experimental (symbols) spinwave frequencies determined from BLS spectra which are shown in Fig. 5.4. Calculations were performed using the exchange constant A = 2.35 0.10 erg cm1 [(a)-(c)]
and A = 0.6 0.10 erg cm1 [(d)-(f)]. The magnetic anisotropies were neglected
and remaining parameters had the values MS = 970 emu cm3 and g = 2.05.

5.2 Determination of exchange constants

67

the exchange constant cannot be obtained from the fit alone. Therefore, we
have performed analytical calculations of BLS intensities and examined the
conditions under which the BLS intensity becomes zero for odd PSSW modes
only. For this purpose, an analytical approach presented in Appendix A was
used, which combines the BLS intensity proposed by Buchmeier et al. [161,162]
and the depth selectivity of the MOKE by Hamrle et al. [163], and is based on
summation of the Kerr amplitudes originating from different depths of a film,
taking into account their different phases. This analytical approach predicts
the BLS intensity of all odd modes to be zero when conditions (A.8) and (A.9)
are fulfilled, i.e.
4 Re(Nz )d/ = odd integer and 4 Im (Nz )d/ = 0 .
However, using the complex refractive index N = 1.1 + 1.1i of Co2 MnSi for the
employed wavelength of =532 nm [164], we obtain that for an incidence angle
= 45 and the film thickness d = 20 nm
4 Re(Nz )d/ = 0.15 and 4 Im (Nz )d/ = 0.18 ,
respectively. As such, the conditions for zero BLS intensity are not fulfilled
in the BLS experiments carried out for Co2 MnSi films. Therefore all PSSW
modes in the investigated Co2 MnSi films should provide a non-zero signal in
the BLS spectra. Hence the exchange constant of Co2 MnSi is determined to be
A = 2.35 0.10 erg cm1 , which was determined from the calculations shown
in Fig. 5.5(a)-(c).
Measurement on samples with different annealing temperatures
Figure 5.6(a) shows selected BLS spectra of Co2 MnSi films annealed at different temperatures, which were recorded at an external magnetic field H=2.0 kOe
applied along the [110] direction of Co2 MnSi and at a transferred wave vector
of qk = 1.35 105 cm1 . In both the Stokes and anti-Stokes parts of the spectra two peaks originating from the magnonic scattering processes are visible.
According to simulations presented in Fig. 5.5(a), the peak at around 20 GHz
can be attributed to the DE mode, while the peak appearing at approximately
32 GHz results from the excitation of the perpendicular standing spin wave. The
characteristic shape of the BLS spectrum remains unchanged for all the investigated samples. A closer inspection of the spectral peak positions, however,
reveals that the frequency of the DE mode decreases by about 1 GHz when annealing temperature is increased from 350 to 500 C, while the frequency of the
PSSW mode increases by about 1.5 GHz (Fig. 5.6(b)) over the same annealing
temperature range. These changes in the BLS frequencies indicate a decrease

68

Experimental results I - Magnetic properties of Co2 -based Heusler films


40

700

BLS intensity (a.u.)

600
500
400

35

PSSW1

DE

300 PSSW1

(b)

350
400
450
500

H=2.0 kOe
j=35

Frequency (GHz)

(a)

DE

200

30
25
20

DE

15

100
0
-40

PSSW1

-30

-20

20

30

40

Frequency (GHz)

10
325 350 375 400 425 450 475 500 525

Annealing temperature (C)

Figure 5.6: (a) BLS spectra of Co2 MnSi films annealed at different temperatures.
Measurements were performed at an external field H = 2.0 kOe applied along the
[110] direction of Co2 MnSi and a transferred wave vector qk = 1.35 105 cm1 . (b)
Dependence of spin-wave frequencies on the annealing temperature of the Co2 MnSi
films.

of the saturation magnetization and an increase of the exchange constant in


the series of investigated Co2 MnSi films. Simulations confirm this behavior
(Fig. 5.7), showing that MS decreases by about 6 % (Fig. 5.7(a)) whereas A
increases by about 9 % (Fig. 5.7(b)) as the annealing temperature is increased
from 350 to 500 C. Accordingly, the exchange stiffness D increases with the
1050
1040

560

2.35

(a)

(b)

(c)

2.30

540

1030

1010
1000
990
980

520

A (erg/cm)

MS (emu/cm )

1020

D (meV D )

2.25
2.20
2.15
2.10

500

480

970
2.05

960
950

460
350 375 400 425 450 475 500

Annealing temperature (C)

2.00

350 375 400 425 450 475 500

350 375 400 425 450 475 500

Annealing temperature (C)

Annealing temperature (C)

Figure 5.7: (a) Saturation magnetization, (b) exchange constant and (c) exchange
stiffness as a function of the annealing temperature. The values in (a) and (b) were
obtained from simulations of BLS frequencies shown in Fig. 5.6(b), the values in (c)
were calculated using Eq.

5.2 Determination of exchange constants

69

increase of the annealing temperature (Fig. 5.7(c)). Since the increase of the
annealing temperature is related to an improvement of the L21 order in the
Co2 MnSi films (Sect. 5.1.2), this result obviously implies that D is sensitive to
the disorder in the crystal lattice.
It should be noted that MS and A values shown in Fig. 5.7 were obtained
after reprocessing of the experimental data with the corrected assumption that
all spin-wave modes are present in the BLS spectra. Therefore, they are different from those published by us in Ref. [A4], where only even modes were
initially assumed to provide a non-zero BLS intensity.

5.2.2

Co2 FeAl

The BLS spectra collected for the Co2 FeSi sample are shown in Fig. 5.8. In (a)
spectra recorded at a transferred wave vector qk = 1.67105 cm1 and different
values of the external magnetic field are displayed, while in (b) BLS spectra
are presented which were measured at H = 1.0 kOe and at different incidence
angles (i.e. different qk ) of the probing light beam. As is clearly visible
from Fig. 5.8(a), the positions of the peaks move to larger frequencies as the
magnetic field increases, confirming the magnonic origin of the observed peaks.
A much stronger dependence of the DE mode on the in-plane direction of the
wave vector evident in Fig. 5.8(b) allows for an unambiguous separation of the
DE mode from the PSSW modes. Remarkably, the PSSW2 mode is difficult
2000

2000

j=45
0.5 kOe
1.0 kOe
1.5 kOe
2.0 kOe
2.5 kOe

1500
DE

1000

PSSW2 DE
PSSW3

PSSW3
PSSW1

500

PSSW1

PSSW4

-40

-30

-20

-10

10

PSSW4

20

Frequency (GHz)

30

40

(b)
BLS intensity (a.u.)

BLS intensity (a.u.)

(a)

H=1.0 kOe
j=20
j=45
j=60

1500
DE

PSSW1

1000
PSSW2

500

PSSW4 PSSW3

PSSW2 PSSW3
DE
PSSW4

PSSW1

0
-40

-30

-20

-10

10

20

30

40

Frequency (GHz)

Figure 5.8: (a) BLS spectra of Co2 FeAl recorded at a transferred wave vector qk =
1.67 105 cm1 and different magnetic fields. (b) BLS spectra of Co2 FeAl measured
in an external field H = 1.0 kOe and at different angle of incidence . In (a) and
(b), the dashed lines are guides to the eye, visualizing the change of spectral peak
positions.

70

Experimental results I - Magnetic properties of Co2 -based Heusler films


50

BLS frequency (GHz)

45

(a)

45

-1

(b)

j=45, q||=1.67H10 cm

H=1.0 kOe

PSSW4

40

40
35

35

30

30

PSSW3

25
25

DE

20

PSSW2

20

15

Stokes
anti-Stokes
simulation

10
5
0.0

0.5

1.0

1.5

H (kOe)

2.0

2.5

PSSW1

15

3.0

10
10

20

30

40

50

60

70

80

j (deg)

Figure 5.9: Comparison of simulations (solid lines) with experimental BLS frequencies
(symbols), which were determined from the BLS spectra presented in Fig. 5.8. (a)
shows the dependence of BLS frequencies on the external magnetic field, whereas
(b) demonstrates the dependence of BLS frequencies on the angle of incidence . In
simulations following parameters were used: A = 1.55 erg cm1 , MS = 1027 emu
cm3 and g = 2.1.

to recognize in the Stokes part of the spectra (i.e. negative frequencies) shown
in Fig. 5.8(a). This is due to the much stronger intensity of the adjacent DE
mode as well as a relatively small spacing between the PSSW2 and DE modes.
In the anti-Stokes region of the spectra, the DE mode has intensities that are
comparable to or smaller than those of the PSSW2 mode. Hence, both of them
are clearly observed in this part of the spectra. The origin of the asymmetry of
the DE mode intensities in the Stokes and anti-Stokes regions will be discussed
in Sect. 5.2.5.
Figure 5.9 shows a fit of peak positions extracted from Fig. 5.8(a) and
(b) along with the experimental data. The fit was obtained using the exchange constant A = 1.55 0.05 erg cm1 , the saturation magnetization
MS = 1027 10 emu cm3 (5.17 B /f.u.) and the Lande factor g = 2.1 0.1.
The corresponding exchange stiffness is D = 370 10 meV
A2 . The value of
the cubic volume anisotropy constant K1 was assumed to be zero in this fitting
procedure. Since including K1 into the theoretical model used for the fitting
only slightly changes the frequency of the spin waves (by about 1 GHz), setting
K1 to zero does not detrimentally affect the determination of MS and A.

5.2 Determination of exchange constants

5.2.3

71

Co2 Cr0.6 Fe0.4 Al

Typical BLS spectra measured on the investigated CCFA film are presented
in Fig. 5.10(a) for several values of the external magnetic field. As is clearly
shown in this figure, the positions of the peaks in both Stokes and anti-Stokes
parts of the spectrum move to higher frequencies with increasing magnetic
field. This field dependence is evidence of the magnonic nature of the peaks.
One of the observed peaks in the BLS spectra stems from the hydbridization
of the DE and the second PSSW mode, as is revealed by simulations of the
BLS frequencies shown in Fig. 5.10(b). The hybridized mode is referred to
as PSSW2+DE mode. It exhibits a large (0.5-1 GHz) splitting between the
experimental Stokes (N) and anti-Stokes (H) frequencies, which is particularly
pronounced for values of the external field in the range 100-400 Oe. Moreover,
a careful investigation of Fig. 5.10(b) reveals that the lower frequency component of the split PSSW2+DE mode is observed only in the Stokes part of
the BLS spectrum, whereas the higher frequency component appears only in
the anti-Stokes part. Worthwhile to mention is also the observation that the
field dependence is slightly different for Stokes and anti-Stokes frequencies. An
explanation of these features of the hybridized mode will be given in Sect. 5.2.5.
From the fit presented in Fig. 5.10(b) we obtain for CCFA the exchange
24

900

BLS frequency (GHz)

H=1500 Oe

PSSW4

PSSW3

PSSW1

500

PSSW3

600

400
300

H=1000 Oe

200
100 H=400 Oe
0
-25

-20

(b)

22
PSSW2+DE

PSSW1

700
PSSW4

BLS intensity (a.u.)

800

PSSW2+DE

(a) H || [100]CCFA

W4

PSS

20
W3
PSS

18
16
14

DE
W2+

PSS

12

W1

PSS

10
8
6

Stokes
anti-Stokes

-15

-10

-5

10

Frequency (GHz)

15

20

25

2
0

500

1000

1500

H (Oe)

Figure 5.10: (a) Selected BLS spectra measured on the CCFA film at a transferred wave vector qk = 1.67 105 cm1 and in external magnetic fields H =
400, 1000, 1500 Oe. The dashed lines are guides to the eye. (b) Comparison of calculated (solid lines) and experimental (symbols) BLS frequencies. Simulations were performed using A = 0.48 erg cm1 , MS = 520 emu cm3 , g = 1.9 and K1 = 20 kerg
cm3 .

72

Experimental results I - Magnetic properties of Co2 -based Heusler films


constant A = 0.48 0.04 erg cm1 , the saturation magnetization MS =
520 20 emu cm3 and the Lande factor g = 1.9 0.1. In the simulations the
cubic volume anisotropy constant K1 = 20 kerg cm3 was included, whose determination is described in Sect. 5.3. The exchange stiffness of CCFA is found
to be D = 203 16 meV
A2 .

5.2.4

Co2 FeSi

Figure 5.11 shows BLS spectra collected for the Co2 FeSi samples fabricated
by PLD. In Fig. 5.11(a), film thickness dependent spectra are shown, which
were measured at an external field H = 2.0 kOe and a transferred wave vector
qk = 1.67 105 cm1 . In Fig. 5.11(b) and (c), H- and -dependent measurements performed for the 60 nm thick Co2 FeSi film are presented. In the 20 nm
thick Co2 FeSi sample, only a single spin wave mode is excited with the frequency of 20 GHz, while in the 60 nm thick Co2 FeSi film an additional mode
appears around 40 GHz. A peak at about 30 GHz, visible in Fig. 5.11(a) for the
60 nm thick sample, turns out to be an artifact peak since it moves to lower
frequencies with the increasing magnetic field (Fig. 5.11(b)). The other two
peaks are confirmed to be of magnonic origin. The assignment of the peaks
to the spin wave modes indicated in Fig. 5.11 is based on the numerical fit
shown in Fig. 5.12. Similarly to the results presented for the other Heusler
compounds, an asymmetry between the Stokes and anti-Stokes frequencies was
6

H=2.0 kOe
j=45

DE

PSSW2

3
PSSW2

d=60 nm

-40

-30

-20

20

2
1

DE

d=20 nm

4
DE

d=60 nm
H=2.0 kOe

(c)
10
8

PSSW1

12

PSSW1

6
4

d=60 nm
j=45

(b)

DE

PSSW2

BLS frequency (a.u.)

(a)

PSSW2

H=2.4 kOe

H=2.0 kOe

j=45

H=1.5 kOe

j=15

H=1.0 kOe

30

Frequency (GHz)

40

-40

-30

-20

20

30

Frequency (GHz)

40

-40

-30

-20

20

30

40

Frequency (GHz)

Figure 5.11: BLS spectra of (a) Co2 FeSi films with thicknesses d = 20 and 60 nm
acquired at an applied external field H = 2.0 kOe and a transferred wave vector
qk = 1.67 105 cm1 (i.e. = 45 ), (b) 60 nm thick Co2 FeSi film measured at
different values of H at qk = 1.67 105 cm1 and (c) 60 nm thick Co2 FeSi film
recorded at H = 2.0 kOe and different angle of incidence . Solid lines are guide to
the eye.

5.2 Determination of exchange constants

73

50

BLS frequency (GHz)

3
SW
PS

W2
PSS

1
PSSW

(a)
45

(b)

(c)

d=60 nm
j=45

d=60 nm
H=2.0 kOe
PSSW2

40

W
PSS

35
30

DE
25
PSSW1

DE
DE

20
Stokes
anti-Stokes
simulation

15
10

W1

H=2.0 kOe
j=45

20 30 40 50 60 70 80 90 100 0.0

d (nm)

PSS

0.5

Stokes
anti-Stokes
simulation

Stokes
anti-Stokes
simulation

1.0

1.5

H (kOe)

2.0

2.5

3.0 10

20

30

40

50

60

70

80

j (deg)

Figure 5.12: Fit to the experimental BLS frequencies which were determined from
spectra shown in Fig. 5.11 as function of (a) films with thicknesses d, (b) external
magnetic field H and (c) angle of incidence of the probing light beam . The fit
yields the exchange constant A = 3.15 erg cm1 , the saturation magnetization MS =
1019 emu cm3 and the Lande factor g = 2.0. Magnetic anisotropies are neglected.

also observed in case of Co2 FeSi. The DE mode is present only in the Stokes
part of the spectrum.
The calculated BLS frequencies shown in Fig. 5.12, which provide the best
agreement with the experimental data, were obtained for the exchange constant
A = 3.15 0.05 erg cm1 , the saturation magnetization MS = 1019 10 emu
cm3 (5.06 B /f.u.) and the Lande factor g = 2.0. Magnetic anisotropies were
neglected in the calculations since they were found to be below the sensitivity
of the BLS technique (see Sect. 5.3.4). The exchange stiffness of Co2 FeSi is
found to be D = 715 20 meV
A2 .

5.2.5

Discussion

We first compare the exchange constants obtained for the investigated Heusler
compounds, and discuss observed trends. Thereafter, the origin of asymmetries
between the Stokes and anti-Stokes frequencies will be addressed.
Exchange constants
The results of our experimental work focused on the determination of exchange
constants of various Co2 -based Heusler compounds are summarized in Table 5.1

Experimental results I - Magnetic properties of Co2 -based Heusler films


A (erg cm1 ) MS (emu cm3 ) D (meV
A2 )

Sample

Co2 MnSi (L21 )

2.35(0.1)

970(8)

575(20)

2.05

Co2 FeSi (L21 )

3.15(0.05)

1019(10)

715(20)

2.0

Co2 FeAl (B2)

1.55(0.05)

1027(10)

370(10)

2.1

CCFA (B2)

0.48(0.04)

520(20)

203(16)

1.9

Table 5.1: Exchange constant A, saturation magnetization MS , exchange stiffness D


and Lande g-factor obtained by the simulation of experimental BLS frequencies for
different Heusler films studied in this thesis.

along with other parameters obtained from the numerical fit of the BLS frequencies. The values of the exchange stiffness D are plotted in Fig. 5.13 versus
the number of valence electrons per atom NV , which is a common way to systematize the magnetic properties of Heusler compounds (Sect. 3.3.1 and 3.3.4).
In this figure, the D values determined within the framework of the present the-

800
Co2FeSi (L21)

700
Co2MnSi (L21)

600
2

D (meV D )

74

500
Co2MnGe

400
300

Co2FeAl (B2)
CCFA (B2)

200
Co2MnAl (B2)

present work
others

100
0
6.9

7.0

7.1

7.2

7.3

7.4

7.5

7.6

NV
Figure 5.13: Exchange stiffness D determined in the present thesis (full symbols) and
by other authors (open symbols) as a function of the number of valence electrons per
atom NV . The dashed lines are guides to the eye connecting L21 and B2 ordered
compounds, respectively.

5.2 Determination of exchange constants

75

sis are represented by full symbols. For completeness, Fig. 5.13 also contains D
values of Co2 MnAl and Co2 MnGe taken from third-party publications [165,166].
These values are represented by open symbols. As is suggested by dashed lines,
the experimental points can be subdivided into two branches specified by the
crystallographic ordering of the investigated films, i.e. either B2 or L21 . In
both branches, a linear increase of D is observed with increasing NV . This
behavior is similar to the Slater-Pauling rule found for the Heusler compounds
and the dependence of their Curie temperatures on NV . In the following, these
two features will be discussed in more detail.
A subdivision of the experimental D values according to the crystal structure is justified by the results of our investigations performed on the series of
Co2 MnSi films annealed at different temperatures. As was shown in Fig. 5.7(c),
the exchange stiffness D increases with increasing annealing temperature, i.e.
with increasing L21 order. This dependence of D on the crystallographic ordering is more strikingly illustrated in Fig. 5.14, where D is plotted as a function of the ordering parameter SL21 and exhibits a linear dependence. Extrapolation to SL21 = 0 , which corresponds to a B2 structure, provides a
value of D = 324 meV
A2 . This value apparently falls onto the B2 branch
shown in Fig. 5.13 as it agrees quite well with the D value of the B2-ordered
Co2 FeAl (D = 370 meV
A2 ).
On the other hand, the segregation of the experimental points into two

550

exp. data
linear fit

D (meV D )

500

450

400

350
0.0

L21

B2
0.2

0.4

0.6

0.8

L21 ordering parameter, S L2

1.0

Figure 5.14: Dependence of the exchange stiffness D on the L21 ordering parameter
for Co2 MnSi.

Experimental results I - Magnetic properties of Co2 -based Heusler films


800
present work
others

Co2FeSi

700
600
2

D (meV D )

76

500

Co2MnSi
Co2FeAl

400

Co2MnGe

300
CCFA

200
Co2MnAl

100

5.64

5.66

5.68

5.70

5.72

5.74

5.76

a (D )
Figure 5.15: Exchange stiffness D as a function of the (bulk) lattice constant a
(Table 3.1). The dashed line is a guide to the eye.

branches might be related to the species of the non-magnetic element. While the
upper branch in Fig. 5.13 contains those compounds where the non-magnetic
element is Si, the compounds of the lower branch have Al (or Ge) as the nonmagnetic constituent. The higher D values of the Si containing compounds
can be understood considering the atomic radii of the non-magnetic elements.
A smaller atomic radius of Si (110 pm) compared to that of Al (125 pm) [167]
leads to a contraction of the unit cell resulting in an enhancement of the exchange interaction (Sect. 3.3.3). The dependence of the exchange stiffness on
the (bulk) lattice constant a is illustrated in Fig. 5.15, and demonstrates that
larger exchange stiffness values are indeed observed for smaller lattice constants.
The linear dependence between the exchange stiffness D and the number
of valence electrons NV observed in both branches of Fig. 5.13 is largely a
result of the increasing NV . The increase of NV in Co2 YZ Heusler compounds
adds electrons to the electronic structure, which are primarily found in the t2g
states of the transition metal Y [82]. This increases the electron density that
participates in exchange, thus resulting in a larger exchange integral J and
related exchange stiffness D. Additionally, a decreasing size of the unit cell
observed with increasing NV (Fig. 5.15) might contribute as well to the linear
dependencies shown in Fig. 5.13. It should, however, be noted that although the
dependence of D on NV appears to be linear for a given non-magnetic element
Z, further Co2 -based Heusler compounds should be investigated to confirm the

5.2 Determination of exchange constants


linear trend and to clarify the exact functional form of this dependence.
The largest exchange stiffness D is found for the Co2 FeSi Heusler compound, with a value of 71520 meV
A2 . This value is 2 2.5 times larger than
the exchange stiffnesses reported for the bcc-Fe (270 meV
A2 ) [44] and bcc-Co
(430 meV
A2 ) [168]. In view of results reported for binary Fe1x Cox alloys [169],
larger exchange stiffnesses in compounds compared to the D values of constituent 3d metals are not further surprising. In the series of Fe1x Cox alloys
studied in Ref. [169], D values exceeding those of Fe and Co were found for several compositions, with a record value of 80050 meV
A2 for Fe53 Co47. However,
the large D value obtained for Co2 FeSi in the present work, which is nearly as
large as the maximum D value found for the Fe1x Cox series, is rather unexpected when taking into account that one quarter of the constituent atoms is
a non-magnetic element, i.e. Si.
Asymmetry of Stokes and anti-Stokes frequencies
To explain the asymmetries that we observe between the Stokes and anti-Stokes
parts of the BLS spectra, we calculated the depth profiles of the dynamic magnetization, using the same model as for the simulations of BLS frequencies [44].
The results for CCFA are shown in Fig. 5.16, where depth profiles of the dynamic magnetization are presented for d = 60 and 80 nm, along with the calculated dependence of the spin-wave frequencies on the CCFA thickness. This
figure shows that in the vicinity of d = 80 nm, which is the thickness of the
investigated CCFA film, a crossing between the DE and PSSW2 modes occurs,
resulting in a hybridization of these modes. Due to the hybridization, a gap of
0.3 GHz is created, which corresponds to the splitting between the Stokes and
anti-Stokes frequencies observed in the experiment.
The hybridization is also visible in the calculated depth profiles of the dynamic magnetization in Fig. 5.16. The presented depth profiles were calculated
for the anti-Stokes modes. The left part of each profile image shows the trajectories of the magnetization vector along the film depth (the trajectory at a given
depth is an ellipse). In the corresponding right part of each image the profiles
of the in-plane and out-of-plane components of the dynamic magnetization are
shown, denoted by solid and dashed lines, respectively. The amplitude of the
in-plane dynamic magnetization is roughly twice as large as the amplitude in
the out-of-plane direction, which is due to the shape anisotropy of the CCFA
film. As is shown for d=60 nm, the DE mode has a large dynamic magnetization near one interface which decays to the second interface. The n-th order
mode of the PSSW has n nodes (i.e. there are n points in the depth of the
CCFA film where the dynamic magnetization is zero) and symmetrical with
respect to both interfaces. For the 80 nm thick film, the profiles of the two hy-

77

Experimental results I - Magnetic properties of Co2 -based Heusler films

W4

5
SW
PS

S
PS

3
SW

W2

20

PS

25

PSS

1
PSSW

d=60nm,DE

d=80nm,PSSW4

30

d=60nm,PSSW2

BLS frequency (GHz)

15

d=80nm,PSSW3

10

DE
Stokes
anti-Stokes
calculations

H=400 Oe

0
0

20

40

60

80

100

120

140

CCFA thickness (nm)


d=60nm,PSSW1 d=80nm,PSSW1 d=80nm,PSSW2/DE d=80nm,DE/PSSW2

ws=w0-wm
kStokes

incoming
& scattered
light
s

emitted
magnon

78

absorbed
magnon

ws=w0-wm
kantiStokes

M,H

Figure 5.16: Calculated spin-wave frequencies as a function of the thickness of the


CCFA film for H = 2.0 kOe, Hk[100]CCFA and a transferred wave vector qk = 1.67
105 cm1 . Symbols represent experimental data corresponding to Fig. 5.10, while
solid lines are simulated frequencies. The parameters used in the simulations are
the same as in Fig. 5.10. The graph is surrounded by calculated depth profiles of
spin-wave modes for d = 60 and 80 nm. The left part of each profile image shows the
trajectory of magnetization when looking onto the magnetization vector. The right
part shows the corresponding profile of the amplitude of the dynamic magnetization
in the out-of-plane (dashed line) and in-plane (solid line) directions, respectively. The
sketch in the bottom-right corner shows the used geometry.

bridized modes (denoted as DE/PSSW2 and PSSW2/DE) have characteristic


features of both the DE and PSSW2 modes. A comparison of those two modes
shows that when the maximum dynamic amplitude for the DE/PSSW2 mode
is at the upper interface, the PSSW/DE has a larger amplitude at the opposite
interface. The picture is reversed for the Stokes modes, i.e. modes that were
bound to the upper interface are now bound to the bottom interface and vice
versa.
In the BLS experiments performed for the 80 nm CCFA film, we are only
sensitive to the dynamic magnetization near the upper interface, since the ty-

5.2 Determination of exchange constants

79

50

d=80 nm, DE

BLS frequency (GHz)

5
W

W4

S
PS

S
PS

3
SW
PS

35

W2

40

PSS

1
PSSW

45

30
d=80 nm, PSSW2

25
20

DE

15
10
0

20

Stokes
anti-Stokes
calculations
40

60

H=1000 Oe
5
-1
q||=1.67A10 cm
80

100

120

140

Co2FeAl thickness (nm)

Figure 5.17: Calculated dependence of spin-wave frequencies on the thickness of the


Co2 FeAl film for H = 1.0 kOe and a transferred wave vector qk = 1.67 105 cm1 .
Symbols represent experimental spin-wave frequencies determined from the BLS spectra shown in Fig. 5.8, solid lines represent simulated frequencies. The parameters used
in the simulations are the same as in Fig. 5.9. On the right part of the figure, depth
profiles of dynamic magnetization are shown for the DE and PSSW2 modes in the
80 nm thick Co2 FeAl film.

pical probing depth for the used laser wavelength is about 20-30 nm [162, 163].
Moreover, the sensitivity to the out-of-plane magnetization component is about
twice as large compared with the in-plane component in the given geometry
( = 45 ). Combining these two points with the discussed mode profiles shows
that the DE/PSSW2 (PSSW2/DE) mode provides a larger signal in the antiStokes (Stokes) part of the BLS spectra. This explains why different spin-wave
modes are observed in the Stokes and anti-Stokes parts of the BLS spectra of
the CCFA film.
The same reasoning applies also to the explanation of asymmetries between
the Stokes and anti-Stokes frequencies observed for Co2 FeAl and Co2 FeSi films.
As was indicated by the calculated dependence of the spin-wave frequencies on
the film thickness in Fig. 5.17, the DE mode of the investigated Co2 FeAl film is
hybridized with the PSSW2 mode, as is explained above for the CCFA sample.
However, while the DE and PSSW2 mode are indistinguishable in CCFA due
to the small difference in frequency between these two modes, the frequency
difference between these same two modes is larger in Co2 FeAl. Therefore, both
modes are distinguishable in the BLS spectra. The fact that in the Stokes
part of the spectra only the DE mode is observed can again be explained by

80

Experimental results I - Magnetic properties of Co2 -based Heusler films


the depth profiles of the dynamic magnetization, which are shown on the right
side of Fig. 5.17. The illustrated profiles were calculated for the Stokes modes,
and demonstrate a much larger out-of-plane magnetization component close to
the upper interface for the DE mode, while this component is nearly vanishing
for the PSSW2 mode. This is the reason why the DE mode provides a much
larger signal in the Stokes region. For the anti-Stokes modes, the calculated
depth profiles are simply reversed as already mentioned in the discussion of
BLS frequencies in CCFA. Since in this case the out-of-plane component of the
dynamic magnetization is slightly larger for the PSSW2 modes, the PSSW2
mode is observed with a larger intensity than the DE mode on the anti-Stokes
sides. A similar picture results for the Co2 FeSi films, which we however, do not
discuss in detail here.

5.3

Magnetic anisotropies

Magnetic anisotropies inherent to the investigated Heusler films were studied by


means of both MOKE magnetometry and BLS spectroscopy. Using the MOKE
setup described in Sect. 4.2.1, we measured magnetization reversal curves at
different in-plane angles between a particular crystallographic direction of the
films and the plane of light incidence. In all of the measurements presented in
the following, the sample orientation was varied in 1 steps in the range between 0 and 360 . Acquired loops were used to determine the coercive field HC
as a function of , which contains the information about the anisotropy fields
(Sect. 2.1.3). This measurement procedure thus provides qualitative information about the magnetic anisotropies of the studied films. For a quantitative
determination of magnetic anisotropies in the form of anisotropy constants,
spin-wave frequencies were studied as a function of the in-plane sample orientation . Experimentally obtained dependencies of the spin-wave frequencies
on were fitted by the model described in Ref. [44] yielding the values of
anisotropy constants.

5.3.1

Co2 MnSi

Figure 5.18 shows an example of hysteresis curves measured using MOKE magnetometry for the Co2 MnSi film annealed at 425 C. The displayed loops were
acquired at = 0, 42 and 45 , where is the angle between the in-plane [110]
direction of the Co2 MnSi film and the plane of incidence. Depending on the
sample orientation, the hysteresis loops reveal a strongly asymmetric shape,
which results from the quadratic MOKE (QMOKE) contribution (Sect. 2.2.1).

5.3 Magnetic anisotropies

81

10

10

(c)

0
-5
-10
-15

-200 -100

100

5
exp. MOKE

0
-5
LMOKE

-10
QMOKE

-15
-20
-25

200

MOKE signal (arb. units)

-20

10

(b)

MOKE signal (arb. units)

MOKE signal (arb. units)

(a)

-200 -100

100

5
0
-5
-10
-15
-20
-25

200

-200 -100

H (Oe)

H (Oe)

100

200

H (Oe)

Figure 5.18: Hysteresis loops measured by MOKE magnetometry at sample orientations of (a) 0, (b) 42 and (c) 45 for the Co2 MnSi(100) film annealed at 425 C
(full line). The symmetrization and antisymmetrization of these loops provide the
LMOKE (dashed line) and QMOKE (dotted line) contributions.

It should be noted that a considerable QMOKE is observed for all of the investigated Co2 MnSi films. This second-order effect will be discussed in more
detail in Sect. 5.5.
Using a symmetrization and antisymmetrization procedure described in
Sect. 4.2.1 and Refs. [A2, 122], the recorded Kerr rotation loops were separated into the LMOKE and QMOKE contributions. Both these contributions
for the loops shown in Fig. 5.18 are demonstrated in the figure as well. The
LMOKE component was used for the determination of the coercive field HC .
The coercive field determined from the LMOKE loops is presented in
Fig. 5.19 as a function of the in-plane sample orientation for Co2 MnSi(100)
films annealed at 350, 425 and 500 C. The polar plots reveal that a four-fold
in-plane magnetic anisotropy is present in the investigated Co2 MnSi samples.
90

90
120

16

60

120

20

12

16

30

150

30

150

8
0

180

12

a
4
180

4
8

330

210

12
16

[110]CMS

30

12
8

60

12

150

HC (Oe)

HC (Oe)

HC (Oe)

180

120

16

8
4

[100]CMS

90

60

(a)

240

300
270

16

210

20

(b)

330

330

210

12
240

300
270

16

(c)

240

300
270

Figure 5.19: Dependence of the coercive field HC on the in-plane sample orientation
for Co2 MnSi(100) films annealed at (a) 350, (b) 425 and (c) 500 C.

82

Experimental results I - Magnetic properties of Co2 -based Heusler films


This four-fold symmetry reflects the cubic symmetry of the crystal structure.
From the rectangular shape of the corresponding LMOKE loops, which is typical for an easy axis magnetization reversal, the h110i axes of Co2 MnSi are
identified as magnetically easy axes. The h110i directions are magnetically hard
axis directions.
All of the polar plots presented in Fig. 5.19 exhibit sharp peaks of HC along
the hard in-plane directions. As will be demonstrated in the following sections,
this feature does not only appear in Co2 MnSi thin films but also in the films of
other Heusler compounds. Its origin is discussed in Sect. 5.4.
A closer look at the polar plots in Fig. 5.19 demonstrates that the variation
of HC with is most pronounced for the film annealed at 350 C, for which
a different HC value is registered for every in the range between 0 and 45 .
For the film annealed at 425 C, the variation of HC with is less pronounced
since HC values are constant in the range of 15 around the h110i directions.
Finally, the variation of HC with becomes even smaller in the film annealed at
500 C, where HC remains constant also in the range of 15 around the h100i
directions. This indicates a reduction of the magneto-crystalline anisotropy
with increasing annealing temperature.
The cubic anisotropy constants were determined for films annealed at 350,
375, 400, 450 and 500 C using BLS spectroscopy. For this purpose, BLS spectra
were recorded for various in-plane sample orientation at a transferred wave
vector of qk = 1.67 105 cm1 and an external magnetic field of H = 300 Oe,
which was large enough to saturate the sample as determined from the MOKE
measurements presented above. The angle between H and the [110] easy
axis direction of the Co2 MnSi films was varied from 0 to 180 in 15 steps
by rotating the sample. Due to the four-fold symmetry of the anisotropy (see
Fig. 5.19), the variation of in the whole range from 0 to 360 was not required.
Figure 5.20(a) shows the frequency of the DE mode as a function of the
in-plane orientation for the Co2 MnSi film annealed at 375 C. The sample
exhibits a clear four-fold magnetic anisotropy which is in agreement with the
results obtained from the MOKE investigations presented above. Maxima of
the surface spin wave frequencies at = n 90 , with n being an integer, show
that the h110i directions are easy axis directions [170]. Furthermore, from
the minima of the DE frequencies at (n + 1/2) 90 we deduce that the h100i
directions are hard axis directions [170].
From the fit of experimental data by the model described in Ref. [44], we
determined the cubic volume anisotropy constant K1 . The values of K1 for
samples annealed at different temperatures are shown in Fig. 5.20(b). The
maximal value of K1 = 90 10 kerg/cm3 was found at 375 C. The increase
of the annealing temperature leads to a drop of the volume anisotropy constant
by a factor of 10, in agreement with the results of MOKE investigations which

5.3 Magnetic anisotropies

83

[010]

15.6

15.2

14.8

14.4

(a)
0

100

experimental data
numerical fit

-K1 (k e rg /c m )

[110]

[100]

Frequency (GHz)

16.0

80
60
40
20
0

30

60

90

120

150

180

Sample orientation (deg)

(b)
350

375

400

425

450

475

500

Annealing temperature (C)

Figure 5.20: (a) Frequency of the DE mode as a function of the angle between the
external magnetic field H = 300 Oe and the [110] easy axis direction for the Co2 MnSi
(100) film annealed at 375 C. (b) Cubic volume anisotropy constant K1 for the
Co2 MnSi(100) films annealed at different temperatures.

showed a decreasing anisotropy as well.

5.3.2

Co2 Cr0.6 Fe0.4 Al

Typical Kerr rotation loops of the investigated CCFA sample are presented in
Fig. 5.21(a) for different in-plane sample orientations . Here, denotes the
angle between the [100] CCFA axis and the plane of light incidence. In contrast to the previously described MOKE investigation of the Co2 MnSi samples,
the loops recorded from the CCFA film show no evidence of the second-order
contributions.
The dependence of the coercivity values determined from the measured
MOKE loops on is shown in Fig. 5.21(b). As can be seen in the figure,
the CCFA film has a four-fold in-plane magnetic anisotropy with coercivity
maxima in h110i directions and minima along the h100i crystal axes. In combination with the shape of the loops at = 0 and 45 , which is rectangular
in the latter case, this suggests that the h110i directions are easy axes, while
h100i are hard axes.
The results of MOKE investigations are confirmed by BLS measurements,
results of which are presented in Fig. 5.22. The corresponding BLS spectra were
recorded in an external magnetic field H = 180 Oe and at a transferred wave
vector qk = 1.67 105 cm1 (i.e. = 45 ). The in-plane sample orientation
was varied between 0 and 45 . As is visible in Fig. 5.22, the spin-wave
frequencies observed for H applied in [100] CCFA direction ( = 0 ) are about

Experimental results I - Magnetic properties of Co2 -based Heusler films


[010] CCFA

20

90

a=0
a=3
a=45

10

20

-10

50

30

150

0 180

-20
-50

[100] CCFA

10
20

-100

60

10

30

(a)

120

30

HC (Oe)

MOKE (mdeg)

330

210

240

300
270

100

(b)

H (Oe)

Figure 5.21: (a) MOKE loops measured for the CCFA(100) film for different sample
orientations . (b) Polar plot of the coercivity HC .

1 GHz smaller with respect to those with H being along the [110] direction.
This indicates that the h100i directions are magnetically hard axes, while h110i
are easy axes [170].
The dependence of the spin-wave frequencies on was fitted by the model
described in Ref. [44]. In Fig. 5.22, the fit is represented by solid lines. From
this fit the cubic volume anisotropy constant is found to be K1 = 20
10 kerg/cm3 . All the remaining fit parameters are identical with those of
Fig. 5.10 in Sect. 5.2.3, which are the exchange constant A = 0.48 erg cm1 ,
24

Stokes
anti-Stokes

22
20

BLS frequency (GHz)

84

PSSW4

18
16

H=180 Oe
PSSW3

14
12
PSSW2+DE

10
8

PSSW1

6
4

[100]

[110]

2
0 5 10 15 20 25 30 35 40 45

a (deg)

Figure 5.22: Dependence of the BLS frequencies on the in-plane sample orientation
, which is the angle between the [100] CCFA axis and the in-plane applied magnetic
field H=180 Oe. Triangles-up (down) refer to Stokes (anti-Stokes) frequencies of the
BLS spectra. Solid lines are the calculated spin-wave frequencies.

5.3 Magnetic anisotropies

85

the saturation magnetization MS = 520 emu cm3 and the Lande factor g = 1.9.
The relatively small K1 value is in agreement with the characteristic dependence
of HC on (Fig. 5.21(b)), i.e. constant values in the range close to the h110i
directions.
Note that the polar plot of HC shown in Fig. 5.21(b) exhibits sharp peaks
along the hard axes similar to the Co2 MnSi samples discussed in the previous
Section.

5.3.3

Co2 FeAl

Figure 5.23(a) shows examples of hysteresis loops for the investigated


Co2 FeAl sample, which were recorded for different in-plane sample orientations . In this case, is defined as the angle between the [110] axis of the
Co2 FeAl film and the incidence plane of the laser beam. For particular values
of , the loops exhibit a slightly asymmetric shape, pointing to the presence
of a small second-order MOKE in the investigated film. The loop measured at
= 0 (i.e. along the [110] axis) has a typical rectangular shape for the easy
axis magnetization reversal, while the loop acquired at = 45 (i.e. along the
[100] axis) exhibits hard axis behavior (reduced remanence and smaller coercive
field).
The polar plot of the coercivity determined from the measured hysteresis loops is presented in Fig. 5.23(b). It shows a four-fold in-plane magnetic
anisotropy with h110i easy and h100i hard axis directions to which a weak
two-fold magnetic anisotropy is superimposed. Moreover, sharp peaks of HC
appear along the h100i directions, as was also observed for the previously discussed Co2 MnSi and CCFA samples.
The results of BLS measurements performed for the determination of the
[100]CFA

90
120

[110]CFA
30

150

10
0
-10

a=0
a=42
a=45

-20

-200 -150 -100

4
0

a
180

4
330

210

-30

(a)

60

20

HC (Oe)

MOKE (mdeg)

30

-50

50

H (Oe)

100

150

240

200

300
270

(b)

Figure 5.23: (a) MOKE loops of the Co2 FeAl(100) film for different sample orientations. (b) Polar plot of the coercivity HC .

86

Experimental results I - Magnetic properties of Co2 -based Heusler films


23.8

exp. data
numerical fit

23.6

[010]

[100]

23.5

[110]

Frequency (GHz)

23.7

23.4
23.3
23.2
23.1

30

60

90

120

150

180

Sample orientation (deg)

Figure 5.24: Frequency of the DE mode as a function of the angle between the
external magnetic field H = 1.0 kOe and the [110] direction for the Co2 FeAl(100)
film.

anisotropy constants are presented in Fig. 5.24, showing the frequency the DE
mode in dependence of the in-plane sample orientation . As above, is
defined with respect to the [110] axis of the Co2 FeAl film. The corresponding
BLS spectra were recorded in an external magnetic field H = 1.0 kOe and at
a transferred wave vector qk = 1.67 105 cm1 (i.e. = 45 ). As is shown in
Fig. 5.24, the spin-wave frequencies observed in the [100] ( = 45 ) and [010]
( = 135 ) directions of the Co2 FeAl film differ from one another, showing that
both these crystallographic axes are magnetically inequivalent. This result is in
agreement with the results of the above discussed MOKE investigations. Since
the [010] direction provides a smaller BLS frequency (Fig. 5.24), this direction
is a hard axis of the investigated Co2 FeAl film [170]. The two-fold anisotropy
found for the Co2 FeAl film is most likely a result of induced strain which might
originate from the lattice mismatch of Co2 FeAl with MgO.
Fitting the experimental data by the model described in Ref. [44] we obtain
the cubic volume anisotropy constant K1 = 675 kerg/cm3 for the investigated
Co2 FeAl film. The uniaxial anisotropy is found to be much smaller with a value
of Ku = 12 5 kerg/cm3 . This is in accord with the results obtained from
MOKE measurements.

5.3.4

Co2 FeSi

Examples of Kerr rotation loops collected from the 60 nm thick Co2 FeSi film
deposited on MgO with a Cr buffer are presented in Fig. 5.25 along with the

5.3 Magnetic anisotropies

87

90

60

120

40

15

20

10

60

30

150

HC (Oe)

MOKE (mdeg)

20

0
-20
a=0
a=6
a=80
a=170

-40
-60

a
5
0

180

10

-60

-40

-20

20

40

60

240

300

80

270

H (Oe)

(a)

330

210

15
20

-80

(b)

Figure 5.25: (a) MOKE loops of the Co2 FeSi(100) film for different sample orientations. (b) Polar plot of the coercivity HC .

dependence of the coercivity HC on the in-plane sample orientation. The polar


plot in Fig. 5.25(b) indicates the presence of an uniaxial magnetic anisotropy
in the investigated film. The direction of the easy axis is rotated by about 20
from the [110] axis of Co2 FeSi. For the 20 nm thick Co2 FeSi sample deposited
by the same technique, we find a similar dependence of HC on which suggests
that a uniaxial anisotropy is also present in the 20 nm thick film. These results
are rather surprising in view of the cubic symmetry of the crystal structure.
Results of BLS investigations performed for the 60 nm thick Co2 FeSi sample
10

BLS intensity (a.u.)

8
DE

PSSW1

6
a=90

a=30

a=15

DE frequency (GHz)

25.0

H=2.0 kOe
j=45
t=60 nm

(a)

24.5

24.0

23.5

(b)

a=0

-40

-30

-20

20

30

Frequency (GHz)

40

23.0

15

30

45

60

75

90 105 120

a (deg)

Figure 5.26: (a) Examples of BLS spectra collected form the 60 nm Co2 FeSi sample
in an external magnetic field H=2.0 kOe at = 45 (i.e. qk = 1.67105 cm1 ) and at
different in-plane orientation . (b) Dependence of the DE frequency on the in-plane
sample orientation .

88

Experimental results I - Magnetic properties of Co2 -based Heusler films


are summarized in Fig. 5.26. The BLS spectra were recorded in an external
magnetic field of H=2.0 kOe at = 45 (i.e. qk = 1.67 105 cm1 ), whereas
the in-plane sample orientation was varied in 15 steps between 0 and 130 .
In Fig. 5.26(a), BLS spectra measured at = 0, 15, 30 and 90 are exemplary
shown. As can be seen in this figure, the frequency of the peak originating from
the excitation of the DE mode does not significantly change upon variation of
. A plot of the DE frequencies determined from the collected BLS spectra
as a function of (Fig. 5.26(b)) exhibits some frequency variations, which are
in the range of 0.1-0.15 GHz. These variation, however, are at the limit of
measurement accuracy of the BLS setup. Therefore, the results presented in
Fig. 5.26(b) suggest that the magnetic anisotropy is negligible in the investigated Co2 FeSi films, i.e. K1 < 10 kerg/cm3 . This is in disagreement with the
strong uniaxial anisotropy evidenced by MOKE (Fig. 5.25(b)).
Hysteresis loops measured for the second set of Co2 FeSi samples, which were
deposited directly on MgO and where the Co2 FeSi layer has different thicknesses
d, are shown in Fig. 5.27(a). The loops presented in this figure were measured
at two different in-plane sample orientations = 22.5 . It can be observed
that the loops become more and more squared with increasing d. Furthermore,
90
120

50

a=22.5
o
a=!22.5

MOKE (mdeg)

11 nm

-25

340

HC (Oe)

25

21 nm

-50

60

360

d=21 nm
30

150

320
300

180

320
340

210

360

(b)

330

240

300
270

-75

90
120

230

-100

42 nm

-150

HC (Oe)

-125

98 nm

(a)

220

-175
-500

H (Oe)

500

1000

d=42 nm
30

150

210
200

180

210
220

-1000

60

230

330

210

(c)

240

300
270

Figure 5.27: (a) MOKE loops for Co2 FeSi (100) films with different thicknesses,
recorded at in-plane sample orientations = 22.5 (dashed line) and = 22.5
(full line). (b),(c) Polar plots of coercivity for the Co2 FeSi (100) films with film
thickness d = 21, 42 nm.

5.3 Magnetic anisotropies


the magnetic field necessary to saturate the sample becomes smaller when d
is increased. Note that the saturation filed is 500 Oe at small thicknesses,
which is a much larger value with respect to all other samples discussed above.
All of the loops presented in Fig. 5.27(a) exhibit an asymmetric shape arising
from a second-order contribution to the MOKE signal. The evolution of the
QMOKE with the thickness of the Co2 FeSi layer will be discussed in more detail
in Sect. 5.5.
After the symmetrization of the measured loops, the LMOKE loops were
used to determine the coercivity HC of the investigated samples. The dependence of HC on is presented in Fig. 5.27(b) and (c) for the film thicknesses
d = 21 and 42 nm, respectively. As is clearly visible from these polar plots, HC
is nearly constant with respect to in-plane orientation . For the film thickness
of 21 nm, the polar plot exhibits a very weak four-fold anisotropy, which modulates HC by less than 1 %. A similar behavior is also observed in the 11 nm
thick sample. For larger film thicknesses, a weak two-fold anisotropy (98 nm) or
a mixture of two-fold and four-fold anisotropy (42 nm (Fig. 5.27(c))) is found,
which again modulates HC by less than 1 %.
It is interesting to note that, in contrast to the first set of samples, the
Co2 MnSi films of the second set provide a single broad peak with an approximate width of 10 GHz.

5.3.5

Discussion

We first discuss the results of magnetic anisotropy studies presented in the preceding Subsections. Thereafter, coercivity values obtained for different samples
under investigation are summarized and discussed.
Magnetic anisotropies
The results of anisotropy investigations described in the previous Subsections
show that, except for the Co2 FeSi samples, all of the studied films possess a
four-fold in-plane magnetic anisotropy with h110i easy and h100i hard directions. The four-fold anisotropy is in accord with the crystal structure of the
films exhibiting a cubic symmetry. Moreover, the values of the cubic volume
anisotropy constants determined by BLS (see Table 5.2) are at least one order of magnitude smaller than those of bulk Fe and Co. A comparison of our
values with K1 values determined for thin films of pure Co and Fe by other
authors appears problematic. This is mainly due to the substantial variation of
reported K1 values with sample characteristics such as thickness, substrate and
fabrication technique. The small K1 values determined in this thesis for Heusler

89

90

Experimental results I - Magnetic properties of Co2 -based Heusler films


Sample

K1 (kerg/cm3 )

Co2 MnSi (B2)

-90(10)

Co2 MnSi (L21 )

-8(1)

Co2 FeAl (B2)

-67(5)

CCFA (B2)

-20(10)

Table 5.2: Cubic volume anisotropy constant K1 obtained from BLS measurements
for different Heusler films studied in this thesis.

compounds are in agreement with the fact that the orbital magnetic moments
are believed to be largely quenched in these compounds (see Sect. 3.3.5), which
results in a weak spin-orbit coupling and thus in a weak magneto-crystalline
anisotropy.
Regarding Co2 MnSi films annealed at different temperatures, we find that
the cubic volume anisotropy constant K1 drops by a factor of 10 when the
annealing temperature is increased from 375 to 500 C (Fig. 5.20(b)). Since
the increase of Ta is related to an improvement of the L21 order in the films
as revealed by XRD investigations (Sect. 5.1.2), this result suggests that the
magneto-crystalline anisotropy is smaller in the more ordered L21 structure
compared to the B2 structure. On the other hand, the decrease of K1 with
increasing annealing temperature might also be related to an interdiffusion
of Cr from the buffer layer which can occur at elevated temperatures. However, a negligible K1 value (< 10 kerg/cm3 ) found by BLS for the L21 -ordered
Co2 FeSi sample also grown on a Cr buffer layer strongly supports the conclusion
that the observed decrease of the magneto-crystalline anisotropy is an effect of
the improving crystal structure. The decrease of K1 with increasing L21 order
might be explained by a stronger quenching of the orbital moments in the more
symmetric L21 structure. Mn-Si disorder present in the B2 structure leads to
a less symmetric coordination of Co atoms, resulting in higher orbital magnetic moments and thus stronger magneto-crystalline anisotropy. It should be
noted that the same trend for K1 was also observed by Yilgin et al. in their
ferromagnetic resonance investigations of a similar Co2 MnSi sample set [171].
Interestingly, the cubic volume anisotropy constant of the predominantly
B2 ordered Co2 MnSi film is nearly five times as large as the corresponding value found for the CCFA film with the B2 structure. Since both the
Co2 MnSi and CCFA samples have a similar structure (i.e. substrate, buffer
and capping layer are of the same materials), it is possible to compare those re-

5.3 Magnetic anisotropies


sults. A larger K1 value found for the Co2 MnSi film is rather unexpected since
the orbital magnetic moments are found to be larger in Fe containing compounds than in those with Mn (Sect. 3.3.5). Therefore, the magneto-crystalline
anisotropy should be larger in the compounds where the transition metal is
Fe. The smaller magneto-crystalline anisotropy found for CCFA is probably
related to the partial substitution of Fe by Cr atoms. This assumption is additionally supported by a larger K1 value found for the Co2 FeAl film with
respect to the CCFA sample (see Table 5.2). The K1 value determined for the
Co2 FeAl film, however, is still smaller than that found for the Co2 MnSi film
with B2 structure. This might be due to distortions in the crystal lattice of
the Co2 FeAl film, as was indicated by the presence of a two-fold anisotropy
superimposed to the four-fold one.
While for Co2 MnSi, Co2 FeAl, and CCFA samples the results of BLS investigations are in accord with those of MOKE measurements (i.e. both techniques
reveal a four-fold magnetic anisotropy), there is a strong disagreement for the
Co2 FeSi sample deposited on the Cr-buffered MgO substrate. The BLS investigations suggest that magnetic anisotropies are negligible in the Co2 FeSi film,
whereas the polar plot of coercivity indicates a uniaxial anisotropy for this
sample. This discrepancy might be due to the fact that BLS measurements
are performed on magnetically saturated samples, while the coercivity is determined in a non-saturated state where usually magnetic domains are present. As
discussed in Sect. 2.1.3, the coercivity does not only depend on the anisotropy
energy, but also on the details of the magnetic microstructure such as critical
field for domain nucleation or strength of domain wall pinning, for example.
Coercivity
The coercivity values obtained for the investigated Heusler films which exhibit
a clear four-fold anisotropy are summarized in Table 5.3 for the h110i and h100i
crystallographic directions. The listed coercive fields are averaged values over
the four respective equivalent directions. For all of these films, the coercivity is quite low, reflecting a weak magneto-crystalline anisotropy discussed in
the previous Subsection. The low coercivity values also show that Co2 -based
Heusler compounds are soft magnetic materials, which makes them attractive
for technological applications.
In Co2 MnSi films, HC does not significantly change if the crystal structure
is changed from B2 to L21 . However, a closer look on the coercivity in the
complete series of Co2 MnSi samples annealed at different temperatures reveals
an increase of HC at intermediate temperatures around 425 C. This behavior
is clearly demonstrated in Fig. 5.28, where HC is plotted versus Ta for three
different in-plane sample orientations . The maximum of HC , appearing at

91

92

Experimental results I - Magnetic properties of Co2 -based Heusler films


Sample

HCkh110i (Oe)

HCkh100i (Oe)

Co2 MnSi (B2)

13.7

16.3

Co2 MnSi (L21 )

14.8

15.7

Co2 FeAl (B2)

9.9

10.2

CCFA (B2)

27.1

29.5

Table 5.3: Coercivity of Heusler samples which provide a four-fold anisotropy in the
polar plots of HC .

425 C for all , can most likely be attributed to an interplay of two counteractive effects, namely a reduction of the magneto-crystalline anisotropy observed
by BLS, and an enhanced number of pinning centers with increasing Ta . While
the former leads to a reduction of HC , the latter results in higher coercivity
values. In the Co2 MnSi samples investigated here, such an increased number of
pinning centers might originate from the diffusion of Cr atoms from the buffer
layer into the Co2 MnSi film.
In contrast to the Heusler films listed in Table 5.3, Co2 FeSi films deposited
directly on MgO support relatively large coercive fields of about 350 Oe, found
in the samples with thicknesses of 11 and 21 nm. The coercivity values of the
complete set of Co2 FeSi samples with different thicknesses are summarized in
Fig. 5.28(b). As is demonstrated in this figure, increasing the thickness from
400

20

Co2MnSi

Co2FeSi
300

a=45
12

HC (Oe)

HC (Oe)

a=0
16

200

100

a=42

(a)
8

(b)
0

350

375

400

425

450

475

Annealing temperature (C)

500

20

40

60

80

100

d (nm)

Figure 5.28: (a) Variation of coercive field HC with the annealing temperature in the
Co2 MnSi films for in-plane sample orientations = 0, 42 and 45 . (b) Dependence
of HC on film thickness in Co2 FeSi samples.

5.4 Magnetization reversal


21 to 98 nm results in a reduction of HC by almost a factor of 5, i.e. from
345 to 70 Oe. The high coercive fields at small thicknesses correlate with large
saturation fields of about 500 Oe. Both features are probably related to the
particularities observed in the surface morphology of similar films [172]. In
scanning electron microscopy images, the surface appears to be covered with
trenches, so that the films seems to consist of differently shaped grains with
dimensions in the range of several hundreds of nanometers. The trenches probably act as pinning centers, inhibiting the domain wall propagation during the
magnetization reversal. As the effect of the surface morphology is less important
in thicker films, we observe a reduction of HC with increasing film thickness.
This reasoning is also supported by small coercivity values in the range of 20 Oe
found for the Co2 FeSi samples with a Cr buffer layer, where no trenches are
observed.
The high saturation fields can be explained as follows. Due to the shape
anisotropy, the magnetization can be expected to have some preferential directions in the differently shaped grains. Therefore higher magnetic fields must be
applied in order to align the magnetization in a particular direction.

5.4

Magnetization reversal

To explain the origin of sharp peaks in HC observed along the h100i directions
for Co2 MnSi, CCFA and Co2 FeAl films, the magnetization reversal process (for
the CCFA sample) was studied in detail using MOKE microscopy (Sect. 4.2.2).
The measurements were performed in collaboration with Dr. R. Schafer at the
IFW Dresden. Kerr images were taken with the external magnetic field H
applied both exactly parallel ( = 0 ) and slightly rotated ( 3 ) out of the
direction of the [100] CCFA axis. In the following these two orientations of H
will be referred to as peak and out-of-peak orientations, respectively.

5.4.1

Magnetization reversal process

Figure 5.29(a) shows a Kerr image of the CCFA sample taken in the outof-peak orientation, after the sample had been saturated in a negative field
and a field of 21 Oe had been subsequently applied. The dots in this and all
subsequently discussed Kerr images are most probably small dust particles on
the sample surface, which do not influence domain propagation or nucleation.
The Kerr image shown in Fig. 5.29(a) was taken with the sensitivity of the
microscope set to the longitudinal component of magnetization, i.e. the magnetization component parallel to the direction of the applied magnetic field.

93

94

Experimental results I - Magnetic properties of Co2 -based Heusler films


sensitivity

H
sensitivity

(a)

(b)

(c)
Figure 5.29: (a),(b) Kerr microscopy images of the CCFA sample taken at H = 21 Oe
at a sample orientation = 3 (out-of-peak orientation) with sensitivity of the
microscope parallel and perpendicular to H. Magnetic contrast in (a) results from
the magnetization component parallel to H. The directions of magnetization in
different domains are indicated by white arrows. In (b) practically no contrast is
present. (c) Sketches of the observed stripy domain structure.

Therefore the dark and bright stripes visible in this figure indicate magnetic
domains with opposite orientation of the longitudinal component of magnetization. For the in-plane transverse magnetization component practically no
contrast was obtained (Fig. 5.29(b)), which indicates a constant magnetization
component in that direction. Hence, magnetization reversal solely happens by
the appearance of a stripy domain structure with a stripe direction parallel
to H, as it is sketched in Fig. 5.29(c). The magnetization within different
domains points along different easy axis directions. As such the domains in
adjacent stripes are separated by 90 domain walls.
Figure 5.30 shows Kerr images recorded with H applied parallel to the [100]
CCFA axis (i.e. peak orientation), whereas the left and right panels show
the results for H = 18 and 27 Oe, respectively. In each case, the sample had
been saturated in a negative field before applying the fields at which the Kerr
images were recorded. At H = 18 Oe, which is the field value slightly before
the magnetization reversal occurs, the CCFA sample exhibits a stripy domain
pattern with a strong magnetic contrast when the microscope is sensitive to
the transverse magnetization component (Fig. 5.30(b)). For the longitudinal
magnetization component, no contrast was found (Fig. 5.30(a)). Therefore,
the domain configuration consists of stripes, as is sketched in Fig. 5.30(c),

5.4 Magnetization reversal

95

sensitivity

sensitivity

(a)

(d)

sensitivity

sensitivity

(b)

(e)

(c)

(f)

Figure 5.30: Kerr microscopy images of the CCFA sample recorded at a sample
orientation = 0 (peak orientation) with H = 18 Oe [(a),(b)] and H = 27 Oe
[(d),(e)]. (c),(f) Sketches of the domain structure at H = 18 and 27 Oe, respectively.

which are again separated by 90 domain walls. In contrast to the out-ofpeak magnetization reversal, however, the stripe direction is now transverse
to H. If the external field is further increased, a jump in the hysteresis loop
occurs. The domain structure consists of domains in which the magnetization is
oriented along the four easy axes directions expected from the cubic anisotropy.
The domains are separated by 90 domain walls. This corresponds to the
formation of a checkerboard domain structure, as is sketched in Fig. 5.30(f).
Upon further increasing the external field, a stripy domain configuration with
a stripe direction perpendicular to H is again observed.
The reversal mechanisms in both the peak and the out-of-peak orientation
are schematically illustrated in Fig. 5.31.

96

Experimental results I - Magnetic properties of Co2 -based Heusler films


(a) 'peak' orientation:

(b) 'out-of-peak' orientation:

[110] CCFA

easy axis
directions

a=0E

27 Oe

-37 Oe

[110] CCFA

easy axis
directions

a.3E

[110] CCFA

[110] CCFA

.-50 Oe -17 Oe

/70 Oe

.-50 Oe -10 Oe

21 Oe

-30 Oe

/70 Oe

Figure 5.31: Schematic representation of the magnetization reversal (a) when H is


parallel to the h100i CCFA direction (peak orientation) and (b) when H is slightly
rotated out of the 100 CCFA direction (out-of-peak orientation).

5.4.2

Influence on coercivity

The peculiar domain structure during the magnetization reversal along the
h100i CCFA directions ( = 0 ) leads to a higher coercivity value compared
to 3 . This can be understood as follows. In the stripy domain structure
which appears prior to the checkerboard domain pattern, the magnetization lies
parallel to the two easy axes, which are closest to the direction of the external
field. Since this configuration is an energetically favorable state, it persists even
upon a certain increase of the external field. Only when a field is reached which
is higher than the nucleation field of the stripy domains, does a partial jump of
the magnetization occur into the energetically equivalent easy axes directions
with opposite orientations, and the checkerboard domain structure is formed.
The higher fields necessary for the formation of the checkerboard domain structure give rise to the coercivity peaks along the h100i CCFA directions.
The same considerations most likely also apply to other investigated Heusler
films which exhibit sharp peaks in their dependence of HC on the in-plane
sample orientation.

5.5

Quadratic magneto-optical Kerr effect

As demonstrated in Sect. 5.3.1, Kerr rotation loops recorded from the investigated Co2 MnSi films exhibit a considerable asymmetry, which provides evidence
of a strong second-order contribution to the MOKE signal in these films. An
asymmetry of a similar magnitude is also found in Co2 FeSi films deposited directly on MgO without the use of a Cr buffer layer. In other films studied in
this thesis, no or only weak second-order contributions are observed.
In this Section we will focus on those systems which provide a large QMOKE
signal, namely Co2 FeSi and Co2 MnSi. For the series of Co2 MnSi samples annealed at different temperatures, we investigated the evolution of the QMOKE

5.5 Quadratic magneto-optical Kerr effect

97

signal with the improvement of the L21 structure. For Co2 FeSi samples, the influence of the film thickness on the QMOKE signal was studied. In both cases,
the 8-directional method described in Sect. 4.2.1 was employed for a separate
determination of QMOKE contributions related to ML MT and ML2 MT2 .

5.5.1

Co2 FeSi

Figure 5.32(a) shows the two QMOKE contributions related to ML MT () and


ML2 MT2 (N) which were obtained for the 21 nm thick Co2 FeSi film at different in-plane sample orientation , and nearly perpendicular incidence angle
( 0.5 ). For comparison, the LMOKE signal in saturation is shown by ()
in this figure as well. As can be seen in Fig. 5.32(a), both QMOKE signals
exhibit an oscillating dependence on . This dependence can be described by
a cosine function (plus constant term) for the QMOKE contribution related to
ML MT , while the QMOKE contribution related to ML2 MT2 shows a sinusoidal
dependence. The observed dependencies of the two QMOKE signals on are
in accord with the analytical expressions for the Kerr signal (Eq. (2.14)) introduced in Sect. 2.2.1. The LMOKE signal does not change upon variation of ,
which is also in agreement with Eq. (2.14). A closer look to Fig. 5.32(b) reveals
that both oscillating dependencies have the same amplitude of 20 mdeg. Ac-

d=21 nm

30

60

LMOKE

MLMT
50

MOKE (mdeg)

MOKE (mdeg)

20
10

ML

0
-10
-20

40
30
20
QMOKE

10
2

ML-MT

(a)

-45

(b)

45

sample orientation (deg)

90

20

40

60

80

100

d (nm)

Figure 5.32: (a) Dependence of different MOKE signals in saturation on the inplane sample orientation determined for the 21 nm thick Co2 FeSi film using the
8-directional method. The incidence angle was 0.5 . (b) Dependence of the
QMOKE and LMOKE signals at saturation on the film thickness d. The QMOKE
signal was determined from the height of peaks in QMOKE loops, which were obtained by the antisymmetrization of MOKE loops measured at = 45 .

98

Experimental results I - Magnetic properties of Co2 -based Heusler films


cording to Eq. (2.14), however, the strengths of the two QMOKE contributions
should differ by a factor of 2. This factor of 2 is effectively cancelled by the
fact that ML MT in saturation is given by ML MT = M 2 cos sin and thus is
equal to (1/2) M 2 for = 45 , with being the angle between the direction
of the applied field and incidence plane. The maximal QMOKE signal reaches
30 mdeg.
The relationship between the QMOKE signal and the thicknesses of the
investigated Co2 FeSi films is presented in Fig. 5.32(b). Note that these data
points were determined from height of the peaks in the QMOKE loops, which
were obtained by the antisymmetrization of MOKE loops measured at the
in-plane sample orientation = 22.5 using the standard MOKE setup (i.e.
= 45 ). Increasing the film thickness to d = 21 nm results in an increase
of the QMOKE signal. Upon a further increase of the film thickness the
QMOKE signal decreases again reaching a value of 5 mdeg for the 98 nm thick
Co2 FeSi film. This value is about three times smaller than the QMOKE signal
of 17 mdeg determined for the 21 nm thick sample. The LMOKE which is also
shown in Fig. 5.32(b) for comparison, increases with increasing film thickness,
and saturates at high thicknesses. The possible origin of these observations will
be discussed in Sect. 5.5.3.

5.5.2

Co2 MnSi

Figure 5.33 shows the ML MT and ML2 MT2 contributions to the QMOKE signal for the Co2 MnSi film annealed at (a) 425 and (b) 500 C in dependence
of the in-plane sample orientation . The data points were obtained by the
application of the 8-directional method to the MOKE signal measured in saturation at an incidence angle = 0.5 of the probing beam. The data exhibit a
clear oscillating dependence on for both QMOKE contributions, which is in
accord with the analytical expression for the Kerr signal given by Eq. (2.14).
However, in contrast to the previously described QMOKE in Co2 FeSi samples,
the contributions proportional to ML MT and ML2 MT2 have different amplitudes. For the Co2 MnSi sample annealed at 425 C, the amplitude of the ML MT
contribution is larger than that of the ML2 MT2 contribution. The situation is
reversed for the film annealed at 500 C.
The other films in the series of Co2 MnSi samples with varying annealing
temperatures (i.e. varying L21 order) exhibit a similar oscillating behavior upon
the variation of . A comparison of QMOKE signals obtained for the complete
set of samples shows that both contributions to the QMOKE signal increase in
amplitude with increasing annealing temperature (Fig. 5.34(a)). The LMOKE,
which is also shown in Fig. 5.34(a) for comparison, does not significantly change

5.5 Quadratic magneto-optical Kerr effect

99

8
6

MLMT

QMOKE (mdeg)

QMOKE (mdeg)

10

MLMT

4
2
Ta =425C

-2

4
2
Ta =500C

0
-2

ML-MT

(a)

-45

-4
-30

-15

15

30

45

ML-MT

(b)

-45

-30

Sample orientation (deg)

-15

15

30

45

Sample orientation (deg)

Figure 5.33: Dependence of QMOKE contributions determined in saturation by the 8directional method on the in-plane sample orientation for the Co2 MnSi film annealed
at (a) 425 and (b) 500 C.

with Ta . Figure 5.34(a) reveals in addition that the increase of the ML2 MT2
amplitude is stronger compared to the ML MT amplitude. This leads to the
fact that the ML MT contribution which is larger at low annealing temperatures
becomes weaker compared to the ML2 MT2 at higher Ta . Moreover, we observe
an increase of both the maximal QMOKE signal and the offset of the QMOKE
contribution proportional to ML MT when annealing temperature increases.

10

max(MLMT)

QMOKE (mdeg)

MOKE (mdeg)

MLMT

1
2

ML-MT

ML

8
7
6
5

-1

(b)

(a)
375

400

425

450

475

Annealing temperature (C)

500

375

400

425

450

475

500

Annealing temperature (C)

Figure 5.34: (a) Dependence of the QMOKE amplitudes on the annealing temperature of the Co2 MnSi films. For comparison, the corresponding dependence of the
LMOKE is shown as well. (b) Dependence of the maximal QMOKE signal on the
annealing temperature.

100

5.5.3

Experimental results I - Magnetic properties of Co2 -based Heusler films

Discussion

Thickness dependence of QMOKE


To understand the thickness dependence of QMOKE observed in Co2 FeSi films
and its different behavior compared to the LMOKE signal, we additionally
measured the first and second order contributions to Kerr ellipticity. The results
of these measurements
are shown in Fig. 5.35 along with the Pythagorean

average = 2 + 2 for both the LMOKE and QMOKE signals. As is clearly


visible from this figure, Kerr ellipticity exhibits an opposite behavior compared
to that of the Kerr rotation demonstrated in Fig. 5.32(b). While the QMOKE
ellipticity increases with increasing thickness saturating at high values of d,
the LMOKE ellipticity reaches a maximum at d = 21 nm and then decreases
again. The reversed dependence of and on the film thickness for QMOKE
and LMOKE leads to a similar behavior of the Pythagorean average. For both
QMOKE and LMOKE signals reaches a maximum in the range of 20-30 nm
and then saturates with further increasing d.
The decrease of at higher thicknesses can be understood considering the
depth sensitivity of the magneto-optical Kerr effect (see also Appendix A). In
accord with Eq. (A.3), the Kerr effect amplitude originating from an ultra-thin
sublayer of thickness t situated in at a depth ti can be expressed as

4iti Nz,CFS
i Ct exp
,
(5.4)

p
2
2
where Nz,CFS = NCFS
Nair
sin2 is the normalized k-vector of light in zdirection, Nair the refractivity index of air and C a complex constant. The
total Kerr effect amplitude tot , which is related to the Pythagorean average
through the relation = |tot |, is given byP
the summation over all contributions
originating from different depths, tot = i i . Due to the exponential term
in Eq. (5.4), Kerr signal originating from a deeper sublayer ti exhibits a larger
damping as well as a larger shift in phase. The phase difference of Kerr signals
originating from different depths ti and tj is ij = 4 Re(Nz,CFS )(ti tj )/.
If the ferromagnetic films are thick and transparent enough, as it is the case
of films studied here, the Kerr effect amplitudes i , j from depths ti , tj may
differ by a phase of . In such a case, i and j cancel each other leading to a
reduction of the resulting Kerr effect amplitude tot .
From the above discussion and the fact that the Pythagorean average
exhibits similar dependence on d for both LMOKE and QMOKE we conclude
that the peculiar behavior of LMOKE and QMOKE signal in Fig. 5.32 is a
purely optical effect, which originates from the depth-sensitivity of the MOKE.

5.5 Quadratic magneto-optical Kerr effect

70

101

LMOKE

MOKE (mdeg)

60
50

,
,

40

Kerr ellipticity e
2
2 (1/2)
(q + e )

30
20
QMOKE

10
0
0

20

40

60

80

100

d (nm)

Figure 5.35: Kerr ellipticity and Pythagorean average of the Kerr ellipticity and

rotation = 2 + 2 for the QMOKE and LMOKE signals.

Annealing temperature dependence of QMOKE


The increase of the QMOKE amplitudes as well as of the maximal QMOKE
signal with the annealing temperature found in the Co2 MnSi films indicates that
the second order contribution to the MOKE signal becomes more pronounced
with the improving L21 order in the films. This is also corroborated by the very
high QMOKE amplitude of 20 mdeg and maximal QMOKE signal of 30 mdeg
observed in the Co2 FeSi(21 nm) film with L21 structure. The origin of the high
QMOKE signal observed in the investigated Co2 MnSi and Co2 FeSi films is not
yet fully understood and requires more systematic studies.
As discussed in Sect. 5.5.1, the amplitudes of both QMOKE contributions
are expected to be equal for a crystal with a cubic symmetry. Different amplitudes found in the Co2 MnSi films point therefore to the presence of a lattice
distortion. Additional XRD investigations, in which mainly the presence of
the tetragonal distortion was checked, reveal indeed a small difference of the
in-plane and out-of-plane lattice constants. The dependence of the lattice constants on the annealing temperature, however, is found to be too weak to deduce
any systematic evolution with Ta , which possibly could explain the changes of
the QMOKE amplitudes shown in Fig. 5.34(a). From these results it seems that
the QMOKE has a much higher sensitivity to small variations in the crystal
lattice compared to XRD.
Finally, we address the fact that practically no or a very weak QMOKE
contribution to the detected MOKE signal is found for other Heusler films
studied in this thesis. In Co2 FeAl(80 nm) and CCFA(80 nm) samples, this is

102

Experimental results I - Magnetic properties of Co2 -based Heusler films

probably due to both the B2 structure of the films and the reduction of the Kerr
signal appearing at higher film thicknesses. A weak QMOKE contribution in the
Co2 FeSi sample deposited on the Cr-buffered MgO substrate is quite surprising
since it exhibits L21 structure according to XRD investigations. Thus it is
expected to provide a large QMOKE signal according to the previous discussion.
Moreover, it is also worthwhile to mention that a significant QMOKE contribution to the measured MOKE signal is by no means a particular feature
of the Heusler films. Second-order MOKE, manifesting itself in an asymmetry
of the hysteresis loops acquired by MOKE magnetometry, has been reported
for various other systems and is rather typical for thin ferromagnetic films consisting of materials with a cubic symmetry of the crystal lattice. Thin Heusler
films, however, are found to exhibit the highest QMOKE signal reported so far.

Chapter 6
Experimental results II Modification of Heusler films by
ion irradiation
In Chapt. 5, we have demonstrated that the exchange stiffness D and the
cubic volume anisotropy constant K1 of the Heusler compounds are significantly
influenced by the structural order. As introduced in Sect. 4.4, structural properties of magnetic thin films and multilayers can be tuned by ion irradiation.
In view of a possible application of this technique to thin films consisting of
Heusler compounds, this thesis also explores the effects of the ion bombardment
on the properties of Heusler films.

6.1

Preliminaries

In the ion irradiation experiments carried out in the present thesis, two different ion species were used. On the one hand, the Heusler films were bombarded
with very light He+ ions of two different energies, namely 30 and 130 keV. On
the other hand, heavy 30 keV Ga+ ions were employed for the irradiation. The
main goal of experiments with He+ was to explore the applicability of the
light-ion irradiation technique as an alternative technique to the conventional
high-temperature annealing for the promotion of a partially disordered Heusler
structure to the L21 phase. These investigations are motivated by the experimental work reported for the binary FePt(Pd) alloy, where an improvement of
the long-range order parameter was induced by 130 keV He+ irradiation com-

104

Experimental results II - Modification of Heusler films by ion irradiation

bined with a mild annealing [12, 13]. It should be noted, however, that in
contrast to the L21 structure of the Heusler films studied within the framework
of this thesis, FePt(Pd) crystallizes in a much simpler L10 structure which is a
fcc lattice with successive crystallographic planes occupied by a different kind
of atom. Irradiation experiments employing heavy Ga+ ions were carried out in
particular with regard to a potential application of the focused ion beam (FIB)
technique for the micro-patterning of Heusler films.
In our irradiation experiments, we focused on two material systems, namely
on Co2 MnSi and Co2 FeSi. The irradiation with He+ ions was performed for
Co2 MnSi, since an extensive study of the structural and magnetic properties
of this compound in dependence of the annealing temperature was present (see
Chapt. 5). In this way a direct comparison of both the high-temperature annealing and the light-ion irradiation techniques would be possible. Ga+ irradiation
was carried out for Co2 FeSi. This choice is mainly based on the simple fact
that L21 ordered Co2 FeSi films were easily available. Before carrying out the
irradiation experiments, the irradiation process was simulated using the software SRIM (Stopping and Range of Ions in Matter) [149, 150], which allowed
for the estimation of a suitable range of ion energies and ion fluences used in
the experiment. In case of He+ irradiation, the starting point for the choice of
ion energies was the aforementioned preexisting experimental work carried out
for the FePt(Pd) system.
In the following, the results of SRIM simulations will be shown in detail
for both utilized ion species. First, however, we will give an overview of the
Heusler samples used in the ion irradiation experiments.

6.1.1

Overview of investigated samples

Co2 MnSi
Co2 MnSi(100) films were fabricated in the group of Prof. Dr. Y. Ando at the
Tohoku University, Sendai. The samples consist of an epitaxial MgO(100)/
Cr(40nm)/Co2 MnSi (30nm) structure covered by either a 1.3nm thick Al or
2nm thick Ta capping layer. For the deposition of both the Cr buffer and the
Co2 MnSi layers inductively coupled plasma assisted magnetron sputtering was
employed. After the growth of the Co2 MnSi layer, the sample was annealed at
350 C resulting in a predominant B2 structure of Co2 MnSi as will be demonstrated in more detail in the following. To ensure equal initial conditions and
thus the comparability of the irradiated films, two 1 in2 Co2 MnSi samples (with
either Al or Ta capping layer) were prepared, which were cut into 55 mm2
pieces before carrying out the irradiation.

6.1 Preliminaries

105

30

j-axis

(b)

Cr (200)

202
Co2MnSi
(400)

MgO
(200)

Co2MnSi
(200)

Intensity (a.u.)

(a)

100

220

40

50

60

220
c-axis

202

70

2q (deg)

Figure 6.1: (a) X-ray -2 scan and (b) (220) pole figure of the as deposited
MgO/Cr(40 nm)/Co2 MnSi(30 nm)/Al(1.3 nm) sample.

Results of XRD structural characterization of the as deposited Co2 MnSi film


covered by Al protective layer are shown in Fig. 6.1. These data were acquired
and analyzed during the research stay in the group of Prof. Dr. Y. Ando at the
Tohoku University. The (220) equivalent reflections were observed with fourfold symmetry in the corresponding x-ray pole figure (Fig. 6.1(b)), providing
evidence of the epitaxial growth of the Co2 MnSi film. The shape of these
reflections, however, is slightly elongated which might be due to the presence
of grains with slight crystallographic disorientation or stacking faults such as
twinned grains for example [115]. Additionally, the observed broadening of
the diffraction peaks might originate from a nonuniform strain present in the
film. The x-ray -2 diffraction pattern (Fig. 6.1(a)) exhibits clear (200) B2
superstructure reflections. Since no (111) reflections, indicative for the L21
phase, were detected in the corresponding pole figure (not shown here), it is
concluded that the fabricated film was predominantly B2 ordered. However,
the presence of a certain amount of A2 type disorder cannot be excluded from
these results. The Co2 MnSi sample capped with a Ta layer provides similar
results of XRD characterization.
Co2 FeSi
The Co2 FeSi(100) film used for the irradiation with Ga+ ions consists of an
epitaxial MgO(100)/Co2 FeSi(11nm)/Al(4nm) structure prepared by RF magnetron sputtering in Mainz by Dr. G. Jakob and collaborators. This sample
is a part of a Co2 FeSi series with four different thicknesses which was mainly
used for the investigation of the QMOKE in Chapt. 5. As already mentioned
therein, the epitaxial growth of the layer stack was determined by means of

106

Experimental results II - Modification of Heusler films by ion irradiation

XRD [153]. The XRD measurements also revealed the L21 structure of the
deposited Co2 FeSi layer. Before the irradiation, a patterned Cu layer was put
on top of the sample defining nine squared areas of 1 mm2 size. To each of these
areas a different fluence of Ga+ ions was applied during the irradiation. One
squared area was left nonirradiated for reference. Deposition and patterning
of the Cu layer was carried out at the Nano-Bio-Center at the University of
Kaiserslautern. For the patterning photolithography was employed.

6.1.2

Simulations of the irradiation process

Prior to ion irradiation an estimation of appropriate irradiation parameters as


well as of the damage impact, which the ion bombardment would have on the
structural properties of the films, was made. For this purpose, simulations of
the irradiation process were performed using the software SRIM [149, 150]. Especially in case of He+ irradiation of Co2 MnSi films, it was of major importance
to avoid implementation of ions into the Heusler layer and to ensure low defect
densities caused by the ion bombardment. Only with these conditions fulfilled
the intended improvement of structural and magnetic properties of the Heusler
layer might be feasible. In all simulations presented in the following, the number of incoming ions was set to 3000, which is high enough to give adequate
statistics.
He+ irradiation of Co2 MnSi
Figure 6.2 shows the results of SRIM simulations performed for 30 keV He+ ions
impinging on a Co2 MnSi sample. In Fig. 6.2(a), a projection of ion trajectories
onto the thickness of the multilayer stack is illustrated showing that most of
the impinging He+ ions are stopped in the MgO substrate. The calculated
distribution of ion ranges in the stack (Fig. 6.2(b)) makes this even more clear
exhibiting a maximum located well inside the substrate. Note that here the
ion range is the depth at which the ion is stopped in the irradiated specimen.
Also the mean ion range of 248 nm, calculated by the SRIM software, is much
larger than the thickness of the Co2 MnSi layer.
For estimation of an appropriate range of ion fluences, detailed calculations
of ion-induced damage cascades were performed. The results of these calculations are shown in Fig. 6.2(c), where the trajectories of knocked out target
atoms are illustrated for the Al (green), Co2 MnSi (blue) and Cr (grey) layers.
According to this figure, displacements of target atoms are mostly concentrated
in a close proximity of ion trajectories, which guarantees low displacement and
thus low defect densities, respectively. Low defect densities are also evidenced

(a)

Co2MnSi
Cr
MgO

107

MgO

Co2MnSi
Cr

6.1 Preliminaries

(b)

Cr

Al

Co2MnSi

Cr

(c)

Co2MnSi

Al

(d)

Figure 6.2: Results of SRIM simulations performed for the irradiation of the
MgO/Cr(40nm)/Co2 MnSi(30nm)/Al(1.3 nm) sample with 30 keV He+ ions. (a) Ion
trajectories in the multilayer stack including the substrate. (b) Distribution of ion
ranges for He+ ions in the multilayer stack. (c) Trajectories of the displaced atoms
in the Al (green), Co2 MnSi (blue) and Cr (grey) layer. (d) Depth distribution of
displaced atoms in different layers of the multilayer stack.

by the calculated depth distribution of displaced target atoms which is presented in Fig. 6.2(d). The number of displaced atoms in the Co2 MnSi layer
is in the order of 4103 per incoming ion and nanometer. Allowing every
atom in the Co2 MnSi layer to be displaced 0.01 times and assuming the density
of Co2 MnSi to be 8.851022 atoms/cm3 , we obtain the result that the applied
ion fluences should be in the range of 21014 ions/cm2 . The total damage introduced into the crystal lattice at this particular fluence can be estimated
taking into account that most displacements result in creation of Frenkel pairs
(Sect. 4.4), the majority of which instantly recombines leaving only about 10 %
of surviving vacancies [13]. This gives a vacancy density of 81019 cm3 . Hence

108

Experimental results II - Modification of Heusler films by ion irradiation

(b)

(a)

MgO

MgO

Co2MnSi

Al

(c)

Cr

Cr

Co2MnSi

Al

(d)

Figure 6.3: Results of SRIM simulations performed for the irradiation of the
MgO/Cr(40nm)/Co2 MnSi(30nm)/Al(1.3 nm) sample with 130 keV He+ ions. (a) Ion
trajectories in the multilayer stack including the substrate. (b) Ion range distribution of the impinging ions. (c) Trajectories of the displaced atoms in the Al (green),
Co2 MnSi (blue) and Cr (grey) layer. (d) Distribution of displaced atoms in different
layers of the multilayer stack.

the Co2 MnSi target is damaged to about 0.1 % in this case.


Supported by the above results, the ion fluences were chosen in the range
between 1014 and 1016 ions/cm2 for the irradiation of Co2 MnSi samples with
30 keV He+ ions. It should be noted, however, that SRIM simulations give only
a very rough estimation of the real situation because this software does not take
into account the changes of the target during the irradiation. Every impinging
ion is rather assumed to move through the lattice with no changes of crystal
integrity. This might result in an underestimation of the damage introduced
into the lattice by ion bombardment.
In addition to 30 keV He+ ions, the Co2 MnSi films were also irradiated with

6.1 Preliminaries

109

He+ ions with an energy of 130 keV. Note that this is the energy for which an
improvement of the L10 order was achieved in the binary FePt(Pd) alloy [13].
The results of corresponding SRIM simulations are presented in Fig. 6.3. Due
to a much higher energy compared to the previously considered 30 keV He+
ions, 130 keV ions travel a larger distance before they lose their kinetic energy
and are stopped in the multilayer stack. Therefore, the maximum of the ion
range distribution is shifted further inside the MgO substrate, and the mean ion
range increases to 992 nm (Fig. 6.3(b)). Moreover, the radial distributions of
He+ ions in both the Co2 MnSi and Cr layers becomes much narrower compared
to the case of 30 keV ions (Fig. 6.3(c)). A consequence of that is a reduced
averaged number of displaced target atoms (Fig. 6.3(d)). Additionally, the
depth distribution of target displacements becomes less uniform exhibiting more
variations with changing depth. In the experiment, the ion fluences were chosen
in same range as in case of 30 keV ions for reasons of comparability. It should
be kept in mind, however, that the damage impact provided by 30 keV ions at
21014 ions/cm2 is now expected at higher fluences.
Changing the cap material from Al to Ta does not dramatically modify
the distributions of He+ ions inside the multilayer stack. For both considered
energies, i.e. 30 and 130 keV, its maximum is still located inside the MgO
substrate. Moreover, the ion ranges are only slightly reduced (from 248 to
243 at 30 keV and from 992 to 988 nm at 130 keV) compared to the case of Al
cap. The calculated defect densities, which are created in the Co2 MnSi and Cr
layers by the ion bombardment, remain nearly unchanged when Al is replaced
by Ta. Therefore, the choice of the cap material should not strongly affect the
comparability of irradiation experiments carried out at Co2 MnSi films covered
either by Al or Ta protective layer.
Ga+ irradiation of Co2 FeSi
Results of SRIM simulations performed for the irradiation of the 11 nm thick
Co2 FeSi layer with 30 keV Ga+ ions are presented in Fig. 6.4. In contrast to
the previously discussed He+ ions, the travelling distance of Ga+ ions is much
shorter in the considered sample. The calculated mean ion range is only 17 nm
so that the majority of impinging Ga+ ions are stopped in the Co2 FeSi layer.
This is clearly demonstrated in Fig. 6.4(a) and (b), where ion trajectories and
ion range distribution are visualized, respectively. Moreover, irradiation with
Ga+ ions leads to much larger defect densities in the Heusler layer compared
to He+ ions. According to SRIM simulations (Fig. 6.4(d)), the averaged number of displaced atoms in the Co2 FeSi layer is 50 per incoming ion and per
nanometer, which is four orders of magnitude larger than the corresponding
value calculated for 30 keV He+ ions. The higher damage impact of Ga+ ions

MgO

Co2FeSi

(a)

Al

MgO

Co2FeSi

Experimental results II - Modification of Heusler films by ion irradiation

Al

110

(b)

MgO

Co2FeSi

(c)

Al

MgO

Co2FeSi

Al

(d)

Figure 6.4: Results of SRIM simulations performed for the irradiation of the
MgO/Co2 FeSi(11nm)/Al(4 nm) sample with 30 keV Ga+ ions. (a) Ion trajectories in
the multilayer stack including the substrate. (b) Ion range distribution of the impinging ions. (c) Trajectories of the displaced atoms in the Al (green), Co2 FeSi (blue) and
Cr (grey) layer. (d) Depth distribution of displaced atoms in the multilayer stack.

is a direct consequence of the predominant nuclear stopping mechanism relevant for heavy ions. Due to large energy transfers to the primarily hit target
atoms, these have enough energy to undergo further collisions and release their
collision partners from their lattice sites. Hence, extended collision cascades
appear involving a much larger volume of the target material where atoms are
in motion (Fig. 6.4(c)).
From the above SRIM simulations, the density of vacancies introduced
into the Co2 FeSi layer by Ga+ irradiation at a fluence of 21014 ions/cm2
is estimated to be 1023 cm3 . Assuming again that only 10 % of the initially created vacancies survive and that the atomic density of Co2 FeSi is
8.921022 atoms/cm3 , this yields a value of 11 % for the damage created in

6.2 He+ irradiation of Co2 MnSi

111

the Co2 FeSi layer. Note that this value is by a factor of 100 larger with respect
to the total damage of 0.1 % estimated in the previous Subsection for the irradiation of Co2 MnSi with 30 keV He+ ions at the same fluence. Increase the
ion fluence by one order of magnitude (i.e. to 21015 ions/cm2 ) would already
result in a 100 % damage, i.e. melting of the target material. Hence, the ion fluences in the range between 1014 and 1015 ions/cm2 should be enough to induce
modifications of structural and magnetic properties of the Co2 FeSi layer.
Based on the above results of SRIM simulations, the ion fluences employed
in the experiment were varied between 31014 and 91016 ions/cm2 .

6.2

He+ irradiation of Co2MnSi

As already mentioned before, He+ irradiation of Co2 MnSi films was performed
with ions of two different energies, namely 30 and 130 keV. In case of 30 keV
ions, the irradiation was carried at RT as well as at elevated temperatures of 150
and 250 C. The irradiation of Co2 MnSi with 130 keV He+ ions was performed
at RT only. During the irradiation the entire film surface was exposed to
the beam. After the irradiation the samples were characterized by various
techniques including XRD, BLS, MOKE and SQUID magnetometry. The films
irradiated with 30 keV He+ ions at RT were additionally investigated by means
of x-ray absorption and circular magnetic dichroism (XAS/XMCD) as well as
photoemission spectroscopy at high energies (HAXPES).
The XAS/XMCD measurements were performed by Prof. H. J. Elmers and
collaborators (University of Mainz) at the beamline UE56/1-SGM at BESSY
II in Berlin. The samples were magnetically saturated by an external magnetic
field of 1.6 T applied perpendicularly to the film surface. The x-ray polarization
was kept constant while the magnetization direction was switched to determine
the XMCD signal. Collecting of x-ray absorption spectra occurred in two different ways [173, 174]. On the one hand, the x-ray absorption was measured
directly in a transmission experiment (TM) giving access to magnetic moments
averaged along the film normal. On the other hand, the x-ray absorption was
determined indirectly by detecting the total electron yield (TEY). Due to the
limited escape depth of the photoemitted electrons originally stemming from
the Auger relaxation of the 2p-core hole, the TEY method is surface sensitive
and allows therefore for a separate analysis of magnetic moments at the upper
film surface.
The HAXPES experiments were performed by Dr. A. Gloskovskii and collaborators (University of Mainz) at the beamline BL47XU of SPring-8 (Japan).
The photon energy was fixed at 7.940 keV. The inelastic mean free path of the

112

Experimental results II - Modification of Heusler films by ion irradiation

8 keV electrons is expected to be about 23 nm in AlOx and 8 nm in Co2 MnSi


[175]. This allows an investigation of the bulk electronic properties of Co2 MnSi
films below the capping layer [176]. For the analysis of the kinetic energy of the
photoemitted electrons a hemispherical analyzer (Scienta; R4000-12kV) with
an overall energy resolution of 250 meV [75] was used.
Characterization of Co2 MnSi films employing a SQUID magnetometer was
carried out by H. Schneider at the University of Mainz.
In the following, we first discuss the effect of RT irradiation with 30 and
130 keV on the Co2 MnSi films. Thereafter, the results of irradiation experiments
carried out at elevated temperatures with 30 keV He+ ions will be shown.

6.2.1

RT irradiation 30 keV He+

XRD investigations
Figure 6.5(a) summarizes -2 scans measured for the Co2 MnSi films which
were irradiated with 30 keV He+ ions at RT and at different fluences. These
data were acquired and analyzed during the research stay in the group of
Prof. Dr. Y. Ando at the Tohoku University. The scans shown in Fig. 6.5(a) look
very similar to -2 pattern of the as deposited Co2 MnSi sample (Fig. 6.1(a)).
In particular both Co2 MnSi peaks are still present in the scans after the ion
irradiation. Moreover, their intensities are not significantly changed upon the
increasing of the applied ion fluence. Also the (220) pole figures recorded for the
irradiated films remain similar to the one recorded for the nonirradiated sample
(Fig. 6.1(b)). From these results it can be concluded that the crystallinity and
epitaxy of the Co2 MnSi layer are largely preserved after the irradiation. It
should be noted, however, that the shape of the MgO peak, which is present
in the -2 scans in Fig. 6.5(a), is substantially modified at higher applied fluences. Going back to the discussion of the origin of the MgO double peak in
Chapt. 5, the appearance of a single peak at higher fluences points to a reduction of intensity of the MgO reflection. This is most likely due to an enhanced
damage of the MgO crystal lattice at higher ion fluences.
In the (111) pole figures measured for the irradiated Co2 MnSi films (not
shown here), no evidence of the (111) equivalent reflections is observed. Therefore we conclude that no transition from the B2 to the L21 structure has occurred in the Co2 MnSi films after the He+ bombardment at chosen parameters.
However, a detailed analysis of -2 scans presented in Fig. 6.5(a) reveals at
the fluences of 11014 and 51014 ions/cm2 a slight increase of the ratio of
(200) and (400) integrated intensities with respect to the nonirradiated film
(Fig. 6.5(b)). A further increase of the ion fluences results in a decrease of

6.2 He+ irradiation of Co2 MnSi

113

Figure 6.5: (a) -2-scans of MgO/Cr(40 nm)/Co2 MnSi(30 nm)/Al(1.3 nm) films irradiated with 30 keV He+ ions at different fluences. (b) Ratio of the (200) and (400)
integrated XRD intensities as a function of the applied ion fluence.

the (200)/(400) ratio, which is followed by an unexpected rising trend at the


largest applied fluence of 41016 ions/cm2 . According to powder pattern simulations (PoderCell software [118]), the (200)/(400) ratio is expected to be about
30 % for both the B2 and L21 ordered Co2 MnSi compound. Therefore, the increase of this ratio at the aforementioned fluences could be an indication of
an improvement of the B2 structure and possibly a local appearance of L21
ordered regions in the irradiated films. Additional investigations by means of
SQUID magnetometry, which will be discussed in the following Subsections,
support the conclusion of an improving order only for the fluences of 11014
and 51014 ions/cm2 .
XAS/XMCD spectra
XAS and XMCD data recorded at the Co and Mn L2,3 edges in a transmission
experiment are presented in Fig. 6.6 comparing irradiated and nonirradiated
samples. The data were provided by Prof. H. J. Elmers and collaborators (University of Mainz). A doublet structure appears in the Mn L2 region (Fig. 6.6(b))
and a weak satellite peak is visible in the Co spectra 3.8 eV above the L3 absorption edge (denoted by a vertical line in Fig. 6.6(a)). Both these features
have been reported for thin Co2 MnSi films by different groups [177179]. In
particular, it has been demonstrated that the intensity of the Co L3 satellite
(also referred to as Heusler bump in literature) directly correlates with the
ordering degree inside the Co2 MnSi layer [178]. Hence, the XAS spectra shown
in Fig. 6.6(a) were further processed to evaluate the intensity of the Co L3

114

Experimental results II - Modification of Heusler films by ion irradiation


2.0

1.0

(a)

(b)

(m ++m- )d/2

1.5

1H10

16

5H1015

0.5

1H10

1.0

15

5H1014

0.5

1H1014
no irrad.
0.0
0.4

L3

L2

L3

0.0

L2

1,2
0.3
0,8

(m+-m- )d/2

0.2
0,4

0.1
0.0

0,0

-0.1
-0,4

Co 2p Y 3d,TM

Mn 2p Y 3d, TM

-0.2
780

790

800

810

Photon energy (eV)

820

640

650

660

Photon energy (eV)

Figure 6.6: Transmission XAS and XMCD spectra at (a) Co and (b) Mn L2,3 absorption edges collected from Co2 MnSi films irradiated with 30 keV He+ ions at different
fluences. Vertical line in (a) indicates the Co L3 shoulder sensitive to the chemical
order in the volume of Co2 MnSi films.

satellite.
Figure 6.7(a) shows the XAS spectra in the energy range of the Co L3
satellite, which were obtained after linear background subtraction from the
spectra presented in Fig. 6.6(a). Note that the analysis of the XAS/XMCD
data was also carried out by by Prof. H. J. Elmers and collaborators (University of Mainz). The spectra are presented for the nonirradiated film as well
as for the films irradiated with 30 keV He+ ions at the fluences of 11014 ,
51014 and 11015 ions/cm2 , respectively. As can be seen in this figure, the
intensity of the satellite is noticeably increased for the fluences of 11014 and
51014 ions/cm2 compared to the nonirradiated film. At 11015 ions/cm2 , a
reduction of the satellite intensity is observed. This behavior is more clearly
illustrated in Fig. 6.7(b), where the relationship between the intensity of the Co
L3 satellite and the ion fluence is resented in the whole range of applied fluences.
The increase of intensity of the Co L3 satellite observed for the films irradiated

6.2 He+ irradiation of Co2 MnSi

115

md

I (a.u.)

(a)
782

(b)
784

hn (eV)

786

1014

1015

1016
2

Fluence (ions/cm )

Figure 6.7: (a) Transmission XAS spectra in the range of the Co L3 satellite peak
after linear background subtraction. (b) Dependence of the Co L3 satellite peak on
the He+ ion irradiation.

with He+ ions at fluences of 11014 and 51014 ions/cm2 , suggests that in this
particular range of applied fluences the local order in the bulk of Co2 MnSi films
has been increased by the irradiation [178]. At fluences above 51014 ions/cm2 ,
however, the irradiation with 30 keV He+ ions most likely leads to a reduction
of the local order, as suggested by the decreasing intensity of the Co satellite
in this range of fluences.
It should be noted that for the Co2 MnSi film irradiated at 41016 ions/cm2 ,
the XAS/XMCD spectra could not be measured in the transmission mode due
to a low luminescence signal. This is probably related with the modification
of the MgO substrate by He+ irradiation at higher fluences which was also
suggested by the results of XRD investigations discusses in the previous Subsection.
Figure 6.8 shows the XAS/XMCD spectra collected using the surface sensitive TEY. At both Co and Mn L2,3 absorption edges significant changes are
observed with respect to the above discussed bulk spectra. Namely, a multiplet
structure appears in the Mn XAS spectra (Fig. 6.8(b)) which is very similar to
that found for MnOx [180]. This feature becomes even more pronounced when
the ion fluence is increased. We conclude therefore that the Al capping layer
was partially sputtered off during the ion irradiation despite a very small mass
of He+ ions. Consequently, the formation of MnO became possible. Moreover,
an additional peak appears in the Co XAS spectra (Fig. 6.8(a)) at approximately 6 eV above the L2,3 lines. Formation of CoO can be ruled out as origin
of this feature since otherwise a multiplet structure similar to that of Mn should
also be present [181, 182]. A tentative explanation for this additional peak at
6 eV above the L2,3 edges would be the presence of Co3+ (3d7 ) in combination

116

Experimental results II - Modification of Heusler films by ion irradiation


1.0

(a)

(b)

(m ++m- )d/2

1.5

1H1016
5H1015

0.5

1H10

1.0

15

5H1014
0.5

1H1014
no irrad.
0.0

L3

L2

(m+-m- )d/2

0.03
0.02
0.01
0.00
-0.01

Co 2p Y 3d,TEY
780

790

800

810

Photon energy (eV)

L2

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
-0.2
-0.3
-0.4

0.04

-0.02

L3

0.0

820

Mn 2p Y 3d, TEY
640

650

660

670

680

Photon energy (eV)

Figure 6.8: Surface sensitive XAS and XMCD spectra at (a) Co and (b) Mn L2,3
absorption edges collected from Co2 MnSi films irradiated with 30 keV He+ ions at
different fluences. Vertical line in (a) indicates the Co L3 shoulder sensitive to the
chemical order in the volume of Co2 MnSi films.

with a large crystal field. Therefore the peak might indicate the formation of
a AlCoO3 at the interface.
The Co L3 satellite at 3.8 eV above the main line is still present in the surface sensitive XAS/XMCD spectra. Its intensity, however, provides a different
dependence on the ion fluence compared to the behavior found in the volume
of the films (Fig. 6.7). The intensity is found to continuously decrease with
increasing fluence (not shown here). This result points to a degradation of the
surface quality, which is in agreement with the additional features appearing in
the Co and Mn XAS spectra after the irradiation.
Magnetic moments
The element specific magnetic moments were determined by Prof. H. J. Elmers
and collaborators (University of Mainz) from the XAS/XMCD spectra discussed in the previous Subsection. For this purpose a sum rule analysis was

6.2 He+ irradiation of Co2 MnSi


Surface

Volume

0.5

m spin (mB )

117

0.4
0.3
0.2
0.1

Co
Mn

(a)

Co
Mn

(b)

m orb /m spin

0.0

(c)

0.10

(d)

Co
Mn

Co
Mn

0.05
0.00
0

10

14

10

15

10
2

Fluence (ions/cm )

16

10

14

10

15

10

16

Fluence (ions/cm )

Figure 6.9: Element-specific spin magnetic moments per 3d-hole and ratios of orbital
and spin magnetic moments in the volume and at the surface of the investigated
films. Full(open) symbols represent Co(Mn) data.

employed [183], which yields the magnetic moments per 3d-hole.


The determined spin magnetic moments for the volume and the surface of
the irradiated Co2 MnSi films are shown in Fig. 6.9(a) and (b), respectively, in
dependence of the applied fluence. As a result of ion irradiation we observe
an increase of the spin magnetic moments on both Co and Mn atoms up to
the fluence of 11015 ions/cm2 and a subsequent decrease for higher fluences
(Fig. 6.9(a)). At the surface, both Co and Mn atoms possess slightly lower
magnetic moments (Fig. 6.9(b)) compared to the volume values and exhibit a
tendency to be reduced by the He+ irradiation. This might be due to a lower
degree of order at the surface and formation of MnO as discussed above.
The ratio of orbital and spin magnetic moments is shown in Fig. 6.9(c) and
(d). No clear dependence between orb /spin and ion fluence is observed except
for Co in the volume of the Co2 MnSi samples (Fig. 6.9(c)) where a peak at a
fluence of 11015 ions/cm2 appears. This points to a reduction of symmetry in
the Co environment. As such the quenching of the orbital moment at the Co
site is partially removed after the ion bombardment in this particular range of
applied fluences.
Assuming the number of 3d-holes for Co and Mn to be 2.24 and 4.52 per
atom [177], respectively, and taking into account an additional correction factor of 1.5 for the number of Mn 3d-holes, which is necessary due to the mixing
of two Mn j levels [184], we determined the saturation magnetization of the
investigated Co2 MnSi films from the above discussed data. In addition, the
saturation magnetization was determined from magnetization curves measured
by a SQUID magnetometer. Corresponding SQUID measurements were per-

118

Experimental results II - Modification of Heusler films by ion irradiation


6

m (mB / f.u.)

m sum (m B /f.u.)

5
4
3
2
1
0
0

volume
surface
SQUID

(a)
1014

1015

1016
2

Fluence (ions/cm )

5
4
3
2
1

(b)

0
-1
-2
-3
-4
-5
-20

no irrad.
1H1014
14
5H10
15
1H10
15
5H10
1H1016
4H1016

-15

-10

-5

10

15

20

H (kOe)

Figure 6.10: (a) Saturation magnetization of Co2 MnSi films irradiated with 30 keV
He+ ions at different fluences, as determined from XAS/XMCD and SQUID measurements. (b) RT magnetization curves measured by SQUID magnetometer for Co2 MnSi
films irradiated with 30 keV He+ ions at different fluences.

formed by H. Schneider at the University of Mainz. The obtained values are


presented in Fig. 6.10(a) in dependence of the applied fluence along with the
measured magnetization curves (Fig. 6.10(b)).
As can be seen in Fig. 6.10(a), the bulk saturation magnetization obtained
from XMCD measurements are systematically larger compared to the values
determined by SQUID. However, both experimental techniques show that the
saturation magnetization prior to ion irradiation is smaller than 5 B /f.u. expected for L21 -ordered Co2 MnSi compound from theoretical calculations. This
indicates presence of disorder in the as deposited Co2 MnSi sample, which is in
accord with results of the XRD characterization discussed in the beginning of
this Section and Sect. 6.1.1. We recall that the fabricated film was found to
have a B2 structure with an admixture of A2-type disorder. Irradiation with
30 keV He+ ions up to a fluence of 11015 ions/cm2 leads to a slight increase
of the volume saturation magnetization, as is visible from data XAS/XMCD
data shown in Fig. 6.10(a). The saturation magnetization values obtained from
SQUID measurements provide a similar dependence on the applied fluence except for the data point at 11014 ions/cm2 . It is more than 1 B /f.u. smaller
compared to the nonirradiated film. Since we do not observe any deviations of
this sample in the characterization by other techniques, this noticeable drop of
its saturation magnetization is in our opinion not related to the ion irradiation.
Most likely, it is a result of an unintentional sample damage during mounting
or transportation.
The saturation magnetization at the surface of the films is systematically

6.2 He+ irradiation of Co2 MnSi

119

smaller with respect to the bulk values determined from XAS/XMCD measurements. This behavior is not farther surprising in view of the discussion in
Sect. 3.2.2. A lower coordination number of magnetic Co and Mn atoms at the
surface results in a modification of the electronic structure which in turn gives
rise to lower saturation magnetization. The irradiation with 30 keV He+ ions
seems to decrease the surface magnetization (Fig. 6.10(a)).
HAXPES investigations

Relative intensity, I / I max

In addition to the XRD and XAS/XMCD investigations described above, we


also looked at the electronic structure of the irradiated Co2 MnSi films by means
of HAXPES. These measurements were carried out by Dr. A. Gloskovskii and
collaborators (University of Mainz). A HAXPES valence band spectrum measured for the film irradiated with He+ ions at 51014 ions/cm2 is shown in
Fig. 6.11 along with the corresponding spectra of the nonirradiated film and
a bulk reference sample. As is clearly visible from this figure, the valence
band spectrum of the nonirradiated film has a wide maximum in the energy
range from about 7 to 0 eV without distinct features. The pronounced peak
of the d states at about 1.3 eV below the Fermi level, which is well resolved
in the valence spectrum of the bulk sample, is largely smeared out, pointing

Co t2g
Mn eg

1.0

0.5

0.0

nonirradiated
14
+
2
5H10 He ions/cm
Co2MnSi bulk
8

Binding energy (eV)


Figure 6.11: HAXPES (h = 7.94 keV) spectra of Co2 MnSi films irradiated with
30 keV He+ ions in comparison to a nonirradiated sample and a cleaved Co2 MnSi bulk
sample. The enhanced intensity of Co and Mn d states at about 1.3 eV, characteristic
for bulk Co2 MnSi, is marked by an arrow.

120

Experimental results II - Modification of Heusler films by ion irradiation

to the presence of disorder in the sample. After irradiation, the valence band
spectrum of the thin film resembles much more closely that of the bulk material providing clear evidence of the improvement electronic structure in the
Co2 MnSi film after the irradiation.
MOKE investigations
The characterization of magnetic properties of the investigated Co2 MnSi films
was completed by MOKE studies in which hysteresis curves were recorded at
different in-plane sample orientations . Depending on the recorded loops
exhibit a more or less pronounced asymmetric shape similar to that of the loops
presented in Sect. 5.3.1, which arises from a strong QMOKE contribution.
The evolution of the LMOKE and QMOKE loops with the ion fluence is
exemplarily shown in Fig. 6.12 for the case of =42 . These loops were obtained
from the measured Kerr rotation loops by symmetrization and antisymmetrization procedure described in Sect. 4.2.1. In the lower range of applied fluences,
the LMOKE loops (Fig. 6.12(a)) are not significantly changed by the He+ ion
irradiation. At larger fluences however, their shape becomes more slanted,
which indicates that higher magnetic fields are necessary to saturate the sample. Moreover, the LMOKE signal is enhanced with respect to the nonirradiated
film and the films exposed to lower ion fluences. This indicates a modification of magneto-optical properties by the He+ irradiation. The QMOKE loops
(Fig. 6.12(b)) provide a similar behavior to that found for the LMOKE loops,

20

(a)

a=42

32

(b)

a=42

10
5
0
no irrad.
1H1014
5H1014
1H1015
5H1015
16
1H10
4H1016

-5
-10
-15

QMOKE (arb. units)

LMOKE (mdeg)

15

28

4H10

16

1H10

16

5H1015

24

1H10

15

5H10

14

20
16

1H1014

12

no irrad.

-20
-800

-400

H (Oe)

400

800

-800

-400

400

800

H (Oe)

Figure 6.12: LMOKE (a) and QMOKE (b) loops obtained after symmetrization and
antisymmetrization of hysteresis curves measured at an in-plane sample orientation
=42 for Co2 MnSi films irradiated with He+ ions at different fluences at RT.

6.2 He+ irradiation of Co2 MnSi

121
90

90
120

60

60

120

180

12
18

180

12
18

330

210

12

HC (Oe)

HC (Oe)

HC (Oe)

18

12
6

18

30

150

30

150

240

240

270

14

(a) no irrad.

a
180

12
330

210

24

30

300

[110]CMS
30

150

12

18

330

210

24

24

60

24

24
18

[100]CMS

90

120

30

24

300
270

(b) 5H10 ions/cm

240

16

300
270

(c) 1H10 ions/cm

Figure 6.13: Dependence of the coercive field HC on the in-plane sample orientation
for the nonirradiated Co2 MnSi film (a) and after irradiation with He+ ions at (a)
51014 ions/cm2 and (b) 11016 ions/cm2 .

i.e. significant changes start to appear in the higher range of applied fluences.
While in the range between 11014 ions/cm2 and 11015 ions/cm2 the height of
the peaks is comparable to that of the nonirradiated sample, it is strongly reduced beyond 11015 ions/cm2 . Considering the results presented in Sect. 5.5,
the observed reduction of the QMOKE signal at higher fluences is a strong hint
for the deterioration of the crystal structure of the films.
The LMOKE loops were used to determine the values of the coercive field
HC and its dependence on the applied ion fluence. Polar plots of HC versus
sample orientation , which are shown in Fig. 6.13, reveal a four-fold magnetic
anisotropy of the as prepared Co2 MnSi sample. The four-fold anisotropy is
preserved after the ion irradiation. However, the variation of coercive field
with appears to be more pronounced for the sample irradiated with He+ ions
30
28

a=0

HC (Oe)

26
24
22
20
18

a=42

16
14

10

14

10

15

10

16

Fluence (ions/cm )

Figure 6.14: Variation of the coercive field HC with the applied ion fluence for inplane sample orientations = 0 and 42 .

122

Experimental results II - Modification of Heusler films by ion irradiation

at 11016 ions/cm2 . According to the results discussed in Chapt. 5, this finding


suggests the increase of anisotropy at higher fluences.
Figure 6.14 additionally demonstrates the dependence of HC on the applied
fluence for two in-plane sample orientations = 0 and 42 . In this figure, an
increase of HC values is clearly visible after the irradiation with He+ ions at fluences up to 51014 ions/cm2 with respect to the nonirradiated sample. This increase is observed for both considered in-plane sample orientations. For fluences
above 51014 ions/cm2 , HC exhibits a decreasing trend (until 51015 ions/cm2 ),
which is followed again by a notable increase at 11016 ions/cm2 . This dependence correlates very well with that found for orbital magnetic moments of Co
(Fig. 6.9(c)).

6.2.2

RT irradiation with 130 keV He+

XRD investigations
The results of structural characterization of Co2 MnSi films irradiated with
130 keV He+ at RT are summarized in Fig. 6.15. Figure 6.15(a) shows 2 scans recorded from the films exposed to different ion fluences. Similar
to the RT irradiation with 30 keV He+ discussed in Sect. 6.2.1, all of the
diffraction patterns exhibit the B2 superstructure reflection, i.e. (200) peak,
whose intensity is not significantly changed upon irradiation. The shape of
the MgO substrate peak is again noticeably modified at the highest applied
fluence of 41016 ions/cm2 . The observed modification of the MgO peak is
most likely a result of an increasing damage of the MgO substrate, as also
discussed for 30 keV He+ ion irradiation (see Sect. 6.2.1). In the (111) pole
figures, which were recorded for all of the irradiated films, no evidence of the
(111) equivalent reflections could be found. Therefore it can be concluded that
no transition from the B2 to the L21 structure was achieved in the investigated
Co2 MnSi films by the RT irradiation with 130 keV He+ ions.
The integrated intensity of the (200) superstructure peak normalized by the
intensity of the (400) reflection is presented in Fig. 6.15(b) as a function of the
applied fluence. No clear trend is observed in this dependency. However, a
remarkable increase of the (200)/(400) ratio with respect to the nonirradiated
sample is found for the fluence 11014 ions/cm2 . As already mentioned in
the previous Section, the calculated (200)/(400) ratio is expected to be about
30 % for Co2 MnSi with a perfect B2 structure. Hence, the observed increase
might be an indication of an improvement of the B2 order induced by the ion
bombardment.

6.2 He+ irradiation of Co2 MnSi

123

Figure 6.15: (a) -2-scans of MgO/Cr(40 nm)/Co2 MnSi (30 nm)/Al(1.3 nm) films
irradiated with 130 keV He+ ions at different fluences. (b) Ratio of the (200) and
(400) integrated XRD intensities as a function of the applied ion fluence.

Magnetic moments
In addition to the above discussed XRD investigation, we also determined the
saturation magnetization of the Co2 MnSi films irradiated with 130 keV He+ ions
at RT. For this purpose SQUID measurements were performed by H. Schneider
at the University of Mainz. The magnetization curves acquired at RT for the
films exposed to different ion fluences are presented in Fig. 6.16(a). The saturation magnetization determined from these curves is shown in Fig. 6.16(b) in
dependence of the applied fluence. Obviously, no changes of the saturation mag-

Figure 6.16: (a) Magnetic hysteresis curves of Co2 MnSi films irradiated at RT with
130 keV He+ ions. The curves were measured at RT employing a SQUID magnetometer. (b) Dependence of saturation magnetization on the applied fluence.

124

Experimental results II - Modification of Heusler films by ion irradiation

netization appear up to a fluence of 51014 ions/cm2 . The values remain identical with that of the nonirradiated film. At fluences above 51014 ions/cm2 ,
a decrease of the saturation magnetization is found. This decrease at higher
fluences is a strong indication of increasing disorder in the lattice. Note that for
a perfect L21 structure the saturation magnetization is expected to be 5 B /f.u.
from theoretical calculations. Interesting to mention is also the fact that the dependence of the saturation magnetization on the applied fluence does not agree
with the fluctuating behavior found for the (200)/(400) ratio (Fig. 6.15(b)).

Irradiation with 30 keV He+ above RT

6.2.3

BLS investigations
Figure 6.17 summarizes BLS spectra measured for the Co2 MnSi films irradiated
with 30 keV He+ ions at (a) 150 and (b) 250 C. The spectra were measured
in an external field H=1.5 kOe and at an incidence angle =15 of the probing laser beam. For both considered temperatures and for all applied fluences,
recorded spectra exhibit two magnonic peaks which originate from the excitation of the DE and PSSW1 modes. In the lower range of applied fluences, He+
irradiation does not significantly modify the spectral position of the peaks. At
higher fluences (& 11015 ions/cm2 ), however, we observe a shift of both the
DE and PSSW1 peaks to lower frequencies. This behavior is found for both
temperatures at which the irradiation was carried out. In addition to the frequency shift, we also observe a decrease of the separation distance between the

(a)

H=1.5 kOe
j=15

PSSW1

1H10

150

T=150C

DE

PSSW1

16

5 H 10

15

1 H 10

15

100 3 H 10 14
1 H 10 14

50

3 H 10

400

(b)

DE

BLS intensity (a.u.)

BLS intensity (a.u.)

200

13

H=1.5 kOe
j=15

DE

T=250C

DE
PSSW1

PSSW1

1H1016

300

5H10

1H10

200

3H10

3H10

-30

-20

20

Frequency (GHz)

30

40

14

13

no irrad.

0
-40

15

14

1H10

100

no irrad.

15

-40

-30

-20

20

30

40

Frequency (GHz)

Figure 6.17: BLS spectra of Co2 MnSi films irradiated with 30 keV He+ ions at
different fluences. (a) shows BLS spectra for samples irradiated at 150 C, (b) for
those irradiated at 250 C.

35

125

Stokes
anti-Stokes

PSSW1

BLS frequency (GHz)

BLS frequency (GHz)

6.2 He+ irradiation of Co2 MnSi

30
25
20

DE

15

(a)

Tirr=150C

10

14

10

15

10

16

35

Stokes
anti-Stokes

PSSW1

30
25
20

DE

15

(b)

Tirr=250C

1014

1015

1016

Fluence (ions/cm )

Fluence (ions/cm )

Figure 6.18: BLS frequencies as a function of the applied fluence for processing
temperatures of (a) 150 and (b) 250 C. The data points were determined from BLS
spectra shown in Fig. 6.18.

two spin-wave modes with increasing ion fluence. These features are shown
more clearly in Fig. 6.18(a) and (b), where the experimental BLS frequencies
are plotted as a function of the applied fluence.
The experimental BLS frequencies shown in Fig. 6.17 were fitted using a
theoretical model described in Ref. [44]. The saturation magnetization MS
and the exchange constant A obtained from these simulations are presented in
Fig. 6.19 along with the calculated exchange stiffness D. Both the saturation
magnetization (Fig. 6.19(a)) and exchange constant (Fig. 6.19(b)) decrease after
2.8
1040

2.6

1020

2.4

960
940
920

560

2.2

980

600

D (meV D )

1000

A (erg/cm)

MS (emu/cm )

1060

2.0
1.8
1.6

520

480

440

900
880
860

1.4

(a)
0

10

14

10

15

10
2

Fluence (ions/cm )

16

1.2

(b)
0

400

10

14

10

15

10
2

Fluence (ions/cm )

16

(c)
0

10

14

10

15

10

16

Fluence (ions/cm )

Figure 6.19: (a) Saturation magnetization, (b) exchange constant and (c) exchange
stiffness as a function of the applied fluence for the Co2 MnSi films irradiated with
130 keV He+ ions at 250 C. Presented data points were obtained from the fits of
BLS frequencies shown in Fig. 6.18.

126

Experimental results II - Modification of Heusler films by ion irradiation

the irradiation. As such also the exchange stiffness (Fig. 6.19(c)) exhibits a
decrease upon the increasing ion fluence, which is particularly pronounced for
fluences above 11014 ions/cm2 . Taking into account the dependence of D on
the L21 ordering degree discussed in Chapt. 5, the dependence of D presented
in Fig. 6.19(c) clearly suggests that no transition from the initial B2 structure
to the L21 structure was evoked in the Co2 MnSi by the irradiation with 30 keV
He+ ions performed at 250 C. Since the BLS frequencies determined for the
films irradiated with 30 keV He+ at 150 C provide a similar dependence on the
applied fluence (Fig. 6.18(a)), the same conclusion can also be made for these
samples.
MOKE investigations
The influence of 30 keV He+ irradiation carried out at 150 C on the LMOKE
and QMOKE loops is shown in Fig. 6.20. The loops are illustrated for the
in-plane sample orientation =42 and were obtained from the measured Kerr
rotation loops by symmetrization and antisymmetrization procedure described
in Sect. 4.2.1. Similar to 30 keV He+ irradiation at RT, the LMOKE loops
exhibit a more slanted shape when the fluence is increased. This change in
shape is accompanied with an enhancement of the LMOKE signal (Fig. 6.20(a)),
which, in contrast to RT irradiation, becomes notable already in the lower
range of applied fluences. The QMOKE loops (Fig. 6.20(b)) remain nearly
unchanged up to 11015 ions/cm2 . At higher fluences, however, the QMOKE
signal is remarkably reduced, which corresponds with the behavior found for
RT irradiation. Samples irradiated at 250 C provide a similar dependence of
22

16

(a)

a=42

20

QMOKE (arb. units)

LMOKE (mdeg)

12
8
4
0

no irrad.
3H1013
1H1014
14
3H10
1H1015
5H1015
1H1016

-4
-8
-12
-16

-800

-400

H (Oe)

400

(b)

a=42
1H10

16

5H10

15

16
14

1H1015

12

3H10

10

1H1014

3H10

18

14

13

no irrad.

6
4

800

-800

-400

400

800

H (Oe)

Figure 6.20: LMOKE (a) and QMOKE (b) loops obtained after symmetrization and
antisymmetrization of hysteresis curves measured at an in-plane sample orientation
=42 for Co2 MnSi films irradiated with He+ ions at different fluences at 150 C.

6.2 He+ irradiation of Co2 MnSi

127

90

16

90
60

16

30

150

8
180

8
12
16

HC (Oe)

HC (Oe)

12

90

120

20

60

30

150

12
8
180

8
12

330

210

330

210

16

20

240

20

300

240

300

270

13

(a) 3H10 ions/cm

HC (Oe)

120

20

10
8
6
4
2
0
2
4
6
8
10

120

[110]CMS
30

150

a
180

330

210

240

270

14

(b) 3H10 ions/cm

[100]CMS

60

300
270

16

(c) 1H10 ions/cm

Figure 6.21: Dependence of the coercive field HC on the in-plane sample orientation
for the Co2 MnSi films irradiated with 30 keV He+ ions at (a) 31013 ions/cm2 , (b)
31014 ions/cm2 and (c) 11016 ions/cm2 . The irradiation was carried out at 250 C.

the LMOKE and QMOKE loops on the applied fluence and are therefore not
presented here.
Plotting coercivity HC versus the in-plane sample orientation reveals that
the four-fold symmetry of the plots is preserved in the whole range of applied
fluences for both processing temperatures. However, the variation of HC with
becomes stronger at higher fluences as is visible from the polar plots of films
irradiated at 250 C which are presented in Fig. 6.21. This suggests an increase
of the magneto-crystalline anisotropy in the Co2 MnSi films which might be a
result of a reduced order in the crystal lattice. On the other hand, however, we
observe a decrease of coercivity with increasing fluence as is clearly shown in
Fig. 6.22 where HC values are plotted as a function of applied fluence for both
24
22

a=0

20

HC (Oe)

18
16
14

a=42

12
10
8

150C
250C

6
0

10

14

10

15

10

16

Fluence (ions/cm )

Figure 6.22: Dependence of coercivity HC on applied fluence for the Co2 MnSi films
irradiated with 30 keV He+ ions at 150 (full symbols) and 250 C (open symbols).
The data are shown for the in-plane sample orientations =0 and 42 .

128

Experimental results II - Modification of Heusler films by ion irradiation

processing temperatures at =0 and 42 . The drop of HC is most pronounced


in the upper range of applied fluences and begins at a lower fluence for higher
processing temperature. The observed behavior of HC is somewhat contradictory to the indicated increase of the anisotropy. An opposite trend would be
expected also because the damage impact of the irradiation is larger at higher
fluences, i.e. more defects are introduced into the crystal lattice of the target
material.

6.2.4

Discussion

The experimental results described above show first of all that the crystallinity
and epitaxy of the Co2 MnSi films remains largely intact after the bombardment
with 30 keV as well as 130 keV He+ ions. At fluences above 11015 ions/cm2 ,
however, the damage impact on the crystal lattice of the Co2 MnSi layer appears to become stronger with respect to the fluences below 11015 ions/cm2 .
This is particularly suggested by the decrease of the saturation magnetization
and larger saturation fields observed for fluences above 11015 ions/cm2 . A
reduction of the QMOKE signal observed in this range of fluences is also a
strong hint of an increased defect density in the Co2 MnSi layer. Moreover, the
crystal integrity of the MgO substrate appears to be deteriorated by the ion
irradiation in the fluence range above 11015 ions/cm2 . These finding are in
agreement with the results of SRIM simulations, which showed that (1) low
defect densities are introduced into the Co2 MnSi lattice by He+ irradiation at
fluences in the range between 11014 and 11015 ions/cm2 and (2) most of the
He+ ions are stopped in the MgO substrate.
For RT irradiation with 30 keV He+ at 11015 and 51015 ions/cm2 , some
improvement of the structural properties in the volume of the Co2 MnSi layer is
evidenced by the XRD, XAS and HAXPES investigations. However, no (111)
reflections are observed with XRD, which shows that L21 long-range order is
not present in the investigated films after the irradiation. Therefore, it can be
concluded that either (1) the B2 structure has been improved by reducing the
A2 disorder present in the fabricated Co2 MnSi film or (2) regions with L21 order
have been locally nucleated in the films after the irradiation. A combination of
both would also be possible. The improvement of B2 order could be an effect
of successive pairwise Co-Mn and Co-Si exchanges due to mobile vacancies
introduced by the irradiation [13]. The ordering probably proceeds from the
preexisting B2 ordered regions and is driven by the energy gain which results
from the reduction of relatively high-energetic Co-Mn and Co-Si swaps [10]
present in the A2 disordered regions. A local appearance of L21 ordered nuclei
would also be a result of successive exchanges induced by mobile vacancies.

6.3 Ga+ irradiation of Co2 FeSi

129

As discussed by Bernas et al. [13] for FePt(Pd) alloy, a transition from a


disordered to an ordered structure requires both ordered nuclei from which
to grow as well as high enough mobility of atoms. Therefore, if the 30 keV
He+ irradiation carried out at RT indeed resulted in the nucleation of local
regions with L21 order, performing the irradiation in combination with a mild
annealing could be expected to result in the formation of the L21 structure on
the long-range scale. However, in our BLS investigations carried out for the
Co2 MnSi films irradiated with 30 keV He+ ions at 150 C and 250 C no increase
of the exchange stiffness D is found, which should occur if the structure of the
films is changed from B2 to L21 (see Chapt. 5). From this result we conclude
that no transition to the L21 structure has been induced by the 30 keV He+
irradiation carried out at elevated temperatures.
Irradiation of the Co2 MnSi films with 130 keV He+ ions does not evoke a
transition to the L21 structure as well. Results of XRD investigations, however,
indicate that an ion-induced improvement of the B2 structure or appearance
of L21 ordered regions on the local scale has most likely also occurred for this
energy at a fluence of 11014 ions/cm2 .
Finally, the surface of the investigated Co2 MnSi films exhibits degradation
upon the irradiation with He+ ions. As suggested by XAS spectra measured
for the samples with Al cap, this is due to oxidation of the Co2 MnSi layer as
well as due to a possible formation of the AlCoO3 at the interface between the
Al cap and the Co2 MnSi layer.

6.3

Ga+ irradiation of Co2FeSi

Irradiation of Co2 FeSi with Ga+ ions was performed at RT using a FIB device located at the IFOS, Kaiserslautern. In contrast to the He+ irradiation,
where the complete sample surface was exposed to the beam, here the ion beam
was scanned over the areas, designated for the irradiation. After irradiation,
the sample was investigated by MOKE magnetometry. The results of these
investigations will be presented in the following.

6.3.1

MOKE investigations

Figure 6.23 shows MOKE hysteresis loops recorded from sample areas irradiated with different ion fluences. The loops are presented for in-plane sample
orientations = 45 , with being the angle between the [100] direction of
the Co2 FeSi film and the plane of light incidence. Ga+ irradiation has several effects on the hysteresis curves. On the one hand, both the coercivity HC

130

Experimental results II - Modification of Heusler films by ion irradiation

MOKE (mdeg)

60
40

a =22.5o
a =-22.5o

20

0 x 1015

0.3 x 1015

-20

0.6 x 1015

-40

1 x 1015

-60
-80

3 x 1015

30 x 1015

10 x 1015

60 x 1015

-100 90 x 1015
-120
-1500 -1000 -500

500

1000 1500

H (Oe)
Figure 6.23: MOKE hysteresis loops recorded from different areas of the Co2 FeSi
sample that were irradiated with different fluences of 30 keV Ga+ ions.

and the LMOKE signal at saturation decrease with increasing ion fluence. At
the highest applied fluence of 91016 ions/cm2 , the ferromagnetic property of
Co2 FeSi seems to be completely lost. On the other hand, the initially notable
asymmetry of the loops is drastically reduced even at the lowest applied fluence. Since this asymmetry originates from the QMOKE contribution to the
Kerr rotation, the reduction of the asymmetry corresponds to the reduction
of the QMOKE contribution. The fact that the loop asymmetry is reverted
when changes sign reflects the four-fold symmetry of the QMOKE in a cubic
crystal.
30
350

nonirradiated sample
25

MOKE (mdeg)

300

HC (Oe)

250
200

nonirradiated sample

150
100
50
0

LMOKE

15
10
5

(a)
0

20

0
1014

1015

1016
2

Fluence (ions/cm )

1017

QMOKE

(b)
0

1014

1015

1016

1017

Fluence (ions/cm )

Figure 6.24: Dependence of (a) HC and (b) the amplitudes of the LMOKE and
QMOKE of the Co2 FeSi sample on the applied ion fluence.

6.3 Ga+ irradiation of Co2 FeSi

131

The dependence of HC as well as the LMOKE and QMOKE amplitudes


on the applied fluence are shown more clearly in Fig. 6.24. For the determination of the QMOKE amplitude the 8-directional method was employed. As
is visible from Fig. 6.24, HC (a) and LMOKE amplitude (b) exhibit a similar dependence on the applied fluence. They decrease slowly up to a fluence
of 61015 ions/cm2 . Subsequently, a faster decrease is observed which at
91016 ions/cm2 finally results in a vanishing HC as well as LMOKE amplitude.
The QMOKE amplitude shown in Fig. 6.24(b) provides a different behavior.
It decreases rapidly from its initial value of 15 mdeg to a value of 3 mdeg for
the smallest applied fluence. Higher fluences lead to a further decrease of the
QMOKE amplitude until it vanishes at a fluence of 11016 ions/cm2 .

6.3.2

Discussion

The behavior of LMOKE and HC observed in the lower range of applied fluences, where only a small decrease of both parameters is recorded, suggests
that the defect density introduced into the Co2 FeSi lattice is quite low. This is
in contradiction with results of SRIM simulations, according to which a 100 %
damage is expected already in the fluence range of 1015 ions/cm2 . The reason
for this discrepancy is most likely a purely ballistic model underlying the SRIM
software. A possible implementation of Ga+ ions into the Co2 FeSi lattice is
thus not taken into account in the simulations of the irradiation process. The
implementation of Ga+ appears probable because of the existence of the related
Co2 FeGa compound.
In this context it is interesting to refer to the ab initio studies of the substitutional series Co2 FeSi1x Gax carried out by Gercsi et al. [11]. Assuming the L21
structure of the investigated compounds, they have shown that half-metallicity
is preserved in the compositional range x 0.5. The calculated DOS looks very
similar to that of Co2 FeSi. Moreover, the magnetic moments are maximally
reduced by 8 % in the considered substitutional series. We can now make a very
rough estimation of x for our samples assuming that all impinging Ga+ ions are
stopped in the Co2 FeSi layer and replace a Si atom. With an atomic density of
Co2 FeSi %CFS =8.921022 atoms/cm3 , the irradiated volume 11107 cm3 and
the fact that Si atoms make up 25 % of all atoms in the irradiated volume, we
obtain for the fluence of 11015 ions/cm2 an x value of 0.04. This value is well
inside the compositional range given above. An increase of the applied fluence
by one order of magnitude results in x = 0.4, which is still within this range.
At fluences beyond 11016 ions/cm2 , however, the upper boundary is clearly
exceeded, so that larger changes in the electronic and magnetic properties of
Co2 FeSi can be expected.

132

Experimental results II - Modification of Heusler films by ion irradiation

Although having a very approximative character, the preceding considerations could explain the initially low decrease of LMOKE amplitude and HC
values followed by a notable drop above 61015 ions/cm2 . The breakdown
of the ferromagnetic order manifesting itself in zero MOKE and coercivity is
probably a result of an increasing damage impact of Ga+ irradiation at higher
flunces.
The immediate decrease of the QMOKE amplitude is likely a result of a
much stronger sensitivity of QMOKE to both the crystallographic order and
chemical composition (Sect. 5.5).

Chapter 7
Summary and outlook
The work reported in this thesis addresses a systematic investigation of the exchange interaction and magnetic anisotropies in thin films of Co2 -based Heusler
compounds with the general composition Co2 YZ. For this purpose, the exchange stiffness D and anisotropy constants K were determined for various
compounds by means of Brillouin light scattering (BLS) spectroscopy. The
information about the type of magnetic anisotropy inherent to the investigated Heusler films was additionally obtained from magneto-optical Kerr effect (MOKE) magnetometry measurements. The particular achievement of
our investigations was to reveal the dependence of D and K on the chemical
composition and atomic ordering of the Heusler films. The latter point is of
major importance since the half-metallicity is believed to be strongly affected
by disorder in the crystal lattice.
The influence of the chemical composition on exchange and anisotropies
was studied for compounds wherein Y=Mn,Fe and Z=Al,Si, as well the quaternary compound Co2 Cr0.6 Fe0.4 Al. The influence of the structural order was
investigated for the Co2 MnSi films exhibiting a varying degree of the L21 order.
The gradual variation of the L21 order in the Co2 MnSi films was achieved by
annealing the samples at different temperatures, after their deposition.
Within the investigated Co2 -based Heusler compounds, the largest exchange
stiffness D is found for the L21 -ordered Co2 FeSi, with a value of 71520 meV
A2 .
This value is extraordinarily large, and is (to the best of our knowledge) only
surpassed by the intermetallic compound Fe53 Co47 . Moreover, we observe a
linear increase of the exchange stiffness D upon increasing the number of valence
electrons NV within the unit cell induced by the change of composition (e.g.
replacement of Mn by Fe at the Y site in the Co2 YZ unit cell). This behavior
is most likely a result of an increased electron density in the t2g states of the
transition metal located at the Y site, and a larger exchange integral J resulting

134

Summary and outlook

from it. The linear increase of D observed in this thesis correlates with the
Slater-Pauling rule found for the Co2 -based Heusler compounds and the linear
dependence of their Curie temperatures on NV . Furthermore, our studies show
that a larger ordering degree in the investigated Heusler compounds is related
to a stronger exchange interaction, which is evidenced by larger D values found
in the L21 -ordered films, compared to those with a less ordered B2 structure.
Most of the investigated Co2 -based Heusler films are found to exhibit a
four-fold magneto-crystalline anisotropy with h110i easy in-plane directions.
The four-fold symmetry of the magnetic anisotropy reflects the cubic symmetry
of the Heusler crystal lattice. For the Co2 FeAl film, we also observe a weak
uniaxial contribution to the magnetic anisotropy, which might be an effect of
strain originating from the lattice mismatch between the Heusler compound and
the substrate. The cubic volume anisotropy constants K1 determined for the
investigated Co2 -based Heusler compounds are found to be at least one order
of magnitude smaller than those of bulk Fe and Co. Moreover, the K1 values
are found to become smaller with increasing ordering degree in the investigated
films. In several L21 -ordered films, the magneto-crystalline anisotropy is even
found to be negligible (< 10 kerg/cm3 ), which points to a vanishing spin-orbit
coupling in the more ordered Heusler compounds.
In the course of our investigations of magnetic anisotropies inherent to
Heusler films, we found out that the ordering degree of the studied films is
related to the anisotropy of the polar representations of coercivity as a function of the in-plane sample orientation. Hence, MOKE magnetometry can be
routinely used in our group for a first quick check of the state of order in the
Heusler samples. In particular, the L21 -ordered films and those with atomic disorder can be clearly differentiated by this technique. However, this technique
only works for systems exhibiting a four-fold magneto-crystalline anisotropy,
and is not applicable to films with uniaxial easy axes induced by preparation
conditions for example. In this case additional investigations by means of BLS,
and x-ray diffractions must be carried out. The latter is also necessary if a
detailed analysis of the types of disorder is required.
Moreover we find that amongst the investigated Co2 -based Heusler compounds, Co2 MnSi and Co2 FeSi exhibit a considerable asymmetry of Kerr rotation loops, which provides an evidence of a strong contribution which is second
order in the spin-orbit coupling to the MOKE signal (QMOKE). The largest
QMOKE signal is observed in the L21 -ordered Co2 FeSi films, with a record
value of 30 mdeg. Moreover, the amplitude of QMOKE signal appears to increase upon increasing the ordering degree of the films, as was particularly
demonstrated for Co2 MnSi with a gradual variation of the L21 order. The origin of such a strong QMOKE contribution to the MOKE signal in the Heusler
compounds is still to be investigated, and requires a systematic study of the

135
dependence of chemical composition as well as preparation conditions of the
films (e.g. induced strain). For this purpose, a new MOKE setup has been
built up by Dr. J. Hamrle, Dr. S. Trudel, and G. Wolf [185], which allows for
a fully automatized application of the 8-directional method necessary for the
determination of the QMOKE amplitudes. Very recently, first measurements
were performed employing this setup and further experiments are planned.
Given the strong dependence of the exchange interaction and magnetic
anisotropies revealed in the course of this thesis, we also explored the structural modification of Heusler films using ion irradiation. In the irradiation
experiments both light 30 and 130 keV He+ and heavy 30 keV Ga+ ions were
used. Experiments utilizing He+ were carried out for Co2 MnSi films with an
initially predominant B2 order, with and without a simultaneous mild anneal.
A particular goal of these experiments was to investigate the feasibility of an
ion-induced enhancement of order in Heusler compounds, similar to the increase of the L10 long-range order achieved by He+ irradiation for the FePt(Pd)
films [12, 140]. Ga+ irradiation was performed for L21 -ordered Co2 FeSi films at
room temperature only.
For the study of the influence of He+ ion irradiation on the structural,
magnetic and electronic properties of Co2 MnSi films various techniques were
employed including x-ray diffraction, BLS spectroscopy and MOKE magnetometry. In some cases additional investigations by means of superconducting
quantum interference device magnetometry, photoemsission at high energies,
and x-ray absorption and circular magnetic dichroism were also carried out.
The results of these investigations show that the crystallinity and epitaxy of
the Co2 MnSi films remain largely unaffected by the bombardment with He+
ions in the whole range of applied fluences. Moreover, at particular fluences,
we observe an improvement of the magnetic and electronic properties of the
Co2 MnSi layer, whereas the probed electronic properties converge towards the
bulk material after the room temperature irradiation. However, no evidence of
the transition to the L21 phase is found. Furthermore, the surface of the films
bombarded by He+ ions is found to be partially destroyed despite the small
mass of the impinging ions. This finding makes the prospective use of lightion irradiation as a tool for the improvement of structural order in the Heusler
films questionable. Also it is not clear if the observed small modifications of the
film properties toward those of a bulk material can be enhanced, thus possibly
leading to the B2/L21 transformation. Comprehensive studies in a wider range
of ion energies and fluences as well as temperatures during the irradiation are
necessary to clarify this issue.
For Ga+ irradiation experiments on Co2 FeSi films, we find a strong destructive impact of the ion beam on the magnetic properties of Co2 FeSi. For larger
fluences, we even observe a breakdown of the ferromagnetic order. Interestingly,

136

Summary and outlook

the LMOKE is found to be more robust than the QMOKE with respect to the
Ga+ irradiation. The LMOKE remains nearly unchanged in a wide range of
applied fluences, while the QMOKE is found to decrease rapidly upon Ga+
irradiation. This result suggests a much stronger sensitivity of the QMOKE to
both the atomic order and chemical composition. As such, it might be used as
probing technique for the quality of Heusler samples. In comparison to other
experimental techniques commonly used for structural characterization (e.g.
XRD) it requires less measurement time and effort for the analysis of the data.
However, first the origin of the large QMOKE in the Heusler films must be
understood.
Small and even vanishing spin-orbit coupling in Co2 -based Heusler compounds suggested by the results of our investigations is of an outstanding importance for technological applications of these materials. In many advanced spin
electronic devices, such as spin torque transfer magnetic RAM (STT-MRAM)
for example [186], materials with a low Gilbert damping are required. Since
this material parameter depends amongst others on the strength of the spinorbit interaction [187], our results show that Co2 -based Heusler compounds
are good candidates for such applications. A low Gilbert damping is also
a necessary prerequisite for potential materials in digital spin wave logic devices [188190]. To date, such applications were restricted to yttrium iron
garnet (YIG), which is known to exhibit a low damping of spin waves. The results obtained in this thesis for Co2 -based Heusler compounds mark this class
of materials as very attractive for this type of applications.

Appendix A
Calculation of BLS intensity
In the backscattering geometry, the BLS intensity from a spin wave mode n,
excited in a single FM layer of thickness d, can be calculated using the expression [162]
I

(BLS,n)

Z d
2

(n)
(n)

= I0 [L(z)mL (z) + P (z)mP (z)]dz ,

(A.1)

(n)

(n)

where mL (z) and mP (z) are the depth dependent longitudinal and polar
components of the dynamic magnetization of a given spin wave mode, respectively. L(z) and P (z) are complex depth sensitivity functions of the off-diagonal
reflectivity coefficient rsp for longitudinal ML (z) and polar MP (z) static magnetization profiles defined by [162, 163]
Z

rsp (M ) = rsp (M = 0) +

[L(z)ML (z) + P (z)MP (z)]dz .

(A.2)

The dependence of L(z) and P (z) on depth z is the same for both of them.
Assuming that the substrate has a refractive index identical to that of the FM
layer, the analytical terms of L(z) and P (z) can be expressed as
L(z) = L(0) exp [4iNz z/] ,

(A.3)

P (z) = L(z) ,

(A.4)

where the complex coefficient is the ratio of P (0)


p and L(0). Nz is the normalized wave vector in the polar direction Nz = (N (fm) )2 (N (air) )2 sin2 ,
where N (f m) and N (air) are the refractive indices of the FM layer and air, respectively, and is the angle of incidence of the probing light beam. Note that

138

Calculation of BLS intensity

if the refractive index of the substrate is different from that of the FM layer, the
analytical expressions become more complex. However, the basic features (i.e.
continuous decay of the amplitude and continuous shift of the phase) remain
valid.
Using open (anti-pinning) boundary conditions at the interface of the FM
layer, the depth profile of the dynamic magnetization of a PSSW mode n is
given by
(n)
mL (z, ) = m0 cos (nz/d) cos (sw ) ,
(A.5)
(n)

mP (z, ) = m0 cos (nz/d) cos (sw + /2) ,

(A.6)

where sw and are the frequency of the spin-wave mode and time, respectively. The magnetization vector follows an elliptical trajectory, with ellipticity
. Therefore, the mL and mP magnetization components are shifted by /2 in
their time dependence.
Combining equations (A.1), (A.3)-(A.6) and integrating over the thickness
of FM layer and averaging over time, we obtain

1
(BLS,n)
2 L(0)dm0
I
= I0 (1 + || ) 2 2
2
2
n
(A.7)
|1 exp (i)(1)n |2 ,
with being a dimensionless parameter defined as = 4Nz d/. Zero BLS
intensity is expected when the last term in Eq. (A.7) is zero. Thus BLS intensity
becomes zero for odd n when the following conditions are fulfilled:
4 Re (Nz )d/ = 2k + 1 ,

(A.8)

4 Im (Nz )d/ = 0 ,

(A.9)

where k is integer. For even n, BLS intensity is zero when


2 Re (Nz )d/ = k ,

(A.10)

4 Im (Nz )d/ = 0 .

(A.11)

Equations (A.8) and (A.9) are the relevant conditions for the discussion of BLS
intensities in Chapt. 5.

List of publications
[A1] J. Hamrle, S. Blomeier, O. Gaier, B. Hillebrands, R. Schafer, M. Jourdan, Magnetic anisotropies and magnetization reversal of Co2 Cr0.6 Fe0.4 Al
Heusler compound, J. Appl. Phys. 100, 103904 (2006).
[A2] J. Hamrle, S. Blomeier, O. Gaier, B. Hillebrands, H. Schneider, G. Jakob,
B. Reuscher, A. Brodyanski, M. Kopnarski, K. Postava, C. Felser, Ion
beam induced modification of exchange interaction and spin-orbit coupling
of Co2 FeSi Heusler compound, J. Phys. D: Appl. Phys. 40, 1558 (2007).
[A3] J. Hamrle, S. Blomeier, O. Gaier, B. Hillebrands, H. Schneider, G. Jakob,
K. Postava, C. Felser, Huge quadratic magneto-optical Kerr effect in the
Co2 FeSi Heusler compound, J. Phys. D: Appl. Phys. 40, 1563 (2007).
[A4] O. Gaier, J. Hamrle, S. J. Hermsdoerfer, B. Hillebrands, Y. Sakuraba,
M. Oogane, Y. Ando, Influence of the L21 ordering degree on the magnetic
properties in the Co2 MnSi Heusler compound, J. Appl. Phys. 103, 103910
(2008).
[A5] J. Hamrle, O. Gaier, S.-G. Min, B. Hillebrands, Y. Sakuraba, Y. Ando,
Determination of exchange constants of Heusler compounds by Brillouin
light scattering spectroscopy: application to Co2 MnSi, J. Phys. D: Appl.
Phys. 42, 084005 (2009).
[A6] O. Gaier, J. Hamrle, S. Trudel, A. Conca Parra, B. Hillebrands,
E. Arbelo, C. Herbort, M. Jourdan, Brillouin light scattering study of
Co2 Cr0.6 Fe0.4 Al and Co2 FeAl Heusler compounds, J. Phys. D: Appl. Phys.
42, 084004 (2009).
[A7] O. Gaier, J. Hamrle, B. Hillebrands, M. Kallmayer, P. Porsch,
G. Schonhense, H. J. Elmers, J. Fassebnder, A. Gloskovskii, C. A. Jenkins,
C. Felser, E. Ikenaga, Y. Sakuraba, S. Tsunegi, M. Oogane, Y. Ando, Improvement of structural, electronic, and magnetic properties of Co2 MnSi
thin films by He+ irradiation, Appl. Phys. Lett. 94, 152805 (2009).

140

List of publications

[A8] J. Hamrle, O. Gaier, S. Trudel, B. Hillebrands, H. Schneider, G. Jakob,


Exchange stiffness in Co2 -based Heusler compounds, submitted to Phys.
Rev. Lett., arXiv:0904.4194.

Bibliography
[1] G. A. Prinz, Magnetoelectronics, Science 282, 1660 (1998).
[2] R. A. de Groot, F. M. Mueller, P. G. van Engen, K. H. J. Buschow, New
Class of Materials: Half-Metallic Ferromagnets, Phys. Rev. Lett. 50, 2024
(1983).
[3] J. K
ubler, G. H. Fecher, C. Felser, Understanding the trend in the Curie
temperatures of Co2 -based Heusler compounds: Ab initio calculations, Phys.
Rev. B 76, 024414 (2007).
[4] G. H. Fecher, C. Felser, Substituting the main group element in cobalt-iron
based Co2 FeAl1x Six , J. Phys. D: Appl. Phys. 40, 1582 (2007).
[5] P. J. Webster, Magnetic and chemical order in Heusler alloys containing
cobalt and manganese, J. Phys. Chem. Solids 32, 1221 (1971).
[6] S. Wurmehl, G. H. Fecher, H. C. Kandpal, V. Ksenofontov, C. Felser, H. J.
Lin, Investigation of Co2 FeSi: The Heusler compound with highest Curie
temperature and magnetic moment, Appl. Phys. Lett. 88, 032503 (2006).
[7] V. Y. Irkhin, M. I. Katsnelson, Ground state and electron-magnon interaction in an itinerant ferromagnet: half-metallic ferromagnets, J. Phys.:
Condens. Matter 2, 7151 (1990).
[8] P. A. Dowben, R. Skomski, Finite-temperature spin polarization in halfmetallic ferromagnets, J. Appl. Phys. 93, 7948 (2003).
[9] L. Chioncel, M. I. Katsnelson, R. A. de Groot, A. I. Lichtenstein, Nonquasiparticle states in the half-metallic ferromagnet NiMnSb, Phys. Rev. B 68,
144425 (2003).
[10] S. Picozzi, A. Continenza, A. J. Freeman, Role of structural defects on
the half-metallic character of Co2 MnGe and Co2 MnSi Heusler alloys, Phys.
Rev. B 69, 094423 (2004).

142

Bibliography

[11] Z. Gercsi, K. Hono, Ab initio predictions for the effect of disorder and
quarternary alloying on the half-metallic properties of selected Co2 Fe-based
Heusler alloys, J. Phys.: Cond. Matter 19, 326216 (2007).
[12] D. Ravelosona, C. Chappert, V. Mathet, Chemical order induced by ion
irradiation in FePt (001) films, Appl. Phys. Lett. 76, 236 (2000).
[13] H. Bernas, J.-P. Attane, K.-H. Heinig, D. Halley, D. Ravelososn, A. Marty,
P. Auric, C. Chappert, Y. Samson, Ordering Intermetallic Alloys by Ion
Irradiation: A Way to Tailor Magnetic Media, Phys. Rev. Lett. 91, 077203
(2003).
[14] J. Stohr, H. C. Siegmann, Magnetism, From Fundamentals to Nanoscale
Dynamics, vol. 152 of Solid-State Sciences, Springer Verlag, Berlin Heidelberg (2006).
[15] M. Getzlaff, Fundamentals of magnetism, Springer Berlin Heidelberg New
York (2008).
[16] H. Ibach, H. L
uth, Festk
orperphysik, Springer Berlin Heidelberg (1985).
[17] J. C. Slater, Electronic Structure of Alloys, J. Appl. Phys. 8, 385 (1937).
[18] L. Pauling, The Nature of the Interatomic Forces in Metals, Phys. Rev.
54, 899 (1938).
[19] R. M. Bozorth, Atomic Moments of Ferromagnetic Alloys, Phys. Rev. 79,
887 (1950).
[20] N. L. Huang, R. Orbach, Biquadratic superexchange, Phys. Rev. Lett. 12,
275 (1964).
[21] U. Falk, A. Furrer, J. K. Kjems, H. U. G
udel, Biquadratic Exchange in
CsMnx Mg1x Br3 , Phys. Rev. Lett. 52, 1336 (1984).
[22] I. Dzyaloshinsky, A thermodymic theory of weak ferromagnetism of antiferromagnetics, J. Phys. Chem. Solids 4, 241 (1958).
[23] T. Moriya, Anisotropic Superexchange Interaction and Weak Ferromagnetsm, Phys. Rev. 120, 91 (1960).
[24] A. Hubert, R. Schafer, Magnetic domains, Springer Berlin Heidelberg New
York (2009).
[25] R. Kassing (Editor), Bergmann/Sch
afer, Lehrbuch der Experimentalphysik, Band 6, Walter de Gruyter, Berlin (2005).

Bibliography

143

[26] E. C. Stoner, E. P. Wohlfarth, A mechanism of magnetic hysteresis in


heterogeneous alloys, Phil. Trans. Roy. Soc. London A240, 599 (1948).
[27] J. Kerr, On rotation of the plane of polarization by reflection from the pole
of a magnet, Phil. Mag. 3, 321 (1877).
[28] P. Yeh, Optics of anisotropic layered media: A new 44 matrix algebra,
Surf. Sci. 96, 41 (1980).
Visn
[29] S.
ovsk
y, Magneto-optical ellipsometry, Czech. J. Phys. B 36, 625
(1986).
[30] K. Postava, J. Pistora, T. Yamaguchi, P. Hlubina, Polarized light in structures with magnetic ordering, in M. Pluta, M. Szyjer (Editors), Lightmetry
2002: Metrology and Testing Techniques using Light, vol. 5064 of Proc.
SPIE, 182 (2003).
[31] K. Postava, J. Pistora,
Quadratic Magneto-Optic
Trans. Magn. Soc. Jpn. 2,

Visn
S.
ovsk
y, D. Hrabrovsk
y, T. Yamaguchi,
Effects in Reflection from Uniaxial Crystals,
151 (2002).

Visn
[32] S.
ovsk
y, Magneto-optical permittivity tensor in crystals, Czech. J.
Phys. B 36, 1425 (1986).
[33] G. Traeger, L. Wenzel, A. Hubert, Computer Experiments on the Information Depth and the Figure of Merit in Magnetooptics, Phys. Stat. Sol. (a)
131, 201 (1992).
[34] Z. Qiu, S. Bader, Surface magneto-optic Kerr effect (SMOKE), J. Magn.
Magn. Mater. 200, 664 (1999).
Visn
[35] S.
ovsk
y, M. N
yvlt, V. Prosser, R. Lopusnk, R. Urban, J. Ferre,
G. Penissard, D. Renard, R. Krishnan, Polar magneto-optics in simple
ultrathin-magnetic-film structures, Phys. Rev. B 52, 1090 (1995).
[36] J. Hamrle, J. Ferre, J. P. Jamet, V. Repain, G. Baudot, S. Rousset, Vicinal
interface sensitive magneto-optical Kerr effect: Application to Co/Au(322),
Phys. Rev. B 67, 155411 (2003).
Visn
[37] K. Postava, D. Hrabrovsk
y, J. Pistora, A. R. Fert, S.
ovsk
y, T. Yamaguchi, Anisotropy of quadratic magneto-optic effects in reflection, J. Appl.
Phys. 91, 7293 (2002).

144

Bibliography

[38] P. Bruno, Y. Suzuki, C. Chappert, Magneto-optical Kerr effect in a paramagnetic overlayer on a ferromagnetic substrate: A spin-polarized quantum
size effect, Phys. Rev. B 53, 9214 (1996).
[39] R. Kubo, J. Phys. Soc. Jpn. 12, 570 (1957).
[40] N. W. Ashkroft, N. D. Mermin, Solid State Physics, Saunders College,
Philadelphia (1987).
[41] J. Hamrle, Magneto-optical determination of the in-depth magnetization
profile in magnetic multilayers, Dissertation, University Paris XI and
Charles University Prague (2003).
[42] F. Bloch, Z. Phys. 61, 206 (1930).
[43] L. D. Landau, E. M. Lifshitz, On the theory of the dispersion of magnetic
permeability in ferromagnetic bodies, Z. Phys. Sov. Union 8, 153 (1935).
[44] B. Hillebrands, Spin-wave calculations for multilayered structures, Phys.
Rev. B 41, 530 (1990).
[45] B. Hillebrands, Brillouin light scattering from layered magnetic structures,
in M. Cardona, G
untherodt (Editors), Light Scattering in Solids VII, vol. 75
of Topics in Applied Physics, Springer Verlag, Berlin Heidelberg (2000).
[46] S. O. Demokritov, B. Hillebrands, A. N. Slavin, Brillouin light scattering
studies of confined spin waves: linear and nonlinear confinement, Phys.
Reports 348, 441 (2001).
[47] H. LeGall, J. P. Jamet, Theory of the Elastic and Inelastic Scattering of
Light by Magnetic Crystals, Phys. Stat. Sol. (b) 46, 467 (1971).
[48] W. Wettling, M. G. Cottam, J. R. Sandercock, The relation between onemagnon light scattering and the complex magneto-optic effects in YIG, J.
Phys. C: Solid State Phys. 8, 211 (1975).
[49] J. F. Cochran, J. R. Dutcher, Calculation of the intensity of light scattered
from magnons in thin films, J. Magn. Magn. Mater. 73, 299 (1988).
[50] F. Heusler, Verh. Dtsch. Phys. Ges. 5, 219 (1903).
[51] H. H. Potter, The x-ray structure and magnetic properties of single crystals
of Heusler alloy, Proc. Phys. Soc. 41, 135 (1929).
[52] A. J. Bradley, J. W. Rodgers, Crystal structure of the Heusler alloys, Proc.
R. Soc. London Ser. A 144, 340 (1934).

Bibliography

145

[53] G. B. Johnston, E. O. Hall, Studies on the Heusler alloys-I. Cu2 Mn Al and


associated structures, J. Phys. Chem. Solids 29, 193 (1968).
[54] P. J. Webster, Heusler alloys, Contemp. Phys. 10, 559 (1969).
[55] Y. Ishikawa, Y. Noda, Spin waves and magnetic interactions in the ferromagnetic Heusler alloys, AIP Conf. Proc. 24, 145 (1975).
[56] D. C. Price, Indirect magnetic coupling between local moments in metals,
J. Phys. F: Metal Phys. 8, 933 (1978).
[57] M. B. Stearns, Hyperfine field and magnetic behavior of Heusler alloys, J.
Appl. Phys. 50, 2060 (1979).
[58] P. G. Van Engen, K. H. J. Buschow, R. Jongebreur, M. Erman, PtMnSb,
a material with very high magneto-optical Kerr effect, Appl. Phys. Lett. 42,
202 (1983).
[59] T. Taira, T. Ishikawa, N. Itabashi, K. Matsuda, T. Uemura, M. Yamamoto,
Spin-dependent tunnelling characteristics of fully epitaxial magnetic tunnel
junctions with a Heusler alloy Co2 MnGe thin film and a MgO barrier, J.
Phys. D: Appl. Phys. 42, 084015 (2009).
[60] T. Tsunegi, Y. Sakuraba, M. Oogane, K. Takanashi, Y. Ando, Large tunnel
magnetoresistance in magnetic tunnel junctionsusing a Co2 MnSi Heusler
alloy electrode and a MgO barrier, Appl. Phys. Lett. 93, 112506 (2008).
[61] T. Ishikawa, S. Hakamata, K. Matsuda, T. Uemura, M. Yamamoto, Fabrication of fully epitaxial Co2 MnSi/MgO/Co2 MnSi magnetic tunnel junctions, J. Appl. Phys. 103, 07A919 (2008).
[62] N. Tezuka, N. Ikeda, S. Sugimoto, K. Inomata, Giant Tunnel Magnetoresistance at Room Temperature for Junctions using Full-Heusler Co2 FeAl0.5 Si0.5
electrodes, Jpn. J. Appl. Phys. 46, L454 (2007).
[63] Y. Sakuraba, M. Hattori, M. Oogane, Y. Ando, H. Kato, A. Sakuma,
T. Miyazaki, H. Kubota, Giant tunneling magnetoresistance in
Co2 MnSi/Al-O/Co2 MnSi magnetic tunnel junctions, Appl. Phys. Lett. 88,
192508 (2006).
[64] N. Tezuka, N. Ikeda, S. Sugimoto, K. Inomata, 175 % tunnel magnetoresistance at room temperature and high thermal stability using Co2 FeAl0.5 Si0.5
full-Heusler alloy electrodes, Appl. Phys. Lett. 89, 252508 (2006).

146

Bibliography

[65] G. E. Bacon, J. S. Plant, Chemical ordering in Heusler alloys with the


general formula A2 BC or ABC, J. Phys. F: Metal Phys. 1, 524 (1971).
[66] P. Eckerlin, H. Kandler, Alloys and Compounds of d-Elements with Main
Group Elements Part 2 (Landolt-B
ornstein-Group III Condensed Matter vol
6), Hellwege, K.-H. and Hellwege, A. M., Berlin: Springer (1985).
[67] P. J. Webster, K. R. A. Ziebeck, Alloys and Compounds of d-Elements
with Main Group Elements Part 2 (Landolt-B
ornstein-Group III Condensed
Matter vol 19C), Wijn, H. P. J., Berlin: Springer (1988).
[68] H. J. Elmers, S. Wurmehl, G. H. Fecher, G. Jakob, C. Felser,
G. Schonhense, Field dependence of orbital magnetic moments in the
Heusler compounds Co2 FeAl and Co2 Cr0.6 Fe0.4 Al, Appl. Phys. A 79, 557
(2004).
[69] R. Y. Umetsu, K. Kobayashi, A. Fujita, K. Oikawa, R. Kainuma,
K. Ishida, N. Endo, K. Fukamichi, A. Sakuma, Half-metallic properties of
Co2 (Cr1x Fex )Ga Heusler alloys, Phys. Rev. B 72, 214412 (2005).
[70] S. Wurmehl, G. H. Fecher, H. C. Kandpal, V. Ksenofontov, C. Felser, H.-J.
Lin, J. Morais, Geometric, electronic, and magnetic structure of Co2 FeSi:
Curie temperature and magnetic moment measurements and calculations,
Phys. Rev. B 72, 184434 (2005).
[71] K. Inomata, M. Wojcik, E. Jedryka, N. Ikeda, N. Tezuka, Site disorder
in Co2 Fe(Al,Si) Heusler alloys and its influence on junction tunnel magnetoresistance, Phys. Rev. B 77, 214425 (2008).
[72] S. Wurmehl, J. T. Kohlhepp, H. J. M. Swagten, B. Koopmans, M. Wojcik,
B. Balke, C. G. F. Blum, V. Ksenofontov, G. H. Fecher, C. Felser, Probing the random distribution of half-metallic Co2 Mn1x Fex Si Heusler alloys,
Appl. Phys. Lett. 91, 052506 (2007).
[73] S. Wurmehl, J. T. Kohlhepp, H. J. M. Swagten, B. Koopmans, C. G. F.
Blum, V. Ksenofontov, H. Schneider, G. Jakob, G. Reiss, Off-stoichiometry
in Co2 FeSi thin films sputtered from stoichiometric targets revealed by nuclear magnetic resonance, J. Phys. D: Appl. Phys. 42, 084017 (2009).
[74] K. Kobayashi, R. Umetsu, A. Fujita, K. Oikawa, R. Kainuma,
K. Fukamichi, K. Ishida, Magnetic properties and phase stability of halfmetal-type Co2 Cr1x Fex Ga alloys, J. Alloys Compd. 399, 60 (2005).

Bibliography

147

[75] B. Balke, G. H. Fecher, H. C. Kandpal, C. Felser, K. Kobayashi, E. Ikenaga,


J.-J. Kim, S. Ueda, Properties of the quaternary half-metal-type Heusler
alloy Co2 Mn1x Fex Si, Phys. Rev. B 74, 104405 (2006).
[76] K. Kobayashi, R. Umetsu, K. Ishikawa, R. Kainuma, K. Ishida, Phase stability of B2 and L21 ordered phases in Co2 YGa (Y=Ti,V,Cr,Mn,Fe) alloys,
J. Magn. Magn. Mater. 310, 1794 (2007).
[77] Y. Miura, K. Nagao, M. Shirai, Atomic disorder effects on half-metallicity
of the full-Heusler alloys Co2 (Cr1x Fex )Al: A first-principles study, Phys.
Rev. B 69, 144413 (2004).
[78] S. Ishida, T. Masaki, S. Fujii, S. Asano, Theoretical search for half-metallic
films of Co2 MnZ (Z=Si, Ge), Physica B 245, 1 (1998).
[79] S. Picozzi, A. Continenza, A. J. Freeman, Co2 MnX (X=Si, Ge, Sn)
Heusler compounds: An ab initio study of their structural electronic, and
magnetic properties at zero and elevated pressure, Phys. Rev. B 66, 094421
(2002).
[80] I. Galanakis, P. H. Dederichs, N. Papanikolaou, Slater-Pauling behavior
and origin of the half-metallicity of the full-Heusler alloys, Phys. Rev. B
66, 174429 (2002).
[81] E. Sasioglu, L. M. Sandratskii, P. Bruno, I. Galanakis, Exchange interactions and temperature dependence of magnetization in half-metallic Heusler
alloys, Phys. Rev. B 72, 184415 (2005).
[82] H. C. Kandpal, G. H. Fecher, C. Felser, Calculated electronic and magnetic
properties of the half-metallic, transition metal based Heusler compounds, J.
Phys. D: Appl. Phys. 40, 1507 (2007).
[83] J. K
ubler, A. R. Williams, C. B. Sommers, Formation and coupling of
magnetic moments in Heusler alloys, Phys. Rev. B 28, 1745 (1983).
[84] M. Oogane, Y. Sakuraba, J. Nakata, H. Kubota, Y. Ando, A. Sakuma,
T. Miyazaki, Large tunnel magnetoresistance in magnetic tunnel junctions
using Co2 MnX (X=Al,Si) Heusler alloys, J. Phys. D: Appl. Phys. 39, 834
(2006).
[85] V. N. Antonov, H. A. D
urr, Y. Kucherenko, L. V. Bekenov, A. N. Yaresko,
Theoretical study of the electronic and magnetic structures of the Heusler
alloys Co2 (Cr1x Fex )Al, Phys. Rev. B 72, 054441 (2005).

148

Bibliography

[86] I. Galanakis, P. Mavropoulos, P. H. Dederichs, Electronic structure and


Slater-Pauling behaviour in half-metallic Heusler alloys calculated from first
principles, J. Phys. D: Appl. Phys. 39, 765 (2006).
[87] W. Kohn, Nobel Lecture: Electronic structure of matterwave functions and
density functionals, Rev. Mod. Phys. 71, 1253 (2005).
[88] H. C. Kandpal, G. H. Fecher, C. Felser, G. Schonhense, Correlation in
the transition-metal-based Heusler compounds Co2 MnSi and Co2 FeSi, Phys.
Rev. B 73, 094422 (2006).
[89] I. Galanakis, Appearance of half-metallicity in the quaternary Heusler alloys, J. Phys.: Condens. Matter 16, 3089 (2004).

[90] K. Ozdogan,
E. Sasioglu, A. B., I. Galanakis, Doping and disorder in
the Co2 MnAl and Co2 MnGa half-metallic Heusler alloys, Phys. Rev. B 74,
172412 (2006).
[91] T. Block, C. Felser, G. Jakob, J. Ensling, B. M
uhling, P. G
utlich, R. J.
Cava, Large negative magnetoresistance effects in Co2 Cr0.6 Fe0.4 Al, J. Solid
State Chem. 176, 646 (2003).

[92] I. Galanakis, K. Ozdogan,


A. B., E. Sasioglu, Effect of doping and disorder
on the half metallicity of full Heusler alloys, Appl. Phys. Lett. 89, 042502
(2006).
[93] S. Wurmehl, G. H. Fecher, K. Kroth, F. Kronast, H. A. D
urr, Y. Takeda,
Y. Saitoh, K. Kobayashi, H.-J. Lin, G. Schonhense, C. Felser, Electronic
structure and spectroscopy of the quaternary Heusler alloy Co2 Cr1x Fex Al,
J. Phys. D: Appl. Phys. 39, 803 (2006).
[94] P. Mavropoulos, I. Galanakis, V. Popescu, P. H. Dederichs, The influence
of spin-orbit coupling on the band gap of Heusler alloys, J. Phys.: Condens.
Matter 16, S5759 (2004).
[95] P. Mavropoulos, K. Sato, R. Zeller, P. H. Dederichs, Effect of spin-orbit
interaction on the band gap of half metals, Phys. Rev. B 69, 054424 (2004).
[96] M. Sargolzaei, M. Richter, K. Koepernik, I. Opahle, H. Eschrig, I. Chaplygin, Spin and orbital magnetism in full Heusler alloys: A density functional theory study of Co2 YZ (Y=Mn,Fe; Z=Al,Si,Ga,Ge), Phys. Rev. B
74, 224410 (2006).
[97] R. Skomski, P. A. Dowben, The finite-temperature densities of states for
half-metallic ferromagnets, Europhys. Lett. 58, 544 (2002).

Bibliography

149

[98] M. Lezaic, P. Mavropoulos, J. Enkovaara, G. Bihlmayer, S. Bl


ugel, Thermal Collapse of Spin Polarization in Half-Metallic Ferromagnets, Phys.
Rev. Lett. 97, 026404 (2006).
[99] L. Chioncel, E. Arrigoni, M. I. Katsnelson, A. I. Lichtenstein, Electron
Correlations and the Minority-Spin Band Gap in Half-Metallic Heusler Alloys, Phys. Rev. Lett. 96, 137203 (2006).
[100] L. Chioncel, Y. Sakuraba, E. Arrigoni, M. I. Katsnelson, M. Oogane,
T. Miyazaki, E. Burzo, A. I. Lichtenstein, Nonquasiparticle States in
Co2 MnSi Evidenced through Magnetic Tunnel Junction Spectroscopy Measurements, Phys. Rev. Lett. 100, 086402 (2008).
[101] I. Galanakis, Surface propeties of the half- and full-Heusler alloys, J.
Phys.: Condens. Matter 14, 6329 (2002).
[102] S. J. Hashemifar, P. Kratzer, M. Scheffler, Preserving the Half-Metallicity
at the Heusler Alloy Co2 MnSi(001) Surface: A Density Functional Theory
Study, Phys. Rev. Lett. 94, 096402 (2005).
[103] K. Nagao, Y. Miura, M. Shirai, Half-metallicity at the (110) interface
between a full Heusler alloy and GaAs, Phys. Rev. B 73, 104447 (2006).
[104] Y. Miura, H. Uchida, Y. Oba, K. Nagao, M. Shirai, Coherent tunneling
conductance in magnetic tunnel junctions of half-metallic full Heusler alloys
with MgO barriers, J. Phys.: Condens. Matter 19, 365228 (2007).
[105] Y. Miura, H. Uchida, Y. Oba, K. Abe, M. Shirai, Half-metallic interface
and coherent tunneling in Co2 YZ/MgO/Co2 YZ (YZ=MnSi, CrAl) magnetic
tunnel junctions: A first-principles study, Phys. Rev. B 78, 064416 (2008).
[106] P. Mavropoulos, M. Lezaic, S. Bl
ugel, Half-metallic ferromagnets for magnetic tunnel junctions by ab initio calculations, Phys. Rev. B 72, 174428
(2005).
[107] J. K
ubler, First principle theory of metallic magnetism, Physica B 127,
257 (1984).
[108] G. H. Fecher, H. C. Kandpal, S. Wurmehl, C. Felser, G. Schonhense,
Slater-Pauling rule and Curie temperature of Co2 -based Heusler compounds,
J. Appl. Phys. 99, 08J106 (2006).
[109] Y. Kurtulus, R. Dronskowski, G. D. Samolyuk, V. P. Antropov, Electronic
structure and magnetic exchange coupling in ferromagnetic full Heusler alloys, Phys. Rev. B 71, 014425 (2005).

150

Bibliography

[110] J. Thoene, S. Chadov, G. Fecher, C. Felser, J. K


ubler, Exchange energies,
Curie temperatures and magnons in Heusler compounds, J. Phys. D: Appl.
Phys. 42, 084013 (2009).
[111] J. K
ubler, Ab initio estimates of the Curie temperature for magnetic compounds, J. Phys. Condens. Matter 18, 9795 (2006).
[112] H. J. Elmers, S. Wurmehl, G. H. Fecher, G. Jakob, C. Felser,
G. Schonhense, Enhanced orbital magnetic moments in the Heusler compounds Co2 CrAl, Co2 Cr0.6 Fe0.4 Al, Co2 FeAl, J. Magn. Magn. Mater. 272,
758 (2004).
[113] K. Miyamoto, A. Kimura, K. Iori, K. Sakamoto, T. Xie, T. Moko, S. Qiao,
M. Taniguchi, K. Tsuchiya, Element-resolved magnetic moments of Heuslertype ferromagnetic ternary alloy Co2 MnGe, J. Phys.: Condens. Matter 16,
S5797 (2004).
[114] I. Galanakis, Orbital magnetism in the half-metallic Heusler alloys, Phys.
Rev. B 71, 012413 (2005).
[115] B. D. Cullity, Elements of X-ray diffraction, Addison-Wesley, Reading,
MA (1956).
[116] M. Kallmayer, A. Conca, M. Jourdan, H. Schneider, G. Jakob, B. Balke,
A. Gloskovskii, H. J. Elmers, Correlation of local disorder and electronic
properties in the Heusler alloy Co2 Cr0.6 Fe0.4 Al, J. Phys. D: Appl. Phys. 40,
1539 (2007).
[117] S. Wurmehl, J. T. Kohlhepp, H. J. M. Swagten, B. Koopmans, Hyperfine
field distribution in the Heusler compound Co2 FeAl probed by 59 Co nuclear
magnetic resonance, J. Phys. D: Appl. Phys. 41, 115007 (2008).
[118] http://www.bam.de/de/service/publikationen/powder cell.htm .
[119] http://www.crystalmaker.com/products.html .
[120] M. Bauer, Grundlagen und Strategien f
ur schnelles Schalten der Magnetisierung, Dissertation, Technische Universitat Kaiserslautern (2000).
[121] R. Lopusnik, Time resolved magneto-optical investigations of magnetization dynamics in iron garnet films, Dissertation, Technische Universitat
Kaiserslautern (2001).

Bibliography

151

[122] T. Mewes, H. Nembach, M. Rickart, B. Hillebrands, Separation of the


first- and second-order contributions in magneto-optic Kerr effect magnetometry of epitaxial FeMn/NiFe bilayers, J. Appl. Phys. 95, 5324 (2004).
[123] J. R. Sandercock, Brillouin scattering study of SbSi using a double-passed,
stabilized scanning interferometer, Opt. Comm. 2, 73 (1970).
[124] B. Hillebrands, Progress in multipass tandem Fabry-Perot interferometry:
A fully automatized, easy to use, self-aligning spectrometer with increased
stability and flexibility, Rev. Scien. Instr. 70, 1589 (1999).
[125] B. Hillebrands, TFPDAS3, Tandem-Fabry-Perot Acquisition Software
Version 02-24-2005, Kaiserslautern .
[126] J. R. Sandercock, Tandem Fabry-Perot Interferometer TFP-1, Operator
Manual, http://jrs-si.ch .
[127] W. Demtroder, Experimentalphysik 2, Springer, 2. Edition (1999).
[128] C. Chappert, H. Bernas, J. Ferre, V. Kottler, J.-P. Jamet, Y. Chen,
E. Cambril, T. Devolder, F. Rousseaux, V. Mathet, H. Launois, Planar
Patterned Magnetic Media Obtained by Ion Irradiation, Science 280, 1919
(1998).
[129] H. Bernas, T. Devolder, C. Chappert, J. Ferre, V. Kottler, Y. Chen,
C. Vieu, J.-P. Jamet, V. Mathet, E. Cambril, O. Kaitasov, S. Lemerle,
F. Rousseaux, H. Launois, Ion beam induced magnetic nanostructure patterning, Nucl. Instrum. Methods B 148, 872 (1999).
[130] T. Devolder, Light ion irradiation of Co/Pt systems: Structural origin of
the decrease in magnetic anisotropy, Phys. Rev. B 62, 5794 (2000).
[131] T. Devolder, J. Ferre, C. Chappert, H. Bernas, J.-P. Jamet, V. Mathet,
Magnetic properties of He+ -irradiated Pt/Co/Pt ultrathin films, Phys. Rev.
B 64, 064415 (2001).
[132] T. Mewes, R. Lopusnik, J. Fassbender, B. Hillebrands, M. Jung, D. Engel, A. Ehresmann, H. Schmoranzer, Suppression of exchange bias by ion
irradiation, Appl. Phys. Lett. 76, 1057 (2000).
[133] A. Mougin, T. Mewes, R. Lopusnik, M. Jung, D. Engel, A. Ehresmann,
H. Schmoranzer, J. Fassbender, B. Hillebrands, Modification of the Exchange Bias Effect by He Ion Irradiation, IEEE Trans, Magn. 36, 2647
(2000).

152

Bibliography

[134] A. Mougin, T. Mewes, M. Jung, D. Engel, A. Ehresmann,


H. Schmoranzer, J. Fassbender, B. Hillebrands, Local manipulation and
reversal of the exchange bias field by ion irradiation in FeNi/FeMn double
layers, Phys. Rev. B 63, 060409(R) (2001).
[135] J. Juraszek, J. Fassbender, S. Poppe, T. Mewes, B. Hillebrands, D. Engel, A. Kronenberger, A. Ehresmann, H. Schmoranzer, Tuning exchange
bias and coercive fields in ferromagnet/antiferromagnet bilayers with ion
irradiation, J. Appl. Phys. 91, 6898 (2002).
[136] J. Fassbender, S. Poppe, T. Mewes, A. Mougin, B. Hillebrands, D. Engel,
M. Jung, A. Ehresmann, H. Schmoranzer, G. Faini, K. J. Kirk, J. N. Chapman, Magnetization Reversal of Exchange Bias Double Layers Magnetically
Patterned by Ion Irradiation, Phys. Stat. Sol. (a) 189, 439 (2002).
[137] J. Fassbender, S. Poppe, T. Mewes, J. Juraszek, B. Hillebrands, K.-U.
Barholz, R. Mattheis, D. Engel, M. Jung, H. Schmoranzer, A. Ehresmann,
Ion irradiation of exchange bias systems for magnetic sensor applications,
Appl. Phys. A 77, 51 (2003).
[138] S. Poppe, J. Fassbender, B. Hillebrands, On the mechanism of irradiationenhanced exchange bias, Europhys. Lett. 66, 430 (2004).
[139] J. McCord, T. Gemming, L. Schultz, J. Fassbender, M. O. Liedke,
M. Frommberger, E. Quandt, Magnetic anisotropy and domain patterning of amorphous films by He-ion irradiation, Appl. Phys. Lett. 86, 162502
(2005).
[140] D. Ravelosona, C. Chappert, H. Bernas, D. Halley, Y. Samson, A. Marty,
Chemical ordering at low temperatures in FePd films, J. Appl. Phys. 91,
8082 (2002).
[141] C.-H. Lai, C.-H. Yang, C. C. Chiang, Ion-irradiation-induced direct ordering of L10 FePt phase, Appl. Phys. Lett. 83, 4550 (2003).
[142] S. O. Demokritov, C. Bayer, S. Poppe, M. Rickart, J. Fassbender, B. Hillebrands, D. I. Kholin, N. M. Kreines, O. M. Liedke, Control of Interlayer
Exchange Coupling In Fe/Cr/Fe Trilayers by Ion Beam Irradiation, Phys.
Rev. Lett. 90, 097201 (2003).
[143] D. Engel, I. Krug, H. Schmoranzer, A. Ehresmann, A. Paetzold, K. Roll,
B. Ocker, W. Maass, Alteration of exchange anisotropy and magnetoresistance in Co/Cu/Co/FeMn spin valves by ion bombardment, J. Appl. Phys.
94, 5925 (2003).

Bibliography

153

[144] D. Engel, H. Schmoranzer, A. Ehresmann, H. C. Mertins, D. Abramsohn, W. Gudat, Soft X-ray resonant magnetic reflection investigations of
FeMn/Co/Cu/Co spin valves modified by He-ion bombardment, Physica B
345, 185 (2004).
[145] J. Schmalhorst, V. Hoink, G. Reiss, D. Engel, D. Junk, A. Schindler,
A. Ehresmann, H. Schmoranzer, Influence of ion bombardment on transport
properties and exchange bias in magnetic tunnel junctions, J. Appl. Phys.
94, 5556 (2003).
[146] J. Fassbender, D. Ravelosona, Y. Samson, Tailoring magnetism by lightion irradiation, J. Phys. D: Appl. Phys. 37, R179 (2004).
[147] J. Fassbender, J. McCord, Magnetic patterning by means of ion irradiation and implantation, J. Magn. Magn. Mater. 320, 579 (2008).
[148] M. Nastasi, J. W. Mayer, J. K. Hirvonen, Ion-Solid Interactions: Fundamentals and Applications, Cambridge University Press (1996).
[149] J. F. Ziegler, J. F. Biersack, J. P. Littmark, The Stopping and Range of
Ions in Solids, Pergamon, New York (1985).
[150] J. F. Ziegler, SRIM manual, www.srim.org (2006).
[151] A. Conca, M. Jourdan, C. Herbort, H. Adrian, Epitaxy of thin films of
the Heusler compound Co2 Cr0.6 Fe0.4 Al, J. Cryst. Growth 299, 299 (2007).
[152] H. Schneider, E. Vilanova, B. Balke, C. Felser, G. Jakob, Hall effect in
laser ablated Co2 (Mn,Fe)Si thin films, J. Phys. D: Appl. Phys. 42, 084012
(2009).
[153] H. Schneider, G. Jakob, M. Kallmayer, H. J. Elmers, M. Cinchetti,
B. Balke, S. Wurmehl, C. Felser, M. Aeschlimann, H. Adrian, Epitaxial
film growth and magnetic properties of Co2 FeSi, Phys. Rev. B 74, 174426
(2006).
[154] S. Okamoto, O. Kitakami, Y. Shimada, Crystal distortion and magnetic
00
moment of epitaxially grown -Fe16 N2 , J. Magn. Magn. Mater. 208, 102
(2000).
[155] N. Tezuka, N. Ikeda, A. Miyazaki, S. Sugimoto, M. Kikuchi, Tunnel magnetoresistance for junctions with epitaxial full-Heusler Co2 FeAl0.5 Si0.5 electrodes with B2 and L21 structures, Appl. Phys. Lett. 89, 112514 (2006).

154

Bibliography

[156] M. Jourdan, A. Conca, C. Herbort, M. Kallmayer, H. J. Elmers,


H. Adrian, Tunneling spectroscopy of the Heusler compound Co2 Cr0.6 Fe0.4 Al,
J. Appl. Phys. 102, 093710 (2007).
[157] W. Wang, H. Sukegawa, R. Shan, T. Furubayashi, K. Inomata,
Preparation and characterization of highly L21 -ordered full-Heusler alloy
Co2 FeAl0.5 Si0.5 thin films for spintronics device applications, Appl. Phys.
Lett. 92, 221912 (2008).
[158] B. Rameev, F. Yildiz, S. Kazan, B. Aktas, A. Gupta, L. R. Tagirov,
D. Rata, D. Buergler, P. Gruenberg, C. M. Schneider, S. Kammerer,
G. Reiss, A. H
utten, FMR investigations of half-metallic ferromagnets,
Phys. Stat. Sol. (a) 203, 1503 (2006).
[159] H. Kijima, T. Ishikawa, T. Marukame, H. Koyama, K. Matsuda, T. Uemura, M. Yamamoto, Epitaxial Growth of Full-Heusler Alloy Co2 MnSi Thin
Films on MgO-Buffered MgO Substrates, IEEE Trans. Magn. 42, 2688
(2006).
[160] W. H. Wang, M. Przybylski, W. Kuch, L. I. Chelaru, J. Wang, Y. F.
Lu, J. Barthel, H. L. Meyerheim, J. Kirschner, Magnetic properties and
spin polarization of Co2 MnSi Heusler alloy thin films epitaxially grown on
GaAs(001), Phys. Rev. B 71, 144416 (2005).
[161] M. Buchmeier, B. K. Kuanr, R. R. Gareev, D. E. B
urgler, P. Gr
unberg,
Spin waves in magnetic double layers with strong antiferromagnetic interlayer exchange coupling: Theory and experiment, Phys. Rev. B 67, 184404
(2003).
[162] M. Buchmeier, H. Dassow, D. E. B
urgler, C. M. Schneider, Intensity
of Brillouin light scattering from spin waves in magnetic multilayers with
noncollinear spin configurations: Theory and experiment, Phys. Rev. B 75,
184436 (2007).
Visn
[163] J. Hamrle, J. Ferre, M. N
yvlt, S.
ovsk
y, In-depth resolution of the
magneto-optical Kerr effect in ferromagnetic multilayers, Phys. Rev. B 66,
224423 (2002).
[164] S. Picozzi, A. Continenza, A. J. Freeman, Magneto-optical properties of
Heusler compounds from a first-principles approach, J. Phys. D: Appl. Phys.
39, 851 (2006).
[165] T. Kubota et al. (unpublished).

Bibliography

155

[166] M. Belmeguenai, F. Zighem, Y. Roussigne, S.-M. Cherif, P. Moch,


K. Westerholt, G. Woltersdorf, G. Bayreuther, Microstrip line ferromagnetic resonance and Brillouin light scattering investigations of magnetic
properties of Co2 MnGe Heusler thin films, Phys. Rev. B 79, 024419 (2009).
[167] J. C. Slater, Atomic Radii in Crystals, J. Chem. Phys. 41, 3199 (1964).
[168] X. Liu, M. M. Steiner, R. Sooryakumar, G. A. Prinz, R. F. C. Farrow,
G. Harp, Exchange stiffness, magnetization, and spin waves in cubic and
hexagonal phases of cobalt, Phys. Rev. B 53, 12166 (1996).
[169] X. Liu, R. Sooryakumar, C. J. Gutierrez, G. A. Prinz, Exchange stiffness
and magnetic anisotropies in bcc Fe1x Cox alloys, J. Appl. Phys. 75, 7021
(1994).
[170] B. Hillebrands, Brillouin light scattering, in Y. Zhu (Editor), Modern
techniques for characterizing magentic materials, Springer (2005).
[171] R. Yilgin, Y. Sakuraba, M. Oogane, S. Mizukami, Y. Ando, T. Miyazaki,
Anisotropic Intrinsic Damping Constant of Epitaxial Co2 MnSi Heusler Alloy Films, Jpn. J. Appl. Phys. 46, L205 (2007).
[172] H. Schneider, C. Herbort, G. Jakob, H. Adrian, S. Wurmehl, C. Felser,
Structural, magnetic and transport properties of Co2 FeSi films, J. Phys. D:
Appl. Phys. 40, 1548 (2007).
[173] M. Kallmayer, H. Schneider, G. Jakob, H. J. Elmers, K. Kroth,
H. Kandpal, U. Stumm, C. Cramm, Reduction of surface magnetism of
Co2 Cr0.6 Fe0.4 Al Heusler alloy films, Appl. Phys. Lett. 88, 072506 (2006).
[174] M. Kallmayer, H. Schneider, G. Jakob, H. J. Elmers, B. Balke, C. Cramm,
Interface magnetization of ultrathin epitaxial Co2 FeSi(110)/Al2 O3 films, J.
Phys. D: Appl. Phys. 40, 1552 (2007).
[175] S. Tanuma, C. J. Powell, D. R. Penn, Calculations of Electron Inelastic
Mean Free Paths. V. Data for 14 Organic Compounds over the 50-2000 eV
Range, Surf. Interface Anal. 21, 165 (1994).
[176] G. H. Fecher, B. Balke, A. Gloskowskii, S. Ouardi, C. Felser, T. Ishikawa,
M. Yamamoto, Y. Yamashita, H. Yoshikawa, S. Ueda, K. Kobayashi, Detection of the valence band in buried Co2 MnSiMgO tunnel junctions by means
of photoemission spectroscopy, Appl. Phys. Lett. 92, 193513 (2008).

156

Bibliography

[177] J. Schmalhorst, S. Kammerer, M. Sacher, G. Reiss, A. H


utten, Interface structure and magnetism of magnetic tunnel junctions with a Co2 MnSi
electrode, Phys. Rev. B 70, 024426 (2004).
[178] N. D. Telling, P. S. Keatley, G. van der Laan, R. J. Hicken, E. Arenholz,
Y. Sakuraba, M. Oogane, Y. Ando, T. Miyazaki, Interfacial structure and
half-metallic ferromagnetism in Co2 MnSi-based magnetic tunnel junctions,
Phys. Rev. B 74, 224439 (2006).
[179] T. Saito, T. Katayama, T. Ishikawa, M. Yamamoto, D. Asakura,
D. Koide, X-ray absorption spectroscopy and x-ray magnetic circular dichroism of epitaxially grown Heusler alloy Co2 MnSi ultrathin films facing a MgO
barrier, Appl. Phys. Lett. 91, 262502 (2007).
[180] B. T. Thole, R. D. Cowan, G. A. Sawatzky, J. Fink, J. C. Fuggle, New
probe for the ground-state electronic structure of narrow-band and impurity
systems, Phys. Rev. B 31, 6856 (1985).
[181] T. J. Regan, H. Ohldag, C. Stamm, F. Nolting, J. L
uning, J. Stohr,
R. L. White, Chemical effects at metal/oxide interfaces studied by x-rayabsorption spectroscopy, Phys. Rev. B 64, 214422 (2001).
[182] T. Katayama, S. Yuasa, S. Saito, Y. Kurosaki, T. Saito, T. Kamino,
K. Kobayashi, Y. Suzuki, H. Manaka, T. Koide, X-ray absorption and x-ray
magnetic circular dichroism studies on a monatomic bcc-Co(001) layer facing an amorphous AlO tunnel barrier, J. Appl. Phys. 100, 023912 (2006).
[183] B. T. Thole, P. Carra, F. Sette, G. van der Laan, X-Ray Circular Dichroism as a Probe of Orbital Magnetization, Phys. Rev. Lett. 68, 1943 (1992).
[184] H. A. D
urr, G. van der Laan, D. Spanke, F. U. Hillebrecht, N. Brookes,
Electron-correlation-induced magnetic order of ultrathin Mn films, Phys.
Rev. B 56, 8156 (1997).
[185] S. Trudel, G. Wolf, H. Schulthei, J. Hamrle, B. Hillebrands, Tandem
magneto-optical Kerr effect magnetometer for the study of quadratic effects
submitted to J. Phys.: Conf. Series.
[186] Y. Huai, Spin-Transfer Torque MRAM (STT-MRAM): Challenges and
Prospects, AAPPS Bulletin 18, 33 (2008).
[187] Kambersk
y, On the Landau-Lifshitz relaxation in ferromagnetic metals,
Can. J. Phys. 48, 2906 (1970).

Bibliography

157

[188] T. Schneider, A. A. Serga, B. Leven, B. Hillebrands, R. L. Stamps,


M. Kostylev, Realization of spin-wave logic gates, Appl. Phys. Lett. 92,
022505 (2008).
[189] K. Lee, S. Kim, Conceptual design of spin wave logic gates based on a
MachZehnder-type spin wave interferometer for universal logic functions,
J. Appl. Phys. 104, 053909 (2008).
[190] J. D. Adam, IEEE 76, 159 (1988).

Curriculum vitae
Oksana Gaier

Personal information:
Date of birth:
22.10.1979
Place of birth:
Alma-Ata (USSR)
Nationality:
German
Marital status:
unmarried
School education:
1986 - 1993
1993 - 1995
1995 - 2000
08.06.2000

Public school Alma-Ata


St.-Matthias-Gymnasium Gerolstein
Eichendorff-Gymnasium Koblenz
Degree: General qualification for university entrance

Diploma and PhD studies:


2000 - 2006
Diploma studies of Physics, Technical University Kaiserslautern
07.11.2002
Intermediate examination, Subsidiary subject: Chemistry
27.03.2006
Diploma in Physics, Subsidiary subject: Chemistry,
Diploma thesis: Plasmon-induced near fields in coupled
nanoparticle systems
Since 2006
PhD studies, Technical University Kaiserslautern,
group of Prof. Dr. Burkard Hillebrands
09. - 12.2007
Research stay at Tohoku University Sendai, Japan,
group of Prof. Dr. Yasuo Ando

Acknowledgements
I would like to thank all who have contributed in a way or another to this thesis.
In particular, I thank Prof. Dr. Burkard Hillebrands for giving me the possibility to work in his group, to be a part of a project of great current interest,
involving national and international partners. The three years of my PhD studies have contributed a lot to my professional and personal development.
I also thank Prof. Dr. Claudia Felser from the Johannes Gutenberg University
of Mainz for accepting the role of second referee of this thesis, and for her contribution to the development of Heusler compounds.
A special thank goes to Prof. Dr. Yasuo Ando from Tohoku University in Sendai
for hosting me and giving me chance to work in his group for a few months,
and to get to know Japan, a country that impressed me a lot.
My gratitude goes also to Dr. Jaroslav Hamrle for his assistance during my PhD
studies, a good collaboration, and a first careful proof-reading of this thesis.
Many thanks go to Dr. Simon Trudel for a second proof-reading of this thesis,
English corrections, and very helpful and constructive suggestions.
I am also very grateful to Dr. Thomas Schneider and Dr. Andreas Beck for the
technical support with LATEX.
Moreover, I would like to thank my colleagues at the Johannes Gutenberg University of Mainz for a good collaboration in the course of this thesis. In particular, Prof. Dr. Hans-Joachim Elmers and Michael Kallmayer for the XAS/XMCD
measurements of the Co2 MnSi films, and all the discussions regarding the analysis of the data and basics of this experimental technique. I also wish to thank
Dr. Gerhard Jakob, Horst Schneider, Dr. Martin Jourdan, Dr. Andres Conca
Parra for the preparation of the Co2 FeSi, Co2 FeAl and CCFA films. A special

thank goes to Horst Schneider for SQUID magnetometry measurements carried out for the He+ irradiated Co2 MnSi films, and discussions about XRD. My
gratitude goes also to Dr. Andrei Gloskovskii and his colleagues for providing
HAXPES measurements included in this thesis.

Equal thanks go to my colleagues at the Tohoku University in Sendai for a good


collaboration, and their kindness during my stay in their group. In particular, I
would like to thank Dr. Yuya Sakuraba, Takahide Kubota and Sumito Tsunegi
for the fabrication of the Co2 MnSi films. I also thank all those who taught me
how to use the x-ray diffractometer.
I wish to thank Sebastian Hermsdorfer for the XRD characterization of the
Co2 MnSi series with different annealing temperatures.
Moreover, I thank Dr. J
urgen Fassbender for irradiation of Heusler films with
+
He ions, and the preliminary discussions that helped me gain a better understanding of the ion-solid interactions. This was very important for the planning
of the irradiation experiments. In this context, I wish also to thank Dr. Steffen
Blomeier for our discussions regarding ion irradiation in the starting phase of
this subproject. Furthermore, I thank Bernhard Reuscher and people from the
IFOS Kaiserslautern for the Ga+ irradiation of Co2 FeSi films.
My gratitude goes also to Dr. Rudolf Schafer for providing the possibility to
investigate the magnetization reversal of CCFA films using a Kerr microscope
located in his group at the IFW Dresden.
I also thank Dr. Aleksander A. Serga, Dr. Thomas Schneider, Sebastian Schafer
and Helmut Schulthei for their technical assistance during the BLS measurements.
Finally, I thank all present and former members of the Magnetism group for
the kind working atmosphere.
Nicht zuletzt mochte ich mich im besonderen Mae bei meinen Eltern, Galina
und Iwan Gaier, bedanken f
ur ihre Unterst
utzung jeglicher Art warend meines
gesamten Physikstudiums und besonders in den drei Jahren meiner Promotion.

You might also like