The Study of Energy Levels and Reactions at Oxide-Oxide Heterojunction Interfaces with Photoelectron Spectroscopy

by

Hao-Ting Kung

A thesis submitted in conformity with the requirements for the degree of Master of Applied Science

School of Graduate Studies, Department of Materials Science & Engineering University of Toronto

© Copyright by Hao-Ting Kung 2020

ii

The Study of Energy Levels and Reactions at Oxide-Oxide

Heterojunction Interfaces with Photoelectron Spectroscopy

Hao-Ting Kung

Master of Applied Science (MASc.)

School of Graduate Studies, Department of Materials Science & Engineering

University of Toronto

2020

Abstract

Oxide-oxide heterojunction interfaces (OHIs) play a crucial role in the development of next

generation thin-film devices due to their high tunability. The formation of OHIs is complex as it

can involve events such as charge transfer, interfacial diffusion, and chemical reactions. These

events contribute to the electronic band structure at an OHI and its ability to perform an intended

application. In this work, multiple transition-metal-oxide/oxide OHIs were formed and

systematically studied with in-situ photoelectron spectroscopy. The chemical compositions and

electronic band structures at the OHIs were found deviating from those of the bulk materials and

the driving forces behind the observed phenomenon are discussed. Evidence of charge transfer

such as band bending and interface dipoles were also observed at the OHIs and the results were

found to be deviating from the existing energy level alignment rules. The findings of this thesis

should be taken into consideration when designing OHIs.

iii

Acknowledgments

First and foremost, I would like to extend my gratitude to Professor Zheng-Hong Lu for providing

me with this incredible opportunity to pursue my passion and conduct research related to

nanotechnology at his state-of-the-art facilities. During my time in the group, Professor Lu

provided me guidance when needed and allowed me to have the freedom to explore my own

projects. Professor Lu’s supervising style gave me opportunities to be exposed to various research

areas outside of my thesis topic, which added great value to my graduate studies learning

experience. I would also like to thank Professor Lu for reaching out to his industrial connections

and helping me find potential opportunities that would allow me to pursue a career in OLED. I

really appreciate it.

I would also like to thank the MSE department for providing me with the opportunity to study and

conduct research at the University of Toronto. The Graduate Research Days was such a wonderful

event and I am honored and grateful that the department invited me to the event where I got to

visit the campus and meet with different professors. I especially want to thank Maria and Fanny

for always being helpful with all the questions I had, which really made life, in general, much

easier.

I had the best groupmates I could possibly ask for. Everyone was friendly and helpful. We truly

worked well as a team. I like to thank all present and past Lu Group members that I had the honor

to work with. I would like to extend my appreciation to Peicheng Li and Antoine Dumont.

Peicheng spent a significant amount of his time training me to use different systems in our labs

and he was always there to provide his valuable input when I was unsure about any aspects of my

projects. Antoine always let me observe him using new pieces of equipment, such as SEM and

AFM. Also, Antoine took time from his own research project to help me with moving and installing

the Mini-Etcher. He made that project significantly easier. Thank you all so much!

Last but certainly not least, I want to thank my family and my girlfriend. When I felt it was

necessary for me to attend graduate school to continue developing my career, my family gave me

their full support on my decision to resign from work. The mental support I received from them

kept me moving forward. In Toronto, my sister and girlfriend have always been supportive and

made Toronto feel like home. Thank you everyone!

iv

Table of Contents

Acknowledgments.......................................................................................................................... iii

Table of Contents ........................................................................................................................... iv

List of Tables ................................................................................................................................. vi

List of Figures ............................................................................................................................... vii

Background and Motivation ........................................................................................................1

1.1 Thin-Film Devices and Interfaces ........................................................................................1

1.2 Complexity of Oxide-Oxide Heterojunction Interfaces (OHIs) ..........................................2

1.3 Applications of OHIs ...........................................................................................................3

1.4 Transition Metal Oxide (TMO) ...........................................................................................5

1.5 Key Material Electronic Structure Parameters ....................................................................6

1.6 Thesis Objective and Structure ............................................................................................8

Experimental Techniques ..........................................................................................................10

2.1 Sample Preparation ............................................................................................................10

2.1.1 Physical Vapor Deposition (PVD) .........................................................................10

2.1.2 Ultra-Violet Ozone Oxidation of the Under-Layer Oxide .....................................11

2.2 Photoelectron Spectroscopy (PES) ....................................................................................12

2.2.1 X-Ray Photoelectron Spectroscopy (XPS) ............................................................15

2.2.2 Ultra-Violet Photoelectron Spectroscopy (UPS) ...................................................17

Study of Energy Levels at OHIs ...............................................................................................20

3.1 Introduction ........................................................................................................................20

3.2 Experimental ......................................................................................................................20

3.3 Results ................................................................................................................................22

3.3.1 Reaction at OHI .....................................................................................................22

v

3.3.2 Formation of Additional Electronic States at OHIs ...............................................24

3.3.3 Energy Levels at OHIs ...........................................................................................25

3.4 Discussion ..........................................................................................................................27

3.4.1 Reaction Thermodynamics at OHIs .......................................................................27

3.4.2 Other Potential Factors that can Cause Mo5+ Formation .......................................31

3.4.3 Interface Dipoles and Band Structures at OHIs .....................................................32

3.5 Summary ............................................................................................................................35

Highly Complex OHIs ..............................................................................................................37

4.1 Introduction ........................................................................................................................37

4.2 Experimental ......................................................................................................................37

4.3 Results & Discussion .........................................................................................................37

4.4 Conclusion .........................................................................................................................44

Conclusion and Future Work ....................................................................................................45

5.1 Conclusion .........................................................................................................................45

5.2 Future Work .......................................................................................................................46

Appendix A: Mo 3d XPS spectra of increasing MoO3 thicknesses on GeO2, and ReO3 ..............48

Appendix B: Mo5+ and Mo6+ Percentages .....................................................................................49

Appendix C: UPS Results ..............................................................................................................50

Appendix D: VBM and Ф Progression with Increasing ReO3 Thickness .....................................51

Appendix E: Under Layer Oxides Core Level Shifts ....................................................................53

References ......................................................................................................................................55

vi

List of Tables

Table 2.1: Area ratios of different spin-orbit splitting XPS peak pairs. ....................................... 17

Table 3.1: Bulk electronic structure parameters for the underlying oxides in this paper. Note IE

stands for ionization energy. ......................................................................................................... 27

Table 3.2: Gibbs free energies of all possible oxidation reactions that can take place at the OHIs

under study. The free energy values listed are presented as per mole of oxygen molecules. ....... 28

Table 3.3: The most thermodynamically favorable Mo5+ formation reactions that can take place

at the OHIs under study. The diffusion coefficient of Al in Al2O3 and Si in SiO2 at room

temperature are also included. ...................................................................................................... 29

Table 4.1: Possible reactions at the MoO3/AgOx and ReO3/AgOx OHIs. .................................... 42

vii

List of Figures

Figure 1.1: General schematic of charge transport and the formation of excitons in OLEDs. ....... 2

Figure 1.2: Schematic of various energy levels of MoO3, oxygen-deficient MoO3, and MoO2.

Note that the shaded areas represent electron occupation. ............................................................. 6

Figure 1.3: Schematic diagram of the generic electronic band structure of a semiconductor with

the key parameters labelled. ............................................................................................................ 8

Figure 2.1: a simplified schematic diagram of a thermal evaporation system.............................. 11

Figure 2.2: PHI 5500 Multi-Technique system simplified layout []. ........................................... 12

Figure 2.3: Illustration of take-off angle in PES. .......................................................................... 13

Figure 2.4: Illustration of the difference between K.E. and K.E. measured. ..................................... 14

Figure 2.5: XPS core level photoelectron excitation schematics. ................................................. 15

Figure 2.6: XPS spectrum of molybdenum 3d XPS peaks (3d3/2 on the left and 3d5/2 on the left).

....................................................................................................................................................... 17

Figure 2.7: UPS photoelectron excitation schematics. ................................................................. 18

Figure 2.8: Typical UPS spectrum of MoO3 with key electronic structure parameters labelled. . 19

Figure 3.1: Mo 3d XPS spectra of increasing MoO3 thicknesses on, from left to right, Al2O3,

SiO2, and ITO. Curve fitting with Mo6+ and Mo5+ was performed and the fitted curves are shown.

....................................................................................................................................................... 23

Figure 3.2: Mo5+ percentages as a function of MoO3 thickness for systems formed by growing

MoO3 on Al2O3, SiO2, and ITO. ................................................................................................... 23

Figure 3.3: The valence band region of UPS spectra of MoO3 grown on various oxide films with

the MoO3 thicknesses increase from top to bottom. MoO3 thicknesses, VBM, and defect bands

are shown. Note that the Fermi level was set at zero and the negative x axis represents the energy

viii

levels below the Fermi level. An example of how VBM is determined from a UPS spectrum is

shown in the MoO3 on Al2O3 plot. Also, the defect bands are marked as d1 or d2. ...................... 24

Figure 3.4: i) Measured work function and ii) valence band maximum values of the growth of

MoO3 on Al2O3, SiO2, and ITO films. Note that the Fermi level is used as the reference point and

is set at zero eV energy. ................................................................................................................ 26

Figure 3.5: Mo5+ % at the OHIs plotted against the Gibbs free energies of the most favorable

oxidation reactions. ....................................................................................................................... 29

Figure 3.6: Indium XPS 3d5/2 peaks of the ITO substrate and 0.5 nm thick MoO3 deposited on the

same ITO substrate. ...................................................................................................................... 31

Figure 3.7: Energy level alignment schematics at all OHIs in this study. Ф is the work function,

Δ is the interface dipole, and ΔEVBM is the VBM position under the Fermi level. ....................... 34

Figure 3.8: a) Work functions measured at the OHIs plotted against their associated work

functions of the underlying oxides. b) Interface dipoles plotted as a function of their associated

work functions of the underlying oxides. The MoO3 thickness is 1 nm for all systems. The

empirical equation of the fitted linear line is shown in the plot.................................................... 35

Figure 4.1: Mo 3d XPS core level peaks with 1 nm, 3 nm, and 8 nm of MoO3 deposited on AgOx.

....................................................................................................................................................... 38

Figure 4.2: UPS spectra of MoO3 with different thicknesses on AgOx. ....................................... 38

Figure 4.3: Typical Re 4f XPS spectrum for ReO3. ...................................................................... 40

Figure 4.4: Re 4f XPS core level peaks with 0.5 nm, 1 nm, 3 nm, and 8 nm of ReO3 deposited on

AgOx. ............................................................................................................................................ 40

Figure 4.5: Oxygen 1s XPS peaks of AgOx samples with different UV ozone exposure times. .. 41

Figure 4.6: XPS oxygen 1s peaks of different thicknesses of ReO3 deposited on AgOx. ............ 42

Figure 4.7: XPS oxygen 1s peaks of different thicknesses of MoO3 deposited on AgOx. ........... 43

1

1

Background and Motivation

The information in this chapter is adapted from a paper published as H.T. Kung et al., "Reaction

and Energy Levels at Oxide–Oxide Heterojunction Interfaces" Adv. Mater. Interfaces, 1901456

(2019).

1.1 Thin-Film Devices and Interfaces

As the world becomes more populated and electronically inter-connected, the demand for faster

computing power and more efficient devices also increases. This increasing in demand means that

high performing thin-film electronic devices are needed. Thin-film electronic devices are made of

multiple layers of thin films and these films can be as thin as several nanometers. Some typical

thin-film devices include transistors, organic light emitting diodes (OLEDs), and solar cells. The

interfaces between these layers have crucial effects on the behavior of the devices as they

significantly affect how charges transfer through them. Using OLEDs as an example, when an

external voltage is applied on the electrodes of a device, holes and electrons are injected into the

device via the anode and the cathode, respectively. Excitons (hole and electron pairs) are formed

in the emissive layer where light is emitted. Holes travel on the highest occupied molecular orbitals

(HOME) for organic semiconductors, or in the valence band (VB) for non-organic semiconductors.

Electrons travel on the lowest unoccupied molecular orbital (LUMO) for organic semiconductors

in the conduction band (VB) for non-organic semiconductors. If energy barriers exist at the

interfaces of the different layers a device, more energy is required for the charge carriers to be

transported. As a result, the positions of these energy levels at the interfaces have a great impact

on the charge transport efficiency. Figure 1.1 is a simplified schematic of the formation of

excitons in OLEDs.

2

Figure 1.1: General schematic of charge transport and the formation of excitons in OLEDs.

In the electronic industries and research fields, great focus has been placed on conventional

semiconductors, such as silicon, and organic semiconductors, as they are commonly found in well-

established thin-film devices like CPU transistors and LEDs. As a result, the interfaces in these

devices have been well studied. However, oxide-oxide heterojunction interfaces (OHIs) are poorly

understand and they exhibit a wide range of unique behaviors.

1.2 Complexity of Oxide-Oxide Heterojunction Interfaces (OHIs)

OHIs are extremely complex as multiple phenomena can occur there. In contrast to conventional

semiconductors, where electrons move through the bulk material as independent free particles,

electrons in metals oxide interact with the regularly spaced oxygen ions, which prevents them from

behaving as free particles. This phenomenon enables oxide semiconductors to have unique and

unexpected properties [1]. Due to their broad applications, OHIs have gained significant interests

recently in the research community, especially OHIs involving transition metal oxides (TMO).

TMOs have other physical parameters such as spin, spin-orbit coupling, and flexible lattice during

the formation of OHIs [2]. As a result, TMO-based OHIs can exhibit different characteristics, such

as insulating, metallic, magnetic, and superconducting, depending on their electronic and crystal

structures [1, 3]. One of the most unique phenomena is the spin and orbital reconstruction at OHIs

[4]. It has been found that this phenomenon is mainly due to the rearrangement of charge because

of interfacial charge transfer and electrostatic interactions [3, 4]. It is well known that at any

interface between two materials, charge can transfer to balance the electron chemical potential [4,

3

5] However, this process is complex at OHIs as electron doping at the interfaces can form a region

where the local carrier density varies with altered crystal and electronic structures [4]. For example,

researchers have discovered that the interface of lanthanum aluminate (LaAlO3) and strontium

titanate (SrTiO3) is highly conductive. By themselves, LaAlO3 and SrTiO3 are insulators; however

at the OHI, a 2-dimensional electron gas (2DEG), which is a sheet of free electrons, exists. It was

found that charge transfer occurs at this OHI in order to balance the electron chemical potential

and hence resulting in an accumulation of charge at the interface. This explanation is also

supported by the discovery of the presence of Ti3+ at the OHI, which shows that the Ti4+ in bulk

SrTiO3 is reduced [6]. A similar phenomenon has also been reported at the insulating

heterostructure of SrTi4+O3 and LaTi3+O3 where charge transfer renders this OHI exhibiting

metallic properties [7].

Electrostatic interactions also play a role in changing the properties at OHIs. If one or both of the

oxide layers are not charge neutral at an OHI (possibly due to the existence of defects), structure

relaxation is necessary to compensate this polarity mismatch and stabilize the interface [4]. At

OHIs, the oxides reconstruct by forming oxygen vacancies and having the TMOs assume different

valence states [3, 4]. Interface reconstruction can also be triggered by lattice strain resulting from

lattice mismatch as oxygen vacancies are formed to minimize the strain [4, 8, 9].

Another important factor that can alter the composition and energy band structure at an OHI is

interfacial reaction. It has been reported that oxides can be reduced or oxidized based on the

materials that they are in contact with [10]. Therefore, if diffusion and reaction occur, the physical

and energy band structure at the interface may be different from the bulk structures which may or

may not be desirable for applications.

1.3 Applications of OHIs

Oxide-oxide heterojunction interfaces (OHIs) are broadly used in many different fields in the

electronic industry and here a few examples are named. In transistors, it has been reported that

OHIs can increase electron mobility and hence, improve device performance. Faber et al. [11]

demonstrated by incorporating an In2O3/ZnO heterojunction in a thin-film transistor, the electron

mobility in the device increased by a factor of 100 compared to single-layer ln2O3 and ZnO

devices. This improvement in electron mobility was found to be due to suppression of electron

injection barrier caused by the accumulation of free electrons at the In2O3/ZnO interface. These

4

free electrons also enable the OHI to be highly conductive which greatly improves charge transport

efficiency from the source to the drain. In OLEDs, the implementation of OHIs have been proven

beneficial as well. It was reported by Li et al. [12] that a highly efficient top-emitting OLED device

with an oxidized aluminum anode was fabricated. It was found that even though aluminum oxide

is normally thought to be electrically insulating, by incorporating a MoO3/Al2O3 heterojunction,

the current density of the device was greatly increased. It is reported that if the MoO3 layer is

absent, the performance of the device is significantly reduced. At the OHI, reduction of MoO3 was

noted and Li at el. speculated that the metal aluminum and MoO3 could have diffused through the

oxide layer and reacted with each other, which resulted in improved conductivity. This discovery

is beneficial to the OLED industry, as it demonstrates that OHIs allows devices to achieve high

performance without the need to remove the native oxide on the aluminum anode.

OHIs also play an important role in the field of catalysis. TMOs are widely used to form

heterogeneous catalysts in oxidation and acid-base reactions due to their high structural diversity

and reactive O2- terminated surfaces [13, 14]. It has been reported that MoO3/SiO2 catalyst is

efficient and eco-friendly for the synthesis of 1, 8-dioxodecahydroacridines [13, 14] and the

oxidation of methane [15]. There are many oxides that can be used to support MoO3 based on

different applications. It was also reported by Parmalinana et al. [16] that reactivity of a catalyst

used in partial oxidation of methane to formaldehyde varies greatly depending on what oxide is

used as the support due to the interaction between the TMO and the support oxide. This interaction

controls reducibility and dispersion of the active phase in the catalyst; therefore, the interfaces

formed by the oxides need to be well understood in order to select the most suitable combinations.

The energy levels at OHIs are especially important in photocatalysis as charge transfer efficiency

greatly affects the photocatalytic activity. It is known that many harmful organic compounds can

be decomposed in a solution with the presence of TiO2 exposed to sunlight [17]. However, TiO2

can only absorb radiation in within the ultra-violet (UV) region. It was reported that WO3/TiO2

and MoO3/TiO2 mixed oxide systems show improved photocatalytic activities compared to that of

a pure TiO2 system. This phenomenon occurs because charge transfer happens between TiO2 and

MoO3 or WO3. Electrons transfer from TiO2 to MoO3 or WO3 and vice versa for holes. This

transfer of charge causes an accumulation of holes in the valence band of TiO2 and electrons in

the conduction band of MoO3 or WO3. As a result, the lifetimes of the electrons and holes are

greatly increased which improves the photocatalytic activities [18].

5

1.4 Transition Metal Oxide (TMO)

In this work, OHIs were formed by growing transition metal oxides (in this case MoO3 and ReO3)

on various oxide substrates. A great emphasis is placed on the OHIs involving MoO3 in this study,

so more detailed background information on MoO3 is provided in this section. There are three

main reasons why TMOs were chosen as the over-layer oxide:

1. MoO3 and ReO3 are TMOs so they possess more than one valence states which enables

any changes to the electronic structure and chemical states taking place at the OHIs

observable with photoelectron spectroscopy.

2. MoO3 and ReO3 can be deposited via physical vapor evaporation; therefore, the film

thickness can be easily monitored and controlled.

3. MoO3 is widely used in the OLED industry as a hole injection layer and it has been

extensively studied by researchers [10, 19, 20]. Therefore, its electronic structure is well

understood which assists us with perform reliable quantitative analyses on the samples

MoO3 is a wide band gap (3.0-3.1 eV) and high work function (6.8-6.9 eV) semiconductor [10].

Even though the Mo cations in the stoichiometric MoO3 exhibits an oxidation state of 6+, oxygen

deficiency causes a small fraction of Mo6+ to be reduced to Mo5+. The presence of Mo5+ forms a

defect band centered at approximately 1 eV below the Fermi level. If the MoO3 is further reduced,

metallic oxide, MoO2 can form which produces another defect band centered at around 2 eV under

the Fermi level. The energy levels of the defect bands are supported by previously reported

experimental results [10, 20] and reported DFT calculations [21, 22, 23]. Figure 1.2 depicts the

energy-level diagram of MoO3, oxygen-deficient MoO3, and MoO2. Note that in this thesis, the

term MoO3 encompasses pure MoO3 with reduced states present.

6

Figure 1.2: Schematic of various energy levels of MoO3, oxygen-deficient MoO3, and MoO2. Note

that the shaded areas represent electron occupation.

1.5 Key Material Electronic Structure Parameters

In this section, the key material electronic structure parameters such as the work function and

ionization potential are introduced, since these parameters are critical for the understanding of the

results and analyses of this work.

The work function is defined as the minimum energy required to remove an electron from a solid

to a point just outside the solid surface [24]. According to the definition, the work function Φ is

equal to the energy level difference between the vacuum level, EVAC and Fermi level, EF:

Φ = E − E (1)

In metals, EF is the boundary between unoccupied and occupied states in a continuous density of

states at absolute 0 K. While in semiconductor, there are generally no states inside the bandgap;

consequently, the Ef in a semiconductor only represents the inflection point in Fermi distribution

of charges and is located between the conduction band minimum, CBM, and valence band

maximum, VBM [24]. Note that the work function of a semiconductor is usually not considered a

7

material’s property as the Fermi level can shift due to factors such as the presence of defects and

impurities.

There are two components in defining the work function, a bulk component and a surface

component. The bulk component, which is the dominant one, is associated to the electron chemical

potential of an electron in a solid [24]. The electron chemical potential is defined as the change in

free energy when an electron is added or removed from the solid. The surface component is

associated with the surface dipole that only exists at the solid-vacuum interface, and the dipole

formed from the redistribution of charges on the surface of the solid [24].

Ionization energy (IE) is defined as the energy required to remove an electron from the highest

occupied energy state to the vacuum level. IE is expressed as the difference between EVAC and

VBM:

IE = EVAC – VBM (2)

Electron Affinity (EA) is defined as the amount of energy gained by the solid when an electron is

added from the vacuum level to the lowest unoccupied energy level. Assuming the amount of

electron holes in the HOMO is negligible, EA is expressed as the difference between EVAC and

CBM:

EA = EVAC – CBM (3)

Figure 1.3 is a generic energy band diagram with key electronic parameters labelled.

8

Figure 1.3: Schematic diagram of the generic electronic band structure of a semiconductor with

the key parameters labelled.

1.6 Thesis Objective and Structure

Even though extensive research has been done on inorganic semiconductors, inorganic

semiconductor-organic, and metal oxide-organic interfaces, there is little reported work on oxide-

oxide interface heterojunctions. One of the most widely used technique for surface and interface

analysis is photoelectron spectroscopy (PES). So far, PES has been used in limited cases of OHIs

[6]. In general, the research community urgently needs a systematic study of multiple OHIs that

may lead to general trends or rules at these complex interfaces. As a result, in this thesis, multiple

OHIs were studied using in-situ X-ray photoelectron spectroscopy (XPS) and ultraviolet

photoelectron spectroscopy (UPS) aiming to provide some guidance in designing thin-film devices

that contain OHIs for their specific applications.

The structure of this thesis is as follows:

9

Chapter 2 provides the background information on the experimental techniques involved

in this study. The techniques include physical vapor deposition (PVD), oxide layer growth,

and PES (XPS and UPS).

Chapter 3 contains the result and discussion from the PES study of multiple OHIs formed

from growing MoO3 and ReO3 on various under layer oxides. The experimental results

cover the chemical compositions and the electronic band structures of the OHIs. This

chapter also explains the trends and phenomena identified, and the energy levels at OHIs

are evaluated using the existing energy level alignment rules.

Chapter 4 explores more complex TMO/AgOx OHIs. These OHIs deviate from the

observations made in Chapter 3 and the possible driving forces behind the deviations are

investigated using PES.

Chapter 5 summarizes the studies discussed in Chapters 3 and 4, and presents the

opportunities for future work based on the content of this thesis.

10

Experimental Techniques

2.1 Sample Preparation

In this section of the chapter, the sample preparation techniques involved in this work are discussed

in detail.

2.1.1 Physical Vapor Deposition (PVD)

Physical vapor deposition (PVD) is the thin-film fabrication technique employed in this work. The

type of PVD utilized is vacuum thermal evaporation (VTE). Material deposition with VTE

involves heating the oxide power to a temperature above the material’s sublimation temperature.

At this temperature, the solid power is sublimated and it travels through ultra-high vacuum (UHV),

and condenses and solidifies on the oxide under layer. The UHV environment allows the deposited

film to contain a minimal amount of impurities and hence, the high-quality samples are highly

reproducible. The thickness of the deposited film is monitored by a quartz crystal microbalance

(QCM). By knowing the density of the evaporant and the configuration of the evaporator, the QCM

is calibrated accordingly.

Due to the simple operating principles of VTE, the thicknesses of deposited films are easily

monitored and controlled. Also, by utilizing shadow masks, thin films with different patterns can

be fabricated. As a result, VTE is utilized in numerous different fields, such as coatings for optical

or wear-prevention purposes, thin-film devices, and aluminum coated foils (e.g. gift wraps) [25].

Furthermore, VTE is still the most commonly used thin-film fabrication technique used in OLED

commercial production despite the rise of printable electronics. Therefore, this process is well

studied and understood. Figure 2.1 is a simplified schematic of a VTE chamber that is used for

oxide layer deposition.

11

Figure 2.1: a simplified schematic diagram of a thermal evaporation system.

In this work, the deposition of MoO3 and ReO3 were performed in the PHI5500 Multi-technique

System metal deposition chamber and organic deposition chamber, respectively. The heaters used

in these two chambers are both mounted Knudsen cells (K-cells). The vacuum pumps used to

maintain UHV for the metal deposition chamber and the organic deposition chamber are an ion

pump and a cryogenic pump, and their base pressures are approximately 10-9 Torr and 10-8 Torr,

respectively.

2.1.2 Ultra-Violet Ozone Oxidation of the Under-Layer Oxide

The oxide under layers are prepared through three different methods. The first method is the above

mentioned VTE technique. This technique is applicable for the metal oxides that have a lower

sublimation temperature, such as MoO3 and ReO3. However, for oxides with higher sublimation

temperatures, a high temperature heating source is needed in order to evaporate them. As a result,

to prepare under-layer oxides, such as Al2O3, a different method was implemented. The second

method involves depositing metals, such as aluminum and silver, via VTE, and then the metals

were oxidized in a ultra-violet (UV) ozone chamber in atmosphere. UV ozone operates by

decomposing oxygen molecules and forming O3, and then the O3 is decomposed into high energy

oxygen atoms, O [26]. These activated oxygen atoms react with the metal and form metal oxide.

The third method applies to semiconductors with existing native oxide on their exposed surfaces,

such as silicon and germanium wafers. This method involves treating the wafers with UV ozone

12

without prior removal of the native oxide layer. The process ensures that any non-oxidized material

at the surface is oxidized.

2.2 Photoelectron Spectroscopy (PES)

The main object of this work is observe how the properties of the OHIs deviate from those of their

bulk materials, which potentially can advance our understanding of these interfaces and help

board-reaching conclusions be reached. PES is an extremely powerful technique when it comes to

thin film and interface analysis as it can provide essential information on the chemical

composition, film thickness, oxidation states, valence band structure, Fermi level, and work

function of a sample.

In this work, PES was performed in the PHI5500 Multi-Technique System, which is shown in

Figure 2.2, with the base pressure of 10-9 Torr.

Figure 2.2: PHI 5500 Multi-Technique system simplified layout [27].

PES is a surface analytical technique that is based on photoelectric effect discovered in 1887 by

Hertz. Photoelectric effect is associated with the emission of electrons from a material exposed to

radiation and this type of emission of electrons can be summarized into three steps [28, 29]:

1. Electrons inside of the material absorb the energy of the incident photons and become

excited. This energy is converted to kinetic energy;

2. Excited electrons move away from their ground states and travel within the material

with a fraction of them reaching the surface;

13

3. Electrons with enough kinetic energy can be emitted from the sample surface into the

vacuum. The emitted electrons are called photoelectrons.

The kinetic energy (K.E.) of a photoelectron is given by

K. E. = ℎ𝑣 – B. E. − Φ (4)

Where h is Planck’s constant, v is the incident radiation frequency, B.E. is the binding energy of

the electron in the material with respect to the Fermi level, and Φ is the work function of the

sample. Some of the photoelectrons are collected by a hemispherical electron analyzer and guided

to a spectrometer which measures the number of electrons at a specific kinetics energy. By varying

the tilt of the sample, photoelectrons ejected at different take-off angles, ϴ, are detected. The take-

off angle is shown in Figure 2.3.

Figure 2.3: Illustration of take-off angle in PES.

It is worth mentioning that the actual measured kinetic energy, K. E. measured, is different from the

K. E. in Equation 4. This difference is due to the sample and the spectrometer have different work

functions (different vacuum levels). This difference can be seen in Figure 2.4.

14

Figure 2.4: Illustration of the difference between K.E. and K.E. measured.

By taking the difference between K.E. and K.E.measured into consideration, K.E.measured can be

calculated with Equation 5.

K.E.measured = hv – B.E. - Φs (5)

Where Φs is the work function of the spectrometer which can be preset by the user.

Using Equation 5, a PES system is able to measure the kinetic energy of the photoelectrons which

allows the binding energy to be determined.

The intensity of the emitted photoelectrons, I, at a certain take-off angle, ϴ, from a certain depth,

d, can be determined using Equation 6.

I = Io exp(-d/(λsin ϴ)) (6)

Where Io is the initial intensity of photoelectrons at depth, d, and λ is inelastic mean free path of

the emitted photo electrons in a particular material.

15

There are two different modes of PES commonly used for characterization of semiconductor thin

films: X-ray photoelectron spectroscopy (XPS) and ultra-violet photoelectron spectroscopy (UPS).

Each technique is discussed in the following sections.

2.2.1 X-Ray Photoelectron Spectroscopy (XPS)

The typical X-ray used in the XPS is Aluminum Kα or Magnesium Kα radiation which has a

photon energy of 1486.7 eV or 1253.7 eV, respectively. In this work, the X-ray source used is

Aluminum Kα. Due to the high energy of X-ray radiation, the incident photons possess a sufficient

amount of energy to ionize the core-level electrons of the sample material. This excitation process

is shown in Figure 2.5. The probing depth of XPS can be up to approximately 10 nm depending

on the inelastic electron mean free path of the sample material [28, 29].

Figure 2.5: XPS core level photoelectron excitation schematics.

XPS is a powerful surface analytical technique because of numerous reasons. First of all, the core-

level photoelectrons have distinct binding energies and they can be used as the fingerprint of each

16

element. As a result, XPS is often used to identify elements and their amounts relative to other

elements present in a sample; hence, it can be used to calculate the elemental composition of a

sample. The atomic percentage of an element in a sample is calculated with the following equation:

𝐶 = /

∑ / (7)

Where 𝐶 is the atomic fraction of an element, A is the total area of an element’s XPS peak(s), and

ASF is the atomic sensitivity factor which is used to standardize the XPS peak areas obtained for

each element. ASF is needed because the cross section of interaction between the incident radiation

and each element is different which affects the probability of the photoelectric effect occurring.

As a result, the total counts of photoelectrons of some elements are inherently higher than others

[29].

XPS is also capable of identifying the different oxidation states and bonds that are present in a

sample. The binding energy of an electron is the energy required to remove an electron from its

ground state to the Fermi level. When an atom is bonded to another atom, its energy states are

slightly distorted and therefore, the binding energies of its core-level photoelectrons deviate from

those of a single atom. A similar concept can be applied to explain the shift in binding energies

based on different oxidation states. Typically, the binding energy of an element increases with

increasing oxidation states. The binding energy of an electron is determined by the electrostatic

interaction between the electron and the nucleus, but the degree of this interaction is reduced by

the shielding caused by the other electrons surrounding the nucleus. As a result, when an ion has

higher oxidation state, the binding energy of a photoelectron is higher as the shielding is weaker

[29].

Figure 2.6 shows an XPS spectrum of molybdenum, Mo, in a MoOx sample. Note that there are

two peaks for the 3d orbital. Other than the s orbitals, all other orbitals generate two possible states

with different binding energies, and this phenomenon is called spin-orbit splitting or j-j coupling

[30]. The nomenclature for the peaks is nlj, where n is the principle quantum number, l is angular

momentum quantum number, and j is l + s, where s is the spin angular momentum number. The

area ratios between a pair of XPS peaks are shown in Table 2.1.

17

Table 2.1: Area ratios of different spin-orbit splitting XPS peak pairs.

Orbital Subshell Peaks Area Ratio

p p1/2 & p3/2 1:2

d d3/2 & d5/2 2:3

f f5/2 & f7/2 3:4

The two fitted curves represent the two different types of Mo in the sample, namely Mo6+ and

Mo5+. Note that the binding energies of the Mo6+ peaks are higher than those of the Mo5+ peaks

due to its higher oxidation state.

Figure 2.6: XPS spectrum of molybdenum 3d XPS peaks (3d3/2 on the left and 3d5/2 on the left).

2.2.2 Ultra-Violet Photoelectron Spectroscopy (UPS)

The typical radiation energy of the UV radiation used in UPS is Helium Iα and its energy is around

21.22 eV. Since UV radiation is much less energetic than X-ray, it can only excite and ionize

electrons in the valance energy levels, as shown in Figure 2.7. As a result, the main function of

UPS is to determine the electronic structure of the valence region of a sample. Since the interaction

cross sections of UV and the electrons are large, UPS enables users to accurately determine

electronic structure parameters such as, Φ and Ef.

18

Figure 2.7: UPS photoelectron excitation schematics.

There are two regions in a UPS spectrum: secondary electron region and valence region. The signal

in the secondary electron region is generated by from the inelastically-scattered electrons, so this

region does not directly provide information about the valence electronic structure. However by

applying the following equation, the work function can be determined:

Φ = ℎ𝑣 − B. E. (8)

Where B.E.SECO is the binding energy of the secondary electron cut-off (SECO) which is

determined by linearly extrapolating the cut-off edge of the secondary electron region.

The valence region shows the energy states in the valence region, hence, it provides information

on Ef and the defect states that potentially exist in the band gap.

By combining the information obtained from both regions of an UPS spectrum, the IE of a sample

can be calculated using the following equation:

19

IE = 𝐸 + Φ (9)

A typical UPS spectrum of MoO3 is shown in Figure 2.8. The key electronic structure parameters

are labelled and the Fermi level is set to zero. Note that this spectrum is a combination of the

valence region and the secondary electron cut-off region; therefore, it is purely for illustration

purpose and it does not represent there are actually energy states above the vacuum level.

Figure 2.8: Typical UPS spectrum of MoO3 with key electronic structure parameters labelled.

20

Study of Energy Levels at OHIs

The information in this chapter is adapted from a paper published as H.T. Kung et al., "Reaction

and Energy Levels at Oxide–Oxide Heterojunction Interfaces" Adv. Mater. Interfaces, 1901456

(2019).

3.1 Introduction

As discussed in Chapter 1 of the thesis, the electronic structure and the chemical composition of a

material can be altered at an OHI due to effects such as charge transfer, interfacial diffusion, and

chemical reactions. These events collectively contribute to the energy band structure of an OHI

which in turn affects its ability to perform an intended application. To study the properties of OHIs,

multiple different OHIs were fabricated and examined with XPS and UPS. The OHIs were

fabricated by growing MoO3 and ReO3 on different underlying oxides, which were aluminum

oxide (Al2O3) grown on aluminum, silicon dioxide (SiO2) grown on p-type silicon wafer,

germanium dioxide (GeO2) grown on p-type germanium wafer, indium tin oxide (ITO), and

rhenium (III) oxide (ReO3) prepared by thermal vapor deposition. In this thesis, a great focus is

placed on the results obtained from the OHIs involving MoO3, and the results from the OHIs

involving ReO3 as the over layer are used as supporting information to avoid providing repetitive

information.

3.2 Experimental

XPS and UPS were performed in a PHI 5500 Multi-Technique system. The X-ray source in XPS

is monochromatic Aluminum Kα with a radiation energy of 1486.7 eV, and the UV source in UPS

is non-monochromatic He Kα with a radiation energy of 21.22 eV. The take-off angles for XPS

and UPS measures are 75◦ and 88◦, respectively. XPS was performed in an ultra-high vacuum

environment with a base pressure of 10-9 Torr. During UPS measurements, a constant bias of -15

V was applied to the sample to measure the work function.

MoO3 and ReO3 deposition and XPS/UPS measurements were perform in-situ to avoid exposing

the samples to the atmosphere. Exposure of MoO3 to the atmosphere can decrease its work function

significantly and render the UPS results invalid [31]. MoO3 was deposited via thermal vapor

evaporation. 99.999% pure MoO3 powder was heated by a Knudsen cell in a 10 cc alumina crucible

and evaporated at 551 ◦C. The deposition rate was kept at approximately 0.3 to 0.4 Å/s with a base

21

pressure of 10-8 Torr. ReO3 was deposited also by thermal vapor evaporation by heating ReO3

powder in a Knudsen cell to approximately 445 ◦C. The deposition rate was around 0.6 Å/s with a

base pressure of 10-7 Torr. The thicknesses of the deposited MoO3 and ReO3 were monitored by

an oscillating quartz thickness monitor.

The Al2O3 thin film was prepared by first depositing 150 nm thick of aluminum film on a silicon

substrate ex-situ (this deposition was performed outside of the PHI 5500 system in a different

vacuum chamber), and the oxide layer was grown by placing the sample in a UV ozone (UVO)

chamber for 15 minutes in atmosphere. This procedure was implemented aiming to replicate the

work reported by Lee et al. [12]. The SiO2 and GeO2 thin films were prepared by placing Si and

Ge wafers in the UVO chamber for 15 minutes after cleaning with acetone and methanol. The ITO

film used in this work was pre-coated on a glass substrate by the supplier and it is commonly used

as the anode of organic optoelectronic devices. The ITO substrate was cleaned in acetone and

methanol and placed in the UVO chamber for 15 minutes for cleaning purpose. The ReO3 thin film

was prepared through in-situ thermal vapor evaporation in the PHI 5500 system. 10 nm of ReO3

was deposited on a piece of cleaned silicon wafer.

The oxide layer thickness for Al2O3, SiO2, and GeO2 were determined using the Strohmeier

Overlayer equation [32] as shown in Equation 9.

𝑑 = 𝜆 𝑠𝑖𝑛𝜃 ln ( + 1) (10)

Where d is the oxide thickness, λ is the inelastic mean free path of the emitted photo electrons in

the metal (m) or the oxide (o), N is the volumetric density, and I is the XPS peak intensity (total

area under the relevant XPS peak).

The parameters used to calculate the thickness of Al2O3 are: λo = 2.8, λm = 2.6, θ = 75°, and =

1.6 [33]. The calculated oxide thickness is approximately 3.4 nm. An alternative version of

Equation 1 was proposed by Lu et al. for SiO2 thickness calculation [34], and based on that

information, the SiO2 thickness was determined to be around 0.9 nm. The parameters needed to

calculate the GeO2 thickness was obtained through the NIST Electron Effective-Attenuation-

Length Database [35]. The following parameters were used: λo = 2.6, λm = 2.38, θ = 75°, and =

1.8. The oxide thickness was determined to be around 2.3 nm.

22

3.3 Results

3.3.1 Reaction at OHI

XPS and UPS were performed on MoO3 over layers (up to 8 nm) on different oxide films without

sample charging observed. Based on the XPS spectra, we determined that the Mo6+ 3d5/2 peak

located at a binding energy of approximately 232.5 eV and the Mo5+ 3d5/2 peak located at around

1 eV lower than the Mo6+ peak. The binding energies of the XPS peaks can vary by ±0.5 eV due

to factors such as the presence of defects and cation impurities which in turn can shift the position

of Fermi level in MoO3 [10]. Figure 3.1 is the fitted Mo 3d XPS spectrum of 1 nm of MoO3

deposited on Al2O3, SiO2, and ITO as examples. All fitted XPS spectra are included in the

Appendix A as Figure A. Note that as part of the curve fitting proves, the full width at half

maximum (FWHM) was fixed to be the same for Mo6+ and Mo5+ peaks, the area ratio between the

d3/2 and d5/2 peaks was set to be 2/3, and the difference in binding energies between Mo6+ and Mo5+

peaks was set to be 1 eV. The Mo6+ and Mo5+ percentages were determined based on curve fitting

at different MoO3 thicknesses. The Mo5+ percentages for the MoO3 deposited on Al2O3, SiO2, and

ITO are summarized and shown in Figure 3.2 as examples. The Mo6+ and Mo5+ percentages for

all OHIs in this study are summarized and shown in Figure B in Appendix B. Note that the

percentages determined via curve fitting have an accuracy range of approximately ±3% due to

human error involved in the fitting process. Note that the XPS curve fitting process was not

performed on the ReO3/oxide OHIs as the positions of ReO3 peaks for different oxidation states

were difficult to locate.

23

Figure 3.1: Mo 3d XPS spectra of increasing MoO3 thicknesses on, from left to right, Al2O3,

SiO2, and ITO. Curve fitting with Mo6+ and Mo5+ was performed and the fitted curves are shown.

Figure 3.2: Mo5+ percentages as a function of MoO3 thickness for systems formed by growing

MoO3 on Al2O3, SiO2, and ITO.

24

By comparing the MoO3 compositions at the OHIs (using the compositions at 1 nm MoO3), we

are able to observe that the degree of MoO3 reduction is more pronounced in the following

increasing order: SiO2 (14.6% Mo5+), ReO3 (15.5% Mo5+), ITO (31.7% Mo5+), GeO2 (32.4%

Mo5+), and Al2O3 (46.9% Mo5+). Note that the Mo6+ percentages on all underlying oxides fail to

reach 100% even at 8 nm thick because oxygen vacancies intrinsically form in MoO3 which results

in a small percentage of Mo5+ being formed [10].

3.3.2 Formation of Additional Electronic States at OHIs

At every MoO3 thickness, UPS was performed to obtain information on the electronic structure of

the OHIs. The valence band regions of the MoO3 grown on Al2O3, SiO2, and ITO are shown in

Figure 3.3, and the measured work functions (Ф) and valence band maximum (VBM) values are

summarized in Figure 3.4. Note that the values in Figure 3.4 have an error range of around ±0.1

eV. The UPS results for the rest of the OHIs in this study are shown in Figure C1 and Figure C2

in Appendix C.

Figure 3.3: The valence band region of UPS spectra of MoO3 grown on various oxide films with

the MoO3 thicknesses increase from top to bottom. MoO3 thicknesses, VBM, and defect bands are

25

shown. Note that the Fermi level was set at zero and the negative x axis represents the energy

levels below the Fermi level. An example of how VBM is determined from a UPS spectrum is

shown in the MoO3 on Al2O3 plot. Also, the defect bands are marked as d1 or d2.

From Figure 3.3, it can be observed that the signature defect band, d1, for Mo5+ is present at all the

studies OHIs with it being more prominent at thinner MoO3 thicknesses. This observation is

consistent with the XPS results summarized in Figure 3.1. Furthermore, as it is indicated in the

figure, a second defect band is observed at the MoO3-Al2O3 interface. This defect band is most

likely associated with the second defect band in the MoO2 electronic structure [10], which suggests

that MoO3 is the most reduced at this OHI. Mo4+ is not noticeable in the associated XPS spectrum

is possibly due to the weak signal of the small amount of Mo4+ present at the interface relative to

the amount of existing Mo6+ and Mo5+.

3.3.3 Energy Levels at OHIs

In Figure 3.4, we observe valence band bending. We notice that the VBMs closer to the OHIs are

lower (further from the Fermi level). As the MoO3 thin films become thicker, the VBMs approach

the bulk value (around 2.7 eV below the Fermi level). The measured work functions near all OHIs

are also lower and they approach the bulk value as the MoO3 thicknesses increase. There are

multiple factors that can affect the work function of MoO3 such as, the amount of defects that are

present [10, 20], and charge transfer at the OHIs [5]. Note that the theoretical work function of 6.8

– 6.9 eV is not observed even at MoO3 thickness of 8 nm because of intrinsic defects that are

present in the samples [10, 20]. Layer-by-layer growth of ReO3 on Al2O3, SiO2, and GeO2 was

performed to confirm the UPS results obtained from the OHIs involves MoO3. Similar VBM and

Ф trends were observed and the results are shown in Appendix D.

26

Figure 3.4: i) Measured work function and ii) valence band maximum values of the growth of

MoO3 on Al2O3, SiO2, and ITO films. Note that the Fermi level is used as the reference point and

is set at zero eV energy.

The bulk electronic structure parameters for the oxides used in this study were measured by UPS

and are listed in Table 3.1. The parameters reported in published literature are also listed. Note

that there are deviations between the measured values and the literature values due to the different

film growth environment and conditions which can affect impurity and defect contents, and the

chemical compositions of the oxides.

27

Table 3.1: Bulk electronic structure parameters for the underlying oxides in this paper. Note IE

stands for ionization energy.

Al2O3 SiO2 GeO2 ReO3 ITO

Measured Literature Measured Literature Measured Literature Measured Literature Measured Literature

Ф

(eV) 2.26

3.18 [12],

3.97 [36],

4.7 [37]

3.54

4 [38],

4.08-4.23

[39]

3.81 3.98-4.31

[40] 6.39

6.7 [41],

6.8 [42] 4.28

4.21 [43],

4.4-4.5 [44],

4.24-5.1 [45]

VBM

(eV) -5.36 -5.55 [12] -5.52

-5.4 [38],

-4.9 [46] -5.17 -4.7 [47] -3.23 -3.3 [41] -3.09

-3.11 [43],

-3.25 [48]

IE

(eV) 7.62 8.72 [12] 9.06 9.4 [38] 8.98 N/A 9.62 10 [41] 7.37 7.32 [43]

3.4 Discussion

In this section, we discuss the potential driving forces behind the formation of Mo5+ at the

examined OHIs. Also, we examine the effects of charge transfer have on the electronic structures

of the OHIs.

3.4.1 Reaction Thermodynamics at OHIs

It is well-known that the Gibbs free energy change of oxide formation can be used to aid predicting

whether oxidation and reduction reactions will occur at thin-film interfaces. Here we consider all

possible oxidation reactions that can take place at the OHIs. The reactions with their associated

Gibbs free energies are listed in Table 3.2.

28

Table 3.2: Gibbs free energies of all possible oxidation reactions that can take place at the OHIs

under study. The free energy values listed are presented as per mole of oxygen molecules.

Designation Reaction ΔGO2 (kJ/mol)

a. 2MoO2 + O2 2MoO3 -275.6 [10, 49]

b. 2Mo2O5 + O2 4MoO3 36 [10]

c. 4/3Al + O2 2/3Al2O3 -1054.9 [49, 50]

d. Si + O2 SiO2 -856.4 [49, 50]

e. Ge + O2 GeO2 -521.4 [49, 50]

f. 2ReO2 + O2 2ReO3 -235.2 [49]

g. 4ReO3 + O2 2Re2O7 -153.16 [49]

h. 4/3In + O2 2/3In2O3 -549.8 [10, 50]

i. Sn + O2 SnO2 -519.9 [49]

At the OHIs, there are three possible oxidation states for Mo: Mo6+, Mo5+, and Mo4+. Reactions

cannot take place spontaneously between MoO3 and the underlying oxides (with the exception of

ReO3 and ReO2) as both cation species are already in the highest oxidation state. However, for

oxides like Al2O3, SiO2, and GeO2, there is the possibility that the underlying non-oxide material

and the deposited MoO3 can diffuse through the oxide layer and make contact with each other. If

such contact occurs, MoO3 can be reduced. For the MoO3-ReO3 and MoO3-ITO interfaces, there

can be residual non-fully oxidized elements present that are able to reduce MoO3. The most

thermodynamically favorable Mo5+ formation reactions that can take place at each OHI is listed in

Table 3.3.

29

Table 3.3: The most thermodynamically favorable Mo5+ formation reactions that can take place

at the OHIs under study. The diffusion coefficient of Al in Al2O3 and Si in SiO2 at room

temperature are also included.

OHI Reaction ΔGreaction (kJ) Diffusion Coefficient (m2/s)

MoO3-Al2O3 4MoO3 + 4/3Al 2Mo2O5 + 2/3Al2O3 -1090.9 Al in Al2O3: 1.55 x 10-5 (at

1540 ◦C) [51]

MoO3-SiO2 4MoO3 + Si 2Mo2O5 + SiO2 -892.4 Si in SiO2: 1.17 x 10-37 (at

1150 ◦C) [52]

MoO3-GeO2 4MoO3 + Ge 2Mo2O5 + GeO2 -557.4 Not Available

MoO3- ReO3 4MoO3 + 2ReO2 2Mo2O5 + 2ReO3 -135.8 N/A

MoO3- ITO 4MoO3 + 4/3In 2Mo2O5 + 2/3In2O3 -585.8 N/A

To observe the relationship between percentage of Mo5+ and the Gibbs free energies of the possible

oxidation reactions that can occur at the OHIs (using the percentages at 1 nm MoO3), Figure 3.5

was plotted based on the values stated in Table 3.3.

Figure 3.5: Mo5+ % at the OHIs plotted against the Gibbs free energies of the most favorable

oxidation reactions.

30

From Figure 6, it can be seen that the extent of Mo5+ formation is more prominent with more

spontaneous redox reactions, with the exception of the MoO3-SiO2 interface. This observation

suggests that the reduction of MoO3 is driven significantly by the spontaneity of the redox reactions

that can take place at the OHIs. To investigate the MoO3-SiO2 interface “anomaly”, we check the

diffusion coefficients of the three following systems: Al in Al2O3, Si in SiO2, and Ge in GeO2.

Using the published empirical equations derived from experimental data, the diffusion coefficient

for Al in Al2O3 [51] is hundreds orders of magnitude higher than the one for Si in SiO2 [52] at

elevated temperatures as shown in Table 3.3. Also, it was reported that significant out-diffusion of

Ge through GeO2 was observed at 200 ◦C [53] so it is likely that out-diffusion of Ge occurs at the

room temperature as well. Due to the low diffusion coefficient of Si in SiO2, it is unlikely that Si

can make contact and reduce MoO3. There is the possibility that MoO3 can diffuse through the

underlying SiO2 layer, but it has been indicated by multiple studies that the oxide layer is dense

[54, 55] and hence, the diffusion of MoO3 through the oxide layer is unlikely. Therefore, we

classified the MoO3-SiO2 interface as non-reactive and all other OHIs as reactive.

For the reactive OHIs, it can be seen in Figure 4 that only a small amount of Mo4+ is formed at the

MoO3-Al2O3 interface (d2 in the UPS spectrum). According to Table 2, the formation of Mo4+ is

also thermodynamically favorable at the MoO3-GeO2 and MoO3-ITO interfaces; however, no

Mo4+ was detected at these interfaces. This observation suggests that at these OHIs, the driving

force from the oxidation of the out-diffused germanium or residual indium is not enough to

completely reduce Mo6+ to Mo4+. This reduction might be more favorable at higher temperatures

where the rate of solid-state diffusion becomes higher, which allows more Mo6+ to be contact with

the reducing elements. A similar phenomenon was observed by Greiner et al. where MoO3 does

not get reduced to MoO2 when it makes contact with metallic nickel even though such reaction is

thermodynamically favorable [10].

Here we take a closer look at the MoO3-ReO3 interface. Similar to MoO3, ReO3 is a transition

metal oxide that can have more than one oxidation states. It has been reported that at temperature

higher than 400 ◦C, ReO3 can break down into Re2O7 and ReO2 [56, 57]. Also even though

inconclusive, the existence of Re5+ is possible [58].As a result, the presence of Re4+ or Re5+ can

reduce MoO3 and form Mo5+.

31

For the MoO3-ITO interface, we examined the possibility of the presence of non-oxidized indium

or tin in the ITO substrate. XPS was performed on the ITO substrate and the indium and tin peaks

were asymmetric exhibiting the characteristics of oxygen-deficient ITO studied by Chen et al.

[59]. XPS was also performed after 0.5 nm of MoO3 was deposited on the ITO substrate and it was

noted that the indium peak became more symmetrical and it shifted slightly to a higher binding

energy signifying the oxidation of indium (the binding energy of metallic indium is around 0.2 eV

lower than indium oxide). Note that the signal of tin was too weak to identify any changes. The

XPS curves of indium is shown in Figure 3.6.

Figure 3.6: Indium XPS 3d5/2 peaks of the ITO substrate and 0.5 nm thick MoO3 deposited on

the same ITO substrate.

3.4.2 Other Potential Factors that can Cause Mo5+ Formation

To explain the formation of Mo5+ at the MoO3-SiO2 interface, we consider driving forces that are

not dictated by thermodynamics.

32

It has been reported by multiple studies that interfacial strain is a potential cause of oxygen

deficiency. It was found that tensile strain caused by lattice mismatch causes oxide film at the

interface to undergo reconstruction by rejecting oxygen atoms [4, 8, 9]. Therefore, it is possible

that tensile strain induced by lattice mismatch exists at the OHI and hence, causes Mo6+ to be

reduced to Mo5+. However, the degree of lattice mismatch is difficult to quantify without

performing modelling.

Another parameter that may have an effect on the degree of Mo5+ formation is the oxygen diffusion

coefficients in the underlying oxides. If oxygen can diffuse through the oxides with ease, it

increases the probability for oxygen to diffuse from MoO3 to the underlying oxides, which causes

the MoO3 to become oxygen deficient. However, the empirical parameters published for the

calculation of oxygen diffusion coefficients are either unavailable or vary significantly depending

on the experimental setups and conditions [51, 53]. As a result, the diffusion coefficients are

difficult to quantify in this paper.

The final factor we consider is charge transfer at the OHIs. The charge transfer can occur at an

interface when two different materials come in to contact with each other [2, 3, 5]. If charge

transfer occurs, the vacuum levels will shift and an interface dipole will form at the interface. For

the OHIs in this paper, we observe a negative interface diploe at every OHI; therefore, electrons

are transferred from the underlying oxides to MoO3 which can reduce Mo6+ to Mo5+. The energy-

level alignment (ELA) and the formation of interface dipoles (Δ) at the OHIs are discussed in detail

in the following section.

3.4.3 Interface Dipoles and Band Structures at OHIs

The model commonly used in the research community and industries for ELA is the classical

Schottky-Mott model which assumes the vacuum levels of the two dissimilar materials align after

contact. However, it was reported by Greiner et al. [5] that when the work function (Ф) of the

underlying layer is greater than the ionization energy (IE) of the over layer, the highest occupied

molecular orbital (HOMO) of over layer organic molecule is pinned to approximately 0.3 eV

below the Fermi level and this is the HOMO pinning region. Furthermore, it was also discovered

that when the work function of the underlying layer is less than the electron affinity (EA) of the

over layer, the HOMO offset binding energy is approximately the band gap of the over layer. This

is also known as the lowest unoccupied molecular orbital (LUMO) pinning region [60]. In the

33

HOMO and LUMO pinning regions, the Fermi levels are pinned due to charge transfer. Vacuum

level alignment is only dominant when the work function of the underlying layer is between the

IE and EA of the over layer [60]. For inorganic semiconductor, the valence band maximum (VBM)

and conduction band minimum (CBM) are equivalent to the HOMO and LUMO of organic

semiconductor respectively. When considering energy level alignment at oxide-oxide interfaces,

VBM and CBM should be considered instead.

The IE and EA of the MoO3 used in this paper are approximately 9.45 eV and 6.45 eV,

respectively. If the universal energy level alignment rule is applicable to the oxide-oxide interface,

all the OHIs in this paper will fall within the conduction band minimum (CBM) pinning region

since all work functions of underlying oxide layer is lower than the EA of the MoO3, and negative

interface dipoles should form.

To visualize the energy level alignment at the OHIs, we plotted the energy level schematics as

shown in Figure 3.7. The energy level parameters used at the interfaces were obtained from

depositing 1 nm of MoO3 on all underlying oxides.

34

Figure 3.7: Energy level alignment schematics at all OHIs in this study. Ф is the work function,

Δ is the interface dipole, and ΔEVBM is the VBM position under the Fermi level.

From the energy level alignment schematics, we have made several important observations. As

previously discussed, a negative interface dipole is formed at every OHIs which means that

electrons are transferred from the underlying oxides to MoO3. This charge transfer results in the

formation at the interface of an n-doped region in MoO3 and a p-doped region in the underlying

oxides. The observed band bending indicates the formation of these regions. For MoO3 near an

interface, we observed that the VBM bends downward which indicates that the region is more

electron rich as the Fermi level is closer to the CBM. Whereas for the underlying oxides, with the

exception of ITO, their VBM bend upward which indicates that the region is electron depleted as

the Fermi level is further away from the CBM. Note that the bending of the VBM of the underlying

oxide was observed via measuring the core-level shifts of the oxide peaks with XPS. It was

observed that the oxide peaks shift to lower binding energies which signifies the VBMs move

closer to their respective Fermi levels. This methodology has been proven to be reliable by Li et

al. [61]. The plots that depict the core level shifts of the under layer oxides can be found in

Appendix E. A possible reason why no band bending was observed on ITO is due to the fact that

ITO is a highly degenerated semiconductor and is conductive, and therefore, the electron

“depleted” region is easily filled by the electrons in the bulk material.

3.3.1 Variation of Work Functions

We also plotted the measured interface dipoles and work functions at the interfaces as a function

of the work functions of the underlying oxides. The plots are shown in Figure 3.8.

35

Figure 3.8: a) Work functions measured at the OHIs plotted against their associated work

functions of the underlying oxides. b) Interface dipoles plotted as a function of their associated

work functions of the underlying oxides. The MoO3 thickness is 1 nm for all systems. The

empirical equation of the fitted linear line is shown in the plot.

According to the studies on the energy levels at the organic-organic interfaces performed by Li et

al., within the CBM pinning region (lowest unoccupied molecular orbital, LUMO, for organic

semiconductors), the slope of the fitted linear line was found to be 1.[36] However, the slope of the

fitted linear line in Figure 8 b) is approximately 0.59. The interface dipole trends between the two

different types of interfaces deviate from each other is likely due to the OHIs being highly

interactive. For organic-organic interfaces, the organic molecules do not strongly interact with

each other and they do not have multiple chemical states like TMOs. Therefore at these interfaces,

charge transfer (and not reaction) is the main driving force behind the shift of energy levels.

Whereas for the OHIs in this study, the formation of Mo5+ due to interfacial reaction, charge

transfer, and other previously discussed driving forces is also a critical factor that affects the

electronic structures at the interfaces. Furthermore, interfacial diffusion at the OHIs likely render

the compositions at the interfaces convoluted.

3.5 Summary

We systematically examined multiple OHIs using photoelectron spectroscopy. OHIs formed by

MoO3 and different oxides were closely examined by XPS and UPS. We discovered some degree

of MoO3 reduction occurs at all of the OHIs in this study and the degree of reduction varies for

36

each underlying oxide. The presence of Mo5+ (and Mo4+ for the MoO3-Al2O3 interface) at the OHIs

change the valence band structure by introducing a new defect band under the Fermi level and

deviating the work functions and VBMs from the bulk values. We classified the OHIs into two

categories: reactive and non-reactive. For the reactive OHIs, we conclude that the primary driving

force behind the reduction of MoO3 is the Gibbs free energies of oxidation of the underlying layers.

For the non-reactive OHI, we conclude that the formation of Mo5+ is a result of charge transfer at

the interfaces and other possible factors such as oxygen deficiency due to interfacial strain.

We also closely examined the valence band structures at the OHIs to observe the effects of charge

transfer. We note that the OHIs deviate from the existing energy level alignment rule. We conclude

that this deviation is due to the reactivity of the interfaces and other factors such as interfacial

diffusion.

37

Highly Complex OHIs

4.1 Introduction

As it is discussed in Chapter 3, OHIs are extremely complex as phenomena such as reaction,

interfacial diffusion, and charge transfer can occur. As a result, the energy levels at these interfaces

deviate from the existing energy level alignment rules and their positions are difficult to predict

using existing models.

The AgOx/MoO3 OHI has been implemented in top-emitting OLEDs [62, 63, 64]. Researchers

found using silver as the anode in a top emitting OLED attractive as it is highly reflective. It was

reported by the researchers that by growing a thin layer of silver oxide using UV ozone on the

silver anode (between silver and MoO3), the performance of the device was greatly improved due

to the decrease in hole injection barrier [62]. However, this OHI is extremely complex as the nature

of the silver oxide film is unpredictable and it can be easily affected based on the oxide growth

condition. In this chapter, the TMO/AgOx OHIs are explored to further illustrate the complexity

of OHIs.

4.2 Experimental

The AgOx under layer samples were prepared by depositing approximately 100 nm thick of silver

on silicon substrates via VTE, and then the samples were oxidized in an UV ozone chamber for

different periods of time, namely 2 minutes, 5 minutes, and 10 minutes. MoO3/AgOx and

ReO3/AgOx OHIs were prepared and examined with PES by growing MoO3 and ReO3 on the AgOx

samples via the layer-by-layer deposition technique in the PHI 5500 Multi-technique System. The

experimental setup and procedure are similar to those detailed in Chapter 3. Refer to Section 3.2

for the details of the experimental setup.

4.3 Results & Discussion

XPS was performed on the samples with different MoO3 and ReO3 thicknesses deposited on AgOx

(2 minute UV ozone exposure time) to observe any potential changes of the compositions of the

samples at different over layer thicknesses. The XPS spectra of normalized Mo 3d peaks are shown

in Figure 4.1 and the UPS spectra of MoO3 with different thicknesses grown on AgOx are shown

in Figure 4.2.

38

Figure 4.1: Mo 3d XPS core level peaks with 1 nm, 3 nm, and 8 nm of MoO3 deposited on AgOx.

Figure 4.2: UPS spectra of MoO3 with different thicknesses on AgOx.

39

Unexpected observations are made from the XPS spectra. From Figure 4.1, it can been seen that

the Mo 3d peaks are narrow for samples with thinner MoO3 films. This observation implies that at

the MoO3/AgOx OHI, there is predominantly Mo6+ and less Mo5+. Note that the slight left shift in

XPS peaks with increasing TMO thickness is due to the increasing in Mo5+ content. As Mo5+ is

formed, the effect of electron shielding strengthened; as a result, the Fermi level shifts up and the

binding energies of core Mo electrons increase. The XPS results are also supported by the UPS

spectra shown in Figure 4.2. It can be seen that the defect band associated to Mo5+ is absent closer

to the OHI and it appears further from the OHI. As it is mentioned in Section 1.4, there is always

some degree of oxygen-deficiency in the MoO3 samples used in this work; therefore, some Mo5+

should be present in bulk MoO3. In order for the OHI to be mostly pure of Mo6+, the traces of Mo5+

in MoO3 must be oxidized. To observe the oxidation of the over layer oxide, the ReO3/AgOx OHI

was also examined with XPS.

The typical Re 4f XPS spectrum is shown in Figure 4.3. Note that the two visible peaks are the

Re 4f5/2 and 4f7/2 peaks which correspond to Re6+ and possibly contain traces of Re4+ and Re5+ [56-

58]. However, at the ReO3/AgOx OHI, it is apparent that the ReO3 is oxidized as one pair of

additional peaks with higher binding energies are formed. The XPS spectra of Re 4f peaks at the

OHI are shown in Figure 4.4.

40

Figure 4.3: Typical Re 4f XPS spectrum for ReO3.

Figure 4.4: Re 4f XPS core level peaks with 0.5 nm, 1 nm, 3 nm, and 8 nm of ReO3 deposited on

AgOx.

In Figure 4.4, it is shown that the two additional peaks are most prominent at lower thicknesses

(closer to the OHI) which suggests that the oxidation process is the most spontaneous at the

interface.

As it is mentioned in Chapter 3, two of the possible driving forces for oxidation and reduction of

the over layer oxides at OHIs are interfacial reaction and non-reactive charge transfer. For non-

reactive charge transfer, the direction of charge transfer at the OHI can be predicted by the

available energy-level alignment rules. Based on UPS measurements, the VBM and Ф of the AgOx

under layer are around -1.42 eV and 4.86 eV respectively (IE = 6.28 eV). The IEs of MoO3 and

ReO3 are approximately 9.45 eV and 9.6 eV respectively based on UPS measurements. As it is

stated in Section 3.4.3, in order for electrons to be transferred from the over layer to the under

layer, the Ф of the under layer must be greater than the IE of the over layer. Based on the

experimental data, the IEs of MoO3 and ReO3 are greater than the Ф of AgOx, as a result, such

charge transfer is not probable at the MoO3/AgOx and ReO3/AgOx OHIs.

41

If MoO3 and ReO3 are oxidized through interfacial reactions, possible reactions need to be

considered. Based on previously published papers, silver with different oxidation states, namely

Ag+ and Ag2+, can form when silver is oxidized in an UV ozone chamber [62, 63]. Due to the XPS

silver core level peaks for different oxidation states are highly convoluted, the oxygen 1s XPS

peaks were examined to verify the presence of different types of silver oxides. The oxygen XPS

spectrum of AgOx with different UV ozone exposure times are shown in Figure 4.5.

Figure 4.5: Oxygen 1s XPS peaks of AgOx samples with different UV ozone exposure times.

Based on Figure 4.4, it is clear that with increasing UV ozone exposure time, the Ag2+ content

increases as the peak at lower binding energy correspond to Ag2O2 and the higher binding energy

peak correspond to Ag2O [63]. The right shift of the peaks with increasing Ag2O2 content can also

be explained by the weakening of electron shielding. With the presence of different types of silver

oxides present, the possible reactions are listed in Table 4.1.

42

Table 4.1: Possible reactions at the MoO3/AgOx and ReO3/AgOx OHIs.

Designation Reaction ΔGO2 (kJ/mol)

a. 2Mo2O5 + 2Ag2O2 4MoO3 + 2Ag2O 3.2 [10, 50]

b. 4ReO3 + 2Ag2O2 2Re2O7 + 2Ag2O -185.96 [50]

For the MoO3/AgOx OHI, at room temperature the oxidation of Mo2O5 by Ag2O2 is not

spontaneous as shown by reaction a. in Table 4.1. However since the Gibbs free energy of the

reaction is low, it is probable that the reaction can occur at an elevated temperature such as during

the MoO3 deposition process. For the ReO3/AgOx OHI, the oxidation of ReO3 by Ag2O2 is

spontaneous as shown by reaction b. in Table 4.1.

To observe the reactions listed in Table 4.1, MoO3 and ReO3 were deposited on AgOx substrates

(10 minute UV ozone exposure time) and the changes to the oxygen 1s peaks were investigated

with XPS. Figure 4.6 depicts the changes in the oxygen peaks with different thicknesses of ReO3

deposited on AgOx.

Figure 4.6: XPS oxygen 1s peaks of different thicknesses of ReO3 deposited on AgOx.

43

The binding energy of the ReO3 oxygen peak was found to be at around 530.5 eV which convolutes

with the Ag2O oxygen peak. However, from Figure 4.5, it is apparent that after even 0.5 nm of

ReO3 is deposited on the AgOx substrate, the right oxygen peak, which corresponds to Ag2O2,

diminishes. This observation is in agreement with the hypothesis of ReO3 being oxidized by

Ag2O2.

Figure 4.7 depicts the changes in the oxygen peaks with different thicknesses of MoO3 deposited

on AgOx.

Figure 4.7: XPS oxygen 1s peaks of different thicknesses of MoO3 deposited on AgOx.

The changes of the oxygen 1s peaks shown in Figure 4.7 are unexpected, as it is shown that both

Ag2O and Ag2O2 oxygen peaks diminish when a thin layer of MoO3 is deposited on the AgOx

substrate (note that the MoO3 oxygen 1s peak is located at around 529.5 eV). The remaining

oxygen peak becomes narrower and more symmetric (especially with the right shoulder

diminishing) as the MoO3 thickness increases which potentially suggests that the amount of Ag2O2

is decreasing; however, the evidence is not concrete solely based on the XPS results. There may

be other processes taking place at the MoO3/AgOx that have yet to be discovered.

44

4.4 Conclusion

In this chapter of the thesis, highly complex MoO3/AgOx and ReO3/AgOx OHIs are explored using

photoelectron spectroscopy. It was discovered at these OHIs, the over layers (MoO3 and ReO3) are

oxidized despite the energy level alignment rules suggesting that electrons should not transfer from

the over layer to the under layer. It was observed by XPS that the AgOx grown by exposing silver

in UV ozone can produce two types of silver oxide, Ag2O2 and Ag2O, and the amount of Ag2O2

increases with increasing UV ozone exposure time. Based on thermodynamics, it is possible that

Ag2O2 can oxidize MoO3 and ReO3 and convert to its more stable form, Ag2O. This process is

observed on the ReO3/AgOx OHI by examining changes of the oxygen 1s peaks as a function of

ReO3 thickness. However, no concrete evidence is available for the MoO3/AgOx OHI that suggests

what the primary driving force is for the reduction of MoO3. This chapter demonstrates that due to

the high complexity of OHIs, the fabrication process needs to be well designed, since the

fabrication process can have unintended impact the composition of the interfaces which in turn

can affect the performance of the final device.

45

Conclusion and Future Work

5.1 Conclusion

In this thesis, numerous OHIs were systematically examined by in-situ photoelectron spectroscopy

with the objective to discover any general trends or rules at these interfaces. Multiple OHIs were

formed by growing TMOs on different oxides. First part of the thesis explores the interfaces

formed by growing MoO3 on Al2O3, SiO2, GeO2, ITO, and ReO3. Reduction of MoO3 was

observed at all examined interfaces and they were classified into two categories: reactive interfaces

and non-reactive interfaces. For the reactive OHIs, the primary driving force behind the reduction

of MoO3 is redox reaction at the interfaces. For the non-reactive interface, a combination of

different factors such as charge transfer (energy-level alignment), and oxygen deficiency result in

the reduction of MoO3. Evidence of energy-level alignment, such as bend banding and interface

dipole, was also observed at these OHIs. It was discovered that the degrees of interface dipole and

the work function have a linear relationship with the under layer oxide’s work function. However,

due to the high complexity of these interfaces (contributed by phenomena such as interfacial

diffusion, reaction, and the formation of TMOs with different oxidation states), OHIs deviate from

the existing energy-level alignment rules.

In the second part of the thesis, MoO3/AgOx and ReO3/AgOx OHIs were explored to demonstrate

the high complexity of OHIs. In contrast to the aforementioned OHIs, evidence of oxidation of the

over layer TMOs was observed with XPS, even though electrons should be transferring from the

under layer to the over layer based on the energy-level alignment rules. It was confirmed that two

types of silver oxides are formed during the UV ozone oxide growth process: Ag2O2 and Ag2O.

The content of Ag2O2 increases with increasing UV ozone exposure time. Ag2O is the more

thermodynamically stable form of silver oxide, hence, Ag2O2 can oxidize the TMOs and be

reduced to Ag2O. This part of the thesis demonstrate the importance of OHI fabrication process

design as the process can bring unintended impact to the composition of the OHI.

As OHIs are become increasingly more common in today’s applications such as thin-film devices

and chemical catalysis, it is crucial to have a good understanding of the properties of these

interfaces. The findings in this thesis have implications to the design and fabrication of electronic

thin-film devices as the interaction that take place at the OHIs need to be taken into consideration

for device manufacturing and optimization.

46

5.2 Future Work

The following areas potentially require further studies:

The only non-reactive OHI examined was the MoO3/SiO2 interface. As a result, the primary

driving force behind the reduction of the over layer TMO was not determined. In order to

obtain such information, more non-reactive OHIs need to be fabricated. However, oxides

have high sublimation temperatures so a high temperature heating source is needed in the

PHI5500 system in order to perform in-situ photoelectron spectroscopy.

A complete energy alignment rule derived using OHIs is currently unavailable. Such rule

was not constructed in this study due to the low variety of available over layer oxides to

the research group. As it is mentioned in the previous point, oxides have high sublimation

temperatures and the only oxides that could be evaporated by the PHI5500 system were

MoO3 and ReO3. As a result, to fabricate OHIs that cover all three regions of the energy-

level alignment rule [5, 60] (CBM pinning, vacuum level aligned, and VBM pinning

regions), a high temperature source is needed.

One of the possible driving forces behind the reduction of the over layer TMO is the

rejection of oxygen atom due to lattice mismatch at the OHI. To further study the

relationship between the extend of formation of Mo5+ and the degree of lattice mismatch,

modelling or computer simulations may be needed to accurately calculate the degree of

lattice mismatch at an OHI.

The MoO3/AgOx OHI is not well understood as the factors that cause the reduction of

MoO3 are not clear. Therefore, techniques such as high resolution transmission electron

microscopy and scanning tunneling microcopy may be utilized to obtain more information.

The unique properties of OHIs can be proven beneficial in a device. As it was

aforementioned in Chapter 1, due to TMOs ability to perform orbital reconstruction, OHIs

can exhibit superconducting or conductive characteristics. As a result, there is the

possibility to incorporate one of the OHIs studied in this thesis in a thin-film transistor

(TFT). As it is shown in Chapter 3, charge transfer occurs at the OHIs (band bending and

interface dipoles); therefore, there is an accumulation of charges at an OHI. This

47

accumulation of charges makes the interface highly conductive which is a desirable

property for the channel in a TFT [11].

48

48

Appendix A: Mo 3d XPS spectra of increasing MoO3 thicknesses on GeO2, and ReO3

Figure A: Mo 3d XPS spectra of increasing MoO3 thicknesses on, from left to right, GeO2, and

ReO3. Curve fitting with Mo 6+ and Mo 5+ was performed and the fitted curves are shown.

49

Appendix B: Mo5+ and Mo6+ Percentages

Figure B: Mo5+ and Mo6+ percentages of MoO3 films deposited on various oxide films as a

function of MoO3 thickness.

50

Appendix C: UPS Results

Figure C1: The valence band region of UPS spectra of MoO3 grown on GeO2 and ReO3 with the

MoO3 thicknesses increase from top to bottom. The defect bands are denoted d1.

Figure C2: Measured work function and valence band maximum values of the growth of MoO3

on GeO2, and ReO3 films.

51

Appendix D: VBM and Ф Progression with Increasing ReO3 Thickness

Figure D1: Valence band maximum values of the growth of ReO3 on Al2O3, SiO2, and GeO2 films.

Note that the Fermi level is used as the reference point and is set at zero eV energy. Note that the

measure VBM values approach the bulk ReO3 VBM value of -3.2 eV as the thicknesses of ReO3

increase.

52

Figure D2: Work function values of the growth of ReO3 on Al2O3, SiO2, and GeO2 films. Note

that the Fermi level is used as the reference point and is set at zero eV energy. Note that the measure

Ф values approach the bulk ReO3 Ф value of 6.39 eV as the thicknesses of ReO3 increase.

53

Appendix E: Under Layer Oxides Core Level Shifts

Figure D1: XPS Al2O3 core level peaks without MoO3 deposited (left) and with 1 nm MoO3

deposited (right)

Figure D2: XPS SiO2 core level peaks without MoO3 deposited (left) and with 1 nm MoO3

deposited (right)

54

Figure D3: XPS GeO2 core level peaks without MoO3 deposited (left) and with 1 nm MoO3

deposited (right)

Figure D4: XPS ReO3 core level peaks without MoO3 deposited (left) and with 1 nm MoO3

deposited (right)

55

55

References

[1] J. Mannhart, D. G. Schlom, Science 2010, 327, 1607.

[2] Z. Zhong, P. Hansmann, Phys. Rev. X 2017, 7, 011023.

[3] P. Zubko, S. Gariglio, M. Gabay, P. Ghosez, and J. M. Triscone, Annu. Rev. Condens. Matter

Phys. 2011, 2, 141.

[4] H. Y. Hwang, Y. Iwasa, M. Kawasaki, B. Keimer, N. Nagaosa, and Y. Tokura, Nat. Mater.

2012, 10, 1038.

[5] M. T. Greiner, M. G. Helander, W. Tang, Z. Wang, J. Qiu, Z. Lu, Nat. Mater. 2012, 11(1),

76.

[6] C. Canellieri, V. Strocov, Spectroscopy of Complex Oxide Interfaces, Springer Berlin, New

York, NY 2018.

[7] A. Ohtomo, D. A. Muller, J. L. Grazul, H. Y. Hwang, Nature 2002, 419, 378.

[8] K. Kawashima, G. Logvenov, G. Christiani and H. U. Habermeier, J. Magn. Magn. Mater.

2015, 378, 539.

[9] S. Heo, C. Oh, J. Son & H. M. Jang, Sci. Rep. 2017, 7, 4681.

[10] M. Greiner, L. Chai, M. Helander, W. Tang, and Z. Lu, Adv. Funct. Mater. 2013, 23, 215.

[11] H. Faber, S. Das , Y. Lin, N. Pliatsikas, K. Zhao, T. Kehagias, T. D. Anthopoulos, Sci. Adv.

2017, 3, 1602640.

[12] J. Lee, P. Li, H. T. Kung, Z. H. Lu, J. Appl. Phys. 2019, 125, 145501.

[13] J. C. Vedrine, Catalysts 2017, 7, 341.

[14] A. Khojastehnezhad, F. Moeinpour, M. Vafaei, J. Mex. Chem. Soc. 2015, 59, 29.

[15] M. A. Banares, J. Fierro, J. B. Moffat, J. Catal. 1993, 142, 406.

[16] A. Parmaliana, F. Arena, J. Catal. 1997, 167, 57.

[17] J. Papp, S. Soled, K. Dwight, A. Wold, Chem. Mater. 1994, 6, 496.

[18] S. Bai, H. Liu, J. Sun, Y. Tian, S. Chen, J. Song, R. Luo, D. Li, A. Chen, C. C. Liu, Appl.

Surf. Sci. 2015, 338, 61.

[19] D. O. Scanlon, G. W. Watson, D. J. Payne, G. R. Egdell, D. S. L. Law, J. Phys. Chem. C.

2010, 114, 10, 4636-4645.

56

[20] M. T. Greiner, L. Chai, M. G. Helander, W. M. Tang, Z. H. Lu, Adv. Funct. Mater. 2012,

22, 4557.

[21] K. Inzani, M. Nematollahi, F. Vullum-Bruer, T. Grande, T. W. Reenaas, S. M. Selbach,

Phys. Chem. Chem. Phys. 2017, 19, 9232.

[22] S. Bandaru, G. Saranya, N. J. English, C. Yam, M. Chen, Sci. Rep. 2018, 8, 10144.

[23] K. Inzani, T. Grande, F. Vullum-Bruer, S. M. Selbach, J. Phys. Chem. C 2016, 120, 8959.

[24] A. Kahn. Mater. Horiz. 2016, 3, 7.

[25] D. M. Mattox, Handbook of Physical Vapor Deposition (PVD) Processing: Film Formation,

Adhesion, Surface Preparation and Contamination Control. Noyes Publications, 1998.

[26] H. L. The, R. M. Tiggelaar, E. Berenschot, A. V. D. Berg, N. Tas, J. C. T. Eijkel, ACS

Nano. 2019, 13(6), 6782-6789.

[27] M T Greiner, M G Helander, Z B Wang, Z H Lu, Rev. Sci. Instrum. 2009, 80(12):125101.

[28] M. Ohring, Materials Science of Thin Films, AP, ISBN 0-12-524975-6, 2002.

[29] S. Hufner, Photoelectron Spectroscopy Principles and Applications, ISBN 978-3-662-

09280-4, 2003.

[30] D. Briggs, XPS: Basic Principles, Spectral Features and Qualitative Analysis, in: D. Briggs,

J.T. Grant (Eds.), Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, IM

Publications, Chichester, 2003, pp. 31-56.

[31] M. T. Greiner, M. G. Helander, Z. Wang, W. M. Tang, Z. H. Lu, J. Phys. Chem. C 2010,

114, 19777.

[32] B. R. Strohmeier, Surf. Interface Anal. 1990, 15, 51.

[33] M. R. Alexander, G. E. Thompson, X. Zhou, G. Beamson, and N. Fairley, Surf. Interface

Anal. 2002, 34, 485.

[34] Z. H. Lu, J. P. McCaffrey, B. Brar, G. D. Wilk, R. M. Wallace, L. C. Feldman, and S. P.

Tay, Appl. Phys. Lett. 1997, 71, 2764.

[35] C. J. Powell, NIST Electron Effective-Attenuation-Length Database (National Institute of

Standards and Technology, Gaithersburg, MD, 2001), p.Version 1.0.

[36] Y. I. Semov, Phy. Stat. Sol.1969, 32

57

[37] A. Dymshits, A. Henning, G. Segev, Y. Rosenwaks, and L. Etgar, Scientific Reports 2015,

5, 8704

[38] M. Krzywiecki, L Grzadziel, Appl. Surf. Sci. 2014, 311, 740.

[39] J. Yang, D. Yan, and T. Jones, Chem. Rev. 2015, 115, 11, 5570-5603

[40] T. I. Lee, Y. Seo, J. Moon, H. J. Ahn, H. Y. Yu, W. S. Hwang, and B. J. Cho, Solid-State

Electron. 2017, 130, 57.

[41] L. G. Gerling, S. Mahato, C. Voz, R. Alcubilla, Appl. Sci. 2015, 5, 695.

[42] Suchitra, J. Pan, and U. V. Waghmare, J. Appl. Phys. 2014, 116, 034304

[43] T. Liu, X. Zhang, J. Zhang, W. Wang, L. Feng, L. Wu, W. Li, G. Zeng, and B. Li, Int. J.

Photoenergy 2013, http://dx.doi.org/10.1155/2013/765938

[44] Y. Park, V. Choong, and Y. Gao, Appl. Phys. Lett. 1996, 68, 2699

[45] M. Fang, C. Zhang, and Q. Chen, Applied Surface Science 2016, 385, 28–33

[46] J. K. Yang, W. S. Kim, H. H. Park, Thin Solid Films 2006, 494, 311

[47] S. K. Sahari, N. A. Abdul Halim, M. Kashif, S. Marini, R. Sapawi, K. Kipli, and N. Junaidi,

Journal of Telecommunication, Electronic and Computer Engineering 2018, 10, 61.

[48] F. Chen, R. Schafranek, S. Li, W. B. Wu, and A. Klein, Journal of Physics D: Applied

Physics 2010, 43, 29

[49] R. A. Robie, and B. S. Hemingway, Geol. Soc. Am. Bull. 1995,

https://doi.org/10.3133/b2131.

[50] D. R. Lide, J. Am. Chem. Soc. 2005, https://doi.org/10.1021/ja041017a.

[51] M. Le Gall, B. Lesage & J. Bernardini, Philos. Mag. A 1994, 70, 761.

[52] M. Uematsu, H. Kageshima, and Y. Takahashi, Appl. Phys. Lett. 2004, 84, 876.

58

[53] S. Ogawa, R. Asahara, Y. Minoura, H. Sako, N. Kawasaki, I. Yamada, T. Miyamoto, T.

Hosoi, T. Shimura, and H. Watanabe, J. Appl. Phys. 2015, 118, 235704.

[54] M. Morita, T. Ohmi, E Hasegawa, M. Kawakami, and M. Ohwada, J. Appl. Phys. 1990, 68,

1272.

[55] M. Morita, T. Ohmi, E. Hasegawa, M. Kawakami, and K. Sums, Appl. Phys. Lett. 1989, 55,

562.

[56] G. Rouschia, Chem. Rev. 1974, 74, 531.

[57] P. A. Shcheglov and D. V. Drobot, Russ. J. Phys. Chem. 2006, 80, 1819.

[58] M. T. Greiner, T. C. R. Rocha, B. Johnson, A. Klyushin, A, K. Gericke, and R. Schlögl, Z.

Phys. Chem. 2014, 228, 521.

[59] A. Chen, K. Zhu, H. Zhong, Q. Shao, and G. Ged, Sol. Energy Mater. Sol. Cells 2014,

120, 157.

[60] Y. Li, P. Li, Z. H. Lu, Adv. Mater. 2017, 29, 1700414.

[61] Y. Li, P. Li, and Z. H. Lu. AIP Adv. 2018, 8, 035218.

[62] C. T. Wang, C. C. Ting, P. C. Kao, S. R. Li, and S. Y. Chu, Appl. Phys. Lett. 2018, 113,

051602.

[63] C. W. Chen, P. Y. Hsieh, H. H. Chiang, C. L. Lin, H. M. Wu, and C. C. Wu, 2003, Appl.

Phys. Lett. 83, 5127.

[64] S. I. Kim, H. W. Oh, J. W. H, B. K. Ju, and C. W. Lee, 2011, Thin Solid Films 519, 6872.