Ramanathan Sh. (Ed.) Thin Film Metal-Oxides: Fundamentals and Applications in Electronics and Energy

Подождите немного. Документ загружается.

108 J. Lu et al.

Fig. 3.12 Resistance and TMR ratio vs. applied magnetic ﬁeld for the LSMO/STO/LSMO tunnel

junction at 4.2 K (Reproduced from [58])

Sun and his collaborators from IBM were among the ﬁrst to develop MTJs with

LSMO [55]. They chose STO as the barrier layer because of the same crystal struc-

ture and a close lattice mismatch. The trilayer structure was an epitaxial junction

LSMO/STO/LSMO grown on (100) STO single crystal substrate by pulsed laser

deposition. This heterostructure junction showed an MR of 84% at 4.2 K. Accord-

ing to the spin-polarized tunneling model, the spin polarization of LSMO was 54%.

Following this heterostructure approach, the IBM group and others made rapid im-

provement in TMR. TMR of 400% at 4.2 K has been reported by Sun et al. [56]

and Viret et al. [57], respectively. Bowen et al. reported TMR of 1;800%at4:2 K

(Fig. 3.12)[58] that transferred to a 95% spin polarization for the LSMO electrodes.

Recently, a TMR ratio over 10,000% was reported on LSMO/STO/LSMO tunnel

junctions grown by rf magnetron sputtering [59].

Despite the very large value at 4 K, TMR ratios for LSMO/STO/LSMO dimin-

ished quickly as the temperature increased and entirely disappeared at 200 K. On

the other hand, the Curie temperature of LSMO used as the electrode was above

room temperature. Therefore, the disappearance of TMR could not be explained by

the reduction in the spin polarization. Sun et al. [56] suggested below 100 K, the

transport mechanism through the STO barrier was simple direct tunneling whereas

above 100 K the tunneling transport was dominated by the variable range hopping

across the barrier through defect states inside the barriers. They speculated that the

magnetic impurities and short spin diffusion length in the STO barrier caused the

high spin scattering rate which greatly reduced the TMR ratio. Recently, Feng et al.

[59] has observed the conductance blockade phenomenon in LSMO (100 nm)/STO

(3 nm)/LSMO (100nm) epitaxial trilayer structures below 100 K (Fig. 3.13). It was

argued that such a conductance blockade with the near 100% spin polarization

3 Novel Magnetic Oxide Thin Films 109

Fig. 3.13 Conductance .dI=dV/ vs. applied bias .V

/ and temperature T for LSMO

(100 nm)/STO (3 nm)/LSMO (100 nm) epitaxial trilayer structures (a) above 100 K; (b)below

100 K. The inset shows Conductance .G

D dI=dV /.V D 0/ as function of temperature

(Reproduced from [59])

from LSMO yielded very high TMR values. This conductance blockade was likely

mediated by space charge, namely oxygen vacancies in the STO barrier layer and

the applied magnetic ﬁeld could remove such a blockade and result in a record high

TMR ratio (over 10,000% at 14T and 10 K). They did not provide any TMR val-

ues at higher temperatures that somewhat suggested that the very high TMR also

diminished quickly above 100 K. It was clear that this blockade was very different

from the Coulomb blockade in terms of the energy regime. However, the mecha-

nism of this conductance blockade remains unclear and it is not yet understood how

the magnetic ﬁeld inﬂuences the blockade.

3.2.4 Magnetite

Magnetite .Fe

/, also known as lodestone, is probably the ﬁrst known magnetic

material. It is a ferrimagnet and its Curie temperature is 860 K. The crystal struc-

ture of Fe

is an inverse Spinel structure (as depicted in Fig. 3.14)inwhich

ions are positioned at tetrahedrally coordinated A-sites and mixed Fe

and

110 J. Lu et al.

Fig. 3.14 The crystal

structure of Fe

(inverse

Spinel AB

)

at octrahedrally coordinated B-sites. Therefore, a more appropriate formula

that reﬂects such a distribution of Fe ions is Fe

.Fe

.Thereisavery

strong antiferromagnetic coupling between the A-site and B-site Fe ions that gives

arise to the ferrimagnetism. The B-site Fe

is responsible for the net magnetic mo-

ment and the saturation moment for Fe

is 480 emu=cm

. At room temperature

magnetite is a poor conductor .200 S=cm/ and the conductivity becomes smaller

as the temperature is reduced.

The half metallicity in Fe

comes from the double-exchange interaction be-

tween Fe

and Fe

in the B-lattice as described in detail by Loos et al. [60]. The

band structure calculations indeed showed a gap in the majority spin at the Fermi

level, where there is no gap in the minority spin [4]. The spin polarization of mag-

netite showed a great dispersion, ranging from as low as 40 to as high as 80%by

spin-polarized photoemission spectroscopy [61–63]. PCAR cannot be used to study

because it undergoes the Verwey transition at 120 K. The Verwey transi-

tion is a ﬁrst-order phase change and Fe

becomes much more insulating below

the transition. The true nature of this transition still remains an open question. Re-

cently, an electrically driven phase transition with a very large hysteresis has been

reported below the Verwey transition temperature [64] that has a very strong re-

semblance to the current induced transition observed in another strongly correlated

oxide, VO

[65]. These novel phenomena not only present opportunities for better

understanding of highly correlated oxides but also to achieve novel functionalities. It

is conceivable that the magnetic properties of Fe

will undergo a drastic change

at around the metal-insulator transition due to the strong correlation between the

magnetic and transport properties in this system. Thus, it would be also an inter-

esting topic to investigate the magnetic ordering in the presence of such current

driven phase transitions. This could provide new perspectives on electron–electron

correlations in such systems.

There have been many efforts to build oxide heterostructure tunnel junctions us-

ing magnetite as the electrode to exploit the half metallicity to achieve very high

TMR values. However, most structures reported in the literature have only used

as one of the magnetic electrodes in contrast to the LSMO/STO/LSMO tri-

layer. In addition, many different oxide barriers have been used. Suzuki’s group has

3 Novel Magnetic Oxide Thin Films 111

Fig. 3.15 Magnetoresistance vs. magnetic ﬁeld at 80 K for a Fe

=CoCr

0:7

0:3

MnO

junction (Reproduced from [66])

demonstrated Fe

=CoCr

=La

0:7

0:3

MnO

trilayer grown on (110) oriented

STO substrates and reported a negative TMR of 25 to 33% at cryogenic tempera-

tures (Fig. 3.15). The low TMR was likely caused by the barrier oxide. The reason

to use CoCr

as the barrier was that it has the spinel structure with a very small

lattice mismatch to Fe

so that it was relatively easy to maintain the epitaxial re-

lationship through the trilayer junction structure. The thickness of the barrier layer

was 6 nm. Instead of the direct tunneling, the inelastic hopping facilitated by lo-

calized states such as oxygen vacancies dominated the transport of spin-polarized

current in the barrier layer. CoCr

is a ferrimagnet below 95 K that presents an

additional spin scattering mechanism. Suzuki’s group has also used other oxide bar-

riers with a spinel structure such as MgTi

and FeGa

[67,68] but these efforts

did not make any signiﬁcant improvement in TMR. We believe that the low TMR

observed in these junctions was likely due to the spin scattering either at the inter-

face between Fe

and barrier or inside the barrier. So far, a barrier material to

form an ideal interface with Fe

is yet to be identiﬁed.

3.3 Diluted Magnetic Oxide Semiconductors

The search for semiconductors with room-temperature ferromagnetism has been

a long-standing goal because of their potential as spin-polarized carrier sources

and possible new spintronic devices [1]. It has been regarded as the “holy grail”

to achieve room-temperature ferromagnetism in semiconductors. Success in ﬁnding

RT diluted magnetic semiconductors a very profound impact on spintronics. The

diluted magnetic semiconductor (DMS) is typically a non-magnetic semiconductor

doped with a small amount of open shell transition metal. The discovery of di-

luted ferromagnetism in transition metal doped III–V semiconductors, in particular

112 J. Lu et al.

Fig. 3.16 Hall resistance .R

Hall

/ vs. magnetic ﬁeld under the gate voltages of 0, C125 and 125 V

at 22.5 K, respectively. R

Hall

curves measured at V

D 0 V before and after application of gate

voltages are virtually identical. The inset is the same curves shown at higher magnetic ﬁelds

(Reproduced from [69])

(Ga,Mn)As, has helped to bring practical semiconductor spintronic devices within

reach. One of the key features of some of these materials is that they exhibit carrier-

mediated ferromagnetism, in which the ferromagnetism is caused by the interaction

of the magnetic ions with the carriers – electrons or holes. The Curie temperature

and other magnetic properties can be modiﬁed by changing the carrier concentra-

tion with electric ﬁelds (gates) or with optical excitation. The electric-ﬁeld control

of ferromagnetism with a ﬁeld-effect transistor structure has already been demon-

stratedbyOhnoetal.[69]. They fabricated (In,Mn)As (with Mn composition 0.03)

hall bar devices with voltage gates to control the carrier concentration in (In,Mn)As,

then measured the hall resistance as function of gate voltage at 22 K(Fig.3.16).

They found that the ferromagnetism could be enhanced and removed depending on

the hole population in the (In,Mn)As as expected.

The introduction of coherent spins into ferromagnetic structures could poten-

tially usher in a new paradigm of quantum magnetoelectronics. As an example,

some proposed quantum computation schemes rely on the controllable interaction

of coherent spins with ferromagnetic materials to produce quantum logic operations

[70]. In addition, this ability to gate the magnetism by changing carrier concentra-

tion presents a new paradigm for memory/logic integrated devices which has the

potential for much lower power dissipation compared with conventional CMOS.

However, it turns to be extremely difﬁcult to obtain room temperature in Mn doped

III–V semiconductor compounds. For example, the record Curie temperature for

heavily Mn doped (Ga,Mn)As is still below 200 K [71,72]. Thus, room-temperature

logic with DMSs is still an elusive target.

3 Novel Magnetic Oxide Thin Films 113

Fig. 3.17 Computed Curie

temperature .T

/ for various

p-type semiconductors

containing 5% of Mn and

3:5  10

holes=cm

(Reproduced from [73])

Calculations based on the Zener mean-ﬁeld model of ferromagnetism by Dietl

et al. predicted room-temperature ferromagnetism in 5% Mn doped p-type ZnO

(Fig. 3.17)[73]. This theory prediction spearheaded an explosive search for the

room-temperature diluted magnetism in a series of oxides.

Matsumoto et al. deposited TiO

ﬁlms in both anatase and rutile phases doped

with transition metals [74]. They found that Co doped anatase TiO

grown on

LaAlO

exhibited room-temperature ferromagnetism with a magnetic moment of

0:32 

=Co (Fig. 3.18). The Curie temperature was estimated to be higher than

400 K. Encouraged by this experimental observation of room-temperature ferro-

magnetism, many researchers have started to add various transition metals into host

semiconducting oxides such as TiO

, ZnO, and SnO

. The two most interested sys-

tems were Co doped TiO

and Mn doped ZnO. However, many early reports did not

present a comprehensive characterization other than the magnetic measurements. In

addition, some results were unable to be reproduced elsewhere. The intensive in-

terests and the conﬂicting experimental results have produced many review articles

that summarized the status in this ﬁeld [75,76].

The nature of diluted magnetism consequently resulted in very small magnetic

moments primarily due to the small percentage of transition metals. Thin ﬁlm sam-

ples contain minute quantities of mass that made the magnetic measurements highly

susceptible to the magnetic contamination. For example, Abraham et al. reported

that the low-temperature ferromagnetism was observed in HfO

thin ﬁlms when

handled with stainless steel tweezers [77]. Thus, it is extremely important to pro-

cess and handle the samples in a “non-magnetic” environment.

Beside the unintentional magnetic contaminants, magnetic secondary phases also

showed magnetic hysteresis and in some occasion the Anomalous Hall Effect (AHE)

which is often used to identify the carrier-mediated ferromagnetism in DMS. Thin

oxide ﬁlms are normally deposited at conditions that are far from equilibrium. Thus,

the transition metal doping often forms nano-clusters of a metallic phase embed-

ded in the oxide matrix that produces confusing magnetic behaviors. However, the

114 J. Lu et al.

Fig. 3.18 (a) Hysteresis loop of Co

0:07

0:93

ﬁlm on LaAlO

at room temperature. Magnetic

ﬁeld was applied parallel to the ﬁlm surface. The inset demonstrates that this ﬁlm is transparent.

(b) Temperature dependence of magnetic moment in a ﬁeld of 20 mT parallel to the surface. T

estimated to be higher than 400 K (Reproduced from [74])

ferromagnetism caused by the secondary phase is not carrier mediated and therefore

the electrical control of the spin polarization cannot be realized. Venkatesan and his

collaborators observed cobalt nano clusters in highly reduced Co doped TiO

ﬁlms

that caused the superparamagnetism and AHE (Fig. 3.19)[78]. This experimental

evidence challenged the idea of using AHE only to test for true DMSs.

Many scientists have realized that observing a hysteresis loop and the anoma-

lous Hall resistance is not sufﬁcient to prove homogeneous diluted magnetism in

these semiconductingoxides. Ando has proposed to use magnetic circular dichroism

(MCD) to determine the correlation between the magnetism and the semiconduc-

tor electronic structure [79]. MCD is observed in magnetic materials in which

3 Novel Magnetic Oxide Thin Films 115

−1

0 100 200 300 400

100

200

300

400

−1500

−20

−10

Inset (d)

300 K

(μΩcm)

H(T)

Inset (c)

x=0.01

x=0.02

(μemu)

T (K)

0 100 200 300 400

T (K)

Inset (b)

(Oe)

x=0.01

x=0.02

x=0.01

380 K

H(Oe)

350 K

5 K

50 K

100 K

150 K

200 K

250 K

300 K

emu)

–1 0 1

−15

Inset (a)

M(μemu)

H/T(Oe/K)

380 K

350 K

300 K

250 K

200 K

0 1500

Fig. 3.19 Hysteresis loops for Ti

0:99

0:01

2•

as a function of temperature. Inset (a) M vs. H/T

at different temperatures. Insets (b)and(c) are the temperature dependence of H

and M

for

1x

2•

ﬁlms. Inset (d) Hall resistivity vs. H of the Ti

0:99

0:01

2•

ﬁlm (Reproduced

from [78])

clockwise-polarized and counterclockwise-polarized light are absorbed differently.

In addition, MCD provides direct information on the spin polarization in a semicon-

ductor band so that one can associate the ferromagnetism to the band structure of the

semiconductor. MCD analysis has already conﬁrmed the intrinsic dilute magnetism

in (In,Mn)As, (Ga,Mn)As, and (Zn,Cr)Te but is not widely applied in analyzing ox-

ide systems. Ando’s group has used MCD to probe the ferromagnetism in transition

metal doped ZnO and Co doped TiO

The electric-ﬁeld control of ferromagnetism, which is highly relevant to practi-

cal device applications, is also an important method to reveal the carrier-mediated

ferromagnetism in a true diluted magnetic oxide. Zhao et al. has demonstrated

that the magnetization of a 60 nm thick Co doped TiO

changed as much as 15%

at room temperature with ˙30 V applied an adjacent ferroelectric PbZr

0:2

0:8

116 J. Lu et al.

Fig. 3.20 The same hysteresis loops for anatase Co

0:07

0:93

after several electric voltage pol-

ing on the PZT layer shown in (a) from 6;000 to 6,000 Oe (b) from 750 to 750 Oe. The two

insets in (a) are the electric polarization vs. applied ﬁeld of the PbZr

0:2

0:8

layer and the satu-

ration magnetization of Co

0:07

0:93

as a function of the applied voltage on PbZr

0:2

0:8

.The

inset in (b) plots the coercive ﬁeld .H

/ of Co

0:07

0:93

as a function of the applied voltage on

PbZr

0:2

0:8

(Reproduced from [80])

layer to pole it [80](Fig.3.20). It is worth mentioning that PbZr

0:2

0:8

is also

piezoelectric so that the strain in Co doped TiO

could also be altered signiﬁcantly

with the reverse poling voltages applied on PbZr

0:2

0:8

In the years following the ﬁrst report on room-temperature ferromagnetism in

TiO

there has been a signiﬁcant experimental and theoretical exploration of a wide

range of oxides considered to be candidates for DMSs. However, the existence of

true diluted magnetic oxides and the origin of the ferromagnetism still remain an

open question but the reward for ﬁnding room-temperature ferromagnetic semicon-

ductors could represent a new era in the technology revolution!

3 Novel Magnetic Oxide Thin Films 117

3.4 Room-Temperature Multiferroic Oxide Thin Films

3.4.1 Introduction

Multiferroic is used to describe materials that possess two or all three of the ferro-

properties: ferroelectricity, ferro/ferrimagnetism, and ferroelasticity. In a broader

sense, it also covers materials with ferro- and antiferro-properties or pure antiferro-

properties. These materials have been known for decades, however interest in

multiferroics has been revitalized in the past few years largely due to the advance-

ments in both ﬁrst principle calculations and experimental techniques, especially

in thin ﬁlm growth [81]. Multiferroic materials will play a signiﬁcant role in de-

veloping systems with large magneto-electric coupling where the manipulation of

magnetization or polarization can be achieved by applying an electric or magnetic

ﬁeld, respectively. This magneto-electric coupling with an extra degree of freedom

will eventually usher a paradigm shift from conventional electronic devices. Many

have already proposed novel device concepts based on multiferroics. However, the

number of naturally occurring multiferroic materials is very small while ones with

ferromagnetism and ferroelectricity are even rarer. Spaldin et al. explained that the

rarity of ferromagnetic ferroelectricity was due to the contra-indication between

magnetism and ferroelectricity (empty d orbital vs. partially ﬁlled d orbital) that

prevented the simultaneous occurrence of ferromagnetism and ferroelectricity [81].

She also offered guidance to design new multiferroics using an alternative mech-

anism either for ferrimagnetism or for ferroelectricity and some efforts have been

made to search for these new multiferroics.

Single-phase materials exhibiting strong magnetism at room temperature and

ferroelectricity are highly desired but also very rare. So far, the only example of

a single-phase material showing both electric and magnetic orderings well above

room temperature is BiFeO

, which is ferroelectric and antiferromagnetic. Another

approach to achieve the desired magneto-electric coupling is to combine a ferro-

electric phase and a ferromagnetic phase into a composite. Here, we will focus on

BiFeO

and the multiferroic composites that are room-temperature multiferroic and

more suitable for practical device applications. For a comprehensive overview in

this ﬁeld, there are some reviews that are available for further reading [82–84].

3.4.2 BiFeO

BiFeO

is the most studied multiferroic materials because of its room-temperature

antiferromagnetism and ferroelectricity. The two critical temperatures for the ferro-

electricity and antiferromagnetism in BiFeO

are 830 and 370

C, respectively.

BiFeO

has a distorted perovskite structure (R3C) with a D 3:96

Aand˛ D

89:4

. A counter-rotation of the oxygen octahedra occurs as illustrated in the

Fig. 3.21 [85]. Fe ions are displaced from the center along the [111] direction