Sunteți pe pagina 1din 70

The Influence of Film Thickness and Stacking Order on the Magnetic Properties of

Perovskite Oxide Bilayers

By

KYLE HOKE
B.S. (Western Washington University) 2015

THESIS

Submitted in partial satisfaction of the requirements for the degree of

MASTER OF SCIENCE

in

Materials Science and Engineering

in the

OFFICE OF GRADUATE STUDIES

of the

UNIVERSITY OF CALIFORNIA

DAVIS
Approved:

Dr. Yayoi Takamura, Chair

Dr. Susan Gentry

Dr. Roopali Kukreja

Committee in charge
2017

–i–




ProQuest Number: 10621413




All rights reserved

INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.






ProQuest 10621413

Published by ProQuest LLC (2017 ). Copyright of the Dissertation is held by the Author.


All rights reserved.
This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.


ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
Contents

Abstract iii

1 Introduction & Background 1


1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2 Previous Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2 Techniques 13
2.1 Pulsed Laser Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Physical Characterization Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.1 X-Ray Reflectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.2 X-Ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2.3 Reciprocal Space Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3 Superconducting Quantum Interference Device Magnetometry . . . . . . . . . . . . 25
2.4 Soft X-Ray Magnetic Spectroscopy Techniques . . . . . . . . . . . . . . . . . . . . . . 26
2.4.1 X-Ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4.2 X-Ray Magnetic Circular Dichroism . . . . . . . . . . . . . . . . . . . . . . . . 28

3 Physical and Magnetic Characterization of Bilayer Perovskite Oxide Samples 31


3.1 X-Ray Reflectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 X-Ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Reciprocal Space Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 Soft X-Ray Magnetic Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.5 SQUID Magnetometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.6 X-Ray Magnetic Circular Dichroism: Hysteresis Loops . . . . . . . . . . . . . . . . . 50

4 Conclusions 57

Bibliography 61

–ii–
KYLE HOKE
September 2017
Materials Science and Engineering

The Influence of Film Thickness and Stacking Order on the Magnetic Properties of Perovskite
Oxide Bilayers

Abstract

Correlated materials form an active field of research because of their wide range of functional

properties such as piezoelectricity, ferroelectricity, and even more exotic properties such as colos-

sal magneto-resistance. These properties and their potential applications are interesting due to the

complex interactions that arise between the lattice, charge, spin, and orbital degrees of freedom.

Furthermore, research has shown that many interesting behaviors arise at the interface of such

materials when deposited in thin film heterostructures. Within the last 20 years, thin film depo-

sition technology has advanced to the point that atomic level precision is achievable and thus, it

is possible to explore and tailor the interactions at such interfaces in a controlled and systematic

way.

Previous work has shown that interfacial exchange interactions known as exchange-bias

arise at the interface between a ferromagnetic (FM) and antiferromagnetic (AFM) material. A

similar effect has been shown to exist between two distinct FM layers where one has a higher

coercive field than the other. In this case, the material with the higher coercive field biases the

material with lower coercive field, leading to a horizontal shift of the magnetic hysteresis loops.

Unlike bilayers consisting of metallic FM layers, interfacial charge transfer was found to play an

important role in the magnetic properties of the oxide bilayers.

In this work, the magnetic behavior in La0.7 Sr0.3 MnO3 (LSMO) and La0.7 Sr0.3 CoO3 (LSCO)

bilayers were investigated as a function of layer thickness and layer order. X-ray reflectivity, x-ray

diffraction, and reciprocal space mapping were used to structurally characterize the samples. X-

ray absorption was used to quantitavetively investigate charge transfer between layers in a sam-

ple, x-ray magnetic circular dichroism was used to measure element specific magnetic hysteresis

–iii–
loops, and superconducting quantum interference device magnetometry was used to gather bulk

magnetic hysteresis loops from the samples. It was found that decreasing the LSMO layer thick-

ness in the bilayers led to less charge transfer at the bilayer interface for the same LSCO layer

thickness. Changing the growth order of the bilayers without changing the layer thicknesses also

suppressed charge transfer at the bilayer interface.

These results show that charge transfer at interfacial layers of correlated oxide heterostruc-

tures has a more complex dependency on layer thickness than previously understood. It is also

apparent from these results that this charge transfer effect depends on both of the layers involved

in forming the interface. This work ultimately contributes to the understanding of interfacial phe-

nomena in complex oxide heterostructures and building reductive models of arbitrary, layered

material systems.

–iv–
Chapter 1

Introduction & Background

Complex oxides are a class of materials that exhibit a wide range of interesting functional

properties, such as ferromagnetism, ferroelectricity, and colossal magnetoresistance. In complex

oxdies, the lattice, electronic, orbital, and spin degrees of freedom are intimately related and have

a large effect on material properties. The interactions between these different degrees of freedom

make complex oxides highly tunable via controllable factors such as strain, chemical composi-

tion, and interface engineering. However, only in recent years has thin film growth technology

developed to the point that epitaxial films with atomic layer precision of the layer thickness can

be created [2]. This advancement has allowed us to create artificial layered materials in which we

can combine the functional properties of different complex oxides and explore novel interfacial

phenomena.

The prototype perovskite structure is ABX3 , shown in Figure 1.1, where the ’A’ site

cations are often rare earth metals, and ’B’ site cations are usually transition metals. Perovskite

oxides are a subset of the complex oxides and have the structure shown in Figure 1.1, but with

oxygen anions in place of the ’X’ atoms. In the case of my samples, and all other samples dis-

cussed in this thesis, the ’X’ anion is always oxygen. The central ’B’ atoms are surrounded by

an octahedron of six oxygen atoms. Figure 1.1 shows that the BO6 octahedra are corner sharing

which leads to B-O-B chains that extend in all three dimensions across the lattice. Thus, the func-

tional properties of these materials are dictated mainly by these ’B’ site atoms. In this thesis, I will

1
mainly focus on dependence of magnetic properties with varying thicknesses and layer ordering

as an example of interface engineering for modifying material properties.

Figure 1.1: Atomic arrangement for the ABX3 perovskite structure.

The magnetic ordering in complex oxides is mainly dictated by indirect exchange inter-

actions since the ’B’ cations are too far apart for direct exchange mechanisms. The nature of the

bonding between any ’B’ site cation and a neighboring oxygen anion will depend heavily on the d

orbital configuration in the transition metal cation due to interaction of the d orbitals in the ’B’ site

cation with the p orbitals of the oxygen anion. According to crystal field theory, this octahedral

environment will result in an energy level splitting of the d orbitals into a degenerate triplet state

called the t2 g orbitals, and degenerate doublet called the eg orbitals [3]. An induced strain from

lattice mismatch in epitaxially grown films can cause the d orbitals to undergo a further splitting

as in Figure 1.2. If the lattice is placed under biaxial tensile strain in the (001) plane, the crystal

will compress in the out-of-plane [001] direction. Conversely, if the lattice is placed in biaxial com-

pressive strain in the (001) plane, then the lattice will expand along the [001] direction. The lattice

compression in the [001] direction favors electron occupancy of the x2 -y2 d orbital, while the z2

orbital is favored when the lattice is stretched along the [001] direction. This lifting of degeneracy

modifies the electron occupation in the d orbital and thus affects the exchange interactions and the

magnetic properties [4].

2
Figure 1.2: d-orbital splitting for an octahedral environment. The left hand side shows the crystal field
energy levels with no distortion present. The right hand side shows how the energy levels are altered
when the octahedra is placed under tensile strain in the z-direction. The orbitals within the eg and t2g
orbitals reverse when the octahedra is placed under a compressive strain instead [5].

The Goodenough-Kanamori-Anderson (GKA) rules can be used to predict the magnetic

ordering of a material; this is sometimes called a ”superexchange” interaction [6]. Figure 1.3 shows

an example of the how the GKA rules are applied to two oxygen mediated Mn3+ cations. The the-

ory predicts that virtual electron hopping between the Mn3+ and the O2− can lower the energy of

the system. Virtual electron hopping, unlike real electron hopping, does not change the oxidation

state and the initial electron count per atom is always restored. Therefore, the initial and final

states appear to be the same after a virtual electron hopping event. Figure 1.3a shows two Mn3+

cations with electrons in the 3dx2 orbitals. The virtual hopping of electrons may only take place

when the spin states of the two Mn cations are anti-parallel, which leads to anti-ferromagnetic

(AFM) ordering. Figure 1.3b is similar to 1.3a, but in this case the lower energy state is one in

3
which parallel spins are favored leading to ferromagnetic (FM) ordering. Figure 1.3c shows the

90◦ exchange interaction where two Mn cations are corner sharing the same oxygen anion. Virtual

hopping of electrons from the two 3dz 2 orbitals to the oxygen 2py and 2pz orbitals lead to a low

energy state that has FM ordering.

Figure 1.3: The GKA rule applied to the magnetic interaction between two Mn3+ cations. (a) AFM
ordering, (b,c) FM ordering [7]. Note that in (b) one of the electrons sits on the x2 − y 2 orbital which is
not the case for (a).

Doping of the samples with elements of different, fixed valence states, such as an alkaline

earth metal, will change the valence states of the B cation depending on the ratio of the rare earth

to alkaline earth elements in the unit cell. For example, in LaCoO3 (LCO) the valence state of the

cobalt ion is +3, lanthanum is +3 and the three oxygen ions are each -2. For doping with Sr, the

strontium ion has a valence state of +2, so the valence state of the cobalt ion must change to main-

tain charge neutrality in the unit cell. Thus, doping complex oxide samples leads to multivalent

configurations that exhibit the double exchange effect, and give rise to ferromagnetic properties.

In the case of mixed valence ’B’ sites, physical electron transfer becomes possible through an ex-

change mechanism called ”double exchange” as illustrated in Figure 1.4. The figure uses Mn3+

4
and Mn4+ coupled to an oxygen anion to demonstrate the electron hopping from one cation to the

other. The magnetic moments of the manganese cations have to be parallel or the electron hopping

event would violate Hund’s rule [4]. Therefore, the magnetic and electrical properties in complex

oxides are closely linked to each other and typically the metal/insulator transition coincides with

the ferromagnetic-paramagnetic (FM-PM) transition.

Figure 1.4: Illustration of the double exchange interaction for multi-valent Mn cations [7].

The specific materials I will discuss in this thesis are La1−x Srx MnO3 and La1−x Srx CoO3

thin films grown on (LaAlO3 )0.3 (Sr2 TaAlO6 )0.7 (LSAT) substrates. The x subscript indicates the

amount of Sr doping in the sample. Figure 1.5 shows the magnetic phase diagram of La1−x Srx MnO3

as a function of the Sr doping in the system. As shown in the magnetic phase diagrams for both

La1−x Srx MnO3 and La1−x Srx CoO3 , there are several different magnetic and electronic phases

present due to the strong interplay between the electronic and lattice degrees of freedom [8].

In this case, the doping of Sr into the perovskite oxide is chosen such that the material is in a

ferromagnetic-metallic (FM-M) state. The exact amount of strontium used is determined by the

Curie temperature, TC . If the material had a FM to PM phase transition below room temperature,

its potential applications would be limited since any device would have to be constantly cooled to

prevent the transition from occurring. Therefore, the doping level is chosen to maximize TC .

5
Figure 1.5: La1−x Srx MnO3 magnetic and electrical phase diagram as a function of Sr doping [8].
La1−x Srx MnO3 shows six different phases: charge ordered insulator (C-I), ferromagnetic insula-
tor (FM-I), ferromagnetic-metal (FM-M), anti-ferromagnetic-metal (AFM-M), paramagnetic-insulator
(PM-I), and paramagnetic-metal (PM-M). TN and TC indicate the Neel Curie temperatures respectively.

La1−x Srx MnO3 and La1−x Srx CoO3 exhibit FM-M behavior when the Sr concentration

(x) increases to about x=0.15 and x=0.18, respectively. The magnetic phase diagrams of these

materials in Figures 1.5 and 1.6 show that paramagnetic (PM), anti-ferromagnetic (AFM), and

insulating phases are also present. The doping levels of the La1−x Srx MnO3 and La1−x Srx CoO3

are equivalent to prevent La/Sr segregation across the bilayer interface.

6
Figure 1.6: La1−x Srx CoO3 magnetic and electrical phase diagram as a function of Sr doping [9]. The
different phases present are: paramagnetic semiconductor (PS), paramagnetic metal (PM), ferromag-
netic metal (FM), and spin glass (SG). MIT=metal-to-insulator transition.

In chapter 2, the growth and characterization techniques used in this work will be de-

scribed. Chapter 3 will present the experiments I completed to address the above question and

the results will be discussed using the techniques outline in chapter 2. Chapter 4 will start with

conclusions and end with a brief discussion of future work in this area.

1.1 Motivation

The work in this thesis strives to understand and tailor magnetic properties in thin, lay-

ered films of perovskite oxides. This thesis builds off the work of Dr. Binzhi Li, a previous gradu-

ate student from Dr. Yayoi Takamura’s research group. In his work, bilayer systems were studied

to reduce the number of unknown factors influencing the properties of interest in a given sample.

For example, if trilayer samples are considered, the interplay of the two separate film interfaces

would have to be accounted for in all measurements.

As Dr. Herbert Kroemer once said, ”The interface is the device” [10]. He was talking

about devices based on semiconductor films, but the notion applies just as easily to devices based

on complex oxides. Interface engineering is at the core of a number of promising applications for

perovskite oxides. For example, investigations into complex oxides are developing the necessary

7
groundwork to use spin-dependent magnetic memory devices [11].

Spin-currents are generated when current passes through a highly spin-polarized FM

metal and switch to another magnetic layer. This device structure, called a ’spin-valve’, is essen-

tially composed of two FM materials separated by a non-magnetic spacer. One has its moments

’pinned’ by an antiferromagnetic (AFM) layer, while the second is ’free’. The pinned layer has

a much higher magnetic coercivity than the unpinned layer. The hysteresis loop of this pinned

layer is shifted away from zero via exchange bias, so it has a single, stable magnetic state at H=0 T

as illustrated in Figure 1.7. This exchange bias shift allows switching of only the unpinned layer

under low applied field. In this way it is possible to create two electrical states, one where the

two layers are parallel (low resistance state), and one where they are anti-parallel (high resistance

state).

Figure 1.7: Schematic diagram of a FM-AFM bilayer with the FM layer pinned. The sketches on the
left show how the bilayer responds to the applied field shown on the right [14].

An example of one of these magnetic storage devices is called ”racetrack memory.” This

8
device utilizes the movement of magnetic domain walls under a spin current to obtain very stable,

long term, high density storage of information [13].

Another application of layered complex oxides is magnetic read heads, utilizing a prop-

erty called colossal magneto-resistance (CMR), where a small applied field to the material can

drastically change the resistance by a few orders of magnitude [12].

The incorporation of complex oxides into devices like ”racetrack” memory and magnetic

read heads is still new. These types of devices require the development of comprehensive models

which can predict their electrical and magnetic properties. The aim of this thesis is to understand

and investigate magnetic properties of complex oxides under different conditions, such as the

application of electric fields and strain.

1.2 Previous Work

The samples in the previous works were all LSCO/LSMO bilayers grown on (001) ori-

ented LSAT substrates. All controllable experimental parameters in this study, such as: substrate

temperatures during deposition, specific target materials, energy flux onto target materials, en-

vironmental temperatures during measurements, and others, were taken from previous work for

consistency. Table 1.1 lists some of the material properties for LSMO, LSCO, and LSAT. LSMO is

a soft FM metal and LSCO is a hard FM metal. LSAT was chosen for the substrate because of the

close match in lattice parameters with LSMO and LSCO, see Table 1.1. These samples are sensitive

to in-plane strain, and lattice matching minimizes this strain.

9
Table 1.1: Materials Properties

LSMO[15] LSCO LSAT[16]

TC (K) 370 220 N/A

Bulk Saturization Magnetization (emu/cm3 ) 600 150 N/A

Coercive Field (T) 0.002 ∼0.6 N/A

Bulk Lattice Parameters (Å) ac =3.89 ac =3.83 a=3.87

Magnetism Type FM FM Diamagnetic

Category Metal Metal Insulator

In Dr. Li’s first paper[17], the exchange interactions in La0.7 Sr0.3 CoO3 /La0.7 Sr0.3 MnO3

(LSCO/LSMO) bilayers were probed. The sample was prepared on a (001) oriented LSAT sub-

strate. A 6 nm thick LSCO layer followed by an LSMO layer of the same thickness was deposited

at a substrate temperature of 700 ◦ C. The film was then cooled to room temperature in 300 Torr

O2 to ensure oxygen stoichiometry. All measurements for this sample were taken at 80 K. This

work discovered that the moments in the LSCO film near the LSCO/LSMO interface switched

with the moments in the LSMO film, which deviates from the typical LSMO only switching. The

proposed cause of this switching was attributed to charge transfer at the interface of the LSCO

and LSMO layers. The evidence of charge redistribution was observed using the x-ray absorption

(XA) spectroscopy shown in Figure 1.8. The proposed charge transfer mechanism involves an

electron transfer from Mn3+ to Co3+ ( Mn3+ + Co3+ → Mn4+ + Co2+ ). These new ionic species at

the interface create a small, softer LSCO layer that is more strongly coupled to the LSMO than the

rest of the LSCO layer. By ’soft’ it is meant that the magnetic moments in the sample align with an

external field more easily than other materials in a relative sense. A magnetically ’hard’ sample is

just the opposite. This strong coupling allows the small interfacial layer to unexpectedly switch

with the LSMO layer at fields well below the transition of the hard LSCO layer.

10
Figure 1.8: The panels show the XA spectra for L2,3 edges. The dotted lines are the weighted differences
of the bilayer and single layer films. The top panel shows Mn XA spectra for the bilayer and an LSMO
film. The dashed line indicates shoulder features of Mn4+ and Mn3+ . The bottom panel is the Co XA
spectra for the bilayer and an LSCO film. The dashed lines indicate the A/B subpeaks on Co L2 edge
[17]. Used with permission from Dr. Binzhi Li.

His second paper studied the magnetic switching behavior of the same LSCO/LSMO

bilayer films as a function of the LSCO film thickness [18]. Specifically, LSCO layers of 2, 4, and

8 nm were grown, followed by a fixed LSMO layer thickness of 6 nm. The charge transfer pre-

viously observed was also found in these samples [17]. XA spectra at the Co L-edges show the

presence of Co2+ which is associated with the magnetically ’softer’ layer at the LSCO/LSMO in-

terface. The magnetically ’harder’ layers in the LSCO show the typical Co3+/4+ signature found

11
in single layer LSCO films. Dr. Li found that the switching behavior, discussed above, is highly

dependent on the LSCO film thickness, and that at a critical thickness of the film, the LSCO layer

undergoes magneto-electronic phase separation (MEPS) in the direction of the film growth, shown

in Figure 1.9. MEPS is an effect where chemically homogeneous materials display inhomogeneous

magnetic and electronic properties [19]. It was shown that this phase separation is due to charge

transfer at the LSCO/LSMO interface. These results demonstrate the unique tunability of mag-

netic properties in complex oxide systems, which is different from common FM metal systems.

The fundamental mechanism responsible for the tunability in complex oxides is the indirect ex-

change interactions between the transition metal ions in the material.

Figure 1.9: Magneto-electronic phase separation schematic showing the magnetic/electronic structure
for samples with increasing LSCO layer thickness (a) → (c). The LSMO layer is a fixed thickness [18].
Used with permission from Dr. Binzhi Li.

The aim of my research is to expand upon this work and answer the question: Does

varying the LSMO layer thickness have the same effect on an LSCO/LSMO bilayer system as

varying the LSCO layer thickness does? I also further explore how switching the growth order of

the LSMO and LSCO layers affects the magnetic properties of the system.

12
Chapter 2

Techniques

The experimental techniques used to investigate the perovskite oxide bilayers are de-

scribed in this chapter. These techniques were used to determine the physical and magnetic prop-

erties of each sample. I started by growing the bilayer systems using a technique known as pulsed

laser deposition (PLD). Physical characterization was then conducted to confirm that the film had

been grown as desired. Once I was sure of the structure of the thin films, then the characterization

of the magnetic properties was performed.

2.1 Pulsed Laser Deposition

PLD is a relatively new advancement in film growth technology[2]. It provides control

over crystal quality that older methods such as thermal evaporation or sputtering do not offer

[20]. There are very few restrictions on the type of material one can deposit using PLD. As long

as the material absorbs at the wavelength of the laser being used, the material may be ablated.

This deposition technique has been used for semiconductors, metals, polymers, and biological

materials [2]. The technique provides atomic layer precision in the layer thickness and roughness

of grown thin films, as well as stoichiometric transfer of the target material in multicomponent

materials.

13
Figure 2.1: Schematic of a pulsed laser deposition system. The main components are the vacuum
chamber, the vacuum pump, target, substrate, laser, gas inlet, and the ablation plume [2].

These features are highly desirable in perovskite oxide heterostructures with only a few

nanometers of individual layer thicknesses. Figure 2.1 shows a schematic of a typical PLD cham-

ber. The system consists of a chamber equipped with a turbomolecular vacuum pump that is

backed by a typical mechanical roughing pump. Together these pumps can reach pressures down

to 10−6 Torr. A KrF laser (248 nm wavelength) is used to ablate a target with the desired stoi-

chiometry, creating a plasma plume which carries the vaporized target material to the substrate

through the momentum of the initial ejection. Both the target and the substrate are held on rotat-

ing platforms to ensure uniform use of the target and uniform deposition of the ablated material

on the substrate. The gas inlet supplies oxygen to the chamber during the deposition; this is often

called a ”process gas.” In the case of perovskite oxides, the oxygen process gas serves to ensure

the stoichiometry of the deposited material and to thermalize the plasma species.

The growth parameters considered in growing a film with PLD include: substrate tem-

perature, laser energy density, process gas species, pressure, substrate-target distance, pulsed laser

frequency, and the wavelength of the laser [2]. Optimization of these parameters was performed

14
by Dr. Li [17, 18, 33, 48, 49]. In studying thin films, it is necessary to understand the possible

growth modes. The main types of growth modes are: layer-by-layer, step flow, island, and a

combination of layer-by-layer and island.

The layer-by-layer and step flow growth modes are the ones that we desire for thin films,

these modes will ensure epitaxy and sharp interfaces that are necessary in our perovskite oxide

thin films. The layer-by-layer growth mode occurs when the contact angle θ, as shown in Figure

2.2, is approximately zero [21]. The contact angle is the angle formed by the plane of the substrate

and the slope of the surface of the film nucleus where it meets the substrate.

Figure 2.2: Schematic of atom nucleation processes on a substrate during thin film deposition [22].

Young’s equation uses thermodynamics to relate the surface energy terms (denoted as γ)

in Figure 2.2 to the contact angle of the nucleus,

γsv = γf s + γf v cos θ, (2.1)

where the γsv , γf s , and γf v denote the surface energy of the substrate-vapor interface, the film-

substrate interface, and the film-vapor interfaces, respectively. We find that the layer growth mode

occurs when θ → 0, which gives,

γsv ≥ γf s + γf v . (2.2)

The LSAT substrate has a cubic symmetry while the LSMO and LSCO have rhombohe-

dral symmetry. These non-cubic symmetries arise from deviations away from the ideal perovskite

15
structure. These deviations are due to the ionic size differences in the A, B, and O ions and can be

quantified using the Goldschmidt tolerance factor,

rA + rO
t= √ . (2.3)
2 (rB + rO )

The perovskite oxide structure is an ideal cubic structure then t = 1, cubic structures are also

obtained if 0.9 < t ≤ 1. If t > 1 then the structure is hexagonal or tetragonal, and if 0.71 < t < 0.9

the structure is rhombohedral. When t < 0.71 different structures, such as structures based on

the trigonal system begin to arise. In order for the ’X’ sites to match up from one unit cell to the

next when the system does not have cubic symmetry, the octahedra must rotate. These octahedral

rotations have been studied in great detail elsewhere [23]. We can however, imagine side length

of the unit cell in the LSMO and LSCO crystal systems is doubled, which makes them effectively

cubic; we call this nonprimitive unit cell ”pseudocubic.” The bulk pseudocubic lattice parameters

of LSAT, LSMO, and LSCO are 3.868 Å, 3.875 Å, and 3.82 Å, respectively [24–26]. The surface

energy at an interface can be thought of as the amount of energy released when separating the

films from the interface. If the lattice parameters differ greatly between two films, i.e. higher

lattice mismatch, then the amount of energy required to bring them together is higher, and hence

the surface energy is higher. When LSCO or LSMO is grown on LSAT (or LSCO on LSMO and

vice versa) the γf s term in Young’s equation effectively goes to zero because the amount of lattice

mismatch is minimal, and equation 2.2 can be satisfied. In addition, the lattice mismatch creates

in-plane biaxial strain which is equal along both in-plane directions at the interface. This biaxial

strain increases with increasing film thickness, and at a certain point misfit dislocations will begin

to form at the interface to relax the building strain. Without a common crystal structure and a

close match in lattice parameters, layer-by-layer growth would not be encouraged.

Given that main materials of interest are perovskite oxides, the process gas used is oxy-

gen. The deposition oxygen partial pressure must be high enough to ensure stoichiometric growth.

The deposition pressure also effects the kinetics of the ablated species. If oxygen partial pressure

is high, oxygen ions will scatter the ablated particles and lower the growth rate. Once the deposi-

tion is complete, the film is cooled in a higher oxygen partial pressure (typically 300 Torr) which

16
allows the oxygen to diffuse through the sample to ensure stoichiometry. The process pressure

and annealing pressure help preserve the oxygen stoichiometry of the target material in the film.

The temperature of the substrate is held at a relatively high temperature throughout the

deposition process, typically, 500 800 ◦ C for epitaxial single crystal growth [27]. If the substrate is

not at high temperature and oxygen partial pressure is lower, then oxygen vacancies will form in

the film. Thus, this process is important for high quality films. If the substrate is at room tempera-

ture during the growth, the film will be amorphous, as the deposited atoms will not have enough

energy to diffuse/move to achieve a crystalline state. At intermediate temperatures, the film may

become polycrystalline or form other types of defects. The diffusion length can be roughly defined

by,

p
lD = DS τ , (2.4)

where DS is the surface diffusion coefficient, and τ is the residence time [2]. The diffusion coeffi-

cient is generally expressed as,

EA

DS ∝ e k B T . (2.5)

An activation energy is required for the atom to diffuse in the film. The important thing to note

is that the diffusion length has an exponential dependence on the temperature. In this study we

have used 700◦ C, as it is optimal for ensuring proper oxygen stoichiometry in growing epitaxial

single crystal thin films [28].

Another growth parameter is the laser energy density, which is the total laser energy

divided by the laser spot size. This must be large enough to vaporize material from the target

and provide enough kinetic energy for the ejected material to reach the substrate. As the energy

density increases, the ejected material density and the probability material will be ejected from the

target increases, which can result in increased film roughness. If the energy density is too low, the

target material will not be ablated and nothing will be deposited. Optimizing these parameters is

important for getting a smooth film that matches the in-plane crystal structure of the substrate as

17
exactly as possible.

There are a few common ways to control the thickness of the film being grown. One

of the methods is to use in situ reflection high-energy electron diffraction (RHEED). This method

uses the diffraction of electrons from the periodic surface atoms. A single point on the diffraction

pattern is used to track times of high and low intensity during the deposition process. The specular

reflection spot is typically used due to high intensity, sensitivity to the diffraction condition, and

the highest intensity when the surface layer is complete [29]. In a layer-by-layer growth mode, as

more material is deposited the surface layer begins to form the next partially complete layer which

results in an intensity drop due to the surface roughness. The intensity of the reflection increases

as the current layer reaches full coverage. The other, simpler, way is to assume a constant growth

rate and count the number of pulses for a deposition, then measure the film thickness to get a

thickness per pulse calibration for a particular energy density and material, which is utilized for

this thesis work.

2.2 Physical Characterization Techniques

Given the dependence between crystal structure and magnetic properties of the samples,

the physical properties must be characterized to ensure high quality. To this end we use several

techniques to determine the strain state, measure the thickness, the in- and out-of-plane lattice

parameters, and the interfacial roughness of the films.

2.2.1 X-Ray Reflectivity

X-ray reflectivity (XRR) uses x-rays incident upon a sample at low angles to measure the

overall film thickness, density, and its interfacial roughness. Figure 2.3 shows the XRR geometry.

The incident and refracted angles are θ and θt , respectively. Using Snell’s Law,

n0 sin θ = n sin θt , (2.6)

where n0 is the index of refraction of a vacuum and n is the index of refraction of the film, we can

18
determine the critical angle below which all x-rays are externally reflected (no refraction occurs).

Figure 2.3: XRR schematic. A, B, C, and D reference the different x-rays in the schematic. t is the film
thickness. θ is the incident and reflected angle of the x-ray. θt is the angle of the refracted x-ray ’B.’

This critical angle (θc ) is related to the electron density (ρ) of the thin film and the incident

x-ray wavelength (λ) via,


θc ≈ 1.64 · 10−3 ρλ. (2.7)

Thus, we can use the XRR data to determine the electron density of the film(s) deposited. The

thickness of the film can be determined using the path length difference, ∆, of the x-rays reflected

from the top and bottom interfaces of the film as shown in Figure 2.3. Constructive interference

between the x-rays occurs when the path length difference is an integer multiple of the incident

wavelength. The film thickness can be expressed as,

(m1 − m2 ) λ
t= . (2.8)
2 (sin ω1 − sin ω2 )
The subscripts 1 and 2 account for the fact that there are two refraction events, one when

the incident x-ray enters the film, and another when it leaves the film. The m’s denote the different

modes at which this relation is satisfied and the ωs are the refracted angles for events one and two,

where ω = θ/2. Plotting the angle of incidence against intensity shows oscillatory behavior due

to the constructive interference condition being met every time the path length difference is an

integer multiple of the incident wavelength. If the film is rough, the intensity of the oscillations

will die out very quickly. The XRR data is fit using a software package such as Brukers Leptos

19
software. The parameters in the fitting model can be used to determine the film thickness, interface

roughness, and film density. Figure 2.4 shows a representative example of an XRR spectra from a

gold film on a silicon substrate.

Figure 2.4: An example of an XRR spectra of a gold film on a silicon substrate at different film thick-
nesses [30]. The frequency of the oscillations increases with increasing film thickness.

2.2.2 X-Ray Diffraction

X-ray diffraction (XRD) is used to characterize the in- and out-of-plane lattice parameters

of the film as well as the substrate. These parameters are used to determine the strain state of the

film. This information can be used to calculate the amount of strain on the film by comparison to

the bulk values. The Bragg equation,

2d sin θ = nλ, (2.9)

is satisfied when, for a given plane spacing, dhkl , the x-rays of wavelength λ are incident on the

sample at an angle θ from the plane. This equation can be derived by a simple geometric argument

if we examine a schematic of two parallel planes, as shown in Figure 2.5. Two parallel, coincident

x-rays will constructively interfere when two adjacent planes of atoms are spaced such that the

20
path length difference of the two x-rays is an integer multiple of the wavelength of the x-rays.

Figure 2.5: X-rays ’1’ and ’2’ are shown to be parallel and coincident. X-ray ’1’ reflects from the surface
plane of atoms while x-ray ’2’ reflects from the adjacent, parallel plane of atoms. They each follow
the law of specular reflection so that all angles shown are the same. The plane spacing dhkl and the
incident angle θ dictate the path length difference of the two x-rays [31].

This measurement is done for the out-of-plane peak, (002), because it provides a high

intensity signal for this sample set. As we will see in the next section, it is useful to examine a

plane with in- and out-of-plane components. The oscillations around the film peak in Figure 2.6

are known as Keissig fringes. Keissig fringes occur when reflected x-rays from the film-substrate

interface constructively interfere with the surface reflected rays. Any mosaicity, non-uniformity

of d-spacing throughout the film thickness, or other imperfections in the crystalline quality of the

film will suppress the fringes. Thus, Keissig fringes in the data are an indication of good crystalline

quality [32].

21
Figure 2.6: XRD pattern around (002) peak of LSCO film on a (001) oriented LSAT substrate [33].

2.2.3 Reciprocal Space Maps

Reciprocal space is the Fourier transform of the physical crystal lattice. Every point in

reciprocal space corresponds to a (hkl) plane in the physical lattice. If the physical lattice vectors

are ~a, ~b, and ~c, then we can define the reciprocal lattice parameters as ~a∗ , ~b∗ , and ~c∗ , which are

defined with respect to the physical lattice. We can then find a plane (hkl) in reciprocal space with

a linear combination of the reciprocal lattice vectors,

~ = h~a∗ + k~b∗ + l~c∗ .


G (2.10)

~ vector is proportional to the inverse of the plane


It can be shown that the magnitude of the G

spacing [31],

~ ∝ 1
|G| . (2.11)
dhkl
In a typical reciprocal space map (RSM) for a thin film, the lattice points of the film are

superimposed on the lattice points of the substrate. In growing epitaxial films, we want the in-

22
plane lattice parameters to match in the film and substrate. Figure 2.7b shows an example of a fully

strained film on a substrate. On the left side of the figure the horizontal axis represents the in-plane

lattice parameters and the vertical axis represents the out-of-plane lattice parameters. The right

side of the figure shows the reciprocal lattice vectors that correspond to the physical lattice. In the

fully strained film, the film peaks are vertically aligned with the substrate peaks, which means

the in-plane lattice parameters match up, while the out-of-plane lattice parameters are distorted.

Figure 2.7a shows the case of a full relaxed film. When a film is fully relaxed, neither the in-plane,

nor the out-of-plane lattice parameters will be equal to those of the substrate lattice parameters.

Each point on the RSM will lie on the same line going radially outward from the origin if both the

film and the substrate are cubic. A partially relaxed film means the lattice parameters will vary

in the out-of-plane direction in the film and thus will significantly alter the magnetic properties of

the film [34].

Figure 2.7: RSM schematic of a sample consisting of a generic film on a substrate. The left hand side is
a schematic drawing of the physical situation of a film on a substrate. The right hand side represents
the corresponding reciprocal space map. Used with permission from Dr. Yayoi Takamura.

23
It is not practical to image an entire map shown in Figure 2.7 using point detectors, as it

requires several hours to gather quality data to image a single peak. However, all the information

we need to characterize the strain state of the system can be found by looking at a single peak, or

a smaller number of peaks. An asymmetric Bragg peak such as (204) or (103) is often chosen for

a reciprocal space map. The choice of an asymmetric peak will show both in- and out-of-plane

features on the RSM; without both it would be impossible to determine the strain state of the film.

Figure 2.8 shows the RSM for the (103) peak; we observe that the film is fully strained because the

film and substrate peaks appear at the same [h00] position, which means that the in-plane lattice

parameters are equal.

Figure 2.8: RSM of LSMO on (001) oriented SrTiO3 (STO) around the (103) Bragg peak. Used with
permission from Dr. Binzhi Li [33].

This information about the strain state of the film, coupled with the information about

the crystalline quality, allow us to characterize whether a grown epitaxial film is fully strained or

not. The fully strained state is desirable because it provides a uniform crystal structure throughout

the film, and we can dictate the lattice parameter of the film by changing the substrate if desired.

24
2.3 Superconducting Quantum Interference Device Magnetometry

The superconducting quantum interference device (SQUID) is an extremely sensitive

magnetometer. In this study I have used the SQUID magnetometer to obtain information about

bulk magnetic properties of grown thin films. In essence, the SQUID is a current-to-voltage con-

verter; a sample is passed through a superconducting ring that is interrupted by one or more

Josephson junctions [35]. A schematic of the SQUID is shown in Figure 2.9.

Figure 2.9: SQUID schematic diagram [35].

The superconducting ring can only trap discreet levels of magnetic flux. The smallest, non-

25
divisible amount of flux in a superconducting ring is a single flux quantum (Φ0 ). It is defined

with fundamental constants, which means it is the same for all superconductors,

h
Φ0 = . (2.12)
2e
Thus, the junctions are able to very accurately measure the magnetic flux through the

superconducting ring by utilizing Lenz’s Law. We can also apply different external magnetic fields

to the sample in the SQUID to measure magnetic hysteresis. The substrates of the samples studied

in this paper are diamagnetic and will produce a background signal which must be accounted

for when analyzing the data by subtracting out a linear background. Using the SQUID, we can

measure the magnetic transition temperature, the saturation magnetization, and film coercivity.

2.4 Soft X-Ray Magnetic Spectroscopy Techniques

The two soft x-ray magnetic spectroscopy techniques I used are x-ray absorption (XA)

and x-ray magnetic circular dichroism (XMCD). I will start by describing XA spectroscopy.

2.4.1 X-Ray Absorption Spectroscopy

XA spectra are characterized by sharp increases in absorption when x-ray energies cor-

respond to values characteristic of the element in question. These characteristic values are the

energies required to excite a core electron to an unoccupied level or eject the electron completely,

which results in an increased absorption of x-rays. These sharp increases in the spectra are known

as absorption edges [36]. The spectroscopic notation of the excitation is different depending on

the origin of the photoelectron, as shown in Figure 2.10. The figure shows that the L-edge signal

is due to 2p electrons being excited into the 3d shell, the K-edge signal is due to 1s electrons being

excited into the 3d or 4p shells. Furthermore, in the presence of spin-orbit coupling, the 2p shell

splits into multiple energy levels for the 3d transition metals I use in this thesis. L-edges were uti-

lized to study changes of the electronic, structural, and spin states of the transition metal cluster,

as it provides higher sensitivity than K-edge spectroscopy.

26
Figure 2.10: XA spectroscopy: K and L edge energy levels. The red arrows represents the excited
electrons start and end energy levels for each edge. The L-edge will occur at lower energy because the
energy level difference between the 3d and 2p/2s shells is much smaller than the difference between
the 4p/3d shells and the 1s shell [37].

The spectra in this thesis were collected at the synchrotron in the Advanced Light Source

(ALS), Lawrence Berkeley National Laboratory. The need to continuously tune the x-ray energy

to the absorption edges of the elements in the sample makes the synchrotron source a necessity

in obtaining these spectra. The experiment was performed in total electron yield (TEY) mode. In

this mode the samples are mounted onto copper sample holders with conductive silver paste and

the absorption is indirectly measured by monitoring the photoelectron current from the sample

surface. The core-holes formed by the excitation of photoelectrons are filled by Auger-electrons

emissions. These Auger emissions create secondary electrons (SE) through inelastic scattering

events. The magnitude of the current from the secondary electrons is proportional to the intensity

of the absorption process. The SE have a finite escape depth of about 4 nm to 10 nm and show

27
an exponential decay of intensity into the sample, away from the surface, making this a surface

sensitive technique [38]. The SE escape depth means the top layer must be less than 10 nm thick

or less in order to probe the buried layer underneath the top layer in the sample. XA spectroscopy

is widely used to determine the local geometric and electronic structure of matter [39].

2.4.2 X-Ray Magnetic Circular Dichroism

XMCD is used to characterize the element specific magnetic properties of thin films.

XMCD is the difference in the XA spectra acquired using right and left circularly polarized (RCP

and LCP) x-rays [38]. The element specificity comes from fact that the incident x-rays are tuned to

the absorption edges of specific elements.

XMCD has been widely used to characterize the magnetic interfacial properties in het-

erostructure thin films. The element specificity allows one to separate out the magnetic behavior

of the different layers in a multilayer film [40]. The magnetic properties of the LSMO and LSCO

thin films are mainly dictated by the d-orbitals of the transition metal atoms in the compound

(Mn and Co), thus we investigated the L-edge because it involves 2p to 3d transitions. Figure 2.11

shows the main ideas of the XMCD technique.

28
Figure 2.11: Working principles of XMCD [41]. Figure discussed in text.

If the transition metal oxide has ferromagnetic ordering, there will be an imbalance in spin-down

and spin-up holes in the 3d orbitals. This leads to an energy shift in the bands by an amount deter-

mined by the exchange energy [4]. The LCP or RCP x-rays will transfer their angular momentum

to the photoelectrons, causing them to preferentially excite in either a spin-down or spin-up state.

The two peaks in the L-edges are due to the spin-orbit coupling in the 2p core shell, and since

the spin-orbit coupling is opposite in L2 as in L3 , we see a flip in the XMCD (difference) spec-

tra between the two edges. The ’L’ notation is basic spectroscopic notation where the L2 and L3

denote the 2p1/2 and 2p3/2 levels, respectively [42]. Figure 2.12 shows the XA spectra with the

corresponding XMCD spectra.

29
Figure 2.12: Example XMCD spectra at the Co L-edges for right (µ+ ) and left (µ− ) circular polarization
[43].

The polarized x-rays will act as a detector for the spin of the photoelectrons. The intensity

of the spectra is proportional to the number of empty d states of a given spin, since spin-flips are

forbidden in electric dipole transitions. Figure 2.12 illustrates the XA spectra for RCP or LCP x-

rays. Based on Figure 2.11, we observe a higher intensity for RCP x-rays, due to spin-up angular

momentum imparted to the core electrons at the L3 edge, and the presence of more spin-up holes

in the d-obitals. At the L2 edge the spin-orbit coupling is opposite so we see the intensities flip as

RCP now excites spin-down photoelectrons.

The XA and XMCD spectra provide a wealth of information about a sample, including:

local binding environment, the valence states, and characteristics of the magnetic ions [38, 40].

With the above mentioned magnetic and physical characterization complete, we can gain consid-

erable insight on the magnetic properties of a thin film bilayer, and the interactions that occur at

the interface between the two films.

30
Chapter 3

Physical and Magnetic Characterization


of Bilayer Perovskite Oxide Samples

LSCO/LSMO and LSMO/LSCO heterostructures were grown epitaxially on (001)-oriented

(LaAlO3 )0.3 (Sr2 AlTaO6 )0.7 (LSAT) substrates by pulsed laser deposition. The film thicknesses and

ordering will be denoted using the notation CxMy, where the C and M correspond to LSCO and

LSMO, respectively; the x and y correspond to the thickness of the layer they succeed. All films

were deposited with a laser pulse repetition rate of 1 Hz, and the temperature of the substrate is

held at 700◦ C throughout the process. After the deposition, the films were cooled to room temper-

ature from 700◦ C. All experimental measurements were conducted at 80 K.

3.1 X-Ray Reflectivity

The x-ray reflectivity (XRR) curves shown in Figure 3.1 were obtained with a Bruker D8

4-circle diffractometer using copper Kα x-rays. The data was fit using the Leptos software from

Bruker. The physical parameters in Tables 3.1, 3.2, and 3.3 show the results from the fits shown in

Figure 3.1. At the Cu x-ray energy, the atomic scattering factors for the LSMO and LSCO layers

are nearly identical, hence the reflectivity measurements can accurately discern the total thickness

of the bilayers, but cannot distinguish between individual layers.

31
Figure 3.1: X-ray reflectivity data for samples M4C4, C4M4, and C8M4. Fits were calculated using
Bruker Leptos software.

Based on the fitting parameters the LSMO layer roughness is around 0.5 nm and is con-

sistent throughout the samples while the LSCO layer roughness is much less. The exception to this

is sample M4C4 in which the LSMO layer was deposited first. The roughness of the LSMO layer is

transferred to the LSCO layer as the LSCO must copy the growth surface. The substrate roughness

remains predictably low in all cases. The thicknesses of the layers are close to the target values

of either 4 or 8 nm, including the roughness as an error on the layer thickness. Interface rough-

ness induced FM coupling, or ”orange peel” coupling, becomes significant when the interfacial

roughness exceeds 0.5 nm [45]. This type of coupling is magnetostatic and occurs between two

magnetic layers with in-plane magnetization, separated by a non-magnetic spacer layer. Given

that the roughness of the interfacial layer reaches a maximum of 0.53 nm in all samples discussed

here, we assume that the effect is negligible. The cost function in Tables 3.1, 3.2, and 3.3 represents

the logarithmic difference between each simulated and experimental data point, and it is used as

a metric for goodness of fit.

32
Table 3.1: M4C4

Layer Thickness (nm) Roughness (nm) Density (g/cm3 )

LSCO 4.25 0.55 6.25

LSMO 4.50 0.53 5.79

LSAT (sub) – 0.06 6.74

Energy 8048 eV

Cost Function Value 8.07 × 10−3

Table 3.2: C4M4

Layer Thickness (nm) Roughness (nm) Density (g/cm3 )

LSMO 4.01 0.59 6.44

LSCO 4.49 0.01 6.70

LSAT (sub) – 0.02 6.74

Energy 8048 eV

Cost Function Value 4.73 × 10−3

Table 3.3: C8M4

Layer Thickness (nm) Roughness (nm) Density (g/cm3 )

LSMO 4.18 0.49 6.44

LSCO 8.39 0.20 6.52

LSAT (sub) – 0.02 6.74

Energy 8048 eV

Cost Function Value 1.58 × 10−3

From the XRR models, we can conclude that samples M4C4, C4M4, and C8M4 are close

to the desired layer thicknesses with relatively smooth interfaces and surfaces. The density of each

33
layer is also within norms which indicates that no significant defects, such as oxygen vacancies,

have formed. The exception to this is sample M4C4, where LSMO layer density is lower than

expected. This is likely due to the large roughness of the LSMO layer which carries over to the

LSCO layer.

3.2 X-Ray Diffraction

The x-ray diffraction data in Figure 3.2 was acquired around the (002) peak using the

same diffractometer to gather the XRR data. A Ge (220) 2-bounce monochrometer was used to

isolate the Kα1 x-rays. The fits were made with Bruker’s Leptos software. The Kiessig fringes

around the film peaks are indicative of smooth interfaces and a uniform d-spacing through the

film thickness [46]. The relatively small volume of the M4C4 and C4M4 bilayers do not produce

much signal, and as the x-rays penetrate mostly into the substrate, a much stronger signal for the

substrate peak is observed. For the film peak, these finite thickness effects make the peaks broad

with low intensity. The bulk lattice parameters of the substrate and films are supplied by the

material manufacturers and confirmed with measurements of single layer samples of thicknesses

great enough to ensure full relaxation of the films. These parameters are similar to one another,

which minimizes the lattice distortions of the films. Consequently, in Figure 3.2, the film peaks are

close to the substrate peak and can be difficult to distinguish, especially when we consider that

there are thickness fringes produced from both layers in the film which contribute to the overall

signal.

34
Figure 3.2: X-Ray diffraction data from the bilayers. The notation of LSCO and LSMO indicate the
approximate location of the individual film peaks. Fits were calculated using Bruker Leptos software.

Table 3.4 shows the tabulated lattice parameters as determined from the Leptos models.

The bulk pseudocubic lattice parameters of the LSMO, LSCO, and LSAT layers are 3.894 Å, 3.83

Å, and 3.868 Å, respectively [24–26]. The lattice mismatch for LSMO and LSCO with respect to

the LSAT substrate is 0.67% compressive strain and 0.98% tensile strain, respectively.

Table 3.4: Lattice parameters from Leptos XRD models

Sample Layer OOP Lattice Parameter (Å) Layer Thickness (nm)

LSCO 3.82 4.25


M4C4
LSMO 3.89 4.50

LSMO 3.896 4.01


C4M4
LSCO 3.79 4.49

LSMO 3.897 4.18


C8M4
LSCO 3.797 8.39

All LSAT 3.868 -

35
While the in-plane lattice parameters of the films are forced to be the same as that of

the substrate, the out-of-plane (OOP) lattice parameters have no such restriction. Between sam-

ples C4M4 and C8M4, Table 3.4 shows that changing the thickness of the LSCO layer has very

little effect on the out-of-plane lattice parameters of either film. However, as shown in Table 3.5,

switching the film order in M4C4 from C4M4 shows that the LSCO layer lattice parameter is sig-

nificantly affected while the LSMO layer lattice parameter remains largely unchanged. This result

is due to LSCO being much more sensitive to the template surface for growths. Because of the

inherent octahedral tilts associated with the rhombohedral structure of both LSMO and LSCO,

LSMO represents a different template surface than LSAT directly. Previous studies have shown

that LSMO quickly returns to the rhombohedral tilt structure regardless of the substrate upon

which it is grown [47].

Table 3.5: The percent difference in out-of-plane lattice parameters for two cases: C4M4 → C8M4 and
C4M4 → M4C4.

C4M4

LSCO LSMO

LSCO 0.18% -
C8M4
LSMO - 0.03%

LSCO 0.79% -
M4C4
LSMO - 0.15%

3.3 Reciprocal Space Maps

The reciprocal space maps (RSM) shown in Figure 3.3 were taken at the Stanford Syn-

chrotron Radiation Lightsource (SSRL) beamline 7-2 on the SLAC National Accelerator Labora-

tory. The small volume of the thin films and their close lattice match with the substrate necessitates

the use of a highly monochromatic and bright x-ray beam present at sychrotron sources to obtain

good quality data. Each of the three films are fully strained as indicated by the film and substrate

36
peaks being aligned vertically at the same ’H’ value. With our x-ray diffraction measurements we

can reliably say that the grown films are fully strained, epitaxial, and of good crystalline quality.

Figure 3.3: Reciprocal space maps of (202) substrate’s film peaks for the bilayers. The vertical alignment
of the substrate and film peaks in each sample indicates that the films are fully strained.

37
3.4 Soft X-Ray Magnetic Spectroscopy

Previous studies performed with a fixed LSMO thickness of 6 nm indicated that for

LSMO:LSCO thickness ratio ≤ 1, the LSCO film experienced magneto-electronic phase separa-

tion where the film displayed three distinct magnetic layers separated vertically. Dr. Li found

that the softer LSCO layers present at the LSMO/LSCO interface showed a Co2+ signature in

the XA spectra, while the harder LSCO layer beneath it demonstrated a mixed Co3+/4+ valence

state. These signatures were determined by comparing Co2+ and Co3+/4+ XA reference spectra

to experimental data from the thin film samples. Dr. Li was able to approximate the amount of

each ion in a given samples by means of a linear combination of the reference spectra fitted to the

experimental data.

In this work, the LSMO thickness was decreased to 4 nm so different film thickness ratios

could be explored. The effect of growth order on the magnetic properties was also investigated.

Figure 3.4 plots a comparison of the Co XA spectra for sample C4M4 and M4C4 together with

reference spectra. The C4M4 sample has a small shoulder on the right hand side of the main peak,

at the L3 edge, around 780.4 eV, and labeled as feature ’C’ in Figure 3.4. This small shoulder is

indicative of a Co2+ signal as seen by comparison to the reference spectra. Interestingly, when

we flip the layer order in sample M4C4, this Co2+ signal completely disappears and the LSCO

layer closely resembles the Co3+/4+ reference spectra from the LSCO single layer sample. A slight

asymmetry of the L2 peak for sample C4M4 can also be observed.

38
Figure 3.4: XA spectra at the Co L edge for samples C8M4, C4M4, and M4C4 in comparison to reference
spectra. The Co2+ reference spectra is from a CoFe2 O4 film that was grown in-house, and the LSCO
reference spectra is from [48]. The dotted lines are included to help the reader identify common peaks
among the spectra.

The XMCD spectra in Figure 3.5 are taken at the Co edge under an applied field of ±1.4

T at 80 K. Again, in sample C4M4 there is evidence of the Co2+ signal in the shoulder to the left of

the main peak; this feature is present in sample C8M4 as well. Sample M4C4 has a relatively small

XMCD signal on the Co edge which aligns with previous work showing that the samples with an

LSCO/air interface need to reach a critical LSCO layer thickness of > 4 nm before the interface will

be magnetic [49]. The out-of-plane lattice parameter of LCSO for sample M4C4 is also nearly 0.8%

different than the C4M4 sample, while there is only a 0.18% difference between C4M4 and C8M4.

This structural difference will result in different magnetic properties as well. For example, in this

case the magnetic moment in the LSCO layer is suppressed as the out-of-plane lattice parameter

decreases.

39
Figure 3.5: XMCD difference spectra at the Co L edge in comparison to reference spectra. Data was
acquired at 80 K. The dotted lines are included to help the reader identify common peaks among the
spectra. Note that the data for sample M4C4 was multiplied by a factor of 5 to enhance its small
moment.

The feature in Figure 3.5 along dotted line ’A’ is common to both of the reference spectra

and is a result of multiplet structure of the Co ion. The XA spectra of the LSMO layer is shown

in Figure 3.6. The Mn3+ valence state is from a single layer sample of Mn2 O3 , and the Mn4+

valence state is from a single layer sample of Li2 MnO3 [50]. Sample M4C4 looks nearly identical

to a single layer LSMO film, reinforcing the idea that charge transfer is suppressed in this sample.

Samples C4M4 and C8M4 begin to round out at common peaks ’A’ and ’B’ indicating the presence

of charge transfer.

40
Figure 3.6: XA spectra at the Mn L edge for samples C8M4, C4M4, and M4C4 in comparison to refer-
ence spectra. The reference spectra used were adapted from [50]. The dotted lines are included to help
the reader identify common peaks among the spectra.

The Mn L edge XMCD spectra for each sample is shown in Figure 3.7, we can see that

samples C8M4 and C4M4 appear to be a mix of the Mn3+ and Mn4+ states. Sample M4C4 again

looks nearly identical to the Mn3+ reference spectra. The difference in magnitude in the moments

of samples C4M4 and C8M4 is likely due to the difference in the amount of charge transfer be-

tween the two samples.

41
Figure 3.7: XMCD difference spectra at the Mn edge in comparison to reference spectra [50].

Figures 3.4 and 3.5 show that we see traces of the same Co2+ signatures in in the C4M4

samples as have previously been seen in bilayer films that have undergone MEPS. Switching the

order of the layers completely eliminates these signatures.

3.5 SQUID Magnetometry

The bulk magnetic properties of the samples were studied with SQUID magnetometry.

Two types of hysteresis loops were acquired, referred to as ’major’ and biased ’minor’ loops. In

the major loop measurement applied was swept from ±2 T to ensure that all the magnetic layers

were saturated in the direction of the magnetic field. In the biased minor loop measurement the

sample is first biased by applying a field large enough to saturate all the layers in the sample,

following which a smaller loop is measured where the maximum field is chosen to only switch

the orientation of the soft layer. These biased minor loops were performed after applying fields

of ±2 T, athen sweeping fields of ±0.15 T, to investigate the presence and strength of exchange

bias between the magnetically hard and soft layers. The LSAT substrate is a diamagnetic material

which results in the presence of a negatively sloping linear response in the hysteresis loop. This

42
background was subtracted from all hysteresis loop measurements.

I compare sample C4M4 to sample C4M6 from Dr. Li’s previous work in Figure 3.8.

The LSMO layer in samples C4M6 is 2 nm thicker than the LSMO layer in samples C4M4. The

figure shows that sample C4M6 has a single magnetic transition around 0.03 T [18]. This behavior

indicates that the two films are strongly magnetically coupled and act as a single magnetic layer.

Previous work showed that increasing the LSCO layer thickness would vertically segregate the

LSCO film into hard and soft layers and two magnetic transitions in the hysteresis loop would be

observed, as expected for a composite material [17]. In accordance with this, sample C4M4 shows

two magnetic transitions indicative of a soft magnetic layer with HC ≈ 0.03 T and a more gradual

transition that begins at 0.05 T and reaches saturation at 0.5 T. The soft magnetic layers shares

the same HC as sample C4M6 (Fig 3.8) which requires ∼1 T to reach saturation. The transitions in

sample C4M4 indicate that reducing the thickness of the LSMO layer to 4 nm is sufficient to induce

the MEPS already seen for thicker LSCO films. For this reason we refer to the LSCO magnetic layer

as ’medium’ LSCO as it is not as hard as the LSCO layer in sample C4M6.

Figure 3.8: Thickness dependence of LSMO layer in SQUID major loops. C4M6 data was used with
permission from Dr. Li.

43
Based on the XAS spectra, charge transfer is present at the LSMO/LSCO interface in

samples C4M6, C2M6, and C8M6. There is, however, less charge transfer in sample C4M4 than in

the CxM6 series as seen in the Co2+ signature in the XA spectra. This is because the Co2+ layer

cannot extend through all 4 nm of LSCO, leaving a ’medium-hard’ LSCO layer.

Figure 3.9: SQUID major loops. There is a second transition at higher field in the C4M4 samples,
similar to the more distinct second transition in the C8M6 sample. This is notable in that it the second
transition appears for a 4 nm thick LSCO film only once the LSMO film thickness is reduced from 6 nm
to 4 nm.

Figure 3.9 compares the general shape of two SQUID major loops. Sample C8M6 is

thicker in both the LSCO and LSMO layers than the C4M4 sample. Sample C4M4 has a low

remnant magnetization like sample C4M6, while C8M6 has a square loop. Figure 3.10 shows hys-

teresis loops for samples C8M4 and M4C4; they have similar shape and have similar values for the

coercivity values. However, the saturation and remnant magnitization are much lower for sample

C8M4 than for M4C4. This is due to the tendency of sample M4C4 to act like a single layer LSMO

film, which is a soft magnetic material. It would also appear that lowering the LSMO layer thick-

ness has an affect on the way the LSCO layer behaves since sample C8M6 shows two transitions

while C8M4 shows one distinct transition, and a second that is smeared out as the moments in the

44
LSCO layer ’drag’ the change in LSMO moments.

Figure 3.10: SQUID major loops. Sample M4C4 is an incomplete data set, but the trend is quite clear
on the positive side of the loop.

Coercivity is chosen as a metric to track the differences amongst each sample as the coer-

civity of the softer transition in the films is a good indicator of the strength the coupling between

the layers. Figure 3.11 tracks the coercivities as a function of the thickness of the LSCO layer.

Initially the coercive field drops quickly, then appears to level out and the LSCO layer thickness

increases.

45
Figure 3.11: Coercivity dependence on LSCO layer thickness.

This same trend continues in Figure 3.12 when the layer order is flipped so that LSMO is deposited

directly onto the substrate. The layer thickness in the pairs of samples in Figure 3.12 are the same.

The coercive field in the LSMO first bilayers seems to be suppressed compared to the LSCO first

counterparts, though more data is needed to justify this claim.

46
Figure 3.12: Coercivity dependence on film order reversal in bilayer thin film samples with equal
thickness of the LSMO and LSCO layers. Samples C20M20 and M20C20 are used with permission
from Dr. Binzhi Li.

The SQUID major loops of sample C4M4 and M4C4 in Figure 3.13 show the similar be-

havior of the loops and the slight shift in coercive field. The coercive field, as a function of both the

LSMO and LSCO layer thicknesses, is plotted in Figure 3.14. The figure suggests that the coercive

field for the soft (or first) transition follows a general trend of decreasing as the thickness of either

layer decreases.

47
Figure 3.13: SQUID major loops for samples C4M4 and M4C4.

Plotting the coercive field from the SQUID data of the varies samples in Figure 3.14 shows a

consistent decrease as the layer thicknesses increase.

Figure 3.14: Coercivity dependence on LSMO and LSCO layer thicknesses.

The relative thicknesses of the samples can lead to an exchange-coupling effect where

48
the LSMO layer is pinned by the harder LSCO layer. The LSCO moments can be saturated in one

direction by applying a large positive or negative bias. This will ”pin” the LSMO moments and

require a much larger applied field to achieve a net magnetic moment of zero. If the field is then

swept back in the opposite direction, the LSMO moments will flip back in line with the LSCO

moments before the applied field has reached zero. If we do this biasing and sweeping in both

directions we get curves similar to the ones in Figure 3.15. The vertical shift has been manually

removed in the loops due to the portion of the LSCO moments that remain fixed in the direction

of the initial biasing field. The horizontal shift in the loops is due to the exchange coupling of

the moments in the soft layer to the hard layer which require a larger field in order to switch

orientations.

Figure 3.15: Minor loops of thick and thin samples with the vertical shift removed to highlight ex-
change bias effects.

If the two minor loops overlap each other when the vertical shift is removed, then the

two layers act independently from one another. If the two loops do not overlap then there is

an exchange spring effect in the films. Sample M4C4 shows some exchange biasing, unlike its

counterpart M20C20, which implies that the difference in thickness must have an effect on the

49
exchange spring behavior of the film. It is unlikely that the much greater thickness of the M20C20

and C20M20 samples have only a small contribution of the overall magnetization coming from

the interfacial layers. If this were true then the C20M20 sample would not have a biasing effect,

which is clearly observed in Figure 3.15.

3.6 X-Ray Magnetic Circular Dichroism: Hysteresis Loops

To compare to the SQUID magnetometry results, we performed XMCD hysteresis loops.

These loops, like the XMCD spectra, are element specific and thus offer information about the

magnetic behavior of each layer individually. The Co-XMCD loops are shown in Figure 3.16 for

sample C4M4, and the layer-flipped sample M4C4. The M4C4 sample has a single smeared out

transition, indicating that the LSCO layer is acting as a single magnetic unit. Sample C4M4 on the

other hand has two distinct transitions, one at approximately 0.03 T, and the second is a gradual

transition that occurs at higher fields. This behavior is indicative of the charge transfer in the

LSCO layer, creating two magnetic layers separated vertically in the film. This suggests that,

instead of ’hard’ and ’soft’ LSCO layers, there are more likely ’medium-hard’ and ’soft’ LSCO

layers in sample C4M4, and that the thickness of the soft LSCO layer is less than it is for sample

C8M6.

50
Figure 3.16: XMCD hysteresis loops at the Co edge for samples M4C4 and C4M4 [18].

Comparing this to the SQUID data is Figure 3.13, it is clear that the coercivity in the bulk

is similar for each sample. Given the obvious difference in the coercivities in Figure 3.16, it is

obvious that the LSMO layer is dominating the bulk magnetic characteristics of the sample. The

SQUID data also shows that the saturation magnetization is lower for sample M4C4 indicating

more competing alignments in the magnetic moments of the sample compared to sample C4M4.

The magnetic behavior in the LSMO layer for the same two samples is shown in Figure

3.17. The C4M4 sample shows a transition at 0.03 T which is very close to the low field transition

in the LSCO layer. Sample M4C4 has the same kind of continuous transition, but saturation takes

place at a much higher field. If sample M4C4 was truly acting as a single continuous layer, both

of the hysteresis loops in Figure 3.16 and 3.17 would be the same. The combination of these two

figures show us that the LSCO and LSMO layers are largely magnetically decoupled from one

another. The hysteresis loop for M4C4 from the SQUID, which resembles the loop for C4M4 in

Figure 3.9, does not have two easily distinguishable steps in it.

51
Figure 3.17: XMCD hysteresis loops at the Mn edge for samples M4C4 and C4M4.

A superposition of the Co and Mn edges for samples M4C4 is shown in Figure 3.18. The

two loops are generally similar in shape, meaning the Co and Mn layers are acting in unison

to some degree. The remnant magnetization in the bulk only appears to be smaller than in the

superposition. This is due to the saturation magnetization being at a higher field than 0.6 T, so the

remnant magnetization in the XMCD loops is artificially raised when the data is normalized.

52
Figure 3.18: XMCD hysteresis loops at the Mn and Co edges for sample M4C4 were superpositioned
and overlayed with the SQUID major loop for sample M4C4.

Figure 3.19 overlays the LSCO layer hysteresis loops of samples C4M4 and C4M6. Sam-

ple C4M6 has a single distinct magnetic transition around 0.03 T while sample C4M4 has two

transitions as discussed above. The exact thickness of the LSMO layer has a profound effect on

the magnetic switching behavior of the underlying LSCO layer, rather than being dictated solely

in the thickness of the LSCO layer itself.

53
Figure 3.19: XMCD hysteresis loops at the Co edge for samples C4M4 and C4M6 [33].

If we look at the magnetic switching characteristics of the LSMO layer in samples C4M4 and

C4M6, we see that they behave almost identically. Changing the thickness of the LSMO layer in

the bilayer samples affects only the magnetic behavior of the LSCO layer.

Figure 3.20: XMCD hysteresis loops at the Mn edge for samples C4M4 and C4M6 [17].

54
Comparing samples C4M4 to C8M4, we see that increasing the LSCO layer thickness

lowers the coercive field in the LSCO layer as seen in Figure 3.21. The C8M4 data for the Co edge

hysteresis loop has been averaged over several loops; it is hard to say with confidence, due to the

error bars involved, if the small step in the C8M4 loop near zero field is similar to the small step

seen in sample C4M4.

Figure 3.21: XMCD hysteresis loops at the Co edge for samples C4M4 and C8M4.

The coercive field for the LSMO layer is also reduced for sample C8M4 compared to sample C4M4,

as seen in Figure 3.22. This behavior may have to do with the interfacial moments in the LSMO

layer coupling to the ’softer’ LSCO layer. The soft LSCO layer is thicker in sample C8M4, and

since the XMCD technique in electron yield mode can only probe the first 10 nm of the sample,

the signal includes less of the hard LSCO layer in the data when compared to sample C4M4.

55
Figure 3.22: XMCD hysteresis loops at the Mn edge for samples C4M4 and C8M4.

The XMCD data presented here shows that changing the LSMO layer thickness has a dis-

tinct effect on the magnetic properties of the LSCO layer in these bilayer systems. The techniques

used to probe the physical and magnetic characteristics of the sample provide a complete picture

of the magnetic behavior at the interface of the bilayer samples. The structural characterization

can provide insight into the physical changes that influence the magnetic behavior in the samples,

and the magnetic characterization provides insight into how the magnetic moments of the distinct

layers interact with one another to form interesting bilayer systems.

56
Chapter 4

Conclusions

In this work, I investigated charge transfer and magnetic interactions in LSMO/LSCO

bilayers. LSMO and LSCO are perovskite oxides that form several different magnetic phases de-

pending on the amount of Sr in the sample. This feature can be manipulated to obtain LSMO and

LSCO with a ferromagnetic-metal magnetic phase. When these materials are combined in a thin

film bilayer system, the magnetic moments interact and manifest different phenomena.

It was found that when changing the LSCO layer thickness, t, in a bilayer system with a

fixed LSMO layer thickness, a critical thickness, t = tc exists. At t ≥ tc the LSCO layer begins to

form different magnetic phases that are layered vertically in the sample. Furthermore, the LSCO

layer splits into a non-magnetic layer, a magnetically hard layer, and a magnetically soft layer,

while the LSMO remains a single magnetic phase. This magnetic phase separation is known as

magneto-electronic phase separation (MEPS) and depends on charge transfer in the Co ions in

the LSCO layer. The Co3+/4+ ions are seen in a single layer LSCO film; this mixed valence state

is magnetically ’hard.’ After charge transfer has occurred, Co2+ ions form at the interface of the

bilayer and are magnetically ’soft’. When t < tc , both layers within the bilayer magnetically

function as a single unit. Previous work suggests that this critical thickness occurs when the fixed

LSMO layer thickness was 6 nm and when 4 nm < t < 6 nm [17, 18]. The work presented here

showed that a different critical thickness exists when the LSMO layer thickness is decreased. With

the fixed LSMO layer thickness at 4 nm, evidence of MEPS is seen when the LSCO layer thickness

57
is 4 nm or greater. This result suggests that the critical thickness depends on the thickness ratio

between the LSCO and LSMO layers, since the critical thickness for LSCO was found to be 6 nm

when the LSMO layer thickness was fixed at 6 nm.

The effect of reversing layer growth order on the magnetic properties of the bilayer sys-

tems was also explored. The SQUID major loops show a 12.5% reduction (∼54 emu/cm3 ) in the

saturation magnetization in the samples with 4 nm layer thicknesses when reversing the growth

order from C4M4 to M4C4; the 20 nm thick samples showed a 17.4% reduction (∼86 emu/cm3 )

in the saturation magnetization when reversing the growth order from C20M20 to M20C20. The

remnant magnetization showed a similar drop when reversing the growth order in the same way;

a 22.6% reduction (∼30 emu/cm3 ) for the 4 nm thick films, and a 26.9% reduction (102 emu/cm3 )

for the 20 nm thick films. This implies that switching the growth order form LSCO/LSMO to

LSMO/LSCO leads to a reduction in these two parameters, regardless of film thickness. The

general shape of the major loops are qualitatively similar, but the XMCD loops begin to show a

distinct difference between sample C8M4 and sample C4M4. Therefore, in order to distinguish the

magnetic properties of the individual layers it is necessary to use element specific analysis, such

as soft x-ray magnetic spectroscopy. At the Co edge, there are two distinct magnetic transitions

in samples with growth order C4M4, while samples with growth order M4C4 show only one con-

tinuous transition indicating that there is either very little, or no MEPS in the LSCO layer. When

the LSCO layer thickness reaches the critical value, t = 4 nm when the LSMO layer thickness is 4

nm, then MEPS occurs and the LSCO layer begins to form a non-magnetic layer of mixed valence

Co3+ /Co4+ at the LSAT/LSCO interface. A soft, magnetically active Co2+ phase is formed at

the LSCO/LSMO interface and these two layers are coupled together and switch together under

applied field. When the LSCO layer thickness is further increased, a magnetically hard, mixed

valence Co3+ /Co4+ phase forms between the non-magnetic layer and the magnetically soft layer.

At the Mn edge, samples with growth order C4M4 have a soft LSMO layer while samples with

growth order M4C4 have a more smeared out hysteresis loop which is evidence that the magnetic

moments in the LSMO layer are being ’dragged’ by the moments in the harder LSCO layer.

The XA spectroscopy data at the Co edge shows that samples with growth order C4M4

58
and C8M4 have a small fraction from Co2+ ions which indicates the presence of charge transfer

in the LSCO layer. This charge transfer creates the MEPS that is responsible for the hard and soft

LSCO layers. Samples with growth order M4C4 do not have a strong indication of Co2+ ions,

which means that little charge transfer happens in the bilayer system. Along with the fact that

we see the magnetic moments in the LSMO layer being dragged in the XMCD Mn loops means

that there is exchange coupling in the M4C4 system, but not charge transfer. However, samples

with growth order C4M4 exhibit both exchange coupling of the LSMO/LSCO interface and charge

transfer at the interface. Thus, simply flipping the layer order suppresses the charge transfer at the

interface in the sample, but leaves some exchange bias coupling in the system. At the Mn edge, the

XA spectra show an indication of Mn4+ ions which again show that some charge transfer is present

in samples with growth order C8M4 and C4M4, while the M4C4 sample shows little indication

that charge transfer has taken place and resemble the XA spectra of a single layer LSMO sample.

This is in agreement with the XA spectra at the Co edge.

The XMCD spectra for samples with growth order C8M4 and C4M4 show much larger

magnetic moments in the LSCO layer compared to samples with growth order M4C4, which is

about 5 times smaller. Previous work has shown that samples with an LSCO/air interface need the

LSCO layer thickness to be greater than 4 nm before the sample will become magnetic [49]. This

conclusion holds for samples with growth order M4C4, as seen by the small magnetic moment

in the sample. The Mn XMCD spectra show strong magnetic moments in the LSMO layer for all

samples. The moment in samples with growth order M4C4 is larger than in samples with growth

order C4M4 due to the lack of charge transfer in the former. Comparing all the samples revealed

trends in the coercive field that are dependent on the thickness of not only the LSCO layer, but

also the LSMO layer. The coercive field in the LSCO/LSMO and LSMO/LSCO bilayers appears to

drop off exponentially as the thickness of the LSMO layer, LSCO layer, or both layers is increased.

This study, in conjunction with Dr. Li’s work, demonstrates the tunability of the magnetic

properties of perovskite oxide bilayers through interface engineering. This work, and others like

it, offers demonstrable evidence that the magnetic properties can not only be controlled through

careful engineering of a system, but the parameter space in which they operate also corresponds

59
to a large range of useful output values [8, 17, 18, 33, 38, 48].

From the initial findings in this thesis, I propose that the critical thickness of the LSCO

layer occurs when the thickness ratio of the LSCO and LSMO layers is 1:1. To further explore

this conclusion, future work must develop samples with several different thickness ratios at sev-

eral different layer thicknesses. In addition to studying the thickness ratios, the substrate/film and

film/air interfaces should also be characterized to determine if their proximity to the LSCO/LSMO

interface has a significant effect on the magnetic properties. These studies can be carried out using

the soft x-ray magnetic spectroscopy techniques described in this thesis.

Another interesting avenue of research would be to investigate how layer thickness af-

fects structural parameters. We know that octahedral tilt patterns are present in complex oxides,

but there is not a quantification of what degree the octahedra are tilted for a given layer thickness,

or how far altered tilt patterns extended away from interfaces. These tilt patterns can be investi-

gated by scanning transmission electron microscopy, or via standard x-ray diffraction techniques

by looking for the so-called half-order peaks in complex oxides with octahedral tilt [51]. However,

it is necessary to use synchrotron radiation to resolve these peaks as they typically have a low

intensity, and Cu Kα x-rays in a standard diffractometer are not bright enough to distinguish the

peaks from the background signal.

60
Bibliography

[1] Michael J. R. Hoch. “The Intriguing Properties of Transition Metal Oxides”. In: AIP Confer-

ence Proceedings 909.1 (2007), pp. 3–8.

[2] Robert W. Eason. Pulsed Laser Deposition of Thin Films: Applications-Led Growth of Functional

Materials. John Wiley & Sons, Inc., 2006.

[3] C Adamo et al. “Effect of biaxial strain on the electrical and magnetic properties of (001) La

0.7 Sr 0.3 MnO 3 thin films”. In: Applied physics letters 95.11 (2009), p. 112504.

[4] Nicola A Spaldin. Magnetic materials: fundamentals and applications. Cambridge University

Press, 2010.

[5] Martin Silberberg. Principles of general chemistry. McGraw-Hill Higher Education, 2012.

[6] J. Stöhr and H.C. Siegmann. Magnetism: From Fundamentals to Nanoscale Dynamics. Springer

Series in Solid-State Sciences. Springer Berlin Heidelberg, 2006. ISBN: 9783540302827.

[7] M. Opel. “Spintronic oxides grown by laser-MBE”. In: Journal of Physics D: Applied Physics

45.3 (2012).

[8] Pavlo Zubko et al. “Interface physics in complex oxide heterostructures”. In: Annu. Rev.

Condens. Matter Phys. 2.1 (2011), pp. 141–165.

[9] J. Wu and C. Leighton. “Glassy ferromagnetism and magnetic phase separation in La1−x Srx CoO3 ”.

In: Phys. Rev. B 67 (17 2003), p. 174408.

[10] H. Kroemer. “Nobel Lecture: Quasielectric fields and band offsets: teaching electrons new

tricks”. In: Review of mordern physics 73.3 (2001).

61
[11] Michael S. Lee et al. “Tailoring Spin Textures in Complex Oxide Micromagnets”. In: ACS

Nano 10.9 (2016), pp. 8545–8551.

[12] AP Ramirez. “Colossal magnetoresistance”. In: Journal of Physics: Condensed Matter 9.39 (1997),

p. 8171.

[13] Stuart SP Parkin, Masamitsu Hayashi, and Luc Thomas. “Magnetic domain-wall racetrack

memory”. In: Science 320.5873 (2008), pp. 190–194.

[14] Josep Nogues and Ivan K Schuller. “Exchange bias”. In: Journal of Magnetism and Magnetic

Materials 192.2 (1999), pp. 203–232.

[15] Hans Boschker et al. “High-temperature magnetic insulating phase in ultrathin La 0.67 Sr

0.33 MnO 3 films”. In: Physical review letters 109.15 (2012), p. 157207.

[16] Joseph H Ngai et al. “Achieving A-Site Termination on La0. 18Sr0. 82Al0. 59Ta0. 41O3 Sub-

strates”. In: Advanced Materials 22.26-27 (2010), pp. 2945–2948.

[17] Binzhi Li et al. “Unconventional switching behavior in La0.7Sr0.3MnO3/La0.7Sr0.3CoO3

exchange-spring bilayers”. In: Applied Physics Letters 105.20 (2014), p. 202401.

[18] Binzhi Li et al. “Tuning interfacial exchange interactions via electronic reconstruction in

transition-metal oxide heterostructures”. In: Applied Physics Letters 109.15 (2016), p. 152401.

[19] Chunyong He. “Magneto-electronic phase separation in doped cobaltites”. PhD thesis. UNI-

VERSITY OF MINNESOTA, 2009.

[20] Krishna Seshan. Handbook of thin film deposition. William Andrew, 2012.

[21] Kenneth A Jackson. Kinetic Processes: Crystal Growth, Diffusion, and Phase Transformations in

Materials. John Wiley & Sons, 2006.

[22] Milton Ohring. Materials science of thin films. Academic press, 2001.

[23] AM Glazer. “The classification of tilted octahedra in perovskites”. In: Acta Crystallographica

Section B: Structural Crystallography and Crystal Chemistry 28.11 (1972), pp. 3384–3392.

[24] DL Proffit et al. “Influence of symmetry mismatch on heteroepitaxial growth of perovskite

thin films”. In: Applied Physics Letters 93.11 (2008), p. 111912.

62
[25] Rajesh V Chopdekar, Elke Arenholz, and Yuri Suzuki. “Orientation and thickness depen-

dence of magnetization at the interfaces of highly spin-polarized manganite thin films”. In:

Physical Review B 79.10 (2009), p. 104417.

[26] MA Torija et al. “Epitaxial La 0.5 Sr 0.5 CoO 3 thin films: structure, magnetism, and trans-

port”. In: Journal of Applied Physics 104.2 (2008), p. 023901.

[27] S Amoruso et al. “Substrate heating influence on the deposition rate of oxides during pulsed

laser deposition in ambient gas”. In: Applied Physics Letters 98.10 (2011), p. 101501.

[28] M Mathews et al. “Magnetization reversal mechanism in La 0.67 Sr 0.33 MnO 3 thin films

on NdGaO 3 substrates”. In: Journal of Applied Physics 107.1 (2010), p. 013904.

[29] H M Christen and G Eres. “Recent advances in pulsed-laser deposition of complex oxides”.

In: Journal of Physics: Condensed Matter 20.26 (2008), p. 264005. URL: http://stacks.iop.

org/0953-8984/20/i=26/a=264005.

[30] M. Yasaka. “X-ray Thin-Film Measurement Techniques”. In: Rigaku Journal 26 (2 2010).

[31] Gregory S Rohrer. Structure and bonding in crystalline materials. Cambridge University Press,

2001.

[32] Stefan Sellner et al. “Strongly enhanced thermal stability of crystalline organic thin films

induced by aluminum oxide capping layers”. In: Advanced Materials 16.19 (2004), pp. 1750–

1753.

[33] B. Li. Interfacial Magnetic Phenomena and Domain Engineering in Complex Oxide Thin Films and

Nanostructures. 2015.

[34] LM Berndt, Vincent Balbarin, and Y Suzuki. “Magnetic anisotropy and strain states of (001)

and (110) colossal magnetoresistance thin films”. In: Applied Physics Letters 77.18 (2000),

pp. 2903–2905.

[35] M. McElfresh. Fundamentals of Magnetism and Magnetic Measurements. 1994. URL: https :

/ / www . mrl . ucsb . edu / sites / default / files / mrl _ docs / instruments /

fundamentals.pdf.

63
[36] Katherine Chamberlain. “The Fine Structure of Certain X-Ray Absorption Edges”. In: Phys.

Rev. 26 (5 1925), pp. 525–536.

[37] Junko Yano and Vittal K Yachandra. “X-ray absorption spectroscopy”. In: Photosynthesis re-

search 102.2-3 (2009), p. 241.

[38] Tobias Funk et al. “X-ray magnetic circular dichroisma high energy probe of magnetic prop-

erties”. In: Coordination Chemistry Reviews 249.1 (2005), pp. 3–30.

[39] Frank De Groot. “High-resolution X-ray emission and X-ray absorption spectroscopy”. In:

Chemical Reviews 101.6 (2001), pp. 1779–1808.

[40] FMF De Groot. “X-ray absorption and dichroism of transition metals and their compounds”.

In: Journal of Electron Spectroscopy and Related Phenomena 67.4 (1994), pp. 529–622.

[41] Joachim Stohr. Magnetic Dichroism Spectroscopy and Microscopy. 2000. URL: http://www-

ssrl.slac.stanford.edu/stohr/ (visited on 04/23/2017).

[42] CJH Schutte et al. “Notations and conventions in molecular spectroscopy: Part 1. General

spectroscopic notation (IUPAC Recommendations 1997)”. In: Pure and applied chemistry 69.8

(1997), pp. 1633–1640.

[43] Gerrit van der Laan and Adriana I. Figueroa. “X-ray magnetic circular dichroismA versa-

tile tool to study magnetism”. In: Coordination Chemistry Reviews 277278 (2014). Following

Chemical Structures using Synchrotron Radiation, pp. 95 –129.

[44] PP Ewald. “Autobiography of a Science.(Book Reviews: Fifty Years of X-ray Diffraction)”.

In: Science 138 (1962), pp. 1254–1255.

[45] DC Parks et al. “Interfacial roughness effects on interlayer coupling in spin valves grown on

different seed layers”. In: Journal of Applied Physics 87.6 (2000), pp. 3023–3026.

[46] D. K. G. de Boer, A. J. G. Leenaers, and W. W. van den Hoogenhof. “Glancing-incidence x-

ray analysis of thin-layered materials: A review”. In: X-Ray Spectrometry 24.3 (1995), pp. 91–

102.

64
[47] EJ Moon et al. “Effect of interfacial octahedral behavior in ultrathin manganite films”. In:

Nano letters 14.5 (2014), pp. 2509–2514.

[48] Binzhi Li et al. “Tuning interfacial exchange interactions via electronic reconstruction in

transition-metal oxide heterostructures”. In: Applied Physics Letters 109.15 (2016), p. 152401.

[49] Binzhi Li et al. “Thickness-dependent magnetic and electrical transport properties of epitax-

ial La0.7Sr0.3CoO3 films”. In: AIP Advances 7.4 (2017), p. 045003.

[50] Ruimin Qiao et al. “Revealing and suppressing surface Mn (II) formation of Na 0.44 MnO 2

electrodes for Na-ion batteries”. In: Nano Energy 16 (2015), pp. 186–195.

[51] SJ May et al. “Quantifying octahedral rotations in strained perovskite oxide films”. In: Phys-

ical Review B 82.1 (2010), p. 014110.

65

S-ar putea să vă placă și