Copyright
By
Wei Guo
2021
The Dissertation Committee for Wei Guo Certifies that this is the approved version
of the following Dissertation:
Transition Metal Oxide Thin Films Integration on SrTiO3
Committee:
Alexander A. Demkov, Supervisor
John G. Ekerdt
Alejandro L. De Lozanne
Maxim Tsoi
Xiaoqin Li
Transition Metal Oxide Thin Films Integration on SrTiO3
by
Wei Guo
Dissertation
Presented to the Faculty of the Graduate School of
The University of Texas at Austin
in Partial Fulfillment
of the Requirements
for the Degree of
Doctor of Philosophy
The University of Texas at Austin
May 2021
Dedication
To all the pains and gains in the past years.
v
Acknowledgements
It has been a long, long journey and it finally comes to an end. Thanks to all the
pains I have been through in the past years. They grinded me. They completed me.
I would like to thank my parents, for unconditionally supporting me in everything.
For my overseas study, we sacrificed a lot of time that we were supposed to be together.
They raised me up with everything they could give. And I am proud to be their son.
I gratefully thank my advisor, Prof. Alex Demkov, for his guidance and help for
the past years. I admire his intelligent and knowledgeable mind. I appreciate his hard
working and workout spirit.
Many thanks to our lab manager, Agham Posadas, for all his training and help he
has given me. He is always reliable in the lab and knows everything! I also thank all my
group members, Patrick, Kristy, Elliot, Tobi, Marc, Lingyuan, Hosung, Ali, Donghan,
Wente, Fatima, Therese, Ilya, Jackie, and group members from Prof. John Ekerdt group,
Ed, Bryce, Shen, Di, and Annie. Thank you for all your help and for being good lab mates!
Also I thank my friends in Prof. Keji Lai’s group, Lu, Di, Zhaodong, Xiaoyu. Especially
Lu as a good listener and gym partner, and a horse rider. My research life would have been
much worse if I did not know all these fun, nice people!
I thank Prof. Vladimir Strokov for his help during my visit to Switzerland. Thanks
to my undergraduate advisor, Prof. Min Xiao and Prof. Chunfeng Zhang, who has led me
into the world of physics research and provided me with opportunities.
I want to give special thanks to my friend Yuntian Song who gave me faith and
support during my darkest time. I hope he continues well to fulfill his ambition and I
believe he will.
vi
I also want to thank my high school physics teacher, Pei Liu, who inspired me and
opened the gate of the physics world to me. I would have never chosen physics as my major
and be a physics Ph.D. if he did not teach me.
And at last, I thank all the time I was alone and not alone. All the sleepless nights I
have been through. All the days filled with frustration and self-doubt. All the experiences
and all the thoughts. I learned more than physics and research in the past seven years. It is
a terminal point of my campus life. But it is just a node in life. The next chapter of my
future is opening. I will keep moving.
vii
Abstract
Transition Metal Oxide Thin Films Integration on SrTiO3
Wei Guo
The University of Texas at Austin, 2021
Supervisor: Alexander A. Demkov
Transition metals (TMs) have an immense range of intriguing physical properties
and phenomena. Transition metals and transition metal oxides (TMOs) play a very
important role in modern day scientific research and industry. TMs and TMOs exhibit an
even broader range of structural, electrical, magnetic, and optical properties when
fabricated as thin films on other materials compared to their the bulk forms. SrTiO3 (STO)
is a widely used cubic perovskite oxide material known for its excellent electronic
properties. When TMs and TMOs are integrated on STO, effects based on their interaction
with STO have lead researchers to explore the physics behind such effects and their
possible industrial device applications. In this thesis, we will mainly focus on the
integration of TMs and TMOs with STO and with epitaxial STO grown on Si.
The interactions of the transition metal Pt and rare earth metal Eu when deposited
on STO by molecular beam epitaxy (MBE) were investigated. For Pt growth on STO, I
investigated the properties of ultrathin Pt as a function of coverage on TiO2-terminated
SrTiO3 substrate at different temperatures. I used in situ x-ray photoelectron spectroscopy
(XPS), ex situ scanning electron microscopy (SEM) and atomic force microscopy (AFM)
to observe the evolution of the electronic structure and surface morphology of Pt. I
compared the electronic structure of Pt and the different growth patterns at low and high
viii
temperatures. I also performed high temperature annealing of low temperature-grown
samples and found a “bubble-up” behavior of the continuous film.
We also performed ultra-high vacuum deposition of Eu metal on STO (001) and
achieved EuO epitaxy on STO via oxygen scavenging. I explored the oxygen scavenging
behavior of Eu using STO films on Si by varying the STO thickness and Eu deposition
temperature. In situ XPS was used to investigate the electronic structure of the nominal
Eu/STO/Si stack. Our XPS results on the Eu/EuO stack revealed an unusual downward
band bending at the interface. This is supported by density functional theory calculations
by Gao. This work has been published in J. Appl. Phys. 121, 105302 (2017) and J. Appl.
Phys. 124, 235301 (2018). Theoretical calculations performed in our group predicted the
existence of a 2-dimensional electron gas (2DEG) at the EuO/STO interface and
demonstrated that the 2DEG location can be controlled if an additional layer of BaTiO3 is
included. To explore this effect on the 2DEG, I performed soft x-ray angle-resolved
photoemission spectroscopy (SX-ARPES) with our collaborators at the Swiss Light
Source. The results are currently being summarized for publication.
For possible applications in Si photonics, I performed a detailed study of dry
oxidation of a Si substrate below a thin epitaxial SrTiO3. Annealing time and temperature
are the key factors to optimize the SiO2 thickness. I developed a theoretical model based
on a modification of the Deal-Grove-Massoud formalism that predicted the thickness of
SiO2 formed underneath STO as a function of time and temperature. The model fits the
experimental data well. This work has been published in J. Appl. Phys. 127, 055302 (2020).
In addition, I performed preliminary studies on free-standing STO membranes. I developed
a fabrication process and performed Raman measurements.
We also proposed a quantum well structure design of BaSnO3/SrTiO3/Al2O3 to
make use of the conduction offset as large as ~3.5eV between BaSnO3 (BSO) and Al2O3.
The whole deposition process is done by MBE and characterized by reflection high energy
electron diffraction (RHEED) and XPS to confirm the BSO film quality. We are still
working on improving the quantum well quality to be able to make multiple quantum well
structures.
ix
Table of Contents
List of Tables .................................................................................................................... xii
List of Figures .................................................................................................................. xiii
Chapter 1: Introduction ........................................................................................................1
1.1 Transition metal and oxide thin films ...................................................................1
1.2 SrTiO3 on Si ..........................................................................................................3
1.2.1 Si ............................................................................................................3
1.2.2 SrTiO3 ....................................................................................................4
1.2.3 SrTiO3 growth on Si ..............................................................................5
1.2.4 SrTiO3/Si and SrTiO3 free-standing membranes ...................................8
1.3 Chapter overview ................................................................................................12
1.4 Reference ............................................................................................................14
Chapter 2: Experiment techniques .....................................................................................19
2.1 Molecular beam epitaxy (MBE) .........................................................................20
2.1.1 Operating principles and instrument illustration .................................20
2.2.2 Preparation for MBE growth ...............................................................23
2.2.3 Maintaining vacuum and cryopump ....................................................24
2.2.4 Effusion cells .......................................................................................25
2.2.5 Electron beam evaporator ....................................................................25
2.2.6 Quartz crystal monitor (QCM) ............................................................26
2.2 Reflection high-energy electron diffraction (RHEED) .......................................26
2.3 X-ray photoelectron spectroscopy (XPS) ...........................................................29
2.3.1 Operating principles .............................................................................30
x
2.3.2 Surface sensitivity ................................................................................32
2.3.4 XPS peak analysis ................................................................................32
2.4 References ...........................................................................................................33
Chapter 3: Temperature dependence of the morphology and electronic structure of
ultrathin platinum on TiO2-teminated SrTiO3 (001) ....................................................34
3.1 Introduction .........................................................................................................34
3.2 Experiment ..........................................................................................................37
3.3 Results and Discussion .......................................................................................38
3.4 Conclusions .........................................................................................................47
3.5 Acknowledgment ................................................................................................48
3.6 References ...........................................................................................................48
Chapter 4: EuO epitaxy on SrTiO3 by oxygen scavenging ................................................50
4.1 Introduction .........................................................................................................51
4.2 Experiments on Oxygen Scavenging of STO by Eu...........................................56
4.3 Results and Discussion .......................................................................................58
4.3.1 High temperature deposition with thick STO film ..............................58
4.3.2 High temperature deposition with thin STO film ................................61
4.3.3 Low temperature deposition ................................................................66
4.3.4 Theoretical verification by DFT ..........................................................69
4.4 Further Research on Field-Effect of 2DEG in EuO/STO ...................................74
4.5 Conclusions .........................................................................................................82
4.6 Acknowledgement ..............................................................................................84
4.7 References ...........................................................................................................84
xi
Chapter 5: Thermal oxidation of Si buried under thin SrTiO3 and free-standing
SrTiO3 membranes .......................................................................................................87
5.1 Introduction to Dry Oxidation of Si ....................................................................87
5.2 Si Oxidation Experiments ...................................................................................90
5.3 Theoretical Model of Si oxidation ......................................................................97
5.4 Fabrication of Free-standing STO Membranes ................................................106
5.5 Measurements of STO membranes ...................................................................108
5.6 Conclusions .......................................................................................................115
5.7 Acknowledgements ...........................................................................................116
5.8 References .........................................................................................................116
Chapter 6: Advanced design of BaSnO3/SrTiO3/Al2O3 quantum wells ..........................120
6.1 Introduction .......................................................................................................120
6.2 Growth and QW design ....................................................................................122
6.3 characterization .................................................................................................126
6.4 Outlook .............................................................................................................128
6.5 References .........................................................................................................129
Chapter 7: Summary and future work ..............................................................................130
7.1 Summary ...........................................................................................................130
7.2 Future work .......................................................................................................132
Bibliography ....................................................................................................................133
xii
List of Tables
Table 1. Summary of the growth conditions and the corresponding lattice parameters
from different growth methods reported in the literature (with published
years). Blanks indicate the corresponding paper did not include detailed
data. (Pressure unit is Torr if not marked differently). .................................11
Table 2. Oxidation parameters of the Massoud model [21] for temperature less than
1000°C. .......................................................................................................103
Table 3. Phonon branch assignments for second-order Raman peaks measured from
STO bulk substrate. .....................................................................................111
xiii
List of Figures
Figure 1.1: Transition metals (yellow) in the periodic table frame with lanthanides
and actinides shown separately (light yellow) ................................................2
Figure 1.2: Atomic crystal structure of ABO3 perovskite oxide. This is a unit cell
schematic with green A-site atoms and blue B-site atoms. All small red
spheres are oxygen atoms. ..............................................................................2
Figure 1.3. (a)Top view of the atomic arrangement of STO on the Si (100) surface,
with the STO unit cell rotated 45º with respect to the Si unit cell (b) Side
view of the STO/Si interface. The dimers of Si are bonded to the half
ML of Sr and to oxygen, allowing STO to be epitaxially grown. ..................6
Figure 1.4. Schematic of a waveguide phase modulator. ...............................................12
Figure 2.1: Schematic of the vacuum system at the Advanced Atomic Design Lab. All
instruments are linked to the main transfer line. ...........................................19
Figure 2.2: Front view of MBE with components labeled.................................................21
Figure 2.3: Back view of MBE with components labeled .................................................22
Figure 2.4: Schematic of the principle of RHEED. The angle of incidence of the beam
is usually less than 5º from the surface. ........................................................27
Figure 2.5: RHEED images of (a) commercial STO substrate bought from MTI
Crystal (b) 5 uc of STO grown on a Si substrate. .........................................28
Figure 2.6: Picture of the VG Scienta R3000 XPS system. (Front view) ..........................30
Figure 2.7: Conventional X-ray source for an XPS instrument. (Figure adopted from
[11])...............................................................................................................31
xiv
Figure 3.1: Pt 4f core level spectrum for deposition at 200 ºC on STO as a function of
Pt thickness. At 0.25 ML coverage (the lowest) the Pt 4f7/2 peak has a
binding energy of 72.4 eV. It gradually shifts by 1.1 eV to the bulk
metallic value (71.3 eV) when Pt reaches 2 ML. ..........................................39
Figure 3.2: Pt 4f7/2 binding energy as a function of Pt thickness for different
deposition temperatures (200–800 ºC). All values shift to the bulk Pt
metal value (~71.3 eV) at different temperatures when Pt is sufficiently
thick. Lower deposition temperature will lead to larger total binding
energy shift while also taking more layers to reach the bulk metal value. ...41
Figure 3.3: Sr 3d, Ti 2p, O 1s binding energy shift as a function of Pt thickness. There
is a ~0.3 eV shift for all three core levels while Pt is accumulating,
which is consistent with an upward band bending........................................41
Figure 3.4: Pt/STO valence band spectrum grown at 800 ºC as a function of thickness.
It shows the Pt contribution to the signal gradually increase in the range
of 0–3 eV. At 2 ML, it shows nonzero intensity at 0 eV which means the
Fermi edge starts to form. .............................................................................42
Figure 3.5: (a) STO surface after water boiling treatment. It shows a terrace structure
with flat steps (b) AFM measurement of 2ML Pt/STO grown at 800 ºC
and corresponding line profile measurement. The Pt nanoclusters (with
~150nm diameter and 0.5 nm height) arrange themselves in rows aligned
along the substrate terrace edges. ..................................................................43
xv
Figure 3.6: (a) SEM image of Pt/STO grown at 200 ºC for 10 ML. At the nanometer
scale we can only see a featureless surface with a clear edge when we
check it at micrometer scale (as shown in the inset). The left area in the
inset has no Pt grown on because of shadowing by the sample holder
during deposition. (b) SEM image of Pt/STO grown at 800 ºC for 1 ML.
It shows uniform nanoclusters with average lateral size of ~80 nm. ............45
Figure 3.7: (a) SEM image of Pt/STO with 10 ML of Pt grown at 200 ºC and
annealed at 800 ºC for an hour in air. The flat surface breaks out and
splits into small nanoclusters. Pt clusters have radii in the 3-5 nm range
with a separation of ~25-30 nm. (b) Histogram of the cluster sizes
obtained from the SEM image. The first two columns are gray because
of the overcounting of the algorithm at very small radius. Red line is the
normal distribution fit of the rest of the columns centered at 3.75 nm. ........47
Figure 4.1: Schematic of EuO/STO epitaxial arrangement: (a) side view (b) top view. ...55
Figure 4.2: RHEED pattern for EuO grown on 10 nm STO/Si at 300 °C. ........................59
Figure 4.3: Aberration-corrected TEM images of the EuO/STO interface with the
atomic model overlaid. Pink circles are Eu, red circles are oxygen, green
circles are Sr, and blue circles are Ti. (a) and (b) HAADF and BF pair of
images for EuO [110]/STO [100] projections. (c) and (d) HAADF and
BF pair of images for EuO [100]/STO [110] projections. ............................60
Figure 4.4: VB spectrum of Eu grown on 10 nm STO/Si. Only Eu2+ and oxygen
features are visible. .......................................................................................61
Figure 4.5: RHEED pattern evolution during Eu deposition on a thin (2 nm) STO
layer...............................................................................................................62
xvi
Figure 4.6: (a) VB spectra of Eu on 2-nm STO/Si. Two types of Eu peaks are visible.
(b) Si 2p signal of Eu deposited on 2-nm STO/Si. Silicon metal signal is
almost gone but SiOx and EuSiy appear. .......................................................63
Figure 4.7: VB spectra of step-by-step Eu growth on 3-nm STO. Growth times are
shown with different colors that correspond with the spectra. .....................65
Figure 4.8: (a) Valence band spectrum of Eu on 2-nm STO/Si at 20 °C. (b) Eu 4d
spectrum with clearly resolved multiplets characteristic of Eu metal. .........67
Figure 4.9: Valence band evolution of Eu deposition on 2-nm STO/Si at 20 °C. .............68
Figure 4.10: (a) Valence-band XPS spectrum of 10 min Eu deposition on 2-nm
STO/Si pseudo-substrate at 20 °C. This is consistent with the spectrum
from bulk EuO. (b) Valence-band XPS spectrum of 60 min Eu
deposition at 20 °C. A Eu/EuO/STO/Si structure is formed. (c) The
attenuated and broadened 4f DOS of the whole heterostructure
(“calculated pseudo-XPS spectrum”). The Eu metal 4f DOS is marked as
dark blue, the 4f DOS of each EuO layer is marked as magenta, the total
EuO 4f DOS is marked as violet, and the total 4f DOS is marked as red.
The inset gives the 4f DOS of each EuO layer on a larger scale. (e) A
three-dimensional representation of the “calculated-pseudo-XPS
spectrum.” Here the spectrum of each layer is shown separately along
the y direction. The peak position of 4f EuO DOS of each layer is
connected with a violet line. (Figure is also published in Ref. [35]) ............71
Figure 4.11: Valence band scan of a EuO/STO/BTO/Si sample after transportation in
ambient. .........................................................................................................76
Figure 4.12: Eu resonant map spectrum at the Eu 3d core level. .......................................77
Figure 4.13: Ti resonant map with strong Eu2+ 4f signal. ..................................................78
xvii
Figure 4.14: Zoom-in Ti resonant map with possible 2DEG signals (circled in red). .......79
Figure 4.15: Oxygen 1s core level comparison for 1) Ge/EuO/STO/BTO/Si sample
before it was taken out of vacuum (red); 2) The same sample in step 1)
left in air for over 2 days (green); 3) The same sample in step 2) after
being heated in vacuum at 350 ºC for 1 hour (blue). Oxygen from EuO
and GeOx are marked above the peaks. ........................................................80
Figure 4.16: Valence band comparison for 1) Ge/EuO/STO/BTO/Si sample before it
was taken out of vacuum (red); 2) The same sample in step 1) after air
exposure for over 2 days (green); 3) The same sample in step 2) after
heating in vacuum at 350 ºC for 1 hour (blue). Eu signal positions are
marked above the peaks. ...............................................................................82
Figure 5.1 Example ellipsometry measurement with the corresponding model fit.
These are the measurement results at 45°. This gives a STO layer with
17.54 nm and SiO2 layer with 7.54 nm thickness. The red one is the
curve and the blue one is the curve. Measured data are shown as open
circles and the fit curves as solid lines. .........................................................92
Figure 5.2: (a) RHEED image for a 10 nm STO/SiO2/Si after the growth and before
anneal. STO pattern is shown along the [110] direction. (b) RHEED
image for a 10 nm STO/SiO2/Si after 800°C anneal for 2 hours. STO
pattern along the [110] direction is still sharp and clear. ..............................94
Figure 5.3: XPS spectrum for the Si 2p region. The major peak around 104.5 eV is
SiO2 and the minor peak around 103 eV is SiOx, Si metal (99.5 eV) is
also still visible. ............................................................................................95
xviii
Figure 5.4: Comparison of the XRD out-of-plane full scans before and after
annealing. STO peaks decrease a little and Si is a little higher. All peaks
do not show any obvious deformation other than intensity variation. ..........96
Figure 5.5: The schematic of oxygen propagation in the SrTiO3/SiO2/Si structure.
The inset shows the atomic structure schematic. ..........................................98
Figure 5.6: Diffusivity of oxygen in SrTiO3 (D1) and SiO2 (D2) in the 300-1200°C
temperature range. Inset: Diffusivity of oxygen in SrTiO3 and SiO2 in the
300-700°C temperature range. The STO low temperature (<700°C) data
is obtained by inverse relationship projection from existing diffusivity at
higher temperature (>700°C) ......................................................................100
Figure 5.7: The SiO2 thickness as a function of oxidation time at different
temperatures from equations (12) and (13). The lines represent the
model and the shapes represent the experimental data. Diamonds are
samples annealed under 800°C, circle is under 750°C and square is
under 700°C. All thickness data have an error bar of ~1 nm. .....................104
Figure 5.8: Schematic of the procedure to fabricate STO membranes ............................107
Figure 5.9: (a) (b) are SEM images from different spots on STO/Si etched sample.
The biggest membrane outlined by yellow is ~15 × 15 μm2 ......................108
Figure 5.10: Raman spectrum of a bulk STO substrate. Characteristic peak positions
are labeled and marked with red arrows. ....................................................110
Figure 5.11: Raman spectrum of 20 nm STO thin film on Si. (Thin film grown by
MBE)...........................................................................................................112
Figure 5.12: Raman signal of STO membranes made from 20 nm STO/Si ....................114
Figure 6.1: Band alignment of a single quantum well. The conduction band offset
between BSO and Al2O3 is 3.5 eV. .............................................................124
xix
Figure 6.2: Schematic of the heterostructures to be grown. ............................................124
Figure 6.3: Simulation results for possible energy states in a single quantum well. .......126
Figure 6.4: RHEED evolution from (a) thick Al2O3 on STO buffer layer on Si (b) 3 uc
STO on previous Al2O3 surface (c) 10 uc BSO on previous STO surface..128
1
Chapter 1: Introduction
1.1 TRANSITION METAL AND OXIDE THIN FILMS
Transition metals play a very important role in modern day scientific research and
in industry. Transition metals (TMs) are commonly taken as the set of elements located in
Group 3 to 12 of the periodic table, as shown in Figure 1.1. Elements with a partially filled
d sub-shell are identified as transition metals. In addition, lanthanide and actinide metals,
with a fully filled d shell but only partially filled f sub-shell, are also sometimes considered
as transition metals. Electrons of the partially filled sub-shell provide not only multiple
stable valence states, but are responsible for the tremendous variety of interesting physical
properties and phenomena in TMs, including magnetism [1], superconductivity [2]-[4],
catalytic properties [5], ferroelectricity [6], multiferroicity [7], [8], and many more [9].
Among all the many kinds of TM compounds, transition metal oxides (TMOs) are
the main focus of this thesis. Multiple oxidation states of TMs enrich the oxide family with
different electronic configurations. The oxidation process transfers the outer s electrons to
bond with oxygen leaving partially occupied d shell as the highest occupied electronic
states. These outer shell electrons determine the physical properties of TMOs [9], [10].
TMOs crystallize in the perovskite, rock salt, rutile, corundum, spinel and other oxide
crystal structures. Perovskite oxides, described by the chemical formula ABO3 (Figure 1.2),
have been widely studied over many decades due to multiple phenomena they exhibit
including, but not limited to, ferroelectricity [6], ferromagnetism [11] high-κ dielectric
constant [12], metal-insulator transitions [13]. We will focus TMs and TMOs deposited on
SrTiO3 in this thesis and study the physical phenomena in these material systems.
2
Figure 1.1: Transition metals (yellow) in the periodic table frame with lanthanides and
actinides shown separately (light yellow)
Figure 1.2: Atomic crystal structure of ABO3 perovskite
oxide. This is a unit cell schematic with green A-site
atoms and blue B-site atoms. All small red spheres are
oxygen atoms.
TMOs exhibit a broad range of structural, electrical,
magnetic or optical properties when fabricated as thin films
instead of bulk crystals [10], [14], [15]. The reduction of the
dimensionality from 3D to 2D provides an extra degree of
freedom for tuning the physical properties and gives even more possibilities when a 2D
film is integrated on different substrates. Symmetry breaking and interfaces of thin film
heterostructures result in phenomena like charge transfer and are the origin of novel
physical phenomena not observed in bulk materials [16], [17]. For example, a
3
heterojunction of two insulating thin films LaAlO3/SrTiO3 shows the unexpected
interfacial superconductivity [18]-[20]. Interfacial magnetism was discovered in the same
materials system even though both oxides are non-magnetic in bulk [19]-[22]. A very
strong polarization enhancement can be obtained when a non-ferroelectric material is
fabricated as a part of a superlattice with a traditional ferroelectric material [23]. The
immense potential of the integration of TMO thin films with semiconductors leads to many
more exciting possibilities both in fundamental physics and in novel applications.
1.2 SRTIO3 ON SI
1.2.1 Si
Si is arguably the most important semiconductor material today. It is widely used
as a platform for the modern electronics industry based on the complementary metal–
oxide–semiconductor (CMOS) technology [24] and the metal–oxide–semiconductor field-
effect transistor (MOSFET), which is probably THE most important and widely
manufactured device in the information age. Si is also an excellent integration platform for
photonic applications due the ability to fabricate Si waveguides with very low losses and
high optical mode confinement [25]-[27]. The seemingly unlimited supply of Si on earth
is another reason that makes Si so important. The fundamental reason of the irreplaceability
of Si is in its structural and electronic properties.
Si has an indirect band gap of 1.12 eV [28] and can be doped to be a conductor with
dopants like As or P for n-type doping, and B for p-type doping [29]. It has a diamond
crystal structure with covalent bonds, which results in a very high melting point of 1414
°C. The oxidation of Si is another key factor for its widespread industrial application.
The native oxide SiO2 has a tightly bonded, nearly defect-free interface with Si. Great
4
interface quality obtained through modern day growth methods affords great transistor
performance in state-of-art industrial production [29]. The wide bandgap of SiO2 (~9 eV
[30]) limits the gate leakage current to an extremely low level when used as a gate
dielectric. The great thermal and mechanical stability [31], high dielectric strength [32],
and convenience of growth, all made SiO2, until very recently, a passivation layer of choice
in the fabrication and manufacturing process of electronic devices.
Due to rapid developments in the semiconductor industry and the miniaturization of
modern semiconductor devices that came along with it, Moore’s Law is no longer followed as
devices approached the fundamental physical limits of traditional materials. For decades,
Si/SiO2 used to be the only option in the semiconductor industry. However, it faced challenges
when the dimension requirements for gate dielectrics scaled down to just several Å in thickness
in recent years [33]. High leakage current at this thickness caused increasing power usage and
device failures, which made people seek new approaches at this length scale [33]. The use of
high-κ gate dielectrics is one of the implemented solutions. The new gate stack can still have a
relatively large dielectric thickness (d) while maintaining a higher capacitance as the
dependency is 𝐶 ∝ 𝜅
𝑑. Intel Corporation launched their 45 nm technology node in 2007 using
hafnium dioxide (HfO2) as gate dielectric [34], [35]. As of today, people are still seeking
alternative methods both material-wise (MoS2 [36], graphene [37]) and architecture-wise, with
various types of FETs (FinFET [38], GAAFET [39]) being studied and developed.
1.2.2 SrTiO3
SrTiO3 (STO) is a widely used cubic perovskite oxide material known for its
excellent electronic properties. STO exhibits water photolysis [40], superconductivity [41],
room temperature ferroelectricity (when strained) [42], and many other fascinating
5
properties. It has a lattice constant of 3.905 Å which is lattice-match to many other complex
oxides, which makes it favorable to serve as a substrate or in a superlattice. The interface
of STO with other complex oxides also exhibits a large array of fascinating phenomena
like ferromagnetism and superconductivity in LaAlO3/SrTiO3 interfacial two-dimensional
electron gas (2DEG) [18]-[22], large positive linear magnetoresistance at the EuO/SrTiO3
interface [43], magnetic ordering in LaMnO3/SrTiO3 superlattices [44], etc. These STO-
based effects lead to the further exploration of the physics behind such effects and spurred
the search for possible industrial device implementations. As Si is still the main
“workhorse” in the modern semiconductor industry, integration of STO onto Si would open
new possibilities to utilize these electronic properties and create more efficient devices with
new functionality. Fortunately, such a process of integration does exist [45]. In this thesis,
I will mainly focus on the integration of TMs and TMOs with STO and with STO on Si.
1.2.3 SrTiO3 growth on Si
The discovery by Mckee et al. in 1998 of a way to epitaxially nucleate STO on Si
directly by molecular beam epitaxy (MBE) has opened the possibility of integrating other
perovskite materials on Si [45]. A typical process involves first removing the surface SiO2
and then depositing 0.5 monolayer (ML) of Sr metal in an MBE chamber [46]-[50]. This
Sr sub-monolayer prevents Si oxidation during the initial nucleation of STO [51]. A thin
(1-4 nm), barely crystalline STO layer is then deposited at low temperature (<300°C) with
the correct stoichiometric ratio of Sr and Ti in the presence of modest oxygen pressure (10-
8 ~ 10-7 Torr) [50]. This layer is then fully crystallized in vacuum at 500-550°C
(crystallization temperature could be higher if STO is slightly off stoichiometry), which
results in SiO2-free epitaxial STO on Si [52]. We show a top view of the atomic
6
arrangement of STO on the Si (100) surface and a side view of the interface in Figure 1.3.
From the top view, the STO unit cell is rotated by 45º with respect to the conventional Si
unit cell due to the lattice parameter of Si being ~√2 times that of STO. Other perovskites,
like BaTiO3 or LaAlO3, which are epitaxially grown on the initial STO layer are expected
to be crystallographically aligned with the STO template. From the side view, the half ML
of Sr is located in the gaps of the dangling bonds of the clean Si surface and serves as a
surface that allows STO to epitaxially crystallize on top and form the perovskite structure.
Figure 1.3. (a)Top view of the atomic arrangement of STO on the Si (100) surface, with
the STO unit cell rotated 45º with respect to the Si unit cell (b) Side view of
the STO/Si interface. The dimers of Si are bonded to the half ML of Sr and
to oxygen, allowing STO to be epitaxially grown.
7
Once the initial template of 4-5 ML of STO is crystallized, additional STO can be
grown with two possible options. If the additional STO is again grown at low temperature
(<300°C) with modest oxygen supply (mid 10-7 Torr) followed by annealing in vacuum to
fully crystallize it, one would get a clean SiO2-free interface between STO and Si. If instead
one grows additional STO, treating the template as if it were a single crystal STO substrate,
additional SiO2 will be formed since oxygen has enough diffusivity in STO at the normal
STO growth temperature of 500-550°C to reach Si. The resulting SiO2 thickness in this
case will increase as the total deposition time is prolonged. For STO layers less than 20 nm
thick, this typically produces about 1-2 nm of SiO2 interlayer between STO and Si [52].
Depending on whether or not the presence of this SiO2 interlayer is critical for an
application, the appropriate process for growing additional STO should be used. Note, that
despite the presence of this amorphous SiO2 interlayer, the resulting STO layer is still
single crystal and in epitaxial registry to Si. The growth of the STO template layer is very
crucial to those epitaxially-grown materials on top. The template crystal quality and strain
highly affect the quality and strain of following epitaxial film, which can introduce some
unexpected features or unwanted defects. Interfacial SiO2, due to high temperature growth,
is also very useful as a film strain engineering tool for STO and other materials deposited
on STO [52]-[54]. Zhang et al. have reported a thermal engineering process for tuning
strain during STO/Si growth [54]. By using the large mismatch of the thermal expansion
coefficients of STO and Si, they were able to produce a continuously tunable STO strain
by varying the growth temperature. Their main idea is to utilize the misfit dislocations
formed during growth and the amorphous SiO2 interlayer between STO and Si to control
residual stress. In-plane tensile stress is built up during the cooling process from the growth
temperature to room temperature. This thermal engineering method provides a practical
8
way to manipulate the STO template and the epitaxial film grown on top since the strain
level of the template layer directly affects the film quality, relaxation process, and lattice
parameters of the epitaxial film.
1.2.4 SrTiO3/Si and SrTiO3 free-standing membranes
With the STO/Si deposition technique and so many STO-based complex material
systems and fascinating phenomena, STO can serve as an excellent bridge material to
connect Si and other TMs or TMOs, enabling the integration of these materials on Si to
open more possibilities of novel devices. We will list several well-studied materials
systems based on STO/Si below.
High mobility interfacial 2DEG in LaAlO3/SrTiO3 was first reported by Ohtomo
and Hwang in 2004 [55]. This discovery attracted huge interest and new interesting
properties of this interface have also been studied, as mentioned before [18]-[22]. STO and
LAO in bulk are both insulating materials and their band gaps are 3.3 eV and 5.6 eV,
respectively [56], [57]. While a lot of papers discuss the LAO/STO system as the next
generation electronic device [58]-[60], Ortmann et al. published a series of papers on the
implementation of the LAO/STO quantum well (QW) structure for novel electro-optic
(EO) devices [61]-[64]. They overcame many fundamental limitations for the integration
of thick STO/LAO QW heterostructures on Si, such as the significant structural distortions
at STO/LAO interfaces, which previously restricted thick heterostructure growth [65].
They managed to make 10 periods of 7 uc LAO/4 uc 1% La:STO QW on Si with STO
buffer layer via molecular beam epitaxy (MBE) and confirmed great crystal quality by
reflection high-energy electron diffraction (RHEED), X-ray diffraction (XRD), and
scanning transmission electron microscopy (STEM) measurements [64]. They also
9
presented calculations based on the LAO/STO QW on Si for novel electro-optical devices
operating at near-infrared optical wavelength [63]. This work has opened a new avenue for
the development of a wide range of novel optical, electro-optical devices, sensors, light
sources and photonics.
Another significant materials system based on STO/Si is ferroelectric perovskites on
STO. The Pockels effect is an electric field-induced change in the refractive index of a crystal.
It was first demonstrated by Friedrich Pockels in 1893 [66]. BaTiO3 (BTO) is a highly
promising ferroelectric material for the fabrication of electro-optic (EO) modulators because it
exhibits one of the largest Pockels coefficients among the EO materials [67]-[71]. It also has
the added benefit of being readily integrated on a Si materials platform with an STO buffer
layer. These two characteristics make epitaxial BTO ideal for use in next generation silicon
photonics application with fast, low-power optical switches, or even for new forms of
computing including neuromorphic and quantum computing [68], [71]. At room temperature,
bulk BTO has a tetragonal crystal structure with space group P4mm and lattice constants a =
3.994 Å and c = 4.0335 Å. It has a remnant ferroelectric polarization of 26 μC/cm2 and a Curie
temperature of 130ºC, above which it becomes a cubic paraelectric material with a lattice
constant of 4.006 Å and space group Pm3m [70]. Being a ferroelectric, BTO has a unique
crystallographic direction in which the ferroelectric polarization points. For EO modulators,
because the polarization direction controls the coupling between light and an external electric
field, it is important to understand how different growth methods and subsequent processing
affect the direction of the ferroelectric polarization. Certain electro-optic devices may require
polarization to be in the plane of the film (in-plane switching liquid crystal devices), while
other applications may require it to be normal to the plane of the film (Mach-Zehnder
modulator).
10
Orientation of BTO is the key factor of the EO response of the film. At room
temperature, there is a ~4% lattice mismatch between BTO and Si (calculated as (aSi-
aBTO)/aBTO). We call a BTO film c-axis-oriented if its c-axis is pointing normal to the substrate
surface plane. On the other hand, we call a BTO film a-axis-oriented if its c-axis lies in the
plane of the film. There is also a large difference between the coefficients of thermal expansion
of BTO (9.0×10-6 K-1) and Si (2.6×10-6 K-1), which leads to an additional in-plane biaxial
tensile stress in the BTO layer during cooling [70]. This additional stress can cause orientation
changes in the BTO polarization. In this section, we summarize several of the most relevant
publications about epitaxial BTO on Si and compare the BTO structural parameters they
obtained with each other.
In Table 1, I have summarized the growth conditions and the corresponding BTO
lattice parameters obtained from the different growth methods [72]. The techniques used to
grow BTO have several things in common: thicker BTO on STO/Si tends to produce more a-
axis orientation; post-deposition annealing of at least 600 ºC in air or oxygen can transform a
c-axis BTO film to an a-axis one; higher oxygen pressure enhances a-axis orientation. In
addition, the quality and the thickness of the STO template layer, and the presence and amount
of an SiO2 interlayer, can significantly affect the manner and degree of BTO film relaxation.
Method Tempera
ture (ºC)
Oxygen
pressure (Torr)
STO thick-
ness (nm)
BTO thick-
ness (nm)
a (nm) c (nm) Orient
ation
MBE (2006) 550-570 4-5×10-8 30 (BST) 10 3.9996±0
.0005
4.025±
0.003
c
MBE (2007) 600 2 30 4.01 4.05 c
MBE (2011) 620 1×10-6 5 40 3.978 4.057 c
MBE (2013) 700 5×10-6 3.9/6.2 8 3.993 3.038 c
MBE (2013) 700 5×10-6 3.9/6.2 40 3.993 4.015 a/c
MBE (2013) 600 1×10-5mbar 4 8 4.038 c
MBE (2014) 440-525 1-5×10-7 4 7 3.996 4.027 c
MBE (2013) 500 1×10-7mbar 2 8 4.032 c
11
MBE (2013) 600 1×10-5mbar 4 130 3.997 4.032 a
MBE (2014) 8 80 3.998 4.03 a
MBE (2015) 4-6 90 4.03 c
4-6 90 3.99(850
ºC RTA)
a
MBE (2017) 630 1.6×10-6 5 100 4.00 4.06 c
40 100 4.00 4.02 a
MBE (2019) 650 5×10-7 15 40 3.996 4.046 c
ALD (2014) 1.6 7-20 3.93 4.02 c
ALD (2019) 2 66 4.007(65
0ºC
RTA)
a
ALD (2020) 3.6 10 3.98 a
Rf sputtering
(2012)
700 1.5-3×10-4 20 120 4.024 c
Rf sputtering
(2013)
500-600 10-5 bar 4 100 3.966 4.062 a
Rf sputtering
(2020)
650 2 Pa 10 60 4.003 4.042 a
PLD (2018) 450 0.02 mbar LNO/CeO2/
YSZ
110 4.04 4.12 c
PLD (2018) 700 0.1 mbar LNO/CeO2/
YSZ
110 3.99 4.02 a
PLD (2020) 650 0.01 mbar 10 100 4.09 c
Sol-gel (2020) 2 20 3.980 4.036 a
Table 1. Summary of the growth conditions and the corresponding lattice parameters
from different growth methods reported in the literature (with published
years). Blanks indicate the corresponding paper did not include detailed
data. (Pressure unit is Torr if not marked differently).
A typical waveguide phase modulator based on BTO/Si system is shown in Figure
1.4. The waveguide is located between the metal contacts, and the electric field is
horizontal, or parallel to the surface. The confinement is achieved by placing a Si
waveguide atop the perovskite (indicated with an orange bar in Figure1.4). As a matter of
fact, the majority of the mode is in Si, and only about 20% is in BTO. The electro-optically
induced index change results in a phase shift that depends on the applied voltage, length of
12
the modulator, wavelength of light, spacing between the electrodes G, and the overlap
integral between the applied electric field and the optical mode. Details of calculations can
be found elsewhere [72]. I published a review paper [72] that reviewed the growth of
epitaxial BTO on Si by a variety of deposition methods including MBE, pulsed laser
deposition (PLD), and RF sputtering. I summarize the resulting BTO film structure and
quality based on the reported characterization results and discussed EO measurements of
basic devices made from this materials platform, where such data is available.
Figure 1.4. Schematic of a waveguide phase modulator.
1.3 CHAPTER OVERVIEW
In this thesis, my research mainly focuses on TM and TMO integration on STO and
STO/Si. Chapter 2 contains an introduction to the experimental equipment used in my
research at the Advanced Atomic Design Lab in the University of Texas at Austin. Our
major deposition method is MBE equipped with RHEED, and an in situ characterization
tool, which is x-ray photoelectron spectroscopy (XPS).
13
Chapter 3 explores the temperature dependence of the morphology and electronic
structure of ultrathin coverage of platinum (Pt) on STO. High and low growth temperatures
applied to Pt results in total different surface morphology and electronic structure
evolution. Some interesting annealing results are also shown.
Chapter 4 describes a series of transition metal depositions on STO [74]. We find
that Eu deposited under UHV condition results an unstable monoxide to form. Previous
work on other metals has been done by former group members [43]. We explore the oxygen
scavenging behavior and manage to epitaxially grow EuO on STO. We vary the STO
thickness and growth temperature to study this interface. Further research is to insert a
BTO film and use angle-resolved photoemission spectroscopy (ARPES) at the Paul Scherer
Institute (PSI), Switzerland, to explore the effect of the ferroelectric BTO layer on the
interfacial 2DEG.
Chapter 5 focuses on the STO/Si pseudo-substrate. We first study the dry oxidation
of buried Si as a function of annealing time and determine the maximum temperature for
which a relatively thin STO layer (10 nm) remains intact. We developed a theoretical model
based on a modification of the Deal–Grove–Massoud formalism that predicts the thickness
of SiO2 formed underneath STO as a function of time and temperature and report a robust
recipe for dry oxidation of Si buried under an epitaxial layer of STO. The following
research is to fabricate a thick SiO2 interlayer on STO/Si template and use selective etch to
remove SiO2 and produce free-standing STO membranes. Free-standing 2D STO may
show different properties than either thin film or bulk STO. In a preliminarily study we
used Raman spectroscopy to characterize STO membranes.
Chapter 6 is our prototype design of BaSnO3/SrTiO3/Al2O3 quantum wells. This
design is based on the large conduction band offset (~3.5 eV) between BSO and Al2O3.
14
The deposition process is done by MBE and characterized by RHEED and XPS to confirm
the film quality.
1.4 REFERENCE
1. B. N. Figgis and J.Lewis, The Magnetochemistry of Complex Compounds in Modern
Co-ordination Chemistry, Ed. J. Lewis and C. Wilkins, Interscience, New York (1960).
2. T. Sakudo, Ferroelectrics, 95, 37 (1989).
3. K. Takada, H. Sakurai, E. Takayama-Muromachi, F. Izumi, R. A. Dilanian, and T.
Sasaki, Nature, 422, 53 (2003).
4. Y. Kamihara, H. Hiramatsu, M. Hirano, R. Kawamura, H. Yanagi, T. Kamiya, and H.
Hosono, J. Am. Chem. Soc. 128, 10012 (2006).
5. K. A. Kovacs, P. Grof, L. Burai, M. Riedel. J. Phys. Chem. A. 108, 11026 (2004).
6. H. N. Lee, H. M. Christen, M. F. Chisholm, C. M. Rouleau, and D. H. Lowndes. Nature,
433, 395 (2005).
7. S. W. Cheong and M. Mostovoy. Nat. Mater. 6, 13 (2007).
8. M. Mostovoy, Nat. Mater. 7, 269 (2008).
9. S. Maekawa, T. Tohyama, S. E. Barnes, S. Ishihara, W. Koshibae, and G. Khaliullin,
Physics of Transition Metal Oxides, (Springer, 2004).
10. A. A. Demkov and A. B. Posadas, Integration of Functional Oxides with
Semiconductors (Springer New York, 2014).
11. N. C. Bristowe, J. Varignon, D. Fontaine, E. Bousquet and Ph. Ghosez, Nat. Comm. 6,
6677 (2015)
12. J. A. Kittl, K. Opsomer, M. Popovici, N. Menou, B. Kaczer, X. P. Wang, C. Adelmann,
M. A. Pawlak, K. Tomida, A. Rothschild, B. Govoreanu, R. Degraeve, M. Schaekers,
M. Zahid, A. Delabie, J. Meersschaut, W. Polspoel, S. Clima, G. Pourtois, W. Knaepen,
C. Detavernier, V. V. Afanas’ev, T. Blomberg, D. Pierreux, J. Swerts, P. Fischer, J. W.
Maes, D. Manger, W. Vandervorst, T. Conard, A. Franquet, P. Favia, H. Bender, B.
Brijs, S. Van Elshocht, M. Jurczak, J. Van Houdt, D. J. Wouters, Microelectron. Engin.
86, 1789 (2009).
13. M. Imada, A. Fujimori, Y. Tokura, Rev. Mod. Phys. 70, 1039 (1998).
14. H. Y. Hwang, Y. Iwasa, M. Kawasaki, B. Keimer, N. Nagaosa, and Y. Tokura, Nat.
Mater. 11, 103 (2012).
15. J. Chakhalian, J. W. Freeland, A. J. Millis, C. Panagopoulos, and J. M. Rondinelli, Rev.
Mod. Phys. 86, 1189 (2014).
15
16. N. Sai, B. Meyer, and D. Vanderbilt, Phys. Rev. Lett. 84, 5636 (2000).
17. H. Yamada, M. Kawasaki, Y. Ogawa, and Y. Tokura, Appl. Phys. Lett. 81, 4793 (2002).
18. N. Reyren, S. Thiel, A. D. Caviglia, L. F. Kourkoutis, G. Hammerl, C. Richter, C. W.
Schneider, T. Kopp, A.-S. Rüetschi, D. Jaccard, M. Gabay, D. A. Muller, J.-M.
Triscone, and J. Mannhart, Science. 317, 5842 (2007).
19. L. Li, C. Richter, J. Mannhart, and R. C. Ashoori, Nat. Phys. 7, 762 (2011).
20. J. A. Bert, B. Kalisky, C. Bell, M. Kim, Y. Hikita, H. Y. Hwang, and K. A. Moler, Nat.
Phys. 7. 767(2011).
21. A. Brinkman, M. Huijben, M. van Zalk, J. Huijben, U. Zeitler, J. C. Maan, W. G. van
der Wiel, G. Rijnders, D. H. A. Blank, and H. Hilgenkamp, Nat. Mater. 6, 493 (2007).
22. B. Kalisky, J. A. Bert, B. B. Klopfer, C. Bell, H. K. Sato, M. Hosoda, Y. Hikita, H. Y.
Hwang, and K. A. Moler. Nat. Commun. 3, 922 (2012).
23. H. N. Lee, H. M. Christen, M. F. Chisholm, C. M. Rouleau, and D. H. Lowndes, Nature,
433, 395 (2005).
24. D. J. Frank, R. H. Dennard, E. Nowak, P. M. Solomon, Y. Taur and Hon-Sum Philip
Wong, Proceedings of the IEEE. 89, 259 (2001).
25. S. Abel, J. Fompeyrine, “Electro-Optically Active Oxides on Silicon for Photonics” in
Thin Films on Silicon (V. Narayanan, M. Frank and A. A. Demkov. eds.) 455 (2016).
26. M. Hochberg and T. Baehr-Jones, Nat. Photon. 4, 492(2010).
27. G. T. Reed, G. Mashanovich, F. Y. Gardes and D. J. Thomson, Nat. Photon. 4, 518
(2010).
28. B. G. Streetman, S. Banerjee, Solid State electronic Devices, 524 (Hall, New Jersey,
2000).
29. J. D. Plummer, M. D. Deal, P. B. Griffin, Silicon VLSI Technology, (Prentice Hall,
Upper Saddle River, 2013).
30. E. Vella, F. Messina, M. Cannas, R. Boscaino, Phys. Rev. B. 83, 174201 (2011).
31. B. El-Kareh, Fundamentals of Semiconductor Processing Technologies, (Springer US,
New York, 1995).
32. E. Harari, J. Appl. Phys. 49, 2478-2489 (1978).
33. H. Iwai, Microelectron. Eng. 86, 1520 (2009).
34. K. Mistry, C. Allen, C. Auth, B. Beattie, D. Bergstrom, M. Bost, M. Brazier, M.
Buehler, A. Cappellani, R. Chau, C.-H. Choi, G. Ding, K. Fischer, T. Ghani, R. Grover,
W. Han, D. Hanken, M. Hattendorf, J. He, J. Hicks, R. Heussner, D. Ingerly, P. Jain,
R. James, L. Jong, S. Joshi, C. Kenyon, K. Kuhn, K. Lee, H. Liu, J. Maiz, B. McIntyre,
16
P. Moon, J. Neirynck, S. Pae, C. Parker, D. Parsons, C. Prasad, L. Pipes, M. Prince, P.
Ranade, T. Reynolds, J. Sandford, L. Shifren, J. Sebastian, J. Seiple, D. Simon, S.
Sivakumar, P. Smith, C. Thomas, T. Troeger, P. Vandervoorn, S. Williams, K.
Zawadzki, Int. El. Devices Meet. 247 (2007).
35. M. T. Bohr, R. S. Chau, T. Ghani, K. Mistry, IEEE Spectrum. 44, 29 (2007).
36. B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti and A. Kis. Nature Nanotech
6, 147(2011).
37. L. Liao, Y. Lin, M. Bao, R. Cheng, J. Bai, Y. Liu, Y. Qu, K. L. Wang, Y. Huang and
X. Duan. Nature 467, 305 (2010).
38. D. Hisamoto, W. Lee, J. Kedzierski, H. Takeuchi, K. Asano, C. Kuo, E. Anderson, T.
King, J. Bokor and C. Hu. IEEE Transactions on Electron Devices. 47 2320 (2000).
39. N. Singh, A. Agarwal, L. K. Bera, T. Y. Liow, R. Yang, S. C. Rustagi, C. H. Tung,
R. Kumar, G. Q. Lo, N. Balasubramanian, and D. L. Kwong. IEEE Electron Device
Letters, 27, 383 (2006)
40. R. Konta, T. Ishii, H. Kato, and A. Kudo, J. Phys. Chem. B. 108, 8992 (2004).
41. K. Ueno, S. Nakamura, H. Shimotani, A. Ohtomo, N. Kimura, T. Nojima, H. Aoki, Y.
Iwasa, and M. Kawasaki, Nat. Mater. 7, 855 (2008).
42. J. H. Haeni, P. Irvin, W. Chang, R. Uecker, P. Reiche, Y. L. Li, S. Choudhury, W. Tian,
M. E. Hawley, B. Craigo, A. K. Tagantsev, X. Q. Pan, S. K. Streiffer, L. Q. Chen, S.
W. Kirchoefer, J. Levy and D. G. Schlom. Nature 430, 758 (2004).
43. K. J. Kormondy, L. Gao, X. Li, S. Lu, A. B. Posadas, S. Shen, M. Tsoi, M. R.
McCartney, D. J. Smith, J. Zhou, L. L. Lev, M.-A. Husanu, V. N. Strocov, and A. A.
Demkov, Sci. Rep. 8, 7721 (2018).
44. W. S. Choi, D. W. Jeong, S. S. A. Seo, Y. S. Lee, T. H. Kim, S. Y. Jang, H. N. Lee,
and K. Myung-Whun, Phys. Rev. B. 83, 195113 (2011).
45. R. A. McKee, F. J. Walker, and M. F. Chisholm, Phys. Rev. Lett. 81, 3014(1998).
46. J. Ramdani, R. Droopad, Z. Yu, J. A. Curless, C. D. Overgaard, J. Finder,K.
Eisenbeiser, J. A. Hallmark, W. J. Ooms, V. Kaushik, P. Alluri, and S. Pietambaram,
Appl. Surf. Sci. 159, 127 (2000).
47. R. A. McKee, F. J. Walker, and M. F. Chisholm, Science 293, 468 (2001).
48. X. Hu H. Li, Y. Liang, Y. Wei, Z. Yu, D. Marshall, J. Edwards, R.Droopad, X. Zhang,
A. A. Demkov, K. Moore, and J. Kulik, Appl. Phys.Lett. 82, 203 (2003).
49. X. Zhang, A. A. Demkov, H. Li, X. Hu, Y. Wei, and J. Kulik, Phys. Rev.B. 68, 125323
(2003).
17
50. Y. Wei, X. Hu, Y. Liang, D. Jordan, B. Craigo, R. Droopad, Z. Yu, A. A. Demkov, J.
Edwards, Jr., and W. Ooms, J. Vac. Sci. Technol. B 20, 1402(2002).
51. K. D. Fredrickson, H. Seo and A. A. Demkov, J. Appl. Phys. 120, 065301 (2016).
52. M. Choi, A. Posadas, R. Dargis, C.-K. Shih, A.A. Demkov, D.H. Triyoso, N.D.
Theodore, C. Dubourdieu, J. Bruley, and J. Jordan-Sweet, J. Appl. Phys. 111, 064112
(2012).
53. K. J. Kormondy, S. Abel, F. Fallegger, Y. Popoff, P. Ponath, A. B. Posadas, M.
Sousa, D. Caimi, H. Siegwart, E. Uccelli, L. Czornomaz, C.Marchiori, J. Fompeyrine,
A. A. Demkov. Microelectron. Eng. 147, 215 (2015).
54. L. Zhang, Y. Yuan, J. Lapano, M. Brahlek, S. Lei, B. Kabius, V. Gopalan, and R.
Engel-Herbert, ACS Nano, 12, 1306(2018)
55. A. Ohtomo and H. Y. Hwang, “A high-mobility electron gas at the LaAlO3/SrTiO3
heterointerface,” Nature, 427, 423 (2004).
56. J. Robertson, J. Vac. Sci. Technol. B. 18. 1785(2000).
57. S.G. Lim, S. Kriventsov, T. N. Jackson, J. H. Haeni, D. G. Schlom, A. M. Balbashov,
R. Uecker, P. Reiche, J. L. Freeouf, and G. Lucovsky, J. Appl. Phys. 91, 4500 (2002).
58. R. Ramesh and D. G. Schlom, MRS Bull. 33, 1006(2008).
59. D. F. Bogorin, P. Irvin, C. Cen, and J. Levy, “LaAlO3/SrTiO3-based device concepts,”
in Multifunctional Oxide Heterostructures, Oxford University Press, 364(2010).
60. L. Bjaalie, B. Himmetoglu, L. Weston, A. Janotti, and C. G. Van de Walle. New J.
Phys. 16, 025005 (2014).
61. J. E. Ortmann, N. Nookala, Q. He, L. Gao, C. Lin, A. B. Posadas, A. Y. Borisevich, M.
A. Belkin, and A. A. Demkov, ACS NANO 12, 7682 (2018).
62. J. E. Ortmann, A. B. Posadas and A. A. Demkov, J. Appl. Phys. 124, 015301 (2018).
63. J. E. Ortmann, M. A. Duncan and A. A. Demkov, Optical Materials Express 9, 2982
(2019).
64. J. E. Ortmann, S. Kwon, A. B. Posadas, M. J. Kim and A. A. Demkov, J. Appl. Phys.
125, 155302 (2019).
65. N. Nakagawa, H. Y. Hwang, and D. A. Muller. Nat. Mater. 5, 204 (2006).
66. F. Pockels, Lehrbuch Der Kristalloptik (Leipzig und Berlin, 1906).
67. B. W. Wessels, Annu. Rev. Mater. Res. 37, 659 (2007).
68. S. Abel, T. Stöferle, C. Marchiori, C. Rossel, M.D. Rossell, R. Erni, D. Caimi, M.
Sousa, A. Chelnokov, B.J. Offrein, J. Fompeyrine, Nat. Commun. 4, 1671(2013).
69. D. P. Kumah, J. H. Ngai and L. Kornblum, Adv. Funct. Mater. 30, 1901597(2019)
18
70. L. Mazet, S. M. Yang, S. V. Kalinin, S. Schamm-Chardon and C. Dubourdieu, Sci.
Technol. Adv. Mater. 16, 036005 (2015).
71. S.Abel, F. Eltes, J. E. Ortmann, A. Messner, P. Castera, T. Wagner, D. Urbonas, A.
Rosa, A. M. Gutierrez, D. Tulli, P. Ma, B. Baeuerle, A. Josten, W. Heni, D. Caimi, L.
Czornomaz, A. A. Demkov, J. Leuthold, P. Sanchis. J. Fompeyrine. Nature Material,
18, 42 (2019).
72. W. Guo, A. B. Posadas, A. A. Demkov, J. Vac. Sci. Technol. A 39, 030804 (2021)
73. R. C. Alferness, IEEE Transactions on Microwave Theory and Techniques, 82,
1121(1982).
74. A. B. Posadas, K. J. Kormondy, W. Guo, P. Ponath, J. Geler-Kremer, T. Hadamek,
and A. A. Demkov, J. Appl. Phys. 121, 105302 (2017)
19
Chapter 2: Experiment techniques
I will now introduce the principles and instrument details of the equipment I have
used in my research. The Advanced Atomic Design Lab has an MBE reactor equipped with
RHEED and an in situ XPS. A schematic of our system is shown in Figure 2.1. I will
explain these in detail in this chapter.
Figure 2.1: Schematic of the vacuum system at the Advanced Atomic Design Lab. All
instruments are linked to the main transfer line.
20
2.1 MOLECULAR BEAM EPITAXY (MBE)
Molecular beam epitaxy (MBE) is a physical vapor deposition process that deposits
high quality thin films epitaxially under ultra-high vacuum (UHV). This is a widely used
method both in the semiconductor industry and in research labs. Günther et al. invented
the principal idea of MBE in 1958 [1]. It attracted much attention and enjoyed fast
development since its invention due to the high demand for high quality III-V
semiconductor devices [2]. Cho and Arthur et al. lead the work at Bell Laboratories and
developed MBE in the late 1960s [3].
2.1.1 Operating principles and instrument illustration
The basic principle of MBE is to evaporate high-purity materials under UHV
environment and generate molecular beams going straight to the substrate wafers. Under
appropriate conditions (temperature, flux, crystalline orientation, etc.), gaseous species
arriving on the wafer will condense and react with each other to form an epitaxial film. The
major advantage of MBE over other deposition tools is the precise control of the deposition
process with relatively low flux rate, benefiting from the UHV environment. MBE can
achieve sub-monolayer deposition control, which allows one to have more control over
film properties and allow for tuning the band gap, lattice constant or crystalline orientation.
Our lab is equipped with a custom-built DCA M600 Oxide-MBE. The major
components are listed below and I label them on a picture of the chamber in Figure 2.1.
21
Figure 2.2: Front view of MBE with components labeled.
22
Figure 2.3: Back view of MBE with components labeled
23
(1) Cryopump, pressure gauges and residual gas composition analyzer
(2) Cryopanel and water cooling system
(3) Manipulator with thermal control and sample transfer
(4) Effusion cells and e-beam evaporator.
(5) Quartz crystal monitor (QCM) for flux measurement.
(6) Oxygen source with plasma generator.
(7) RHEED system for monitoring the surface crystallinity.
2.2.2 Preparation for MBE growth
Samples need pre-treatment to be suitable for a vacuum process. The preparation
usually consists of: 1) ultrasonic degrease in acetone, isopropanol and deionized (18
MΩ/cm) water for 5-10 minutes each to remove surface grease and dust as much as
possible, 2) ozone treatment for 15-30 min to remove surface residual carbon.
These steps are very basic procedures for sample preparation. Different substrates
require different additional treatments. For example, the TiO2-terminated STO surface [4]
requires boiling water soaking of the sample and up to 4 hours oxygen anneal to form
atomically flat TiO2-terminated surface; A clean Ge surface requires ex situ wet etching
using HCl as the etchant and H2O2 as the oxidant [5].
After the sample is well-cleaned and treated, it is loaded into the load lock of the
vacuum system and pumped down to the same vacuum level as the system. It is then
transferred to other chambers for the subsequent experiments.
24
2.2.3 Maintaining vacuum and cryopump
The vacuum level in the MBE main chamber is within the ultra-high vacuum
(UHV) regime, meaning pressures ~10-9 Torr or lower. Our MBE has a base pressure of
3×10-10 Torr when all cells are cold and the chamber is well-baked out.
Maintaining vacuum is crucial for the operation of an MBE as it determines the
level of atom flux control one can achieve. The mean free path of a gaseous species is the
distance it travels on average without collisions with other atoms or molecules of the gas.
Only a sufficiently high vacuum level will guarantee that the molecular flux will not be
interrupted or scattered while it travels from the effusion cells to the wafer. This usually
requires the mean free path to be higher than 0.5-1 m for MBE. Under normal atmospheric
pressure (760 Torr), the typical mean free path is ~100 nm. In a UHV environment with
10-10 Torr, one obtains a ~100 km mean free path. At first glance, it seems wasteful to
maintain UHV instead of a lower level of vacuum, but the growth process has a lot of
heating, gas injection and other steps, all of which will decrease the vacuum level of the
chamber. A typical oxide MBE operation vacuum level with hot effusion cells and oxygen
flow provided is in the high vacuum regime (8×10-4 ~8×10-8 Torr). This vacuum level
yields mean free paths in the tens of cm to 1 km range, so we can still be sure the mean
free path longer than the distance from the cells to the substrate, as long as we maintain the
vacuum level below ~5×10-5 Torr.
Our MBE system uses a cryogenic pump to maintain UHV. A cryogenic pump
works by having a large area cold surface connected to the vacuum chamber to condense
and adsorb gaseous species. This cold surface is cooled by compressed helium refrigeration
resulting in temperatures of 10-15 K temperature enabling one to condense most gases,
effectively removing them from the chamber to maintain UHV. The walls of the MBE
25
chamber are shielded with a cryo-panel that circulates chilled ethanol. The cooling of the
stainless-steel chamber walls helps keep the pressures low during operation when sources
are hot by preventing adsorbed atoms and molecules from re-evaporating.
2.2.4 Effusion cells
Effusion cells (also called Knudsen cells) are miniature furnaces with heating coils
and inner crucibles. High purity metals are loaded in the crucible and evaporated by heating
the crucible to a sufficiently high temperature, which depends on the material (up to ~1800
ºC in the case of Ti). A water-cooling system and radiation shield maintains the exterior
temperature to avoid heating the rest of the reactor during operation. .
Our MBE chamber is equipped with 6 effusion cells, which are of 3 types. For Ti
and La, which have very high melting temperatures, we use high-temperature effusion cells
that allow one to use operating temperatures of 1400 ºC - 1800 ºC. For Ba, Sr and Eu, low-
temperature effusion cells with pyrolytic boron nitride (PBN) crucibles are enough for
operating temperatures below 1000 ºC. Metallic Al requires a special cold-lip cell to prevent
liquid Al from creeping out of the crucible.
All the cells have compressed air-controlled shutters that serve as “caps”. When
closed, shutters physically block line of sight from the effusion cell to the substrate,
effectively stopping any deposition from that cell when that element is not needed during
the growth.
2.2.5 Electron beam evaporator
For those metals with extremely high melting temperatures, which exceed the
temperature range of effusion cells, the electron beam evaporator or so-called E-gun is
used.
26
A beam of electrons from a tungsten filament is focused and accelerated (~7.75 kV)
towards the target crucibles by electrostatic and magnetic fields. The electron beam will
heat the metal in the crucible, melting and evaporating it. The MBE in our lab is equipped
with a 4-pocket E-gun. A viewport located above the pockets (see Fig. 2.3) allows one to
monitor them and to adjust the electron beam position and sweep on the metal in the pocket
for optimum coverage.
2.2.6 Quartz crystal monitor (QCM)
QCM is the tool we use to measure the atomic or molecular beam flux. It is located
right below the manipulator and substrate position, to yield flux readings as close as
possible to the flux arriving at the substrate. The QCM determines the flux by measuring
the accumulated film thickness deposited on a quartz crystal [6]. The operating principle
of the sensor is that deposited film on the sensor crystal will cause a mass change and cause
a certain shift in the resonance frequency of the quartz oscillator [7]. We can then determine
the film thickness that accumulates in a certain time interval, from which we can calculate
the flux.
Flux control is a very important factor in MBE deposition, since the stoichiometry
of the film and hence the film quality is largely determined by it.
2.2 REFLECTION HIGH-ENERGY ELECTRON DIFFRACTION (RHEED)
Reflection high-energy electron diffraction (RHEED) is a surface diffraction
technique that is commonly used in deposition tools, especially MBE, to monitor the
surface crystallinity during growth. The principle of RHEED is to use a beam of highly
energetic electrons that strike the sample at a very small angle relative to its surface. The
27
surface crystal structure of the sample diffracts the incident electrons, causing an
interference pattern to form, which can be seen on the phosphor screen.
A schematic of principle of RHEED is shown in Figure 2.4.
Figure 2.4: Schematic of the principle of RHEED. The angle of incidence of the beam is
usually less than 5º from the surface.
In Figure 2.4, the incident beam is diffracted by the crystal structure of the sample.
Due to the small incident angle and the high energy of the incident electrons, the
penetration depth of the beam is only several atomic layers and only atoms near the top
surface will contribute to the diffraction. An infinite 3D lattice in real space corresponds to
a 3D array of points in reciprocal space. Considering just the ideal surface as a two-
dimensional lattice, this translates to reciprocal space as a 2D array of reciprocal lattice
rods. When these reciprocal rods intersect the Ewald sphere corresponding to the electron
energy, it produces a diffraction pattern that is projected onto a phosphor screen. A charge-
coupled device (CCD) camera is then used to capture the image on the screen and send it to
28
the software. Our manipulator can rotate the sample so we can change the orientation of the
surface and get diffraction patterns from different crystallographic directions like [100], [111]
or [110]. A more detailed momentum and energy analysis and discussion of the working
principles can be found elsewhere [8], [9].
RHEED images taken from real samples are shown in Figure 2.5. Here I present the
RHEED patterns for a commercial STO single crystal substrate and for an MBE-grown STO
thin film on Si.
Figure 2.5: RHEED images of (a) commercial STO substrate bought from MTI Crystal
(b) 5 uc of STO grown on a Si substrate.
In Figure 2.5 (a), the pattern consists of sharp dots sitting along a circular arc, which
is the ideal diffraction pattern from an atomically flat surface with large coherence length.
The pattern of dots is determined by the intersection of the reciprocal rods and the Ewald
sphere. Atomically flat surfaces typically have narrow reciprocal lattice rods such that the
diffraction spots are smaller and shaper. In Figure 2.5 (b), the pattern has elongated into
streaks and is not as sharp as in (a). This shows a surface that is not as flat as that of a
commercially bought substrate but still relatively flat. Due to disorder and smaller
29
coherence length due to uncorrelated surface steps, the reciprocal rods are wider and have
finite length, resulting in this slightly less sharp streaky pattern.
RHEED is the most important monitoring tool in MBE deposition. We can adjust
the growth conditions (flux and growth temperature) in real-time to ensure high film quality
using RHEED. For example, the element flux does drift especially for long time
depositions and the crystal quality of the top surface can get worse with time. RHEED is
very sensitive to surface quality changes and we can make the corresponding adjustment
to the flux to maintain the right stoichiometry.
2.3 X-RAY PHOTOELECTRON SPECTROSCOPY (XPS)
X-ray photoelectron spectroscopy (XPS) is a powerful characterization tool for
identifying existing elements in samples and their chemical state, composition, electronic
structure and density of the electronic states. The principle of XPS is based on the
photoelectric effect explained by Einstein in the early 1900s for which he received a Nobel
Prize [10]. Our lab has an in situ VG Scienta R3000 XPS system connected to the common
transfer line along with the MBE. We can measure the samples after growth without
breaking the vacuum and avoid possible sample reaction with air or other contaminants. A
front view picture of our XPS system is shown in Figure 2.6 with the main components
labeled.
30
Figure 2.6: Picture of the VG Scienta R3000 XPS system. (Front view)
2.3.1 Operating principles
The photoelectric effect happens when electromagnetic radiation is absorbed by a
material. The electrons in the atoms of the material absorb the energy of the radiation and
are ejected from the material if the photon energy is high enough to overcome the binding
energy due to the atomic nucleus. The binding energy of the emitted electron is determined
by the following equation: Ebinding + ϕ = Ephoton - Ekinetic. Here ϕ is the work function, which
is the minimum energy needed to remove an electron from the material, and Ekinetic is the
measured kinetic energy of the emitted electron.
31
In laboratory XPS systems, Ephoton is usually a fixed value corresponding to the
energy emitted by the x-ray source of the system. By measuring the kinetic energy of the
emitted electrons (and correcting for ϕ), we can calculate the binding energy of emitted
electrons. The electron energy analyzer and detector allows one to get a count of how many
electron are emitted at a given energy, producing an energy spectrum. Figure 2.7 shows
the schematic of a conventional XPS instrument.
Figure 2.7: Conventional X-ray source for an XPS instrument. (Figure adopted from [11])
X-rays are generated from the x-ray source, which is then usually monochromated
and focused onto the sample. Electrostatic focusing lenses then help to collect the emitted
electrons from a range of angles then send them through a hemispherical analyzer to filter
electrons by kinetic energy and then count them.
32
2.3.2 Surface sensitivity
XPS usually requires UHV conditions, especially for surfaces sensitive to
oxidation. Recalling that the vacuum level determines the mean free path, UHV conditions
will guarantee that: 1) the emitted electrons are not scattered by gas molecules along the
path from sample to detector; 2) attenuation and distortion of the spectra by surface
contamination is avoided [11]. High vacuum conditions (<10-5 Torr) usually satisfies
condition 1, but the sample surface will also adsorb the residual gas molecules at these
pressures in minutes and the contamination cannot be readily ignored under such vacuum
conditions. The surface contamination rate under high vacuum can be as fast as 1 s/ML,
meaning it only takes 1 second to accumulate 1 ML of residual molecules, assuming unity
sticking coefficient. Under UHV conditions, the contamination rate is much slower – about
104 s/ML in UHV [11]. This is a few times longer than a typical set of XPS scans thus
ensuring the sample surface remains contamination-free during the scans.
2.3.4 XPS peak analysis
Here we explain a typical peak analysis of an XPS experiment. We will see many
results of the analysis described here in the following chapters.
Our XPS system uses an Al Kα X-ray source with a photon energy of 1486.6 eV.
When we scan through the whole binding energy range from 0 eV to 1486.6 eV, the emitted
electrons will come from all the core levels from all the elements in the material (except
for H and He), as well as from the valence band. Their intensities (normalized with
photoionization cross-section) reflect the composition of the material. Characteristic core
level peaks are usually the strongest peaks in a pure element spectrum. We analyze these
characteristic core levels to determine the electronic structure and oxidation state of that
element in the sample.
33
With the exception of 1s spectra, core level spectra are doublets corresponding to
different total angular momentum quantum numbers. This arises from spin-orbit splitting.
For example, a 3d core level has n = 3, l = 2 and s = 1
2 . The total angular momentum j =
l ± s will then be 3/2 or 5/2. Since the relative peak intensities of doublets are determined
by the ratio of their degeneracy 2j+1 [11], the theoretical intensity ratio of the doublets is
2:3 for a 3d core level.
2.4 REFERENCES
1. K. G. Günther, Z. Naturforschg. A. 13 (12), 1081-1089 (1958).
2. A. Y. Cho, F. K. Reinhart, Appl. Phys. Lett. 21, 355 (1972).
3. J. R. Arthur Jr., J. Appl. Phys. 39, 4032 (1968).
4. R. C. Hatch, M. Choi, A. B. Posadas, and A. A. Demkov, J. Vac. Sci. Technol. B. 33,
061204 (2015).
5. P. Ponath, A. B. Posadas, R. C. Hatch and Alexander A. Demkov, J. Vac. Sci. Technol.
B. 31, 031201 (2013).
6. J.E. Mahan, Physical Vapor Deposition of Thin Films, Wiley-VCH, 2000.
7. M. Rodahl, F. Höök, A. Krozer, P. Brzezinski, B. Kasemo, Rev. Sci. Instrum. 66, 3924
(1995).
8. A. Ichimiya, P. I. Cohen, Reflection High Energy Electron Diffraction Cambridge
University Press (2004).
9. Y. Horio, Y. Hashimoto, A. Ichimiya, Appl. Surf. Sci. 100, 292 (1996).
10. A. Einstein, Ann. Phys. 322, 132 (1905).
11. S. Hofmann, Auger- and X-Ray Photoelectron Spectroscopy in Materials Science: A
User-Oriented Guide, Springer Berlin Heidelberg, Berlin, Heidelberg, (2013).
34
Chapter 3: Temperature dependence of the morphology and electronic
structure of ultrathin platinum on TiO2-teminated SrTiO3 (001)
In this chapter, we investigate properties of ultrathin Pt as a function of coverage
(up to 10 monolayers) on TiO2-terminated SrTiO3 (001) substrate at different temperatures
(200–800 ºC). In situ x-ray photoelectron spectroscopy, scanning electron microscopy, and
atomic force microscopy are used to observe the electronic structure and surface
morphology evolution of Pt. The authors find that although Pt will not wet SrTiO3 in the
thermodynamic sense, it forms a continuous film when deposited at 200 ºC due to the low
surface mobility. At 800 ºC, even at very low coverage, Pt forms nanoclusters showing
bulk-like metallic features in the photoemission spectra. We compare the observed
electronic structure evolution of Pt and the different growth patterns at low and high
temperatures with available theoretical calculations.
The contents of this chapter were published in: W. Guo, A. B. Posadas, and A. A.
Demkov, J. Vac. Sci. and Technol. B. 35, 061203 (2017); and A. B. Posadas, K. J.
Kormondy, W. Guo, P. Ponath, J. Geler-Kremer, T. Hadamek, and A. A. Demkov, J. Appl.
Phys. 121, 105302 (2017).
3.1 INTRODUCTION
Platinum (Pt) is a noble transition metal that has excellent chemical and thermal
stability as well as high electrical conductivity (9.4×106 S/m). These make Pt a popular
contact material, especially on oxides, because of its inertness toward oxidation [1]–[4]. In
addition, it is also a very efficient co-catalyst on titania and various other oxide catalysts
for enhancing the reduction half-reaction in solar water splitting to generate hydrogen [5],
[6]. On the other hand, SrTiO3 (STO) is a widely used substrate material in oxide epitaxy.
35
It has the cubic perovskite structure with a lattice constant of 0.3905 nm [7], [8]. In
addition, STO has a rather large dielectric constant (~300) and can readily be made
conductive by doping [8]. When these two materials are placed in contact, a Schottky
barrier (SB) with values ranging from 0.6 to 1.3 eV is formed at the interface, with upward
band bending in the STO [1], [3], [9]. Pt has a very high work function of ~5.7 eV [10],
while the electron affinity of STO is ~3.9 eV [8]. With a lattice parameter of 0.3923 nm
for Pt and 0.3905 nm for STO and both materials having a cubic structure, Pt and STO
appear to be well lattice-matched and suitable for epitaxial growth. However, it turns out
that Pt (001) does not normally grow epitaxially on STO (001) and instead prefers to grow
in the (111) orientation even on the lattice-matched STO (001) TiO2-terminated surface
[11].
There are a number of metals such as Al, Ti, Eu and Nb that can scavenge oxygen
from STO to form oxygen vacancies even at moderate temperature (200–400 ºC) and under
relatively high oxygen pressure (up to 10-2 Torr) [4]. However, Pt is one of the small set of
metals that has good stability to oxidation when deposited on STO. Through a combination
of the low oxide enthalpy of formation and high work function of Pt, electrons prefer to
transfer from STO into the metal, and hence STO is not reduced. This makes Pt on STO an
ideal system to study the formation of the initial Schottky barrier without the complications
of surface reactions.
Although there are many papers discussing the Pt/STO heterostructure, particularly
where Pt is used as an electrode [1], they mostly focus on micrometer-thick Pt films. There
are not as many studies about the first few monolayers (MLs) of Pt and its development
into a fully metallic layer under different growth temperatures. Chung and Weissbard
showed that the SB formation is a gradual process during the Pt accumulation in the first
36
monolayer, accompanied by charge transfer from STO to Pt [12]. They also reported that
the Pt core level binding energy (BE) shifts to lower values as the coverage of Pt increases
because of gradual charge transfer from STO to Pt [13]. The same type of a SB forms at
the Pt/BaTiO3/(Ba,Sr)TiO3 interface and a similar shift of the Pt core level has also been
reported for this system [3], [14]. Copel et al. measured the band bending in STO when Pt
is deposited on it and described the metallicity of Pt as it accumulates on STO. Theoretical
calculations using density functional theory (DFT) have been performed to study the SB
formation at the Pt/STO interface. [11], [15], [16]. Seo and Demkov explored the growth
of Pt on STO within the first ML using DFT and reported the electronic density of states
for different Pt coverage [11]. They suggested that Pt nanoclusters will start to form from
submonolayer coverage, which can be verified by measuring the core level binding energy
of Pt. By performing thin Pt layer by layer growth at different temperatures, we can verify
this theoretical result and obtain more understanding about the Schottky barrier formation,
electronic structure evolution, and surface morphology of Pt.
The surface morphology of Pt on STO has also been recently reported [17]-[19].
Polli et al. grew 20 nm Pt by molecular beam epitaxy (MBE) on differently terminated
STO substrates and showed by atomic force microscopy (AFM) and conventional
transmission electron microscopy [18] that the TiO2 termination resulted in a more uniform
and ordered orientation and size distribution of Pt clusters. Christensen et al. showed
scanning electron microscopy (SEM) images of Pt nanoparticles [19]. They point out that
with a Pt average coverage of 1 ML, the clusters aggregate into nanoparticles without
coalescing.
In this chapter, we present a systematic study of depositing 0.25, 0.5, 1, 2, 4, and
10 ML of Pt metal on a TiO2-terminated (001) STO surface in ultrahigh vacuum at
37
temperatures ranging from 200 to 800 ºC, and investigate the evolution of the electronic
structure by in situ x-ray photoelectron spectroscopy (XPS). We also use SEM and AFM
to investigate the morphology of the surface.
3.2 EXPERIMENT
The deposition is performed in a customized DCA Instruments M600 MBE
chamber with base pressure below 6 × 10-10 Torr. Nb:STO (001) (0.7 wt. % doped)
substrates of 5 × 5 mm2 from MTI Crystal are treated using the water boiling process to
yield TiO2-terminated surfaces [20]. The substrates are degreased ultrasonically in acetone,
isopropanol, and deionized (18 MΩ/cm) water for 5 min each, then soaked in boiling
deionized water for half an hour to dissolve excess surface SrO. The substrate is then
annealed in flowing oxygen for 4 h in a tube furnace at 950 ºC to form a well-ordered TiO2-
terminated surface. This is then followed by a 725 ºC anneal in the vacuum chamber for 1
h to remove volatile surface contaminants from air exposure. This process results in a single
TiO2-terminated surface with extremely low surface defect density [20]. Pt is evaporated
by an electron beam source operated at an emission current of ~110mA with an electron
energy of 7.75 keV. A quartz crystal monitor is used to measure the deposition rate of Pt.
The rate is kept around 0.1 nm/ min (~0.5 ML of Pt per minute) by adjusting the emission
current of the electron beam source. Reflection-high-energy electron diffraction with 21
keV electrons is used to monitor the surface quality and the orientation of Pt.
Pt deposition is commenced after the substrate temperature reaches the desired
value. An amount of Pt equivalent to a cumulative coverage of 0.25, 0.5, 1, 2, 4, and 10
ML is deposited on the substrate. After each target coverage is deposited, the sample is
cooled down and transferred in situ to the XPS chamber to measure the core level and
38
valence band spectra. The sample is then transferred back to the growth chamber for the
next deposition, and so on. Our XPS system uses monochromated Al Kα radiation as the
x-ray source and a VG Scienta R3000 hemispherical electron energy analyzer. The core
level spectra of Pt 4f, Sr 3d, Ti 2p, and O 1s were scanned for each coverage. A Hitachi
S5500 SEM is used to image the surface of samples grown at 200 and 800 ºC to determine
the effect of temperature on surface morphology. An Asylum MFP-3D atomic force
microscope (AFM) is used to image and quantify the surface morphology of the films.
3.3 RESULTS AND DISCUSSION
In Figure 3.1, we show the Pt 4f core level binding energy as a function of coverage
for deposition at 200 ºC. The Pt 4f7/2 peak is at ~72.4 eV at a Pt coverage of 0.25 ML and
gradually shifts to ~71.3 eV after a coverage of 2 ML is reached. This indicates a transition
of the Pt from a “partially oxidized” state to a fully metallic state. This result is consistent
with recent DFT calculations [11] that pointed out that, initially, Pt atoms on the TiO2-
terminated STO surface are located at the “hollow” site surrounded by four O atoms and
forming Pt–O bonds. This results in a partially oxidized state for the initial Pt causing the
core level to be shifted to higher BE. As more Pt is added, the Pt–Pt bonds gradually start
to become dominant. Finally, the fully metallic Pt layer forms and the BE corresponds to
that in bulk Pt metal. This transition has also been observed by Chung and coworkers [12],
[13].
39
Figure 3.1: Pt 4f core level spectrum for deposition at 200 ºC on STO as a function of Pt
thickness. At 0.25 ML coverage (the lowest) the Pt 4f7/2 peak has a binding
energy of 72.4 eV. It gradually shifts by 1.1 eV to the bulk metallic value
(71.3 eV) when Pt reaches 2 ML.
When examining the BE change as a function of deposition temperature, we find
the behavior shown in Figure 3.2. When depositing Pt at 200 ºC, the total 4f7/2 BE shift
with thickness, going from 0.25 to 2 ML of Pt, is about 1.0 eV. But when we increase the
growth temperature, the total shift is reduced, becoming less than 0.5 eV when depositing
at 800 ºC. This trend can be understood a follows. At lower temperature, Pt is less likely
to move away from the energetically preferred oxygen hollow site [11], [21]. At higher
temperature, however, as the surface mobility increases, Pt will form particles or clusters
more and more easily with increasing coverage as it does not wet STO [16]. At low
temperature, initially, Pt atoms are bonded to oxygen, with the BE of this partially oxidized
40
metal shifted significantly from that of the metal. This results in a large energy shift when
plotted as a function of coverage. As the temperature goes up, Pt atoms gain enough energy
to move across the surface and form metal clusters. Thus, even at low coverage the BE is
closer to the bulk metallic value. Thus, for the same Pt coverage, higher growth temperature
results in a lower BE value. The BE of Pt 4f7/2 at 2 ML coverage grown at 800 ºC almost
reaches the bulk metallic value, while it takes several more layers for the lower deposition
temperatures. At this coverage, the deposited Pt is not yet the same as the bulk Pt BE
because the Pt forms as nanoclusters (or nanoparticles). This effect is commonly explained
as the incomplete screening of metal clusters [22], with smaller metal clusters having larger
BE shifts. We have also measured the core level spectra of Sr 3d, Ti 2p, and O 1s. Their
binding energies as a function of Pt coverage are shown in Figure 3.3. All of them show a
uniform systematic shift of about 0.3 eV to lower binding energy while Pt is accumulating.
This is consistent with the shift being caused by upward band bending in STO. We find
that a Schottky barrier of ~0.65 eV is built up gradually as Pt is accumulating.
41
Figure 3.2: Pt 4f7/2 binding energy as a function of Pt thickness for different deposition
temperatures (200–800 ºC). All values shift to the bulk Pt metal value
(~71.3 eV) at different temperatures when Pt is sufficiently thick. Lower
deposition temperature will lead to larger total binding energy shift while
also taking more layers to reach the bulk metal value.
Figure 3.3: Sr 3d, Ti 2p, O 1s binding energy shift as a function of Pt thickness. There is
a ~0.3 eV shift for all three core levels while Pt is accumulating, which is
consistent with an upward band bending.
A high-count scan of the valence band spectrum as a function of coverage grown
at 800 ºC is shown in Figure 3.4. It can be seen that the Pt signal emerges gradually in the
region around 0–3 eV as the film thickness increases to 2 ML, indicating the presence of
individual Pt states at low coverage. At 2 ML, we see significant nonzero intensity at the
Fermi level and the onset of a metal-like valence band shape. This means that at 2 ML
42
coverage, Pt clusters are big enough to have a Fermi edge. We note that the growth
temperature also affects the size of Pt clusters, with higher temperatures yielding bigger
clusters due to the increased surface mobility. We use AFM to analyze the surface structure
of 2 ML Pt grown at 800 ºC on STO. In Figure 3.5, we can see the structure of the Pt
clusters. Our treatment to form TiO2-terminated STO surface produces a terrace structure
[23] as shown in Figure 3.5(a) and the Pt clusters in the same terrace align in a row
following the STO surface morphology. The average diameter of a cluster is 150 nm and
the average height is 0.5 nm. The number of Pt atoms contained in one such cluster is on
the order of ~106. This is roughly the critical size of Pt clusters needed to develop a Fermi
edge and have a bulk metal core level binding energy.
Figure 3.4: Pt/STO valence band spectrum grown at 800 ºC as a function of thickness. It
shows the Pt contribution to the signal gradually increase in the range of 0–3
eV. At 2 ML, it shows nonzero intensity at 0 eV which means the Fermi
edge starts to form.
43
Figure 3.5: (a) STO surface after water boiling treatment. It shows a terrace structure with
flat steps (b) AFM measurement of 2ML Pt/STO grown at 800 ºC and
corresponding line profile measurement. The Pt nanoclusters (with ~150nm
diameter and 0.5 nm height) arrange themselves in rows aligned along the
substrate terrace edges.
44
To gain further insight, we acquired SEM images for Pt films grown at 200 and 800
ºC, which are shown in Figure 3.6. For 200 ºC deposition, it is hard to infer a clear surface
structure for the first several monolayers of Pt with the resolution of our equipment. Even
for a 10 ML Pt/STO sample, there is still no clear cluster or particle structure seen at the
nanometer-length scale [as shown in Figure 3.6(a)]. Instead, the image shows a flat,
featureless coverage with a visible edge at the area blocked by sample holder when we
check it at micrometer scale [as shown in Figure 3.6(a) inset]. Although Pt does not wet
the STO surface [16], at low growth temperature, the mobility of Pt atoms is quite low, and
most of Pt atoms deposited on the substrate will stay where they arrive and will not move
around. As more Pt is deposited on the surface, a film with a flat surface forms. However,
at 800 ºC, things are quite different. Pt atoms have enough thermal energy to freely move
across the surface and form nanosize clusters with diameter ~80 nm at an average Pt
coverage of 1 ML [Figure 3.6(b)]. This agrees with Christensen’s high-resolution SEM
image of 1 ML Pt on STO which shows a similar nanocluster structure [19]. When we heat
up a sample grown at 200 ºC to high temperature (800 ºC for an hour), the surface
morphology drastically changes. The SEM image of such a sample is displayed in Figure
3.7(a) and shows that the initially flat film breaks and Pt clusters of various sizes form. We
use Hough transform to analyze the distribution of the cluster size and obtain the histogram
shown in Figure 3.7(b). Note that the algorithm is not accurate for very small clusters
(below 3 nm) due to the resolution of the images. However, we are still able to determine
that the vast majority of Pt clusters have radii in the 3–5 nm range with a separation of
~25–30 nm.
45
Figure 3.6: (a) SEM image of Pt/STO grown at 200 ºC for 10 ML. At the nanometer scale
we can only see a featureless surface with a clear edge when we check it at
micrometer scale (as shown in the inset). The left area in the inset has no Pt
grown on because of shadowing by the sample holder during deposition. (b)
SEM image of Pt/STO grown at 800 ºC for 1 ML. It shows uniform
nanoclusters with average lateral size of ~80 nm.
46
47
Figure 3.7: (a) SEM image of Pt/STO with 10 ML of Pt grown at 200 ºC and annealed at
800 ºC for an hour in air. The flat surface breaks out and splits into small
nanoclusters. Pt clusters have radii in the 3-5 nm range with a separation of
~25-30 nm. (b) Histogram of the cluster sizes obtained from the SEM
image. The first two columns are gray because of the overcounting of the
algorithm at very small radius. Red line is the normal distribution fit of the
rest of the columns centered at 3.75 nm.
3.4 CONCLUSIONS
We investigated the deposition of Pt on atomically smooth, TiO2-terminated (001)-
oriented STO substrates for coverages up to 10 ML in the temperature range from 200 to
800 ºC. In situ XPS, AFM and SEM were used to explore the core level binding energy,
valence band spectrum, and surface morphology as functions of coverage and deposition
temperature. We found that owing to the relatively low mobility of Pt at low temperature,
initially, Pt atoms tend to stay at the low energy oxygen hollow site of the STO surface and
thus exhibit a partially oxidized state in core level XPS in agreement with DFT calculations
[11]. As coverage increases the Pt valence band develops along with the formation of a flat
Pt film, culminating in the emergence of the Fermi edge after 2 ML. However, when the
growth temperature is increased, Pt has enough mobility to move across the surface and
form clusters. Higher growth temperature results in larger Pt clusters that demonstrate
metallic XPS features at lower coverage than if the deposition were done at lower
temperature. SEM clearly shows nanoclusters forming at high growth temperature and a
continuous film at low temperature. The critical Pt cluster size required to exhibit a clear
Fermi edge and have bulk metallic core level is found to be ~106 atoms per cluster by AFM.
48
3.5 ACKNOWLEDGMENT
The authors thank Hosung Seo for insightful discussions. This work was supported
by the Air Force Office of Scientific Research under Grant Nos. FA9550-12-10494 and
FA9550-14-1-0090.
3.6 REFERENCES
1. Q. Fu and T. Wagner, Surf. Sci. Rep. 62, 431 (2007).
2. M. Mrovec, J.-M. Albina, B. Meyer, and C. Els€asser, Phys. Rev. B 79, 245121
(2009).
3. R. Schafranek, S. Payan, M. Maglione, and A. Klein, Phys. Rev. B 77, 195310
(2008).
4. A. B. Posadas, K. J. Kormondy, W. Guo, P. Ponath, J. Geler-Kremer, T. Hadamek,
and A. A. Demkov, J. Appl. Phys. 121, 105302 (2017).
5. M. S. Chen and D. W. Goodman, Science 306, 252 (2004).
6. M. Kitano, K. Tsujimaru, and M. Anpo, Top. Catal. 49, 4 (2008).
7. B. Reihl, J. G. Bednorz, K. A. M€uller, Y. Jugnet, G. Landgren, and J. F. Morar,
Phys. Rev. B 30, 803 (1984).
8. A. A. Demkov and A. B. Posadas, Integration of Functional Oxides with
Semiconductors (Springer, New York, 2014).
9. M. Copel, P. R. Duncombe, D. A. Neumayer, T. M. Shaw, and R. M. Tromp, Appl.
Phys. Lett. 70, 3227 (1997).
10. J. Robertson, J. Vac. Sci. Technol. B 18, 1785 (2000).
11. H. Seo and A. A. Demkov, Phys. Rev. B 92, 245301 (2015).
12. Y. W. Chung and W. B. Weissbard, Phys. Rev. B 20, 3456 (1979).
13. M. K. Bahl, S. C. Tsai, and Y. W. Chung, Phys. Rev. B 21, 1344 (1980).
14. F. Rao, M. Kim, A. J. Freeman, S. Tang, and M. Anthony, Phys. Rev. B 55, 13953
(1997).
15. A. Asthagiri and D. S. Sholl, J. Chem. Phys. 116, 9914 (2002).
16. H. Seo, A. B. Posadas, and A. A. Demkov, J. Vac. Sci. Technol., B 30, 04E108
(2012).
17. J. Son, J. Cagnon, and S. Stemmer, J. Appl. Phys. 106, 043525 (2009).
49
18. A. D. Polli, T. Wagner, T. Gemming, and M. R€uhle, Surf. Sci. 448, 279 (2000).
19. S. T. Christensen, B. Lee, Z. Feng, M. C. Hersam, and M. J. Bedzyk, Appl. Surf.
Sci. 256, 423 (2009).
20. R. C. Hatch, M. Choi, A. B. Posadas, and A. A. Demkov, J. Vac. Sci. Technol., B
33, 061204 (2015).
21. H. Brune, Surf. Sci. Rep. 31, 125 (1998).
22. G. K. Wertheim and S. B. DiCenzo, Phys. Rev. B 37, 844 (1988).
23. M. Choi, “Monolithic integration of functional perovskite structures on Si” Ph.D.
thesis (The University of Texas at Austin, Austin, TX, 2014).
50
Chapter 4: EuO epitaxy on SrTiO3 by oxygen scavenging
The EuO/SrTiO3 heterojunction is a promising combination of a ferromagnetic
material and a two-dimensional electron system. We explore the deposition of Eu metal on
STO/Si pseudo-substrates, with varying STO thickness, under ultrahigh vacuum
conditions. By varying the thickness of the STO layer (2-10 nm) and the deposition
temperature (20-300 °C), we investigate the process by which oxygen is scavenged from
STO by Eu. In situ x-ray photoelectron spectroscopy is used to investigate the electronic
structure of the nominal Eu/STO/Si stack. We find that as a result of Eu deposition,
epitaxial EuO is formed on thick STO (6-10 nm), leaving behind a highly oxygen-deficient
SrTiO3-δ layer of ∼4 nm in thickness. However, if the thickness of the STO layer is
comparable to or less than the scavenging depth, the crystal structure of STO is disrupted
and a solid state reaction between Eu, Si, and STO occurs when the deposition is done at a
high temperature (300 °C). On the other hand, at a low temperature (20 °C), only a 1-2 nm-
thick EuO interlayer is grown, on top of which the Eu metal appears to be stable. By
analyzing the Eu/EuO interface, we also find that electrons transfer from Eu metal into
EuO and induce an unexpected downward band bending at the interface. We use density
functional theory (DFT) to calculate the atomic and electronic structure of the interface and
find that the theoretical and experimental results agree with each other. For further study
on the manipulation of the 2DEG at the EuO/STO interface, we insert a BTO layer under
STO and explore the field-effect arising from the BTO polarization on the 2DEG using soft
x-ray angle-resolved photoemission spectroscopy (SX-ARPES). This study elucidates the
EuO growth process under different conditions and provides better understanding and
control of this system.
51
The major content of this chapter was published in: W. Guo, A. B. Posadas, S. Lu,
D. J. Smith and A. A. Demkov, J. Appl. Phys. 124, 235301 (2018). And L. Gao, W. Guo, A.
B. Posadas and A. A. Demkov, Phys. Rev. Materials 3, 094403(2019). Collaborators at
Arizona State University are responsible for the TEM results. ARPES measurements were
greatly supported by Prof. Strocov at the Paul Scherrer Institute (PSI).
4.1 INTRODUCTION
The ferromagnetic semiconductor EuO is a promising material for spintronic
devices [1]-[3] due to its very large magnetic moment (7 μB per Eu atom) and large spin-
splitting of the conduction band leading to almost 100% spin polarization of carriers [4],
[5]. The semiconducting behavior combined with the high magnetic moment makes it an
ideal spin-filter material below its Curie temperature of 69 K [6]. EuO has been epitaxially
integrated onto graphene [7], silicon [1], [8], GaN [1], and several oxide substrates [9] to
provide ferromagnetism in these systems. In addition, EuO also exhibits large
magnetoresistance, strong magneto-optical effect, and unusual transport properties [1], [8]-
[11]. These features make EuO a fascinating oxide for fundamental research, in addition to
the possibility of enabling new kinds of device applications involving spin transport.
The integration of EuO on heterostructures exhibiting a two-dimensional electron
gas (2DEG) is a potentially interesting approach to explore the interaction of strong
magnetism with the sheet charge, particularly for the case of the 2DEG at STO interfaces
[9], [12], [13]. By controlling the deposition of an oxide with a large negative enthalpy of
formation like EuO on STO [14], one can stabilize a highly confined conductive layer of
oxygen-deficient STO at the interface [15]. This heterostructure then offers a way to
combine strong ferromagnetism and a 2DEG in one system. Kormondy et al. have recently
52
reported the observation of a large linear positive magnetoresistance in the 2DEG at the
EuO/SrTiO3 interface [9]. Lömker et al. also report a magnetically tunable two-
dimensional electron system in the same structure [12]. Prinz et al. created quantum wells
with ultrathin EuO layers and widened the bulk EuO bandgap from 1.19 eV to ∼1.4 eV in
ultrathin films by the quantum confinement effect [16].
Growing thin films of high quality, stoichiometric EuO is difficult because of the
high stability of the competing Eu2O3 phase [6], [9]. Thin-film growth of EuO has been
reported in several studies [6]-[9], [12], [17], [18] and it is crucial to control the oxygen
partial pressure in order to form stoichiometric EuO and prevent over-oxidation [18].
Alternatively, stoichiometric EuO can be grown on certain substrates under Eu adsorption-
controlled conditions [7], [8], [19]. EuO integration on Si has been reported by several
groups [6], [8], [18], [20], and the EuO/Si interface has been predicted to be
thermodynamically stable if the two materials are in contact [21]. Lettieri et al. reported an
epitaxial EuO films grown on Si by molecular beam epitaxy (MBE) using an SrO buffer
[18]. Sr deoxidation is used to remove native SiO2 [22], [23] and 5 monolayers (ML) of
SrO are grown to provide a lattice-matched surface for EuO growth (0.3% mismatch).
Caspers et al. grew polycrystalline EuO directly on HF-treated, H-passivated Si substrates
by precisely controlling the Eu and oxygen flux [6]. These authors also grew epitaxial EuO
[001] on Si [001] by passivating the Si surface with a 13-Å silicon sub-oxide (SiOx) buffer
layer prior to EuO deposition [20]. However, this process also led to interfacial silicide
formation. Direct EuO growth on Si was reported by Averyanov et al. who used sub-
monolayer reconstructions of Eu on Si to create a Zintl-like template to avoid Si oxidation
[8]. There are also reports of epitaxial EuO growth on STO with buffer layers such as BaO
or SrO to prevent over-oxidation of EuO due to oxygen out-diffusion from STO [24], [25].
53
Common to all reported growth methods, EuO growth is typically performed at
temperatures between 300 and 450 °C. Furthermore, controlling the oxygen pressure is
always crucial to form high quality, stoichiometric EuO. Posadas et al. found that EuO can
be grown directly on an STO substrate by Eu deposition in ultra-high vacuum (UHV)
without providing any oxygen [14]. It has been shown that when metals with large enthalpy
of oxide formation (such as Al and Eu) are deposited on STO substrates, oxygen can be
scavenged from the top STO layers and form oxides on top even without additional oxygen.
The oxygen vacancies created in STO create a conductive layer on the STO side of the
interface. This conductive layer in STO due to oxygen vacancies has also been reported in
other studies [9], [14], [26].
Eu, with very large oxide formation enthalpy and a low work function of 2.5 eV, is
expected to scavenge significant amounts of oxygen from STO to form EuO, leaving
behind a highly conductive oxygen-deficient SrTiO3-δ layer [14]. Kormondy et al. used this
process to grow epitaxial EuO on STO single-crystal substrates at 300 °C in UHV. The
spin-polarized 2DEG formed at the EuO/STO interface displayed positive linear
magnetoresistance when the EuO was below its Curie temperature. From electron energy
loss spectroscopy (EELS) analysis in cross-sectional scanning transmission electron
microscopy, the oxygen-deficient SrTiO3-δ layer was observed to be ∼4 nm deep [9]. This
thickness appears to be the “oxygen scavenging depth” of Eu when deposited on STO at a
temperature of 300 °C. It is expected that this depth will vary strongly with temperature
since oxygen scavenging involves atomic diffusion. Therefore, it is important to investigate
the effect of temperature, and also STO layer thickness, on the scavenging of oxygen in
STO by Eu. In particular, it would be interesting to establish what happens when the STO
54
thickness becomes comparable to or smaller than the oxygen scavenging depth. In other
words, how much oxygen can Eu scavenge from STO before STO starts to be degraded?
EuO has a cubic rock-salt structure with a lattice constant of 5.14 Å [27]. The lattice
matching of EuO to a cubic perovskite structure requires a 45° in-plane rotation, essentially
lining up with the rock salt-like layers of the perovskite. The nominal matching is therefore
between [100] direction of the perovskite and [110] direction of rock salt. Schematics of
the matching interface structure are displayed in Figure 4.1. In the case of EuO matching
with STO (a = 3.905 Å), the lattice mismatch is ∼7%.
55
Figure 4.1: Schematic of EuO/STO epitaxial arrangement: (a) side view (b) top view.
In this chapter, we report on the fabrication of EuO/SrTiO3/Si heterostructures by
depositing metallic Eu under UHV conditions on STO/Si pseudo-substrates with varying
STO layer thickness. Different STO thicknesses allow us to place a limit on the amount of
available oxygen for reaction with Eu. We investigate the oxygen scavenging mechanism
56
by varying both the STO thickness (from 2 to 10 nm) and the temperature of Eu deposition
(from 20 °C to 300 °C). Crystallinity of the growing film is monitored by in situ reflection
high-energy electron diffraction (RHEED). We analyze the composition of the resulting
structure by in situ x-ray photoelectron spectroscopy (XPS). DFT is also used to calculate
the electronic structure of the Eu/EuO interface, which we then compare to experimental
XPS results. Further study on the field-effect of the 2DEG in EuO/STO has also been
started but more experiments are needed to draw any conclusions from this study.
4.2 EXPERIMENTS ON OXYGEN SCAVENGING OF STO BY EU
The deposition is performed in a customized DCA 600 MBE system with a base
pressure of 6 × 10−10 Torr. The Si substrates (p-type with a resistivity of 0.1-0.2 Ω cm from
University Wafer) are diced into 20 × 20 mm2 pieces and degreased ultrasonically in
acetone, isopropanol, and deionized water (18MΩ/cm) for 5 min each and then exposed to
UV/ozone to remove carbon from the surface. The Si substrates are then annealed in the
vacuum chamber at 700 °C for 10 min followed by the growth of a 2-nm-thick STO
epitaxial film at 250 °C which is crystallized by annealing in vacuum at 500 °C [21], [28],
[29]. The STO films are grown such that there is no interfacial SiO2 present by separating
the oxygen gas introduction with the STO crystallization anneal [28]–[31]. Additional STO
layers can be grown without SiO2 formation by depositing disordered STO at 250 °C using
co-deposition under moderate oxygen pressure (∼mid 10-7 Torr), and then annealing in
UHV at 500 °C to further crystallize the STO film. It is important to prevent SiO2 formation
to ensure that the only oxygen source for scavenging by Eu is from the STO layer.
We prepared STO/Si pseudo-substrates with STO layer thicknesses of 2, 3, 4, 6,
and 10 nm. For each STO layer thickness, different coverages of Eu are deposited from an
57
effusion cell with the cell temperature fixed at 530 °C. This cell temperature has been
determined to yield an Eu metal evaporation rate of ∼2.5 Å/min. An accurate measurement
of the arrival rate of Eu is done by measuring the thickness (using x-ray diffraction finite
size oscillations) of epitaxial Eu2O3 on GaN substrates following growth for a fixed amount
of time [32]. As the sticking coefficient of Eu is unity under Eu2O3 growth conditions, the
arrival rate of Eu on the substrate can be accurately determined.
The Eu metal is deposited under two substrate temperatures designated here as high
(300 °C) and low (20 °C). RHEED with 21 keV electrons is used to monitor the surface
crystallinity changes going from the bare STO surface to EuO during growth. In situ x-ray
photoelectron spectroscopy (XPS) with monochromated Al Kα radiation and a VG Scienta
R3000 hemispherical electron energy analyzer are used to check the core levels of relevant
elements as well as the valence band (VB).
The Eu metal is deposited in a step-by-step fashion and checked with XPS after
each step. Each Eu deposition step lasts for a few minutes, with a deposition pause after
each step in order to transfer the sample into the in situ XPS chamber for analysis, and then
moved back to the MBE growth chamber to deposit the next step. This process is repeated
until the total Eu deposition is finished. This procedure allows us to see the evolution of
the chemical and electronic structure of the sample as Eu is deposited. The core levels Eu
3d, Eu 4d, Si 2p, Ti 2p, and O 1s are measured as a function of Eu coverage to obtain
information about the scavenging process.
58
4.3 RESULTS AND DISCUSSION
4.3.1 High temperature deposition with thick STO film
We previously showed that EuO can be epitaxially grown on STO single crystal
substrates by oxygen scavenging from direct Eu metal deposition at 300 °C with a ∼4 nm
scavenging depth [9]. In the present work with EuO/SrTiO3/Si heterostructures, we first
perform Eu deposition on 10 nm STO on Si at 300 °C. This STO thickness appears to be
sufficient to provide enough oxygen to Eu metal in order to form stoichiometric EuO with
several nm thickness, similar to our earlier results on STO single crystal substrates [9]. The
Eu metal was deposited for ∼30 min, which is expected to form roughly 75 Å of Eu metal,
assuming unity sticking coefficient. Eu metal has a body-centered-cubic structure with 4.58
Å lattice constant, and EuO has a rock-salt structure with a lattice constant of 5.14 Å [27].
Assuming the entire 75 Å Eu metal layer is oxidized to EuO by oxygen scavenging, this
would result in a EuO thickness of ∼54 Å. RHEED patterns after Eu deposition along the
[110] and [100] STO azimuths are shown in Figure 4.2. The pattern observed is associated
with the rock-salt surface indicating the formation of epitaxial EuO. During Eu deposition,
the RHEED pattern for the STO layer first becomes blurry and fades away after 5 min
(∼12.5 Å of Eu metal) as newly arrived Eu covers STO in a disordered manner. As Eu
deposition is continued, the RHEED pattern for EuO gradually appears, becoming sharper
and brighter as deposition proceeds. This is a clear indication that oxygen atoms have been
scavenged from the STO layer through the growing EuO layer up to the surface to form
crystalline EuO.
59
Figure 4.2: RHEED pattern for EuO grown on 10 nm STO/Si at 300 °C.
To better understand the EuO/STO interface, we performed detailed electron
microscopy observation for several EuO films grown on STO single-crystal substrates by
oxygen scavenging. Representative images are shown in Figure 4.3. The samples were
prepared in cross-section geometry using standard mechanical polishing and argon ion-
milling, and images were recorded with an ARM200F scanning transmission electron
microscope operated at 200 keV. The beam convergence angle was set at 20 mrad, and the
collection angles were 0-22 mrad for large-angle bright-field (BF) imaging and 90-150
mrad for high-angle annular-dark-field (HAADF) imaging.
60
Figure 4.3: Aberration-corrected TEM images of the EuO/STO interface with the atomic
model overlaid. Pink circles are Eu, red circles are oxygen, green circles are
Sr, and blue circles are Ti. (a) and (b) HAADF and BF pair of images for
EuO [110]/STO [100] projections. (c) and (d) HAADF and BF pair of
images for EuO [100]/STO [110] projections.
Figures 4.3(a) and 4.3(b) are a pair of HAADF and BF images of the EuO/STO
interface for EuO [110]/STO [100] crystal projections. The abrupt interface and the high
quality EuO epitaxial growth are clearly apparent. Figures 4.3(c) and 4.3(d) show another
HAADF/BF pair of images for EuO [100]/STO [110] projections, again confirming the
excellent crystallinity of the epitaxial EuO layer.
In situ XPS scans confirm a pure Eu2+ signal (no Eu0 or Eu3+) consistent with
stoichiometric EuO from both the Eu 3d and 4d core levels (not shown), and also from the
valence band (VB) as shown in Figure 4.4. The EuO VB shows a strong peak at ∼2 eV
from the Eu 4f states and a smaller feature at higher binding energy associated with O 2p
61
in EuO. It is obvious from the XPS data that no Eu3+ oxidation state is present, which would
normally be found in the region from 6 to 8 eV. This result means that EuO is not over-
oxidized and that Eu oxidation from scavenging is self-limited to the +2 oxidation state at
this deposition temperature. These results confirm that crystalline EuO can be grown on
thick (10 nm or more) STO/Si pseudo-substrates, with the same behavior as on a bulk STO
substrate.
Figure 4.4: VB spectrum of Eu grown on 10 nm STO/Si. Only Eu2+ and oxygen features
are visible.
4.3.2 High temperature deposition with thin STO film
Next, Eu metal is deposited on thinner STO layers at the same temperature (300
°C). For STO layers with a thickness of 6 nm, the same oxygen-scavenging process as that
observed on bulk STO occurs, and EuO is epitaxially formed for several nm. However,
when the STO layer becomes thinner than the scavenging depth (∼4 nm) at this
62
temperature, the RHEED pattern evolution is different. A pattern similar to Figure 4.2
emerges briefly but vanishes later during growth. The sequence of RHEED patterns
observed is shown in Figure 4.5. The same sequence of RHEED patterns is also observed
for growth on thinner STO layers (2-3 nm).
Figure 4.5: RHEED pattern evolution during Eu deposition on a thin (2 nm) STO layer.
The disappearance of the EuO RHEED pattern indicates either the loss of EuO
crystallinity or the accumulation of some other non-crystalline phase on the surface. By
checking the XPS of Eu deposited on <4 nm STO film, we find the presence of another
feature at ∼1 eV binding energy in the VB spectrum [Figure 4.6(a)], which is associated
with EuSiy. This is also confirmed in Eu 3d and 4d core level spectra (not shown), which
also have an extra EuSiy feature that causes a ∼0.4 eV shift of the leading edge to lower
binding energy [19].
63
Figure 4.6: (a) VB spectra of Eu on 2-nm STO/Si. Two types of Eu peaks are visible. (b)
Si 2p signal of Eu deposited on 2-nm STO/Si. Silicon metal signal is almost
gone but SiOx and EuSiy appear.
64
Surprisingly, the normally buried Si 2p signal also now appears in the XPS scans
and shows both silicide (Si− at 98.2 eV) and silicate (Si+ at 100.8 eV) features. Only a small
amount of elemental Si remains (Si0), indicating a relatively thick Si reaction layer [Figure
4.6(b)]. We overlay the unreacted EuO VB spectra and Si metal signal for comparison in
Figure 4.6 with dashed lines, so that the shifts are more obvious. The appearance of Eu
silicide indicates that Eu reacts with the Si substrate at some point during growth, most
likely when the EuO RHEED pattern disappears.
To further explore how this reaction takes place, we performed a step-by-step Eu
deposition on thin STO (3 nm) to see the evolution of the electronic structure. The evolution
of the VB spectrum is shown in Figure 4.7.
65
Figure 4.7: VB spectra of step-by-step Eu growth on 3-nm STO. Growth times are shown
with different colors that correspond with the spectra.
For the first 10 min (∼25 Å Eu metal), the spectra show only EuO in the VB (Figure
4.7, red line) and elemental Si in the Si 2p scans (not shown here). The RHEED pattern
also corresponds to that of crystalline EuO during this time. When more Eu is deposited,
the RHEED pattern becomes blurry and the VB spectrum starts to show a gradually
increasing EuSiy signal, indicated by the leading edge shifting to lower binding energy.
The presence of possible Eu metal (Eu 5d feature) cannot be excluded from the VB
spectrum alone because it is too small to distinguish from the broad EuSiy signal. The Eu
4d signal shows that there is likely no Eu metal present because the characteristic multiplet
splitting in the metallic Eu 4d spectrum is not observed [32]. At the same time, the Si 2p
spectrum also starts to show both silicide and silicate features. From these results, it is
inferred that Eu oxygen scavenging happens during the first 15-20 min (∼50 Å Eu metal).
After this point, there is not enough oxygen in the thin STO for it to maintain its structure.
We know at 300 °C, Eu has the ability to scavenge oxygen from STO up to 4 nm
deep. When the STO film is thinner than 4 nm, excess Eu results in complete oxygen
depletion of STO. The STO is unable to maintain its structure and is essentially
decomposed by losing too much oxygen. At this temperature, Si atoms are able to diffuse
outward and react with EuO (as well as SrO + Ti). The existence of EuSiy and silicates is
clearly seen from the Eu and Si core level and VB spectra. The likely reaction that could
occur is EuO + SrTiO3 + Si→TiSi2 + SrSiOx + EuSiy. The disappearance of the EuO
RHEED pattern is likely because the reaction forms a mixture of several compounds that
physically break up the EuO layer so that it is no longer crystalline. This is an intriguing
result because it appears to contradict the prediction that EuO is thermodynamically stable
in contact with silicon [19]. Other groups have grown EuO on buffered Si at temperatures
66
higher than 300 °C [6], [8], [18], which implies that the decomposition of STO is acting as
some kind of catalyst for the reaction of EuO and Si.
4.3.3 Low temperature deposition
The degree of oxygen scavenging from STO by an overlayer must depend on the
mobility of oxygen atoms/ions through the oxygen-deficient STO and through the growing
metal oxide layer. Because oxygen transport is required, this phenomenon must be strongly
dependent on the substrate temperature. We also perform Eu deposition on thin STO (<4
nm) at room temperature (20 °C) to try and suppress the scavenging process and confirm
the temperature dependence. Initially, a RHEED pattern associated with crystalline EuO
still shows up with clear and sharp features, which means that the usual oxygen scavenging
process occurs first, even at a low temperature. This is expected because oxygen atoms do
not initially have to travel far. The crystalline EuO pattern, however, quickly becomes very
diffuse and dim and stays like that for the rest of the deposition. XPS analysis of this sample
showed no EuSiy and SiOx signals in the Si 2p spectrum, indicating that the thin STO layer
remained intact and prevented Si from diffusing outward. The VB spectrum also showed a
clear Eu metal signal [Figure 4.8(a)]. The feature around 0 eV [Figure 4.8(a), black peak],
which is separated from the main Eu 4f peak at 2 eV, comes from Eu 5d electrons at the
Fermi level, which only appears in Eu metal [33], [34]. The Eu 4f peak can be decomposed
into two components, coming from Eu metal and Eu2+. This observation is also supported
by recent density functional theory (DFT) calculations of the Eu/EuO interface by Gao et
al. [35]. The Eu 4d core level spectrum further confirms the existence of Eu metal because
of the clearly separated multiplet peaks around 127-132 eV corresponding to different total
angular momentum J quantum numbers [32] [Figure 4.8(b)].
67
Figure 4.8: (a) Valence band spectrum of Eu on 2-nm STO/Si at 20 °C. (b) Eu 4d
spectrum with clearly resolved multiplets characteristic of Eu metal.
68
Figure 4.9: Valence band evolution of Eu deposition on 2-nm STO/Si at 20 °C.
Step-by-step Eu deposition on 2 nm STO/Si at 20 °C indicates that Eu metal shows
up after ∼20 min (∼50 Å Eu metal) of Eu deposition (Figure 4.9). Only crystalline EuO is
formed in the first 10 min as determined by XPS VB spectra and RHEED with an Eu 4f
peak located at a binding energy of ∼2.2 eV. After ∼20 min, the Eu 5d metal feature starts
to appear and increases in intensity [Figure 4.9, orange line]. This means that a 1-2 nm-
thick EuO layer forms, beyond which point Eu cannot scavenge more oxygen from the
STO layer and only Eu metal accumulates on the surface. The Eu metal 4f peak centered
at a binding energy of ∼1.9 eV continuously increases after 20 min while the Eu2+ 4f peak
at 2.5 eV remains and decreases, eventually causing the leading edge of the combined 4f
peak to shift to lower binding energy. The O 2p feature in the VB is gradually buried by
69
the Eu metal signal which has broad peaks in the VB spectrum in the region 3-10 eV. It is
also worth mentioning that Eu metal is very reactive to any residual oxygen present. Eu
metal films will show a strong Eu3+ component due to the residual oxygen even when
stored in ultra-high vacuum (1 × 10−9 Torr) after a day or two.
We also looked at the Si 2p and O 1s spectra in the step-by-step Eu growth at 20
°C (not shown). The Si 2p signal only showed elemental Si peaks with a continuously
decreasing intensity with virtually no chemical shift. This confirmed no reaction with the
Si at this temperature. The O 1s signal maintained its intensity in the first 10min (∼25 Å
Eu metal) and started to decrease after 10min. This indicated that EuO stopped forming
after 10 min and Eu metal started to bury the EuO after this time. Combining all these
results, we conclude that at a low temperature, oxygen scavenging is limited by the very
low mobility of oxygen atoms/ions. Only a small amount of oxygen is scavenged from
STO right at the interface, allowing the STO layer to maintain its structure during excess
Eu deposition. Without any oxygen supply, more Eu can only stack in metallic form. We
also performed the same experiment at 150 °C, and the results appear to be the same as
that at 20 °C. Further experiments confirm that the oxygen scavenging process is
suppressed below 150 °C growth temperature.
4.3.4 Theoretical verification by DFT
Gao et al. [35] used DFT to study the atomic and electronic structure of the Eu/EuO
interface, as we have seen from low temperature Eu deposition on STO. Calculations
predict that electrons transfer from Eu metal into EuO and induce an unexpected downward
band bending at the interface. Accounting for spectral broadening and attenuation of the
signal from subsurface layers, the calculated layer-resolved total density of states agrees
70
well with experimental XPS in the valence-band region. The total 4f spectrum contains
contributions from both Eu and EuO, with the latter component significantly broadened as
a result of band bending. This bending and charge transfer originate from Eu Fermi-level
pinning at the EuO charge neutrality level, which has been suggested to be located above
the conduction-band bottom.
Calculation details are not shown here and people who interested can refer to
original paper published by Gao et al. [35]. We present below the related theoretical
verification and corresponding XPS spectra and fitting results.
Figures 4.10(a) and 4.10(b) show the experimental XPS data for Eu step-by-step
deposition on STO/Si pseudo-substrate at 20 °C. In Figure 4.10(a), we show the valence-
band (VB) XPS spectrum for a sample after 10 min of Eu deposition on STO. Only rocksalt
EuO forms at this time as a result of oxygen scavenged from STO by Eu atoms. The EuO
is crystalline as indicated by RHEED, as we mentioned in previous sections. The Eu2+ 4f
peak is observed at a binding energy of 2.2 eV, and the smaller and broader peak centered
at 5 eV is from the O 2p band. From 6 to 15 eV the spectrum is flat without any peaks,
indicating that there is no Eu3+ present. This is important, as the formation of the sesqui-
oxide Eu2O3, which is not ferromagnetic, could reduce the spin-filtering efficiency of a
EuO barrier [28].
71
Figure 4.10: (a) Valence-band XPS spectrum of 10 min Eu deposition on 2-nm STO/Si
pseudo-substrate at 20 °C. This is consistent with the spectrum from bulk
EuO. (b) Valence-band XPS spectrum of 60 min Eu deposition at 20 °C. A
Eu/EuO/STO/Si structure is formed. (c) The attenuated and broadened 4f
DOS of the whole heterostructure (“calculated pseudo-XPS spectrum”). The
Eu metal 4f DOS is marked as dark blue, the 4f DOS of each EuO layer is
marked as magenta, the total EuO 4f DOS is marked as violet, and the total
4f DOS is marked as red. The inset gives the 4f DOS of each EuO layer on a
larger scale. (e) A three-dimensional representation of the “calculated-
pseudo-XPS spectrum.” Here the spectrum of each layer is shown separately
along the y direction. The peak position of 4f EuO DOS of each layer is
connected with a violet line. (Figure is also published in Ref. [35])
In Figure 4.10(b), the VB spectrum of a sample after 60 min of Eu deposition is
shown. We can see the presence of the metallic Eu spectrum. Due to the low deposition
72
temperature resulting in low oxygen mobility through EuO, Eu is unable to continuously
scavenge oxygen from STO. This limits the thickness of EuO to about 2 nm with Eu metal
accumulating on top of the EuO. Compared with Figure 4.10(a), which shows the spectrum
of pure EuO, several other features can be seen due to the presence of Eu metal. There is a
clear, nonzero intensity at the Fermi energy appearing as a small step, corresponding to the
partially occupied Eu 5d band, which is a clear signature of the metal. This feature emerges
at 30 min deposition time and increases in relative intensity as more Eu is deposited. The
Eu 4f feature becomes asymmetric and is now required to be fit with two components
instead of one, as shown in more detail in Figure 4.10(b). The 4f component at 2.2 eV is
the Eu2+ signal from EuO, while the other 4f component at 1.9 eV is interpreted as an Eu
metal 4f signal. As more Eu metal is deposited, the 4f peak clearly shifts toward lower
binding energy, because the Eu component at 1.9 eV is increasing in intensity while the
EuO component at 2.2 eV is being progressively buried. This is a clear indication that at
some point oxygen is no longer able to diffuse through existing EuO at this temperature
and Eu metal begins to accumulate atop the EuO layer. There are also additional broad and
weak features in the 4 - 14 eV binding energy range that appear in the metallic phase that
quickly bury the EuO oxygen 2p peak. These are likely related to Eu 6s electrons in Eu
metal as seen in synchrotron measurements [36]. It should be noted that because oxygen is
supplied via diffusion from STO, there is likely a slight reduction in oxygen content in the
film as the EuO grows thicker. It is therefore likely that a slightly oxygen-deficient EuO
layer is present right at the interface up to the composition range of the EuO phase, which
is reported to be not more than 1 mol% [37], [38]. However, since oxygen solubility in Eu
is extremely small [39], any oxygen not reacting to form EuO is likely to segregate to the
(oxygen-deficient) EuO side of the interface, resulting in Eu metal starting to accumulate
73
once the scavenging depth is reached. This self-limiting behavior of EuO thickness is also
observed in the recent detailed study of the EuO oxygen scavenging process by Lomker
and Muller [40].
To facilitate the comparison with the experiment, we broaden the calculated 4f
DOS of each layer using Gaussian convolution. Based on the width of experimental XPS
spectra, we choose a FWHM parameter σ = 0.90 eV. Considering the inelastic scattering
in an XPS measurement, we attenuate the 4f DOS for each layer by a factor of 𝑒−𝑑
𝐿 where
d is the depth of the layer and L is the inelastic mean free path. The inelastic mean free
path of a 4f electron (kinetic energy of ∼1480 keV) is 23 and 31 Å for EuO and Eu,
respectively, as calculated using the TPP-2M formula [41]. We then add up the attenuated
and broadened DOS of each layer as the “calculated pseudo-XPS spectrum” and show it in
Figure 4.10(c). The 4f DOS of each EuO layer (magenta lines) clearly shows the band
bending. The peak position of the interfacial EuO 4f DOS is 2.7 eV and that of the bulk-
like EuO 4f DOS is at 2.3 eV. This leads to the overall broadening of the EuO 4f spectra.
For the Eu metal 4f state, the peak position is 2.1 or 0.4 eV higher compared to that of the
total EuO 4f DOS. This is close to the experimental measurement (0.3 eV), and the 0.1 eV
discrepancy can be ascribed to the EuO band-gap difference between the calculation and
experiment. As can be seen from the experimental XPS spectrum, the Eu metal component
is much more dominant, as EuO is buried and the corresponding signal has been largely
attenuated due to the longer travel distance of the photoelectrons. We observe that for the
structure of two Eu metal layers over 8.5 layers of EuO, the EuO-to-Eu signal intensity
ratio from the pseudo-XPS spectrum is ∼0.7.
On the other hand, it also influences the spacing between the total 4f peak and Eu
metal 4f peak. In experiment, the peak position of the total 4f spectrum is very close to that
74
of the Eu metal 4f spectrum, while in the calculation, the former is separated by 0.2 eV
from the latter. We show the spectrum of each EuO layer separately in Figure 4.10(d).
Following the violet line indicating the peak position of the EuO 4f contribution, the band
bending across the EuO layers can be clearly seen. Overall, accounting for the contributions
by both Eu metal and EuO, the total 4f peak spectral width increases to 1.1 eV. This is
comparable to the 0.8 eV width observed in experiment.
4.4 FURTHER RESEARCH ON FIELD-EFFECT OF 2DEG IN EUO/STO
Kormondy et al. have reported the spin-polarized two-dimensional t2g electron gas
at the EuO/STO interface [9]. Soft x-ray angle-resolved photoemission spectroscopy (SX-
ARPES) revealed the t2g nature of the carriers. They showed that this 2DEG displays a very
large positive linear magnetoresistance below the Curie temperature of EuO of 70K.
However, it would be desirable to shift this sheet charge into EuO, where the spin split
conduction band would ensure 100 % spin-polarization along with the increase of the Curie
temperature, possibly above liquid nitrogen temperature [42].
Ponath et al. demonstrated a ferroelectric field effect and carrier density modulation
in an underlying Ge, by switching the ferroelectric polarization of an epitaxial c-axis-
oriented BTO film grown on Ge (001) by MBE [43]. They verified the effect of the
polarization switching on the conductivity by microwave impedance microscopy (MIM).
Lee et al. performed a theoretical calculation of the LaAlO3/EuO interface and
demonstrated that electrostatic doping by an electric field in the polar oxide leads to a fully
spin-polarized 2DEG at the interface [42]. These results hint at the possibility that by
integrating an EuO/STO junction on top of ferroelectric BTO, we may gain control over
the interface charge distribution, thus achieving an unexplored way of modulating the
75
2DEG, and possibly shifting the Curie temperature to the range of practical applications.
This idea was theoretically demonstrated by Li et al. recently [44].
For further research of the 2DEG, we performed a preliminary experimental study
on the field-effect in the 2DEG at the EuO/STO interface. We built an EuO/STO/BTO/Si
heterostructure by MBE and used the SX-ARPES instrument at the Paul Scherrer Institute
(PSI) [45]. The BTO is 40 nm-thick and the STO is 5-10 nm-thick. EuO was deposited
with 1-3 nm thickness with a 1-2 nm Ti capping layer on the top to protect EuO from over-
oxidation during transportation.
ARPES experiments are performed at the SX-ARPES end station [45] of the
ADRESS beamline at the Swiss Light Source (PSI, Switzerland) [46]. The principle of SX-
ARPES is the same as XPS, using the photoelectric effect and getting the core level energy
information by analyzing photoelectron energy distributions. Additionally, SX-ARPES
introduces a precise scanning angle control of the sample enabling one to achieve
momentum space resolution with increased probing depth. By combining photoelectron
kinetic energy and momentum analysis, SX-ARPES can get a detailed energy band map of
the sample. For our measurement, circularly polarized x-rays were incident on the sample
at a grazing angle of ~20°. The sample was cooled down to 10-20 K to reduce the thermal
effects and decrease the coherent k-resolved spectral component at high photoexcitation
energies [47]. The combined (beamline and analyzer) energy resolution was ~100 meV.
X-ray absorption spectroscopy (XAS) was performed first to determine the absorption
peaks for key core levels like Ti 2p and Eu 3d. Core levels were scanned after XAS to
check the basic core level electronic structure information. After we determined the core
level peaks positions, resonant maps were scanned for Eu and Ti to find any potential
2DEG. The resonant map uses a range of photon energies (a core level energy range) with
76
continuous energy values to excite the sample. This measurement will have direct
photoemission around the Fermi edge just like usual XPS. If the incident photon has the
resonant energy with a core level, it will also excite one electron to the outer shell, which
will undergo Auger decay. This Auger electron will enhance the total intensity around the
Fermi edge. This allows one to detect very weak energy band structures. We show our
results to date in the following figures. ARPES instrument details can be found in Refs. 45,
46.
The VB scan is shown in Figure 4.11. EuO is very easy to over-oxidize to Eu3+
during transportation even with capping. We see only ~15% of Eu2+ is left compared with
Eu3+. The unreacted VB scan is shown Figure 4.4 where only Eu2+ 4f dominates the
spectrum. These Eu3+ signals will heavily cover the Eu2+ we expect to see.
Figure 4.11: Valence band scan of a EuO/STO/BTO/Si sample after transportation in
ambient.
77
Figure 4.12 shows the Eu resonant spectrum within the energy window of the Eu
3d core level. The strongest signal corresponds to Eu3+ at 1133 and 1137 eV. Eu2+ resonated
at photon energy=1131 eV which is weak compared with Eu3+. There is no resonant signal
at the Fermi edge (~0 eV) when we zoom in, meaning there is no hole gas at Fermi level
in EuO. We then scanned the Ti resonant map to see the Ti L-edge resonant signal.
Figure 4.12: Eu resonant map spectrum at the Eu 3d core level.
Figure 4.13 shows the Ti resonant map around the Fermi level. It clearly shows a
strong Eu2+ 4f signal. This signal is too strong and resonant along with the whole energy
window of Ti 2p. Due to the high Eu2+ 4f intensity, the details at the Fermi edge are
overwhelmed. In Fig. 4.14, we show a zoom-in figure of Figure 4.13 and adjust the contrast
to highlight the possible 2DEG signals, which is circled in red. There are two very weak
signals resonant at 465.5 and 460.2 eV, which correspond to Ti 2p peak positions, but the
78
intensity is just too low that we need to adjust to a very high contrast to see them. This
means that either the signal comes from a depth out of the scope of soft x-rays (4-5 nm),
or there isn’t too much 2DEG in the sample. There are three possible reasons for the poorly
revealed 2DEG signal: 1) Most of the EuO is over-oxidized and the heavy Eu3+ signal
covers the interface signals; 2) The use of a soft x-ray light source limits the measurement
depth to 4-5 nm from the surface, which is where the EuO/STO interface is in the current
structure; 3) We use a Ti capping layer and Ti signals from the capping layer heavily cover
the 2DEG signal, as the t2g carriers in 2DEG also comes from Ti. We tried to perform a
quick k-space scan, but it did not show any fine band structure at the Fermi edge.
Figure 4.13: Ti resonant map with strong Eu2+ 4f signal.
79
Figure 4.14: Zoom-in Ti resonant map with possible 2DEG signals (circled in red).
The main challenge we face is to protect the EuO and avoid over-oxidation during
transportation and keep the EuO/STO interface as close as possible to the surface to have
stronger signals from the interface. It requires a capping layer as thin as possible during the
measurement, but still capable of protecting EuO from oxidation. We have tried various
capping layers and found that a GeOx/Ge capping layer works well to protect EuO.
This new capping process uses a thin, amorphous Ge capping layer (~2 nm)
deposited on EuO at room temperature and the sample is then taken out of the vacuum
system. Oxygen in air will quickly oxidize the top half of the Ge and form GeOx of ~1 nm.
Low oxygen transparency of GeOx at room temperature will protect the deeper Ge,
maintaining its metallic state and preventing the over-oxidation of the EuO layer. The
80
advantage of GeOx is that it can be thermally desorbed by heating the sample to ~300-400
ºC in vacuum. This results in an interface very close to the top surface that should have
stronger interface signals. This method solves all three possible reasons we mentioned for
the previous weak signals. We show below XPS spectra for each step to demonstrate the
feasibility of this method.
Figure 4.15: Oxygen 1s core level comparison for 1) Ge/EuO/STO/BTO/Si sample before
it was taken out of vacuum (red); 2) The same sample in step 1) left in air
for over 2 days (green); 3) The same sample in step 2) after being heated in
vacuum at 350 ºC for 1 hour (blue). Oxygen from EuO and GeOx are
marked above the peaks.
Figure 4.15 shows the oxygen 1s core level for the same sample at different points
in the process. Step 1) is to cap EuO with amorphous Ge metal at room temperature.
81
Oxygen peak intensity is low because the Ge covers EuO. Step 2) is to take the sample out
of vacuum and let it rest for over 2 days in ambient air to oxidize the surface Ge and form
GeOx. The surface layer is a mixture of GeO and GeO2 and has two oxidation states. Thus,
the peak intensity at 529.5 and 531.2 eV both increased. Step 3) is to load the sample back
into the vacuum system and heat the sample at 350 ºC for 1 hour. These are practical
heating conditions to ensure that EuO does not interact with STO and that most of the GeOx
can be removed. The blue curve in Fog. 4.15 shows a very small fraction of GeOx and the
oxygen signal from EuO increases by 2-3 times. This confirms that about half of the Ge
capping has been removed by oxidation and vacuum heating, with the EuO being closer to
the surface. This is very important for SX-ARPES signal enhancement on our samples.
82
Figure 4.16: Valence band comparison for 1) Ge/EuO/STO/BTO/Si sample before it was
taken out of vacuum (red); 2) The same sample in step 1) after air exposure
for over 2 days (green); 3) The same sample in step 2) after heating in
vacuum at 350 ºC for 1 hour (blue). Eu signal positions are marked above
the peaks.
To confirm that our EuO is not over-oxidized, we show the VB scan for the same
sample in each steps. There is only Eu2+ signal in step 1) and the oxygen 2p peak is covered
by VB signal from the Ge capping. After step 2), the Eu2+ signal decreased significantly
and the Eu3+ signal emerged. Considering that oxygen must diffuse from the outside
through the top Ge layer, the EuO next to Ge capping layer will oxidize first. The interface
between EuO and STO remains in the 2+ state. After thermal treatment, theEu2+ signal
increases since the Ge capping is now thinner. This shows that while the thin Ge capping
layer is not 100% protective for all the Eu2+ state, it still prevents over-oxidation of about
half of the 2.5-nm-thick EuO layer. The EuO/STO interface region itself should be well-
protected by the thin Ge capping with this procedure.
We tested this procedure by international shipping of samples with Ge capping.
Under normal transportation conditions, samples are packed in gel pack box with N2 gas
flowing. At least 70% of Eu2+ signal remains after 5 days of international shipping, which
means our protection is feasible for EuO samples, especially thicker ones. This is a great
breakthrough since the protection of EuO has been an issue for EuO characterization and
applications in the past.
4.5 CONCLUSIONS
The oxygen scavenging process for Eu on STO substrates has been explored by
depositing pure Eu metal at different temperatures on different thicknesses of STO films
integrated on Si (001). At 300 °C, Eu takes oxygen from STO films thicker than 4 nm and
83
forms EuO, which is similar to what happens on bulk STO substrates [16]. However, when
the STO films are thin (in the range from 2 to 4 nm), Eu takes too much oxygen and
destabilizes the STO layer, allowing for a solid-state reaction with Si that results in the
formation of EuSiy and silicates.
On the other hand, below 150 °C, Eu can only take a very limited amount of oxygen
and forms up to a ∼20 Å thick layer of EuO. Beyond this point, Eu no longer takes oxygen
from STO and Eu metal accumulates on the surface, as evidenced by the core level signal
changes in photoemission. Once the metal and oxide are brought into contact, the Fermi
level of Eu lines up with the charge neutrality level of EuO arising from the spin-up
evanescent state. Consequently, charge transfer of the 5d electrons from the Eu metal ∼15
Å inside EuO induces a downward band bending near the interface. This downward band
bending is confirmed experimentally by analyzing the XPS spectra of the Eu 4f peaks and
their broadening at the Eu/EuO interface.
Exploring the field-effect of the EuO/STO interface provides a new possibility that
we may gain control over the interface charge distribution thus achieving a new way of
modulating the 2DEG, and possibly shifting the Curie temperature to the range of practical
applications. The ARPES measurement for such a system encountered technical problems
with extremely weak signals and we are trying to work around these issues.
The EuO/SrTiO3 is a promising platform for combining a ferromagnetic material
and a 2-dimensional electron system. The oxygen scavenging process of growing EuO on
STO has the advantage of not needing to control oxygen to form stoichiometric EuO under
certain conditions. Our study elucidates the behavior of this growth process under different
conditions and thus provides for better understanding and control of this system.
84
4.6 ACKNOWLEDGEMENT
This research was partially supported by the National Science Foundation (NSF)
through the Center for Dynamics and Control of Materials: an NSF MRSEC under
Cooperative Agreement No. DMR-1720595 and by the Air Force Office of Scientific
Research (AFOSR) under Grant No. FA9550-18-1-0053. All calculations were performed
at the Texas Advanced Computing Center (TACC) by Lingyuan Gao. I greatly thank Prof.
Strocov for his support when I visited PSI.
4.7 REFERENCES
1. P. G. Steeneken, L. H. Tjeng, I. Elfimov, G. A. Sawatzky, G. Ghiringhelli, N. B.
Brookes, and D.-J. Huang, Phys. Rev. Lett. 88, 47201 (2002).
2. A. Schmehl et al., Nat. Mater. 6, 882 (2007).
3. A. A. Melville, T. Mairoser, A. Schmehl, D. E. Shai, E. J. Monkman, J. W. Harter, T.
Heeg, B. Hollander, J. Schubert, K. M. Shen, J. Mannhart, and D. G. Schlom, Appl.
Phys. Lett. 100, 222101 (2012).
4. B. T. Matthias, R. M. Bozorth, and J. H. Van Vleck, Phys. Rev. Lett. 7, 160 (1961).
5. J. Schoenes and P. Wachter, Phys. Rev. B. 9, 3097 (1974).
6. C. Caspers, M. Müller, A. X. Gray, A. M. Kaiser, A. Gloskovskii, C. S. Fadley, W.
Drube, and C. M. Schneider, Phys. Rev. B. 84, 205217 (2011).
7. A. G. Swartz, P. M. Odenthal, Y. Hao, R. S. Ruoff, and R. K. Kawakami, ACS Nano
6, 10063 (2013).
8. D. V. Averyanov, Y. G. Sadofyev, A. M. Tokmachev, A. E. Primenko, I. A. Likhachev,
and V. G. Storchak, ACS Appl. Mater. Interfaces 7, 6146–6152 (2015).
9. K. J. Kormondy, L. Gao, X. Li, S. Lu, A. B. Posadas, S. Shen, M. Tsoi, M. R.
McCartney, D. J. Smith, J. Zhou, L. L. Lev, M.-A. Husanu, V. N. Strocov, and A. A.
Demkov, Sci. Rep. 8, 7721 (2018).
10. M. Matsubara, A. Schroer, A. Schmehl, A. Melville, C. Becher, M. Trujillo-Martinez,
D. G. Schlom, J. Mannhart, J. Kroha, and M. Fiebig, Nat. Commun. 6, 6724 (2015).
11. M. Matsubara, A. Schmehl, J. Mannhart, D. G. Schlom, and M. Fiebig, Phys. Rev. B
81, 214447 (2010).
12. P. Lömker, T. C. Rödel, T. Gerber, F. Fortuna, E. Frantzeskakis, P. LeFèvre, F. Bertran,
M. Müller, and A. F. Santander-Syro, Phys. Rev. Mater. 1, 062001(R) (2017).
85
13. L. Gao and A. A. Demkov, Phys. Rev. B 97, 125305 (2018).
14. A. B. Posadas, K. J. Kormondy, W. Guo, P. Ponath, J. Geler-Kremer, T. Hadamek, and
A. A. Demkov, J. Appl. Phys. 121, 105302 (2017).
15. K. J. Kormondy, A. B. Posadas, T. Q. Ngo, S. Lu, N. Goble, J. Jordan-Sweet, X. P. A.
Gao, D. J. Smith, M. R. McCartney, J. G. Ekerdt, and A. A. Demkov, J. Appl. Phys.
117, 95303 (2015).
16. G. M. Prinz, T. Gerber, A. Lorke, and M. Müller, Appl. Phys. Lett. 109, 202401 (2016).
17. T. Mairoser, J. A. Mundy, A. Melville, D. Hodash, P. Cueva, R. Held, A. Glavic, J.
Schubert, D. A. Muller, D. G. Schlom, and A. Schmehl, Nat. Commun. 6, 7716 (2015).
18. J. Lettieri, V. Vaithyanathan, S. K. Eah, J. Stephens, V. Sih, D. D. Awschalom, J. Levy,
and D. G. Schlom, Appl. Phys. Lett. 83, 975 (2003).
19. R. W. Ulbricht, A. Schmehl, T. Heeg, J. Schubert, and D. G. Schlom, Appl. Phys. Lett.
93, 102105 (2008).
20. D. V. Averyanov, Y. G. Sadofyev, A. M. Tokmachev, A. E. Primenko, I. A. Likhachev,
and V. G. Storchak, J. Appl. Phys. 113, 17C505 (2013).
21. K. J. Hubbard and D. G. Schlom, J. Mater. Res. 11, 2757 (1996).
22. Y. Wei, X. Hu, Y. Liang, D. C. Jordan, B. Craigo, R. Droopad, Z. Yu, A. A. Demkov,
J. John, L. Edwards, and W. J. Ooms, J. Vac. Sci. Technol. B 20, 1402 (2002).
23. J. Lettieri, J. H. Haeni, and D. G. Schlom, J. Vac. Sci. Technol. A 20, 1332 (2002).
24. H. Miyazaki, T. Ito, H. Im, K. Terashima, S. Yagi, M. Kato, K. Soda, and S. Kimura,
Jpn. J. Appl. Phys. 48, 055504 (2009).
25. H. Miyazaki, H. Momiyama, T. Hajiri, T. Ito, K. Imura, M. Matsunami, and S. Kimura,
J. Phys. Conf. Ser. 391, 012047 (2012).
26. B. I. Edmondson, S. Liu, S. Lu, H. W. Wu, A. B. Posadas, D. J. Smith, X. P. A. Gao,
A. A. Demkov, and J. G. Ekerdt, J. Appl. Phys. 124, 185303 (2018).
27. EuO Crystal Structure: Datasheet from “PAULING FILE Multinaries Edition—2012”
in Springer Materials,
see: http://materials.springer.com/ isp/crystallographic/docs/sd_0553841.
28. R. A. McKee, F. J. Walker, and M. F. Chisholm, Phys. Rev. Lett. 81, 3014 (1998).
29. M. Choi, A. Posadas, R. Dargis, C. Shih, A. A. Demkov, D. H. Triyoso, N. D.
Theodore, C. Dubourdieu, J. Bruley, and J. Jordan-Sweet, J. Appl. Phys. 111, 064112
(2012).
30. H. Li, X. Hu, Y. Wei, Z. Yu, X. Zhang, R. Droopad, A. A. Demkov, J. Edwards, Jr.,
K. Moore, and W. Ooms, J. Appl. Phys. 93, 4521 (2003).
86
31. K. D. Fredrickson, H. Seo, and A. A. Demkov, J. Appl. Phys. 120, 065301(2016).
32. T. Hadamek, D. Shin, A. B. Posadas, A. A. Demkov, S. Kwon, Q. Wang, and M. Kim,
Appl. Phys. Lett. 111, 142901 (2017).
33. S. P. KowaIczyk, N. Edelstein, F. R. McFeely, L. Ley, and D. A. Shirley, Chem. Phys.
Lett. 29, 491 (1974).
34. J. K. Lang, Y. Baer, and P. A. Cox, J. Phys. F. 11, 121 (1981).
35. L. Gao, W. Guo, A. B. Posadas, and A. A. Demkov. Phys. Rev. Materials 3, 094403
(2019).
36. X.-X. Wang, H.-N. Li, W.-H. Zhang, and F.-Q. Xu, J. Phys.: Condens. Matter 19,
096001 (2007).
37. M. W. Shafer, J. B. Torrance, and T. Penney, J. Phys. Chem. Solids 33, 2251 (1972).
38. A. Mauger and C. Godart, Phys. Rep. 141, 51 (1986).
39. P. R. Subramanian, Eu-O (europium-oxygen), in Binary Alloy Phase Diagrams, 2nd
ed., edited by T. B. Massalski (ASM International, Materials Park, OH, 1990), Vol. 2,
pp. 1665–1666.
40. P. Lomker and M. Muller, Phys. Rev. Mater. 3, 061401(R)(2019).
41. S. Tanuma, C. J. Powell, and D. R. Penn, Surf. Interface Anal. 36, 1 (2004).
42. J. K. Lee, N. Sai, and A.A. Demkov, Phys. Rev. B 82, 235305 (2010).
43. P. Ponath, K. Fredrickson, A. B. Posadas, Y. Ren, X. Wu, R. K. Vasudevan, M. B.
Okatan, S. Jesse, T. Aoki, M. R. McCartney, D. J. Smith, S. V. Kalinin, K. Lai, and A.
A. Demkov, Nature Comm. 6, 6067 (2015).
44. W. Li, L. Gao and A. A. Demkov, Phys. Rev. B 102, 035308 (2020).
45. V. N. Strocov, X. Wang, M. Shi, M. Kobayashi, J. Krempasky, C. Hess, T. Schmitt
and L. Patthey, J. Synchrotron Radiat. 21, 32(2014).
46. V. N. Strocov, T. Schmitt, U. Flechsig, T. Schmidt, A. Imhof, Q. Chen, J. Raabe, R.
Betemps, D. Zimoch, J. Krempasky, X. Wang, M. Grioni, A. Piazzalunga and L.
Patthey, J. Synchrotron Radiat. 17, 631(2010).
47. J. Braun, J. Minár, S. Mankovsky, V. N. Strocov, N. B. Brookes, L. Plucinski, C. M.
Schneider, C. S. Fadley, and H. Ebert, Phys. Rev. B 88, 205409(2013)
87
Chapter 5: Thermal oxidation of Si buried under thin SrTiO3 and
free-standing SrTiO3 membranes
In this chapter, we first present the dry oxidation of Si (001) beneath a thin epitaxial
SrTiO3 (STO) layer using furnace annealing in flowing oxygen. A 10-nm layer of STO is
epitaxially grown on Si with no SiO2 interlayer. For such a structure, an annealing
temperature of 800ºC was found to be the limiting temperature to prevent silicate formation
and disruption of the interface structure. The effect of annealing time on the thickness of
the SiO2 layer was investigated. In situ x-ray photoelectron spectroscopy (XPS) and
reflection-high-energy electron diffraction (RHEED) were used to ensure that the quality
of STO is unchanged after the annealing process. The experimental annealing data is
compared with a theoretical oxygen diffusion model based on one due to Deal, Grove and
Massoud. The model fits the experimental data well, indicating that oxygen diffusion
through the STO layer is not the limiting factor. One can therefore readily control the
thickness of the SiO2 interlayer by simply controlling the annealing time in flowing
oxygen. Using this ability to oxidize the underlying Si, we then show some preliminary
results on the fabrication of free-standing STO membranes, which requires thick SiO2
interlayers in order to properly detach the STO from the Si on which it is grown.
A portion of this chapter is published in: W. Guo, A. B. Posadas, A. A. Demkov, J.
Appl. Phys. 127, 055302 (2020). Our collaborators in Prof. Li’s group at UT Austin are
responsible for the Raman spectroscopy measurements.
5.1 INTRODUCTION TO DRY OXIDATION OF SI
STO is a widely used substrate for metal oxide thin film growth [1], [2]. It has a
rather large dielectric constant (~300) [2] making it attractive for dielectric applications.
The lattice constant of STO (3.9 Å) also makes it a suitable substrate material for the
88
epitaxtial growth of many oxides with perovskite, rocksalt, and spinel crystal structures
because of the small lattice mismatch to many such materials, such as LaAlO3 [2]-[4],
BaTiO3 [5]-[7], EuO [8], [9], γ-Al2O3 [10], etc. The discovery by Mckee et al. in 1998 that
one can epitaxially nucleate STO on Si directly without forming SiO2 has opened epitaxial
oxide thin films to the potential for technological development [11]. This process has been
further developed and studied both experimentally and theoretically by several groups [11]-
[14]. Even though STO has excellent dielectric properties making it attractive for use as a
gate oxide in field effect transistors [15], the conduction band offset at the STO/Si
interface is essentially zero, making such an application moot [16], [17]. However, STO
on Si can serve as a bridge material for integrating other oxides epitaxially on Si by serving
as a pseudo-substrate [6], [7], [9], [14]. It is therefore important to understand the behavior
of this system under various temperature and oxygen pressure environments, particularly
with respect to the oxidation of the underlying Si.
Silicon oxidation is a well-studied process that has been long-discussed following
the establishment by Deal and Grove of the oxygen diffusion model for the Si surface in
1965 [18]. This oxygen diffusion and reaction model of Si oxidation agrees well with the
experimental data at different temperatures for relatively thick SiO2 layers (>50 nm). The
Deal-Grove model predicts the resulting oxide thickness for a given temperature as a
function of time. However, significant deviations were found for thinner oxide films.
Massoud improved the Deal-Grove model for the thin SiO2 layer regime by adding an
exponential decay term to the original linear-parabolic model [19]-[21]. Owing to the good
dielectric properties of SiO2 and a well-behaved Si-SiO2 interface, controlled oxidation of
Si has played a fundamental role in the fabrication of Si devices such as metal-oxide-
semiconductor field-effect transistors (MOSFET) and single-electron devices [22]-[26].
89
With epitaxial STO covering the Si surface, the oxidation behavior of Si may not
be the same and it is important to know how the processing conditions, such as temperature
and oxygen pressure, affect the oxidation rate of silicon with an STO overlayer present. A
typical process of forming STO on Si involves first removing the surface SiO2 and then
depositing 0.5 monolayer of Sr in a molecular beam epitaxy (MBE) chamber. This
submonolayer Sr prevents Si oxidation during the initial nucleation of STO [12], [13]. It is
then followed by 4-5 unit cells (uc) of STO deposition at low temperature (<300°C) which
is fully crystallized at ~500°C. Once the initial template is crystallized, additional STO can
be grown, treating the template as an STO substrate. Continuing the growth at oxygen
pressures where STO is fully oxidized (~5×10-7 Torr) and temperatures where STO is
crystalline as deposited (~500°C) typically results in some oxygen diffusing through STO
and partially oxidizing the underlying Si. There have been several studies of Si oxidation
and interlayer reaction at the STO/Si interface [14], [15], [27]. It has been shown that STO
is not fully thermodynamically stable in direct contact with Si at very high temperature
under ultrahigh vacuum conditions [28]. Choi et al. grew STO/Si by MBE and used post-
deposition annealing in oxygen (10-7 to 10-5 Torr) to control the strain relaxation of STO
[14]. Cross-section transmission electron microscopy (TEM) was used to monitor the
interlayer structure and thickness under different oxygen partial pressures and annealing
times. Goncharova et al. discussed the thermal stability of the STO/Si interface [15]. They
considered the possible reactions that could happen between the layers at different
annealing temperatures such as STO + Si → SiO + SrO + O2 + TiSi2. A thin interlayer
composed of SrSiOx, SrO, and TiSix was found at annealing temperatures as low as 550°C.
At 850°C or even higher temperatures, the STO film decomposes completely, leaving
behind only TiSix islands [29]. Yong et al. have discussed the thermal stability and possible
90
interface reactions of the STO/SiO2/Si interface [27]. For example, Si + SrTiO3 → SrSiO3
+ TiSi2, and Si + TiO2 → SiO2 + TiSi2. Optical microscopy and scanning electron
microscopy (SEM) were used to observe the surface morphology changes of the STO film
after annealing at ~800°C. Eisenbeiser et al. used TEM to show the interfacial layer
between STO and Si after the growth [24].
Here, we study the dry oxidation behavior of the buried Si as a function of annealing
time, and determine the maximum temperature for which a relatively thin STO layer (10
nm) remains intact. The STO is found to be stable during the oxidation anneal at 800°C for
up to 10 hours. We start with the MBE growth of STO on Si (001) and then perform a
flowing oxygen anneal in a tube furnace to oxidize the Si underneath. The STO thickness
is fixed at 10 nm (25 unit cells (uc)) for this study. Too thin an STO layer (<10 uc) results
in STO and Si reacting. The practical annealing temperature is found to be 800°C. Below
this temperature, dry oxidation is very slow and impractical; above this temperature, the
STO in contact with Si is not thermally stable. We developed a theoretical model based on
a modification of the Deal-Grove-Massoud formalism that predicts the thickness of SiO2
formed underneath STO as a function of time and temperature, and report a robust recipe
for dry oxidation of Si buried under an epitaxial layer of STO.
5.2 SI OXIDATION EXPERIMENTS
The STO/Si growth is performed in a customized DCA 600 MBE system with a
base pressure of 6×10-10 Torr. P-type doped Si substrates of 20×20 mm2 size are cut from
a prime Si wafer and degreased ultrasonically in acetone, isopropanol, and deionized (18
MΩ/cm) water for 5 minutes each and then exposed to UV/ozone to remove carbon from
the surface. The Si substrates are then annealed in ultrahigh vacuum at 700°C for 10 min
91
followed by Sr-assisted de-oxidation before the growth of a 2-nm-thick STO epitaxial film
[11]-[14]. Prior to the growth of STO, 1/2 ML of Sr is formed on the Si (001) surface. This
Zintl layer has 2×1 symmetry and serves as a template for further deposition [11]. The
initially amorphous STO film is formed by co-deposition of Sr and Ti at low temperature
(200°C) under low oxygen pressure (8×10-8-5×10-7 Torr) and is annealed in vacuum at or
above 550°C to crystallize. This procedure results in no interfacial SiO2 as shown by x-ray
photoemission and transmission electron microscopy [12]. There are two possible ways of
growing additional STO. The first one is to deposit additional amorphous STO near room
temperature in oxygen and anneal it in vacuum to the STO crystallization temperature
(550°C). This will not result in formation of the SiO2 interlayer. Another way is to perform
a co-deposition of Sr and Ti under modest oxygen pressure (~mid 10-7 Torr) at 550°C [14]
(as if growing on an STO substrate). Because of the high oxygen diffusivity in STO,
oxygen can diffuse through and oxidize Si underneath without disrupting the crystal
structure of the already crystallized STO. This results in a very thin SiO2 layer (~2 nm)
between STO and Si. In this study, we use the first method to prepare a SiO2-free interface
prior to annealing. Reflection-high-energy electron diffraction (RHEED) with 21 keV
electrons is used to record the STO crystallinity during growth. In situ x-ray photoelectron
spectroscopy (XPS) using monochromated Al Kα radiation and a VG Scienta R3000
hemispherical electron energy analyzer is used to check stoichiometry of STO. After
annealing, the STO/SiO2/Si samples are then measured using J.A. Woollam M-2000DI
spectroscopic ellipsometer to determine the thickness of SiO2. The instrument uses a
combination deuterium / quartz tungsten halogen lamp as the light source and covers a
wavelength range from 190-1650 nm. The data is collected at three different angles of
incidence (45°, 50°, and 55°). The fit is performed using the built-in software,
92
CompleteEase, with three layers (Si substrate, thermal oxide SiO2, and bulk STO) over the
entire wavelength range of the instrument. For the bulk STO optical constants, we use the
parametrized optical constants determined by the Zollner group at New Mexico State
University [30]. All three angles of incidence are fit simultaneously using N, C, S fit
weighting. We show an example of a fit of the ellipsometry data in Figure 5.1. The
measured data are shown as circles and the fit curves as solid lines. The red one is the
curve and the blue one is the curve.
Figure 5.1 Example ellipsometry measurement with the corresponding model fit. These
are the measurement results at 45°. This gives a STO layer with 17.54 nm
and SiO2 layer with 7.54 nm thickness. The red one is the curve and the
blue one is the curve. Measured data are shown as open circles and the fit
curves as solid lines.
93
After a 10 nm-thick STO layer is grown in the MBE system (RHEED is shown in
Figure 5.2(a)), the sample is taken out of UHV and ultrasonically cleaned by IPA and DI-
water and inserted into a GSL-1700X-S tube furnace from MTI Corporation. Dry oxygen
flows through the furnace tube during the entire annealing cycle with approximately 1
L/min flow rate at atmospheric pressure. We use dry oxidation to avoid a possible reaction
between STO and water [31]. STO/Si samples are placed in the center of the furnace tube
with the STO side facing up. The tube is heated gradually from room temperature to the
final temperature at 10°C/min. The sample is kept at this temperature for the designated
period of time after which the sample is then cooled to room temperature over a period of
two hours. We verified that the STO film still has good crystallinity after the anneal by
RHEED, which is shown in Figure 5.2(b).
94
Figure 5.2: (a) RHEED image for a 10 nm STO/SiO2/Si after the growth and before
anneal. STO pattern is shown along the [110] direction. (b) RHEED image
for a 10 nm STO/SiO2/Si after 800°C anneal for 2 hours. STO pattern along
the [110] direction is still sharp and clear.
To check the interface composition, we grew a relatively thin STO/Si sample with
5 nm of STO and checked it with XPS. The Si 2p spectrum after the 800°C anneal is shown
in Figure 5.3. It shows a major SiO2 peak and a minor SiOx peak, with no TiSi2 signal
visible.
95
Figure 5.3: XPS spectrum for the Si 2p region. The major peak around 104.5 eV is SiO2
and the minor peak around 103 eV is SiOx, Si metal (99.5 eV) is also still
visible.
It is known that it takes 15 hours to grow a 10 nm-thick SiO2 layer on bare Si [32]
at 700°C. The SiO2 layer will grow even more slowly at lower temperature (<700°C)
because of the reduced oxygen diffusion rate. Most Si oxidation data in the literature are
from the practically useful 800°C-1300°C temperature range. However, there are also
papers reporting degradation of the STO/Si structure at high temperature (~1000°C) when
STO reacts with SiO2 and Si [15], [29]. Therefore, the practical temperature range to have
an acceptable oxidation rate without destroying STO is ~700-900°C. The annealing
96
temperature we use is 800°C, which on the one hand is not high enough to degrade the
STO layer but on the other hand, allows the Si oxidation time to remain practical. Under
this annealing temperature, we compared the out-of-plane XRD full scan spectra before
and after annealing for 4 hours (Figure 5.4). We find that the STO film only shows a little
degradation, with the FWHM of rocking curve of the STO (002) peak becoming slightly
broadened from 0.33° to 0.54°.
Figure 5.4: Comparison of the XRD out-of-plane full scans before and after annealing.
STO peaks decrease a little and Si is a little higher. All peaks do not show
any obvious deformation other than intensity variation.
97
5.3 THEORETICAL MODEL OF SI OXIDATION
The standard model of Si oxidation has been proposed almost half a century ago by
Deal and Grove [18]. It is based on describing three steps that result in oxidation: (i) oxygen
(or oxidant species in general) transport from the gas phase to the oxide surface where it is
adsorbed, (ii) oxygen transport through the oxide layer towards Si, and (iii) the interface
oxygen reaction with Si and formation of a new layer of SiO2. However, the model has a
well-known difficulty in predicting the initial stage of oxidation for thin films. Massoud et
al. made modifications to the Deal-Grove model to overcome this issue [19]-[21]. In the
case where an STO overlayer is present, we have to modify the model further as the initial
oxide in our case is not SiO2 but STO, and at later stages oxygen has to diffuse through
both materials in order to reach Si.
Following the Deal-Grove-Massoud logic, we set the oxygen concentration in
different regions and connect them by the oxygen diffusion flux. In Figure 5.5, we show a
schematic of the oxygen propagation through the structure. Here C* is the oxygen
concentration of the gas, C0 is the oxygen concentration at the surface of STO, C1 is the
oxygen concentration at the STO/SiO2 interface, and C2 is the oxygen concentration at the
SiO2/Si interface. X1 and x2 are the thickness of STO and SiO2, respectively. F1 is the
incoming oxygen flux, F2 is the flux inside STO, F3 is the flux inside SiO2, and F4 is the
oxidation rate at the Si interface with SiO2. The steady-state condition is discussed in the
original paper [18].
98
Figure 5.5: The schematic of oxygen propagation in the SrTiO3/SiO2/Si structure. The
inset shows the atomic structure schematic.
In the original Si oxidation model [18], one has:
𝐹1 = ℎ(𝐶∗ − 𝐶0) (1)
where h is the transport coefficient (from gas into STO). The diffusion flux
𝐹2 = −𝐷1𝑑𝐶
𝑑𝑥 (2)
99
is given by Fick’s Law, and because the gradient is linear, we have:
𝐹2 = 𝐷1(𝐶0−𝐶1)
𝑥1, 𝐹3 = 𝐷2
(𝐶1−𝐶2)
𝑥2, 𝐹4 = 𝑘𝐶2, (3)
where D1 is the diffusivity of oxygen in STO, D2 is the diffusivity of oxygen in
SiO2, and k is the reaction rate of Si oxidation.
Since all fluxes are uniform, we set 𝐹1 = 𝐹2 = 𝐹3 = 𝐹4. Solving this system of
equations, we obtain:
𝐶0
𝐶∗ =1+
𝑘𝑥1𝐷1
+𝑘𝑥2𝐷2
1+𝑘
ℎ+
𝑘𝑥1𝐷1
+𝑘𝑥2𝐷2
, 𝐶1
𝐶∗ =1+
𝑘𝑥2𝐷2
1+𝑘
ℎ+
𝑘𝑥1𝐷1
+𝑘𝑥2𝐷2
, 𝐶2
𝐶∗ =1
1+𝑘
ℎ+
𝑘𝑥1𝐷1
+𝑘𝑥2𝐷2
, (4)
The diffusivities of oxygen in STO and silica are compared in Figure 5.6. We
combine high temperature (>700°C) diffusivity data of STO [33] from literature and
extrapolate to lower temperatures (300-700°C) to find estimated diffusivities at those
temperatures. One can see that in the temperature range from 300°C to 1200°C, oxygen
diffusivity in STO, D1, is always at least three orders of magnitude larger than that in silica,
D2. Thus one can neglect the contribution coming from the term 𝑘𝑥1
𝐷1 in the following
discussion. Therefore:
𝑑𝑥2
𝑑𝑡=
𝐹
𝑁1=
𝑘𝐶∗
𝑁1⁄
1+𝑘
ℎ+
𝑘𝑥2𝐷2
=𝐵
𝐴+2𝑥2 , (5)
N1 here is the oxygen needed to oxidize a unit volume of Si, A and B are defined
as:
100
𝐴 = 2𝐷2 (1
𝑘+
1
ℎ) , 𝐵 = 2
𝐶∗𝐷2
𝑁1. (6)
Figure 5.6: Diffusivity of oxygen in SrTiO3 (D1) and SiO2 (D2) in the 300-1200°C
temperature range. Inset: Diffusivity of oxygen in SrTiO3 and SiO2 in the
300-700°C temperature range. The STO low temperature (<700°C) data is
obtained by inverse relationship projection from existing diffusivity at
higher temperature (>700°C)
After the modifications of the Massoud model [19], [20], we obtain:
𝑑𝑥0
𝑑𝑡=
𝐵
𝐴+2𝑥0+ 𝐶1𝑒−𝑥0/𝐿1 + 𝐶2𝑒−𝑥0/𝐿2 (7)
101
where A, B, Ci, and Li, can be fitted from the experimental values and are listed in
Massoud’s paper [19], [20].
The steady-state picture can be justified as follows. Consider a sudden change in
oxygen concentration in the silica layer, the total time it takes for the concentration to come
back to a stable value is ∆t:
∆𝑡 =total oxygen needed to come back to original concentration
oxygen flux (8)
Assuming that the total amount of oxygen needed is the same as the amount of
oxygen that must flow out of the layer, the numerator is simply 1
2(𝐶1 − 𝐶2)𝑥2, and the
oxygen flux is 𝐹 = 𝐷2(𝐶1−𝐶2)
𝑥2. Therefore,
∆𝑡 =1
2(𝐶1−𝐶2)𝑥2
𝐷2(𝐶1−𝐶2)
𝑥2
=𝑥2
2
2𝐷2. (9)
Here x2 is on the order of nm and D is on the order of ~108 nm2/hr, so ∆t is of the
order of 10-8 hour. Thus we can assume that the flow of oxygen through the layer is
established very quickly and can be assumed to be steady for all practical experimental
conditions.
This suggests that we can treat the problem as oxidation of bare Si since STO is
essentially transparent to oxygen diffusion compared to SiO2. Using the Massoud model
[20], [21], we have
𝑑𝑥0
𝑑𝑡=
𝐵
𝐴+2𝑥0+ 𝐶1𝑒−𝑥0/𝐿1 + 𝐶2𝑒−𝑥0/𝐿2 . (10)
Comparing with the experimental data, the formula can be re-written as:
102
𝑑𝑥0
𝑑𝑡=
𝐵+𝐾1𝑒−𝑡/𝜏1+𝐾2𝑒−𝑡/𝜏2
𝐴+2𝑥0 (11)
Solving this differential equation, we obtain:
𝑥0 =𝐴
2+ √(
𝐴
2)
2
+ 𝐵𝑡 + 𝑀1 [1 − exp (−𝑡
𝜏1)] + 𝑀2[1 − exp (−
𝑡
𝜏2) + 𝑀0
𝑀0 = 𝑥𝑖2 + 𝐴𝑥𝑖 , 𝑀1 = 𝐾1𝜏1, 𝑀2 = 𝐾2𝜏2, (12)
where xi is the initial thickness of SiO2, with the parameters given by:
𝐵 = 𝐶𝐵exp (−𝐸𝐵
𝑘𝑇),
𝐵
𝐴= 𝐶𝐵/𝐴exp (−
𝐸𝐵/𝐴
𝑘𝑇)
𝐾1 = 𝐾10 exp (−
𝐸𝐾1
𝑘𝑇), 𝐾2 = 𝐾2
0 exp (−𝐸𝐾2
𝑘𝑇)
𝜏1 = 𝜏10exp (
𝐸𝜏1
𝑘𝑇), 𝜏2 = 𝜏2
0exp (𝐸𝜏2
𝑘𝑇) (13)
The constants are taken from Massoud [21] and are listed in Table 2 for
convenience.
103
Crystal Orientation (100) (111) (110)
CB [nm2/min] 1.70×1011 1.34×109 3.73×108
EB [eV] 2.22 1.71 1.63
CB/A [nm/min] 7.35×106 1.32×107 4.73×108
EB/A [eV] 1.76 1.74 2.10
𝐾10 [nm2/min] 2.49×1011 2.70×1019 4.07×108
EK1 [eV] 2.18 1.74 1.54
𝐾20 [nm2/min] 3.72×1011 1.33×109 1.20×108
EK2 [eV] 2.28 1.76 1.56
𝜏10 [min] 4.14×10-6 1.72×10-6 5.38×10-9
Eτ1 [eV] 1.38 1.45 2.02
𝜏20 [min] 2.71×10-7 1.56×10-7 1.63×10-8
Eτ2 [eV] 1.88 1.90 2.12
Table 2. Oxidation parameters of the Massoud model [21] for temperature less than
1000°C.
Based on these parameters in (12) and (13) we can estimate the silica thickness (x0)
for a given set of annealing time and temperature. We plot the silica thickness as a function
of oxidation time at different temperatures in Figure 5.7. The red diamond shapes in Figure
5.7 are the experimental values from the tube furnace oxidation at 800°C measured by
ellipsometry. We also performed experiments at 750°C and 700°C to verify the prediction
of the model for other temperatures. We want to emphasize that the model described above
is derived based on the Deal-Grove-Massoud model using established parameters
appropriate for Si/SiO2 with no adjustable parameters. We use these established parameters
to compare the model to the measured interfacial SiO2 thicknesses and find reasonable
104
agreement between the model predictions and the data, especially at 800°C, with somewhat
worse fits for the lower temperatures measured.
Figure 5.7: The SiO2 thickness as a function of oxidation time at different temperatures
from equations (12) and (13). The lines represent the model and the shapes
represent the experimental data. Diamonds are samples annealed under
800°C, circle is under 750°C and square is under 700°C. All thickness data
have an error bar of ~1 nm.
The uncertainty in the SiO2 thickness comes from the particulars of the fitting of
the ellipsometry data and the fluctuations present in the annealing process, particularly the
oxygen flow rate. The fluctuations of the oxygen flow rate were found to cause drifts in
the surface temperature of the sample resulting in oxidation thickness variations. We
performed three anneals with the same condition and measured the SiO2 thickness of each.
105
The samples have the same structure before annealing and we controlled the oxygen flux
at the same level. From this, we determine the random error in the SiO2 thickness under
800°C from the oxidation process to be 0.8 nm. Additionally, there is uncertainty due to
the particulars of the ellipsometry fitting of up to 1 nm due to different wavelength range
of fitting and different fit weighting. By choosing proper parameters, the error from the
data fitting can be reduced to 0.1~0.2 nm. From this, we set the error bar in our plot to be
1 nm. Lower temperatures (under 800 °C) and longer anneal times (> 1 hour) would be
expected to result in slightly larger errors as a result of the accumulation of uncertainty
from the temperature variation due to the fluctuation in oxygen flow rate. This could
explain why the model works well for 800°C and not so well for the lower temperatures.
At 800°C, the oxidation rate is not prohibitively slow, while the temperature would
not cause the STO and Si to react and form SrSiO3 or TiSi2 as has been previously reported
[15], [29], [30]. The data agrees reasonably well with the model, indicating that one can
estimate the dry oxidation time needed to obtain a desired thickness of SiO2 while
maintaining a thin single crystal STO film on top. We have explored temperatures higher
than 800ºC in the range of 900-1200ºC. Unfortunately, at these higher temperatures the
STO layer breaks up and reacts with the Si substrate. However, dry oxidation at 800°C is
a robust method to produce interfacial SiO2 layers in the thickness range of 10-50 nm. We
have also explored other STO thickness (5, 20, 40 nm) and have found that the STO film
thickness does not matter if the STO thickness remains under 100 nm. This is confirmed
by the model, which shows that STO thickness is not important over a wide range of values.
106
5.4 FABRICATION OF FREE-STANDING STO MEMBRANES
Based on the ability to control the SiO2 interlayer of STO/Si samples, we are able
to make free-standing STO membranes by etching away the SiO2 layer, which works if the
SiO2 layer is thick enough.
Two-dimensional oxide materials exhibit physical properties that are emergent or
different from the bulk. This opens new possibilities for applications. STO is a widely-used
substrate in thin film growth, making it a very good bridge material for epitaxial growth of
various oxides like LaAlO3, BaTiO3, etc. Here, we are interested in any potential new
physics phenomena that may arise in these quasi-2-dimentional STO membranes.
Recently, Hwang’s group [34], [35] successfully fabricated free-standing STO
membranes via liftoff using a water-soluble template. They used PLD to grow Sr3Al2O6
(SAO) templates, and showed that this material can dissolve in water within seconds. A
group from Nanjing University managed to use the same method to produce monolayer-
thick free-standing STO membranes with mm lateral size [36].
The SAO water-soluble method is very attractive and have attempted to grow it in
our MBE system. While we were able to grow SAO compound on STO substrate and
confirmed that it is water-soluble, we have not been able to obtain a sufficiently good
crystalline SAO layer for epitaxial STO to grow on it. We instead developed another liftoff
process as follows. We first grow ~20 nm thick STO on Si substrate and then use oxygen
annealing to create a 2 - 10 nm-thick SiO2 layer in between STO and Si. We use a diamond
scriber to make scratches on the samples and use 4M NaOH solution to etch away the SiO2
and Si around the scratch. We perform the etching on a heating plate with a magnetic
spinner. Warming up solution and using a spinner to agitate it will speed up the etching
process and minimize potential damage of the solution to STO. This process is indeed able
107
to yield free-standing STO membranes along the scratch edge. We show the schematic of
the procedure in Figure 5.8.
Figure 5.8: Schematic of the procedure to fabricate STO membranes
As Figure 5.8 shows, STO membranes are hanging along the edges of the scribed
areas. We optimize STO and SiO2 thickness to achieve better etch results. Heating
temperature and spinning speed are also optimized. We managed to get STO membranes
with sizes of 5-8 m wide and 10-20 m in length. We show the SEM picture of the
membranes in Figure 5.9. We can see that the membranes are hanging or curled back along
the STO “cliff” from the scratched area. We use yellow lines to mark a membrane with 15
× 15 μm2 in Figure 5.9(b). This size level is the largest we have made using our fabrication
procedure. The typical size we obtain for 10-40 nm-thick STO membranes is ~5×10 µm2.
108
Figure 5.9: (a) (b) are SEM images from different spots on STO/Si etched sample. The
biggest membrane outlined by yellow is ~15 × 15 μm2
5.5 MEASUREMENTS OF STO MEMBRANES
STO membranes are expected to have somewhat different phonon vibrational
modes when compared with STO bulk and thin films since they have different boundary
109
conditions. We use Raman spectroscopy to measure these phonon modes and try to
understand the new vibrational behavior of the membranes.
Raman spectroscopy uses the inelastic scattering of photons when a material is
irradiated by an intense a beam of light. Photons will interact with molecular vibrations,
phonons or other excitations in the material and shift them to higher or lower energies. The
shifted energies will provide information on vibration modes in the material. We use a
Micro-Raman instrument to perform Raman spectroscopy on our STO membranes and
compare the results with bulk and thin film STO Raman spectra. Prof. Li’s group in UT-
Austin performed the measurements shown below. Here, I present only some preliminary
results. There is a lot more research that can done in the future in this area.
First, we performed Raman on a bulk STO substrate to calibrate the system since
this spectrum is well-known and studied [37]-[42]. Bulk STO crystal is centrosymmetric
at room temperature. The degrees of freedom consist of one Flu triply degenerate acoustic
mode, three Flu and one F2u triply degenerate optical modes [38]. Flu and F2u are not Raman
active, and no first-order Raman peaks should exist in defect-free STO single crystal [38].
However, impurities and defects in bulk crystal will still bring first-order Raman-active
modes in the spectrum [37], [41]. We show the data for bulk STO in Figure 5.10. Every
characteristic peak is labeled with position values and marked by red arrows. Here, the 175
cm-1 peak corresponds to the first order TO2 mode [37], [41] with a clear Fano asymmetry.
The rest of the peaks are assigned to possible second order modes as shown in Table 3 [38],
[39]. The bulk STO Raman spectrum measured in our system is consistent with data
published in the literature.
110
Figure 5.10: Raman spectrum of a bulk STO substrate. Characteristic peak positions are
labeled and marked with red arrows.
Energy shift (cm-1) Assignment
250 2TA
2TO1
TO1+TA
315 TO2+TA
TO2+TO1
TO4-TO2
2LA
111
362 2TO2
TO4-TA
TO4-TO1
LO1+TA
620 TO4+TA
TO4+TO1
LO2+TO3
675 2TO3
TO1+LO3
TO4+LA
TA+LO3
712 TO4+TO2
Table 3. Phonon branch assignments for second-order Raman peaks measured from STO
bulk substrate.
We then performed a Raman measurement on a 20 nm as-grown STO thin film on
Si substrate. STO thin films were studied in the early 2000s and compared with the bulk
STO spectrum [37], [40]. Forbidden zero-center optical phonons are clearly observed in
STO thin films. This was attributed to the interaction of the polar mode with polarization
fluctuations in micropolar regions [37]. We produced and measured an MBE-grown STO
thin film sample on Si as a comparison. The results are shown in Figure 5.11 with all peaks
labeled by their energy positions. The sharp, saturated signal at 521 cm-1 is the only Si-
related peak.
112
Figure 5.11: Raman spectrum of 20 nm STO thin film on Si. (Thin film grown by MBE)
First order peaks appear at 163, 265, 482, 551, 792 cm-1 for TO2, TO3, LO2, TO4,
LO4, respectively [37], [43]. There is a temperature dependence for first order Raman peaks
and lower temperatures will give sharper line shapes [37], [43]. Our measurement is
performed at room temperature and our first order peaks are not as well-expressed as they
are at low temperature. The main characteristic peak is at 302 cm-1 and can be assigned to
either 2LA or TA+TO2. This is the only second order peak in a thin film that still has strong
intensity. Other second order peaks that appear in the bulk STO Raman spectrum have
decreased to small peaks. As shown in Table 3 above, the 367 cm-1 peak corresponds to
2TO2 and other assignments at the same position. The 617 cm-1 peak corresponds to
TO4+TA and other assignments at the same position. The 670 cm-1 peak corresponds to
113
2TO3 and other assignments at the same position. The 712 cm-1 peak corresponds to
TO4+TO2. The 282 cm-1 peak corresponds to LO3-LO1 [39], which is not present in the
bulk STO Raman spectrum. The closest assignment for the 198 cm-1 peak is 2TA [39],
which represents a shift of 20 cm-1. The closest assignment for the 433 cm-1 peak is
TO3+TA or LO2+LA, but this peak is normally present only in measurements at 4 K on
bulk STO [39]. Peaks at 763, 866 cm-1 don’t have proper first or second order assignments.
The main reasons for different behaviors between bulk and thin film STO are the
stress, lattice mismatches or vacancies [37], [40]. TO phonon peaks in the thin film
measured spectrum indicate a lowering of the crystal symmetry in the films, inversion
breaking, and/or translational symmetries [37]. With different boundary conditions and
defect distribution, we are naturally curious what a quasi-2-dimensions free-standing
membrane with will have in its Raman spectrum. Using the process described in the last
section, we managed to make sufficiently large STO free-standing membranes for Raman
measurement. By focusing the laser on the free-standing STO film, we are able to get a
spectrum with a weak STO signal, as shown in Figure 5.12. The low intensity is due to the
small thickness of membranes (20 nm). Better optimization of the equipment can also
improve the total intensity and lower the noise level in the future.
114
Figure 5.12: Raman signal of STO membranes made from 20 nm STO/Si
Here, the 521 cm-1 peak is the Si peak. The major peak at 302 cm-1 in supported
thin film STO almost disappears (only a very tiny peak remains). The STO membrane
Raman spectrum is dominated by a 280 cm-1 peak corresponding to LO3-LO1 [39]
(identical peak with STO thin film Raman). First order peaks show up at 265, 481, 791 cm-
1 for TO3, LO2, LO4, respectively [37], [43]. The TO4 mode disappears in the membrane
spectrum compared with the supported thin film STO Raman spectrum. For second order
peaks as shown in Table 3 [37], [38], the 240 cm-1 peak corresponds to 2TA and other
assignments at the same position; the 367 cm-1 peak corresponds to 2TO2 and other
assignments at the same position; the 604 cm-1 peak corresponds to TO3+LO1 [38]; the 617
cm-1 peak corresponds to TO4+TA and other assignments at the same position.
115
Interestingly, the 240 cm-1 peak disappears in supported thin film spectrum but is present
in the membrane spectrum. There is a new, unidentified peak at 571 cm-1 in the membrane
spectrum with relatively high intensity and have not found any assignments for it. The 866
cm-1 peak appears in both STO thin film and membrane spectra but there isn’t a proper
assignment for this peak, either. These need further experiments to reveal the nature of
these modes.
The STO membrane measurement has weak signals with a high noise level. The
range of energy shifts observed is also limited by the optical setup to 150-900 cm-1. Further
optimization of the measurement process is needed to obtain a better and wider spectrum.
5.6 CONCLUSIONS
In this chapter, we first presented dry oxidation of STO/Si heterostructures and
demonstrated that the underlying Si can be safely oxidized at a relatively high temperature
(800ºC) with the STO crystallinity not significantly degraded. We deposited 10-nm of
epitaxial STO on Si and performed flowing oxygen anneals at 800ºC. The SiO2 thickness
is measured by ellipsometry and compared with our Deal-Grove-like oxidation model and
found good agreement between the data and the model. We can use this model to predict
the temperature and time needed to obtain the desired SiO2 thickness for Si that is covered
by a thin layer of STO. This additional knob for controlling the layer structure can enable
one to integrate more complicated oxide structures on this STO/Si pseudo-substrate,
especially for applications requiring complete decoupling between the STO and Si.
The study of Si oxidation under STO provides a precise controllable method of
SiO2 formation in STO/Si and we managed to make free-standing STO membranes based
on that capability. We used NaOH etch to make free-standing STO membranes with typical
116
size of ~5×10 µm2. Raman spectroscopy was performed on STO thin film and membranes.
Preliminary results show weak signals from STO membranes. In the future we will work
on the optimization of the Raman setup to reduce noise and amplify the STO signal. We
can also use different thicknesses of STO to study the phonon modes of a confined quasi-
2D STO membrane. With the high-k features of STO, we can also try to design new
transistor devices using our STO membranes. For example, we have come up with an
architecture replacing graphene or hBN membranes by STO with a higher dielectric
constant to build tunnel field-effect transistors [43], [45].
5.7 ACKNOWLEDGEMENTS
This research was partially supported by the National Science Foundation through
the Center for Dynamics and Control of Materials: an NSF MRSEC under Cooperative
Agreement No. DMR-1720595 and by the Air Force Office of Scientific Research under
Grant FA9550-18-1-0053. We thank Prof. Li’s group for performing the Raman
measurements.
5.8 REFERENCES
1. B. Reihl, J. G. Bednorz, K. A. Műller, Y. Jugnet, G. Landgren, and J. F. Morar, Phys.
Rev. B 30, 803 (1984).
2. A. A. Demkov and A. B. Posadas, Integration of Functional Oxides with
Semiconductors (Springer, New York, 2014).
3. J. E. Ortmann, N. Nookala, Q. He, L. Gao, C. Lin, A. B. Posadas, A. Y. Borisevich, M.
A. Belkin, and A. A. Demkov, ACS NANO 12, 7682 (2018).
4. J. E. Ortmann, A. B. Posadas and A. A. Demkov, J. Appl. Phys. 124, 015301 (2018).
5. P. Ponath, A. B. Posadas, M. Schmidt, A.-M. Kelleher, M. White, D. O’Connell, P.
Hurley, R. Duffy and A. A. Demkov, J. Vac. Sci. and Technol. B 36 (2018).
6. K. D. Fredrickson, V. V. Vogler-Neuling, K. J. Kormondy, D. Caimi, F. Eltes, M.
Sousa, J. Fompeyrine, S. Abel, and A. A. Demkov, Phys. Rev. B 98, 075136 (2018).
117
7. K. J. Kormondy, Y. Popoff, M. Sousa, F. Eltes, D. Caimi, M. D. Rossell, M. Fiebig, P.
Hoffmann, C. Marchiori, M. Reinke, M. Trassin, A. A. Demkov, J. Fompeyrine, and
S. Abel, Nanotechnology 28, 075706 (2017).
8. K. J. Kormondy, L. Gao, X. Li, S. Lu, A. B. Posadas, S. Shen, M. Tsoi, M. R.
McCartney, D. J. Smith, J. Zhou, L. L. Lev, M. Husanu, V. N. Strocov, and A. A.
Demkov, Scientific Reports 8, 7721 (2018).
9. W. Guo, A. B. Posadas, Sirong Lu, David J. Smith, and Alexander A. Demkov, J. Appl.
Phys. 124, 235301 (2018).
10. K. Kormondy, A. Posadas, T. Q. Ngo, S. Lu, J. L. Jordan-Sweet, X. Guo, D. J. Smith,
M. R. McCartney, J. G. Ekerdt, and A. A. Demkov, J. Appl. Phys. 117, 095303 (2015).
11. R. A. McKee, F. J. Walker, and M. F. Chisholm, Phys. Rev. Lett. 81, 3014 (1998).
12. H. Li, X. Hu, Y. Wei, Z. Yu, X. Zhang, R. Droopad, A. A. Demkov, J. Edwards, Jr.,
K. Moore, and W. Ooms, J. Appl. Phys. 93, 4521 (2003).
13. Y. Wei, X. Hu, Y. Liang, D. C. Jordan, B. Craigo, R. Droopad, Z. Yu, A. Demkov, J.
John, L. Edwards, and W. J. Ooms, J. Vac. Sci. Technol. B 20, 1402 (2002).
14. M. Choi, A. B. Posadas, R. Dargis, C. K. Shih and A. A. Demkov, D. H. Triyoso, N.
David Theodore, C. Dubourdieu J. Bruley and J. Jordan-Sweet, J. Appl. Phys. 111,
064112, (2012).
15. L. V. Goncharova, D. G. Starodub, E. Garfunkel, T. Gustafsson, V. Vaithyanathan, J.
Lettieri, and D. G. Schlom, J. of Appl. Phy. 100, 014912 (2006);
16. X. Zhang, A. A. Demkov, H. Li, X. Hu, Y. Wei, and J. Kulik, Phys. Rev. B 68, 125323
(2003).
17. F. Amy, A. S. Wan, A. Kahn, F. J. Walker, and R. A. McKee, J. Appl. Phys. 96, 1635
(2004).
18. B. E. Deal, and A. S. Grove, J. Appl. Phys. 36, 3770 (1965).
19. H. Z. Massoud, J. D. Plummer, and E. A. Irene, J. Electrochem. Soc. 132, 1745 (1985).
20. H. Z. Massoud, J. D. Plummer, and E. A. Irene, J. Electrochem. Soc. 132, 2693 (1985).
21. H. Z. Massoud, and J. D. Plummer, J. Appl. Phys. 62, 3416 (1987).
22. A.F. Tasch, T.C. Holloway, K.F. Lee and J.F. Gibbons. Electron. Lett. 15, 435(1979).
23. W. Hiinsch, Th. Vogelsand, R. Kircher and M. Orlowski, Solid-state Electron 32,
839(1989).
24. K. Eisenbeiser, J. M. Finder, Z. Yu, J. Ramdani, J. A. Curless, J. A. Hallmark, R.
Droopad, W. J. Ooms, L. Salem, S. Bradshaw, and C. D. Overgaard. Appl. Phys. Lett.
76, 1324 (2000).
118
25. S. Guha and V. Narayanan, Phys. Rev. Lett. 98, 196101 (2007).
26. J. W. Park, D. F. Bogorin, C. Cen, D. A. Felker, Y. Zhang, C. T. Nelson, C. W. Bark,
C. M. Folkman, X. Q. Pan, M. S. Rzchowski, J. Levy, and C. B. Eom, Nat. Commun.
1, 94 (2010).
27. G. J. Yong, Rajeswari M. Kolagani, S. Adhikari, W. Vanderlinde, Y. Liang, K.
Muramatsu, and S. Friedrich, J. Appl. Phys. 108, 033502 (2010);.
28. K. J. Hubbard and D. G. Schlom, J. Mater. Res. 11, 2757(1996).
29. A. Posadas, R. Dargis, M. Choi, A. Slepko, J.J. Kim, D.J. Smith, and A.A. Demkov, J.
Vac. Sci. Technol. B. 23, 1071 (2011).
30. S. Zollner, A.A. Demkov, R. Liu, P.L. Fejes, R.B. Gregory, P. Alluri, J. Curless, Z.
Yu,J. Ramdani, R. Droopad, T.E. Tiwald, J.N. Hilfiker, J.A. Woollam, J. Vac. Sci.
Technol. B 18, 2242-2254 (2000).
31. H.S. Kato, S. Shiraki, M. Nantoh, M. Kawai, Surf. Sci. 544, 722(2003)
32. http://www.lelandstanfordjunior.com/thermaloxide.html
33. P. Pasierb, S. Komornicki, M. Rekas, J. Phys. Chem. Solid. 60, 1835(1999)
34. D. Lu, D. J. Baek, S. S. Hong, L. F. Kourkoutis, Y. Hikita and H. Y. Hwang, Nature
Materials, 15, 1255(2016)
35. S. S. Hong, J. H. Yu, D. Lu, A. F. Marshall, Y. Hikita, Y. Cui, H. Y. Hwang, Sci. Adv.
3, 5173(2017)
36. D. Ji, S. Cai, T.R. Paudel, et al. Nature 570, 87–90 (2019).
37. A. A. Sirenko, I. A. Akimov, J. R. Fox, A. M. Clark, Hong-Cheng Li, Weidong Si, and
X. X. Xi, Phys. Rev. Lett. 82, 4500(1999)
38. W. G. Nilsen and J. G. Skinner, J. Chem. Phys. 48, 2240 (1968)
39. C. H. Perry, Jeanne H. Fertel and T. F. McNelly, J. Chem. Phys. 47, 1619 (1967)
40. D. A. Tenne, A. K. Farrar, C. M. Brooks, T. Heeg, J. Schubert, H. W. Jang, C. W. Bark,
C. M. Folkman, C. B. Eom, and D. G. Schlom, Appl. Phys. Lett. 97, 142901 (2010)
41. L. Hilt Tisinger, R. Liu, J. Kulik, X. Zhang, J. Ramdani, and A. A. Demkov. J. Vac.
Sci. Technol. B, 21, 53 (2003)
42. R. Ouillon, J-P. Pinan-Lucarre, P.Ranson, P. Pruzan, S. K. Mishra, R. Ranjan and D.
Pandey, J. Phys.: Condens. Matter 14, 2079(2002)
43. Y.L. Du, G. Chen, M.S. Zhang, Solid State Commun. 130, 577(2004)
44. Choi, K., Raza, S.R.A., Lee, H.S., Jeon, P.J., Pezeshki, A., Min, S.W., Kim, J.S., Yoon,
W., Ju, S.Y., Lee, K., Im, S. Nanoscale 7, 5617 (2015)
119
45. S. Kang, N. Prasad, H.C.P. Movva, A. Rai, K. Kim, X. Mou, T. Taniguchi, K.
Watanabe, L.F. Register, E. Tutuc, and S.K. Banerjee, Nano Lett. 16, 4975 (2016)
120
Chapter 6: Advanced design of BaSnO3/SrTiO3/Al2O3 quantum wells
BaSnO3 and SrSnO3 have triggered huge interest in the area of transparent
conductive oxides as they have superior electron conductivity and wide optical band gaps.
Here, we propose an advanced quantum well structure design of BaSnO3/STO/Al2O3
utilizing the large conduction band offset (~3.5 eV) between BaSnO3 and Al2O3. The
deposition process is done by MBE and the heterostructures are characterized by RHEED
and XPS to confirm the film quality and composition. The work in this chapter is a
prototype implementation of this quantum well design. Further electrical and optical
measurements are needed to verify the actual physical properties of this system.
6.1 INTRODUCTION
Lightly La-doped BaSnO3 (BSO) and SrSnO3 (SSO) have very high room-
temperature conductivity and wide band gap, which makes them attractive for novel
transparent conductors and high-power electronic devices [1]-[10]. Since the early 20120s,
many people have studied the materials both theoretically and experimentally, particularly
the stannate growth process. Kim et al. [1] have reported unprecedentedly high mobility at
room temperature for a PLD-grown 4% La-doped BSO sample. Mobility in single crystal
samples reached as high as 320 cm2/V•s while retaining their optical transparency;
meanwhile, the mobility was still only up to 70 cm2/V•s in epitaxial films at that time.
Later, Liu et al. [2] used first-principles calculations to explore the origin of the superior
mobility in alkaline-earth stannates. Their small electron effective masses result from the
large size of Sn ions and the mainly Sn s-orbital derived conduction band edge. Bharat
Jalan’s group performed many studies of the stannate system. They used hybrid MBE with
a metal-organic Sn source to grow stoichiometric BSO [3], which they characterized with
RHEED, scanning transmission electron microscopy (STEM), electron energy-loss
121
spectroscopy (EELS), and energy dispersive x-ray spectroscopy (EDX) to confirm the film
quality, epitaxy, crystal structure, and composition. Conduction band offsets were also
determined [4] for BSO/STO and BSO/LAO interfaces to be 3.1 eV and 3.0 eV,
respectively. They also reported adsorption controlled BSO growth by hybrid MBE in 2017
[5]. Films grown within this limited growth window are shown to yield La-doped BSO
films with mobilities of 105 cm2/V•s. Low mobility and charge compensation are induced
in Ba- and Sn-deficient films, with a stronger Sn-deficiency dependence. They also
analyzed the defect-driven localization crossover in La-doped SSO films [6]. They pointed
out that substrate-induced dislocations in the film strongly influence the electron phase
coherence length, which causes two-dimensional to three-dimensional weak localization
crossover. They used epitaxial strain to engineer the SSO film phase and mobility [7] and
achieved over 300% mobility enhancement at room temperature compared with unstrained
low-temperature orthorhombic polymorph.
Susanne Stemmer’s group also reported several MBE-grown BSO films with high
mobility. They used PrScO3 as a lattice-matched substrate and obtained 150 cm2/V•s
mobility at room temperature [8]. Structural images and band gap data of their stannate
films were reported in a later publication [9]. Paik et al. also published an adsorption
controlled BSO growth process by standard MBE [10]. BSO films grown on DyScO3
substrate showed a 183 cm2/V•s mobility at room temperature and 400 cm2/V•s at 10 K.
Stannates have demonstrated excellent mobility and optical transparence. The wide
bandgap is also suitable for building up large band offsets with certain materials. We can
design quantum wells (QWs) by building heterostructures of materials with large band
offsets. By inserting a thin layer of one material between two layers of another material
with even larger band gap, we can make a simple quantum well. The energy in the wells
122
becomes discretized in order to satisfy the Schrodinger equations in this confined potential.
The fabrication of novel TMO QW devices opens up new possibilities for advanced
functionalities [11]-[15]. Jackson et al. showed interface-induced magnetism in
GdTiO3/SrTiO3 and SmTiO3/SrTiO3 quantum wells [11]. Need et al. showed that, in
GdTiO3/SrTiO3 quantum wells, the net ferromagnetism inherent to the Mott insulator
GdTiO3 matrix propagates into the nominally nonmagnetic STO quantum wells [12]. Choi
et al. fabricated a PbZrO3/PbTiO3 quantum well [13] with superior dielectric constant (800)
at a stacking period of 1 uc/1 uc (PZO1/PTO1). There are relatively fewer reports of large
band offset quantum wells that allow mid-infrared intersubband absorption. Zhao et al. use
ZnO/ZnMgO quantum wells to achieve mid-infrared intersubband absorption [14].
Ortmann et al. managed to integrate LAO/STO superlattices on Si and showed the ability
to control intersubband absorption energy by changing the width of the STO well layers
[15].
In this chapter, we develop an MBE approach to the growth of BSO and SSO in
our DCA M600 Oxide-MBE system. We also propose a novel BSO/STO/Al2O3 quantum well
structure that can utilize the large band offset between BSO and Al2O3 and the high mobility
of BSO.
6.2 GROWTH AND QW DESIGN
We first grow epitaxial BSO in our MBE system on STO substrates. BSO has a
decent lattice match with STO and grows well on it, and we can also use the STO/Si
pseudo-substrate to integrate BSO on Si. Details of STO on Si deposition are described in
Chapters 4 and 5. Ba is evaporated from effusion cells and Sn comes from an e-beam
evaporator in form of SnO2. Using SnO2 in the e-gun yields a 1-2×10-6 Torr base oxygen
123
pressure during deposition, coming from the SnO2 decomposition. Additional oxygen is
needed to maintain a total oxygen pressure at 2-3×10-5 Torr to ensure Sn is fully oxidized.
The real Ba flux is also observed to decrease slightly when the oxygen pressure exceeds
10-5 Torr likely due to source oxidation. This requires one to increase the Ba flux 1.2-1.3
times above the normal flux value to maintain the same atom arrival rate. The deposition
temperature is naintained between 750-800 ˚C. Higher temperatures will result in Sn re-
evaporation and failure to form the correct phase. At lower temperatures, BSO will not
crystallize well.
The tricky part of using an e-gun for stannate deposition is the unstable e-gun flux.
The SnO2 flux decays too quickly, usually 20-40% over a 5 min period when the rate is ~1
monolayer/min. To guarantee the crystal quality, the flux needs to be sufficiently stable to
maintain the stoichiometric condition. Since we are using SnO2 as Sn source, the total
oxygen pressure is directly related to the Sn flux, and this has been verified to have a linear
dependence by flux measurement using a quartz crystal monitor. We use the real-time
measured chamber pressure as a parameter to represent the Sn flux, and actively maintain
the pressure by minor adjustments to the e-gun emission current to compensate the
decaying flux during the entire deposition process.
We propose a BSO/Al2O3 quantum well structure based on the large band offset
between Al2O3 (7.6 eV) and BSO (3.1 eV). Al2O3 can be epitaxially grown on BSO with
good crystallinity (as -Al2O3), but the reverse is not true due to surface energy issues. To
get around this, we introduce an STO layer on top of Al2O3 to allow BSO to wet and
crystallize as a flat layer. The band alignment of the stack is shown in Figure 6.1 and a
schematic of the proposed structure is shown in Figure 6.2.
124
Figure 6.1: Band alignment of a single quantum well. The conduction band offset
between BSO and Al2O3 is 3.5 eV.
Figure 6.2: Schematic of the heterostructures to be grown.
125
The valence band offsets (VBOs) marked in Figure 6.1 were extracted from XPS
measurements of various stacks consisting of BSO, Al2O3 and STO layers. First, we
measure a core level and the valence band maximum (VBM) binding energies in STO and
BSO bulk-like films. Second, we measure the same core level binding energies in a
BSO/STO stack. Here we use Ti 2p in STO and Ba 3d in BSO. The VBO is calculated as
follows:
VBOSTO/BSO = (ETi 2p- EVBM)STO - (EBa 3d- EVBM)BSO - (ETi 2p- EBa 3d)STO/BSO
With the same method, we get VBOSTO/Al2O3 = 2.3 eV and VBOBSO/Al2O3 = 1 eV. We
also know the experimental band gaps of each material, which allows one to obtain the
conduction band offsets (CBO), as marked by green lines in Figure 6.1. We can see that
the CBO for BSO and Al2O3 is very large at 3.5 eV.
Based on the CBO and the designed well width in this structure, we performed a
simulation using a Poisson-Schrodinger solver. The well consists of 3 uc of STO and 10
uc of BSO with infinitely thick Al2O3 barriers. The simulation was performed by Suyeong
Jang in our group, and the results are shown in Figure 6.3.
126
Figure 6.3: Simulation results for possible energy states in a single quantum well.
Based on the energy levels in the well, we see that one can get transitions whose
energies range from visible light to infrared light. This is a very promising quantum well
structure with a wide range of energy absorption possibilities.
6.3 CHARACTERIZATION
We monitored the RHEED pattern from the surface of the sample during the
deposition procedure to check the crystallinity and surface quality of each quantum well
layer as it is being formed. We also used XPS to check the stoichiometry of the materials
and to measure core level energies in determining the band offsets.
In Figure 6.4, we show the RHEED pattern evolution as each layer of the
heterostructure is deposited. This structure is a 10 uc BSO/3 uc STO/thick Al2O3 stack. The
pattern brightness and sharpness decreases from Al2O3 to BSO which means crystal quality
gets worse as more layers are put on top. There is still work needed to optimize the growth
127
process in order to maintain a good crystal quality if we want to repeatedly build a
superlattice of this stack.
(a)
(b)
(c)
128
Figure 6.4: RHEED evolution from (a) thick Al2O3 on STO buffer layer on Si (b) 3 uc
STO on previous Al2O3 surface (c) 10 uc BSO on previous STO surface
We also show the VB spectrum of a BSO/STO /Al2O3 stack in Figure 6.5. This
sample has 5 uc BSO on 3 uc STO on thick Al2O3, and shows a combination of STO VB
and BSO VB features.
Figure 6.5: VB spectrum of BSO/STO/Al2O3
6.4 OUTLOOK
We are still working on improving the quantum well crystal quality in order to be
able to repeatedly stack these layers for intersubband absorption measurements. The key is
to maintain the BSO crystal quality on the Al2O3 surface with an STO buffer. Transport
measurements of La-doped BSO films grown in our MBE also need to be performed to
optimize the process, especially the doping. This quantum well structure has great potential
for advanced quantum well devices once the process can be optimized.
129
6.5 REFERENCES
1. H. J. Kim, U. Kim, H. M. Kim, T. H. Kim, H. S. Mun, B. G. Jeon, K. T. Hong, W.
J. Lee, C. Ju, K. H. Kim and K. Char. Appl. Phys. Express 5, 061102(2012)
2. H. R. Liu, J. H. Yang, H. J. Xiang, X. G. Gong and S. H. Wei, Appl. Phys. Lett.
102, 112109 (2013)
3. A. Prakash, J. Dewey, H. Yun, J. S. Jeong, K. A. Mkhoyan, and B. Jalan, J. Vac.
Sci. Technol. A 33, 060608 (2015)
4. S.A. Chambers, T. C. Kaspar, A. Prakash, G. Haugstad and B. Jalan, Appl. Phys.
Lett. 108, 152104 (2016)
5. A. Prakash, P. Xu, X. Wu, G. Haugstad, X. Wang and B. Jalan, J. Mater. Chem. C
5, 5730(2017)
6. T. Wang, L. R. Thoutam, A. Prakash, W. Nunn, G. Haugstad and B. Jalan, Phys.
Rev. Mater. 1, 061601(2017)
7. T. Wang, A. Prakash, Y. Dong, T. Truttmann, A. Bucsek, R. James, D. D. Fong, J.
Kim, P. J. Ryan, H.Zhou, T. Birol and B. Jalan, ACS Appl. Mater. Interfaces 10,
43802(2018)
8. S. Raghavan, T. Schumann, H. Kim, J. Y. Zhang, T. A. Cain and S. Stemmer, APL
Materials 4, 016106 (2016)
9. T. Schumann, S. Raghavan, K. Ahadi, H. Kim and S. Stemmer. J. Vac. Sci. Technol.
A 34, 050601 (2016)
10. H. Paik, Z. Chen, E. Lochocki, A. Seidner H., A. Verma, N. Tanen, J. Park, M.
Uchida, S. Shang, B. Zhou, M. Brützam, R. Uecker, Z. Liu, D. Jena, K. M. Shen,
D. A. Muller and D. G. Schlom, APL Mater. 5, 116107 (2017)
11. C. A. Jackson, S. Stemmer, Phys. Rev. B: Condens. Matter Mater. Phys. 88,
180403(2013)
12. R. F. Need, B. J. Isaac, B. J. Kirby, J. A. Borchers, S. Stemmer, and Stephen D.
Wilson, Phys. Rev. Lett. 117, 037205(2016)
13. T. Choi, J. Lee, Thin Solid Films 475, 283(2005)
14. K. Zhao, G. Chen, B. Li and A. Shen, Appl. Phys. Lett. 104, 212104 (2014)
15. J. E. Ortmann, N. Nookala, Q. He, L. Gao, C. Lin, A. B. Posadas, A. Y.
Borisevich, M. A. Belkin, and A. A. Demkov, ACS NANO 12, 7682 (2018).
130
Chapter 7: Summary and future work
7.1 SUMMARY
In this thesis, I have presented major results of my studies of transition metal and
transition metal oxide integration with SrTiO3 and SrTiO3/Si templates.
Pt deposition on STO was studied using different thickness and under multiple
temperatures. We use in situ XPS, ex situ AFM and SEM to explore the core level binding
energy, valence band spectrum, and surface morphology as functions of coverage and
deposition temperature. At low temperature, Pt atoms tend to stay at the low energy oxygen
hollow site owing to the relatively low mobility of Pt. In essence Pt can cover STO even
though surface energies make it thermodynamically unfavorable. High growth temperature
will give Pt enough mobility to move across the surface and form clusters. Pt clusters will
exhibit metallic XPS features even at very low coverage. SEM imaging clearly shows
nanoclusters forming at high growth temperature and a continuous film at low temperature
for the same coverage.
Eu deposition on STO in UHV results in EuO epitaxial growth on STO. We studied
the scavenging process at different temperatures on different thicknesses of STO films
integrated on Si. At 300°C, Eu takes oxygen from STO films that are thicker than 4 nm and
forms EuO. For thinner STO, Eu takes too much oxygen and destabilizes the STO layer,
allowing for a solid-state reaction with Si that results in the formation of EuSiy and silicates.
Low temperature growth of Eu only forms up to a ∼20 Å thick layer of EuO. Beyond this
point, Eu no longer takes oxygen from STO and Eu metal accumulates on the surface. The
EuO/Eu metal interface was also studied theoretically and experimentally. A downward
band bending is confirmed near the interface. The study of the ferroelectric field-effect of
the EuO/STO interface is very preliminary. It is currently limited by technical difficulties
131
of EuO over-oxidation when transporting samples. In the process, we found a workable
solution using a GeOx capping layer. More research needs to be done to optimize this
process.
We also performed studies on STO/Si. We first considered dry oxidation of STO/Si
heterostructures and demonstrate that the underlying Si can be safely oxidized at a
relatively high temperature (800ºC) with the STO crystallinity not significantly degraded.
We built a Deal-Grove-like oxidation model and found good agreement between the data
and the model. This gives us the ability to predict the temperature and time needed to obtain
the desired SiO2 thickness for Si that is covered by a thin layer of STO. This additional
knob for controlling the layer structure can enable one to integrate more complicated oxide
structures on this STO/Si pseudo-substrate, especially for applications requiring complete
decoupling between the STO and Si. With thick SiO2 interlayer available, we were able to
do selective etching of the underlying SiO2 to make free-standing STO membranes. Using
NaOH etching, we obtained free-standing STO membranes with a typical size of ~5×10
µm2. Raman spectroscopy was used to see if there were any differences between STO thin
films and membranes. Preliminary results were obtained but with weak STO membranes
signals.
We proposed a novel BSO/STO/Al2O3 quantum well structure with a large
conduction band offset (~3.5 eV) and the possibility of near- to mid-infrared absorption. We
showed the energy band diagram and the band offsets for such a structure. RHEED and
XPS were used to characterize the surface quality and the stoichiometry of the
heterostructure. We are working on improving the growth process to maintain good surface
crystallinity and make repeated stacks for further energy absorption measurements a
possibility.
132
7.2 FUTURE WORK
TMs and TMOs have a wide range of interesting properties and phenomena to
explore. When they are integrated with STO and Si, additional industrial possibilities are
opened up.
For our EuO/STO interface 2DEG research, we are improving the capping layer to
achieve a better ARPES measurement. This study provides a new possibility to control the
interface charge distribution thus achieving an unexplored way of modulating the 2DEG,
and possibly shift the Curie temperature of EuO to the range of practical applications.
STO free-standing membranes are promising for studying physical properties of
reduced dimensionality and also for potential future transistors. We are working on the
optimization of the Raman setup to reduce noise and amplify the STO membrane signal.
We can use different thicknesses of STO membranes to study the phonon modes of such a
quasi-2D structure. We can also explore the design of new transistor devices using our STO
membranes with their high-k values.
The quantum well of BSO/STO/Al2O3 needs to improvements in the crystal quality
in order to make repeated wells for further energy absorption measurements. Extensive
transport measurements of different doping levels of La-doped BSO samples are also
needed.
133
Bibliography
• A. A. Demkov and A. B. Posadas, Integration of Functional Oxides with
Semiconductors (Springer New York, 2014).
• A. A. Melville, T. Mairoser, A. Schmehl, D. E. Shai, E. J. Monkman, J. W. Harter, T.
Heeg, B. Hollander, J. Schubert, K. M. Shen, J. Mannhart, and D. G. Schlom, Appl.
Phys. Lett. 100, 222101 (2012).
• A. A. Sirenko, I. A. Akimov, J. R. Fox, A. M. Clark, Hong-Cheng Li, Weidong Si, and
X. X. Xi, Phys. Rev. Lett. 82, 4500(1999)
• A. Asthagiri and D. S. Sholl, J. Chem. Phys. 116, 9914 (2002).
• A. B. Posadas, K. J. Kormondy, W. Guo, P. Ponath, J. Geler-Kremer, T. Hadamek,
and A. A. Demkov, J. Appl. Phys. 121, 105302 (2017)
• A. Brinkman, M. Huijben, M. van Zalk, J. Huijben, U. Zeitler, J. C. Maan, W. G. van
der Wiel, G. Rijnders, D. H. A. Blank, and H. Hilgenkamp, Nat. Mater. 6, 493 (2007).
• A. D. Polli, T. Wagner, T. Gemming, and M. R€uhle, Surf. Sci. 448, 279 (2000).
• A. Einstein, Ann. Phys. 322, 132 (1905).
• A. G. Swartz, P. M. Odenthal, Y. Hao, R. S. Ruoff, and R. K. Kawakami, ACS Nano
6, 10063 (2013).
• A. Ichimiya, P. I. Cohen, Reflection High Energy Electron Diffraction Cambridge
University Press (2004).
• A. Mauger and C. Godart, Phys. Rep. 141, 51 (1986).
• A. Ohtomo and H. Y. Hwang, “A high-mobility electron gas at the LaAlO3/SrTiO3
heterointerface,” Nature, 427, 423 (2004).
• A. Posadas, R. Dargis, M. Choi, A. Slepko, J.J. Kim, D.J. Smith, and A.A. Demkov, J.
Vac. Sci. Technol. B. 23, 1071 (2011).
• A. Prakash, J. Dewey, H. Yun, J. S. Jeong, K. A. Mkhoyan, and B. Jalan, J. Vac. Sci.
Technol. A 33, 060608 (2015)
• A. Prakash, P. Xu, X. Wu, G. Haugstad, X. Wang and B. Jalan, J. Mater. Chem. C 5,
5730(2017)
• A. Schmehl et al., Nat. Mater. 6, 882 (2007).
• A. Y. Cho, F. K. Reinhart, Appl. Phys. Lett. 21, 355 (1972).
• A.F. Tasch, T.C. Holloway, K.F. Lee and J.F. Gibbons. Electron. Lett. 15, 435(1979).
• B. E. Deal, and A. S. Grove, J. Appl. Phys. 36, 3770 (1965).
134
• B. El-Kareh, Fundamentals of Semiconductor Processing Technologies, (Springer US,
New York, 1995).
• B. G. Streetman, S. Banerjee, Solid State electronic Devices, 524 (Hall, New Jersey,
2000).
• B. I. Edmondson, S. Liu, S. Lu, H. W. Wu, A. B. Posadas, D. J. Smith, X. P. A. Gao,
A. A. Demkov, and J. G. Ekerdt, J. Appl. Phys. 124, 185303 (2018).
• B. Kalisky, J. A. Bert, B. B. Klopfer, C. Bell, H. K. Sato, M. Hosoda, Y. Hikita, H. Y.
Hwang, and K. A. Moler. Nat. Commun. 3, 922 (2012).
• B. N. Figgis and J.Lewis, The Magnetochemistry of Complex Compounds in Modern
Co-ordination Chemistry, Ed. J. Lewis and C. Wilkins, Interscience, New York (1960).
• B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti and A. Kis. Nature Nanotech
6, 147(2011).
• B. Reihl, J. G. Bednorz, K. A. M€uller, Y. Jugnet, G. Landgren, and J. F. Morar, Phys.
Rev. B 30, 803 (1984).
• B. T. Matthias, R. M. Bozorth, and J. H. Van Vleck, Phys. Rev. Lett. 7, 160 (1961).
• B. W. Wessels, Annu. Rev. Mater. Res. 37, 659 (2007).
• C. A. Jackson, S. Stemmer, Phys. Rev. B: Condens. Matter Mater. Phys. 88,
180403(2013)
• C. Caspers, M. Müller, A. X. Gray, A. M. Kaiser, A. Gloskovskii, C. S. Fadley, W.
Drube, and C. M. Schneider, Phys. Rev. B. 84, 205217 (2011).
• C. H. Perry, Jeanne H. Fertel and T. F. McNelly, J. Chem. Phys. 47, 1619 (1967)
• Choi, K., Raza, S.R.A., Lee, H.S., Jeon, P.J., Pezeshki, A., Min, S.W., Kim, J.S., Yoon,
W., Ju, S.Y., Lee, K., Im, S. Nanoscale 7, 5617 (2015)
• D. A. Tenne, A. K. Farrar, C. M. Brooks, T. Heeg, J. Schubert, H. W. Jang, C. W. Bark,
C. M. Folkman, C. B. Eom, and D. G. Schlom, Appl. Phys. Lett. 97, 142901 (2010)
• D. F. Bogorin, P. Irvin, C. Cen, and J. Levy, “LaAlO3/SrTiO3-based device concepts,”
in Multifunctional Oxide Heterostructures, Oxford University Press, 364(2010).
• D. Hisamoto, W. Lee, J. Kedzierski, H. Takeuchi, K. Asano, C. Kuo, E. Anderson, T.
King, J. Bokor and C. Hu. IEEE Transactions on Electron Devices. 47 2320 (2000).
• D. J. Frank, R. H. Dennard, E. Nowak, P. M. Solomon, Y. Taur and Hon-Sum Philip
Wong, Proceedings of the IEEE. 89, 259 (2001).
• D. Ji, S. Cai, T.R. Paudel, et al. Nature 570, 87–90 (2019).
135
• D. Lu, D. J. Baek, S. S. Hong, L. F. Kourkoutis, Y. Hikita and H. Y. Hwang, Nature
Materials, 15, 1255(2016)
• D. P. Kumah, J. H. Ngai and L. Kornblum, Adv. Funct. Mater. 30, 1901597(2019)
• D. V. Averyanov, Y. G. Sadofyev, A. M. Tokmachev, A. E. Primenko, I. A. Likhachev,
and V. G. Storchak, ACS Appl. Mater. Interfaces 7, 6146–6152 (2015).
• D. V. Averyanov, Y. G. Sadofyev, A. M. Tokmachev, A. E. Primenko, I. A. Likhachev,
and V. G. Storchak, J. Appl. Phys. 113, 17C505 (2013).
• E. Harari, J. Appl. Phys. 49, 2478-2489 (1978).
• E. Vella, F. Messina, M. Cannas, R. Boscaino, Phys. Rev. B. 83, 174201 (2011).
• EuO Crystal Structure: Datasheet from “PAULING FILE Multinaries Edition—2012”
in Springer Materials,
see: http://materials.springer.com/ isp/crystallographic/docs/sd_0553841.
• F. Amy, A. S. Wan, A. Kahn, F. J. Walker, and R. A. McKee, J. Appl. Phys. 96, 1635
(2004).
• F. Pockels, Lehrbuch Der Kristalloptik (Leipzig und Berlin, 1906).
• F. Rao, M. Kim, A. J. Freeman, S. Tang, and M. Anthony, Phys. Rev. B 55, 13953
(1997).
• G. J. Yong, Rajeswari M. Kolagani, S. Adhikari, W. Vanderlinde, Y. Liang, K.
Muramatsu, and S. Friedrich, J. Appl. Phys. 108, 033502 (2010);.
• G. K. Wertheim and S. B. DiCenzo, Phys. Rev. B 37, 844 (1988).
• G. M. Prinz, T. Gerber, A. Lorke, and M. Müller, Appl. Phys. Lett. 109, 202401 (2016).
• G. T. Reed, G. Mashanovich, F. Y. Gardes and D. J. Thomson, Nat. Photon. 4, 518
(2010).
• H. Brune, Surf. Sci. Rep. 31, 125 (1998).
• H. Iwai, Microelectron. Eng. 86, 1520 (2009).
• H. J. Kim, U. Kim, H. M. Kim, T. H. Kim, H. S. Mun, B. G. Jeon, K. T. Hong, W. J.
Lee, C. Ju, K. H. Kim and K. Char. Appl. Phys. Express 5, 061102(2012)
• H. Li, X. Hu, Y. Wei, Z. Yu, X. Zhang, R. Droopad, A. A. Demkov, J. Edwards, Jr.,
K. Moore, and W. Ooms, J. Appl. Phys. 93, 4521 (2003).
• H. Miyazaki, H. Momiyama, T. Hajiri, T. Ito, K. Imura, M. Matsunami, and S. Kimura,
J. Phys. Conf. Ser. 391, 012047 (2012).
• H. Miyazaki, T. Ito, H. Im, K. Terashima, S. Yagi, M. Kato, K. Soda, and S. Kimura,
Jpn. J. Appl. Phys. 48, 055504 (2009).
136
• H. N. Lee, H. M. Christen, M. F. Chisholm, C. M. Rouleau, and D. H. Lowndes. Nature,
433, 395 (2005).
• H. Paik, Z. Chen, E. Lochocki, A. Seidner H., A. Verma, N. Tanen, J. Park, M. Uchida,
S. Shang, B. Zhou, M. Brützam, R. Uecker, Z. Liu, D. Jena, K. M. Shen, D. A. Muller
and D. G. Schlom, APL Mater. 5, 116107 (2017)
• H. R. Liu, J. H. Yang, H. J. Xiang, X. G. Gong and S. H. Wei, Appl. Phys. Lett. 102,
112109 (2013)
• H. Seo and A. A. Demkov, Phys. Rev. B 92, 245301 (2015).
• H. Seo, A. B. Posadas, and A. A. Demkov, J. Vac. Sci. Technol., B 30, 04E108 (2012).
• H. Y. Hwang, Y. Iwasa, M. Kawasaki, B. Keimer, N. Nagaosa, and Y. Tokura, Nat.
Mater. 11, 103 (2012).
• H. Yamada, M. Kawasaki, Y. Ogawa, and Y. Tokura, Appl. Phys. Lett. 81, 4793 (2002).
• H. Z. Massoud, and J. D. Plummer, J. Appl. Phys. 62, 3416 (1987).
• H. Z. Massoud, J. D. Plummer, and E. A. Irene, J. Electrochem. Soc. 132, 1745 (1985).
• H.S. Kato, S. Shiraki, M. Nantoh, M. Kawai, Surf. Sci. 544, 722(2003)
• http://www.lelandstanfordjunior.com/thermaloxide.html
• J. A. Bert, B. Kalisky, C. Bell, M. Kim, Y. Hikita, H. Y. Hwang, and K. A. Moler, Nat.
Phys. 7. 767(2011).
• J. A. Kittl, K. Opsomer, M. Popovici, N. Menou, B. Kaczer, X. P. Wang, C. Adelmann,
M. A. Pawlak, K. Tomida, A. Rothschild, B. Govoreanu, R. Degraeve, M. Schaekers,
M. Zahid, A. Delabie, J. Meersschaut, W. Polspoel, S. Clima, G. Pourtois, W. Knaepen,
C. Detavernier, V. V. Afanas’ev, T. Blomberg, D. Pierreux, J. Swerts, P. Fischer, J. W.
Maes, D. Manger, W. Vandervorst, T. Conard, A. Franquet, P. Favia, H. Bender, B.
Brijs, S. Van Elshocht, M. Jurczak, J. Van Houdt, D. J. Wouters, Microelectron. Engin.
86, 1789 (2009).
• J. Braun, J. Minár, S. Mankovsky, V. N. Strocov, N. B. Brookes, L. Plucinski, C. M.
Schneider, C. S. Fadley, and H. Ebert, Phys. Rev. B 88, 205409(2013)
• J. Chakhalian, J. W. Freeland, A. J. Millis, C. Panagopoulos, and J. M. Rondinelli, Rev.
Mod. Phys. 86, 1189 (2014).
• J. D. Plummer, M. D. Deal, P. B. Griffin, Silicon VLSI Technology, (Prentice Hall,
Upper Saddle River, 2013).
• J. E. Ortmann, A. B. Posadas and A. A. Demkov, J. Appl. Phys. 124, 015301 (2018).
• J. E. Ortmann, M. A. Duncan and A. A. Demkov, Optical Materials Express 9, 2982
(2019).
137
• J. E. Ortmann, N. Nookala, Q. He, L. Gao, C. Lin, A. B. Posadas, A. Y. Borisevich, M.
A. Belkin, and A. A. Demkov, ACS NANO 12, 7682 (2018).
• J. E. Ortmann, S. Kwon, A. B. Posadas, M. J. Kim and A. A. Demkov, J. Appl. Phys.
125, 155302 (2019).
• J. H. Haeni, P. Irvin, W. Chang, R. Uecker, P. Reiche, Y. L. Li, S. Choudhury, W. Tian,
M. E. Hawley, B. Craigo, A. K. Tagantsev, X. Q. Pan, S. K. Streiffer, L. Q. Chen, S.
W. Kirchoefer, J. Levy and D. G. Schlom. Nature 430, 758 (2004).
• J. K. Lang, Y. Baer, and P. A. Cox, J. Phys. F. 11, 121 (1981).
• J. K. Lee, N. Sai, and A.A. Demkov, Phys. Rev. B 82, 235305 (2010).
• J. Lettieri, J. H. Haeni, and D. G. Schlom, J. Vac. Sci. Technol. A 20, 1332 (2002).
• J. Lettieri, V. Vaithyanathan, S. K. Eah, J. Stephens, V. Sih, D. D. Awschalom, J. Levy,
and D. G. Schlom, Appl. Phys. Lett. 83, 975 (2003).
• J. R. Arthur Jr., J. Appl. Phys. 39, 4032 (1968).
• J. Ramdani, R. Droopad, Z. Yu, J. A. Curless, C. D. Overgaard, J. Finder,K.
Eisenbeiser, J. A. Hallmark, W. J. Ooms, V. Kaushik, P. Alluri, and S. Pietambaram,
Appl. Surf. Sci. 159, 127 (2000).
• J. Robertson, J. Vac. Sci. Technol. B 18, 1785 (2000).
• J. Schoenes and P. Wachter, Phys. Rev. B. 9, 3097 (1974).
• J. Son, J. Cagnon, and S. Stemmer, J. Appl. Phys. 106, 043525 (2009).
• J. W. Park, D. F. Bogorin, C. Cen, D. A. Felker, Y. Zhang, C. T. Nelson, C. W. Bark,
C. M. Folkman, X. Q. Pan, M. S. Rzchowski, J. Levy, and C. B. Eom, Nat. Commun.
1, 94 (2010).
• J.E. Mahan, Physical Vapor Deposition of Thin Films, Wiley-VCH, 2000.
• K. A. Kovacs, P. Grof, L. Burai, M. Riedel. J. Phys. Chem. A. 108, 11026 (2004).
• K. D. Fredrickson, H. Seo and A. A. Demkov, J. Appl. Phys. 120, 065301 (2016).
• K. D. Fredrickson, V. V. Vogler-Neuling, K. J. Kormondy, D. Caimi, F. Eltes, M.
Sousa, J. Fompeyrine, S. Abel, and A. A. Demkov, Phys. Rev. B 98, 075136 (2018).
• K. Eisenbeiser, J. M. Finder, Z. Yu, J. Ramdani, J. A. Curless, J. A. Hallmark, R.
Droopad, W. J. Ooms, L. Salem, S. Bradshaw, and C. D. Overgaard. Appl. Phys. Lett.
76, 1324 (2000).
• K. G. Günther, Z. Naturforschg. A. 13 (12), 1081-1089 (1958).
• K. J. Hubbard and D. G. Schlom, J. Mater. Res. 11, 2757 (1996).
138
• K. J. Hubbard and D. G. Schlom, J. Mater. Res. 11, 2757(1996).
• K. J. Kormondy, A. B. Posadas, T. Q. Ngo, S. Lu, N. Goble, J. Jordan-Sweet, X. P. A.
Gao, D. J. Smith, M. R. McCartney, J. G. Ekerdt, and A. A. Demkov, J. Appl. Phys.
117, 95303 (2015).
• K. J. Kormondy, L. Gao, X. Li, S. Lu, A. B. Posadas, S. Shen, M. Tsoi, M. R.
McCartney, D. J. Smith, J. Zhou, L. L. Lev, M.-A. Husanu, V. N. Strocov, and A. A.
Demkov, Sci. Rep. 8, 7721 (2018).
• K. J. Kormondy, S. Abel, F. Fallegger, Y. Popoff, P. Ponath, A. B. Posadas, M.
Sousa, D. Caimi, H. Siegwart, E. Uccelli, L. Czornomaz, C.Marchiori, J. Fompeyrine,
A. A. Demkov. Microelectron. Eng. 147, 215 (2015).
• K. J. Kormondy, Y. Popoff, M. Sousa, F. Eltes, D. Caimi, M. D. Rossell, M. Fiebig, P.
Hoffmann, C. Marchiori, M. Reinke, M. Trassin, A. A. Demkov, J. Fompeyrine, and
S. Abel, Nanotechnology 28, 075706 (2017).
• K. Kormondy, A. Posadas, T. Q. Ngo, S. Lu, J. L. Jordan-Sweet, X. Guo, D. J. Smith,
M. R. McCartney, J. G. Ekerdt, and A. A. Demkov, J. Appl. Phys. 117, 095303 (2015).
• K. Mistry, C. Allen, C. Auth, B. Beattie, D. Bergstrom, M. Bost, M. Brazier, M.
Buehler, A. Cappellani, R. Chau, C.-H. Choi, G. Ding, K. Fischer, T. Ghani, R. Grover,
W. Han, D. Hanken, M. Hattendorf, J. He, J. Hicks, R. Heussner, D. Ingerly, P. Jain,
R. James, L. Jong, S. Joshi, C. Kenyon, K. Kuhn, K. Lee, H. Liu, J. Maiz, B. McIntyre,
P. Moon, J. Neirynck, S. Pae, C. Parker, D. Parsons, C. Prasad, L. Pipes, M. Prince, P.
Ranade, T. Reynolds, J. Sandford, L. Shifren, J. Sebastian, J. Seiple, D. Simon, S.
Sivakumar, P. Smith, C. Thomas, T. Troeger, P. Vandervoorn, S. Williams, K.
Zawadzki, Int. El. Devices Meet. 247 (2007).
• K. Takada, H. Sakurai, E. Takayama-Muromachi, F. Izumi, R. A. Dilanian, and T.
Sasaki, Nature, 422, 53 (2003).
• K. Ueno, S. Nakamura, H. Shimotani, A. Ohtomo, N. Kimura, T. Nojima, H. Aoki, Y.
Iwasa, and M. Kawasaki, Nat. Mater. 7, 855 (2008).
• K. Zhao, G. Chen, B. Li and A. Shen, Appl. Phys. Lett. 104, 212104 (2014)
• L. Bjaalie, B. Himmetoglu, L. Weston, A. Janotti, and C. G. Van de Walle. New J.
Phys. 16, 025005 (2014).
• L. Gao and A. A. Demkov, Phys. Rev. B 97, 125305 (2018).
• L. Gao, W. Guo, A. B. Posadas, and A. A. Demkov. Phys. Rev. Materials 3, 094403
(2019).
• L. Hilt Tisinger, R. Liu, J. Kulik, X. Zhang, J. Ramdani, and A. A. Demkov. J. Vac.
Sci. Technol. B, 21, 53 (2003)
139
• L. Li, C. Richter, J. Mannhart, and R. C. Ashoori, Nat. Phys. 7, 762 (2011).
• L. Liao, Y. Lin, M. Bao, R. Cheng, J. Bai, Y. Liu, Y. Qu, K. L. Wang, Y. Huang and
X. Duan. Nature 467, 305 (2010).
• L. Mazet, S. M. Yang, S. V. Kalinin, S. Schamm-Chardon and C. Dubourdieu, Sci.
Technol. Adv. Mater. 16, 036005 (2015).
• L. V. Goncharova, D. G. Starodub, E. Garfunkel, T. Gustafsson, V. Vaithyanathan, J.
Lettieri, and D. G. Schlom, J. of Appl. Phy. 100, 014912 (2006);
• L. Zhang, Y. Yuan, J. Lapano, M. Brahlek, S. Lei, B. Kabius, V. Gopalan, and R.
Engel-Herbert, ACS Nano, 12, 1306(2018)
• M. Choi, “Monolithic integration of functional perovskite structures on Si” Ph.D. thesis
(The University of Texas at Austin, Austin, TX, 2014).
• M. Choi, A. B. Posadas, R. Dargis, C. K. Shih and A. A. Demkov, D. H. Triyoso, N.
David Theodore, C. Dubourdieu J. Bruley and J. Jordan-Sweet, J. Appl. Phys. 111,
064112, (2012).
• M. Copel, P. R. Duncombe, D. A. Neumayer, T. M. Shaw, and R. M. Tromp, Appl.
Phys. Lett. 70, 3227 (1997).
• M. Hochberg and T. Baehr-Jones, Nat. Photon. 4, 492(2010).
• M. Imada, A. Fujimori, Y. Tokura, Rev. Mod. Phys. 70, 1039 (1998).
• M. K. Bahl, S. C. Tsai, and Y. W. Chung, Phys. Rev. B 21, 1344 (1980).
• M. Kitano, K. Tsujimaru, and M. Anpo, Top. Catal. 49, 4 (2008).
• M. Matsubara, A. Schmehl, J. Mannhart, D. G. Schlom, and M. Fiebig, Phys. Rev. B
81, 214447 (2010).
• M. Matsubara, A. Schroer, A. Schmehl, A. Melville, C. Becher, M. Trujillo-Martinez,
D. G. Schlom, J. Mannhart, J. Kroha, and M. Fiebig, Nat. Commun. 6, 6724 (2015).
• M. Mostovoy, Nat. Mater. 7, 269 (2008).
• M. Mrovec, J.-M. Albina, B. Meyer, and C. Els€asser, Phys. Rev. B 79, 245121 (2009).
• M. Rodahl, F. Höök, A. Krozer, P. Brzezinski, B. Kasemo, Rev. Sci. Instrum. 66, 3924
(1995).
• M. S. Chen and D. W. Goodman, Science 306, 252 (2004).
• M. T. Bohr, R. S. Chau, T. Ghani, K. Mistry, IEEE Spectrum. 44, 29 (2007).
• M. W. Shafer, J. B. Torrance, and T. Penney, J. Phys. Chem. Solids 33, 2251 (1972).
140
• N. C. Bristowe, J. Varignon, D. Fontaine, E. Bousquet and Ph. Ghosez, Nat. Comm. 6,
6677 (2015)
• N. Nakagawa, H. Y. Hwang, and D. A. Muller. Nat. Mater. 5, 204 (2006).
• N. Reyren, S. Thiel, A. D. Caviglia, L. F. Kourkoutis, G. Hammerl, C. Richter, C. W.
Schneider, T. Kopp, A.-S. Rüetschi, D. Jaccard, M. Gabay, D. A. Muller, J.-M.
Triscone, and J. Mannhart, Science. 317, 5842 (2007).
• N. Sai, B. Meyer, and D. Vanderbilt, Phys. Rev. Lett. 84, 5636 (2000).
• N. Singh, A. Agarwal, L. K. Bera, T. Y. Liow, R. Yang, S. C. Rustagi, C. H. Tung, R.
Kumar, G. Q. Lo, N. Balasubramanian, and D. L. Kwong. IEEE Electron Device
Letters, 27, 383 (2006)
• P. G. Steeneken, L. H. Tjeng, I. Elfimov, G. A. Sawatzky, G. Ghiringhelli, N. B.
Brookes, and D.-J. Huang, Phys. Rev. Lett. 88, 47201 (2002).
• P. Lomker and M. Muller, Phys. Rev. Mater. 3, 061401(2019).
• P. Lömker, T. C. Rödel, T. Gerber, F. Fortuna, E. Frantzeskakis, P. LeFèvre, F. Bertran,
M. Müller, and A. F. Santander-Syro, Phys. Rev. Mater. 1, 062001(R) (2017).
• P. Pasierb, S. Komornicki, M. Rekas, J. Phys. Chem. Solid. 60, 1835(1999)
• P. Ponath, A. B. Posadas, M. Schmidt, A.-M. Kelleher, M. White, D. O’Connell, P.
Hurley, R. Duffy and A. A. Demkov, J. Vac. Sci. and Technol. B 36 (2018).
• P. Ponath, A. B. Posadas, R. C. Hatch and Alexander A. Demkov, J. Vac. Sci. Technol.
B. 31, 031201 (2013).
• P. Ponath, K. Fredrickson, A. B. Posadas, Y. Ren, X. Wu, R. K. Vasudevan, M. B.
Okatan, S. Jesse, T. Aoki, M. R. McCartney, D. J. Smith, S. V. Kalinin, K. Lai, and A.
A. Demkov, Nature Comm. 6, 6067 (2015).
• P. R. Subramanian, Eu-O (europium-oxygen), in Binary Alloy Phase Diagrams, 2nd
ed., edited by T. B. Massalski (ASM International, Materials Park, OH, 1990), Vol. 2,
pp. 1665–1666.
• Q. Fu and T. Wagner, Surf. Sci. Rep. 62, 431 (2007).
• R. A. McKee, F. J. Walker, and M. F. Chisholm, Phys. Rev. Lett. 81, 3014(1998).
• R. A. McKee, F. J. Walker, and M. F. Chisholm, Science 293, 468 (2001).
• R. C. Alferness, IEEE Transactions on Microwave Theory and Techniques, 82,
1121(1982).
• R. C. Hatch, M. Choi, A. B. Posadas, and A. A. Demkov, J. Vac. Sci. Technol. B. 33,
061204 (2015).
141
• R. Konta, T. Ishii, H. Kato, and A. Kudo, J. Phys. Chem. B. 108, 8992 (2004).
• R. Ouillon, J-P. Pinan-Lucarre, P.Ranson, P. Pruzan, S. K. Mishra, R. Ranjan and D.
Pandey, J. Phys.: Condens. Matter 14, 2079(2002)
• R. Ramesh and D. G. Schlom, MRS Bull. 33, 1006(2008).
• R. Schafranek, S. Payan, M. Maglione, and A. Klein, Phys. Rev. B 77, 195310 (2008).
• R. W. Ulbricht, A. Schmehl, T. Heeg, J. Schubert, and D. G. Schlom, Appl. Phys. Lett.
93, 102105 (2008).
• R. F. Need, B. J. Isaac, B. J. Kirby, J. A. Borchers, S. Stemmer, and Stephen D.
Wilson, Phys. Rev. Lett. 117, 037205(2016)
• S. Abel, J. Fompeyrine, “Electro-Optically Active Oxides on Silicon for Photonics” in
Thin Films on Silicon (V. Narayanan, M. Frank and A. A. Demkov. eds.) 455 (2016).
• S. Abel, T. Stöferle, C. Marchiori, C. Rossel, M.D. Rossell, R. Erni, D. Caimi, M.
Sousa, A. Chelnokov, B.J. Offrein, J. Fompeyrine, Nat. Commun. 4, 1671(2013).
• S. Guha and V. Narayanan, Phys. Rev. Lett. 98, 196101 (2007).
• S. Hofmann, Auger- and X-Ray Photoelectron Spectroscopy in Materials Science: A
User-Oriented Guide, Springer Berlin Heidelberg, Berlin, Heidelberg, (2013).
• S. Kang, N. Prasad, H.C.P. Movva, A. Rai, K. Kim, X. Mou, T. Taniguchi, K.
Watanabe, L.F. Register, E. Tutuc, and S.K. Banerjee, Nano Lett. 16, 4975 (2016)
• S. Maekawa, T. Tohyama, S. E. Barnes, S. Ishihara, W. Koshibae, and G. Khaliullin,
Physics of Transition Metal Oxides, (Springer, 2004).
• S. P. KowaIczyk, N. Edelstein, F. R. McFeely, L. Ley, and D. A. Shirley, Chem. Phys.
Lett. 29, 491 (1974).
• S. Raghavan, T. Schumann, H. Kim, J. Y. Zhang, T. A. Cain and S. Stemmer, APL
Materials 4, 016106 (2016)
• S. S. Hong, J. H. Yu, D. Lu, A. F. Marshall, Y. Hikita, Y. Cui, H. Y. Hwang, Sci. Adv.
3, 5173(2017)
• S. T. Christensen, B. Lee, Z. Feng, M. C. Hersam, and M. J. Bedzyk, Appl. Surf. Sci.
256, 423 (2009).
• S. Tanuma, C. J. Powell, and D. R. Penn, Surf. Interface Anal. 36, 1 (2004).
• S. W. Cheong and M. Mostovoy. Nat. Mater. 6, 13 (2007).
• S. Zollner, A.A. Demkov, R. Liu, P.L. Fejes, R.B. Gregory, P. Alluri, J. Curless, Z.
Yu,J. Ramdani, R. Droopad, T.E. Tiwald, J.N. Hilfiker, J.A. Woollam, J. Vac. Sci.
Technol. B 18, 2242-2254 (2000).
142
• S. A. Chambers, T. C. Kaspar, A. Prakash, G. Haugstad and B. Jalan, Appl. Phys. Lett.
108, 152104 (2016)
• S. Abel, F. Eltes, J. E. Ortmann, A. Messner, P. Castera, T. Wagner, D. Urbonas, A.
Rosa, A. M. Gutierrez, D. Tulli, P. Ma, B. Baeuerle, A. Josten, W. Heni, D. Caimi, L.
Czornomaz, A. A. Demkov, J. Leuthold, P. Sanchis. J. Fompeyrine. Nature Material,
18, 42 (2019).
• S. G. Lim, S. Kriventsov, T. N. Jackson, J. H. Haeni, D. G. Schlom, A. M. Balbashov,
R. Uecker, P. Reiche, J. L. Freeouf, and G. Lucovsky, J. Appl. Phys. 91, 4500 (2002).
• T. Choi, J. Lee, Thin Solid Films 475, 283(2005)
• T. Hadamek, D. Shin, A. B. Posadas, A. A. Demkov, S. Kwon, Q. Wang, and M. Kim,
Appl. Phys. Lett. 111, 142901 (2017).
• T. Mairoser, J. A. Mundy, A. Melville, D. Hodash, P. Cueva, R. Held, A. Glavic, J.
Schubert, D. A. Muller, D. G. Schlom, and A. Schmehl, Nat. Commun. 6, 7716 (2015).
• T. Sakudo, Ferroelectrics, 95, 37 (1989).
• T. Schumann, S. Raghavan, K. Ahadi, H. Kim and S. Stemmer. J. Vac. Sci. Technol. A
34, 050601 (2016)
• T. Wang, A. Prakash, Y. Dong, T. Truttmann, A. Bucsek, R. James, D. D. Fong, J.
Kim, P. J. Ryan, H.Zhou, T. Birol and B. Jalan, ACS Appl. Mater. Interfaces 10,
43802(2018)
• T. Wang, L. R. Thoutam, A. Prakash, W. Nunn, G. Haugstad and B. Jalan, Phys. Rev.
Mater. 1, 061601(2017)
• V. N. Strocov, T. Schmitt, U. Flechsig, T. Schmidt, A. Imhof, Q. Chen, J. Raabe, R.
Betemps, D. Zimoch, J. Krempasky, X. Wang, M. Grioni, A. Piazzalunga and L.
Patthey, J. Synchrotron Radiat. 17, 631(2010).
• V. N. Strocov, X. Wang, M. Shi, M. Kobayashi, J. Krempasky, C. Hess, T. Schmitt
and L. Patthey, J. Synchrotron Radiat. 21, 32(2014).
• W. G. Nilsen and J. G. Skinner, J. Chem. Phys. 48, 2240 (1968)
• W. Guo, A. B. Posadas, A. A. Demkov, J. Vac. Sci. Technol. A 39, 030804 (2021)
• W. Guo, A. B. Posadas, Sirong Lu, David J. Smith, and Alexander A. Demkov, J. Appl.
Phys. 124, 235301 (2018).
• W. Hiinsch, Th. Vogelsand, R. Kircher and M. Orlowski, Solid-state Electron 32,
839(1989).
• W. Li, L. Gao and A. A. Demkov, Phys. Rev. B 102, 035308 (2020).
143
• W. S. Choi, D. W. Jeong, S. S. A. Seo, Y. S. Lee, T. H. Kim, S. Y. Jang, H. N. Lee,
and K. Myung-Whun, Phys. Rev. B. 83, 195113 (2011).
• X. Hu H. Li, Y. Liang, Y. Wei, Z. Yu, D. Marshall, J. Edwards, R.Droopad, X. Zhang,
A. A. Demkov, K. Moore, and J. Kulik, Appl. Phys.Lett. 82, 203 (2003).
• X. Zhang, A. A. Demkov, H. Li, X. Hu, Y. Wei, and J. Kulik, Phys. Rev.B. 68, 125323
(2003).
• X. X. Wang, H. N. Li, W.-H. Zhang, and F.-Q. Xu, J. Phys.: Condens. Matter 19,
096001 (2007).
• Y. Horio, Y. Hashimoto, A. Ichimiya, Appl. Surf. Sci. 100, 292 (1996).
• Y. Kamihara, H. Hiramatsu, M. Hirano, R. Kawamura, H. Yanagi, T. Kamiya, and H.
Hosono, J. Am. Chem. Soc. 128, 10012 (2006).
• Y. W. Chung and W. B. Weissbard, Phys. Rev. B 20, 3456 (1979).
• Y. Wei, X. Hu, Y. Liang, D. C. Jordan, B. Craigo, R. Droopad, Z. Yu, A. A. Demkov,
J. John, L. Edwards, and W. J. Ooms, J. Vac. Sci. Technol. B 20, 1402 (2002).
• Y.L. Du, G. Chen, M.S. Zhang, Solid State Commun. 130, 577(2004)