X-ray magnetic circular

dichroism study of oxide-based

magnetic materials and

half-metallic alloys

Doctor Thesis

Vijay Raj Singh

Department of Physics, University of Tokyo

July, 2011

Abstract

The study of spintronics materials such as diluted magnetic semiconduc-

tor (DMSs), multiferroic and half-metallic alloys is one of the most attractive

fields in science from the viewpoints of both academic research and applica-

tions. In order to clarify the origin of ferromagnetism of these spintronics, it is

necessary to investigate the electronic structure. In this thesis, we have inves-

tigated the electronic structure of spintronics materials using x-ray absorption

spectroscopy (XAS) and x-ray magnetic circular dichroism (XMCD).

The first discovery of room-temperature ferromagnetism in Co-doped

TiO2 by Matsumoto et al. [1] has arisen great interest in the search for such

materials and a number of studies have been carried-out to investigate whether

the ferromagnetism is carrier-mediated or not [2-3], but the issue still remains

controversial. XMCD at the Co 2𝑝 → 3𝑑 absorption (Co 𝐿2,3) edge is an ideal

technique to clarify this issue because it is an element-specific magnetic probe.

Our previous XMCD study has revealed that the ferromagnetism is not due

to segregated Co metal clusters but is due to Co2+ ions in the TiO2 matrix

[4]. However, the XMCD signal intensities were an order of magnitude lower

than that expected from the bulk magnetization [4]. We performed XAS and

XMCD studies on rutile Co-doped TiO2 by the surface-sensitive total electron

yield (TEY) mode and the bulk-sensitive total fluorescence yield (TFY) mode

and found that Co ions in the bulk indeed have a large moment of 0.8-2.2 𝜇𝐵

/Co [5]. Then we extended the same approach to anatase Co-doped TiO2 and

studied the correlation between magnetism and transport properties.

Further we performed the XAS and the XMCD studies of (1-𝑥) BiFeO3-

𝑥BiCoO3 (BFCO) thin films (where 𝑥 = 0 to 0.30) grown on LaAlO3(001)

substrates using a chemical solution deposition technique. The XAS results

indicated that the Fe ions were in the Fe3+ state and that the Co ions were in

the Co3+ state. XMCD results showed that the Fe ions were in ferromagnetic

state and the Co ions were in the paramagnetic state at room-temperature.

The XMCD measurements also revealed that antiferromagnetically coupled

Fe3+ ions were at the 𝑂ℎ and 𝑇𝑑 sites. The magnetic moment of the Fe ions

increased up to 20% Co content and after that it decreased. However, the Co

magnetic moment was nearly independent of Co content unlike Fe, and the

peak at 20% Co showed only a minor influence. The magnetization deduced

from XMCD is larger than that obtained by SQUID measurements, indicates

the enhancement of ferromagnetism within ∼ 5 nm from the surface, probed

by the total electron yield (TEY) method.

Finally we studied the magnetic and electronic states of Co2Mn𝛽Si0.88(CMS)/MgO and Co2Mn𝛽Ge0.38(CMG)/MgO magnetic tunnel junctions by

means of XMCD measurements. In particular, the Mn composition (𝛽) depen-

dences of the Mn and Co magnetic moments were investigated. As 𝛽 increases

in the CMG films, the spin magnetic moment of Mn decreases, consistent with

Picozzi et al.’s calculations [6] which predicts that the Mn magnetic moment

couples antiferromagnetically at MnCo site, leading to a reduction of saturation

magnetization. The Mn 𝐿2,3-edge XAS showed for higher 𝛽 values a Mn2+-

like multiplet structure in MnO, however, for lower values of 𝛽 we did not

obtain Mn2+-like multiplet structure. The Co spin magnetic moment for all

the samples was obviously larger and/or equal to theoretical value of 1.06 𝜇𝐵

[7]. For the Co-rich region, there is the possibility of the existence of CoMn

antisite because experimental value 𝑚spin(Co)=1.3 𝜇𝐵 is larger than that the

calculated value of the bulk, which is 1.06 𝜇𝐵 for Co at the regular Co site,

CoCo, consistent with Picozzi et al. [6]. These Co-rich films composition im-

ply the presence of Co antisites that would lower the spin polarization at the

Fermi level. For the Mn-rich region, 𝑚spin(Co) slightly decreases which is also

consistent with theoretically predicted value. A Co2+-like multiplet structure

in CoO was not observed in any films, indicating that the Co atoms were not

oxidized. For CMS, the 𝑚spin(Mn) behavior was similar to CMG and in this

case we did not observe oxidation of Mn. However, 𝑚spin(Co) was almost in-

dependent of 𝛽. Because the amount of CoMn in Co-rich CMS may be more

or less than that obtained in Co-rich CMG [8].

References

[1] Y. Matsumoto, M. Murakami, T. Shono, T. Hasegawa, T. Fukumura,

M. Kawasaki, P. Ahmet, T. Chikyow, S.-Y. Koshihara and H. Koinuma, Sci-

ence 291, 854 (2001).

[2] H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chikyow, T.

Hasegawa, H. Koinuma and M. Kawasaki, Nat. Mat. 3, 221 (2004).

[3] S. R. Shinde1, S. B. Ogale, J. S. Higgins, H. Zheng, A. J. Millis, V.N.

Kulkarni, R. Ramesh, R. L. Greene, and T. Venkatesan, Phys. Rev. Lett. 92,

66601 (2004).

[4] K. Mamiya, T. Koide, A. Fujimori, H. Tokano, H. Manaka, A. Tanaka,

H. Toyosaki, T. Fukumura, and M. Kawasaki, Appl. Phys. Lett. 89, 062506

(2006).

[5] V. R. Singh, Y. Sakamoto, T. Kataoka, M. Kobayashi, Y. Yamazaki,

A. Fujimori, F.-H. Chang, D.-J. Huang, H.-J. Lin, C. T. Chen, H. Toyosaki,

T. Fukumura and M. Kawasaki, J. of Phys.: Conds. Matt. 23, 176001 (2011).

[6] S. Picozzi, A. Continenza and A. J. Freeman, Phys. Rev. B 69, 094423,

(2004).

[7] S. Picozzi, A. Continenza and A. J. Freeman, Phys. Rev. B 66, 094421

(2002).

[8] D. Asakura, T. Koide, S. Yamamoto, K. Tsuchiya, T. Shioya, K.

Amemiya, V. R. Singh, T. Kataoka, Y. Yamazaki, Y. Sakamoto, A. Fujimori,

T. Taira and M. Yamamoto, Phys. Rev. B 82, 184419 (2010).

5

Contents

1 Introduction 1

1.1 Spintronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Diluted magnetic semiconductors . . . . . . . . . . . . . . . . . 1

1.2.1 Theoretical models for carrier-mediated ferromagnetism . 3

1.2.2 Possibility of ferromagnetism in Co-doped TiO2 dilute

magnetic oxide . . . . . . . . . . . . . . . . . . . . . . . 8

1.3 Multiferroics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.3.1 Multiferroic materials and magnetoelectric effect . . . . . 14

1.3.2 Single phase multiferroic materials: a brief history [59] . 14

1.3.3 Multiferroic composites . . . . . . . . . . . . . . . . . . . 14

1.3.4 Magnetoelectric effects . . . . . . . . . . . . . . . . . . . 15

1.3.5 Physical properties of BiFeO3 . . . . . . . . . . . . . . . 16

1.3.6 Substitution studies of transition-metal ions in BiFeO3 . 19

1.4 Heusler Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

1.4.1 Physical properties of Heusler alloys . . . . . . . . . . . . 21

1.4.2 Heusler half-metals in devices . . . . . . . . . . . . . . . 27

1.4.3 Heusler alloys: disorder and interfaces . . . . . . . . . . 29

2 Experimental Methods and Principles 31

2.1 Principles of x-ray magnetic circular dichroism and sum rules . . 31

2.1.1 X-ray absorption spectroscopy and x-ray magnetic cir-

cular dichroism . . . . . . . . . . . . . . . . . . . . . . . 31

2.1.2 XMCD sum rules . . . . . . . . . . . . . . . . . . . . . . 33

2.1.3 Analysis of x-ray magnetic circular dichroism spectra . . 34

2.2 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.2.1 NSRRC BL-11A . . . . . . . . . . . . . . . . . . . . . . 34

2.2.2 KEK-PF BL-16A . . . . . . . . . . . . . . . . . . . . . . 36

2.2.3 SPring-8 BL23SU . . . . . . . . . . . . . . . . . . . . . . 36

i

3 X-ray magnetic circular dichroism study of ferromagnetic Ti1−𝑥Co𝑥O2−𝛿

thin films 40

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . 42

3.3.1 Results on rutile Co-doped TiO2 . . . . . . . . . . . . . 42

3.3.2 Results on anatase Co-doped TiO2 . . . . . . . . . . . . 49

3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

4 X-ray magnetic circular dichroism study of ferromagnetic BiFe1−𝑥Co𝑥O3

thin films 56

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

4.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

4.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . 58

4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

5 Effect of off-stoichiometry in Heusler alloy thin films on spin-

dependent tunneling characteristics of Co2Mn𝛽𝑍/MgO (𝑍=

Ge, Si) magnetic tunnel junction studied by x-ray magnetic

circular dichroism 68

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

5.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

5.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . 71

5.3.1 Magnetic properties of Co2Mn𝛽Si0.88/MgO MTJs as a

function of Mn composition 𝛽 . . . . . . . . . . . . . . . 71

5.3.2 Magnetic properties of Co2Mn𝛽Ge0.38 MTJs as a func-

tion of Mn composition 𝛽 . . . . . . . . . . . . . . . . . 77

5.3.3 Mn and Co antisite defects in Co2MnSi and Co2MnGe . 83

5.3.4 Atom exchange in Co2MnSi and Co2MnGe . . . . . . . . 84

5.3.5 Comparison between Co2MnSi and Co2MnGe . . . . . . 86

5.3.6 Formula unit composition model for nonstoichiometric,

Ge-deficient Co2Mn𝛽Ge0.38 films . . . . . . . . . . . . . . 87

5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

6 Summary and Outlook 91

ii

Chapter 1

Introduction

1.1 Spintronics

Spintronics (a neologism meaning “spin transport electronics”), also known

as magnetoelectronics, is an emerging technology that exploits both the intrin-

sic spin of the electron and its associated magnetic moment, in addition to its

fundamental electronic charge, in solid-state devices [1, 2]. Spintronics is ex-

pected to improve upon traditional electronics and photonics devices, allowing

for enhancement in the form of reduced power consumption, faster device op-

eration, and new form of information computation. Thus new multifunctional

devices including spin valves, quantum bits for quantum computing, spin po-

larized light-emitting diodes and spin polarized field-effect transistors can be

and partially have already been realized [3]. Increased functionalities are also

expected, such as integrated magnetic and electronics operations on the same

chip.

1.2 Diluted magnetic semiconductors

Dilute magnetic semiconductors (DMS) are semiconductors doped with a

small amount of transition metal ions that introduce local magnetic moments.

The coupling between the localized moments and delocalized band-electrons

renders unique properties of DMS, such as a giant spin-splitting of electronic

states and indirect ferromagnetic exchange interactions between magnetic mo-

ments [1]. The latter is controlled by the manipulation of carriers by means

of, e.g., doping, electric fields, optical excitation, and quantum structures,

which are all key technologies within the well-established field of conventional,

charge-based electronics. Therefore, the tunable ferromagnetism attainable in

DMS is one of the leading areas of semiconductor spintronics.

1

The reported DMS materials are summarized in Table 1.1. Most of the

early dilute magnetic semiconductors such as tellurides, selenides and sulfides

were based on Mn-doped II-VI semiconductors. The valence match (i.e. iden-

tical charge state) of the cation of the II-VI host semiconductors to the dopant

(Mn), if they can be doped with charge carriers, makes it easy to prepare

samples with a large amount of Mn [2, 3]. The model materials (i.e. II-VI

materials) in which localized spins and delocalized holes can be introduced

and controlled independently, while dimensional effects can be tested by using

quantum heterostructures [4]. The previous studies showed that the dominant

magnetic interaction between Mn spins is antiferromagnetic in the II-VI type

DMS. It has also been proven difficult to create 𝑝- or 𝑛-type carriers to me-

diate ferromagnetic interactions, resulting in paramagnetic, antiferromagnetic

or spin glass behavior [2, 3]. Irrespective of their effects on fueling magnetism

research, II-VI DMS already found their applications in flat panel displays,

since efficient electroluminescence can be obtained by doping Mn and optical

isolators which allow the transmission of light in only one direction. Ferro-

magnetism was observed at temperatures below 2 K in Mn-based zinc-blende

III-V compounds such as CdMnTe after the result of carrier induced ferro-

magnetism in Mn-based zinc-blende III-V compounds [5]. Recently, Kuroda

et al. [6,7] suggested that Cr rich phases of ZnCrTe showed room temperature

ferromagnetism, causing a stimulation of II-VI DMS. However, its origin at

room-temperature is controversial so far.

In III-V DMS, divalent transition metal ions (Mn) substitute for trivalent

cations, thus generating holes whereas in II-VI DMS, additional doping of 𝑝-

type or 𝑛-type elements is required. The holes introduced by the magnetic ions

mediate ferromagnetic interaction between the magnetic ions in III-V DMS.

The reported Curie temperatures of III-V DMS are generally higher than II-

VI DMS but are still too low below room-temperature and thus for industrial

applications. For (GaMn)As, the highest record Curie temperature in III-V

DMS is 173 K [8]. Recently, room-temperature ferromagnetism have been

reported by several oxides and nitrides. As a matter of fact, the studies by

mean-field Zener model based on RKKY exchange interaction developed by

Dietl et al. [9] aimed at searching robust room-temperature DMS were trig-

gered, which predicts 𝑝-type GaN and 𝑝-type ZnO as promising candidates for

room-temperature DMS. Also because oxides and nitrides were already com-

mon materials for light-emitting devices in industry, this prediction triggering

tremendous works to dope transition metal ions into oxides and nitrides. How-

ever, oxides are naturally 𝑛-type because of oxygen vacancies formed during

the growth, while the mean field Zener model concerns 𝑝-type wide band gap

materials. As an alternative theoretical model, Coey et al. [10] suggested that

2

oxygen vacancies forming a spin-split impurity band may play a vital role in

forming carrier-induced ferromagnetism. It is also reported, however, that

oxygen vacancies act as active sites for forming clusters [11]. In fact, there are

comparable amounts of reports that indicate that the magnetism originates

from segregated clusters magnetic materials in oxides against and reports that

claim intrinsic magnetism. It should be noted that in the latter case a careful

study of the structural properties of the materials is often lacking [12,13].

Table 1.1: Representative dilute magnetic semiconductors.Material Class Material References

II-VI 𝑝-Cd1−𝑥Mn𝑥Te:N [5]

II-VI 𝑝-Zn1−𝑥Mn𝑥Te:N [4]

II-VI Zn1−𝑥Cr𝑥Se [14]

II-VI Zn1−𝑥Cr𝑥Te [15]

IV-VI Pb1−𝑥−𝑦Sn𝑦Mn𝑥Te [16]

III-V In1−𝑥Mn𝑥As [17]

III-V Ga1−𝑥Mn𝑥As [18]

III-V GaMnN [19]

III-V GaCrN [20]

III-V GaMnP:C [21]

III-V GaMnSb [22]

IV Ge1−𝑥Mn𝑥 [23]

Oxide Co-TiO2 [24]

Oxide Mn-ZnO [25]

Oxide Co-SnO2 [26]

Oxide Fe-SnO2 [27]

Oxide Cr-In2O3 [28]

1.2.1 Theoretical models for carrier-mediated ferromag-

netism

1.2.1.1 Mean-field and Zener model [29]

Dietl et al. [9] have shown that, for carrier densities lower than in metals,

the double exchange or RKKY cannot be the mechanism leading to carrier-

3

induced ferromagnetism in Mn-based III-V and II-VI DMSs. Dietl et al. [9]

proposed the Zener model based on ferromagnetic interactions mediated by

free carriers in tetrahedrally coordinated semiconductors. In the Zener model

[32], the direct interaction between 𝑑 shells of the adjacent Mn atoms (super-

exchange) leads to a tendency for an antiferromagnetic alignment of the 𝑑 shell

spins because the Mn-𝑑 shell is half-filled. On the other hand, the indirect

coupling of spins through the conduction electrons tends to align the spins

of the incomplete 𝑑 shells in a ferromagnetic manner. It is only when this

dominates over the direct super-exchange coupling between adjacent 𝑑 shells

that ferromagnetism is present. Accordingly, the mean-field approach basically

assumes that the ferromagnetism occurs through interactions between the local

moments of the Mn atoms mediated by free holes in the material. The spin-spin

coupling which is also assumed to be a long range interaction allows the use of

a mean-field approximation. The mean-field model calculates the effective spin

density due to the Mn ion distribution. The direct Mn-Mn interactions are

antiferromagnetic so that the Curie temperature TC, for a given material with

a specific Mn concentration and hole density, is determined by a competition

between the ferromagnetic and antiferromagnetic interactions. It should be

noted that the Zener theory does not take into account the itinerant character

of the magnetic electrons and the quantum oscillations of the electron spin

polarization around the localized spins. Later, both of these were established

to be critical concepts for the theory of magnetic metals [9]. The effect of

the quantum oscillations averages to be zero in semiconductors because the

mean distance between the carriers is greater than that between the spins. If

the quantum oscillations of the electron spin polarization around the localized

spins are taken explicitly into account then the Zener model becomes equivalent

to RKKY. In Fig. 1.1 the limitations of mean field theory are summarized

schematically [34].

4

Figure 1.1: A schematic phase diagram as a function of the exchange coupling

strength (abscissa) relative to the band Fermi energy (𝜖𝐹 ) and the carrier

concentration (ordinate) relative to the Mn concentration for carrier-induced

ferromagnetism in diluted magnetic semiconductors [34].

1.2.1.2 Bound magnetic polaron model

A drawback of the mean field Zener model is that charge carriers are treated

as free carriers. It does therefore not explain the experimentally observed

transport properties of insulating and ferromagnetic (GaMn)As, in particular

the observation of a Mott variable range hopping behavior at low temperatures

[37].The bound magnetic polaron (BMP) model is the opposite approach of

the mean field Zener model, which treats the carriers as quasi-localized states

in an impurity band. In this limit, a localized hole in (GaMn)As exhibits

antiferromagnetic exchange interaction with a number of magnetic impurities

within its localization radius, leading to the formation of a bound magnetic

bound polaron, illustrated in Fig. 1.2. In contrast to the antiferromagnetic

exchange interactions between the local spins and the holes leading to the

existence of BMP, the interaction between magnetic polarons is ferromagnetic.

Since the ratio of the exchange and thermal energy governs the effective radius

of the magnetic polaron, BMPs overlap at sufficiently low temperature. This

gives rise to a ferromagnetic exchange interaction between percolated BMPs

5

at low temperature. The disorder effects play a crucial role in the magnetic

properties [38] if the hole localization radius is less than the distance between

BMPs.

Figure 1.2: Interaction of two bound magnetic polarons (BMP). The shaded

region indicates overlap affected by fields from the two BMPs. The small and

large arrows show impurity and hole spins, respectively [37].

1.2.1.3 Donor impurity band model

The finding promising candidates for room temperature DMS, such as Mn-

doped 𝑝-ZnO and 𝑝-GaN, was of much interest in the scientific community

after the mean field Zener model prediction. Beside the above predicted can-

didates, several oxides such as TiO2, ZnO, SnO2, and In2O3 [24, 25, 27, 28]

have been reported to show room-temperature ferromagnetism. However, the

mean field Zener model predicted 𝑝 type DMS systems to be ferromagnetic,

while the reported materials are 𝑛-type, except for a few cases so a controversy

arises. Furthermore, many of the reported dilute magnetic oxides have Curie

temperatures above 300 K. Later, Coey et al. [10] suggested the donor impu-

rity band model to describe the properties of defect (e.g. oxygen vacancies)

derived 𝑛-type dilute magnetic oxides, which is the extension of the BMP the-

ory above described. Many oxides are 𝑛-type, due to oxygen vacancies, and

have a high dielectric constant. The main ingredients of the donor impurity

band model are as follows. Shallow donors associated with the defects form

6

BMPs, via which ferromagnetic ordering of magnetic moments of dopants is

mediated. The polarons overlap at a sufficiently high BMP concentration, thus

leading to a spin-split impurity band in the band gap and ferromagnetic order-

ing throughout the material. The ferromagnetic coupling between magnetic

ions via an impurity band is illustrated in Fig. 1.3 [10]. According to this

theory, if the donor electron resides in the vicinity of a magnetic impurity a

high Curie temperature is possible, even if the hybridization between the 3𝑑

levels and the conduction band states is just 1 ∼ 2 % [10]. There are two

possibilities for BMP formation by considering that the 3𝑑 levels of transition

metals in the series from Ti to Cu are below the conduction band. The first

occurs near the beginning of the 3𝑑 series, where the majority 3𝑑 level crosses

the Fermi level as shown in the impurity band in Fig. 1.3 (c), and the second is

towards the end of the 3𝑑 series where the minority 3𝑑 level crosses the Fermi

level as shown in Fig. 1.3 (b). This donor impurity band model relies on the

donor formation from defects, which however, are also favored sites of forming

metallic clusters. In addition, the weak 𝑠-𝑑 exchange interaction renders the

model rather impractical. The exchange interaction between band electrons

and the 3𝑑-electrons of the magnetic ions attributed to potential exchange and

kinetic exchange due to the hybridization. The potential exchange interaction

always exists which is induced by the repulsive Coulomb interaction between

the band and 𝑑-electrons. This process tends to align the spins of the band elec-

trons parallel to that of the dopant magnetic moments. The kinetic exchange

contribution stems from a reduction of kinetic energy by delocalization. The

kinetic exchange contribution is due to the hybridization of 3𝑑 levels with the

𝑠- and 𝑝-bands. At the Γ point, 𝑠-𝑑 hybridization is symmetry forbidden [39]

but 𝑝-𝑑 hybridization is always allowed, which may be a reason why 𝑝-type

materials are favored for DMS research. In summary, it is by now generally

accepted that the impurity band model cannot explain the Curie temperatures

above room temperature that have been observed in transition metal doped

magnetic oxides and nitrides [40]. In 𝑛-type DMS with Mn2+/3+ impurities

the exchange mechanism is rather close to the kinematic exchange proposed

by Zener for mixed-valence Mn ions [41].

7

Figure 1.3: The band structure diagram of an oxide with 3𝑑 impurities and a

spin-split donor impurity band. (a) A position of the 3𝑑 level for which the

Curie temperature is low and the splitting of the impurity band is small. (b)

and (c) show cases in which the minority (b) or majority (c) 3𝑑-states interact

with the spin-split donor impurity band [10].

1.2.2 Possibility of ferromagnetism in Co-doped TiO2

dilute magnetic oxide

1.2.2.1 Extrinsic ferromagnetism in Co-doped TiO2 dilute magnetic

oxide

After the mean field Zener model prediction, the first reported dilute mag-

netic oxide with high Curie temperature is the Co-doped TiO2 followed by an

avalanche of reports on the observation of room temperature ferromagnetism

in various oxides such as ZnO, SnO2, In2O3 etc. [12]. Only Co-doped TiO2

has been reported consistently to show ferromagnetism at room temperature

among all dilute magnetic oxides (DMOs). TiO2 has three polymorphs, namely

rutile, anatase and brookite, as illustrated in Fig. 1.4 [43]. Ferromagnetism

has been reported in the rutile and anatase phases. The phase of the TiO2

thin films can be determine from the choice of the substrate. The anatase

phase is grown with lattice mismatch -0.26% and -3.1% On LaAlO3 (LAO)

and SrTiO3 (STO) substrate, respectively. Anatase films are obtained inde-

pendent of the growth method. Films produce the rutile phase if grown on Si

or Al2O3 substrate. Since rutile is known to be the most stable phase, it is

often observed as outgrowths formed in anatase thin films. Fig. 1.5 shows a

TEM image of rutile nanocrystals in a pure (i.e. undoped) anatase TiO2 thin

film grown on LAO [44]. Related to this issue, there are some known factors

to play a role for distributing Co homogeneously. The low growth rate (0.01

8

nm/s) in molecular beam epitaxy (MBE) leads to a layer-by-layer growth and

homogeneous Co distribution in the film as suggested by Chambers et al. [44].

In contrast, a higher rate (0.04 nm/s) leads to a large density of rutile phases

to which Co segregates above 823 K. Post-annealing is reported to result in

redistribution of Co atoms but also an increased clustering of Co within the

film [43]. Another possible factor for influencing the Co distribution could

be the oxygen vacancies. With decreasing oxygen pressure during growth, an

increasing tendency of Co to cluster is reported [43].

Figure 1.4: Structure of TiO2: (A) rutile, (B) anatase, and (C) brookite.

Initially, Chambers et al. [44] suggested carrier mediated ferromagnetism

in anatase Co-doped TiO2 thin films by showing that ferromagnetic behavior is

enhanced by increasing the carrier concentration. The anomalous Hall effects

(AHE) and magneto optical dichroism in rutile, and anatase Co-doped TiO2

also support carrier mediated ferromagnetism [45–48]. AHE is the well known

ferromagnetic response of carriers in ferromagnetic materials. In Fig. 1.6,

it is shown that rutile Co-doped TiO2 represents the anomalous Hall effect,

while its magnetic field dependence is similar to that of the magnetization

measured by magnetometry [49]. The AHE and magneto optical dichroism

measured by Toyosaki et al. [50] in rutile samples, and found a correlation

between them as a function of carrier concentration and external magnetic

field. These measurements suggest that carriers enhance the ferromagnetic

exchange interactions between isolated Co magnetic moments in Co-doped

TiO2.

9

Figure 1.5: Bright-field TEM image of 50-nm pure TiO2 at LaAlO3(001) sub-

strate. Rutile nanocrystals are indicated by R1 and R2 [44].

Figure 1.6: For rutile Co-doped TiO2, the magnetic field dependence of the

Hall resistivity at 300 K. The inset figure is the magnetic field dependence of

the magnetization for the same sample at 300 K [49].

Shinde et al. [51] suggested, that observing an AHE is not a robust test for

confirming carrier mediated ferromagnetism because the co-occurrence of super

10

Figure 1.7: XMCD spectra of anatase Co-doped TiO2 for different post an-

nealing times in comparison with those of Co metal: 0 (as grown), 2, 10, and

20 min. (a) Co 𝐿2,3-edge XAS spectra (Co 10%) (b) Co 𝐿2,3-edge XAS spec-

tra recorded with right- and left circular polarization (Co 10%) (c) Resulting

difference spectra that is, X-ray magnetic circular dichroism (XMCD).

Figure 1.8: For a rutile Co-doped TiO2 thin film (3% Co) (a) Co 𝐿3-edge region

with right- and left circular polarization. (b) Co 𝐿3-edge XMCD. Alleged

multiplet features are denoted by arrows [53].

11

paramagnetic Co clusters and the AHE is possible in Co-doped TiO2 films.

Since XMCD is a useful probe since it reflects element specific contributions

to magnetism so Kim et al. [52] investigated the origin of ferromagnetism of

anatase Co-doped TiO2 with x-ray absorption spectroscopy (XAS) and x-ray

magnetic circular dichroism (XMCD). They found that the ionic multiplet

structure of the Co 𝐿-edge is smeared out gradually with increasing annealing

time, and finally the spectral shape becomes identical to that of Co metal,

suggesting that Co metal clusters are the cause for ferromagnetism as shown in

Fig. 1.7 (a), In Fig. 1.7 (c), the weak XMCD signal increases with annealing

time at 400∘C, indicating that most Co is segregated during the annealing

process. In contrast, Mamiya et al. [53] reported multiplet features in Co 𝐿-

edge XMCD spectra in Fig. 1.8, suggesting that Co2+ ions, and not metallic

Co which is characterized by featureless spectra, contribute to magnetism.

1.2.2.2 Intrinsic ferromagnetism in Co-doped TiO2 dilute magnetic

oxide

Several theories proposed to explain the origin of room temperature ferro-

magnetism in dilute magnetic oxides. However, the local spin density approx-

imation (LSDA) is well known to overestimate the 𝑠𝑝-𝑑 hybridization and the

energy of the transition metal 3𝑑 level relative to the band edges, due to the

underestimation of the band gap. Consequently, this approach leads to a lot

of conflicting results in DMS studies [2, 10]. Coey et al. [10] proposed an im-

purity band model which may be a possible explanation for carrier mediated

magnetism, but a recent calculation shows that oxygen vacancies in oxides

induce deep levels, and cannot lead to long range ferromagnetic exchange in-

teraction [54, 55]. Quilty et al. [56] suggested a strong hybridization between

the conduction band and 𝑡2𝑔-states of a high spin Co2+ ion in rutile Co-doped

TiO2,. The direct 𝑑 − 𝑑 hybridization as illustrated in Fig. 1.9. by X-ray

photoelectron spectroscopy (XPS) measurements showed that Co2+ is with

high-spin state in which an unoccupied 𝑡2𝑔 state is expected to hybridize with

the Ti 3𝑑 𝑡2𝑔 derived conduction band. The authors observed a shift of the

conduction band, which may be expected due to its exchange splitting with

increasing Co doping up to 10%.

For proving intrinsic DMS behavior, carrier-mediated ferromagnetism

such as tunable ferromagnetism and control of magnetization direction by elec-

tric fields should be demonstrated, along with the measurement of AHE. So far,

these have not yet been demonstrated and the mystery the origin of magnetism

in dilute magnetic oxides still remains. If the reported room-temperature di-

lute magnetic oxides are real DMS, they should show electric tunability of fer-

romagnetism, large tunnel magnetoresistance effects, strong magneto-optical

12

effects, and the existence of a spin-split band of carriers [10].

Figure 1.9: The band diagram which shows the high-spin Co2+ state and the

resulting strong 𝑡2𝑔-𝑡2𝑔 coupling between the Ti 3𝑑 and Co 3𝑑 𝑡2𝑔 states [56].

13

1.3 Multiferroics

1.3.1 Multiferroic materials and magnetoelectric effect

Multiferroics are materials in which at least two of the ferroelectric, ferro/anti

ferromagnetic and ferroelastic phases coexist. Though the mechanisms that

allow ferroelectricity and ferromagnetism seem to be incompatible, there are

a select few materials in which ferroelectricity and ferromagnetism are both

present, namely BiFeO3, Cr2O3, yttrium- iron-garnets, boracites, rare-earth

ferrites and manganese-based perovskites. In these materials, the ferroelectric

and ferro/antiferromagnetic phases are coupled in such as way as to produce

a cross phenomenon known as the magnetoelectric (ME) effect. This allows

manipulation of the magnetic phase with an external electric field and/or ma-

nipulation of the electric phase with external magnetic field. The integration

of the ME effect into device technology would have substantial implications,

however the above mentioned single phase materials exhibit prohibitively small

ME effect [57, 58].

1.3.2 Single phase multiferroic materials: a brief history

[59]

Nickel iodine boracites, Ni3B7O13I were discovered as first ferromagnetic

ferroelectric material. Many more multiferroic boracites compounds were syn-

thesized by the above compound, these compounds have complex structures

with many atoms per formula unit and more than one formula unit per unit

cell. The isolation can be prevented in these materials by the large number of

inter-ionic interactions, both of the essential factors causing multiferroism and

of the nature of the coupling between the magnetic, electric, and structural

order parameters. Nickel iodine boracites can be thought of as the “Rochelle-

salt” of magnetic ferroelectrics. It has wide applicability or to contribute to

our increased understanding in the field.

Other Perovskites: A number of other perovskites are known to have

ferroelectric and magnetic (mostly of the Antiferromagnetic type) ordering.

These include the manganites of the small rare earth elements and yttrium

and a few compounds in which Bi is the large cation. Table 1.2 lists some of

the known Multiferroic materials adapted from [58]

1.3.3 Multiferroic composites

There is only few number of single-phase materials which exhibit the coexis-

tence of strong ferro/ferrimagnetism and ferroelectricity at room temperature.

14

Table 1.2: Examples of materials that exhibit ME effect. Nota-

tion: FE-Ferroelectric, AFE-Antiferroelectric, FM-Ferromagnetic, AFM-

Antiferromagnetic and WFM-weak Ferromagnetic [58].

Compound Type of electric order Type of magnetic order TC (K) TN (K)

Pb(Fe2/3W1/3)O3 FE AFM 178 363

Pb(Fe1/2Nb1/2)O3 FE AFM 387 143

Pb(Co1/2W1/2)O3 FE WFM 68 9

Pb(Mn2/3W1/3)O3 AFE? AFM 473 203

Pb(Fe1/2Ta1/2)O3 FE AFM 233 180

Eu1/2Ba1/2TiO3 FE FM 165 4.2

BiFeO3 FE AFM 1123 650

BiMnO3 AFE FM 773 103

YMnO3 FE AFM 913 80

YbMnO3 FE AFM/WFM 983 87.3

HoMnO3 FE AFM/WFM 873 76

ErMnO3 FE AFM 833 79

Ni3B7O13I FE WFM 64 64

Ni3B7O13Br FE WFM 398 30,40

Co3B7O13I FE WFM 197 38

Van Suchtelen et al. [60] proposed that composites of piezoelectric and mag-

netostrictive phases can be electromagnetically coupled via stress mediation

.

1.3.4 Magnetoelectric effects

The coupling between magnetic and electric properties of a material gives rise

to magneto-electric effects. The Magnetoelectric (ME) effects were very popu-

lar in the beginning of 19th century after their discovery by Curie and Rontgen.

However, the progress understanding ME effects seems to have stopped since

about 1970 because of the lack of materials and degrees of freedom to modify

these effects.

BiFeO3 (BFO) is the only single phase material which shows simulta-

neously ferroelectric and ferromagnetic properties at room temperature. BFO

was first synthesized by Royen and Swars. Various studies have been done to

15

this compound, mostly on ceramics motivated by the potential high magneto-

electric property. However, for many researchers it was the matter of contro-

versies to discuss relationship between the structural and physical properties

of BFO.

1.3.5 Physical properties of BiFeO3

1.3.5.1 Structural Properties of BiFeO3

The structure of BFO is characterized by two distorted perovskite unit cells

(a𝑟 = 3.96 A, 𝛼𝑟 = 0.6∘) connected along their body diagonal, denoted by

the pseudocubic < 111 >, to form a rhombohedral unit cell as shown in Fig.

1.10 (a) [61, 62]. The ferroelectric state is realized by a large displacement

of the Bi ions relative to the FeO6 octahedra. This structure results in two

important considerations. First, the ferroelectric polarization lies along the

pseudocubic < 111 > leading to the formation of eight possible polarization

variants, corresponding to four structural variants [63, 64]. Second, the anti-

ferromagnetic ordering of BFO is G-type, in which the Fe magnetic moments

are aligned ferromagnetically within the(111) plane and antiferromagnneti-

cally between adjacent (111) plane. Additionally, BFO is known to exhibit

a spin cycloid structure in the bulk [64] and the preferred orientation of the

antiferromagnnetically aligned spins lies within the (111) plane, perpendicular

to the ferroelectric polarization direction with six equivalent easy axes within

that plane. The magnitude of the ionic shifts are Bi: 0.62Aalong [111], Fe:

0.23Aalong [111], O: 0.30Aalong [111] (all values ± 0.03A). The nature of the

oxygen shifts is more easily visualized by considering the atomic positions and

shifts on a (111) rhombohedral plane. It is seen that the oxygen shifts are

essentially along a line between the projections of two Bi atoms on this plane.

In the ideal perovskite structure, Bi atoms would lie in this (111) plane, but

in BFO Bi shifts by 0.62Anormal to the plane.

The unit cell of BFO is not the unit molecular cell as shown in Fig. 1.10

(b), but may be represented by a rhombohedron, having twice the volume of

the unit molecular cell, and generated by taking three face diagonals meeting

at a vertex of the cube shown as three intersecting edges of the rhombohedron.

16

(a) (b)

Figure 1.10: (a) The crystal structure of BFO and its ferroelectric polarization

(arrow) and antiferromagnetic plane (shaded planes) [62]. (b) Hexagonal unit

cell of BFO [65].

1.3.5.2 Magnetic properties of BiFeO3

BFO has a G-type antiferromagnetic configuration, where each Fe3+ is

surrounded by six antiparallels nearest neighbors. However, original neutron

study did not have the resolution to obtain information on the exact spin ori-

entation. Sosnowska et al. [64] proposed a modified G-type antiferromagnetic

structure where the spin of Fe3+ is subjected to a long-range modulation as

shown in Fig. 1.11.

BFO was discovered in 1950s and till now it has been the subject of nu-

merous investigations. Due to its simple perovskite structure, BFO is a model

material system for investigating the nature of interactions between structural,

electrical and magnetic order parameters. It is expected to have large polar-

ization and piezoelectric coefficients because of its high Curie temperature and

large distortion: but this has previously not been observed. It should have a

noticeable magnetization due to spin canting: but has only been observed in

single crystal under ultra high magnetic field. Furthermore, the coexisting fer-

roelectric and antiferromagnetic order parameters offer an additional degree of

freedom, via the magnetoelectric (ME) exchange. However, the ME exchange

interaction remains an invention in BFO.

17

Figure 1.11: Portion of the BFO lattice and the arrows indicate the Fe3+

moment direction of the proposed model. The spiral period is reduced for

illustration purpose. Figure is taken from Sosnowska et al. [64].

1.3.5.3 Electrical properties of BiFeO3

Due to the low resistivity of samples the electrical characterization on bulk

BFO has been very difficult. The controversy about whether it is ferroelectric

or antiferroelectric was finally settled based on the hysteresis loop measured

by Teague et al. [66]. He performed an experiment in liquid nitrogen which

shows lower leakage current due to lower charge carrier density and mobility.

The measured spontaneous polarization was 3.5 𝜇C/cm2 along the < 100 >

direction, which represents 6.1 𝜇C/cm2 in the < 111 > direction. This value

is much smaller than what would be expected for a ferroelectric material.

The leakage problem, likely due to defects and non-stoichiometry, has been

hampering more comprehensive studies about the bulk BFO and has limited

applications of this material. To solve this problem, recent work has focused

on solid solutions of BFO with other ABO3 materials, such as BaTiO3, which

can prevent second phase formation and increase sample resistivity. For ex-

ample, Ueda et al. have reported a remnant polarization of 2.5 𝜇C/cm2 from

(Bi0.7Ba0.3)(Fe0.7Ti0.3)O3 film [67–69]. The various measured values for the

18

polarization in BFO are summarized in following table-

Table 1.3: The values for the polarization in BiFeO3 is taken from litrature,

in chronological order with the oldest at the top.P [𝜇C/cm2] Sample type Ref.

6.1 Bulk single crystals [66]

2.5 (Bi0.7Ba0.3)(Fe0.7Ti0.3)O3 films (300 nm) on Nb-doped SrTiO3 [67]

2.2 Polycrystalline films (200 nm) [70]

50-90 Thin films (400-100 nm) on SrRuO3 /SrTiO3 [68]

35.7 Polycrystalline films (350 nm) [71]

8.9 Bulk ceramics [72]

158 Polycrystalline films (300 nm) [73]

The controversies concerning BFO can be summarized as follows-

(1) Due to the samples high conductivity, electrical characterization of

bulk single crystal/ceramics has been difficult. The spontaneous polarization

value measured Teague et al. [66] is much smaller than what would be expected

for a ferroelectric material with such a high Curie temperature and large dis-

tortion. Later, scientists tried to mix other ABO3 materials into BFO forming

solid solutions; this helps to increase the sample resistivity. Ueda et al [67]

reported a P𝑟 of ∼ 2.5 𝜇C/cm2 from BFO/BaTiO3 thin films. More recently,

BFO thin films with high resistivity have also been made [74].

(2) Early neutron diffraction study revealed a 𝐺-type antiferromagnetic

spin order for BFO with a small canting between neighboring antiparallel Fe3+

ions [70]. It was concluded that BFO should show weak ferromagnetic prop-

erty at room temperature. But no such behavior has been reported. Later,

Sosnowska et al. [64, 70] reported that the antiferromagnetic order of BFO is

subjected to a spiral modulation that cancels out the net magnetization. By

breaking this cycloidal structure, one could release the magnetization due to

canting [75,76].

1.3.6 Substitution studies of transition-metal ions in BiFeO3

As we know that BFO is a ferroelectric and an antiferromagnet with spa-

tially modulated spin structure. This structure does not allow net magneti-

zation and inhibits the observation of a notable linear magnetoelectric effect.

This problem can be solved by substitution of Fe3+ by other transition-metal

ions. Unfortunately, 𝐵-site ion doping decreases magnetic ordering temper-

ature drastically [77]. Moreover, a high pressure synthesis is necessary to

prepare the ceramic samples [77]. An alternative way to induce a net mag-

19

netization in the BFO is a rare-earth (RE) substitution in the 𝐴 sublattice

of the ABO3 structure. Gd3+, Tb3+, etc., possess a large magnetic moment,

causing significant increase in magnetization even in the lightly doped com-

pounds. More recently, enhancement of magnetization has been reported in

Bi1−𝑥RE𝑥FeO3 (RE=Nd3+,Sm3+), materials [78, 79]. Moreover, the appear-

ance of net magnetization has been observed in La3+ and Ba2+-doped samples

(i.e., under diamagnetic doping) [80, 81]. It has been suggested that the sub-

stitution suppresses an incommensurate spin configuration and causes a small

spin canting [82].

In addition to studies on pure BFO, doping of BFO with a foreign atom

at either 𝐴- or 𝐵-site of the perovskite (ABO3) structured BFO lattice, has

been shown to play an important role in altering its properties. For exam-

ple, substitution of Bi, the bigger 𝐴-site cation, with lanthanide elements such

as lanthanum (La), neodymium (Nd) has been shown to result in remark-

able improvement of the properties of BFO thin films although results varied

from one group to another. On the other hand, Fe, smaller 𝐵-site cation,

has been substituted by elements such as Cr which resulted in an increase

in the polarization but also led to substantial increase in the coercive field.

By using smaller Nd3+ ions (radius=0.983A) in substitution for larger Bi3+

ions (radius=1.03A) in the BFO composition, it is possible to create single-

phase multiferroic Bi1−𝑥Nd𝑥FeO3 (𝑥=0-0.15) ceramics possessing an improved

spatial uniformity of magnetic structure, and a rather linear ferromagneto-

electric behavior [78, 83]. Compared with La3+ ion (radius=1.032A), Sm3+

ion (radius=0.958A) possesses a much smaller radius. This suggests that the

effect of 𝐴-site substitution may further be enhanced if Sm3+ ion is used in

substitution for Bi3+ ion (radius=1.030A) in the ordinary BFO composition.

In Bi1−𝑥Sm𝑥FeO3 system, residual magnetization (M𝑟) may be enhanced by

increasing the 𝐴-site Sm substitution. These modifications in the functional

properties are expected to occur due to the changes in the electronic or crys-

tal structure of BFO, as reported previously for bulk BFO [85]. Nakamura et

al. [84] studied the effect of Co-doping, using a reasonably large range (up to

30 atom%) as compared to previous reports [85–88], on the structural prop-

erties of BFO thin films on LaAlO3 (001) substrate synthesized by chemical

solution deposition and here they report on the structural changes observed

in the BFO thin films induced by Co doping on 𝐵-site which also affected the

electrical and magnetic behavior of these films. To minimize the substrate

induced texture effects, we employed SrRuO3 for growing epitaxial thin films

of BFO which also serves as a base electrode.

20

1.4 Heusler Alloys

The Heusler compounds have a very old story regarding magnetism, starting

more than 100 years ago with the invention of the ternary metallic compound

Cu2MnAl by A. Heusler in 1903 [89]. Remarkably, without none of its elements

is ferromagnetic even though this alloy is a ferromagnet. Further investigations

showed that the general composition 𝑋2𝑌 𝑍 exists as a class of isostructural

ternary metallic alloys, where 𝑋 denotes a transition metal element such as

Ni, Co, Fe or Pt, 𝑌 is a second transition metal element, e.g., Mn, Cr or Ti

and 𝑍 is an atom from 3rd, 4th or 5th group of the periodic table such as Al,

Ge, Sn or Sb. More than thousands different Heusler compounds have been

synthesized until now, a widespread evaluation of experimental work until the

year 1987 can be found in Ref [90]. Due to their very versatile magnetism

Heusler compounds have attracted considerable amount of interest. Actually

the Heusler compound is the predicted half-metallic ferromagnetic nature for

some of these alloys which is the driving force for the intense study of these

alloys [91–95]. Half-metals can be considered as hybrids between metals and

semiconductors since we know that Half-metals are ferromagnetic materials.

The Heusler compounds possess a wide energy gap for minority spin direction

at the Fermi level (𝐸𝐹 ) which give a complete spin polarization at 𝐸𝐹 . This

makes them ideal candidates for applications in spintronics.

This chapter gives a short review of structural, magnetic and elec-

tronic properties of the Heusler compounds in general with emphasis on the

half-metallic Heusler alloys, especially the Co-based ones, since Co2MnSi and

Co2MnGe are the alloys studied in this thesis. After the basic properties of

the bulk Heusler compounds are discussed, some of the recent results and open

problems are presented.

1.4.1 Physical properties of Heusler alloys

1.4.1.1 Structural Properties of Heusler alloys

Heusler alloys are ternary intermetallic compounds. At the stoichiometric

composition, the full Heusler alloys 𝑋2𝑌 𝑍 and the half Heusler 𝑋𝑌 𝑍 struc-

tures are 𝐿21 and 𝐶1𝑏, respectively. The 𝑋 and 𝑌 elements are magnetic

elements; the atom 𝑍 is non-magnetic element. The unit cell have four inter-

penetrating fcc sublattices at (000) and (1/2, 1/2, 1/2) for 𝑋, (1/4, 1/4, 1/4)

for 𝑌 and (3/4,3/4,3/4) for 𝑍 atom. For half-Heusler compounds the site (1/2,

1/2, 1/2) is empty. The two structures are closely related with vacant sites.

The 𝐶1𝑏 structure can be obtained from the 𝐿21 one by replacing the half of

the 𝑋 sites in an ordered manner as shown in Fig 1.12.

21

Figure 1.12: Crystal structures of Heusler alloys (a) 𝐿21 full-Heusler and (b)

𝐶1𝑏 half-Heusler ordered structures. The structure consists of 4 interpene-

trating fcc lattices. One of the four sub lattices is empty in the case of the

half-Heusler alloys. Disordered Heusler phases: (c) 𝐵2 disorder due to the

𝑌 -𝑍 exchange and (d) 𝐴2 disorder caused by the 𝑋-𝑍 or 𝑋-𝑌 intermixing.

One notices that if all atoms are identical, the lattice is simply bcc.

The Heusler compounds are the low temperature equilibrium phase for

the ordered 𝐿21 structure. In the majority of Heusler alloy like the original

Heusler phase Cu2MnAl there exist several structural modifications with dif-

ferent degree of site disorder of the atoms on the 𝑋, 𝑌 and 𝑍 sites. At high

temperatures the crystal structure of Heusler compound is bcc with random

occupation of the atoms on the lattice sites of a simple bcc lattice with half

of the lattice parameter compared to the 𝐿21 structure. For the Heusler com-

pounds, an intermediate structure with 𝐵2 symmetry often occurs. 𝐵2 has

the same lattice parameter as the 𝐿21 phase, but the distribution of 𝑌 and

𝑍 atoms are random on the corresponding sublattices, whereas the two 𝑋

sublattices remain intact.

22

1.4.1.2 Magnetic properties of Heusler alloys

Heusler alloys possess attractive magnetic properties because it is very in-

teresting materials. Anyone can study in the same family of alloys a series

of interesting diverse magnetic phenomena like itinerant and localized mag-

netism, antiferromagnetism, heavy-fermion behavior, helimagnetism and Pauli

paramagnetism [96–99].

The majority of Heusler alloys with a magnetic element at the 𝑌 site

order ferromagnetically, on the other hand, a number of antiferromagnetic

compounds also exist, e.g. Ni2MnAl or Pd2MnAl [100, 101]. The major in-

put to the magnetic moments in the Heusler phases usually stems from the

atoms at the 𝑌 site. If the 𝑋 sites are also occupy by the magnetic atoms,

their moment is usually quite small or even vanishing. For example the Ni

atoms are non-magnetic as mentioned in the above Ni2MnAl compound. A

few Heusler compound with rather large magnetic moments both on the 𝑋

and the 𝑌 sites are also exist. For such case the ferromagnetic Curie tem-

perature 𝑇C becomes exceptionally high and the ferromagnetic state is very

stable. The best examples which show the Co moment of about 1 𝜇𝐵 and the

Curie temperature of 985 K [102] and 1100 K [103] are provided by the Heusler

phases Co2MnSi and Co2FeSi, respectively, the highest 𝑇C values known for the

Heusler alloys. The mechanism which stabilizes the ferromagnetism is a strong

next-nearest neighbor ferromagnetic exchange interaction between the spins at

the 𝑋 and the 𝑌 site [91, 104]. If a non-magnetic element is at 𝑋 site, the

leading exchange interaction between the 𝑌 spins is of weaker super-exchange

type due to hybridization, mediated by the electrons of the non-magnetic 𝑍

atoms. Depending on the valence of 𝑍 the magnetic interaction can have either

sign [104].

Table 1.4: From generalized gradient approximation (GGA) lattice constant,

total spin moment and predicted spin magnetic moments of Co2MnGe are

calculated and their experiment values, taken from Ref. [94].

a[nm] 𝜇tot[𝜇𝐵] 𝜇Mn[𝜇𝐵] 𝜇Co[𝜇𝐵] 𝜇Ge[𝜇𝐵]

experiment 0.574 [101] 5.11 - - -

theory 0.574 5.0 2.98 1.02 -0.03

In Table 1.4, the spin magnetic moments for Co2MnGe are listed. It

possess a spin moment of ∼ 1.0 𝜇𝐵 because the Co atoms are ferromagnetically

coupled to the Mn spin moments and. The Ge atoms have orbital hybridization

which is two orders of magnitude smaller than the Co moment and hence it

has a very small negative moment. The orbital moments absolute values are

23

negligible with respect to the spin magnetic moments [94, 105] because they

are completely quenched.

Heusler compounds such as Cu2MnAl with a magnetic moment only on

the 𝑌 site are considered as good examples of localized 3𝑑 metallic magnetism.

Since there are no Mn-Mn nearest neighbors in the ideal 𝐿21 structure, the

magnetic moments remain essentially localized at the Mn position and the

Mn 3𝑑 wave functions overlap only weakly. However, the magnetic moments

only on the 𝑋 sites for the compound Co2TiSn exhibits weak itinerant ferro-

magnetism with strongly delocalized magnetic moments [106]. As it is clear

from the crystallographic structure as shown in Fig 1.12, the overlapping of

the nearest neighbor X atoms are making of the 3𝑑 wave functions and the

delocalized character of the 𝑑 electrons much larger than in the case of only

the atoms at the 𝑌 site being magnetic. By replacing the Co atoms to Ni

in Co2TiSn, this delocalization effect proceeds further, making the compound

Ni2TiSn a Pauli paramagnet [107].

1.4.1.3 Electronic properties of Heusler alloys

By electron energy-band calculations the spectacular property of full spin

polarization at the Fermi level 𝐸𝐹 was first predicted in 1983 for the NiMnSb

[108]. It have been predicted that PtMnSb and CoMnSb posses this property.

NiMnSb, PtMnSb and CoMnSb have been half-metals [108], since only for ma-

jority spin there is metallic conductivity and for the minority spin the conduc-

tivity is of semiconducting type. In a ferromagnetic transition metal alloys, this

half-metallicity is a very rare property, since usually 𝑠 or 𝑝 bands with a small

exchange splitting cross the Fermi energy and contribute states of both spin

directions. The half-Heusler alloys PtMnSb, NiMnSb and CoMnSb remained

the only ferromagnetic alloys with half-metallic character for several years.

The half-metallic Heusler alloys, Co2MnSi, Co2MnGe and Co2Mn(Sb𝑥Sn1−𝑥)

was found theoretically in the starting of 1990 [93, 109, 110]. The calculated

indirect band gap for the minority carriers is smaller in these materials than

in the half-Heusler compounds [105], for Co2MnSi and Co2MnGe one derives

𝐸gap=0.81eV and 𝐸gap=0.54 eV, respectively. This spin-projected density of

states for Co2MnGe is shown in Picozzi et al. [94].

The origin of the gap in the minority spin band is quite subtle, how-

ever recently band-structure calculations allowed to disclose the fundamental

mechanism for the formation of the gap. The 𝑑− 𝑑 hybridization between the

transition atoms composing Heusler alloys is essential for the formation of the

gap at 𝐸𝐹 . The gap in the case of half-Heusler compounds (e.g., NiMnSb) is

created by the hybridization and bonding-antibonding splitting between the

Mn 𝑑 and the Ni 𝑑 states. However, the gap in the case of full Heusler al-

24

loys (e.g., Co2MnGe) originates from the hybridization between the 𝑑 states

of the two Co atoms and subsequent interaction of these hybrids with the Mn

𝑑 states [91].

The experimental proof of these Heusler alloys is a long and still on-

going controversial issue for the half-metallicity. The electron transport mea-

surement to test the existence of a gap in the spin down electron band was

the first attempts to prove the half-metallicity [111, 112]. Since in the half-

metal for temperatures small compared to the gap in the minority spin band

there is only one spin direction at the Fermi level available, it is expected

that electronic scattering processes involving spin flips and longitudinal spin

wave excitations are inhibited. Thus one should suppose increasing electron

mobility and a change of power law describing the temperature dependence

of the resistivity, when the gap for the minority spin band becomes larger

than the thermal energy. Actually such type of behavior for temperatures

below 80 K has been detected for NiMnSb compound. Additionally the Hall

coefficient shows an anomalous temperature dependence in this temperature

range, strongly suggesting across a gap a thermal excitation of charge car-

riers coexisting with metallic conductivity [111, 112]. Finally, by analyzing

the current-voltage (𝐼 − 𝑉 ) characteristic below the superconducting gap of

a point contact between a Nb superconductor and a bulk PtMnSb sample,

which is dominated by Andreev reflections at the ferromagnet/superconductor

interface, spin polarization of 90% at the Fermi level derived [113].

The spin polarized neutron diffraction measurements on the Co2Mn(Si,

Ge, Sn) samples have been employed to determine the degree of spin polariza-

tion at the Fermi level [114]. This methods probe the spatial distribution of

the magnetization and so it depend sensitively on the spin polarization. The

results imply a finite density of states in the minority spin 𝑑 band of man-

ganese. Hence the spin polarization approaches to larger value, but not 100%.

Recently superconducting/ferromagnetic measurements on a Co2MnSi single

crystal gave a spin polarization of 55% [115]. Similarly, the degree of spin

polarization determined from spin resolved photoemission spectra and it was

always found to be definitely below 100% [116,117].

During the first years after the discovery of the half-metallic character

in the Heusler compounds were considered as exotic ferromagnets for mainly

academic interest. With the development of new ideas of data storage and pro-

cessing designed the attitude have changed completely to use both the charge

and the spin degree of freedom of the conduction electrons, currently call spin-

tronics [118–120]. The non-volatility, increased processing speed and decreased

electric power consumption these are advantages by adding the spin degree of

freedom to conventional electronic devices in these alloys [118, 121]. There is

25

a strong belief in the spintronics community that in future these new concepts

have the perspective to complement or even substitute conventional Si tech-

nology. How valuable it would be for spintronic devices was rapidly realized by

spintronics community to have a ferromagnet available with only one conduc-

tion electron spin direction at the Fermi level. With an electrode possessing

100% spin polarization, the generation of a fully polarized current for spin

injection into semiconductors would be possible [122] and in metallic thin film

systems spin filtering and spin accumulation would be most effective [123]. The

giant magnetoresistance (GMR) [124] as well as the tunneling magnetoresis-

tance (TMR) [125] of a device prepared of two half-metallic electrodes should

be high, since electrical current for one spin direction is totally blocked to first

order, if the two electrodes have opposite magnetization directions.

The novel concepts of spintronics started an upsurge of interest in fer-

romagnetic half-metals in the literature. In addition the half and full Heusler

alloys, there are some binary oxides (e.g., CrO2 and Fe3O4) [113], manganites

(e.g. La0.7Sr0.3MnO3) [113], transition metal chalcogenides (e.g., CrSe) and

pnictides (e.g., CrAS) in the zinc-blende or wurtzite structures [126], diluted

magnetic semiconductors (e.g. GaAs and ZnO doped with magnetic transition

metal ions) [127,128], and the fully spin polarized Heusler alloys.

For several technical reasons and/ or applications the Heusler alloys

seems to be very attractive. Their crystal structure and lattice constant are

closely related to the diamond and zinc-blende structures of most industri-

ally relevant semiconductors and the lattice mismatch is low, for instance for

Co2MnSi with GaAs it is less than 0.4% [129]. The preparation of Heusler

thin films are compatible with current semiconductor technology and can be

carried out by conventional metal film preparation methods. An additional

advantage of these alloys is their high ferromagnetic Curie temperatures, even

at 300 K the half-metallic Heusler alloys are close to ferromagnetic satura-

tion. This is of particular importance, since the temperature dependence of

the spin polarization scales with the corresponding magnetic moment of the

material [130].

Recent deep theoretical investigations using electronic energy band struc-

ture calculations increased the number of Heusler compounds with predicted

half-metallic properties to more than 20, among them Co2CrAl, Fe2MnSi,

Co2CrAl, Co2Cr0.6Fe0.4Al, and Co2FeSi, to mention a few of the new com-

pounds [91, 92]. The experimental work in the literature mostly has been in-

tensed on the classical Heusler half-metallic phases NiMnSb and PtMnSb, the

alloys Co2MnSi and Co2MnGe and the recently exposed compound Co2(CrFe)Al

[131], leaving a huge field for advance experimental investigations.

However, as already stated above the half-metallicity of the Heusler

26

compounds is a subtle property which is easily lost in a real sample.

1.4.2 Heusler half-metals in devices

Its promising application in spintronic devices is the main motivation be-

hind the experimental research on the fully spin polarized Heusler compounds.

Among the most promising materials to be integrated into technologically rel-

evant magnetic tunnel junctions (MTJs) as magnetic electrodes are the half-

metallic Heusler alloys. The effective amplitude of the magnetoresistance in a

simple layered system, consisting of two ferromagnetic metals, separated by a

thin insulating layer, which serves as a tunnel barrier [133], can be expressed

as:

TMR= 𝑅𝐴-𝑅𝑃/𝑅𝑅= 2𝑃1𝑃2/1-𝑃1𝑃2,

where 𝑅𝐴 and 𝑅𝑃 represent the resistance of the two ferromagnetic layers

with their magnetizations aligned antiparallel or parallel to each others and

𝑃1 and 𝑃2 are the spin polarizations at the Fermi level for electrodes 1 and 2,

respectively. Thus, by employing high spin polarized ferromagnetic electrodes

on both side of MTJ, high TMR values can be achieved.

Thin films of Heusler compounds are obviously needed but bulk samples

are not very useful. However, thin film preparation in general, especially thin

film heterostructures preparation, often imposes limits on the process param-

eters and this might severely interfere with the needs to have a high degree of

spin polarization. In order to obtain a large spin polarization, it is important to

have a perfect crystal structure with a small number of grain boundaries. This

can be best achieved by keeping the substrate at high temperature during the

thin film deposition. However, at high temperatures, most Heusler phases grow

in the Vollmer-Weber mode (three-dimensional islands), thus there might be a

strong roughening of the surfaces when using high preparation temperatures,

which for spintronic devices should be avoided. Furthermore, high preparation

temperatures are forbidden in thin film heterostructures combining different

metallic, semiconducting or insulating layers with the Heusler compounds to

avoid excessive inter-diffusion at the interfaces.

TMR devices and thin films was first investigated systematically using

the half-Heusler compounds PtMnSb and NiMnSb [133, 134]. The result was,

however, not encouraging. The spin polarization of NiMnSb integrated in a

MTJ was measured to be 25% at 4.2 K corresponding to a TMR amplitude

of 19.5% [134]. At room temperature the TMR value was 9%. Later, it was

found that Co2MnSi- and Co2MnGe-based TMR devices have shown much

better performance. For tunnel junctions of textured Co2MnSi with an Al

oxide tunnel barrier, a maximum TMR ratio of 108% at 20 K and 33% at room

27

temperature was achieved, corresponding to 72% and 41% of spin polarization,

respectively [135–137]. Recently, Yamamoto et al. [138] used high quality

epitaxially grown Co2MnSi electrodes and they were able to obtain a further

improvement to 1135% (92% spin polarization) at 4.2 K and 236% at room-

temperature. This is actually the highest TMR value observed for junction

using a MgO tunnel barrier. MTJs using a Co2MnGe electrode were recently

developed by the same group. The epitaxial tunnel junctions using MgO as

tunneling barrier showed strongly temperature dependent characteristics with

TMR ratios of 220% at room temperature and 650% at 4.2 K. For the newly

predicted half-metal Co2Cr0.6Fe0.4Al as a magnetic electrode using a MgO

tunneling barrier, a TMR ratio of 74% at 55 K was found [139–141]. The

maximum TMR value obtained for an Al oxide barrier are 52% and 83% at

5 K [142]. The spin polarization was found to be 81% [143]. Among the

MTJs with an amorphous Al oxide tunnel barrier, MTJ with the 𝐵2 ordered

Co2MnAl obtained a large TMR ratio of 83% at 2 K suggesting that one 𝐵2

ordered Co2MnAl may exhibit a high spin polarization [144]. In junctions

using a Co2FeSi electrode, smaller values for the TMR ratio were obtained, i.e.

41% at room temperature and 60% at 5 K.

Regarding the GMR effect in antiferromagnnetically coupled metallic

multilayers or in spin valves consisting of Heusler compounds, only a few ex-

periments have been performed. Current-in-plane (CIP) GMR effect at room

temperature has been measured in a [Co2MnGe/ Rh2CuSn]10 multilayer and

was found to be very small, giving a value of only 0.26%. The situation is even

worse for spin value structures employing the same material combination. This

result is in good agreement with GMR measurements on [Co2MnGe/V]𝑁 mul-

tilayers [105]. The GMR values are far below the values obtained in transition

metal multilayer system, which can be as large as 150% at room temperature

theoretically [145].

To date, efficient electrical spin injection into semiconductor has been

demonstrated only from magnetic semiconductors [146, 147] and conventional

ferromagnetic metals [148,149]. In principle, full spin polarized Heusler alloys

are ideal candidates for epitaxial contacts. In addition they are an alterna-

tive solution to the conductivity mismatch [122]. Results from spin injection

experiments from the epitaxially grown half-Heusler NiMnSb into a spin LED

have shown injected spin polarization up to 2.2% at 80 K. However, this is

rather not a good thing, since even a MnSb reference injector works better.

Using the alloy Co2MnGe as spin injector gives a more encouraging results.

The injected spin polarization at 2 K is calculated to be 27% [150].

28

1.4.3 Heusler alloys: disorder and interfaces

Heusler-based magnetic alloys are promising candidates as discussed above

in TMR results. The experimentally determined spin polarization of TMR

results is always smaller than theoretical predictions. This occurrence leads

to the doubt that at least for a few monolayers at the interfaces the full spin

polarization is lost.

For spintronic devices, a very delicate problem encounter at the inter-

faces of the Heusler compounds with other materials. The spin polarization

of the first few monolayers at the interfaces is of extreme importance for spin

injection into semiconductors or a tunneling magnetoresistance. The large

spin polarization in the bulk of a Heusler compound does not assurance of

a good spintronic material, unless it keeps its spin polarization down to the

interface. Hence in order to reach high spin polarization we have to overcome

all problems in real devices. The perfect 𝐿21 point symmetry is disturbed by

site disorder within the sublattices of the Heusler compounds and this may

cause to destroy the half-metallicity. An important question is, which type of

disorder is most unfavorable for the spin polarization. Therefore the effects of

numerous type of defects in Heusler alloys Co2MnSi and Co2MnGe have been

studied by theoretical model calculations [92]. According to experiment, the

most common defects are: (1) Mn antisites where a Co atom is replaced by

Mn, (2) Co antisites where a Mn atom is replaced by Co, and (3) Co-Mn ex-

change where a Mn-Co nearest neighbor pair shows exchanged sites compared

to the ideal bulk.

It is of special significance to restore the half-metallic ferromagnetism

(HMF) behavior at the interfaces with an insulator or semiconductor for high-

performance spintronic devices. The loss of half-metallicity were first car-

ried out for NiMnSb/ semiconductor interfaces [151, 152], except in the case

of NiMnSb/CdS by theoretical model calculations. Further calculations re-

vealed the presence of interface states at almost all Heusler/semiconductor

contacts [153, 154]. Here for a few atomic layers close to the interface, the

half-metallicity is destroyed and completely restored far away from it.

The half-metallic materials suffer from a tendency of the surface to

adopt a different composition than the bulk except the fundamental problem

of surface states. Segregation occurs in order to minimize the surface energy

and this surface segregation is determined by a difference in the free energy

[156]. Experimentally and theoretically both for the case of the half-Heusler

compounds, it has been proven that segregation, too, has the tendency to

destroy the half-metallic behavior [105,155].

From the above discussion, It is clear that not only the bulk but the

surface/interface properties have to be taken into account for device design.

29

A controlled surface and interface engineering is required. However, the use

of Heusler alloys in device applications is still promising; if material combi-

nations can be found that preserve the half-metallicity even at the interface.

Theoretical calculations have already verified that this is possible in the case

of the NiMnSb(111) surface [151]. To summary of the introductory part, there

are challenges and difficulties, but the gate to Heusler based spintronic devices

has been opened.

30

Chapter 2

Experimental Methods and

Principles

2.1 Principles of x-ray magnetic circular dichro-

ism and sum rules

2.1.1 X-ray absorption spectroscopy and x-ray magnetic

circular dichroism

The measurements of photo-absorption by excitation of a core-level electron

into unoccupied states as a function of photon energy is called x-ray absorption

spectroscopy. The photo-absorption intensity is given by

𝐼(ℎ𝜈) =∑𝑓

∣⟨𝑓 ∣𝑇 ∣𝑖⟩∣2𝛿(𝐸𝑖 − 𝐸𝑓 − ℎ𝜈), (2.1)

where 𝑇 is the dipole transition operator. In the 3𝑑 transition-metal com-

pounds, transition-metal 2𝑝 XAS spectra reflect the 3𝑑 states such as the

valence, the spin state and the crystal-field splitting.

There are two measurement modes for X-ray absorption spectroscopy

(XAS), the transmission mode and the yields mode. In the transmission mode,

the intensity of x-ray is measured in front of and behind the sample and the

ratio of the transmitted x-ray is determined. The transmission mode is stan-

dard for hard x-rays, while, for soft x-rays, the transmission mode is difficult

to perform because of the strong interaction of soft x-rays with the sample.

An alternative to the transmission-mode experiments has been provided

by measuring the decay products of the core hole. The core hole gives rise

to an avalanche of electrons, photons, and ions escaping from the surface of

the sample. This is the yield-mode experiment and is standard for soft x-

rays. The yield mode can be classified into the Auge electron yield, the total

31

fluorescence yield , the ion yield and the total electron yield (TEY). The total

fluorescence yield (TFY) mode suffers from self-absorption because of its long

probing depth and data analysis may become complicated.

In the Auger electron yield mode, one detects Auger electrons of a specific

Auger decay channel of the core hole. The mean free path of 500 eV electron

is of the order of 20 A. Since the mean free path of photon is of the order of

1000A, the Auger electron yield in the soft x-ray range effectively surveys the

region of about 20Adepth from the surface.

Instead of Auger decay, the fluorescence decay is also used for absorption

measurements. Because the fluorescence yield mode has a large detection depth

(>1000A), it is particularly suited for the studies of bulk electronic structure.

When the absorption process takes place at the surface, the atoms which

absorbs the x-ray can be ionized by Auger decay and can escape from the

surface. Detecting the escaping ions as a function of x-ray energy, one can

obtain the signals related to the absorption cross section. This is the ion yield

mode. This is a highly surface sensitive method, whose detection depth is of

the order of 2A.

The TEY mode is the most widely used yield detection technique because

of the easy of detection and the large signal. The difference from the Auger

electron yield mode is that the energy of the outgoing electrons are not selected

and simply the all escaping electrons are counted. Estimated detection depth

of the TEY mode using x-rays for transition metal oxide is about 40 A. In the

present work, we have employed the TFY and the TEY mode.

When the relativistic electrons in the storage ring are deflected by the

bending magnets that keep them in a closed circular orbit, they emit highly

intense beams of linearly polarized x-rays in the plane of the electron orbit

(bremsstrahlung) but they emit circularly elliptically polarized light out of

the plane. Currently, an number of alternative sources for circularly polarized

synchrotron radiation are under development. The most notable are so-called

insertion devices like helical wigglers [157] and crossed [158] undulator, which

are complex arrays of magnets with which the electrons in a storage ring are

made to oscillate in two directions perpendicular to their propagation direction,

with the result that they emit circularly polarized light.

Using circularly polarized x-rays in XAS, x-ray magnetic circular dichro-

ism (XMCD) is defined as the difference in absorption spectra between right-

handed and left-handed circularly polarized x-rays when the helicity of the

x-rays are parallel and antiparallel to the magnetization direction of the mag-

netic materials such as ferromagnet or ferrimagnet. XMCD is sensitive to

magnetic polarization, and therefore enable us to study the magnetic proper-

ties of particular orbitals on each element.

32

Figure 2.1: Schematic diagram of x-ray magnetic circular dichroism (XMCD).

(a) Experimental set up for XMCD measurements. (b) Transition probability

of 2𝑝 → 3𝑑 absorption with circularly polarized x rays for less-than-half filled

3𝑑 electronic configuration. (c) Circularly polarized x-ray absorption spectra.

2.1.2 XMCD sum rules

XMCD reflects the spin and orbital polarization of local electronic states.

Using integrated intensity of the 𝐿2,3-edge XAS and XMCD spectra of a

transition-metal atom, one can separately estimate the orbital [159] and spin

[160] magnetic moments by applying XMCD sum rules given by

𝑀orb = −4∫𝐿3+𝐿2

(𝜇+ − 𝜇−)𝑑𝜔3∫𝐿3+𝐿2

(𝜇+ + 𝜇−)𝑑𝜔(10−𝑁𝑑), (2.2)

𝑀spin + 7𝑀T = −6∫𝐿3(𝜇+ − 𝜇−)𝑑𝜔 − 4

∫𝐿3+𝐿2

(𝜇+ − 𝜇−)𝑑𝜔∫𝐿3+𝐿2

(𝜇+ + 𝜇−)𝑑𝜔(10−𝑁𝑑), (2.3)

where 𝑀orb and 𝑀spin are the spin and orbital magnetic moments in units

of 𝜇𝐵/atom, respectively, 𝜇+(𝜇−) is the absorption intensity for the positive

(negative) helicity, 𝑁𝑑 is the 𝑑 electron occupation number of the specific

transition-metal atom. The 𝐿3 and 𝐿2 denote the integration range. 𝑀T is

the expectation value of the magnetic dipole operator, which is small when

33

the local symmetry of the transition-metal atomic site is high and is neglected

here with respect to 𝑀spin.

2.1.3 Analysis of x-ray magnetic circular dichroism spec-

tra

First we normalized XAS spectra by incident-photon-flux and then for giving

information relating to the absorption edge of the element of interest, the

absorption spectra contain a background of absorption from the other elements

in the sample. Undesirable contaminants and oxides may add to the unwanted

background information. Other factors which may affect the spectra include

diminishing beam intensity (from the gradual decay of storage ring current),

and secondary electron events (such as scattering and vibrational processes).

Generally absorption from the other elements is only very weakly de-

pendent upon energy and may be accounted for by removing a linear function

and/or two-step-like function [161]. However, it is not unusual to encounter

spectra with nonlinear backgrounds. These can prove problematic, because to

fit a function to the background relies on some sense of aesthetic and is very

much subject to human judgment.

It was necessary to fit and remove a background signal from most of

the spectra before it was possible to perform integration, and thereby use sum

rules or determine the branching ratios. The background removal was done

with the greatest possible care, yet it should be noted that doing this can easily

and drastically change the calculated spin and orbital magnetic moments. A

reasonable fit is made to the background using a two-step-like function [161]

as shown in Fig. 2.2.

2.2 Experimental setup

2.2.1 NSRRC BL-11A

A Dragon beam line 11A at National Synchrotron Radiation Research Center

(NSRRC) has been constructed for photoemission, x-ray absorption, XMCD

and magnetic linear dichroism (MLD) measurements. A schematic diagram

of the beamline is shown in Fig 2.3. The light source is a bend magnet.

Synchrotron radiation with horizontal acceptance of 12 mrad was reflected

by horizontally and vertically focusing spherical mirrors (HFM and VFM)

to a grating (6m-GSM). The monochromatic light was reflected by toroidal

refocusing mirror (RFM), and introduced to the end station. Two vertical

plane mirrors (VPM) between the gratings and the exit slit to extend the

34

0.8

0.6

0.4

0.2

0XA

S in

ten

sity

(ar

b. u

nit

s)

(a)

120

100

80

60

40

20

0

600

400

200

0

(b)

XA

S in

ten

sity

(ar

b. u

nit

s)

Integ

ral of X

AS

r=69

6.46

-40

-30

-20

-10

0

10

810800790780770

-100

-80

-60

-40

-20

0

20(c)

XM

CD

inte

nsi

ty (

arb

. un

its)

Photon energy (eV)

Integ

ral of X

MC

D

p=-

114.

85

q=-

73.1

09

Figure 2.2: 𝐿2,3-edge XAS and XMCD spectra of Cobalt (a) XAS spectra

(solid black lines are linear background, blue line is two-step type background,

Red solid line spectrum is raw data and red dotted line spectrum is after

background subtraction) (b) and (c) are the XAS and XMCD spectra and their

integrations calculated from the spectra shown in (a). The doted line shown

in (c) is two-step-like function for edge-jump removal before integration. The

𝑝 and 𝑞 shown in (b) and 𝑟 shown in (d) are three integrals need in sum-rule

analysis.

35

lowest photon energy to 10 eV. In this beamline, six spherical gratings are used

to cover an energy range from 10 eV to 1700 eV. In practical measurements,

the photon energy was scanned using a grating which have 1200 lines/mm and

covers the photon energy range 400 − 1200 eV. Photon flux is 1 × 1010 with

the energy resolution 𝐸/Δ𝐸 = 10, 000. The degree of circular polarization

(𝑃𝑐) was evaluated to be 𝑃𝑐= ± 55% ± 5% on BL-11A of the NSRRC. The

measurement chamber is located at the end station of the beamline as shown

in Fig 2.3.

2.2.2 KEK-PF BL-16A

BL-16A at the Photon Factory (PF) has been constructed for XMCD mea-

surements. A schematic diagram of the beamline is shown in Fig 2.4. The

light source is a double-array undulator of APPLE-II type. The variation of

the phase between the two magnet arrays leads to change of the polarization

of the light. It generates left- and right-handed circular polarized light. This

undulator covers photon energy range of ℎ𝜈 = 0.3 to 1 keV in the circular

polarization mode by the first harmonic radiation. The photon energy was

scanned using a varied-line-spacing plane grating (VLSPG) grazing-incidence

monochromator (600 lines/mm). Photon flux is better than 1× 1011 with the

energy resolution 𝐸/Δ𝐸 = 8, 000. The degree of circular 𝑃𝑐 was evaluated to

be 𝑃𝑐= ± 95% ± 4% on BL-16A of the Photon Factory.

2.2.3 SPring-8 BL23SU

BL-23SU at SPring-8 1 has been designed and constructed for various spec-

troscopic studies on actinide compounds, semiconductor surfaces and biological

materials, etc., in the soft x-ray region [162–164]. A schematic diagram of the

beamline is shown in Fig 2.6. The light source is a double-array undulator of

APPLE-II type. The variation of the phase between the two magnet arrays

leads to change of the polarization of the light. It generates horizontally and

vertically linear, left- and right-handed circular polarized light. This undulator

covers photon energy range of ℎ𝜈 = 0.28 to 3 (0.5−3) keV in the linear (circu-

lar) polarization mode by the first harmonic radiation. The photon energy was

scanned using a varied-line-spacing plane grating (VLSPG) grazing-incidence

monochromator (600 lines/mm). Photon flux is better than 1× 1011 with the

energy resolution 𝐸/Δ𝐸 = 10, 000. The degree of circular 𝑃𝑐 was evaluated

to be 𝑃𝑐= ± 95% ± 4% on BL-16A of the Photon Factory. The XAS-MCD

station are located at the end station ST3 shown in Fig 2.5. At each station,

1The name of “SPring-8” comes from Super Photon ring for 8 GeV.

36

(a)

Figure 2.3: Measurement system in BL-11A at NSRRC. (a) Schematic layout of

beamline. (b) Experimental geometry of XMCD measurements. (c) Overview

of the measurement system at BL-11A.

37

Figure 2.4: Measurement system in BL-16A at PF-KEK. (a) Schematic lay-

out of beamline. (b) Experimental geometry of XMCD measurements. (c)

Overview of the measurement system at BL-16A.

preparation chamber for sample surface cleaning was connected to the mea-

surement chamber to enable transfer without breaking the ultra-high vacuum.

38

Figure 2.5: Measurement system in BL-23SU at SPring-8. (a) Schematic lay-

out of beamline [2.14]. (b) Experimental geometry of XMCD measurements.

(c) Overview of the measurement system at BL-23SU.

39

Chapter 3

X-ray magnetic circular

dichroism study of

ferromagnetic Ti1−𝑥Co𝑥O2−𝛿

thin films

3.1 Introduction

Semiconductors partially substituted with magnetic ions are called diluted

magnetic semiconductors (DMSs) are expected to be useful in spintronics

devices, where electron spins can be controlled by electric field and/or by

photons. Ferromagnetic DMS’s with Curie temperatures (𝑇C’s) higher than

room temperature are highly desirable for the development of spintronic de-

vices. To date, much work in this area has been done, mainly on II-VI

and III-V compounds doped with magnetic ions such as (Cd,Mn)Te [165]

and (Ga,Mn)As [166, 167], but their 𝑇C’s are far below room temperature.

Recently, room temperature ferromagnetism in Co-doped titanium dioxide

(TiO2) [168–171] has attracted much attention. Room-temperature ferromag-

netism was also reported in such materials as (Zn,Cr)Te [172], (Ga,Mn)N [173]

and (Al,Cr)N [174]. The near edge x-ray absorption fine structure study of

Co-doped TiO2 by Griffin et al. [175] claims that ferromagnetism is due to

𝑑-𝑑 double exchange mediated by tunneling of 𝑑 electrons within the impurity

band. Some studies that also claim the ferromagnetism of Co-doped TiO2 is

due to Co metal clusters [176–179]. The recent theoretical study by Calderon

et al. [180], electric field-induced anomalous Hall effect (AHE) study by Ya-

mada et al. [181] and x-ray photoemission spectroscopy study by Ohtsuki et

al. [182] suggested the ferromagnetism of Co-doped TiO2 is due to carrier me-

diated. However, direct information about the magnetization as a function

40

of carrier density has been lacking. Soft x-ray magnetic circular dichroism

(XMCD) at the Co 2𝑝 → 3𝑑 absorption (Co 𝐿2,3) edge is a powerful technique

to clarify this issue because it is an element-specific magnetic probe [179]. Our

previous XMCD study on rutile Co-doped TiO2 has revealed that the ferro-

magnetism is not due to segregated Co metal clusters but is due to Co2+ ions

in the TiO2 matrix [170]. However, the XMCD signal intensities were an order

of magnitude lower than that expected from the bulk magnetization [170].

In a more recent work [171], we performed x-ray absorption spectroscopy

(XAS) and XMCD studies on rutile Co-doped TiO2 not only by the surface-

sensitive total electron yield (TEY) mode but also the bulk-sensitive total

fluorescence yield (TFY) mode and found that Co ions in the bulk indeed have

a large magnetic moment of 0.8-2.2 𝜇𝐵/Co. In this work we have extended

the same approach to anatase Co-doped TiO2 and studied correlation between

magnetism and transport properties. Magnetization measurements of anatase

Ti1−𝑥Co𝑥O2−𝛿 thin films reveal ferromagnetic hysteresis behavior in the M-H

loop at room temperature with a saturation magnetization. In the bulk region

probed by the TFY mode, strong XMCD spectra with similar spectral line

shapes were obtained for all the samples. The magnetization and the XMCD

intensity increased with carrier density, consistent with the carrier-induced

origin of the ferromagnetism.

3.2 Experimental

Rutile Ti1−𝑥Co𝑥O2−𝛿 (101) epitaxial thin films with 𝑥 = 0.03, 0.05 and 0.10

were synthesized by the pulsed laser deposition method on r-Al2O3 (101) sub-

strates at 673K at different oxygen pressures, 𝑃O2= 10−6 or 10−7 Torr [183].

The samples fabricated in oxygen pressure 𝑃O2=10−6 and 10−7 Torr are named

as low-𝛿 and high-𝛿, respectively, since the number of oxygen vacancies in-

creases with decreasing oxygen pressure. The Fermi wavelength for high-𝛿 and

low-𝛿 is in the range of 1.06 nm and 3.46 nm. The carrier densities 𝑛𝑒 for high-𝛿

and low-𝛿 were in the range of 7 × 1021 to 2 × 1020 cm−3. Segregation of sec-

ondary phases were not observed under careful inspection by x-ray diffraction

(XRD), AFM, scanning electron microscopy (SEM), and transmission electron

microscopy (TEM). Its ferromagnetism at room temperature was confirmed by

Hall-effect measurements, magnetization measurements, and magnetic circular

dichroism (MCD) measurements in the visible region [169,183,184].

Anatase Ti1−𝑥Co𝑥O2−𝛿 (001) epitaxial thin films with 𝑥 = 0.05 were

synthesized by the pulsed laser deposition method on LaAlO3 (001) substrates

at 523 K and oxygen pressures (𝑃O2) of 5 × 10−7, 1 × 10−6 and 2 × 10−6

Torr. The resistivity increases in this order and these samples are hereafter

41

referred to metallic, intermediate, insulating samples, respectively. The carrier

densities 𝑛𝑒 were 4.1 × 1019, 1.1 × 1019 and 4.0 × 1018 cm−3, respectively and

the Fermi wavelength is 5.92 nm, 9.13 nm and 12.78 nm, respectialy. Reflection

high-energy electron diffraction was monitored during the in-situ growth. An

intensity oscillation was observed at the initial stage of the growth. The surface

morphology of the resulting films of∼40 nm thickness observed by ex-situ AFM

showed atomically flat surfaces consisting of steps and terraces. Segregation

of secondary phases were not observed under careful inspections by XRD and

TEM [181]. Ferromagnetism at room temperature was confirmed by Hall-effect

measurements and magnetization measurements.

XAS and XMCD measurements were performed at the BL-11A beamline

of the National Synchrotron Radiation Research Center, Taiwan. In XMCD

measurements, magnetic fields (H) were applied parallel to the direction of

anatase (001) and for rutile it was applied to the sample along out-of-plane.

The monochromator resolution was E/ΔE>10000, the circular polarization of

x-rays was 55%. The base pressure of the chamber was about 10−9 Torr and

the sample temperature was maintained at 300 K. The samples were placed in

an ultra-high-vacuum (UHV) experimental chamber. XAS and XMCD spectra

were obtained in the TEY and TFY modes without surface treatment in order

to avoid possible destruction of the sample surfaces. The probing depths of

the TEY mode and TFY mode were ∼5 and 100 nm, respectively.

3.3 Results and Discussion

3.3.1 Results on rutile Co-doped TiO2

Figure 3.1(a)-(b) shows magnetic properties of Ti1−𝑥Co𝑥O2−𝛿 at 300 K with

different x and electron carrier density (𝑛𝑒). It is clear from Fig 3.1 that the

magnetization M(H) was in the range 1.0-1.5 𝜇𝐵/Co with coercive force around

several tens of Oersted, and increases with 𝛿 or 𝑛𝑒. In M(H) measurements,

magnetic field was applied to the sample along out-of-plane i.e. direction of

rutile (101). AHE measurements for Ti1−𝑥Co𝑥O2−𝛿 with different 𝑛𝑒 and x also

showed the same magnetic field dependences. The resultant magnetic “phase

diagram” shows that higher 𝑛𝑒 and x induce the ferromagnetic phase as shown

in Fig 3.1(b).

In Figure 3.2, we show the Co 𝐿2,3-edge XAS and XMCD spectra of

Ti1−𝑥Co𝑥O2−𝛿 (x=0.03, 0.05 and 0.10 with low- and high-𝛿) thin films taken in

the TEY mode. In the figure, 𝜇+ and 𝜇− stand for the absorption coefficients

for photon helicity parallel and antiparallel to the Co majority spin direction,

respectively. The XMCD spectra (Δ𝜇= 𝜇+ - 𝜇−) have been corrected for the

42

-2

-1

0

1

2

M [m

B/C

o]

-20 -10 0 10 20m0H [x103 Oe]

x = 0.03, high-δx = 0.05, low-δx = 0.05, high-δx = 0.10, high-δ

(a) T=300K

1018

1019

1020

1021

1022

1023

Ele

ctro

n d

ensi

ty (

cm-3

)0.100.080.060.040.020

x in Ti1-xCoxO2-d

Ferromagnetism

Paramagnetic

(b) T=300K

-0.03

0.00

0.03

-20 0 20

M/M

' [2T

]

µ0H (mT)

Figure 3.1: (Color online)(a) Magnetization vs. magnetic field curves of rutile-

type Ti1−𝑥Co𝑥O2−𝛿. Inset shows M (H) curve around zero magnetic field at

300 K [171]. (b) Magnetic phase diagram as a function of electron carrier

density 𝑛𝑒 and Co concentration deduced from ordinary Hall effect at 300 K

for Ti1−𝑥Co𝑥O2−𝛿. Solid and open symbols denote ferromagnetic and param-

agnetic samples, respectively. Circle, square, triangle and diamond symbols

correspond to 𝑃O2=10−7, 10−6, 10−5 and 10−4 Torr, respectively, during syn-

thesis [169].

degree of circular polarization of the incident light. The XAS spectra of the

rutile-type Ti1−𝑥Co𝑥O2−𝛿 thin films showed multiplet features. Here, we follow

Mamiya et al. [170]. The XMCD spectra show clear multiplet features that

correspond almost one-to-one to those in the XAS spectra. The line shapes of

the XAS and XMCD spectra are almost the same as Mamiya et al. [170]. The

estimated magnetic moment is in the range of the 0.15-0.24 𝜇𝐵/Co consistent

with Mamiya et al. [170], while the saturation moments deduced from the

SQUID magnetization measurements is 1.0-1.2 𝜇𝐵/Co. In contrast to Kim

et al. [179], the present experiment clearly revealed multiplet features in the

XMCD spectra corresponding to those in XAS without annealing, consistent

with the ferromagnetism arising from Co2+ ions which are coordinated by

O2− ions [170]. The experimental data, i.e., XAS and XMCD spectra, show

qualitatively the good agreement with the calculated spectra for the Co2+

high-spin configuration in the 𝐷2ℎ crystal field [170].

Figure 3.3 show the results of the Co 𝐿2,3 XAS and XMCD spectra of

the same samples taken in the TFY mode. From the figure, it is clear that

43

400

300

200

100

0

800790780Photon Energy (eV)

µ+

µ−

x = 0.03, high-δ

x = 0.05, high-δ

x = 0.05, low-δ

x = 0.10, high-δ

(a) Ti1-xCoxO2-δ, Co L2,3 XAS TEY, 300K

Inte

nsi

ty (

arb

. un

its)

100

80

60

40

20

0

x = 0.05, high-δ TEY, 300K

(b) XAS

Inte

nsi

ty (

arb

. un

its)

µ+

µ−

-4

-3

-2

-1

0

784782780778Photon Energy (eV)

(c) XMCD

Figure 3.2: (Color online) (a) Co 𝐿2,3 XAS and (b)-(c) XAS and XMCD

spectra of rutile-type Ti1−𝑥Co𝑥O2−𝛿 samples for x=0.05, high-𝛿 taken in the

TEY mode at T = 300 K and H = 1 T.

the XMCD intensities are much higher than those taken in the TEY mode.

The large difference between the bulk-sensitive TFY and surface-sensitive TEY

modes indicates that there is a magnetically dead layer of ∼ 5 nm, at the sur-

faces of the samples, which is consistent with recent measurements of the film

thickness dependence of AHE [185]. The spectral line shapes of the XAS and

XMCD spectra of the x = 0.05, high-𝛿 sample shows a clear multiplet feature

while the other samples show relatively weak multiplet features. A possi-

ble origin of the weakness of the multiplet features of other samples might

be the limited S/N ratio. The fine structure, which is indicative of Co2+,

is more pronounced in the TEY than in the TFY modes. This is because

in this compound, there should be at least as many oxygen vacancies as Co

and electrostatic interaction from the oxygen vacancies may affect Co position

significantly [186] and hence we incorporated the random crystal field in our

calculations. In the surface region, which is observed by the TEY mode, the

44

400

300

200

100

0

800790780Photon Energy (eV)

x = 0.03, high-d

x = 0.05, low-d

x = 0.05, high-d

x = 0.10, high-d

µ+

µ−

(a) Ti1-xCoxO2-d, Co L2,3 XAS

TFY, 300K X

AS

Inte

nsi

ty (

arb

. un

its)

70

60

50

40

30

20

x = 0.05 , high-d

TFY, 300K (b) XAS

In

ten

sity

(ar

b. u

nit

s)

µ+

µ−

-10

-5

0

784782780778776Photon Energy (eV)

(c) XMCD

Figure 3.3: (Color online) (a) Co 𝐿2,3 XAS and (b)-(c) XAS and XMCD

spectra of rutile-type Ti1−𝑥Co𝑥O2−𝛿 samples for x=0.05, high-𝛿 taken in the

the TFY mode at T = 300 K and H = 1 T. (d),(e) Comparison of XAS and

XMCD spectra shown in (b) and (c) with cluster-model calculation [187]

45

position of Co atoms may be optimized in a similar way because of the oxida-

tion and less structural constraint at the surface, resulting in relatively uniform

crystal field. On the other hand, in the bulk, probed by the TFY mode, the

position of Co atoms might be frozen in various local structures. The random

crystal fields in the 3D crystal lattice make the TFY spectra broad and differ-

ent from the TEY spectra. The XAS and XMCD spectra taken in the TFY

mode also show good agreement with the calculated spectra for the Co2+ in

bulk Ti1−𝑥Co𝑥O2 with random crystal fields [187].The parameters used in the

calculations listed in Table 3.1.

3.0

2.5

2.0

1.5

1.0

0.5

0

Cluster model

SQUID

TFY

x = 0.03, high-δ

TEY

(a)

Mag

net

izat

ion

(µ B

/ C

o)

T=300K

Cluster model

SQUID

TFY

x = 0.05, low-δ

TEY

(b)T=300K

3.0

2.5

2.0

1.5

1.0

0.5

01.00.80.60.40.20

Cluster model

TFY

SQUID

x = 0.05, high-δ

TEY

(c)

Magnetic field (T)

T=300K

1.00.80.60.40.20

Cluster model

x = 0.10, high-δ

SQUID

TFY

TEY

(d)

T=300K

Figure 3.4: (Color online) M-H relation of SQUID magnetization and magne-

tization estimated from the XMCD spectra Ti1−𝑥Co𝑥O2−𝛿.

Figure 3.4 shows the magnetization estimated from the XMCD spectra

46

Table 3.1: Electronic structure parameters for rutile Co-doped TiO2 thin film

used in the cluster-model calculations in units of eV : the charge-transfer energy

Δ, the on-site 3𝑑-3𝑑 Coulomb energy 𝑈𝑑𝑑, and the 3𝑑-2𝑝 Coulomb energy 𝑈𝑑𝑐

on the Co ion, the hopping integral between Co 3𝑑 and O 2𝑝 𝑉E𝑔 , and the

crystal-field 10Dq.Crystal symmetry Δ 𝑈𝑑𝑑 𝑈𝑑𝑐 𝑉E𝑔 10Dq weight(%)

𝐷2ℎ low spin 4 5 7 1.1 1.1−1.2 38%

𝑂ℎ low spin 3 6 7.5 1.1 1.1−1.2 38%

𝑂ℎ high spin 2 5 7.5 1.1 0.8−0.9 24%

taken in the TFY and TEY modes using optical sum rules [188] compared it

with M-H curves from magnetization measurements as well as cluster model

calculations. For the validity of sum rule in this case we divided the obtained

spin-magnetic moment by a correction factor 0.92 [189]. The magnetic mo-

ment obtained from cluster model calculation is 1.48 𝜇𝐵/Co. Nevertheless,

the Co magnetic moment is found to be obviously much larger in the bulk

region than in the surface region. Since the TFY suffers from self-absorption

so the magnetic moment obtained by optical sum-rule is not so accurate like

TEY mode. The magnetization estimated from the XMCD spectra taken in

the bulk-sensitive TFY mode are similar to those estimated from the SQUID

measurements as well as cluster model calculations, which strongly suggest

that the Co ions in the bulk region are responsible for the ferromagnetism

while the surface layer of the film looks like magnetically dead layer, as con-

firmed by the XMCD taken in the surface-sensitive TEY mode. TiO2 has

an extraordinary chemical stability, hence we can rule out possible surface

degradation as a cause of decrease in surface magnetization. From surface

characterization techniques such as AFM and reflection high energy electron

diffraction, we have not observed any change in the surface state. Also, from

spectroscopic techniques, we have not observed a significant time dependence

of XMCD and x-ray photoemission spectroscopy. Figure 3.1(a) shows mag-

netic hysteresis of Ti1−𝑥Co𝑥O2−𝛿 with different x. The coercive force is so

small that the hysteresis is difficult to resolve by XMCD setup. The magnetic

anisotropy with out-of-plane easy axis is not so strong in Ti1−𝑥Co𝑥O2−𝛿. As

reported by Fukumura et al. [169] , even the out-of-plane anisotropy depends

on Co content and carrier density. Thus, it is difficult to draw unified explana-

tion of the magnetic anisotropy at present. The experimental absorption and

optical MCD spectra of the Co-doped anatase and rutile TiO2 [190, 191] are

similar to the spectra obtained from the first principle calculations [192]. The

first principles calculations are found to be qualitatively or even semiquantita-

47

tively consistent with the experimental absorption and MCD measurements,

indicating that the band-structure of the Co-doped TiO2 is similar to that of

the host TiO2 [192]. The experimental MCD signals of Co-doped anatase and

rutile TiO2 are well identified by theoretical calculation [192] and they do come

from the energy band structure of the system, giving further support on the

intrinsic ferromagnetism in the Co-doped TiO2 system.

Fig 1.1 of chapter 1 shows the phase diagram for (Ga,Mn)As. In the

simple estimation we found “Y-axis = Co-band exchange/Fermi energy” for

all rutile samples were in the range of 1.5 to 16.7. Since we know that effective

mass (𝑚★) of rutile is 20𝑚0, where 𝑚0 is mass of electrons [193] and Co-band

exchange is a constant which is of order 0.1 eV for 𝑛-type carriers [56]. The

“X-axis = Carrier concentration/Co concentration” for these samples were in

the range of 0.4 to 6. Since the a few values are out of the range of the phase

diagram for (Ga,Mn)As so it is necessary to predict phase diagram for rutile

Co-doped TiO2 in future.

48

3.3.2 Results on anatase Co-doped TiO2

2

1

0

-1

-2

-1.0 -0.5 0.0 0.5 1.0

Metallic Intermediate Insulating

M [m

B/C

o]

m0H [ T ]

Ti0.95Co0.05O2-δ

(a) Anatase

T= 300K

-1.0 -0.5 0.0 0.5 1.0

Ti0.95Co0.05O2-δ

(b) Rutile

m0H [ T ]

Metallic

T= 300K

Figure 3.5: (Color online) M-H curves of Ti0.95Co0.05O2−𝛿 at 300 K. (a) Metal-

lic, intermediate and insulating anatase samples [195]. (b) Metallic rutile sam-

ple.

Figure 3.5(a) shows the magnetization curves of anatase Ti1−𝑥Co𝑥O2−𝛿

(𝑥 = 0.05) at 300 K for various carrier densities (𝑛𝑒). The 𝑛𝑒 for metallic,

intermediate and insulating samples were 4.1 × 1019, 1.1 × 1019 and 4.0 ×1018 cm−3, respectively. That of metallic rutile thin films which has the carrier

density of 7 × 1021 cm−3 is also shown in Fig 3.5(b). The saturation mag-

netization of the anatase sample is 0.6-2.1 𝜇𝐵/Co with a coercive force of ∼100 to 200 Oe. In the M(H) measurements, magnetic field was applied parallel

to the the direction of anatase (001). Anomalous Hall-effect (AHE) measure-

ments for anatase Ti1−𝑥Co𝑥O2−𝛿 with various 𝑛𝑒 also show similar magnetic

field dependences [181]. From Fig. 3.5, it is clear that the magnetization of the

anatase thin films is larger than the rutile thin films, which may be attributed

to the fact that anatase films in this study have a mobility ∼ 2-11 cm2V−1s−1

which is two orders of the magnitude higher than the mobility of rutile thin

films [169] and also anatase films have larger conductivity than rutile films.

The systematic behavior of conductivity and saturation magnetization (𝑀𝑠)

could be explained by the carrier-induced ferromagnetism mechanism; namely,

as more carriers are induced at lower (𝑃O2), the magnetic interaction becomes

stronger and results in ferromagnetism. According to the RKKY model [34],

which explains the exchange interaction between the magnetic impurity atoms

in DMS as mediated by free charge carriers, the increase in charge carrier

density can enhance the magnetic ordering.

49

300

250

200

150

100

50

0

795790785780775

XA

S In

ten

sity

(ar

b. u

nit

s)

Insulating

Intermediate

Metallic

(a)

µ+ µ-

Photon energy (eV)

Ti0.95Co0.05O2-δ

XAS in TEY mode 120

100

80

60

40

20

0 XA

S In

ten

sity

(ar

b. u

nit

s)

Metallic

(b) XAS in TEY mode

5

0

-5

784782780778Photon energy(eV)

XM

CD

Inte

nsi

ty (

arb

. un

its)

(c) XMCD in TEY mode

Metallic

300

250

200

150

100

50

0795790785780775

Co2+

high spin D2h

Calc.

Co2+

high spin Oh

Calc.

Metallic Expt.

(d)

Photon energy (eV)

XA

S In

ten

sity

(ar

b. u

nit

s)

-20

-10

0

795790785780775

Co2+

high spin D2h

Calc.

Co2+

high spin Oh

Calc.

MetallicExpt.

(e)

Photon energy (eV)

XM

CD

Inte

nsi

ty (

arb

. un

its)

Figure 3.6: (Color online) Co 𝐿2,3-edge of anatase Ti0.95Co0.05O2−𝛿 taken in the

TEY mode at T = 300 K and H = 1 T. (a) XAS. (b),(c) XAS and XMCD spec-

tra of the metallic anatase Ti0.95Co0.05O2−𝛿 sample. (d),(e) Comparison of the

XAS and XMCD spectra shown in (b) and (c) with cluster-model calculation.

50

Table 3.2: Electronic structure parameters for the anatase Co-doped TiO2

thin film used in the cluster-model calculations in units of eV to analyze the

spectra taken in the TEY mode. Δ: Charge-transfer energy, 𝑈𝑑𝑑: On-site 3𝑑-3𝑑

Coulomb energy, 𝑈𝑑𝑐: 3𝑑-2𝑝 Coulomb energy, 𝑉E𝑔 : Hopping integral between

the Co 3𝑑 and O 2𝑝 orbitals of 𝐸𝑔 symmetry, 10Dq: Crystal-field splitting.

Crystal-field symmetry Spin Δ 𝑈𝑑𝑑 𝑈𝑑𝑐 𝑉E𝑔 10Dq

𝐷2ℎ High 4 5 7 1.1 0.9

𝑂ℎ High 2 5 7.5 1.1 0.8

In Fig. 3.6(a), we show the Co 𝐿2,3-edge XAS (metallic, intermediate and

insulating thin films) and Fig. 3.6(b)-(c) XAS and XMCD spectra of (metallic

thin film) anatase Ti1−𝑥Co𝑥O2−𝛿 obtained in the TEY mode. In the figure,

𝜇+ and 𝜇− refer to the absorption coefficients for photon helicity parallel and

antiparallel to the Co majority spin direction, respectively. The XMCD spectra

Δ𝜇 = 𝜇+ - 𝜇− have been corrected for the degree of circular polarization. The

XAS and XMCD spectra of the metallic anatase Ti1−𝑥Co𝑥O2−𝛿 sample showed

multiplet features as shown by Fig 3.6(a)-(c) and agree with our cluster model

calculations using the parameter values listed in Table 3.2, as shown in Fig.

3.6(d)and (e). The multiplet features of the XMCD spectra show almost one-

to-one correspondence to those in the XAS spectra. The XAS and XMCD

spectra taken in the TEY mode for the intermediate samples also show clear

multiplet features. The spectral line shapes of the XAS and XMCD spectra for

the metallic and intermediate anatase Ti1−𝑥Co𝑥O2−𝛿 samples are also similar

to those of rutile Co-doped TiO2 results which were reported in our previous

work [170, 171]. For the insulating sample, we observed an XAS spectrum

similar to those of the metallic and intermediate samples. The estimated

magnetic moments for all samples obtained from XMCD in the TEY mode

were < 0.3 𝜇𝐵/Co. These values are larger than the 0.1 𝜇𝐵/Co which is

reported by Mamiya et al. [170]. But they are still smaller than the saturation

magnetic moments 0.6-2.1 𝜇𝐵/Co deduced from magnetization measurements.

Figures 3.7(a),(b) and (c) shows the Co 𝐿2,3 XAS and XMCD spectra of

the same samples taken in the TFY mode. From the figure, it is clear that the

XMCD intensities are much higher than those taken in the TEY mode. The

large difference between the bulk-sensitive TFY mode with ∼ 100 nm probing

depth and the surface-sensitive TEY mode with ∼5 nm probing depth suggests

that there is a magnetically dead layer of ∼ 5 nm thickness or more at the sur-

face of the samples as in the case of rutile [170,171]. The presence of a surface

dead layer of ∼ 5 nm thickness is consistent with the recent measurements

of the film-thickness dependence of AHE [195]. The spectral line shapes of

51

120

80

40

0

Metallic

XA

S In

ten

sity

(ar

b. u

nit

s) (b) XAS in TFY mode

-20

-10

0

10

784782780778 XM

CD

Inte

nsi

ty (

arb

. un

its)

Photon energy(eV)

(c) XMCD in TFY mode

Metallic

350

300

250

200

150

100

50

0

795790785780775

Metallic

Intermediate

Insulating

XA

S In

ten

sity

(ar

b. u

nit

s)

(a) XAS in TFY mode

µ+, µ-

Photon energy (eV)

Ti0.95Co0.05O2-δ

300

200

100

0

800795790785780775

Co2+

high spin D2h

Calc.

Co2+

high spin Oh

Calc.

35% D2h LS+ 35% Oh LS+30% Oh HS

MetallicExpt.

(d)

Photon energy (eV)

XA

S In

ten

sity

(ar

b. u

nit

s)

Calc.

-50

-40

-30

-20

-10

0

10

20

800795790785780775

Co2+

high spin D2h

Calc.

Co2+

high spin Oh

Calc.

35% D2h LS+ 35% Oh LS+ 30% Oh HS Calc.

Photon energy (eV)

(e)

MetallicExpt.

XM

CD

Inte

nsi

ty (

arb

. un

its)

Figure 3.7: (Color online) Co 𝐿2,3-edge of anatase Ti0.95Co0.05O2−𝛿 taken in

the TFY mode at T = 300 K and H = 1 T. (a) XAS. (b),(c) XAS and

XMCD spectra of anatase Ti0.95Co0.05O2−𝛿 for metallic sample in the TFY

mode. (d),(e) Comparison of XAS and XMCD spectra shown in (b) and (c)

with cluster-model calculation [196].

52

Table 3.3: Electronic structure parameters for anatase Co-doped TiO2 thin film

used in the cluster-model calculations in units of eV to analyze. Δ: Charge-

transfer energy, 𝑈𝑑𝑑: On-site 3𝑑-3𝑑 Coulomb energy, 𝑈𝑑𝑐: 3𝑑-2𝑝 Coulomb

energy, 𝑉E𝑔 : Hopping integral between the Co 3𝑑 and O 2𝑝 orbitals of 𝐸𝑔

symmetry, 10Dq: Crystal-field splitting.Crystal-field symmetry Spin Δ 𝑈𝑑𝑑 𝑈𝑑𝑐 𝑉E𝑔 10Dq Weight(%)

𝐷2ℎ Low 4 5 7 1.1 1.1−1.2 35

𝑂ℎ Low 3 6 7.5 1.1 1.1−1.2 35

𝑂ℎ High 2 5 7.5 1.1 0.8−0.9 30

the XAS and XMCD spectra of all the samples taken in the TFY mode show

broad features with spectral line shapes similar to those of rutile Co-doped

TiO2 [171]. Both magnetization and XMCD intensity increased with carrier

density. This is consistent with spin alignment arises due to the interaction

of local spins with the spin polarized free carriers, in which carrier-mediated

ferromagnetism and ferromagnetic ordering is realized. The electrically in-

duced ferromagnetism in anatase Ti1−𝑥Co𝑥O2−𝛿 also supports theoretically as

well as experimentally the idea that ferromagnetism originates from a carrier-

mediated mechanism [180, 181] rather than a non-carrier mediated one [175].

The broadening of the TFY spectra may be due to the randomly displaced

positions of Co atoms, which leads to in various local structures as suggested

by the anomalous X-ray scattering study of Matsumura et al. [186]. The ex-

perimental XAS and XMCD spectra show qualitatively good agreement with

the calculated spectra for the Co2+ in random crystal fields [196], where the

calculations were done using the various electronic structure parameters as

listed in Table 3.3.

Fig 1.1 of chapter 1 shows the phase diagram for (Ga,Mn)As. In simple

estimation we found “Y-axis = Co-band exchange/Fermi energy” for metallic,

intermediate and insulating samples were 2.35, 5.71 and 10.98. Since we know

that effective mass (𝑚★) of anatase is 𝑚0, where 𝑚0 is mass of electrons [193]

and Co-band exchange is a constant which is of order 0.1 eV for 𝑛-type car-

riers [56]. The Fermi energy (𝐸𝐹 ) of these samples were 0.0425 eV, 0.02 eV

and 0.009 eV, respectively. The “X-axis = Carrier concentration/Co concen-

tration” for metallic, intermediate and insulating samples were 0.04, 0.01 and

0.004, respectively. Since the value 10.98 is out of the range of the phase di-

agram for (Ga,Mn)As so it is necessary to predict phase diagram for anatase

Co-doped TiO2 in future.

Figure 3.8 shows magnetization versus magnetic field curves estimated

from the XMCD spectra obtained in the TEY and TFY modes using sum

53

3.0

2.5

2.0

1.5

1.0

0.5

0.01.00.80.60.40.20.0

Cluster model

XM

CD

Inte

nsity

(µ

B/C

o)

(a) Metallic

SQUIDTFY

TEY

Magnetic field (T)1.00.80.60.40.20.0

Cluster model

(b) Intermediate

Magnetic field (T)

SQUIDTFY

TEY

1.00.80.60.40.20.0

Cluster model

(c) Insulating

SQUID

TFY

TEY

Magnetic field (T)

Figure 3.8: (Color online) Magnetization as a function of magnetic field ob-

tained from the XMCD intensities of anatase Ti0.95Co0.05O2−𝛿 compared with

M-H curves obtained using a SQUID.

rules [170], as compared with the M-H curves measured using a SQUID. We

have divided the obtained spin-magnetic moment by a correction factor of

0.92 given by Teramura et al. [189]. The Co magnetic moment is found to be

obviously much larger in the bulk region than in the surface region. These

results are also consistent with the x-ray photoemission spectroscopy study by

Yamashita et al. [197]. Since we know that TFY suffers from self-absorption

and therefore it will saturate the XAS signal. This saturated XAS signal will

reduce XMCD signal. Because of this very fact, we can conclude that the real

value of magnetic moment in bulk should be even higher than the measured

TFY value in metallic and intermediate samples which are reported in the

present work. Accordingly, our observation by using the TEY and TFY modes

are validated. The magnetic moment obtained from cluster-model calculation

(Table 3.3) is 1.6 𝜇𝐵/Co, which is similar to the magnetization of ∼ 2 𝜇𝐵/Co

deduced from the TFY results and the SQUID measurement as shown in Fig

3.8. These results suggest that the Co ions in the bulk region are responsible

for the ferromagnetism in anatase Ti1−𝑥Co𝑥O2−𝛿.

54

3.4 Conclusion

In conclusion, we have studied the high temperature ferromagnetism observed

in rutile-type Ti1−𝑥Co𝑥O2−𝛿 films using x-ray magnetic circular dichroism at

the Co 𝐿2,3 edges (both in the TEY and TFY mode). These results represent

that the high temperature ferromagnetism is originated from the Co2+ atoms,

most probably charge carriers induce the ferromagnetism. The magnetic mo-

ment of the Co ions as long as 0.82-2.25 𝜇𝐵/Co was first observed by the bulk

sensitive TFY method. The magnetic moment value deduced with the TEY

mode ( 0.15-0.24 𝜇𝐵/Co) indicates the presence of a magnetically dead layer

of ∼ 5 nm thickness on the sample surface.

We have also provided experimental evidence for carrier-induced ferro-

magnetism of cobalt-doped anatase TiO2 thin films using XMCD at the Co

𝐿2,3 edges in both the TEY and TFY modes. The large magnetic moment of

the Co ions, 0.6-2.4 𝜇𝐵/Co, was observed by the bulk-sensitive TFY method.

The carrier-induced origin of ferromagnetism at room-temperature in anatase

Ti1−𝑥Co𝑥O2−𝛿 is confirmed on the basis of the element-specific XMCD study

at the surface as well as in bulk. In the bulk-sensitive TFY mode, the position

of Co2+ atoms seems to be displaced from the Ti4+ sites, resulting in more

random crystal fields. Good agreement is demonstrated not only in magneti-

zation and AHE but also in the magnetic field dependences of XMCD. The

magnetic moment values deduced with the TEY mode was < 0.3 𝜇𝐵/Co, in-

dicating the presence of a magnetically dead layer of ∼5 nm thickness at the

sample surfaces.

55

Chapter 4

X-ray magnetic circular

dichroism study of

ferromagnetic BiFe1−𝑥Co𝑥O3

thin films

4.1 Introduction

Multiferroics have attracted tremendous interest in recents years, which

simultaneously show spontaneous electric and magnetic ordering in the same

phase, [198–201] because of their potential applications in the fields of infor-

mation storage, spintronics, and sensor, and also because of their underlying

fascinating fundamental physics [198]. There are not many natural multiferroic

materials exist because of the incompatibility between the conventional cation

off-center distortion mechanism in ferroelectrics and the formation of magnetic

moments [199] at the cation sites. Most of multiferroics have, however, very

different ferroelectric and magnetic ordering temperatures and the magnetic

ordering is usually antiferromagnetic. BiFeO3 (BFO) is a rare multiferroic ma-

terial which is both ferroelectric (𝑇C ∼1103 K) and antiferromagnetic (𝑇N ∼643 K) at room temperature, and exhibits weak ferromagnetism because of the

canted spin structure [202]. The magnetic structure is a nearly 𝐺-type antifer-

romagnet, i.e., the Fe magnetic moments are coupled ferromagnetically within

the pseudocubic (111) planes and coupled antiferromagnetically between ad-

jacent planes. However, the additional incommensurate spin modulation leads

to the cancellation of macroscopic magnetization. This incommensurate spiral

spin structure can be suppressed by applying high magnetic field [204], chem-

ical substitution [205] or epitaxial strain in the case of thin films [206], which

prohibits the linear magnetoelectric effect from being observed [203].

56

It has been predicted that spontaneous magnetization can be induced in

BFO by mixed valence or chemical substitution either changing the Fe-O-Fe

bond angle or a statistical octahedra distortion of the FeO6 [198,201,207]. In

the case of thin films, it has been shown that heteroepitaxially strained BFO

films are ferromagnetic at room temperature and show a remarkably large

magnetoelectric effect [208]. The structure and properties of the bulk single

crystals form have been extensively studied [209–214], and they have been

shown to possess a rhombohedrally distorted perovskite structure (a = b = c

= 5.63 A, 𝛼 = 𝛽 = 𝛾 = 59.4∘) at room temperature. Wang et al. [206] recently

reported multiferroic behavior, with ferromagnetic and ferroelectric polariza-

tions that are both large at room temperature, in thin strained films of BFO.

They reported that a 70-nm film shows both an enhanced ferroelectric polariza-

tion (90 𝜇C.cm−2) and a substantial magnetization (1 𝜇𝐵/Fe). This remains

the only report of a robust room temperature multiferroic and suggests the

potential for novel devices that exploit the anticipated strain-mediated mag-

netoelectric coupling between the two ordered ground states. Sakamoto et al.

have reported that the resistivity of BFO improves markedly with substitution

of a small amount of Mn ions for Fe. The electrical properties of Mn-doped

BFO have been attributed to its characteristic electronic structure, including

its band gap, degree of hybridization, and the valence state [215]. So far the

electronic structure and magnetic properties of Co-doped BFO (BFCO) has

not been clarified experimentally.

X-ray absorption spectroscopy (XAS) and magnetic circular dichroism

(XMCD) are effective techniques for probing the electronic and magnetic prop-

erties of complex solids. The optical process involved in XAS and XMCD is

a local one because of the localized core state and the dipole selection rules.

Several studies have utilized these techniques to examine the electronic and

magnetic states of such materials [210–213]. These studies have focused on

multiferroic BFO-based materials and investigated the electronic and magnetic

structure at the Fe 2𝑝 core absorption edge [210–213].

In this thesis, we investigated the electronic structure and the magnetic

properties of BFCO thin films grown on single-crystal LaAlO3 (LAO) (001)

substrates using XAS and XMCD. XAS and XMCD enabled us to study the

element-specific electronic properties of the BFCO thin films. Based on our

experimental results, we discuss the origin of ferromagnetism at room temper-

ature in BFCO thin films.

57

4.2 Experimental

BFCO thin films with a thickness of ∼ 260 nm were grown on LAO (001) sub-

strates using a chemical solution deposition technique. Prior to the deposition

a 100-nm-thick SrRuO3 (SRO) was grown using the pulsed laser deposition

method to serve as a base electrode. These substrates were spin-coated (for

30 s at 4000 rpm) using a solution containing Bi, Co, and Fe organometallic

compounds (Koujundo Chemical Laboratory Co. Ltd.) in a xylene carrier in

the desired ratio at a concentration of 0.2 mol/kg. The samples were then

heated in air at 250∘C for 5 min. Repeating this coating-heating sequence

13 times yielded BFCO films that were about 260 nm thick. After the films

were annealed for 60 min in a 5% ozone-oxygen mixture in a tube furnace at

420∘C, the BFCO/SRO(001)/LAO(001) samples were subjected to a second

annealing stage at 435∘C for 30 min to improve their crystallinity. Structural

characterization was carried out using x-ray diffraction (XRD), which demon-

strated a clear single phase [214]. This XRD experiments confirmed that, as

the Co content grows, BFCO moves from rhombohedral (R) to tetragonal (T)

symmetry and both phases (R and T) exists for Co content= 0.10 to 0.20.

However, it converted to T phase after Co content= 0.20. Transmission elec-

tron spectroscopy (TEM) images show that the existence of secondary phases

in the form of cluster.

The XAS and XMCD measurements were performed at the undulator

beam line BL-23SU of SPring-8, Japan. The measurements were performed

at room temperature in an applied magnetic field over the range up to ± 3T.

The propagation vector of photons with circular polarization was parallel to

the magnetic field and perpendicular to the sample surface. The spectra were

collected in the total electron yield (TEY) mode which has a probing depth of

∼5 nm. The monochromator resolution was E/ΔE>10000, and the circular

polarization of the x-rays was 95%. The base pressure of the chamber was

about 10−9 Torr. The XAS and XMCD spectra were obtained in the TEY

mode without surface treatment.

4.3 Results and Discussion

Figure 4.1 shows the Fe 2𝑝-3𝑑 XAS spectra of the BFO sample consisting

of 𝐿3 (2𝑝3/2) and 𝐿2 (2𝑝1/2) regions. This Fe 2𝑝-3𝑑 XAS spectrum of BFO is

compared with the Fe 2𝑝 XAS spectra of 𝛼-Fe2O3, Fe3O4, and FeO [216,217].

The line shape of the XAS spectrum of the BFO sample shows a two-peak

structure at the 2𝑝3/2 (𝐿3) edge is similar to 𝛼-Fe2O3, indicating that the Fe

ions in BFO were in the trivalent state (𝑂ℎ) site [212, 218]. This result is

58

250

200

150

100

50

0

-50

730725720715710705Photon energy(eV)

XA

S in

ten

sity

(ar

b. u

nit

s)

Fe 2p3/2

Fe 2p1/2

a-Fe2O3, Fe3+

BiFeO3 thin film

Fe 2p-3d XAS at 300K

Fe3O4 Mixed Fe3+

& Fe2+

FeO, Fe2+

Regan et al

Regan et al

Regan et al

Figure 4.1: (Color online) Fe 2𝑝-3𝑑 XAS spectra of the BiFeO3 thin film com-

pared with those of 𝛼-Fe2O3, FeO, Fe3O4 [216,217].

consistent with the electron spin resonance study of BFO [219]. Therefore, the

XAS spectrum of the BFO sample indicated that the valence state of Fe is 3+.

Our experimental results do not include the presence of Fe2+ ions, as has been

suggested by the previous XAS and XMCD studies of BFO thin films [211].

Figures 4.2(a) and (b) show the Fe 2𝑝-3𝑑 XAS spectra of BFO and

BFCO samples with opposite magnetization directions recorded using circular

polarized x-rays. Their difference spectra, i.e., XMCD spectra, are shown in

Figs. 4.2(c). Here, the XAS spectra obtained in applied magnetic fields of

+3.0 and -3.0 T are denoted by 𝜇+ and 𝜇−, respectively. Three sharp peaks

occurred in the XMCD spectra around ℎ𝜈 = 708.5, 709.7, and 710.5 eV. Figure

4.2(d) shows that the Fe 2𝑝-3𝑑 XMCD spectrum of BFCO films were different

from those of the FeO [220], indicating that the magnetism in these samples

did not arise from Fe metal segregation. Also XMCD spectra of BFCO films

are compared with those of GaFeO3 (GFO), where the Fe3+ ions are located

at the 𝑂ℎ sites [221], 𝛾-Fe2O3 nanoparticles, where the Fe3+ ions are located

at both octahedral (𝑂ℎ) and the tetrahedral (𝑇𝑑) sites [222] and Fe3O4, where

Fe3+ ions are located at both the 𝑇𝑑 and 𝑂ℎ sites and Fe2+ ions are located at

59

140

120

100

80

60

40

20

0

XA

S in

tens

ity (

arb.

uni

ts) BiFeO3 thin film

Fe 2p-3d XAS at 300K

s+

s-H= ±3.0 T

(a)

350

300

250

200

150

100

50

0730725720715710705

XA

S In

tens

ity (

arb.

uni

ts)

Co=30%

Co=20%

Co=15%

Co=05%

Co=0%

Fe 2p-3d XAS at 300K, H= ±3T

BiFe1-xCoxO3 (x = 0 to 30%)

σ+

σ−

(b)

Photon Energy(eV)

-20

-10

0

10

Co=30% Co=20% Co=15% Co=5.0% Co=0.0%

XM

CD

Inte

nsity

(ar

b. u

nits

)

Fe 2p-3d XMCDBiFe1-xCoxO3 (x = 0 to 30%)

(c)

150

100

50

0

-50

-100

730725720715710705

A

B

C

Fe 2p-3d XMCD at 300K

GaFeO3 J.-Y. Kim et al.,

γ−Fe2O3 nano-particles

S.B. Profeta et al.,

BiFe1-xCoxO3

Fe3O4 J.Chen et al.,

XM

CD

Inte

nsity

(ar

b. u

nits

)

Photon Energy(eV)

Fe3+

(Oh)

Fe3+

(Oh), Fe2+

(Oh)

Fe3+

(Td)

Fe3+

(Oh)Fe

3+(Oh)

Fe3+

(Td)

Fe3+

(Oh)Fe

3+(Oh)

(d)

Figure 4.2: (Color online) Fe 2𝑝-3𝑑 XAS and XMCD spectra of the

BiFe1−𝑥Co𝑥O3 thin films. (a) XAS spectra in magnetic fields of ± 3T at

300 K. (b) XAS spectra of the BiFe1−𝑥Co𝑥O3 (𝑥 = 0 to 0.30) thin films in

magnetic fields of ± 3T at 300 K. (c) XMCD spectra of the BiFe1−𝑥Co𝑥O3

(𝑥 = 0 to 0.30) thin film in magnetic fields of ± 3T at 300 K. (d) Compari-

son of the experimental XMCD spectra of BFCO with the XMCD spectra of

GaFeO3 [221], 𝛾-Fe2O3 [217], Fe3O4 [222] and FeO [220].60

the 𝑂ℎ sites [217]. The XMCD spectrum of Fe3O4, which displays overlapping

contributions of the Fe3+ and Fe2+ ion, and that of GFO with Fe3+ ions at the

𝑂ℎ sites are different from the spectra of BFCO thin films. On the other hand,

the spectral line shape of the 𝛾-Fe2O3 thin films, where the XMCD signals arose

from the antiferromagnetically coupled Fe3+ (𝑂ℎ) ions and Fe3+(𝑇𝑑), was nearly

identical to that of the BFCO thin films. This indicates that the magnetism

in the BFCO thin films is mainly originated from antiferromagnetially coupled

Fe3+ ions located at both the 𝑂ℎ and 𝑇𝑑 sites. Since in BFCO, Fe ions do not

occupy 𝑇𝑑 sites, this result suggests that the observed ferromagnetism and/or

superparamagnetic behavior in the BFCO thin films is mainly arose from 𝛾-

Fe2O3 like species as a secondary phase, as has been discussed as the origin of

the magnetic moment in BFCO thin films [223].

Figure 4.3(a) shows the Fe 2𝑝-3𝑑 XMCD spectrum of BiFe0.80Co0.20O3

thin films measured in various applied magnetic fields. It can be seen that

the XMCD peak intensity remains high down to an applied field of 0.1 T

[Figs. 4.3(a) and (b)], indicating that the ferromagnetism exists in this sam-

ple. In Fig. 4.3(a), peaks from the Fe3+ ions located at the 𝑇𝑑 and 𝑂ℎ sites

had different signs. This clearly implies the presence of Fe3+ (𝑇𝑑)- Fe3+ (𝑂ℎ)

antiferromagnetic coupling. This XMCD line shape is independent of the mag-

netic field, indicating that the ferromagnetism originates from the antiferro-

magnetic coupled Fe3+ ions at the 𝑂ℎ and 𝑇𝑑 sites. The Fe3+ ions occupy

two sets of nonequivalent positions are (𝑇𝑑 and 𝑂ℎ sites) in unequal numbers

and are oriented in the antiparallel directions, so that there is a net magnetic

moment [218], i.e., the ferromagnetism is caused by the difference in the num-

ber of Fe atoms between the up and down spins on the 𝑇𝑑 (𝑂ℎ) and 𝑂ℎ (𝑇𝑑)

sites respectively. Indeed, room-temperature ferromagnetism in 𝛾-Fe2O3 arises

from such Fe3+-Fe3+ antiferromagnetic coupling [224]. The magnitude of the

magnetic moment was quantitatively estimated using the XMCD spin sum

rule [220] to be ∼ 0.03 𝜇𝐵/Fe for BFO, consistent with our SQUID measure-

ment. This magnetic moment is also consistent with the canted moment in

bulk BFO (0.03 𝜇𝐵/Fe) [225]. Therefore, the magnetic moment of Fe in BFCO

samples may arises from 𝛾-Fe2O3 as a secondary phase, although we could not

observe peaks from 𝛾-Fe2O3 in the XRD spectra. This may be because of

the nanosized dimensions of 𝛾-Fe2O3 [223]. It is to note that existence of the

secondary phases were been confirmed by TEM by our collaborator.

The magnetic moment of Fe in the BiFe0.80Co0.20O3 thin film obtained

from the sum rule and the magnetization data are shown in Figs. 4.4(b).

Since the value of saturation magnetization was ∼1.25𝜇𝐵/Fe for 𝛾-Fe2O3, the

proportion of the 𝛾-Fe2O3 phase in the BiFe0.80Co0.20O3 sample was was es-

timated to be more than ∼11%. The M-H curve obtained from the XMCD

61

-30

-20

-10

0

10

20

30

730725720715710705

B

AC

3.0T2.0T1.0T0.5T0.1T

XM

CD

Inte

nsity

Photon Energy(eV)

(a) Fe 2p-3d XMCD

Fe3+

(Td)

Fe3+

(Oh)Fe3+

(Oh)

BiFe0.80Co0.20O3 thin film at 300K

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.03.02.52.01.51.00.50.0

Mag

netic

mom

ent (

µB

/Fe)

Magnetic field (T)

(b) BiFe0.80Co0.20O3 thin film at 300K

Fe

Figure 4.3: (Color online) Magnetic field dependence of the Fe 2𝑝-3𝑑 XMCD

spectra of the BiFe0.80Co0.20O3 thin film. (a) Spectra measured at various

magnetic fields. (b) Magnetic moment of Fe obtained from the XMCD intensity

as a function of magnetic field.

62

1.4

1.2

1.0

0.8

0.6

0.4

0.2

302520151050

(b) H= ±3T

Fe

% Co in BFO

Mag

netic

mom

ent (

µB

/ ion

)

XMCD

T=300K

0.3

0.2

0.1

0.0

Mag

neiz

atio

n (µ

B/ f

.u.)

SQUID

(a) H=±1

T=300K

Figure 4.4: (Color online) (a) Magnetization of BiFe1−𝑥Co𝑥O3 (𝑥 = 0 to 0.30)

thin films as a function of Co content in BiFeO3 obtained from SQUID. (b)

Magnetization of Fe in BiFe1−𝑥Co𝑥O3 (𝑥 = 0 to 0.30) thin films as a function

of percent of Co content in the BFCO thin films obtained from XMCD.

63

measurements was larger than those from the SQUID measurements. A pos-

sible reason for this discrepancy is that the surfaces of the films had a higher

magnetization than the bulk. Since TEY mode is surface sensitive and probing

depth of ∼5 nm however, SQUID is bulk sensitive. The shape of Fe magnetic

moments with Co content obtained from XMCD as well as SQUID are same

which is clear from Fig.4.4. From the shape of M-H curve we calculated the

Curie constant of Fe from the paramagnetic part of the XMCD intensity, and

found the Curie constant to be ∼ 591 J.K/mol T2, which is much larger than

the Curie constant of Fe3+(S = 5/2) ∼ 32 J.K/mol T2. This means that there

is strong ferromagnetic interaction between the Fe ions. Hence, we may also

consider that BFCO/𝛾-Fe2O3 forms a nanocomposite and can show a strong

magnetoelectric coupling response, because BFCO is strongly ferroelectric and

𝛾-Fe2O3 is strongly ferromagnetic.

Figure 4.5(a) shows the Co 2𝑝 XAS spectrum of BFCO. Because the Co

concentration was low, the intensities were low. The XMCD data are shown

in Fig. 4.5(b). The spectrum is derived from the 𝐿3 (2𝑝3/2) edge occuring at

∼ 642 eV and the 𝐿2 (2𝑝1/2) edge occurring at ∼ 653 eV. It can be clearly

seen that the Co 2𝑝 XAS spectrum of BFCO thin films was similar to that of

LaCoO3 (with Co3+) at high temperatures (650 K) but different from that of

CoO [226] (with Co2+), whose spectra are shown in Fig. 4.5(c). Thus the result

indicates that the Co ions are mainly in the trivalent high-spin states [227].

In addition, the spectra of other Co content are identical, indicating that the

valence state of the Co ions did not change with Co content. Similarly, the

XMCD peak intensity was also nearly independent of Co content. It can be

seen that the XMCD peak intensity approached zero as the magnetic field

goes to zero, as shown in Fig. 4.5(d), indicating that the Co ions were in the

paramagnetic. We calculated the Curie constant of Co from the data shown

Fig. 4.5(d) and found the value of the Curie constant was ∼ 67 J.K/mol T2

at 300 K which is smaller than the Curie constant of Co3+ (S = 2) of ∼ 100

J.K/mol T2. This means that most of the Co ions have an antiferromagnetic

interaction, and the participation of Co ions in the magnetism of BFCO thin

films is low.

The magnetic moment of the Fe and Co ions estimated using the XMCD

spin sum rule [219] are shown in Fig. 4.6(a). The Co magnetic moment was

nearly independent of Co content unlike Fe, and the peak at 20% Co content

showed only a minor influence.

Now we discuss the mechanism of ferromagnetism in the BFCO thin

films. The presence of magnetism in the BFCO thin film, at room temperature

is explained by the existence of nearest-neighbour Fe-Co pairs which cause

the ferromagnetic alignment of the Fe and Co 𝑑-moments under 180∘ super-

64

300

250

200

150

100

50

0

800795790785780775

XA

S In

tens

ity(a

rb. u

nits

)

Co=05%

Co=15%

Co=20%

Co=30%

(a) Co 2p-3d XAS at T= 300K

H= ±3T

σ+

σ−

Photon Energy(eV)

12

10

8

6

4

2

0

-2

800795790785780775

Co=05%

Co=15%

Co=20%

Co=30%

XM

CD

Inte

nsity

(arb

. uni

ts)

Photon Energy(eV)

(b) Co 2p-3d XMCD at T= 300K

400

200

0

-200

800795790785780775

LaCoO3, T=650K

LiCoO2 , T=300K

CoO, T=300K

Co metal, T=300K

BiFe0.80Co0.20O3

(c) Co 2p-3d XAS at 300K

Photon Energy(eV)

Co3+

, S=2

Co3+

, S=2

Co2+

XM

CD

Inte

nsity

(arb

. uni

ts)

0.14

0.12

0.10

0.08

0.06

0.04

0.02

0.003.02.01.00.0

Magnetic field (T)

Mag

netic

mom

ent (

µB

/Co)

(d) BiFe0.80Co0.20O3 thin film

Co

T= 300K

Figure 4.5: (Color online) Co 2𝑝-3𝑑 XAS and XMCD spectra of the

BiFe1−𝑥Co𝑥O3 (𝑥 = 0 to 0.30) thin films (a) XAS spectra in magnetic field

± 3T at 300 K. (b) XMCD spectra. (c) XMCD spectra of the BiFe1−𝑥Co𝑥O3

thin films compared with the XMCD spectra of LaCoO3 [227], LiCoO2, CoO

and Co metal [226]. (d) Magnetic moment of Co obtained from XMCD inten-

sities as a function of magnetic field.

65

0.3

0.2

0.1

0.0

Mag

neiz

atio

n (µ

B/ f

.u.)

SQUID

(a) H=±1

T=300K

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0302520151050

(b) H= ±3T

Fe

Co

% Co in BFO

Mag

netic

mom

ent (

µB/ i

on)

XMCD

T=300K

Figure 4.6: (Color online) Magnetization of BiFe1−𝑥Co𝑥O3 (𝑥 = 0 to 0.30) thin

films as a function of Co content in BiFeO3. (a) Results obtained from SQUID.

(b) Results obtained from XMCD.

66

exchange interaction. Since both Fe and Co ions are formally in the 3+ valance

states, BFCO is an example of a 𝑑5-𝑑6 orbital combination, and the magnetic

coupling between Fe3+ and Co3+ was predicted to be ferromagnetic in terms of

super-exchange [228–230]. If some of the Fe3+ (𝑑5)ions in the cubic structure

are replaced by Co3+ (𝑑6), the antiferromagnetic part of the nearest neighbor

coupling -now between Fe3+ (𝑑5) and Co3+ (𝑑6)−is slightly reduced and, in fact,

ferromagnetic interaction wins. This can be explained by the last electron of

𝑡2𝑔 of the Co3+, which will jump to 𝑡2𝑔 of the Fe3+ and will return back to 𝑡2𝑔of Co3+ and so it will gain ferromagnetism.

4.4 Conclusion

We have performed XAS and the XMCD measurements on BFCO thin films,

which exhibit ferromagnetism at 300 K. The XAS and XMCD line shape of

Fe ions in BFCO thin films are independent of the magnetic field, indicating

that the ferromagnetism originates from the antiferromagnetic coupled Fe3+

ions at the 𝑂ℎ and 𝑇𝑑 sites and ferromagnetism behavior in the BFCO thin

films is mainly arose from 𝛾-Fe2O3 like species as a secondary phase (i.e. fer-

romagnetism behavior in the BFCO thin films is mainly arose from extrinsic

behavior). The magnetic moment of Fe increases with Co content up to 20%

and after that it decreases. The Co ions are in the trivalent high-spin states

and are largely paramagnetic. The Co magnetic moment is nearly independent

of Co content, unlike Fe, and the peak at 20% Co has only a minor influence.

Surface Fe ions in these films also showed a significant enhancement compared

with a bulk sample probed by XMCD.

67

Chapter 5

Effect of off-stoichiometry in

Heusler alloy thin film on

spin-dependent tunneling

characteristics of

Co2Mn𝛽𝑍/MgO (𝑍= Ge, Si)

magnetic tunnel junctions

studied by x-ray magnetic

circular dichroism

5.1 Introduction

Magnetic tunnel junctions (MTJs) have recently been widely studied as mag-

netic storage devices and magnetic sensors. To improve their performance, it is

very important to understand the nature of spin-dependent conduction and de-

velop MTJs having a high tunnel magnetoresistance (TMR) ratio with highly

spin-polarized ferromagnetic electrodes. In this context, a fully epitaxial MTJ

made of an MgO tunnel barrier and half-metallic ferromagnetic electrodes is

very promising because the single-crystalline MgO enables coherent tunneling

through the Δ1 band that is conserving the highly polarized electron spins of

the half-metallic ferromagnetic electrodes. Fully epitaxial Fe/MgO/Fe MTJs,

in which Fe electrodes feature a half-metallic nature for the Δ1 band electrons,

have experimentally shown a TMR ratio of 180% (Ref. [231]) at room temper-

68

ature (RT) owing to coherent tunneling of the Δ1-band electrons. This value

is much higher than those in conventional MTJs consisting of polycrystalline

electrodes and an amorphous AlO𝑥 tunnel barrier.

Co-based full Heusler alloys Co2𝑌 𝑍 such as Co2MnGe and Co2MnSi

are also promising candidates for ferromagnetic electrodes in MTJs, because

theories have predicted that they are ideal half metals [93, 94, 232, 233]. In

fact, TMR ratios of several hundred percent in Co-based full Heusler alloys

with an MgO tunnel barrier have recently been reported [234–238]. Here,

the TMR ratio is defined as [TMR ratio]=(𝑅𝐴𝑃 -𝑅𝑃 ) /𝑅𝑃 in terms of the

resistances for the parallel (𝑅𝑃 ) and antiparallel (𝑅𝐴𝑃 ) geometries [239]. The

(anti) parallel geometry means that electrode 1 is (anti) ferromagnetically

coupled with electrode 2 across the thin nonmagnetic tunnel barrier. Assuming

ideal half metals without interface state (Jullieres model), [240] the formula

can be written as [TMR ratio]=2𝑃1𝑃2 / (1-𝑃1𝑃2), where 𝑃1 and 𝑃2 are the spin

polarizations at the Fermi level (𝐸𝐹 ) of electrodes 1 and 2, respectively [175].

In this model, if the Co-based full Heusler alloys are perfectly half metallic,

i.e., if 𝑃1 and 𝑃2 are equal to 1, the TMR ratio must be infinite. The TMR

ratio of several hundred percent in real Heusler alloy/MgO MTJs implies that

the half metallicity might have been deteriorated for some reason.

A disorder-free 𝑋2𝑌 𝑍 full Heusler alloy has the 𝐿21 crystal structure

having four fcc sublattices. When the elements 𝑌 and 𝑍 are randomly located,

the crystal structure changes to 𝐵2. When 𝑌 , 𝑍, and 𝑋 are disordered, it

changes to 𝐴2. The crystal structure change from 𝐿21 to 𝐵2 or 𝐴2 is one

of the possible reasons to reduce the spin polarizations at 𝐸𝐹 . In addition

a numerical study [241] suggests that lattice distortions and the existence of

impurities at the interfaces could also reduce the spin polarization. Hence,

high-quality interfaces are the key for obtaining high TMR ratios. Thus, it is

very important to characterize the interfacial magnetic and electronic states

of Heusler alloy/MgO MTJs.

Asakura et al. [37] studied the CMG film-thickness dependence of XMCD

and magnetic moments of seven CMG/MgO samples with various CMG thick-

nesses ranging from 1 to 172 monolayers (MLs). They reported that the XAS

and XMCD spectral shapes for thick samples (𝑡CMG ≥ 4 ML) were similar

to those for bulk CMG, [243] and neither the Mn nor Co atoms were oxi-

dized. We have found that about 70% of the Mn atoms in the 1-ML sam-

ple were oxidized. The lattice distortions and disorder in the ultrathin sam-

ples would be related to oxidation. In contrast, Co atoms in the ultrathin

samples were not oxidized and more strongly spin polarized. The enhanced

𝑚spin(Co) for the ultrathin samples could be due to the Fe/Co interfacial ef-

fect as in CMS/MgO heterostructures [244] The existence of Co antisites is

69

suggested by considering theories on Co antisites [91, 92] and the observed

𝑚spin(Co) of 1.40-1.77 𝜇𝐵, which was larger than theoretical values for ideal

compounds. CMG composition-dependent XMCD studies was desirable to

fully understand the electronic states of CMG with an MgO barrier so we

studied non-stoichiometic Co2Mn𝛽Si0.88 (CMS) and Co2Mn𝛽Ge0.38(CMG) thin

films with different compositions of Mn. Nonstoichiometry in Co2𝑌 𝑍 certainly

leads to the introduction of defects in the Co2𝑌 𝑍 host. The effect of defects

in Co2𝑌 𝑍 on the spin-dependent electronic structure has been investigated

theoretically [92,95,245–250]. Picozzi et al. [92] predicted from first principles

calculation that half-metallicity in Co2MnSi is lost for CoMn antisites, where a

Mn site is replaced by a Co atom, because of the appearance of minority-spin

gap states near 𝐸𝐹 , while half-metallicity is retained for MnCo antisites, where

a Co site is replaced by a Mn atom. In this study, we investigated the magnetic

and electronic structures of Mn and Co atoms in CMS/MgO and CMG/MgO

samples with various Mn compositions. XMCD is a very powerful tool to ex-

tract the information about the interfacial magnetic and electronic states of

CMS and CMG-MTJs fabricated with various Mn compositions. It also ex-

plains the origin of the observed magnetization in term of defects, associated

with non-stoichiometry in the prepared CMS and CMG electrodes.

5.2 Experimental

The samples which we studied had layer structures were as follows: (from

the substrate side) MgO buffer layer (10 nm)/ CMS or CMG (30nm)/MgO

barrier (2 nm) /AlO𝑥 (1 nm) capping. The preparation of the samples is

described in detail elsewhere [138]. Each sample layer was successively de-

posited in an ultrahigh vacuum chamber with a base pressure ∼ 6 x 10−8

Pa through the magnetron sputtering for electron beam evaporation of CMS

for MgO. The CMS films were deposited at RT by rf magnetron sputtering

from a stoichiometric CMS target and the films deposited on the MgO buffer

were subsequently annealed in situ at 600∘C (325∘C for CMG) for 15 min.

The transmission electron microscopy (TEM) images show very smooth and

abrupt interfaces and all layers were grown epitaxially. It is also clear that

it has 𝐿21 structure [138]. The X-ray absorption spectroscopy (XAS) and

XMCD measurements were performed at BL-16A of Photon Factory, Japan.

The monochromator resolution was 𝐸/Δ𝐸 > 10000, the circular polarization

of X-rays was 87%, the base pressure of the chamber was about 10−9 Torr

and the sample temperature was maintained at 300 K. The samples were in-

troduced into an ultra high vacuum experimental chamber. XAS and XMCD

spectra were obtained in the TEY modes without surface preparation in order

70

to avoid possible destruction of the sample surface. The probing depth of the

TEY mode was ∼ 5 nm and the XMCD was measured with a magnetic field

(± 3 T) applied perpendicular to the films at 300 K and 20 K.

5.3 Results and discussion

5.3.1 Magnetic properties of Co2Mn𝛽Si0.88/MgO MTJs

as a function of Mn composition 𝛽

Figure 5.1(a) shows the photon flux-normalized polarization dependent XAS

spectra (𝜇+ and 𝜇−) at the Mn 𝐿3,2 edges (2𝑝3/2,1/2 → 3𝑑 absorption). Fig

5.1(b) displays the Mn 𝐿3,2-edge XMCD (Δ𝜇 = 𝜇+-𝜇−) spectra. Here 𝜇+ and

𝜇− stand for the absorption coefficient for the photon helicity parallel and

antiparallel to the Mn 3𝑑 majority spin. The background subtraction of the

XAS and CMCD spectra are shown in Chapter 2 [see Fig 2.2]. In the XAS

spectra for Mn composition 𝛽 from 0.69 to 1.29, a shoulder-like structure was

observed on the higher energy side of the Mn 𝐿3 peak, and the Mn 𝐿2 peak was

split into a doublet. These features are characteristics observed for Co2MnSi

and Co2MnGe [243]. These features are due to interplay of two effects, namely,

the exchange and Coulomb interactions between the core holes and unpaired

electrons in the valence band, and strong hybridization between the 3𝑑 and

surrounding electronic states [243]. Here Mn atom was not oxidized in the

interfacial region which is clear from XAS spectrum. The XAS and XMCD

intensities at both 20 K and 300 K showed 𝛽 dependence, that is, the XAS and

XMCD intensities are constant with 𝛽 from 0.69 to 1.0 and then decreased for

𝛽 beyond 1.0 both at 20K and 300K.

By applying the sum rules [252,253], we obtained the spin- and orbital-

magnetic moments of the Mn atom from the XAS and XMCD spectra. We

assumed the contribution of the dipole operator term 𝑇𝑧 in the spin sum rule

to be negligibly small because the Mn site in CMS has 𝑇𝑑 symmetry [37]. Since

there are overlapping regions of 𝐿3 and 𝐿2 due to strong exchange interaction

between the Mn 2𝑝 core hole and the Mn 3𝑑 electrons in Mn 𝐿2,3 XAS and

XMCD, we divided the obtained spin-magnetic moment by a correction factor

0.68 given by Teramura et al. [254]. We assumed, on the basis of a band-

structure calculation, a 3𝑑 hole number (𝑛ℎ) of 4.5 for Mn. [135,255]. As shown

in Fig 5.2, the deduced spin magnetic moment 𝑚spin(Mn) was 3.07 𝜇𝐵 at 300

K and 3.32 𝜇𝐵 at 20 K for 𝛽 =1.29. For 𝛽 =1.29, 𝑚spin (Mn) was nearly the

as same the theoretical value of 3.04 𝜇𝐵 [92] for bulk. Also the trend of 𝑚spin

(Mn) with 𝛽 was consistent with theoretical calculation by Picozzi et al. [92]

which predicts the Mn magnetic moment is now coupled antiferromagnetically

71

500

400

300

200

100

0

670660650640630

XA

S In

tens

ity (

a.u.

)

b=0.69

b=0.99

b=1.15

b=1.29

(a) B=± 3T T=300 K

µ+

µ-

Mn L2,3 XAS

Mn L2Mn L3

Co2MnbSi0.93

Photon energy (eV)

-60

-40

-20

0

648644640636Photon energy (eV)

XM

CD

Inte

nsity

(a.

u.)

b=0.69 b=0.99 b=1.15 b=1.29

(b)

Mn L3

Mn L2,3 XMCD

Figure 5.1: Mn 𝐿3,2-edge XAS and XMCD of Mn-rich Co2MnSi samples with

various Mn composition (a) XAS taken at 20 K and 300 K with B = ± 3

T. 𝜇+ and 𝜇− are the absorption coefficients for photon helicity parallel and

antiparallel to the Mn 3𝑑 majority spin, respectively. (b) XMCD spectra.

72

at MnCo site, leading to a reduction of saturation magnetization. The Mn

orbital magnetic moment, 𝑚orb (Mn), was found to be ∼ 0.30 𝜇𝐵 at 300 K and

∼ 0.40 𝜇𝐵 at 20 K for all the samples.

Figure 5.3(a) shows photon flux-normalized, polarization dependent XAS

spectra at the Co 𝐿3,2 edges of CMS. Fig 5.3(b) displays the Co 𝐿3,2 XMCD

spectra. All the samples showed a shoulder like structure observed in the

higher energy region of the Co 𝐿3-edge XAS. This feature is common to bulk

samples [243]. The XMCD signals are almost same for all 𝛽 values and, which

indicates that there is no effect of excess Mn on Co spin magnetic moment. We

could not find CoO-like multiplet structure for any sample [255]. This means

that Co atom was not oxidized even in the interfacial region.

To determine the Co magnetic moment, we again used the sum rules

[252, 253] as in the case of the Mn 𝐿2,3 edges. Since the Co atoms are also in

the highly symmetric 𝑇𝑑 crystal field, we again neglect the 𝑇𝑧 term in the spin

sum rule [37]. Here We assumed, on the basis of a band-structure calculation,

a 3𝑑 hole number (𝑛ℎ) of 2.2 for Co [135, 255] to the sum rule. As shown in

Fig 5.4, we determined the Co spin magnetic moment 𝑚spin(Co)= 1.1 𝜇𝐵 at

300 K and 1.2 𝜇𝐵 at 20 K for 𝛽=1.29 sample. The 𝑚spin (Co) was 1.1 𝜇𝐵 at

300 K and 1.2 𝜇𝐵 at 20 K for 𝛽=0.69 and all the experimental values of 𝑚spin

(Co) are almost same as bulk sample whose value is 1.06 𝜇𝐵 [92]. The orbital

magnetic moment 𝑚orb (Co) was in the range of 0.2 𝜇𝐵 to 0.1 𝜇𝐵 for all the

samples at 300 K however, it was 0.3 𝜇𝐵 to 0.2 𝜇𝐵 for all samples at 20 K. The

first principle calculation for CMS/MgO MTJs by Miura et al. [256], revealed

that the 𝑚spin(Co) of CMS did not change between the bulk and interracial

regions, consistent with 𝑚orb(Co) at the interface of CMS discovered by our

XMCD measurements.

Next, we discuss the effects of antisites defects in the non-stoichiometric

CMG/MgO thin films. The deduced 𝑚spin(Mn)for all the CMS samples de-

creases and TMR ratio [138] for all the CMS samples increases with Mn com-

position (𝛽) which is consistent with Picozzi et al. [92]. This result suggest

that Mn anti-site defects are forming. It was reported that the 𝑚spin(Co) of

Co-rich CMS/MgO was 1.1 𝜇𝐵 [251] for a sample at room temperature which is

same as a theoretical value of 1.06 𝜇𝐵 which is also consistent with our results.

The CMS film composition of Co: Mn: Si=2: 0.77: 0.93 is Co-rich similar to

the Co-rich CMG that we have studied here. Consequently, the Co-rich CMS

may be more or less the same amount of CoMn as in Co-rich CMG [37]. Picozzi

et al. reported that in-gap states could theoretically exist within the minority

spin gap for CMS with CoMn. The spin polarization of Co-rich CMS estimated

from the TMR ratio at 4.2 K for CMS/MgO MTJs assuming Julliere’s model,

𝑃CMS, was as low as 0.75. [235] The comparison between Co-rich CMG and Co-

73

4

3

2

1

0

mS

pin

(µB/M

n) Spin magnetic moment of Mn

300K 20K Sato et al. PRB 2010 at 300K

Co2MnbSi0.93

4

3

2

1

01.21.11.00.90.80.7

b

mO

rbita

l(µ B

/Mn)

Orbital magnetic moment of Mn

300K 20K

Figure 5.2: Mn composition (𝛽) dependence of the Mn spins magnetic moment

(a) and the Mn orbital magnetic moment (b). They have been determined by

using the spin and orbital sum rules [252,253].

74

500

400

300

200

100

0

810800790780770

b=0.69

b=0.99

b=1.15

b=1.29

(a) B=± 3T T=300 K

Co L2,3 XAS

µ+

µ-

Co L2Co L3

Co2MnbSi1.06

XA

S In

tens

ity (

a.u.

)

Photon energy (eV)

-40

-30

-20

-10

0

10

788784780776Photon energy (eV)

b=0.69 b=0.99b=1.15b=1.29

Co L3

Co L2,3 XMCD(b)

XM

CD

Inte

nsity

(a.

u.)

Figure 5.3: Co 𝐿3,2-edge XAS and XMCD of Mn-rich Co2MnSi samples with

various Mn composition (a) XAS taken at 20 K and 300 K with B = ±3

T. 𝜇+ and 𝜇− are the absorption coefficients for photon helicity parallel and

antiparallel to the Mn 3𝑑 majority spin, respectively. (b) XMCD spectra.

75

4

3

2

1

0

mS

pin

(µB/C

o)

Spin magnetic moment of Co

300K 20K Sato et al. PRB 2010 at 300K

Co2MnbSi0.93

4

3

2

1

01.21.11.00.90.80.7

b

mO

rbita

l(µ B

/Co)

Orbital magnetic moment of Co

300K 20K

Figure 5.4: Mn composition (𝛽) dependence of the Co spins magnetic moment

(a) and the Co orbital magnetic moment (b). They have been determined by

using the spin and orbital sum rules [252,253]

76

rich CMS can be summarized as follows: (i)the deviation in composition from

2:1:1 for CMG is larger than that for CMS, (ii) the 𝑃CMG of 0.74 (Ref. [261]) is

comparable to the 𝑃CMS of 0.75, [235] and (iii) the experimental 𝑚spin(Co) of

the present XMCD results and the theories for CMS and CMG are reasonably

matching. Importantly, the creation of CoMn antisites in Mn-rich CMS elec-

trodes would be suppressed because a MnCo antisite has a much lower energy

than a CoMn antisite. The suppression of CoMn antisite formation would lead

to a decreased density of gap states around 𝐸𝐹 for Mn-rich CMS. To put it

briefly, CoMn antisites which are harmful to the half-metallicity of CMS would

be suppressed with an increasing Mn composition, resulting in a decreased

density of gap states around 𝐸𝐹 of CMS electrodes. For Mn- rich region, there

is no effect on 𝑚spin(Co) of the composition of Mn because the Co-rich CMS

may be more or less the same amount of CoMn as in Co-rich CMG.

5.3.2 Magnetic properties of Co2Mn𝛽Ge0.38 MTJs as a

function of Mn composition 𝛽

Figure 5.5(a) shows the photon flux-normalized polarization dependent XAS

spectra (𝜇+ and 𝜇−) at the Mn 𝐿3,2 edges (2𝑝3/2,1/2 → 3𝑑 absorption). Fig

5.5(b) displays the Mn 𝐿3,2-edge XMCD (Δ𝜇 = 𝜇+-𝜇−) spectra. Here 𝜇+ and

𝜇− stand for the absorption coefficient for the photon helicity parallel and

antiparallel to the Mn 3𝑑 majority spin. The background subtraction of the

XAS and CMCD spectra are shown in Chapter 2 [see Fig 2.2]. In the XAS

spectra for Mn composition 𝛽 from 0.67 to 1.80, a shoulder-like structure was

observed on the higher energy side of the Mn 𝐿3 peak, and the Mn 𝐿2 peak was

split into a doublet. These features are characteristics observed for Co2MnSi

and Co2MnGe [243]. In the XAS spectra for Mn composition (𝛽) 1.40 and

1.60, showed a Mn2+-like multiplet structure in MnO, in contrast to lower

value of 𝛽. The XAS and XMCD intensities at both 20 K and 300 K showed

𝛽 dependence, that is, XAS and XMCD intensities decreases with 𝛽 = 0.67 to

1.80 at 20 K and 300 K.

By applying the sum rules [252,253], we obtained the spin- and orbital-

magnetic moments of the Mn atom from the XAS and XMCD spectra. We

assumed the contribution of the dipole operator term 𝑇𝑧 in the spin sum rule to

be negligibly small because the Mn site in CMG has 𝑇𝑑 symmetry [37]. Since

there are overlapping regions of 𝐿3 and 𝐿2 due to strong exchange interaction

between the Mn 2𝑝 core hole and the Mn 3𝑑 electrons in Mn 𝐿2,3 XAS and

XMCD, we divided the obtained spin-magnetic moment by a correction factor

0.68 given by Teramura et al. [254]. We assumed, on the basis of a band-

structure calculation, a 3𝑑 hole number (𝑛ℎ) of 4.5 for Mn. [135, 255]. As

77

600

400

200

0

670660650640630

b=0.67

b=0.85

b=1.20

b=1.40

b=1.60

b=1.80

Inte

nsity

(ar

b. u

nits

)

(a) B=±3TT=300K Mn L2,3 XAS

Mn L3 Mn L2

Co2MnbGe0.38

µ+

µ-

Photon energy (eV)

-60

-40

-20

0

648644640636

Photon energy (eV)

Mn L3

b=0.67

b=0.85 b=1.20 b=1.40 b=1.60 b=1.80

(b) XMCD

Figure 5.5: Mn 𝐿3,2-edge XAS and XMCD of CMG samples with various Mn

composition (a) XAS taken at 20 K and 300 K with B = ±3 T. 𝜇+ and 𝜇− are

the absorption coefficients for photon helicity parallel and antiparallel to the

Mn 3𝑑 majority spin, respectively. (b) XMCD spectra.

78

shown in Fig 5.6, the deduced spin magnetic moment m𝑠𝑝𝑖𝑛 (Mn) was 2.94 𝜇𝐵

at 300 K and 3.08 𝜇𝐵 at 20 K for 𝛽 =1.2. For 𝛽=1, 𝑚spin (Mn) was nearly

the as same the theoretical value 3.04 𝜇𝐵 for bulk [92]. Also the trend of 𝑚spin

(Mn) with 𝛽 was consistent with theoretical calculation by Picozzi et al. [92]

which predicts the Mn magnetic moment is now coupled antiferromagnetically

at MnCo site, leading to a reduction of saturation magnetization. The Mn

orbital magnetic moment, 𝑚orb (Mn), was found to be 0.3 ∼ 0.2 𝜇𝐵 at 300 K

and 0.40 ∼ 0.3 𝜇𝐵 at 20 K for all the samples.

Figure 5.7(a) shows photon flux-normalized, polarization dependent XAS

spectra at the Co 𝐿3,2 edges of CMG. Fig 5.7(b) displays the Co 𝐿3,2 XMCD

spectra. All the samples showed a shoulder like structure observed in the

higher energy side of the Co 𝐿3-edge XAS. This feature is common to bulk

samples [243]. We could not find CoO-like multiplet structure for any sample

[255]. This means that Co atom was not oxidized even in the interfacial region.

To determine the Co magnetic moment, we again used the sum rules

[252, 253] as in the case to the Mn 𝐿2,3 edges. Since the Co atoms are also in

the highly symmetric 𝑇𝑑 crystal field, we again neglect the 𝑇𝑧 term in the spin

sum rule [37]. Here We assumed, on the basis of a band-structure calculation,

a 3𝑑 hole number (𝑛ℎ) of 2.2 for Co [135, 255] to the sum rule. As shown

in Fig 5.8, we determined the Co spin magnetic moment 𝑚spin(Co)= 1.2 𝜇𝐵

at 300 K and 1.4 𝜇𝐵 at 20 K for 𝛽 =1.2 sample. The 𝑚spin (Co) was 1.3

𝜇𝐵 at 300 K and 1.4 𝜇𝐵 at 20 K for 𝛽=0.67 and all the experimental values

of 𝑚spin(Co) are obviously larger and/or equal to the theoretically predicted

value of 1.06 𝜇𝐵 [91, 92]. For Co-rich region, there is the possibility of the

existence of CoMn because 𝑚spin(Co)=1.5 𝜇𝐵 is larger than 1.06 𝜇𝐵 for Co

at the regular Co site, CoCo. Picozzi et al. [92] investigated theoretically the

effect of antisite defects in CMG and CMS. They reported that 𝑚spin(Co) of

the Co antisites in CMG, CoMn, was 1.35 𝜇𝐵. For Mn-rich region, 𝑚spin(Co) is

almost same as theoretical predicted value, that is, there is no effect of excess

Mn on 𝑚spin(Co). The orbital magnetic moment 𝑚orb (Co) was in the range

of 0.2 𝜇𝐵 to 0.3 𝜇𝐵 for all the samples at 300 K however, it was 0.4 𝜇𝐵 to 0.3

𝜇𝐵 for all samples at 20 K.

Next, we discuss the effects of antisites defects in the non-stoichiometric

CMG/MgO thin films. The deduced 𝑚spin(Mn)for all the CMG samples de-

creases and TMR ratio [138] for all the CMG samples increases with Mn com-

position (𝛽) which is consistent with Picozzi et al. [92]. This result suggest that

Mn anti-site defects are forming. For 𝛽=1.40 and 1.60, 𝑚spin(Mn) decreases

vary rapid because films show oxidation which is clear from XAS spectra. The

deduced 𝑚spin(Co) for all the CMG samples was obviously larger than theoret-

ical value of bulk sample 1.06 𝜇𝐵 and the experimental XMCD value of 1.04

79

4

3

2

1

0

300 K 20 K Theory (for CMG bulk) Asakura et al.,PRB 2010 at 300K

Spin Magnetic Moment of Mn

msp

in(µ

B/M

n)

Co2MnbGe0.38

4

3

2

1

01.81.61.41.21.00.8

Orbital Magnetic Moment of Mn

300 K 20 K Asakura et al.,PRB 2010 at 300K

msp

in(µ

B/M

n)

b

Figure 5.6: Mn composition (𝛽) dependence of the Mn spins magnetic moment

(a) and Co orbital magnetic moment (b). They have been determined by using

the spin and orbital sum rules [252,253].

80

600

400

200

0

810800790780770

b=0.67

b=0.85

b=1.20

b=1.40

b=1.60

b=1.80

inte

nsity

(ar

b. u

nits

)

Co L3 Co L2

Co L2,3 XAS

(a) B=±3TT=300K

µ+

µ-

Co2MnbGe0.38

Photon energy (eV)

-40

-30

-20

-10

0

10

790785780775770Photon energy (eV)

Co L3

b=0.67

b=0.85 b=1.20 b=1.40 b=1.60 b=1.80

(b) XMCD

Figure 5.7: Co 𝐿3,2-edge XAS and XMCD of CMG samples with various Mn

composition (a) XAS taken at 20 K and 300 K and B =±3 T. 𝜇+ and 𝜇− are

the absorption coefficients for photon helicity parallel and antiparallel to the

Mn 3𝑑 majority spin, respectively. (b) XMCD spectra.

𝜇𝐵 for the bulk samples reported by Miyamoto et al. [243]. Note that their

XMCD study was done at 𝐵 =1.4 T and T=45 K and that they scraped the

bulk CMG samples to obtain clean surfaces and used a Co 3𝑑 hole number

of 3.0 [243]. Picozzi et al. [92] investigated theoretically the effect of antisite

defects in CMG and CMS. They reported that 𝑚spin(Co) of the Co antisites

in CMG, CoMn, was 1.35 𝜇𝐵, which was larger than 1.06 𝜇𝐵 for Co at the

regular Co site, CoCo. Here, the Co antisites mean Co atoms occupying Mn

sites in regular CMG. For Co-rich region 𝑚spin(Co) of ∼ 1.20 𝜇𝐵 therefore this

indicates that there is possibility of the existence of CoMn. Thus Picozzi et

al. [92] agrees with our experimental results. We confirmed by in-situ reflec-

tion high-energy electron-diffraction (RHEED) that the present CMG samples

on which we measured XMCD had the disorder-free 𝐿21 structure. However,

since the film composition was Co:Mn:Ge=2: 0.77: 0.38 deviating from 2: 1:

1, some Co atoms possibly occupy the Mn site. The calculated total DOS for

each of the majority- and minority-spin bands of CMG was reported in next

81

4

3

2

1

0

Spin Magnetic Moment of Co 300K 20K Asakura et al.,PRB 2010 at 300K Theory (for CMG bulk)

msp

in(µ

B/C

o)

Co2MnbGe0.38

4

3

2

1

01.81.61.41.21.00.8

300 K 20 K Asakura et al., PRB 2010 at 300K

mO

rbita

l(µ B

/Co)

Orbital Magnetic Moment of Co

b

Figure 5.8: Mn composition (𝛽) dependence of the Co spins magnetic moment

(a) and Co orbital magnetic moment (b). They have been determined by spin

and orbital sum rules [252,253].

82

section and in-gap states were found to exist within the minority-spin gap only

when CoMn existed. The possible existence of CoMn in CMG may lead to a

decrease in the spin polarization. This consideration is consistent with that

the spin polarization of Co-rich CMG estimated from the TMR ratio at 4.2

K for CMG/MgO/CoFe MTJs assuming Julliere’s model, 𝑃CMG, was as low

as 0.74. [261]. However, in Mn-rich region the 𝑚spin(Co) is almost same as

theoretical predicted value.

5.3.3 Mn and Co antisite defects in Co2MnSi and Co2MnGe

(a) Total DOS for defective and Ideal (b) Ideal (c) Defective

Figure 5.9: LDA calculation of antisite defects in Co2MnSi(a) Total DOS for

defective (solid bold line) and ideal (dashed line) Co2MnSi with Mn antisite in

Co2MnSi; majority spin is shown positive and minority spin is negative. The

Mn-antisite PDOS (multiplied by a factor of 3) is also shown (gray shaded

area). Magnetic moments for the ideal(b) and defective (c) systems for the

Mn antisite in Co2MnSi around the defect. Values in bracket denote magnetic

moments for equivalent atoms in Co2MnGe. [92]

Mn antisite defects are most likely to occur in Co2MnSi (Co2MnGe). In

this case the total density of state (DOS) shows a shift of 0.04 eV towards

higher binding energies in the minority spin channel, resulting in a small in-

crease of the spin gap. For this type of antisites disorder the half-metallicity

is maintained. However, the Mn magnetic moment is now coupled antifer-

romagnetically to the Mn antisites, which implies a reduction of saturation

magnetization. Fig 5.9 shows which region close to the defect as compared to

the ideal case, together with the magnetic moments. Since the point defect-

induced changes are efficiently screened by the conduction electrons, only the

83

nearest neighbor spins are affected

(a) Total DOS for defective and ideal (b) Ideal (c) Defective

Figure 5.10: LDA calculation of antisite defects in Co2MnSi(a) Total DOS for

defective (solid bold line) and ideal (dashed line) Co2MnSi with Co antisite in

Co2MnSi. The inset shows the minority DOS at 𝐸𝐹 projected on the different

neighbors (denoted as roman numbers) as one moves away from the Co antisite

defect. Magnetic moments for the ideal (b) and defective (c) systems for the

Co antisite in Co2MnSi around the defect. Values in bracket denote magnetic

moments for equivalent atoms in Co2MnGe [92].

Although Co antisites are theoretically expected to occur in concentra-

tions typically two orders of magnitude smaller than for the case of the Mn

antisite, experimentally these two defects are found to have the same den-

sity [262]. The Mn atom sitting on the Co position leads to a sharp peak in

the electron DOS located in proximity to the Fermi level [see Fig. 5.10 (a)] and

therefore destroys the half-metallicity. The calculated spin polarization for the

case presented in Fig 5.10 (a) is as low as 6%. The defect-induced states at the

Fermi level are spatially localized, as shown in the inset of Fig 5.10(a). The

analysis of magnetic moment [Fig 5.10(b)-(c)] shows that in the case of an Co

antisite defect the magnetic moments remain virtually unchanged and and are

couple ferromagnetically to the surrounding Co spins.

5.3.4 Atom exchange in Co2MnSi and Co2MnGe

Let us now discuss on atomic interchanges and consider Mn-Co exchange;

this defect can also be viewed as the sum of two different Mn and Co atomic

antisites that tend to aggregate. This defect shows a comparatively high for-

mation energy. However, this is of the same order of magnitude as the sum

of the separated defects (Mn and Co antisites) - which might show that point

84

(a) Total DOS for defective and ideal

(b) Ideal

(c) Defective

Figure 5.11: (a) Total DOS for defective (solid bold line) and ideal (dashed

line) Co2MnSi with Co-Mn swap in Co2MnSi. Magnetic moments for the

ideal(b) and defective (c) systems for the Co-Mn swap in Co2MnSi around the

defect. Values in bracket denote magnetic moments for equivalent atoms in

Co2MnGe [92].

defects have more or less the same probability to cluster, leading to this kind

of disorder, or to remain isolated [92]. The calculated minority DOS is shifted

to higher energies, along with a defect-induced peak located -0.2 eV below the

Fermi level (see Fig. 5.11 ), The majority DOS remains essentially unaffected.

Hence the half-metallic character is kept by the Co-Mn swaps. However, the

total magnetic moment per unite cell is drastically reduced by about 4 𝜇𝐵.

As shown by Table 5.1, the highest formation energy among the cases

studied is shown to be the Mn-Si exchange. Therefore, the rate of a noticeable

concentration of this kind of defect can be ruled out without any doubt. This

is consistent with some experiments which explain that the Si site is fully

occupied by Si, demonstrating that any disorder in Co2MnSi does not occupy

Si. On the other hand, Picozzi et al. [92] conclusions are at disagreement with

other first-principles calculations for the NiMnSb half-Heusler alloy. According

to Orgassa et al. [263], the Mn-Sb (analogous to the Mn-Si here considered)

disorder seems likely.

As far as the electronic and magnetic properties are concerned, a com-

parison between the total DOS [see Fig. 5.12 (a)] in the ideal and defective

cells shows only slight defect induced changes. In particular, in the energy

region around 𝐸𝐹 , the DOS is very similar for both minority- and majority

spin components, resulting in the same band-gap and half-metallic character

85

Table 5.1: Formation energy (in eV) and total magnetic moments (in Bohr

magnetons) for the different defects in Co2MnGe and Co2MnSi hosts [92].

Δ𝐸Co2MnSi 𝑀tot(Co2MnSi) Δ𝐸Co2MnGe 𝑀tot(Co2MnGe)

Co antisite 0.80 38.01 0.84 38.37

Mn antisite 0.33 38.00 0.33 38.00

Co-Mn swap 1.13 36.00 1.17 36.00

Mn-Si swap 1.38 40.00 – –

(a) Total DOS for defective and ideal

(b) Ideal

(c) Defective

Figure 5.12: (a) Total DOS for defective (solid bold line) and ideal (dashed line)

Co2MnSi with Mn-Si swap in Co2MnSi. Magnetic moments for (b) ideal and

(c) defective systems for the Mn-Si swap in Co2MnSi around the defect [92].

as for the pure bulk. As a result, the Mn-Si exchange systems illustrate a total

magnetic moment equal to that of the ideal system. This is consistent with

the atomic magnetic moments shown in Fig. 5.12 (b)-(c): within 0.04 𝜇𝐵, the

defect only results in a exchange between the Mn and Si magnetic moments.

Again, this illustrates that the first coordination shell, which in this case is ac-

curately the same as the one in the bulk for both the exchanged atoms, is the

most appropriate in the formation of the bonds, local charge and spin-density

rearrangement, and resulting magnetic moments [92].

5.3.5 Comparison between Co2MnSi and Co2MnGe

In order to compare the behavior of the defects considered in stoichiometric

CMS and CMG, Picozzi et al. [92] report in bracket in Figs. 5.9, 5.10, and

5.11 the value of the atomic magnetic moments in the region around the defect

site. There are two different kinds of hosts which globally show a very similar

behavior in terms of magnetic moments, suggesting that in the determination

86

of the magnetic properties the larger lattice constant or the smaller minority

band gap of the Ge-based compound are not really relevant. The most sensible

differences are shown quantitatively by the Co-antisite case: there is about a

0.4 𝜇𝐵 difference between the total magnetic moments in stoichiometric CMS

and CMG (See Table 5.1). Figure 5.10 demonstrates that this difference is

almost entirely due to differences in the central Co antisite, which changes

from about 20.9 𝜇𝐵 in stoichiometric CMS to about 21.4 𝜇𝐵 in stoichiometric

CMG, due to a very small difference in the energy position (<0.05 eV) of the

defect-induced peak with respect to 𝐸𝐹 [92].

From first principle calculation the DOS with respect to 𝐸𝐹 - and the

related energy position of the conduction-band minimum in the minority-spin

component- is very similar: for both stoichiometric CMS and CMG, the peak

is basically coincident with 𝐸𝐹 in the case of the Co antisite, whereas it lies

at about 20.2 eV in the case of the Co-Mn exchange. Except for very small

differences that can be traced back to differences in the ideal bulk hosts [264],

the calculated DOS for antisite and exchange defects in stoichiometric CMG

are not shown, due to very similar behavior to stoichiometric CMS.

5.3.6 Formula unit composition model for nonstoichio-

metric, Ge-deficient Co2Mn𝛽Ge0.38 films

We now discuss possible defects induced in the prepared nonstoichiometric,

Ge-deficient Co2Mn𝛽Ge0.38 films. Note that the Co:Ge ratio in the prepared

CMG films was 2:0.38, i.e., the Ge ratio was strongly deficient with respect to

the Co ratio. According to the theoretically predicted formation energies of

various kinds of defects for Co2MnSi [92, 249], with the assumption that the

relative magnitude of these formation energies for Co2MnGe is essentially the

same as that for Co2MnSi, we introduce a formula unit composition model for

strongly Ge-deficient CMG films with 𝛽 < 1.62 as Co2[Mn1−𝑥Co𝑥][Ge1−𝑦Mn𝑦],

where [Mn1−𝑥Co𝑥] and [Ge1−𝑦Mn𝑦] represent the nominal Mn and Ge sites,

respectively. The detailed reasons for this model are as follows: For Ge sites,

MnGe antisites, where a Ge site is replaced with a Mn atom, are likely to

be induced because a MnGe antisite has much lower formation energy than a

vacancy at a Ge site and a CoGe antisite [249]. Thus, Ge sites are fully occupied

by the normal Ge atoms and MnGe antisites. For Mn sites, CoMn antisites are

likely to be induced because a CoMn antisite has lower formation energy than

a vacancy at a Mn site [249]. Thus, the normal Mn atoms and CoMn antisites

should occupy Mn sites.

For 𝛽 > 1.62, MnCo antisite should be assumed in addition to MnGe

antisite because of the increased Mn composition with respect to the Co com-

87

position. Indeed, a MnCo antisite has much lower formation energy than a

vacancy at the Co site. Thus, our formula unit composition model for 𝛽 >

1.62 is [Co2−𝑥Mn𝑥][Mn][Ge1−𝑦Mn𝑦].

The Co:Mn:Ge ratio in the above models must be also equal to that in the

representation of Co2Mn𝛽Ge0.38 with a given 𝛽 value. From this requirement,

the values of 𝑥 and 𝑦 are determined. The details are as follows. For 𝛽 <

1.62, the formula unit (f.u.) composition model is Co2[Mn1−𝑥Co𝑥][Ge1−𝑦Mn𝑦].

In this model, the number of Co atom is (2+x), Mn atom is (1-x+y) and Ge

atom is (1-y) for a formula unit. The ratio of Co atom, Mn atom and Ge atom

must satisfy the relation of Co : Mn : Ge = 2: 𝛽 : 0.38. Thus, we obtain

the following relations, (2+𝑥) : (1-𝑥+𝑦) : (1-𝑦) = 2 : 𝛽 : 0.38. From these

relations we get two equations, (2+𝑥) : (1-𝑥+𝑦) = 2 : 𝛽, (2+𝑥) : (1-𝑦) = 2 :

0.38 or (1-𝑥+𝑦) : (1-𝑦) = 𝛽 : 0.38.

There are two unknown parameters, 𝑥 and 𝑦, and two equations. So,

the values of 𝑥 and 𝑦 can be determined by these two equations. However, we

must note that the 𝑥 and 𝑦 should be larger than 0 (𝑥, 𝑦 > 0)

For 𝛽 > 1.62, the formula unit model is [Co2−𝑥Mn𝑥][Mn][Ge1−𝑦Mn𝑦]. In

this model, the number of Co atoms per f.u. is (2-𝑥), that of Mn atoms is

(1+𝑥+𝑦) and that of Ge atoms is (1-𝑦). So, Co : Mn : Ge = (2-𝑥) : (1+𝑥+𝑦)

: (1-𝑦) = 2 : 𝛽: 0.38. Similarly, the two parameters 𝑥 and 𝑦 are determined

by two equations, (2-𝑥) : (1+𝑥+𝑦) = 2 : 𝛽 and (2-𝑥) : (1-𝑦) = 2 : 0.38 or

(1+𝑥+𝑦) : (1-𝑦) = 𝛽 : 0.38. Table 5.2 shows the formula unit compositions

(or chemical compositions) for nonstoichiometric, Ge -deficient Co2Mn𝛽Ge0.38films with various 𝛽 values provided by the models

Table 5.2: Formula unit compositions (chemical compositions) for nonstoichio-

metric Co2Mn𝛽Ge0.38 films according to the models..

𝛽 Formula unit model 𝑥 𝑦 Formula unit composition

0.67 Co2[Mn1−𝑥Co𝑥][Ge1−𝑦Mn𝑦] 0.623 0.502 Co2[Mn0.377Co0.623] [Ge0.498Mn0.502]

0.85 Co2[Mn1−𝑥Co𝑥][Ge1−𝑦Mn𝑦] 0.477 0.529 Co2[Mn0.523Co0.477] [Ge0.471Mn0.529]

1.03 Co2[Mn1−𝑥Co𝑥][Ge1−𝑦Mn𝑦] 0.346 0.554 Co2[Mn0.654Co0.346] [Ge0.446Mn0.554]

1.2 Co2[Mn1−𝑥Co𝑥][Ge1−𝑦Mn𝑦] 0.235 0.575 Co2[Mn0.765Co0.235] [Ge0.425Mn0.575]

1.4 Co2[Mn1−𝑥Co𝑥][Ge1−𝑦Mn𝑦] 0.116 0.598 Co2[Mn0.884Co0.116][Ge0.402Mn0.598]

1.6 Co2[Mn1−𝑥Co𝑥][Ge1−𝑦Mn𝑦] 0.010 0.618 Co2[Mn0.990Co0.010][Ge0.382Mn0.618]

1.8 [Co2−𝑥Mn𝑥][Mn][Ge1−𝑦Mn𝑦] 0.086 0.636 [Co1.914Mn0.086]Mn[Ge0.364Mn0.636]

This model provides formula units ranging from Co2[Mn0.377Co0.623] [Ge0.498Mn0.502]

(where 𝑥 is 0.62 ) for Co2Mn0.67Ge0.38 to Co2[Mn0.884Co0.116][Ge0.402Mn0.598]

(where 𝑥 is 0.12) for Co2Mn1.40Ge0.38. Even at 𝛽 = 1.03, for which the Co:Mn

ratio is close to 2:1, the model provides a formula unit of Co2[Mn0.654Co0.346]

88

[Ge0.446Mn0.554] that features a still high 𝑥 value of 0.35 for Co2Mn1.03Ge0.38.

This is because the Ge ratio is strongly deficient for Co2Mn𝛽Ge0.38. The similar

way we can also explain for Co2Mn𝛽Si0.88.

Thus, it was shown that the formula unit model based on the formation

energies of various kinds of defects reasonably explains the observed depen-

dence of saturation magnetization (𝜇𝑠) on 𝛽 [138]. This finding supports our

interpretation that the Mn composition dependence of the TMR ratio and

spin magnetization observed universally for CMG-MTJs is due to suppressed

minority-spin gap states around 𝐸𝐹 which are caused by the decreased CoMn

sites in Mn-rich CMG electrode. As we know that Picozzi et al. [92] works

for stoichiometric only down to Mn0.9 and Ge0.9. Since we have seen that our

result from 𝛽=0.67 to 𝛽=1 is consistent with Picozzi et al. [92] and similar ten-

dency was observed by Yamamoto et al. [138] for 𝛽=0.67 to 𝛽=1.40, thus, our

results support to Picozzi et al. [92] as well as Yamamoto et al. [138]. Hence

we justify that chemical stoichiometry and non-stoichiometry CMS and CMG

samples match with stoichiometric (only down to Mn0.9 and Ge0.9) as predicted

by Picozzi et al. [92] and the non-stoichiometry by Yamamoto et al. [138] and

thus my experimental results open the door for theoretician to investigate the

electronic structures of non-stoichiometric, Ge-deficient Co2Mn𝛽Ge0.38 from

first-principles.

5.4 Conclusion

We have studied the magnetic and electronic states of the CMS/MgO and

CMG/MgO interfaces by means of XMCD. By investigating the 𝛽 dependence

of XMCD of CMS/MgO and CMG/MgO interfaces. We obtained that the de-

duced 𝑚spin(Mn) for all the CMS samples decreases with Mn composition (𝛽)

which is consistent with Picozzi’s calculations which predicts the Mn magnetic

moment is now coupled antiferromagnetically at MnCo site, leading to a reduc-

tion of saturation magnetization. The deduced 𝑚spin(Co) for all the samples

was almost equal to a theoretical value of 1.06 𝜇𝐵. The Co spin magnetic mo-

ment was almost constant of the composition of Mn. Thus the Co-rich CMS

may be more or less the same amount of CoMn as in Co-rich CMG.

The deduced 𝑚spin(Mn) for all the CMG samples decreases with Mn

composition (𝛽), which is also consistent with Picozzi et al.’s calculations which

predicts the Mn magnetic moment is now coupled antiferromagnetically at

MnCo site, leading to a reduction of saturation magnetization. However, the

deduced 𝑚spin(Co) for all the CMG samples was also obviously larger and/or

equal to a theoretical bulk value of 1.06 𝜇𝐵. For Co-rich region, there is the

possibility of the existence of CoMn because 𝑚spin(Co)=1.5 𝜇𝐵 is larger than

89

1.06 𝜇𝐵 for Co at the regular Co site, CoCo. For Mn-rich region, 𝑚spin(Co) is

almost same as theoretical predicted value.

90

Chapter 6

Summary and Outlook

In this thesis, we have presented the results of soft x-ray spectroscopic

studies of TiO2-DMSs, BiFe3-multiferroic and Co2Mn𝑍-Heusler alloys. For

the Co-doped TiO2 thin film, present experimental results suggest that carrier-

mediated FM mechanism plays a key role to achieve high-𝑇C. For the TiO2-

DMS thin films, we find that the doped TM ions are in a Co2+ high-spin

state at surface. However, in bulk region the doped TM ions are in a Co2+

mixed-spin state.

In Chapter 3, to clarify the origin of the room-temperature ferromag-

netism of the Ti1−𝑥Co𝑥O2−𝛿 system, we have performed XAS, XMCD on rutile-

type Ti1−𝑥Co𝑥O2−𝛿 at the Co 𝐿2,3 edges (both in the TEY and TFY mode).

These results represent that the high temperature ferromagnetism is origi-

nated from the Co2+ atoms, most probably charge carriers induce the ferro-

magnetism. The magnetic moment of the Co ions as long as 0.82-2.25 𝜇𝐵/Co

was first observed by the bulk sensitive TFY method. The magnetic moment

value deduced with the TEY mode ( 0.15-0.24 𝜇𝐵/Co) indicates the presence

of a magnetically dead layer of ∼ 5 nm thickness on the sample surface. The

results give a strong evidence for intrinsic ferromagnetism in this compound

and unveil deep underlying physics of the room temperature ferromagnetism,

and will contribute to the implementation of semiconductor spintronics devices

operable at room temperature.

we have also provided experimental evidence for carrier-induced ferro-

magnetism of cobalt-doped anatase TiO2 thin films using XMCD at the Co

𝐿2,3 edges in both the TEY and TFY modes. The large magnetic moment of

the Co ions, 0.6-2.4 𝜇𝐵/Co, was observed by the bulk-sensitive TFY method.

The carrier-induced origin of ferromagnetism at room-temperature in anatase

Ti1−𝑥Co𝑥O2−𝛿 is confirmed on the basis of the element-specific XMCD study

at the surface as well as in bulk. In the bulk-sensitive TFY mode, the position

of Co2+ atoms seems to be displaced from the Ti4+ sites, resulting in more

91

random crystal fields. Good agreement is demonstrated not only in magneti-

zation and AHE but also in the magnetic field dependences of XMCD. The

magnetic moment values deduced with the TEY mode was < 0.3 𝜇𝐵/Co, in-

dicating the presence of a magnetically dead layer of ∼ 5 nm thickness at the

sample surfaces.

In Chapter 4, we have performed XAS and the XMCD measurements

of epitaxial BFCO thin films prepared by chemical solution deposition on

LaAlO3(001) substrate, which exhibit ferromagnetism at 300 K. The XAS

and XMCD line shape of Fe ions in BFCO thin films are independent of the

magnetic field, indicating that the ferromagnetism originates from the anti-

ferromagnetic coupled Fe3+ ions at the 𝑂ℎ and 𝑇𝑑 sites and ferromagnetism

behavior in the BFCO thin films is mainly arose from 𝛾-Fe2O3 like species as

a secondary phase (i.e. ferromagnetism behavior in the BFCO thin films is

mainly arose from extrinsic behavior). The magnetic moment of Fe increases

with Co content up to 20% and after that it decreases. The Co ions are in

the trivalent high-spin states and are largely paramagnetic . The Co magnetic

moment is nearly independent of Co content, unlike Fe, and the peak at 20%

Co has only a minor influence. Surface Fe ions in these films also showed a

significant enhancement compared with a bulk sample probed by XMCD. This

may be because of XMCD can detect the thickness of ∼ 5nm on the surface

and therefore detected larger magnetization than SQUID data.

In Chapter 5, We studied the magnetic and electronic states of Co2Mn𝛽Si0.88/MgO

magnetic tunnel junctions by CMS film-composition (𝛽) dependent XMCD

measurements. The XAS and XMCD spectral shapes for these samples were

similar to those for bulk CMS and neither the Mn nor Co atoms were oxi-

dized. By investigating the 𝛽 dependence of XMCD and magnetic moments of

CMSs, we have revealed that the magnetic states of the Mn and Co atoms at

interfacial region, facing to the MgO barrier. We obtained that the deduced

𝑚spin(Mn) for all the CMS samples decreases with Mn composition (𝛽) which

is consistent with Picozzi’s calculations which predicts the Mn magnetic mo-

ment is now coupled antiferromagnetically at MnCo site, leading to a reduction

of saturation magnetization. The deduced 𝑚spin(Co) for all the samples was

almost equal to a theoretical value of 1.06 𝜇𝐵. The Co spin magnetic moment

was almost constant of the composition of Mn. Thus the Co-rich CMS may

be more or less the same amount of CoMn as in Co-rich CMG.

We studied the magnetic and electronic states of Co2Mn𝛽Ge0.38/MgO

magnetic tunnel junctions by CMG film-composition (𝛽) dependent XMCD

measurements. The XAS and XMCD spectral shapes for these samples were

similar to those for bulk CMG and and neither the Mn nor Co atoms were

oxidized. The deduced 𝑚spin(Mn) for all the CMG samples decreases with

92

Mn composition (𝛽), which is also consistent with Picozzi et al.’s calculations

which predicts the Mn magnetic moment is now coupled antiferromagnetically

at MnCo site, leading to a reduction of saturation magnetization. However, the

deduced 𝑚spin(Co) for all the CMG samples was also obviously larger and/or

equal to a theoretical bulk value of 1.06 𝜇𝐵. For Co-rich region, there is the

possibility of the existence of CoMn because 𝑚spin(Co)=1.5 𝜇𝐵 is larger than

1.06 𝜇𝐵 for Co at the regular Co site, CoCo. For Mn-rich region, 𝑚spin(Co) is

almost same as theoretical predicted value.

Finally, future prospects of spintronics research are mentioned. Based

on a number of theoretical predictions [9, 265], room-temperature ferromag-

netism has been reported in DMS such as the TiO2-, ZnO-, and GaN-based

DMSs [183, 266–268]. In particular, there is growing interest in systems with

native defects such as DMS nano-particles because DMS nano-particles have

several advantages for the practical DMS as follows. Firstly, DMS nano-

particles have high surface-to-volume ratios and are amenable for surface mod-

ifications, which could be used to alter the electronic configuration [269]. Ac-

cording to the recent theoretical calculation of the DMS nano-particles [269],

the valence states of TM ions in the DMS nano-particles are changed by the

surface defects and thus magnetic interaction between the substitutional TM

ions changes. Secondly, detrimental substrate effects, such as mismatches in

thermal conductivities and lattice constants, are much weaker in the nano-

particles growth. Thirdly, the processing conditions used in nano-particles

synthesis are often different from thin-film deposition conditions, offering op-

portunities to tune the structural and chemical characteristics, which were

shown to exert significant influence on the magnetism of DMSs [270]. Finally,

DMS nano-particles are potential nano scale building blocks for spintronics

if they show robust magnetism [271]. Subsequently a number of experiments

on DMS nano-particles such as ZnS [272], ZnO [272] and TiO2 [176] revealed

ferromagnetic properties. In this thesis, we have indicated that the TM ions

in the DMS thin-films show small magnetization at surface, which are due

to surface dead layer and show large magnetization in bulk, and they play a

crucial role in ferromagnetism. The present work suggests that not only the

magnetic dopants themselves but also the defects are important to understnad

ferromagnetism of the DMS nano-particles. Hence, systematic study of the ef-

fects of particle’s size, shape, and surface structure on the magnetic properties

of various DMSs nano-particles such as ZnS, TiO2 and ZnO are desired.

In this thesis, we have also indicated that the Co-doped BiFeO3 (BFCO)

thin-films by chemical solution deposition method which show strong ferro-

magnetism at room-temperature. In this material, the most of the ferromag-

netism comes from 𝛾-Fe2O3 nanoparticles. Since BCFO is strong ferroelectric

93

so we may consider that BFCO/𝛾-Fe2O3 forms a nanocomposite and can show

a strong magnetoelectric coupling response. Hence, systematic study of the

magnetoelectric coupling response of various type of composites are necessary.

In addition to DMSs, we have studied the electronic and magnetic struc-

ture of the Heusler alloys. We find that XMCD measurements is a very pow-

erful tool to investigate the electronic configuration of these alloys. Spin-

resolved photoemission studies would also be desirable to directly observe the

half-metallic nature of Heusler alloys. We hope that the present work will pro-

mote soft x-ray spectroscopy on DMS’s and leads to further understandings of

physics of diluted magnetic semiconductor.

94

Acknowledgments

I would like to thank my sincere gratitude to my thesis supervisor Prof.

Atsushi Fujimori for his advice, guidance, patience and tremendous help with

during the project. His constant encouragement, and freedom to work with

anything and at any time made my life much easier and it helped me to work

more efficiently. It is a great honor for me to have been a part of his team at the

diluted magnetic semiconductor (DMS) laboratory, the successful completion

of the work has been only possible due to their excellent guidance, meticulous

observation and critical analysis.

I would also like to express my great thanks to Dr. Teppei Yoshida.

His valuable advices were indispensable to this work. I have learned a lot

from his energetic action and thoughtfulness. I would like to say my special

thanks to Dr. Takashi Kataoka and Mr.Yuta Sakamoto who have introduced

me the enchanted world of DMS. They always helped me and gave spurs to

me when I was in trouble and was ready to give up. They have always been

taking care of me throughout my PhD. I have been stimulated from his strong

will and energetic attitude to the work. I would also like to say my special

thanks to Dr. Walid Meleab and Mr. Shin-ichi Aizaki who were providing me

continuous encouragement. They always helped and took care of me. I have

been stimulated from their strong will and energetic attitude to the work. His

valuable advice was indispensable to this work.

The experiment at BL-16A of Photon Factory was supported by a number

of people, Prof. Tsuneharu Koide, Prof. Kenta Amemiya, and Dr. Daisuke

Asakura. I would like to thank their valuable technical support during the

beamtime. The experiment at SPring-8 was supported by a number of people.

I would like to thank the members : Dr. Yukiharu Takeda, Dr. Takuo Ohkochi,

Dr. Kouta Terai, Dr. Shin-ichi Fujimori, Dr. Tetsuo Okane, Dr. Yuji Saitoh

and Prof. Hiroshi Yamagami. The experiment at Taiwan Light Source were

supported by a number of people. I would also like to thank the members :

Mr. Fan-Hsiu Chang, Dr. Hong-Ji Lin, Prof. Di-Jing Huang and Prof. C. T.

Chen. I really appreciate their helpful technical support during the beamtime.

I would like to express my great thanks to Dr.H. Toyosaki, Dr.Y. Ya-

95

mada, Prof. T. Fukumura and Prof.M. Kawasaki. They provided us with the

high-quality samples of the Co-doped rutile TiO2 as well as Co-doped anatase

TiO2 thin films and warm encouragement. I am also very thankful to Dr.Y.

Nakamura, Prof.M. Azuma and Prof. Y. Shimakawa for providing us the inter-

esting samples BiFeO3:Co thin films and valuable discussions. I am also very

thankful to Dr. T. Ishikawa and Prof. M. Yamamoto for providing us the in-

teresting samples Co2MnSi and Co2MnGe thin films and valuable discussions.

I would like to thank Dr. Hirohiko Okamoto and Dr. Tohru Yamanouchi

for the guidance for the SQUID magnetometer and his support during the

magnetization measurements.

I am grateful to the members of Fujimori Group : Dr. T. Kadono,

Mr. Shin-ichiro Ideta, Mr.Viendra Kumar Verma, Mr. Keisuke Ishigami,

Mr. Leo Cristobal C Amblode II, Mr.Yo Yamazaki, Mr. Ichiro Nishi, Mr.

Wataru Uemura, Mr. Goro Shibata, Mr. Hakuto Suzuki and Mr. Takayuki

Harano. Ms. Miki Ueda, Ms. Kaoru Fukutomi, and Ms. Emiko Murayama for

their supports. Especially, Mr. Keisuke Ishigami, Mr. Yo Yamazaki, and

Mr. Ichiro Nishi helped me on many scenes in the research and their energetic

attitudes toward the research have always encouraged me. I have to express

my great thanks Mr. Virendra Kumar Verma who is a partaker of my joys

and sorrows in the life as a researcher. I have been stimulated from his strong

mind and brilliance. I was really glad to go through my life in Fujimori-group

with them. Not only they gave me valuable discussions but also helped me to

enjoy my life in Japan.

I would also like to greatly acknowledge the financial support from the

Ministry of Education, Culture, Sports, Science and Technology (MEXT),

Japan.

My stay in Tokyo makes a lot of friends who shared with me everything

and a lot of discussions. List is very long, in no particular order Dr. Md.

Rizwan, Dr. Md. Waseem Akhtar, Dr. Md. Rafi, Dr. Zainul aabdin Khan,

Dr. Gautam Singh, Dr. Anand Prakash, and Mr. Deepu. I would also like

to thank my friends Er. Praveen Kumar, Er. Vipin Tripathi, Er. Harish Kr.

Gangwar, Er. Ranbir Singh, Er. Prashant Gupta, Dr. Vinod Kumar, Mrs.

Alka Gupta and many more (sorry if I miss someone’s name!) who made my

days at University of Tokyo really enjoyable and cheerful.

Finally I would like to have an exclusive thanks to my family, especially

my parents for their love, support, and constant guidance and care throughout

my life. May God bless you two; my mom and dad, as I could never ever

thought of any two persons sacrificing and dedicating themselves to my success

so trustworthy and humbly and without any expectation what so ever. I wish

I could return all these kindness. My brother, sisters, sister-in-law, brother-

96

in-law and my consort (Ms. Chandrakiran Singh) I thank them very much for

your encouragement, company and support throughout the years. I will never

forget the support of Mr. D. R. Singh (elder brother), without his support it

would not have been possible to complete my Ph.D. He is really superb and

fantastic with his sense of humor. This kind of brother is born only once upon

a time. I still have to thank the rest of my friends, whom I can not name all,

but I am sure I owe them a lot for all the good times together.

Tokyo, June 2011

Vijay Raj Singh

97

References

[1] T. Dietl and H. Ohno, Materials Today 9, 18 (2006).

[2] T. Dietl, Semicond. Sci. Technol. 17, 377 (2002); J. K. Furdyna, J. Appl.

Phys 64, R29 (1988).

[3] H. Ohno, Science 281, 951 (1998).

[4] D. Ferrand, J. Cibert, A. Wasiela, C. Bourgognon, S. Tatarenko, G. Fish-

man, T. Andrearczyk, J. Jaroszynski, S. Kolesnik, T. Dietl, B. Barbara,

and D. Dufeu, Phys. Rev. B 63, 085201 (2001).

[5] A. Haury, A. Wasiela, A. Arnoult, J. Cibert, S. Tatarenko, T. Dietl, and

Y. Merle d’Aubigne, Phys. Rev. Lett. 79, 511 (1997).

[6] N. Samarth, Nature Mater. 6, 403 (2007).

[7] S. Kuroda, N. Nishizawa, K. Takita, M. Mitome, Y. Bando, K. Osuch,

and T. Dietl, Nature Mater. 6, 440 (2007).

[8] T. Jungwirth, K. Wang, J. Masek, K. Edmonds, J. Konig, J. Sinova, M.

Polini, N. Goncharuk, A. MacDonald, M. Sawicki, R. Campion, L. Zhao,

C. Foxon, and B. Gallagher, Phys. Rev. B 72, 165204 (2005).

[9] T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, Science 287,

1019 (2000).

[10] J. M. D. Coey, M. Venkatesan, and C. B. Fitzgerald, Nature Mater. 4,

173 (2005).

[11] D. H. Kim, J. S. Yang, K. W. Lee, S. D. Bu, T. W. Noh, S.-J. Oh, Y.-W.

Kim, J.-S. Chung, H. Tanaka, H. Y. Lee, and T. Kawai, Appl. Phys. Lett.

81, 2421 (2002).

[12] A. Chambers, Surf. Sci. Rep. 61, 345 (2006).

[13] Y. J. Lee, M. P. de Jong, and R. Jansen, Appl. Phys. Lett. 96, 082506

(2010).

98

[14] G. Karczewski, J. Supercond. 16, 55 (2003).

[15] H. Saito, V. Zayets, S. Yamagata, and K. Ando, Phys. Rev. Lett. 90,

207202 (2003).

[16] T. Story, R. R. Galazka, R. B. Frankel and P. A. Wolff, Phys. Rev. Lett.

56, 777 (1986).

[17] H. Ohno, H. Munekata, T. Penney, S. von Molnar, and L. L. Chang Phys.

Rev. Lett. 68, 2664 (1992).

[18] H. Ohno, A. Shen, F. Matsukura, A. Oiwa, A. Endo, S. Katsumoto, and

Y. Iye, Appl. Phys. Lett. 69, 363 (1996).

[19] S. Sonoda, S. Shimizu, T. Sasaki, Y. Yamamoto, and H.J. Hori, J.

Cryst.Growth 237, 1358 (2002).

[20] S. E. Park, H.-J. Lee, Y. C. Cho, S.-Y. Jeong, C. R. Cho, and S. Cho,

Appl. Phys. Lett 80, 4187 (2002).

[21] N. Theodoropoulou, A. F. Hebard, M. E. Overberg, C. R. Abernathy,

and S. J. Pearton, S. N. G. Chu, and R. G. Wilson, Phys. Rev. Lett. 89,

107203 (2002).

[22] X. Chen, M. Na, M. Cheon, S. Wang, H. Luo, B. D. McCombe, X. Liu,

Y. Sasaki, T. Wojtowicz, J. K. Furdyna, S. J. Potashnik, and P. Schiffer,

Appl. Phys. Lett. 81, 511 (2002).

[23] Y. D. Park, A. T. Hanbicki, S. C. Erwin, C. S. Hellberg, J. M. Sullivan,

J. E. Mattson, T. F. Ambrose, A. Wilson, G. Spanos, and B. T. Jonker,

Science 295, 651 (2002).

[24] Y. Matsumoto, M. Murakami, T. Shono, T. Hasegawa, T. Fukumura, M.

Kawasaki, P. Ahmet, T. Chikyow, S. Koshihara, and H. Koinuma, Science

291, 854 (2001).

[25] P. Sharma, A. Gupta, K. V. Rao, F. J. Owens, R. Sharma, R. Ahuja, J.

M. Osorio Guillen, B. Johansson, and G. A. Gehring, Nature Mater. 2,

673 (2003).

[26] S. B. Ogale, R. J. Choudhary, J. P. Buban, S. E. Lofland, S. R. Shinde,

S. N. Kale, V. N. Kulkarni, J. Higgins, C. Lanci, J. R. Simpson, N. D.

Browning, S. Das Sarma, H. D. Drew, R. L. Greene, and T. Venkatesan,

Phys. Rev. Lett. 91, 077205 (2003).

99

[27] J. M. D. Coey, A. P. Douvalis, C. B. Fitzgerald, and M. Venkatesan, Appl.

Phys. Lett. 84, 1332 (2004).

[28] J. Philip, A. Punnoose, B. I. Kim, K. M. Reddy, S. Layne, J. O. Holmes,

B. Satpati, P. R. LeClair, T. S. Santos, and J. S. Moodera, Nature Mater.

5, 298 (2006).

[29] Hadis Morkoc and Umit Ozgur, Zinc Oxide: Fundamental Materials and

device technology,(Wily-VCH Press, Berlin, 2009).

[30] P. Kacman, Semiconductor Science and Technology, 16, R25 (2001).

[31] C. Zener, Physical Review, 83, 299 (1951).

[32] C. Zener, Physical Review, 81, 440 (1951).

[33] C. Zener, Phys. Rev. 81, 440 (1950).

[34] A. H. MacDonald, R. Schiffer, and N. Samarth, Nature Mater. 4, 195

(2005).

[35] J. Schliemann, J. Konig, H. Lin, and A. H. MacDonald, Appl. Phys. Lett.

78, 1550 (2001).

[36] T. Jungwirth, J. Sinova, J. Masek, J. Kucera, and A. H. MacDonald, Rev.

Mod. Phys.78, 809 (2006).

[37] A. Kaminski, and S. Das Sarma, Phys. Rev. Lett. 88, 247202 (2002).

[38] M. Berciu, and R. N. Bhatt, Phys. Rev. Lett. 87, 107203 (2001).

[39] G. M. Dalpian, and S. Wei, Phys. Rev. B 73, 245204 (2006).

[40] J. M. D. Coey, Curr. Opin. Solid St. M. 10, 83 (2006).

[41] P. M. Krstajic, F. M. Peeters, V. A. Ivanov, V. Fleurov, K. Kikoin, Phys.

Rev. B 70, 195215 (2004).

[42] M. Yokoyama, H. Yamaguchi, T. Ogawa, and M. Tanaka, J. Appl. Phys.

97, 10D317 (2005).

[43] R. Janisch, P. Gopal, and N. A. Spaldin, J. Phys. Condens. Matter 17,

R657 (2005).

[44] S.A. Chambers, C.M. Wang, S. Thevuthasan, T. Droubay, D.E. Mc-

Cready, A.S. Lea, V. Shutthanandan, and C.F. Windisch Jr, Thin solid

films 418, 197 (2002).

100

[45] H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chikyow, T.

Hasegawa, H. Koinuma, and M. Kawasaki, Nat. Mater. 3, 221 (2004).

[46] K. Ueno, T. Fukumura, H. Toyosaki, M. Nakano, and M. Kawasaki, Appl.

Phys. Lett. 90, 072103 (2007).

[47] R. Ramaneti, J. C. Lodder, and R. Jansen, Appl. Phys. Lett. 91, 012502

(2007).

[48] T. Yamasaki, T. Fukumura, Y. Yamada, M. Nakano, K. Ueno, T. Makino

and M. Kawasaki, Appl. Phys. Lett. 94, 102515 (2009).

[49] T. Fukumura, H. Toyosaki, and Y. Yamada, Semicond. Sci. Technol. 20,

S103 (2005).

[50] H. Toyosaki, T. Fukumura, Y. Yamada, and M. Kawasaki, Appl. Phys.

Lett. 86,182503 (2005).

[51] S. R. Shinde, S. B. Ogale, J. S. Higgins, H. Zheng, A. J. Millis, V. N.

Kulkarni, R. Ramesh, R. L. Greene, and T. Venkatesan, Phys. Rev. Lett.

92, 166601 (2004).

[52] J.-Y. Kim, J.-H. Park, B.-G. Park, H.-J. Noh, S.-J. Noh, J. S. Yang, D.-H.

Kim, S. D. Bu, T.-W. Noh, H.-J. Lin, H.-H. Hsieh, and C. T. Chen, Phys.

Rev. Lett. 90, 017401 (2003).

[53] K. Mamiya, T. Koide, A. Fujimori, H. Tokano, H. Manaka, A. Tanaka, H.

Toyosaki, T. Fukumura, and M. Kawasaki, Appl. Phys. Lett. 89, 062506

(2006).

[54] S. Lany, H. Raebiger, and A. Zunger, Phys Rev. B 77, 241201(R) (2008).

[55] L. H. Ye, and A. J. Freeman, Phys Rev. B 73, 081304(R) (2006).

[56] J. W. Quilty, A. Shibata, J. -Y. Son, K. Takubo, T. Mizokawa, H.

Toyosaki, T. Fukumura, and M. Kawasaki, Phys. Rev. Lett. 96, 027202

(2006).

[57] Van E. Wood and A. E. Austin, Magnetoelectric interaction phenomena

in crystals, edited by A. J. Freeman and H. Schmid, Gordon and Breach

Science Publishers,(1975).

[58] G. A. Smolenskii and I. E. Chupis, Sov. Phys. Usp. 25, 475, (1982).

[59] I. E. Dzyaloshinsky, J. Exptl. Theor. Phys. (USSR) 37, 881, (1959) [Sov.

Phys. JETP 37, 628(1960)].

101

[60] J. Van Suchtelen, Philips Res. Rep. 27, 28, (1972).

[61] F. Kubel, and H.Schmid, Acta Crystallogr., Sect. B: Struct. Sci. B46,

698, (1990).

[62] C.Michel, J. -M. Moreau, G. D. Achenbach, R. Gerson, W. J. James Solid

State Commun. 7, 701, (1969).

[63] F. Kubel and H.Schmid, Acta Crystallogr., Sect. B: Struct. Sci. B46, 698

(1990).

[64] I. Sosnowska, T. Peterlin-Neumaier and E. Steichele, J. Phys. C: Solid

State Phys. 15, 4835, (1982).

[65] B. Ruette, M S thesis, Virginia Tech (2003).

[66] James R. Teague, Robert Gerson and W. J. James, Solid State Commu-

nications 8, 1073, (1970).

[67] K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett. 79, 988, (2001).

[68] J. Wang, J.B. Neaton, H. Zheng, V. Nagarajan, S.B. Ogale, B. Liu,

D. Viehland, V. Vaithyanathan, D.G. Schlom, U.V. Waghmare, N.A.

Spaldin, K.M. Rabe, M. Wutting, Science 299, 1791, (2003).

[69] M. Fiebig, T. Lottermoser, D. Frohlich, A.V. Goltsev, R.V. Pisarev, Na-

ture 419, 81, (2002).

[70] V. R. Palkar, J. John, and R. Pinto, Appl. Phys. Lett. 80, 1628, (2002).

[71] K. Y. Yun, M. Noda, M. O. Kuyama, Appl. Phys. Lett. 83, 3981, (2003)

.

[72] Y. P. Wang, L. Zhou, M. F. Zhang, X. Y. Cheng, J.-M. Liu, Z.G. Liu,

Appl. Phys. Lett. 84, 1731, (2004).

[73] K. Y. Yun, D. Ricinschi, T. Kanashima, M. Noda, M.Okuyama, Jpn. J.

Appl. Phys., Part 2 43,L647, (2004).

[74] J. Jackson, S. B. Palmer, J. Phys. D: Appl. Phys. 27, 1581, (1994).

[75] I. Sosnowska, M. Loewenhaupt, W. I. F. Davie, R. M. Ibberson, Physica

B 180, 117, (1992).

[76] F. Popov, A. M. Kadomtseva, S. S. Krotov, D. V. Belov, G. P. Vorob’ev,

P. N.Makhov, and A. K. Zvezdin, Low Temperature Physics, 27, 478,

(2001).

102

[77] J.F.Nye, Physical Properties of Crystals,(Oxford university Press, Oxford,

1960).

[78] K. C. Kao, Dielectric Phenomena in Solid,(Elsevier Academic Press, USA,

2004).

[79] W. Cochran, Advan. Phys. 9, 387, (1960); 10, 401, (1961).

[80] P. W. Anderson, Fizika dielektrikov (Akad. Nauk, SSSR, Moscow, (1959).

[81] C. Kittel, Introduction to Solid State Physics, (1976).

[82] A. J. Dekker, Solid State Physics (Macmillan India Limited Press, India,

(2000).

[83] C.S.Smith, Macroscopic Symmetry and Properties of Crystals in Solid

State Physics, Vol 6 (Academic Press New York) (1958).

[84] Y. Nakamura, M. Kawai, M. Azuma, and Y. Shimakawa, Jpn. J. Appl.

Phys. 49, 051501, (2010).

[85] I. E. Dzyaloshinsky, J. Exptl. Theor. Phys. (USSR) 37, 881, (1959).

[86] Van Suchtelen, Philips Res. Rep. 27, 28, (1972).

[87] M. Fiebig, J. Phys. D: Appl. Phys. 38, 123, (2005) .

[88] H. Schmid, Ferroelectrics 162, 317, (1994).

[89] F. Heusler. Verh. Dtsch. Phys. Ges 5, 219, (1903).

[90] P. J. Ziebeck and K. R.A. Webster, Landolt-Bornstein New Series II/19c.

(1986).

[91] I. Galanakis, P. H. Dederichs and N. Papanikolaou, Phys. Rev. B 66,

174429, (2002).

[92] S. Picozzi, A. Continenza and A. J. Freeman, Phys. Rev. B 69, 094423,

(2004).

[93] S. Ishida, S. Fujii, S. Kashiwagi, and S. Asano, J. Phys. Soc. Jpn. 64,

2152 (1995).

[94] S. Picozzi, A. Continenza, and A. J. Freeman, Phys. Rev. B 66, 094421

(2002).

[95] Y. Miura, K. Nagao and M. Shirai, Phys. Rev. B 69, 144413, (2004).

103

[96] P. J. Webster and K. R. A. Ziebeck, Alloys and Compounds of d-Elements

with Main Group Elements - Part 2. H. P. J. Wijn, Landolt-Bornstein,

New Series, Group III, 19/c (Springer, Berlin), (1988).

[97] P. J. Webster, K. R. A. Ziebeck, and K.-U. Neumann, Magnetic Properties

of Metals., New Series, Group III, 32/c (Springer, Berlin), (2001).

[98] J. Pierre, R. V. Skolozdra, J. Alloys Comp. 101, 262, (1997).

[99] J. Tobola, J. Pierre, S. Kaprzyk, R. Skolozdra, and M. Kouacou J. Phys.:

Condens. Matter 10, 1013, (1998).

[100] K. R. A. Ziebeck and P.J. Webster, J. Phys. F 5, 1756, (1975).

[101] P. J. Eewster and R.S. Tebbel, J. Appl. Phys. 39, 471, (1968).

[102] P. J. Eewster, J. Phys. Chem Sol. 32, 1221, (1971).

[103] K. H. J. Buschow and P.G. Van Engen, J. Magn. Mag. Mat. 25, 90,

(1983).

[104] J. Kuebler, A. R. Williams, and C. B. Sommers Phys. Rev. B 28, 4,

(1983).

[105] I. Galanakis, P. H. Dederichs, Half-metallic alloys-Fundamental and Ap-

plications, Lecture Notes in Physics. Springer, (2005).

[106] K. H. J. Buschow and M. Erman, J. Mag. Magn. Mater. 30, 374, (1983).

[107] J. Pierre, R. V. Skolozdra, and Yu. V. Stadnyk, J. Mag. Magn. Mater.

128, 93, (1993).

[108] R. A. de Groot, F. M. Mueller, P. G. van Engen, and K. H. J. Buschow

Phys. Rev. Lett. 50, 2024, (1983).

[109] S. Fujii, S. Ishida, and S. Asano, J. Phys.: Cond. Matter 2, 8583, (1990).

[110] S. Ishida, S.Kashiwagi, S. Fuji, and S. Asano, Physicaa B 210, 140,

(1995).

[111] K. E.H.M. Hansen, P. E. Mijnarends, L. P. L. M. Rabou, K. H. J.

Buschow, Phys. Rev. B 42, 1533, (1990).

[112] M. J. Otto, R. A. M. van Woerden, P. J. van der Valk, J. Wijngaard, C.

F. van Bruggen and C. Haas, J. Phys.: Cond. Matter 1, 2351, (1989).

104

[113] R. J. Soulen Jr., J. M. Byers, M. S. Osofsky, B. Nadgorny, T. Ambrose,

S. F. Cheng, P. R. Broussard, C. T. Tanaka, J. Nowak, J. S. Moodera, A.

Barry and J. M. D. Coey, Science 282, 85, (1998).

[114] P.J.Brown, K.-U. Neumann, P.J. Webster and K.R.A. Ziebeck, J. Phys.:

Cond. Matter 12, 1827, (2000).

[115] L. Ritchie, G. Xiao, Y. Ji, T. Y. Chen, C. L. Chien, M. Zhang, J. Chen,

Z. Liu, and G. Wu, Phys. Rev. B 68, 104430, (2003).

[116] D. Ristoiu, J. P. Nozieres, C. N. Borca, T. Komesu, H.-k. Jeong and P.

A. Dowben, Europhys. Lett. 49, 624, (2000).

[117] W. Zhu, B. Sinkovic, E. Vescovo, C. Tanaka and J. S. Moodera, Phys.

Rev. B 64, 060403(R) (2001).

[118] G. A. Prinz, Science 282, 1660, (1998).

[119] D. Awschalom and J.M. Kikkawa, Phys. Today 52, 33, (1999).

[120] I. Zutic, J. Fabian, S. D. Sarma, Rev. Mod. Phys. 76, 323, (2004).

[121] S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Daughton, S. von

Molnar, M. L. Roukes, A. Y. Chtchelkanova and D. M. Treger, Science

294, 5546, (2001).

[122] G. Schmidt, D. Ferrand, L. W. Molenkamp, A. T. Filip and B. J. van

Wees, Phys. Rev. B 62, R4790, (2000).

[123] C. Cacho,Y. Lassailly1, H.-J. Drouhin, G. Lampel, and J. Peretti, Phys.

Rev. Lett., 88, 066601 (2002).

[124] B. Dieny, J. Mag. Magn. Mater. 136, 335, (1994).

[125] J. Moodera, Lisa R. Kinder, Terrilyn M. Wong, and R. Meservey, Phys.

Rev. Lett. 74, 3273 (1998).

[126] I. Galanakis and N. Papanikolaou, Phys. Rev. B 67, 104417, (2003).

[127] A. Stroppa, S. Picozzi, A. Continenza, and A. J. Freeman, Phys. Rev. B

68, 155203 (2003).

[128] H. Akai, Phys. Rev. Lett. 81, 3002, (1998).

[129] S. Picozzi, A. Continenza, and A. J. Freeman, Half-metallic alloys-

Fundamental and Applications, Lecture Notes in Physics. Springer,

(2005).

105

[130] R. Meservey, D. Paraskevopoulos, and P. M. Tedrow Phys. Rev. Lett.

37, 858,, (1976).

[131] C. Felser, B. Heitkamp, F. Kronast, D. Schmitz, S. Cramm, H. A. Durr,

H.-J. Elmers, G. H. Fecher, S. Wurmehl, T. Block, D. Valdaitsev, S. A.

Nepijko, A. Gloskovskii, G. Jakob, G. Schonhense and W. Eberhardt J.

Phys.: Cond. Matter 15, 7019, (2003).

[132] M.P. Raphsel, B. Ravel, Q. Huang, M. A. Willard, S. F. Cheng, B. N.

Das, R. M. Stroud, K. M. Bussmann, J. H. Claassen, and V. G. Harris,

Phys. Rev. B 66, 104429, (2002).

[133] M. Julliere, Phys. Lett. 54A, 225, (1975).

[134] C. T. Tanaka, J. Nowak, and J. S. Moodera, J. Appl. Phys. 86, 6239,

(1999).

[135] J. Schmalhorst, S. Kammerer, M. Sacher, G. Reiss, A. Hutten and A.

Scholl, Phys. Rev. B 70, 024426, (2004).

[136] A. Hutten, Phys. Stat. Sol(a) 201, 3271, (2004).

[137] A. Hutten, Half-metallic alloys-Fundamental and Applications, Lecture

Notes in Physics. Springer, (2005).

[138] M. Yamamoto, T. Ishikawa, T. Taira, G.-F. Li, K.-ichiMatsuda and T.

Uemura. J. Phys.: Condens. Matter 22, 164212, (2010).

[139] M. Yamamoto, T. Marukame, T. Ishikawa, K. Matsuda, T. Uemura and

M. Arita, J. Phys. D: Appl. Phys. 39, 824 (2006).

[140] T. Marukame, T. Kasahara, K-I Matsuda, T. Uemura and M. Yamamoto

Jpn. J. Appl. Phys. 44, L521 (2005).

[141] T. Marukame, T. Ishikawa, S. Hakamata, K-I. Matsuda, T. Uemura and

M. Yamamoto IEEE Trans. Magn. 43, 2782 (2007).

[142] K. Inomata, S. Okamura, A. Miyazaki, M. Kikuchi, N. Tezuka, M. Wo-

jcik and E. Jedryka, J. Phys. D: Appl. Phys. 39, 816, (2006).

[143] E. Clifford, M. Venkatesan, R. Gunning, J. M. D. Coey, Solid State

Commun. 131, 61, (2004).

[144] M. Oogane, Y Sakuraba, J Nakata, H Kubota, Y Ando, A Sakuma and

T Miyazaki, J. Phys. D: Appl. Phys. 39, 834, (2006).

106

[145] E. E. Fullerton, M. J. Conover, J. E. Mattson, C. H. Sowers, and S. D.

Bader, Appl. Phys. Lett. 63, 1699, (1993).

[146] R. Fiederling, M. Keim, G. Reuscher, W. Ossau, G. Schmidt, A. Waag

and L. W. Molenkamp, Nature 76, 323, (1999)

[147] Y. Ohno, D. K. Young, B. Beschoten, F. Matsukura, H. Ohno and D.

D. Awschalom, Nature 402, 790, (1999).

[148] X. Jiang, R. Wang, S. van Dijken, R. Shelby, R. Macfarlane, G. S.

Solomon, J. Harris, and S. S. P. Parkin, Phys. Rev. Lett. 90, 256603,

(2003).

[149] J. Strand, B. D. Schultz, A. F. Isakovic, C. J. Palmstrom, and P. A.

Crowell, Phys. Rev. Lett. 91, 036602, (2003).

[150] X.Y.Dong, C. Adelmann, J. Q. Xie, C. J. Palmstrom, X. Lou, J. Strand,

P. A. Crowell, J.-P. Barnes, and A. K. Petford-Long, Appl. Phys. Lett.

86, 102107, (2005).

[151] G. A. de Wijs and R. A. de Groot, Phys. Rev. B 64, 020402, (2001).

[152] A. Debernadi, M. Peressi, and A. Baldereschi, Mat. Sci. Eng. C 23, 743,

(2003).

[153] I. Galanakis, Ph Mavropoulos and P H Dederichs, J. Phys. D: Appl.

Phys., 39, 765, (2006).

[154] S. Picozzi, A. Continenza, and A. J. Freeman, J. Appl. Phys 94, 4723,

(2003).

[155] S. J. Jenkins, Phys. Rev. B 70, 245401, (2004).

[156] In P.A. Dowben and A. Miller, editors, Surface Segregation Phenomena,

Page 145, CRC Press, Boston, (1990)

[157] S. Yamamoto, H. Kawata, H. Kitamura, M. Ando, N. Saki, and N. Sh-

iotani, Phys. Rev. Lett. 62, 2672, (1988).

[158] H. Onuki, N. Saito, and T. Saito, Appl. Phys. Lett. 52, 173, (1988).

[159] B. T. Thole, P. Carra, F. Sette, and G. van der Laan, Phys. Rev. Lett.

68, 1943, (1992).

[160] P. Carra, B. T. Thole, M. Altarelli, and X. Wang, Phys. Rev. Lett. 70,

694, (1993).

107

[161] C.T. Chen, Y. U. Idzerda, H.-J. Lin, N. V. Smith, G. Meigs, E. Chaban,

G. H. Ho, E. Pellegrin, and F. Sette, Phys. Rev. Lett. 75, 152, (1995).

[162] A. Yokoya, T. Sekiguchi, Y. Saitoh, T. Okane, T. Nakatani, T. Shimada,

H. Kobayashi, M. Takao, Y. Teraoka, Y. Hayashi, S. Sasaki, Y. Miyahira,

T. Harami, and T. A. Sasaki, J. Synchrotron Rad. 5, 10, (1998).

[163] Y. Saitoh, T. Nakatani, T. Matsushita, A. Agui, A. Yoshigoe, Y. Teraoka,

and A. Yokoya, Nucl. Inst. Meth. A 474, 253, (2001).

[164] J. Okamoto, K. Mamiya, S.-I. Fujimori, T. Okane, Y. Saitoh, Y. Mura-

matsu, A. Fujimori, S. Ishikawa, and M. Takano, AIP Conf. Proc. 705,

1110, (2004).

[165] A. E. Turner, R. L. Gunshor, and S. Datta, Appl. Opt. 22, 3152, (1983).

[166] H. Ohno, F. Matsukura, and Y. Ohno, Solid State Commun. 119, 281,

(2001).

[167] H. Ohno, J. Magn. Magn. Mater. 200, 110, (1999).

[168] H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chikyow, T.

Hasegawa, K. Koinuma, and M. Kawasaki, Nat. Mat 3, 221 (2004).

[169] T. Fukumura, H. Toyosaki, K. Ueno, M. Nakano, and M. Kawasaki, New

J. Phys. 10 , 055018, (2008).

[170] K. Mamiya, T. Koide, A. Fujimori, H. Tokano, Appl. Phys. Lett. 89,

062506, (2006).

[171] V. R. Singh, Y. Sakamoto, T. Kataoka, M. Kobayashi, Y. Yamazaki, A.

Fujimori, F.-H. Chang, D.-J.Huang, H.-J. Lin, C.T. Chen, H.Toyosaki,

T. Fukumura and M.Kawasaki, J. Phys.: Condens. Matter 23, 176001,

(2011).

[172] H. Saito, V. Zayets, S. Yamagata, and K. Ando, Phys. Rev. Lett. 90,

207202, (2003).

[173] M. L. Reed, N. A. El-Masry, H. H. Stadelmaier, M. K. Ritums, M. J.

Reed, C. A. Parker, J. C. Roberts, and S. M. Bedair, Appl. Phys. Lett.

79, 3473, (2001).

[174] S. G. Yang, A. B. Pakhomov, S. T. Hung, and C. Y. Wong, Appl. Phys.

Lett. 81, 2418, (2001).

108

[175] K. A. Griffin, A. B. Pakhomov, C. M. Wang, S. M. Heald, K. M. Krish-

nan, Phys. Rev. Lett 94, 157204, (2005).

[176] S. R. Shinde, S. B. Ogale, S. Das Sarma, J. R. Simpson, H. D. Drew,

S. E. Lofland, C. Lanci, J. P. Buban, N. D. Browning, V. N. Kulkani, J.

Higgins, R. P. Sharma, R. L. Greene, and T. Venkatesan, Phys. Rev. B

67, 115211, (2003).

[177] P. A. Stampe, R. J. Kennedy, Y. Xin, and J. S. Parker, J. Appl Phys.

92, 7114, (2002).

[178] D. H. Kim, J. S. Yang, K. W. Lee, S. D. Bu, T. W. Noh, S.-J. Oh, Y.-W.

Kim, Y.-S. Chung, H. Tanaka, H. Y. Lee, and T. Kawai, Appl. Phys. Lett.

81, 2421, (2002).

[179] J. -Y. Kim, J. H. Park, B. -G. Park, H. -J. Noh, S. -J. Oh, J. S. Yang, D.

-H. Kim, S. D. Bu, T.-W. Noh, H. -J. Lin, H. -H. Hsieh and C. T. Chen,

Phys. Rev. Lett. 90, 017401 (2003).

[180] M. J. Calderon, S. Das. Sarma, Ann. Phys. 322, 2618, (2007).

[181] Y.Yamada, K.Ueno, T. Fukumura, H. T. Yuan, H. Shimotani, Y. Iwasa,

L. Gu, S. Tsukimoto, Y. Ikuhara, M. Kawasaki, accepted for Science.

[182] T. Ohtsuki, A. Chainani, R. Eguchi, M. Matsunami, Y. Takata, M.

Taguchi, Y. Nishino, K. Tamasaku, M. Yabashi, T. Ishikawa, M. Oura,

Y. Senba, H. Ohashi, and S. Shin, Phys. Rev. Lett. 106, 047602, (2011).

[183] H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chikyow, T.

Hasegawa, H.Koinuma, and M. Kawasaki, Nat. Mat. 3 , 221 (2004).

[184] H. Toyosaki, T. Fukumura, Y. Yamada, and M. Kawasaki, Appl. Phys.

Lett. 86, 182503 (2005).

[185] M. Nakano, T. Fukumura, K. Ueno, and M. Kawasaki, to be submitted.

[186] T. Matsumura, D. Okuyama, S. Niioka, H. Ishida, T. Satoh,1 Y. Mu-

rakami, H.Toyosaki, Y. Yamada, T. Fukumura, and M. Kawasaki, Phys.

Rev. B 76, 115320 (2007).

[187] The calculated spectra are the weighted sum of 38 % of 𝐷2ℎ low-spin,

38 % of 𝑂ℎ low spin and 24 % of 𝑂ℎ high-spin crystal field. This result

again suggests that the majority of the ferromagnetic Co ions in bulk

Ti1−𝑥Co𝑥O2 have Co2+ with random crystal fields.

109

[188] C. T. Chen, Y. U. Idzerda, H.J. Lin, N. V. Smith, G. Meigs, E. Chaban,

G. H. Ho, E. Pellegrin and F. Sette, Phys. Rev. Lett. 75, 152 (1995).

[189] Y. Teramura, A. Tanaka, and T. Jo, J. Phys. Soc. Jpn. 65, 1053 (1996).

[190] T. Fukumura, Y. Yamada, K. Tamura, K. Nakajima, T. Aoyama, A.

Tsukazaki, M. Sumiya, S. Fuke, Y. Segawa, T. Chikyow, T. Hasegawa,

H. Koinuma, and M. Kawasaki, Jpn. J. Appl. Phys., Part 2 42, L105

(2003); Y. Yamada, H. Toyosaki, A. Tsukazaki, T. Fukumura, K. Tamura,

Y. Segawa, K. Nakajima, T. Aoyama, T. Chikyow, T. Hasegawa, H.

Koinuma, and M. Kawasaki, J. Appl. Phys. 96, 5097 (2004).

[191] H. Toyosaki, T. Fukumura, Y. Yamada, and M. Kawasaki, Appl. Phys.

Lett. 86, 182503 (2005).

[192] H. Weng, J. Dong, T. Fukumura, M. Kawasaki, and Y. Kawazoe, Phys.

Rev. B 73, 121201(R) (2006).

[193] H. P. R. Frederikse, J. Appl. Phys. 32, 2211 (1961).

[194] The calculated spectra are the weighted sum of 35% of 𝐷2ℎ low spin

(LS), 35% of 𝑂ℎ low spin (LS), 30% of 𝑂ℎ high spin (HS).

[195] M. Nakano, Y. Yamada, T. Fukumura, K.Ueno, and M. Kawasaki, to be

submitted.

[196] The calculated spectra are the weighted sum of 35% of 𝐷2ℎ low spin

(LS), 35% of 𝑂ℎ low spin (LS), 30% of 𝑂ℎ high spin (HS).

[197] N.Yamashita, T. Sudayama, T. Mizokawa, Y.Yamada, T. Fukumura and

M. Kawasaki, Appl. Phys. Lett. 96, 021907 (2010).

[198] M. Fiebig, J. Phys. D 38, R123 (2005).

[199] N. A. Hill, J. Phys. Chem. B 104, 6694 (2000).

[200] T. Kimura, T. Goto, H. Shintani, K. Ishizaka, T. Arima, and Y. Tokura,

Nature (London) 426, 55 (2003).

[201] W. Eerenstein, N. D. Mathur, and J. F. Scott, Nature (London) 442,

759 (2006).

[202] G. A. Smolenskii, I.Chupis, Sov. Phys. Usp 25,475 (1982).

[203] Y. F. Popov, A. K. Zvezdin, G. P. Vorobfev, A. M. Kadomtseva, V. A.

Murashev, and D. N. Rakov, JETP Lett.57, 69 (1993).

110

[204] Y. F. Popov, A. M. Vorob, G. P. Ev and A. K. Zvezdin Ferroelectric

162,135 (1994).

[205] A.V. Murashov, D. N. Rakov,V. M. Ionov, I. S. Dubenko,Y. V. Titov

and V. S. Gorelik Ferroelectric 162,11 (1994).

[206] F. Bai, J. Wang, M. Wang, J. Li, N. Wang Appl. Phys. Lett.86, 32511

(2005).

[207] W. Prellier, M. P. Singh and P. M.Urugavel J. Phys: Condens Mattet

17, R803 (2005).

[208] J. Wang et.al., Science 299, 1719 (2003) .

[209] N. Itoh, T. Shimura, W. Sakamoto and T. Yogo, Ferroelectrics 356,19

(2007).

[210] T. Higuchi, T. Hattori, W. Sakamoto, N. Itoh, T. Shimura, T. Yogo, P.

Yao, Y.-S. Liu, P.-A. Glans, C. Chang, Z. Wu, and J. Guo, Jpn. J. Appl.

Phys. 47, 7570 (2008).

[211] J. Wang, J. B. Neaton2, H. Zheng1, V. Nagarajan, S. B. Ogale, B. Liu,

D. Viehland, V. Vaithyanathan, D. G. Schlom, U. V. Waghmare, N. A.

Spaldin, K. M. Rabe, M. Wuttig and R. Ramesh, Science 307, 1203b

(2005).

[212] P. Yu, J.-S. Lee, S. Okamoto, M. D. Rossell, M. Huijben, C.-H. Yang,

Q. He, J. X. Zhang, S.Y. Yang, M. J. Lee,Q. M. Ramasse, R. Erni, Y.-H.

Chu, D. A. Arena, C.-C. Kao, L.W. Martin, and R. Ramesh, Phys. Rev.

Letts. 105, 027201 (2010).

[213] S. Ryu, J.-Y. Kim, Y.-H. Shin, B. ?G. Park, J. Y. Son, and H. M. Jang,

Chem. Mater. 21, 5050 (2009) .

[214] Y. Nakamura, M. Kawai, M. Azuma, and Y. Shimakawa, Jpn. J. Appl.

Phys. 49, 051501 (2010) .

[215] W. Sakamoto, H. Yamazaki, A. Iwata, T. Shimura and T. Yogo, Jpn. J.

Appl. Phys. 45, 7315 (2006).

[216] T. J. Regan, H. Ohldag, C. Stamm, F. Nolting, J. Luning, and J. Stohr,

R. L. White, Phys. Rev. B, 64, 214422 (2001).

[217] J. Chen, D. J. Huang, A. Tanaka, C. F. Chang, S. C. Chung, W. B. Wu,

and C. T. Chen, Phys. Rev. B 69, 085107 (2004).

111

[218] W. Eerenstein, F. D. Morrison, J. Dho, M. G. Blamire, J. F. Scott, N.

D. Mathur, Science 307, 1203a (2005).

[219] J.W. Lin, Y. H. Tang, C. S. Lue, and J. G. Lin, Appl. Phys. Letts. 96,

232507 (2010).

[220] C. T. Chen, Y. U. Idzerda, H.-J. Lin, N. V. Smith, G. Meigs, E. Chaban,

G. H. Ho, E. Pellegrin, and F. Sette, Phys. Rev. Letts. 75, 152 (1995).

[221] J.-Y. Kim, T. Y. Koo, and J.-H. Park: Phys. Rev. Letts. 96, 047205

(2006).

[222] S. B. Profeta, M. A. Arrio, E. Tronc, I. Letard, C. C. D. Moulin, and P.

Sainctavit, Phys. Scr. T115, 626 (2005).

[223] P. Singh, Y.A. Park, K. D. Singh, N. Hur, J. H. Jung, W.-S. Noh, J.-Y.

Kim, J. Yoon, Y. Jo, Solid State Commun.150 431(2010).

[224] J. M. D. Coey, A. P. Douvalis, C. B. Fitzgerald, and M. Venkatesan,

Appl. Phys. Lett. 84, 1332 (2004).

[225] C. Ederer and N. A. Spaldin, Phys. Rev. B, 71, 060401(R) (2005).

[226] S. S. Lee, J. H. Kim, S. C. Wi, G. Kim, J.-S. Kanga, Y. J. Shin, S. W.

Han, K. H. Kim, H. J. Song and H. J. Shin, J. Appl. Phys. 97, 10A309

(2005).

[227] M.W. Haverkort, Z. Hu, J. C. Cezar, T. Burnus, H. Hartmann, M.

Reuther, C. Zobel, T. Lorenz, A. Tanaka, N. B. Brookes, H. H. Hsieh,

H.-J. Lin, C. T. Chen, and L. H. Tjeng, Phys. Rev. Letts. 97, 176405

(2006).

[228] J. B. Goodenough, Phys. Rev. 100, 564 (1955).

[229] J. B. Goodenough, J. Phys. Chem. Solids 6, 287 (1958).

[230] J. Kanamori, J. Phys. Chem. Solids 10, 87 (1959).

[231] S. Yuasa, T. Nagahama, A. Fukushima, Y. Suzuki, and K. Ando, Nature

Mater. 3, 868 (2004).

[232] R. A. de Groot, F. M. Mueller, P. G. van Engen, and K. H. J. Buschow,

Phys. Rev. Lett. 50, 2024 (1983).

[233] M. I. Katsnelson, V. Yu. Irkhin, L. Choicel, A. I. Lichtenstein, and R.

A. de Groot, Rev. Mod. Phys. 80, 315 (2008).

112

[234] S. Hakamata, T. Ishikawa, T. Marukame, K.-i. Matsuda, T. Uemura, M.

Arita, and M. Yamamoto, J. Appl. Phys. 101, 09J513 (2007).

[235] T. Ishikawa, T. Marukame, H. Kijima, K.-i. Matsuda, T. Uemura, M.

Arita, and M. Yamamoto, Appl. Phys. Lett. 89, 192505 (2006).

[236] T. Marukame, T. Ishikawa, S. Hakamata, K.-i. Matsuda, T. Uemura,

and M. Yamamoto, Appl. Phys. Lett. 90, 012508 (2007).

[237] T. Taira, T. Ishikawa, N. Itabashi, K.-i. Matsuda, T. Uemura, and M.

Yamamoto, Appl. Phys. Lett. 94, 072510 (2009).

[238] T. Marukame, T. Ishikawa, K.-i. Matsuda, T. Uemura, and M. Ya-

mamoto, J. Appl. Phys. 99, 08A904 (2006).

[239] T. Miyazaki and N. Tezuka, J. Magn. Magn. Mater. 151, 403 (1995).

[240] M. Julliere, Phys. Lett. A 54, 225 (1975).

[241] M. J. Carey, T. Block, and B. A. Gumey, Appl. Phys. Lett. 85, 4442

(2004).

[242] D. Asakura,T. Koide, S. Yamamoto, K. Tsuchiya, T. Shioya, K.

Amemiya, V. R. Singh, T. Kataoka, Y. Yamazaki, Y. Sakamoto, A. Fuji-

mori, T. Taira and M. Yamamoto, Phys. Rev. B 82, 184419 (2010).

[243] K. Miyamoto, K. Iori, A. Kimura, T. Xie, M. Taniguchi, S. Qiao, and

K. Tsuchiya, Solid State Commun. 128, 163, (2003).

[244] T. Saito, T. Katayama, T. Ishikawa, M. Yamamoto, D. Asakura, T.

Koide, Y. Miura, and M. Shirai, Phys. Rev. B 81, 144417,(2010).

[245] I. Galanakis, K. Ozdoan, B. Akta and E. Asioglu, Appl. Phys. Lett. 89,

042502, (2006).

[246] K. Ozdoan, I. Galanakis, E. Asioglu and B. Akta, Solid State Commun.

142, 492, (2007).

[247] K. Ozdoan, E. aolu and I. Galanakis, Phys. Stat. Sol. (RRL) 1, 184,

(2007).

[248] I. Galanakis, K. Ozdo?an, E. Asioglu and B. Akta, Phys. Stat. Sol. (a)

205, 1036, (2008).

[249] B. Hulsen, M. Scheffler and P. Kratzer, Phys. Rev. B 79, 094407, (2009).

[250] T. Marukame and M. Yamamoto, J. Appl. Phys. 101, 083906, (2007).

113

[251] T. Saito, T. Katayama, A. Emura, N. Sumida, N. Matsuoka, T. Ishikawa,

T. Uemura, M. Yamamoto, D. Asakura, and T. Koide, J. Appl. Phys. 103,

07D712 (2008).

[252] B. T. Thole, P. Carra, F. Sette, and G. van der Laan, Phys. Rev. Lett.

68, 1943, (1992).

[253] P. Carra, B. T. Thole, M. Altarelli, and X. Wang, Phys. Rev. Lett. 70,

694 (1993).

[254] Y. Teramura, A. Tanaka, and T. Jo, J. Phys. Soc. Jpn. 65, 1053, (1996).

[255] J. Grabis, A. Bergmann, A. Nefedov, K. Westerholt, and H. Zabel, Phys.

Rev. B 72, 024437, (2005).

[256] Y. Miura, H. Uchida, Y. Oba, K. Abe, and M. Shirai, Phys. Rev. B 78,

064416, (2008).

[257] M. Yamamoto, T. Ishikawa, T. Taira, G. -F. Li, K.-I. Matsuda and T.

Uemura, J. Phys.: Condens. Matter 22, 164212 (2010).

[258] M. Raphael, B. Ravel, M. Willard, S. Cheng, B. Das, R. Stroud, K.

Bussmann, J. Claassen, and V. Harris, Appl. Phys. Lett. 79, 4396 (2001).

[259] B. Ravel, J.O. Cross, M.P. Raphael, V.G. Harris, R. Ramesh, and V.

Saraf, Appl. Phys. Lett. 81, 2812 (2002).

[260] S. Ishida, T. Masaki, S. Fujii, and S. Asano, Physica B 245, 1 (1998).

[261] S. Hakamata, T. Ishikawa, T. Marukame, K.-i. Matsuda, T. Uemura, M.

Arita, and M. Yamamoto, J. Appl. Phys. 101, 09J513, (2007).

[262] B. Ruette, Ph. D. thesis, Bochum (2006).

[263] D. Orgassa, H. Fujiwara, T.C. Schulthess, and W.H. Butler, Phys. Rev.

B 60, 13 237 (1999); J. Appl. Phys. 87, 5870 (2000).

[264] S. Picozzi, A. Continenza, and A.J. Freeman, Phys. Rev. B 66, 094421

(2002).

[265] K.Sato and H. K.-Yoshida Jpn. J. Appl. Phys. 40, L334, (2001).

[266] K. Uchida,H. Tabata, and T. Kawai, Appl. Phys.Lett 79, 988, (2001).

[267] P. Sharma, P. Sharma, A. Gupta, K. V. Rao, F. J. Owens, R. Sharma,

R. Ahuja, J. M. O.. Gillen, B. Johasson and G. A. Gehring, Nat. Mater.

2, 673, (2003).

114

[268] G. T. Thaler, M. E. Overberg, B. Gila, R. Frazier, C. R. Abernathy, S.

J. Pearton, J. S. Lee, S. Y. Lee, Y. D. Park, Z. G. Khim, J. Kim, and F.

Ren, Appl. Phys.Lett 80, 3964, (2002).

[269] N. Ganguli, I. Dasgupta, and B. Sanyal, Appl. Phys.Lett 94, 192503

(2009).

[270] P. V. Padovanovic and Daniel R. Gamelin, Phys. Rev. Lett 91, 157202

(2003).

[271] C. Klingshirn, Phys. Status. Solidi B 244, 3027, (2007).

[272] I. Sarkar, M. K. Sanyal, S. Takeyama, S. Kar, H. Hirayama, H. Mino, F.

Komori, and S. Biswas, Phys. Rev. B 79, 054410, (2009).

[273] D. Karmakar, S. K. Mandal, R. M. Kadam, P. L. Paulose, A. K. Rajara-

jan, T. K. Nath, A. K. Das, I. Dasgupta, and G. P. Das, Phys. Rev. B

75, 144404, (2007).

115