oxide heterostructures - adi abyoga
TRANSCRIPT
1
INTEGRATION OF 2-DIMENSIONAL ELECTRON GASES IN OXIDE HETEROSTRUCTURES WITH SILICON
By ADI SENO ABYOGA
Student Number : S1232991
Examination Committee : Dr. Ir. Mark Huijben
Prof. Dr. Ir. Gertjan Koster Dr. Stefan Kooij Jaap Geessinck
ADVANCE TECHNOLOGY FACULTY OF SCIENCE & TECHNOLOGY
INORGANIC MATERIALS SCIENCE UNIVERSITEIT TWENTE - NETHERLAND
July 2015
2
PREFACE
Many thank you, I deliver to God Almighty, because over the blessing of His grace I can complete this
Final Project. The final project was created in order to meet one of the requirements of program of
study Bachelor's degree, study field of The Advance Technology Application, The Faculty of Science and
Technology at The University of Twente, The Netherlands. I realized that without helping and guidance
of various parties since the lectures program until doing research and making report of this Final Project,
it is very difficult for me to finish the all of programs. Therefore I say many thanks to:
1. Prof. Dr. Hans Hilgenkamp as Dean Faculty of Science and Technology, University of Twente, the
Netherlands.
2. Dr. Ir. Mark Huijben, and Prof. Dr. Ir. Gertjan Koster as my supervisor and chairman, who helped
me understand the assignment better and guided me throughout the whole assignment.
3. Dr. Stefan Kooij, as my external committee member.
4. Jaap Gessinck as my daily supervisor who guided me through the experiments and patiently
helped me to better understand the assignment.
5. My father, Dr. IR. Suprapto Soemardan, MSc, my mother Dra. Dyah Nur Pinudji, my brother Aryo
Abyoga Adhyaksa, b. Eng, ST., MSc, which has also helped me a lot both moral and material.
6. Om Andreas Visschedijk and Tante Therese Minarni, as my host parents in Netherland which
helped me through the hard days in the Netherlands
7. And finally, my girlfriend Sivinadia who supported me through the whole assignment, Samuel,
Sandro, and all my friends in the Netherlands, and Indra Priawan, Syarief Nahdi, Abubakar
Adeni, and Hamsir Azhar, and all my friends in Indonesia for the constant supports.
This final project report may be still far from perfect, it needs to be a continuous refinement to be
additional rides science as one of my devotion to God Almighty as a servant and as the Great Creator.
May God the Almighty God always guide his servants in applied science as one of the great creations to
the path of righteousness and glory of human beings in the world.
Enschede, the Netherlands, June 30th 2015
The Author,
ADI SENO ABYOGA
3
ABSTRACT
Recently, it has been shown a precise control over the deposition of two insulating perovskite oxides,
which creates a two-dimensional electron gases at the interface between them. These materials are
then developed much further, and therefore, a realization of integrating to Silicon platform in order to
develop multifunctional nanoelectronics. In this study, the author demonstrate different combination of
treatments to the interface of Si//STO deposited with 10 unit cells of LAO by Pulsed Laser Deposition on
Si (001). The pre-annealing treatment to the substrate showed the best crystallinity so far. Reflection
High-Energy Electron Diffraction and X-ray Diffraction pattern indicate the deposition of 10 unit cells was
successful. Atomic force microscopy showed smooth surface on the pre-annealing treatment. The room
temperature electrical transport properties, and liquid helium temperature Hall measurement is done
and are comparable to recent studies. In addition, behavior at liquid helium temperature shows a
magnetoresistance effect present at the interface. These properties open up to possibilities of a more
precise pre-treatment control into a Silicon platform.
4
CONTENT LIST
Title 1
Preface 2
Abstract 3
Content List 4
Figure List 6
Table List 9
Appendices List 10
Chapter-1 INTRODUCTION 11
1.1 Background 11
1.2 Outline of Assignment 13
Chapter-2 THEORITICAL BACKGROUND 14
2.1 Materials 14
2.2 Perovskite Oxides 15
2.3 SrTiO3 (STO) 16
2.4 LaAlO3 (LAO) 16
2.5 LAO/STO Interface 17
2.6 Integration of 2DEGs on Si Platform 19
Chapter 3 – EXPERIMENTAL PROCEDURE 21
3.1 Standard SrTiO3 substrate treatment for TiO2 termination 21
3.2 Si Integrated With SrTiO3 22
3.3 Treatment Prior To Deposition 23
3.3.1 Cleaning Process 23
3.3.2 Annealing and Etching Procedure 23
3.4 Film Characterization Prior Deposition 25
3.4.1 Atomic Force Microscopy 25
3.4.2 X-Ray Diffraction measurement 26
3.5 Film Deposition By Pulsed Laser Deposition (PLD) 28
3.6 Electronic Transport Measurement 31
3.6.1 Contacts 31
3.6.2 Measurement Setup 31
Chapter 4 – RESULTS AND DISCUSSION 37
4.1 Growth Monitoring By RHEED 37
5
4.2 RHEED Signals Occurring During the Preparation of n-Type LAO/Si//STO
Interfaces 40
4.3 Film Characterization 41
4.3.1 AFM 41
4.3.2 X-Ray Diffraction 47
4.4 Electronic Transport Properties 52
4.4.1 Deposited 10 uc LAO - Annealing Treatment 54
Chapter 5 – CONCLUSION & SUGGESTION 61
REFERENCES 62
APPENDICES 67
6
FIGURE LIST
Figure-2.1 Schematic Compositions of Perovskite Oxides, p.14
Figure 2.2 Schematic sketch of the perovskite structure of ABO3, p.15
Figure 2.3 Sketch of polar catastrophe in the interface between LaAlO3 and SrTiO3 at (001), p.18
Figure 2.4 Final system of the heterostructure, p20
Figure 3.1 Configuration of the substrates inside the oven, p.24
Figure 3.2 Scattering x-rays from periodic lattice A and C, p.26
Figure 3.3a Schematic of the X-ray diffractometer (XRD in Munchen University of Technology), p.27
Figure 3.3b Illustration of the Euler angles and the angles Ω from XRD, p.27
Figure 3.4 The working principle of PLD [Stefan Thiel]. P.28
Figure 3.5 The PLD-ICE Machine (from University of Twente), p.29
Figure 3.6 Schematic of PLD-ICE in the software, p.29
Figure 3.7 Schematic of order putting the substrate, p30
Figure 3.8a Conventional resistivity measurement, p.32
Figure 3.8b Typical van der pauw measurement, p.32
Figure 3.9 Schematics of a sample mounted on resistivity puck specifically designed for Model
6000 PPMS Controller, p.33
Figure 3.10 Sketch of the experimental setup used for the transport measurement, p.34
Figure 4.1 RHEED integrated inside PLD for show total pulses were achieved during the deposition
of 10 unit cells of LAO, p.37
Figure 4.2 Schematic of RHEED (from Thesis of Mark Huijben), p.37
Figure 4.3a The RHEED pattern of 10 unit cells deposition of LAO (from thesis of Stefan Patrick
Thiel), p.38
Figure 4.3b The RHEED pattern of 10 unit cells before deposition of LAO to Si-STO with etched and
annealed treatment, p.39
Figure 4.3c The RHEED pattern of 10 unit cells after deposition of LAO to Si-STO with etched and
annealed treatment, p.39
Figure 4.4 RHEED oscillations during the deposition of 10uc of LAO on TiO2-terminated STO-Si,
peak intensity at (00) is plotted as a function of time, p.40
Figure 4.5a Clear scanned image of AFM of STO substrate, p.41
Figure 4.6 AFM image of surface topography Si-STO substrate, p.42
7
Figure 4.6a. The 2nd AFM image of surface topography Si//STO substrate with 0.240 nm RMS value
and 1.6 nm peak to peak value, p.44
Figure 4.6b The 2nd AFM image of surface topography Si//STO substrate with 0.08 nm and 0.5 nm
peak to peak value shows annealed treatment of Si//STO substrate (right), p.44
Figure 4.6c AFM image of surface topography with 0.3 nm RMS value 2.7 nm peak to peak value
shows annealed and etched treatment, p.44
Figure 4.6d AFM image of surface topography with 0.8 nm RMS value and 7.3 nm peak to peak
value shows etched and annealed treatment, p.44
Figure 4.7a AFM image of surface topography of Si//STO/LAO with annealing treatment with RMS
value of 0.09 nm and peak to peak value of 0.9 nm (left) and with annealing and then
etching treatment with RMS value of 0.36 nm and peak to peak value of 2.4 nm (right),
p.45
Figure 4.7c AFM image of surface topography of Si//STO/LAO with etching and annealing treatment
with RMS value of 0.7 nm and 4.3 nm peak to peak value, p.45
Figure 4.8 Scan of STO substrate, which shows the (100) peak at ≈22.7516o, (200) peak at
≈46.5096o, and (300) peak at ≈72.5891o (validated by the literature [1] using logarithmic
scale), p.48
Figure 4.9a Cleaned Si-STO as-received substrate with non-quartz zero background plate, p.49
Figure 4.9b Cleaned-and-annealed Si-STO substrate with quartz zero background, p.49
Figure 4.10 Diffractogram of Si//STO/LAO 10uc, the side peaks seems to be increased and
broadened, p.50
Figure 4.11 Combined diffractogram of Si-STO substrate before and after the deposition of 10 unit
cells of LAO. The picture is not so clear, but blue diffractogram is before deposition,
treated with annealing procedure, and red one is after the deposition, with etching and
annealing procedure, p.51
Figure 4.12 LAO peak observed around STO (001) peak, p.51
Figure 4.13 Diffractogram of Si//STO/LAO 10 uc after reducing the background noise much further
and stripping of Kα2 wavelength, p.52
Figure 4.14 Typical Van der Pauw measurements configurations, specifically this model is designed
for the annealed sample. The longitudinal measurements that were used, are vertically
or horizontally going from one edge to the other, while hall measurement that was
used, going diagonally, p.54
Figure 4.15 Temperature Dependent Resistance of Annealed sample from 300 K (up-left) and Sheet
Resistance from 250K (up-right), serves as a comparison is system of STO//LAO
presented by Brinkman et al (bottom-left and right), p.58
8
Figure 4.16 Plot of the sheet carrier density as a function of temperature for a sample treated with
pre-annealing on the Si//STO substrate. According to the sign of the Hall voltage, the
charge carriers are electrons, p59
Figure 4.17 Plot of Temperature dependence of ln(ns-n0) for Si//STO/LAO 10 unit cells where ns is defined as = -1/RH*e and n0 is the temperature limit of ns, p.59
Figure 4.18 Hall Resistance vs T annealed sample. Linearity is observed throughout the whole
temperature range, p.60
Figure 4.19 Magnetoresistance hysteresis of Annealed sample, p60
9
TABLE LIST
Table 3.1. Configuration of sample connections for measuring resistance across the sample, p. 32
Table 3.2. Configuration of sample connection for hall measurement and longitudinal
measurement of resistivity sample treated with annealed, p.33
Table 3.3 Configuration of sample connection for hall measurement and longitudinal
measurement of resistivity sample etched annealed and annealed etched, p.33
10
APPENDIX LIST
Appendix 1 As-received Si//STO substrate, p. 67
Appendix 2 Pre-treatment: Etching and Annealing Si//STO/LAO, p.67
Appendix 3 Pre-treatment: Annealing and Etching Si//STO/LAO, p.68
Appendix 4 Table: Sheet Resistances, Hall Coefficient, Sheet Densities, and Mobilities varied around
temperatures from 150K down to 2K, p. 69
11
Chapter-1. Introduction
1.1 Background
Many of the electronic devices nowadays based their functionality on the behavior of the interface of
their materials. Semiconductor is one of the examples [2]. In many years, research has shown that
semiconductors success is not based on their bulk characteristics instead it's the transport of charge
carriers across or along the interfaces between different materials. When the first bipolar transistor was
created by three scientists of Bell Labs, they found out that for a transistor to work, the addition of
impurities is necessary [2]. With the addition of impurities, the junctions of material with different
properties can be made that will give different characteristics of nonlinear voltage current. These
phenomena can be seen on the interfaces of the junction.
When they first add the impurities, it was done in macro engineering [3] where they alloyed the
semiconductor with the materials that in the end will give desirable properties. Then new technology is
developed where they use diffusion of the doping atoms into semiconductor material from gaseous
source [4]. And finally it came along the latest technology whereby growing the surface of an oxide on
top of the exposed semiconductor materials [5].
There are several aspects that are essential. First, the odd behavior in which they found at the interface
of those materials led to a new branch of quantum physics which was surface physics. Second, the
further development of electron mobility in a semiconductor is not only applied to the semiconductor
itself, but many of the other working devices nowadays. The third thing is the miniaturization of these
devices where down-scaling of such device to a nanometer scale [2].
The understanding of these three aspects is important to this assignment. Within a surface or interface
of materials, the bulk property of materials can be altered in many ways. If one take the relative
contribution of electronic interface reconstruction where spins and charges are manipulated on the
interface of two layers, thus there will be distribution of charge in those interfaces. In addition, an
effective electron doping is created at the interface [6]. Another example is that the possibility to alter
the crystal lattice and position of atoms in the interface by having a structural deformation. The entire
industry of semiconducting materials is possible due to the manipulation of point defects [7]. For
example: the doping of the Si, or oxygen vacancies for fast ion conduction pathways. Or even, the
research has shown that it is possible to alter a stoichiometry composition of one material in order to
have a more stabilized interface [8]. Combination of such interface effects allow to design desirable
transport properties of materials.
Oxide materials, however, show rather intriguing behavior rather than semiconductors. Oxides are
delicate materials where they show a variety of amazing electronic and ionic phenomena. These
phenomena can be shown in their intrinsic functionalities for example ferromagnetism [9], or Ferro-
electricity [10]. Moreover, oxide materials are a versatile material where one can also tune in other
parameters such as the oxygen vacancy [11].
12
The challenge lies in manipulating interface of oxide materials, such that the bulk property will be
affected as well. Thus, a realization comes where it is possible to make a hetero-structure of these oxide
materials. In the past few years, the development of many advanced technology led to a very good
experimental setup for creating, manipulating and characterizing the oxide materials. Therefore, it was
called Molecular Beam Epitaxy (MBE) and Pulsed Laser Deposition (PLD) that has shown a promising
result in creating the oxides layer with very good quality in a sense of layer-by-layer atomic surface [12].
Another major development of recent technology is that clean room now allows a control for unit cell
that creates hetero-structures of the oxides. Different types of oxides can be created in same structure
which will show different behavior at interacting interfaces. A control of the geometry of the unit cell
also allows for a better structure which will have a very-matched epitaxy. The Epitaxy here refers to the
deposition of over-layer crystal on another crystal, which can be either the same substrate
(homoepitaxy) or different substrate (heteroepitaxy). A good and matched epitaxy is desirable in order
to have a smooth terrace thus having a smooth interface as well. Sometimes property of materials can
be anisotropic (being directionally dependent), which is required in the interacting interface having a
smooth terrace.
During the last few years there were many researches cultivated such hetero-structure with different
type of oxides. In 2009, the Advanced Materials Journal published an article that explains the structure-
property relation of SrTiO3/LaAlO3 by single terminating the oxides into their constituting layer by layer
sequence [6]. The deposition of these stacking layers is done by either PLD or MBE, putting one layer
adjacent to another. The amazing thing was that the interface showed a conducting behavior, which was
very odd since the bulk property of either STO or LAO is the insulating.
Ohtomo and Hwang from the Bell Laboratories were the only people that reported the first observation
of high-mobility electron gas at the interface between the two insulating perovskite oxides of LaAlO3 and
SrTiO3 [12]. Their discovery about the fundamental mechanisms in which the oxide interface is
manipulated through electronic reconstruction where the spreading charges across polar/nonpolar
interface causes an effective electron doping on the interface, is very important to this system. Thus,
generating free electrons that are ready to conduct. These electrons are called two-dimensional
electron gas (2DEG). The extension of this statement means that the electron only confined only in two
dimensions, able to move only in lateral plane. The interface of these materials are very interesting, but
this is only limited to a research based application. Integrating such 2DEGs on a Si platform would surely
create a new novel of electronic devices. Fortunately, this has been done before with another system of
heteroepitaxy structures, which was LaTiO3 and SrTiO3 [5]. But, will this kind of experiment be
reproducible with SrTiO3 and LaAlO3 interface? The expectation will surely answer yes.
The experiments are done in University of Twente. The deposition process is limited only to Si//STO as
received substrate (it is expected that the interface is insulating), which then deposited with 10 unit cells
of LAO. STO as a single crystal has clear step-and-terrace structure, which allows for a closely matched
epitaxy when deposited with LAO. The challenge lies in enhancing the cyrstallinity of Si//STO substrate
since grown film is expected to have lots of amount of defects. Different pre-treatments are expected
to enhance its crystallinity, such as pre-annealing to the substrate, and also wet chemical etching by BHF
13
treatment. In addition, different combinations of pre-treatments will have different effect on the
crystallinity.
The research questions that the author would like to answer is then, "How should be kind of treatment
given to get kind of electronic transport properties? Is it one is better than the other? What would be
the characterization of the surface look like? And in the end, does the reproducibility of generating
2DEG on Si platform successful in this kind of system of heterostructure?"
1.2 Outline of the Assignment
Chapter 2 will start with a brief introduction of the perovskite oxides. Individual substrate properties of
SrTiO3 as well as LaAlO3 will be explained. The theory behind the interacting interface of LaAlO3/SrTiO3 is
further explained where the observed conductivity between these materials can be intuitively
understood by electronic reconstructions caused by polarity discontinuity, along with general views of
possible mechanism from many studies. The first goal is then to grow a heterostructure of these
insulating materials with a growth control in unit cells in a clean room laboratory where it often has
been done in NanoLab of University of Twente. Lastly, the heterostructure is then integrated on Si
platform.
Chapter 3 will explain further the applied experimental methods for fabricating the substrates. It will
start with the explanation of the treatment of the as-substrates and the single termination process as a
theory since the as-received substrate is Si-STO and LAO. Furthermore, annealing and etching treatment
is explained further. The film growth and its deposition of 10 unit cells of LAO will be monitored by
Pulsed Laser Deposition and Reflection High Energy Electron Diffraction. In addition, the characterization
will be done by using the Atomic Force Microscopy, the X-Ray Diffraction will be used to check the
crystalline quality and the Quantum Design Physical Property Measurement System will be used to
investigate the transport properties.
Chapter 4 will report the experiment results mainly result of transport properties and open discussion of
results for further studies, whether it can be enhanced by another pretreatment, or something else. Last
chapter as chapter 5 will close this report with a conclusion.
14
Chapter 2 - Theoretical Background
2.1 Materials
The interface effects in heterostructures can be intuitively understood through knowledge from the bulk
properties of the constituting materials. This chapter briefly explains the oxides material that were used
in the experiments, explaining further the structure and properties. It starts with perovskite oxides, and
then the constituting materials of the heterostructure, which are SrTiO3 (STO) and LaAlO3 (LAO). The
Researchers have shown that both of the materials are band insulators. Then it describes the interface
effect of LAO/STO, and also explains in general the reason behind the observed conductivity, and in
particular the polarity discontinuity model. Finally, integration to Si-platform is necessary in order to
make a working device.
The next subchapter will explain the explanation of perovskite oxides. The transition metal oxides are
versatile in their electrical properties in which it can be altered using many ways.
2.2 Perovskite Oxides
The first mineral of perovskite was found by Russian mineralogist, L.A. Perovski , which is CaTiO3 [13]
[14]. The compounds of perovskite exist abundantly in nature. The basic structure of perovskite is
constituted of ABX3 where A is the first metal cation, B is the second metal cation, and X is an anion that
ranges from O2-, Cl-, Br-, I-, and other halogen. The variety cations of A and B has various oxidation state,
it could be divalent, trivalent, or even in the example of CaTiO3, A is divalent and B is tetravalent [6]. The
two examples are LaAlO3 and SrTiO3 where La is trivalent and Al is trivalent, while Sr is divalent and Ti is
tetravalent. For further example, figure 2.1 gives the ions occupying A and B lattice site.
Figure 2.1. Schematic Compositions of Perovskite Oxides [15]
Perovskite structures are very diverse in terms of the chemical composition. Depending on the
constituting cations of A and B, vacancies stabilization, and the oxygen stoichiometry, various degrees of
freedom of perovskite structures can be achieved [1]. For example, the anion that can be placed in X is
O, where ABO3 can condense in various crystal structures. The ideal perovskite, however, is simple cubic
which what most people generally imply. Perovskite structure shows many interesting behavior such as
ferroelectric, piezoelectric, or has a very high dielectric constant [16].
15
The ideal cubic cell of perovskite structure has ion A located at the cube corner positions (0, 0, 0); ion B
located at the body center position (1/2, 1/2, 1/2); and ion O is located at the face centered positions
(1/2, 1/2, 0). The simple cubic structure is shown in Figure 2.2a. It formed an octahedron of BO6 where B
is located in the middle of octahedron and O lies in the corner around B (see Figure 2.2b). They formed
an extended (3D) network of corner-sharing BO6 octahedral. The label A can be any larger cation than B,
ranging from calcium, potassium, sodium, strontium, lead, cerium (or various other rare metals), and so
on, occupying the 12-fold coordinated sites between the octahedral [17].
It is the best to describe a simple cubic structure in Figure 2.2c, which is aligned along the (001) direction
that can be seen as a stacking sequence of alternating AO and BO2 layers along the c-axis [1][6]. The
process is called single termination process where the original crystal structure of ABO3 is made in such
a way, that one layer is comprised of AO layer, and the other is BO2 layer. Single termination process is
needed in order to achieve extremely clean surface, single crystalline, and defect free substrate, where
the surface of the substrate is atomically well defined. A good initial surface quality means has low
kinetics which is favorable for growth condition of one layer to another. Specific treatments are needed
in order to achieve a good single terminated substrates, one of the example is chemical treatment using
BHF (buffered Hydro-Fluoric Acid) which will be explained further in chapter 3.
Figure 2.2: Schematic sketch from the thesis of Thiel of the perovskite structure of ABO3. a). the cubic unit cell has B ions
in the corners, an A ion in the center, and O ions in the middle of its edges. b). Perovskite structure with BO6 octahedral
network highlighted. c). The ABO3 compounds can also be viewed as a layered sequence of AO-BO2 sequence which will be
most useful description in the assignment [1].
16
2.3 SrTiO3 (STO)
SrTiO3 has been used for many research regarding oxides layers where it serves as a base ground for the
constituting heterostructure. There are many applications regarding the usage of SrTiO3. For example,
SrTiO3 is used as a high performance capacitor for analog applications [18], it is also possible to grow a
high Tc superconductors [19]. Furthermore, SrTiO3 is a non-polar compound and an inert compound
which is perfect for the single termination process into AO-BO2 layer and it is also non-reacting with the
deposited materials [1][6].
SrTiO3 has an indirect band gap of around 3.25 eV and a direct band gap of 3.75 eV which is considered
as a band insulator [20][21]. Band gap is an energy range in which there is no electron exists. In order for
electron to conduct or make transition from valence band to conduction band, electron has to receive
some portion of energy from external source exceeding the band gap threshold. Because the value of
the energy band is very large, it is then called as a band insulator.
At room temperature, SrTiO3 sits at simple cubic structure where the lattice parameter is 3.905 Å. The
Ti atoms located at the corners and the Sr atoms at the centers of the cube [6]. As it is mentioned
before, simple cubic structure performed an octahedral of BO6 where B is located in the middle of
octahedral and O lies in the corner around B, in this case BO6 is TiO6 octahedral that has perfect 900
angles extending in three dimensions. It has six equal Ti-O bonds at 1.952 Å, and each of Sr atom is
surrounded by 12 equidistant oxygen atoms at 2.760 Å [6].
The SrTiO3 compound has second order phase transition from cubic to tetragonal at a temperature
between 105 to 110 K, this is still under argument, since several articles mentioned either at 105 K or at
110 K, due to rotation of neighboring TiO6 octahedral in opposite directions [15]. In the ionic limit, SrTiO3
can be described as Sr2+Ti4+O2-3.
2.4 LaAlO3 (LAO)
LaAlO3 is the second material that is needed to build the heterostructure. It has a band gap of around
5.6 eV which is also an insulating material. At room temperature, the structure is a rhombohedrally
distorted perovskite structure. Rhombohedral crystal structure is not explained in detail since this
assignment will deal only with simple cubic structure, when the deposition process is done, the
substrate is heated to a temperature of 850oC, and then LAO will undergo a transition to simple cubic
structure at 813K. LAO is a dielectric material with εr=24 at the temperature range of 300K which can be
functioned in microwave superconducting resonators, filters, and antennae due to its relatively low
losses at microwave frequencies [22].
The structure of LAO can be described as having the same BO6 octahedral structure with anti-phase
rotation [6]. Its lattice parameter is 3.791 Å. There will be some small lattice mismatch. But in fact, the
small mismatch between LAO and STO interface is not so big, which is only 3%, although there will be
renormalization at the surface, it still allows the epitaxial growth of LAO films on STO.
17
2.5 LAO/STO Interface
The origin of the charge carriers at the interface of LAO/STO is still under debate for the past years.
Ohtomo and Hwang proposed the electronic reconstruction model to explain the interface physics of
LAO/STO interface [12]. However, there is other explanation that could contribute to this, for example
Boschker et al [23] published results that argue the model with surface reconstruction model. In this
subchapter, a general view of several mechanisms is presented along with their current research.
The first one is based on the polar catastrophe. Nakagawa et al [24] in his paper theoretically explained
that the phenomena can be intuitively understood by single terminating each material into its
constituting AO-BO2 layers that has different charge. While STO has a neutral charge of Sr2+O2- and
Ti4+O22-, LAO on the other hand, are respectively +1 and -1 charge on La3+O2- and Al3+O2
2-.
In perovskite heterostructures, this stacking sequence is maintained hence there will be a polarity
discontinuity arises at their interface between LAO/STO. Then a diverging electric field potential arises
due to the electric fields between the oppositely charged layers in LAO that needs to be compensated
by electronic reconstruction when the energy is no longer able to accommodate by internal deformation
[25]. It has been found that the conducting behavior is found when the stacking sequence is in the order
of SrO-TiO2-LaO-AlO2 (n-type) and the other stacking of TiO2-SrO-AlO2-LaO (p-type) remains insulating.
The Ti ion allows for mixed valence charge compensation, resulting in the net transfer of nominally 0.5
electrons per two dimensional unit cells from LAO to STO across the interface. Similarly, on the other
stacking sequence, it must require extra holes per two dimensional unit cells. In addition, Figure 2.3
from Nakagawa et al which is presented below to see the sketch of polar catastrophe mechanism. This
can be intuitively understood by article of Chen et al [26] where he gave the example of charge transfer
between bulk vanadate to bulk manganite which is happen through the band alignment.
One way to understand the physical background of this phenomenon is by using a device called SHG
(Second Harmonic Generation). It's a nonlinear optical process in which photons with the same
frequency interact with nonlinear material, and then they are effectively combined to generate new
photons with twice the energy. The required surface and details are explained further in reference [27]
and also by Savoia et al [28].
This model has not been observed experimentally [23][29]. Through quantitative calculation, it can be
achieved that 3-4 unit cells are the critical thickness where the interface showed conducting behavior,
charge transfer also occurred for the n-type LaO/TiO2 but it has not been observed that there were
potential build up which suggest that, it may be not the electronic reconstruction that takes place, but
another contribution such as chemical or structural reconstruction.
Hans Boschker in his thesis suggested that instead of electronic reconstruction, surface reconstruction
takes place that takes into account the geometry of the constituting layers [23]. He did an experiment to
test whether the potential build up was present in the interface of LAO/STO. By doing XPS (X-ray
Photoelectron Spectroscopy), he managed to find potential build up, which can be seen by broadening
of the peaks of the XPS. The XPS experiment showed that the potential buildup in the LAO layer is
18
smaller than 0.5eV, much smaller than the band gap of STO, hence electronic reconstruction is not
viable anymore. Instead surface reconstruction takes place that depends on the growth condition, which
will create impurities that contribute in the induced polarization which then conduct the electron [23].
Figure 2.3. Sketch of polar catastrophe in the interface between LaAlO3 and SrTiO3 at (001). The oxidation numbers are
denoted by superscript in each element, ρ is the alternating net charges, E is the non negative electric field, and V is the
potential. In (a), the interface between LAO and STO (LaO-TiO2 stacking or n-type) is still unreconstructed, leading to a
non-negative electric field which create potential that diverge with the increasing thickness. (c) In order to construct
neutral atom, the system is then transferred half an electron from LaO layers to above and below. The electron is
transferred to empty 3d levels at bottom of the conduction band belonging to adjacent Ti atoms since the 3d states is
localized, and Ti can change its formal oxidation state from Ti3+ to Ti4+. The diverging potential is then avoided. In (b) the
potential diverges negatively in p-type stacking (AlO2-SrO stacking) , and in order to avoid it, (d) removal of half and
electron from SrO plane is necessary.
Wolter Siemons proposed another idea that included oxygen vacancy is the main cause of the
phenomenon. In the sense of the upper limit in the intrinsic case, the charge that is necessary to prevent
the polarization catastrophe is equal to half an electron per unit cell [30] . So the influence of extrinsic
doping will surely affect this number, and thus cannot apply when the situation is at an enormous
amount of charge densities. The article furthermore argued whether it can be proved by experiment the
influence of oxygen vacancies in STO substrate. The sputtering effect that takes place in the La ions to
the oxygen ions from the substrate results in the increase of conductivity on top of STO layer. The article
did an experiment on varying the oxidation level, several characterization experiments are done under
these different oxidation levels, which then conclude that large sheet of charge density observed at
these hetero-interfaces due to oxygen vacancies donating the electron in STO substrate.
19
It is therefore important to understand different perspectives in which these phenomena are built
under. Although further explanation is not given, it is in general clear what might cause the effect of this
conductivity between these two insulating layers.
2.6 Integration of 2DEGs on Si Platform
The phenomena behind these interfaces are fascinating and have much attracted attention to explore
integration such property to Si platform. The advanced developing of film thin deposition technique
allows people to control the growth of heterostructure with a very good quality. This could lead to a
new novel of electronic device that has less energy usage since thermal conductivity of silicon is higher
than those typical oxides such as STO which has 12 W/mK while Silicon has 149 W/mK.
Nevertheless, integrating such properties into devices is not as easy as it seems. Normal deposition of
LAO to STO is easier due to small mismatch. When STO is deposited on Si, large mismatch is observed
since Si has 5.431Å and STO has 3.9Å for the lattice constant. It can be concluded that the surface is not
as smooth as the normal deposition of LAO to STO. Some obstacles are expected can be present as
shown in [31]. For example, the control of termination process of STO to generate 2DEG is crucial, while
TiO2-STO presented metallic 2DEG behavior, the interface between LAO and SrO terminated is
insulating. This means that the as-substrates received of Si-STO that we got has to be in TiO2 terminated,
and at this point, it is not known yet whether the Si-STO is completely TiO2 terminated. Another thing is
the crystalline quality of STO platform. Big lattice mismatch will have an effect on the strain induced
state (which is not explained in detail in this assignment) which will cause defects. Some defect will
localize carriers which hamper the transport properties [32], but the others can scatter the carriers [33].
Therefore, several parameters mentioned above can be served as an indicator whether the substrate is
smooth or not. Its goal is then to have different treatments to four substrates of the Si-STO samples to
compare and then make a conclusion in the end which combination of treatments will give the best
transport properties.
There are four samples that were treated with different combinations of treatments. One as-received
substrate of Si-STO served as a comparison, treated only with cleaning process and deposited with 10
unit cells of LAO, and the other three are as follow: one Si-STO substrate is annealed and deposited, one
Si-STO substrate is annealed then etched with BHF and deposited, and last one is then etched first with
BHF and annealed and then deposited. In order to check the crystallinity and surface quality, then it is
important to conduct AFM and XRD for all of the substrates which is also done in MESA+ Building in the
University of Twente.
Previous research which was done by Jin et al has shown that 2DEG is produced on similar kind of
heterostructure [5]. Promising results has been shown since 2DEG is generated on Si (001) platform. The
heterostructure, however, is very different. The substrate, Si (001) wafer is deposited first with 4.5 unit
cells of buffer layer STO, then 10 unit cells of LaTiO3 (LTO), and then 15 unit cells of STO. The presence of
buffer layer helps to accommodate the difference in lattice constant between LTO and Si (001) platform,
moreover, the lattice mismatch between LTO and STO is much smaller than between LAO and STO, LTO
has a lattice constant of 3.97Å, hence, smoother surface is expected in this heterostructure compare to
20
this assignment. It is expected also by addition of the buffer layer, as well as having smaller lattice
mismatch, the transport properties is then also much better. Further discussion will be explained in
chapter 4. In short, the system would look like in figure 2.4 below
Figure 2.4 Final system after deposition
21
Chapter 3 - Experimental Procedure
This chapter will discuss the experimental methods in which the fabrication and structural
characterization take place for Si//STO/LAO substrate. The chapter will cover the techniques for pre-
treatment of the substrate and film deposition, AFM (Atomic Force Microscopy), and XRD (X-Ray
Diffraction). At the end of this chapter, the setup for the electronic transport measurement is
presented, in which the device that will be used is QD-PPMS (Quantum Design - Physical Property
Measurement System).
The termination process in which STO is treated, and Si-STO deposition process is not done by our
research group, STO with TiO2 termination were bought from Crystec in Berlin (Germany), while STO-Si
were given by Dr. Mark Huijben (Professor of University of Twente). The as-received substrates are
cleaned, annealed, and etched with different sequence. Characterization is done in order to check the
crystallinity and interface quality. It is important to notice the theory of treatment behind the as-
received substrate of both STO and Si-STO.
3.1 Standard SrTiO3 substrate treatment for TiO2 termination.
For this purpose of the assignment, the substrate that is used is (100) oriented single-crystalline SrTiO3.
The substrates were bought from Crystec in Berlin (Germany) with a dimension of 5x5x0.5 mm3 and a
miscut from 0.10 to 0.20. The miscut later will cause a small problem due to the small step and it will
cause a harder single termination process.
The initial as-supplied substrates are first cleaned in order to remove dust and containment. The
substrate is cleaned in acetone for 10 minutes by ultrasonic bath, and then dried off with nitrogen gun
above a blue clean room tissue. Notice that the appropriate tissue in case of dealing with cleaning of
thin films or substrate, it has to be made from thick fiber, able to absorb fluids and attach the dust
particle stronger. Then STO is dipped again with acetone and ethanol (isopropanol could also work) in
around 10 minutes.
At this moment, the STO has a mixed surface termination of SrO and TiO2. Fill another beaker with DI
water and put the STO in the ultrasonic bath for 10 to 30 minutes, in order to hydrolyze the SrO. On the
other hand, the TiO2 is chemically stable and remain unchanged by the DI water. The forming of SrO
later will be etched in the next step, which then will result in TiO2 single terminated STO.
For the etching procedure, BHF (Buffered Hydro Fluoride acid) is used with the ratio of NH4F : HF = 87.5 :
12.5, this is the most commonly BHF solution used specifically for the etching procedure of many thin
films. The STO is brought in an ultrasonic bath with the BHF solution. The procedure needs a delicate
handle since BHF solution is quite acidic, after that, the substrate is rinsed with DI water and dried with
blue industrial tissue and nitrogen gun, and the remaining waste has to be dumped away in the
appropriate waste can.
The last procedure is the annealing method in order to have an improved crystallinity and remove
residues from previous treatment steps. The tube oven has to be cleaned prior to the annealing
22
procedure with an oxygen gas flow of 150 L/h, which is the most numbers used. The temperature is set
to 950oC for 2 hours and then kept constant for the next 1.5 hours before the oven is switched off to
cool.
3.2 Si integrated with SrTiO3
Interface effect of crystalline oxides has proven to be conducting, creating two dimensional electron
gases at the LAO/STO interface. Thus, the realization comes to a point where the integration of such
heterostructure into a Si-based platform could create a multifunctional device. The desirable methods
are needed to combine such heterostructure with high stability. Nevertheless, the high reactivity of Si
with many elements including oxygen, presented a big obstacle, forming amorphous silica phase when
Si is exposed with an oxygen environment as well as the lattice mismatch between Si and the grown
films.
Molecular Beam Epitaxy (MBE) is another method for a deposition process that allows a high control of
the purity of the grown films. Its deposition rate is typically less than 3000 nm per hour, which is
comparable to other deposition techniques. In addition, the absence of carrier gas makes an increase of
the purity of the grown films.
The first epitaxial growth of STO on Si has been shown by McKee et al [34], depositing layer by layer of
substrate starting from one monolayer until five monolayer. Monolayer is a single, closely-packed atom
arrangement. Each subsequent monolayer is differed by its constituting element, starting from alkaline
earth element, which was strontium, and then introduced the oxygen on the third monolayer and
fourth, which in the end gave an epitaxial growth of STO on Si (001). This is based on the concept of
layer-by-layer energy minimization corresponding with interfacial electrostatics (ions near-neighbor
interactions). Ion sizes and charges of the constituting stacking sequence give contribution to the
initiation of heteroepitaxial growth. In this case, heteroepitaxy can be achieved by having a TiO2 plane of
the perovskite structure. This is the requirement for single-orientation, epitaxial growth of STO on Si.
The samples considered in this assignment were fabricated on n-type conducting Si (001) by solid source
MBE. Strontium (Sr) and Titanium (Ti) metal effusion cells and needle valve controlled high purity
molecular oxygen source were used for STO deposition. The substrate is comprised of 80 mono-layers
of STO (around 320Å to 370Å or 32 nm to 37 nm). The difference in charge forces the electron stick onto
the surface. The front side is polished, which can be seen by its reflective surface, while the back side is
not polished. It is expected that Si-STO as-received substrate is insulating, therefore it is important to
deposit LAO layers on top to create 2DEG. It is important to notice that the as-received substrate of Si-
STO is a TiO2 single terminated, but since the lack of information, this assignment will do combination of
treatments to make sure that everything is single terminated.
23
3.3 Treatment Prior to Deposition
3.3.1 Cleaning Process
The cleaning process is done in the chemical laboratory of MESA+ Building in the University of Twente,
and all the substrates from the single crystal of STO and 4 substrates of Si-STO are treated the same way
to remove dust particles that might be lying on the surface of the substrate. The substrate should be
checked first under microscope to see whether it's clean or not. Cleaning with acetone and ethanol is
the standard cleaning procedure.
The cleaning procedure as follow: Two beakers were prepared and cleaned beforehand with demi
water. The beakers were each prepared for acetone and ethanol. The choice of alcohol can be either
ethanol or isopropanol. Isopropanol has more capability of cleaning dust particle and dissolve organic
substance than ethanol since it content less water, hence isopropanol is more expensive as well.
Acetone is a good substance for cleaning since it's very polar, that can dissolves most organic substances
easily, as well as having the ability of being miscible to water, so it can be used simultaneously with
water. Specifically in this experiment, the choice of alcohol that will be used in this process is ethanol.
The blue industrial tissue is put inside the fume hood. Then it is best to mark either the beaker or blue
industrial tissue to indicate which one is acetone and ethanol. Then acetone and ethanol were poured to
each of the beaker. First step is to put the substrate into acetone beaker, and ultra-sonicating it for 5
minutes, then after 5 minutes, rinse the substrate with a bit of ethanol since acetone vaporize very fast,
then next step is to put substrate for another ultra-sonication for 5 minutes inside the ethanol beaker.
After 5 minutes, the substrate was dried with nitrogen gun, and to check whether the surface was clean
enough or not, light microscope below 10x zoom is used. To double check, it is possible also to clean the
substrate with soft tissue poured with ethanol, and dried it with nitrogen gun. All of substrates, the STO
single crystals and Si-STO were treated the same way.
3.3.2 Annealing and Etching Procedure
The annealing procedure is done inside the chemical laboratory of MESA+ Building in University of
Twente. The oven is a drying type quartz tube oven. A glass tube oven mostly has small capacity and
works under reduced pressure. The heating part is made of quartz. Commonly, drying type glass tube
oven has a desiccant agent to absorb the evaporated moisture from the sample.
Prior to the annealing and etching procedure, the best way is doing check under microscope whether it
is really clean after the cleaning process, if there still any dirt spec, it would be best to scrub the sample
using a soft fiber tissue rinsed a bit with ethanol and then dried with Nitrogen gun. After that, the quartz
tube has to be cleaned as well with ethanol, and then dried with a Nitrogen gun.
24
There are 3 substrates that were annealed, sample that is marked number one is treated with an etching
procedure prior to the annealing procedure, while samples that were marked with number three and
four are just normal sample which has to be annealed. The temperature is set at 850oC, which take time
around 30 minutes or more to reach such temperature, and then it stayed constant for 60 to 90
minutes. The cooling down take a longer time, that is usually overnight. It is important not to forget to
set the right oxygen flow, which is 150 mL/h. Since we are dealing with perovskite oxides, oxygen is
needed as a background to keep in the composition and kinetics constant.
Figure 3.1 shows the configuration order of the substrates inside the oven.
The etching procedure as follows: the sample marked with number one is cleaned again, with ethanol,
as usual prior to the etching procedure, Si-STO has to be put inside a glass beaker which was filled with
DI water inside an ultrasonic bath for 30 minutes to hydrolyze the Si-STO sample. BHF is used to etch the
hydrolyzed molecule (SrOH). The BHF is buffered hydrofluoric acid, which has some ratio between
Ammonium Fluoride and hydrofluoric acid (NH4F : HF), most common one is usually 1:3 ratio.
There were 4 beakers that are used for etching procedure, one was glass beaker, and the others are
plastic beakers which will not get dissolved by BHF. The two other plastics beakers are filled with DI
water, and one is filled with BHF while the glass beaker is filled with ethanol (ethanol somewhat
improve the cleaning and able to create a smoother surface). Prior to etching, the researcher has to
wear protective equipment such as glasses, blue gloves in the inside and another plastic glove, and an
apron. BHF needs a very careful handling since it is very acidic and dangerous.
The substrate is then etched for 30 seconds inside the plastic beaker filled with BHF, and then put inside
beaker which has water to stop the etching and dilute the remaining acid, and then put inside last
beaker which has water in it to further clean it. It is important to stir at every beaker in order to get
optimal result. After that, the sample is put inside the ethanol.
The effect of annealing and etching procedure will be discussed further in chapter 4, results and
discussions.
25
3.4 Film Characterization Prior to Deposition
3.4.1 Atomic Force Microscopy
In order to check the single termination process is properly or not, substrate has to be analyzed with
AFM and XRD prior to the deposition process. The Atomic Force Microscopy, we called as AFM, will not
give the exact composition of complete TiO2 termination, but it can be easily seen that one substrate is
single terminated or not through the step that is preserved in the scanned image. Nevertheless, the
treatment method is very common and has been used many times for treatment of single termination
process of STO, and its validity has been proved through many experiments from previous researches.
AFM is a measurement device that makes image of the topography of the surface with very high
precision. It uses a cantilever with a sharp tip, brought close to the surface. As the name implied, force is
measured between the substrate and the tip. The tip is scanned over the whole surface to create
topographical image of the surface [35].
There are several modes that can be used in AFM. For example, the contact mode is used when the tip
is in contact with the surface. The tip will get deflected when it encounters obstacles along the scanning
process. Since the cantilever is made of piezoelectric material, which induces voltage, this voltage is
then used as a measurement of topography.
The mode used in this assignment is called the tapping mode. In tapping mode, the cantilever is not in
contact with the surface rather it oscillates near the surface close to its resonance frequency. Distance
between surface and cantilever is such that the top touches the surface only at the lowest point of
deflection. The closer distance the tip gets to the surface (toward an uphill slope) lesser the deflection,
and force becomes attractive which reduces the amplitude. The feedback loop is used to lift the tip
accordingly until the amplitude reaches its set point again, which is held constant. The topographical
image that is produced is based on the measure of the height the tip which moving accordingly to.
A laser and photo-detector is used to measure the amplitude of the cantilever. The laser is pointed at
the cantilever and reflected off to a photo-detector, for the setup of the AFM. As the cantilever
oscillates, the laser beam will scan along a line on the detector.
26
The experiment is mainly done using Vecco (Digital Instruments) in tapping-air standard mode.
Antimony (n) doped Si cantilever and probes with resonance frequencies of 320 kHz and force constants
of 42 N/m were used. The cantilevers were back coated using 40 nm of Al for high-reflectivity and had a
tip radius curvature of 3.8 um (Bruker). Although the scanning size is limited to 2x2 µm2 with this device,
the topographic features of only few Å high can be achieved. To reduce the effect of vibrations on the
measurement the AFM was usually operated inside a chamber on a stone place, and then closed to
reduce the noise from outside. The scanning area was 2x2 µm2. The height of z-axis is changed from
standard of 14.6 µm to 1 µm for better resolution. Samples or line was changed from 256 to 512 for the
same reason. The same parameters are used throughout every sample. The results of scanned images
substrates will be discussed further in chapter 4.
3.4.2 X-Ray Diffraction measurement
XRD is a tool used for identifying the atomic and molecular structure of a crystal, in which the crystalline
atoms cause a beam of incident X-rays to diffract into many specific directions. By measuring the angles
and intensities of these diffracted beams, a crystallographer can produce a three-dimensional picture of
the density of electrons within a crystal. From this electron density, the mean positions of the atoms in
the crystal can be determined, as well as their chemical bonds, their disorder and various other
information. In this chapter, a general overview of XRD is given.
The wavelength of Xrays is typically around 1Å[36], which means that it is comparable to interatomic
spacing in solids. In other words, when incident beam of x-rays hit a crystal surface, the incident beam
will get diffracted at some angles θ. Spacing can then be deduced from separating the diffraction
maxima, or in other words, measuring relative intensities of different orders information about the
structure of the lines. Suppose there is a periodic lattice of a crystal structure, the integral number of
wavelengths coming from left side indicated by number 1,2,3 which then will get scattered indicated by
1', 2', and 3'. The constructive interference is then scattered from lattice points of A and C, if the
distances BC and CD are equal, in other words, BC+CD is the additional distance in which in phase
condition for constructive interference takes place.
Figure 3.2. Scattering x-rays from periodic lattice A and C. (Courtesy of http://pubs.usgs.gov/of/2001/of01-041/htmldocs/xrpd.htm)
27
Specifically, the substrate is aligned so the Bragg condition is fulfilled for each of the as-substrate
subsequent monolayers along the vertical direction. If the substrate is completely mono crystalline, the
only one peak will be shown on the detector. Suppose there is a periodic lattice of a crystal structures,
that consists of not only one plane of atoms, but two, since BC = CD = d sinθ.
The Bragg condition in real space satisfies that:
This is true for constructive interference. The miller indices then indicate which planes for specific
reflection shows.
The schematic of XRD is shown in Figure 3.3a and Figure 3.3b and 3.5b [37]. According to Figure 3.5,
there are two tubes that create XRD, on the left there is source tube which shoot the X-ray laser or
specifically, the laser that is used is Cu-K-alpha 1 radiation with a wavelength of 0.154098 nm and on the
right there is a detector tube which receives the diffraction pattern made by the lattices.
The scan mode that is used throughout this assignment is Gonio scan or what is also known as θ-2θ
scan. The sample and detector moving coupled at different velocities, and maintaining the relationship
of θ=ω. Gonio scan grants the proper focusing in Bragg-Brentano Geometry or when the divergent and
diffracted beams are focused at fixed radius from sample position. This is very useful to perform
experiments where the objective is to indicate other crystalline phases incorporated in crystal system or
heterostructures [37], which will be shown further in chapter 4. The aligning process itself will be
explained further in Appendix (note that this standard procedure comes from the author’s supervisor,
doing the gonio scan mode, it could be that for different types of scan mode, the alignment will be
different).
Figure 3.3a and 3.3b from practicum manual of XRD of Munchen University of Technology. Figure 3.3a shows a schematic
of the X-ray diffractometer, and figure 3.3b shows the illustration of the Euler angles and the angles Ω (corresponding to
the same ω angle in our case) and 2θ.
28
3.5 Film Deposition by Pulsed Laser Deposition (PLD)
In this subchapter, the deposition of thin films on the terminated substrates is discussed. For this
assignment, Pulsed Laser Deposition (PLD) was used. This method will be described in general, along
with the particular procedures and parameters that are used for the growth of LAO films on Si-STO
substrates.
PLD is quite common in the research base area of thin film deposition. There are many advantages of
PLD, such as being able to ablate wide range of materials, to control the growth rate and quality, and
with some optimization, to transfer as well defined stoichiometry of the material. However, PLD can be
quiet costly, and slow in deposition rates as well as small in coated areas, which hamper the use of PLD
in scale of big industry.
Figure 3.4 is taken from thesis of Stefan
Thiel, the working principle of PLD [3].
As it can be seen from Figure 3.4, high energy laser pulses entered the ultra-high vacuum chamber, and
directed onto the target material. The threshold energy of the laser has to be higher than the material in
order to move the particles from the materials when they collided. The ablated materials which is called
the plume which expand away rapidly (around 106 cm/s), deposited on to the heated substrate when
the plume collided, and then condensed back. In principle, it is quite a simple process, but it can be
tricky as well since the parameters of the growth can be tuned, such as laser influence, background gas
pressure, and substrate temperature which allow the film properties to be manipulated. In addition, it
allows for stoichiometry transfer of a material if the parameters are adjusted accordingly. The PLD
machine is operated by software which is called PLD-ICE control.
29
Figure 3.6 shows schematic of PLD-ICE in the software. By clicking the valve, it will either turn green which means closed
or red which means opened.
The main chamber and transfer chamber is connected via a load lock. The Load Lock system allows the
transfer of carousel or heater can be mounted and exchanged without breaking the vacuum pressure.
To open and take the carousel or heater out, we first click the Rough Load Lock in order to adjust the
pressure from the inside pressure, decreasing pressure and closing the fore-line valve, so that the main
Figure 3.5. The PLD-ICE Machine (is taken from University of Twente),
30
turbo chamber is not connected. The main turbo chamber produces a high pressure due to having a very
high rotary movement (≈1000 rpm). Once it is done and closed, we pressed the Vent Load Lock in order
to adjust the pressure to be same as in the outside, and then we can take it by opening the small panel
of load lock. The LL in the schematic above indicates Load Lock.
Prior to every deposition the target, a stoichiometric LAO single crystal, is polished on fine sandpaper
(grit 500, Buehler CarbimetTM) with a grinder, then wiped on a hard lens of blue industrial tissue, rinsed
with ethanol and blown dry with nitrogen. Then the target is transferred to the deposition chamber and
mounted on the turn-able carousel, which carries it to four different targets for the in situ growth of
complex heterostructures, although only LAO target was used. The aligning process is proved to be
difficult task, avoiding as many impacts as possible to the screen, and the holder of the heater. Another
thing is difficult that the carousel needs to be firmly fixed, otherwise the mass cannot be supported by
the long tube holder, and it will be dropped inside the main chamber.
Our Si-STO substrate is then glued to a heater by silver paste (Demetron), then heated to a temperature
of 100oC and let it dried for 30 minutes, providing good thermal contact.
Figure 3.7 shows the schematic of order putting the
substrate. 1 is treated with cleaning process, 2 is
treated with cleaning, etching, and annealing process, 3
is treated with cleaning, annealing, etching process, and
4 is treated with cleaning and annealing.
However, its problem is when we were trying to mount the heater. Unfortunately, the mount holder is
somewhat pushed away two of the substrates marked with number 1 and 2. We contacted Frank, and
we opened the LL valve, reducing all pressure to ambient pressure. It was lucky that substrate number 2
was found, but not substrate number 1, keep in mind that for the rest of the report that there won't be
any substrate number 1 (which was the as-received cleaned substrate after the deposition). It is
suspected that it might get absorbed by the pressure. Aligning needs to be performed with follow the
instruction from the software.
Next step is aligning the laser. There are several steps of aligning laser. First of all, the laser needs to be
exactly at two points showed in the Figure 3.8. A KrF excimer laser with λ=248nm, pulse width around
20 to 30 ns (LPX200, Lambda Physik). The pulse energy was 27.9 mJ and in the end we got an average of
31
28 mJ, and an energy density of 1.3 J/cm2. The change in aperture allow to adjust amount of ablated
materials.
As a final cleaning step, a pre-ablation is performed using the same parameters later used for the film
deposition. During pre-ablation, the shutter is maneuvered to exactly in front of the substrate. The
oxygen pressure was set to 5 x 10-5 mbar. Since we are dealing with perovskite oxides, oxygen is needed
as a background pressure in order to change the kinetics of the particle. There is a small chance that a
reaction will be heated much further and explode, and oxygen to minimize the risk by oxidizing. The
substrate then will be heated to 850oC by ramp it up, and the pressure is also set in the same value.
When the temperature is reached, the heater will be kept stable, then laser is ablated and film growth is
performed.
The laser pulse rate is set at 1 Hz, and distance between target to substrate is also set around 3 to 5 cm,
changing the substrate distance is also a factor for growth. Kinetics will be changed when the distance is
varied. For example, less debree can be produced when the distance is too far away, but that does mean
that the deposition could be slower. The cool down over several hours is started, during the current
through the resistive heating block is reduced step by step.
For the growth of LAO films need 5 hours for cooling down with a ramp up of 10oC/minute. In this
assignment, the parameters noted in the previous paragraphs will be referred as the standard values for
the deposition of LAO (p(O2)≈ 5 x 10-5 mbar, T≈850o C, energy density ≈ 1.3 J/cm2. The next paragraph
will discuss the aligning process and in situ growth monitoring by RHEED.
3.6 Electronic Transport Measurements
3.6.1 Contacts
A good contact has to be made and special handling is needed since the electronically conducting
interface of LAO and STO is encapsulated by insulators. Several methods have been reported. For
example, Ohtomo and Hwang used the laser annealed, Ohmic contacts to reach the buried interface [12,
p. 3]. Another method reported was a sputtered gold pads using Ti adhesion layer on the sample by
Kalabukhov et al [38]. However, the contact method of the sample used is wire bonding contacts that
was done many times previously in this research group of University of Twente [39] using aluminium
wire.
3.6.2 Measurement Setup
In typical resistivity measurements, 4 electrical wires are attached to a rectangular sample as shown in
Figure 3.10a. A current is fed from contact A to B, and voltage is measured across contacts C and D,
meanwhile resistivity can be derived from voltage drop across contacts C and D, the applied current, and
the geometry of sample. Unfortunately, our sample does not have that kind of specific geometry, but a
thin film of 5x5x0.5 mm3. Instead, a typical Van Der Pauw measurement is used where four wires are
connected in the corner. The advantage of Van Der Pauw method is that it can be applied to any
arbitrarily shape of sample. The common technique is used to find resistivity of material, the doping
type, sheer carrier density, and the mobility [40][41].
32
Figure 3.8 a conventional resistivity measurement Figure 3.8 b typical van der pauw measurement
The basis measurement is that it has to be in such way that current flow along one edge, and voltage
along the other edge. A resistance is then calculated using Ohm's Law:
In Van der Pauw's paper, to measure the sheet resistances from an arbitrary shape, can be measured by
two resistances, first resistance along vertical edge, another along the horizontal edge. The Van der
Pauw formula of sheet resistance is then:
Normally, reciprocal measurements can be obtained, , also its reversed polarity
measurements can be obtained. But since the time is limited, then only one type of measurement is
done along the same direction. The sample connections are then made in Table 3.1.
Table 3.1. Configuration in sample connections for measuring resistance across the sample
St and Sb indicate that sample is put at top box, and the other is bottom box. At the top box is put
sample treated with annealed, and sample treated with etched and annealed, while at the bottom box
put sample as-received substrate without treatment, as well as the annealed and etched sample. In
plane longitudinal and transverse measurement resistivity can be observed by this sample connection.
Then, Hall measurements can be conducted using different connections in the sample connections, as
shown in Table 3.2 for “as-received and annealed” sample and Table 3.3 for “etched annealed and
annealed and etched” sample.
I+ I- V+ V-
RxSt Ch1 5 4 3 6
RySt Ch2 5 3 4 6
RxSb Ch3 12 13 14 11
RySb Ch4 12 14 13 11
33
Table 3.2. For hall measurement and longitudinal measurement of resistivity sample treated with annealed
Table 3.3 for hall measurement and longitudinal measurement of resistivity sample etched annealed and annealed etched.
Figure 3.9 Setup for electronic transport measurements; Schematics of a sample mounted on resistivity puck specifically designed for Model 6000 PPMS Controller.
I+ I- V+ V-
RL1 Ch1 12 13 14 11
RL2 Ch2 12 14 13 11
RH1 Ch3 12 11 14 13
RH2 Ch4 14 13 11 12
I+ I- V+ V-
RL2 Ch1 11 14 13 12
RH2 Ch2 12 11 14 13
RL3 Ch3 4 5 3 6
RH4 Ch4 5 6 3 4
34
Wire-bonding the contacts are done by Mark Huijben (the author’s supervisor), samples were mounted
on connector boards. Resistivity puck is specifically designed for the PPMS Model 6000 controller, where
PPMS device able to monitor the helium level, temperature, magnetic field, as well as other parameters.
The helium level has to stay at level above 60%, therefore the cooling process is not hampered when the
helium level reaches under 50%.
Setup of transport measurements is shown schematically in Figure 3.12. Electronic sample properties
are analyzed using computer controlled current and voltage sources (PPMS Model 6000 Controller) is
equipped with multi meters, which are connected to the sample probe. Different program is used to
measure the temperature dependent resistivity from 300 K down to 4 K. The temperature is controlled
automatically using the PPMS from a program that is written in computer. As can be seen in the
schematics, sample has to be inserted inside the small tube, where the tube is filled with helium to cool
down. After the sample is put inside, option for the PPMS mode can be chosen, that is Resistivity option.
Figure 3-10 Sketch of the experimental setup used for the transport measurement.
When the board is mounted inside the helium tube, it usually adjusts its excitation current of its active
channels, but it can be controlled and set. In this assignment, the current limit is set to 5 µA, voltage
limit at xx volts, and power limit of xx, in order to get a good measurement resolution. The step size is
then 2.44 nA (according to PPMS hardware manual). Calibration mode is set to Fast Mode, reading the
last current and voltage to directly calculate the sample resistance and reports the resistance without
adjusting the reading. The sequence of cooling process is as follow: from 300 K down to 10 K with
different cooling rate where at low T, the rate goes much slower in order to get better resolution and
reading in low temperature regime.
35
From the signs of the Hall voltage acquired later, it can be obtained type of charge carriers whether it's
electron (n-type) or a holes (p-type). As it has been said before, the Silicon platform that was used is n-
type, so it is expected that the experiment will clarify this. The sheet carrier density is obtained from:
With B, the applied perpendicular magnetic field, and q the charge of the carriers.
The configurations in figure 4.14 also can be used as combination of one another. The possible reversed
polarization as well as the reciprocity will give a better numbers in the end. Each permutation, 2 (two)
measurements are done, having eight total measurements in total supposedly, and averaged by dividing
the sum of all measurements by 8 (eight). But, because the limited time given in this assignment, only
one configuration that possible as mentioned before in chapter 3, experimental procedure, having a
total of 2 (two) measurements configured in Rx and Ry. The van der pauw equation is then becomes:
The sheet resistance is obtained through matlab file given by the author's supervisor, Mark Huijben, as
attached in the appendix, the matlab file can be seen there.
It has been shown before that sheet resistance is direction dependent [42], when measuring the sheet
resistance, measuring in x-direction will give a significant different value than measuring resistance in y-
direction.
Sheet resistance can also be calculated by using the equation by geometrical way:
Supposedly, to calculate sheet resistance with this equation, one has to make an assumption that:
Where: Rver=Ry and Rhor=Rx.
Normally this assumption can't be made, since it needs eight other measurements to make sure that the
value of Rver is approximately the same with Rhor. In the end, the calculation of sheet resistance is
calculated using only the matlab file.
The hall coefficient is calculated through this equation:
36
Where VH is the hall voltage, t is the thickness of the sample, “I” is the current fed through the sample, B
is the magnetic field, n is the sheet carrier density, and q is the elementary charge.
Calculating hall coefficient can get the information of sheet density, the type of charge carrier, as well as
its mobility. But it is not possible to get the hall voltage since the current is not known from the data of
PPMS measurement. One intuitive way to calculate the value of hall coefficient is to simply getting the
slope of plot between Resistance (or resistivity in this case) with the varying magnetic field since:
So it is possible to rewrite the equation above by simply substituting the Resistance, and crossing off the
thickness of the sample since it is negligible. The equation then becomes:
And R corresponds to the resistivity obtained through the measurement with the varying magnetic field.
Be aware that the units and dimensions are very important here. The slope that is obtained from the
plot R versus B corresponds to the hall coefficient, which then it is possible to extract the information to
obtain sheet density. The mobility can then be calculated from equation:
37
Chapter-4. Results and Discussions
4.1 Growth Monitoring by RHEED
The deposition process until finish took time about three minutes and 13 seconds. An amount of layers
or ten unit cells were deposited and carefully controlled by the software. By pressing 'layer finished'
button every time the desired layer is finished, the growth can be controlled very well. As can be seen in
the Figure 4.1, to reach 194 pulses are needed 193.4 seconds average. This growth can be easily
monitored by having in situ Reflection High-Energy Electron Diffraction (RHEED) integrated inside the
geometry of PLD.
I
Figure 4.2 shows the geometry of RHEED integrated inside the PLD system. An electron gun is located
perpendicular to the line of heater and carousel, and on the other side is a Phosphor screen with a CCD
camera to observe the diffraction pattern made by the scattering electron from the crystal surface. As it
can be seen also from figure below, a small diffraction pattern is observed on the camera, with three
spots on top of one big spot.
Figure 4.2 is taken from Thesis of Mark Huijben
which shows the schematic of RHEED [29]
Figure 4.1 shows how many pulses were achieved during the deposition of 10 unit cells of LAO
38
The electron gun was carefully aligned and directed to the crystal surface with an acceleration of 35 kV
and 1 A, controlled by the RHEED power generator. Prior to the depositions, the electron beam direction
is adjusted electrically to the center of the extension tube for optimal performance. There were some
obstacles at first when aligning the RHEED due to the deflection of x and y is not straight enough.
Another problem was that to enable RHEED in high pressure condition, it is necessary to keep the
distance of travelling path of the electrons as minimum as possible [43].
A well oriented crystalline structure normally shows a surface characteristic that has a symmetric spot
pattern. As has been mentioned before, the difference between Si and STO lattice mismatch is way
larger than those compared to the normal deposition of LAO to STO. Here we can see the comparison of
the normally RHEED pattern of deposition of 10 unit cells LAO to STO, before the deposition of 10 unit
cells LAO to Si-STO substrate and after the deposition of 10 unit cells LAO to Si-STO substrate. Based on
our observation, The RHEED pattern at number two substrate, indicates combination result from
cleaning, etching, and annealing treatment.
The RHEED is aligned in such a way that the beam is almost parallel to (001) film direction with a grazing
angle around 0 to 0.3o. In principle, once the specular reflection image of the incoming electron beam is
observed clearly, as the long range order along the beam direction is increased, it will eventually evolve
into a scattering pattern of the substrate. The flatness of atomic surface is then determined a clear
coherent scattering of the electron due to surface of crystal.
For example, the RHEED pattern of Figure 4.3a from Thesis of Stefan Patrick Thiel [15] is a good example
where the STO substrate was oriented, so that the electron beam was incident at a grazing angle and
almost parallel to crystal (001) direction. The figure also suggests that it was RHEED pattern of
essentially perfect crystalline order extending over thousands of unit cells. The RHEED pattern shown in
Figure 4.3a shows the normal deposition of LAO to STO.
Figure 4.3 d RHEED pattern of 10 unit cells deposition of LAO from thesis of Stefan Patrick Thiel, serves as a comparison.
39
Figure 4.3 b before the deposition of 10 unit cells of LAO to Si-STO with etched and annealed treatment.
Figure 4.3 c after the deposition of 10 unit cells of LAO to Si-STO with etched and annealed treatment.
Figure 4.3b and 4.3c are results from this assignment. Figure 4.3b shows the RHEED pattern of Si-STO
substrate before the deposition of LAO and Figure 4.3c is the RHEED pattern of LAO to Si-STO substrate
taken during the deposition process, all of them was deposited at the same amount of 10 unit cells of
LAO. The direct beam is observed in all figures located right below the pattern. A specular reflection and
the adjacent first order spot can be clearly observed in Figure 4.3a, indicated the symmetric spot
pattern, thus a well oriented crystalline pattern. Figure 4.3b has a more elongated pattern, meanwhile
Figure 4.3c on the other hand, has a very distinctive specular spot, where one spot is more elongated
than another, and on top of that the adjacent first and second order spot can be clearly observed which
also experienced an elongation.
A typical RHEED image after a thin film deposition of 10uc of LAO on STO is shown in Figure 4.3a in
thesis Thiel proves that the perfectly flat substrate surface and its single-crystallinity is maintained
throughout the LAO film growth, which proceeds layer-by-layer. Also no reconstruction induced
additional spots are found. Surface irregularities and steps lead to a broadening of the reciprocal rods,
40
and streaks appear in the RHEED pattern of Figure 4.3b and 4.3c. When the surface gets rougher and
more three-dimensional the diffraction pattern also contains spots from transmission-diffraction
through islands and large particles on the surface, which gives rise to an elongated rectangular pattern.
It can also explain that in real space, LAO is epitaxial with Si-STO substrate, but the lattice spacing of
second phase is slightly off.
Another thing that might cause the elongation is not on the irregularity surfaces, it might be that the
orientation is a bit off. It can be seen from Figure 4.3b and 4.3c, that there is a tilt in the pattern, unlike
Figure 4.3a which shows a regularly spaced plane of atoms above the specular spot. In addition, usually
even if there is a misfit in lattice constant, symmetry is usually observed, the assymetric pattern can be
also caused by the miss-orientation.
4.2 RHEED Signals Occurring During the Preparation of n-Type LAO/Si-STO Interfaces
Through monitoring of RHEED intensity, controlling the deposition and taking time into account is
possible. The intensity of diffracted electrons due to surface of the substrate is measured at one or
several screen pixels, and the values are continuously monitored on a chart or graph as shown in Figure
4.4. RHEED intensity oscillations are observed to provide information about nucleation, and completion
of the topmost monolayer [44].
Figure 4.4 shows the RHEED oscillations seen during the deposition of 10uc of LAO on TiO2-terminated STO-Si, peak
intensity at (00) is plotted as a function of time.
The starting of intensity is high, and then dropped significantly after the first deposition. The intensities
at the second maxima are still slightly smaller, so that usually here the RHEED intensity is increased
manually by increasing the filament current in the electron gun. Starting from third maxima, the
41
intensity is start increased again but not in significant value. After several more oscillations the height of
the maxima reaches an approximately constant value. Oscillations observed are not clear since the
intensity is not increased from the power source. The drop of intensity can be caused by difference in
position of optimal diffraction condition.
By using a laser pulse rate of 1 Hz, during PLD, one unit cell is deposited in about 19.3 seconds
corresponding to a typical growth rate of ≈ 1.6 nm/min. This timescale allows stop the film deposition at
the desired moment during the evolution of the signal, usually at a certain maximum in intensity (as can
be seen in Figure 4.4). As noted before, the period of the oscillations always corresponds to the growth
of one unit cell. Unfortunately, there was an error on the machine so we could not get the full-width at
half maximum (FWHM) of the specular spot intensity peak. We could not really show the maxima and
minima in FWHM, hence there is not complete picture about the complete coverage or half covered
layers.
4.3 Film Characterization
4.3.1 AFM
In this section, AFM images are discussed in a much detail for every substrate. As a comparison, STO
substrate is scanned with AFM in order to have a good basis for the surface roughness and preserving
step-and-terrace structure of 1 unit cell. Si-STO as-received substrates are also scanned with only
cleaning procedure, then several images of different combination of treatments before and after the
deposition of 10 unit cells of LAO were also taken and discussed further.
-+
Figure 4.5 a shows a clear scanned image of AFM of STO substrate served as a basis of comparison whether a substrate is atomically smooth or not, and whether it has a very clear step-and-terrace structure along with its height profile (0.4 nm step is preserved).
RMS value and peak to peak value is given in order to have a better picture for the surface roughness.
Large number means the large deviation normal to the real surface with respect to its ideal form [45]
Figure 4.5 shows an AFM image of STO substrate. The image of the film shows very smooth step-and-
terrace structure, step of one unit cell is achieved as can be seen from the height profile at about one
42
unit cell high (≈3.9Å). The image is filtered by matching its median height to reduce the noise. Its RMS
value calculated from Gwyddion software is 0.12 nm (the graph not shown in here) with peak to peak
value of 0.8 nm. The usual RMS value for an as-received substrate of STO film is around 1 nm, with
considering there is no treatment as be done previously [46]. Taking a diagonal line of one corner to
another, the contribution taken is only from the amplitude, not taken the spatial contribution (tip) into
account for the RMS value. If it is confined to an area for the roughness measurement to a single
atomically flat surface, then the RMS value that can be obtained is 0.065 nm which really is same with
peak to peak value in Gwyddion. The peak to peak value is called Rt, which is obtained by measure the
absolute value between the highest peak and lowest one. Different in peak to peak value indicates that
defects which are observed in the surface. STO single crystalline substrate indeed has a better substrate
quality than STO film grown on top of Si (001). It can be explained further with quality factor, but the
extension result of this assignment has not yet reached that point.
Feature of half this height were not observed, which confirms that only one termination is present. The
terrace orientation and width are determined by unavoidable misalignment (for our STO substrate is in
between 0.10to 0.20 or only few hundred of nm) of substrate's cutting plane with the STO crystal planes.
Hence the terrace geometry remains the same over a complete substrate.
STO can serve as a good basis to understand how smoothness of a surface of single crystalline structure.
Our substrate, however, is STO film deposited on top of Si (001) using MBE. It is expected that Si-STO
does not have smooth surface compare to the STO substrate, and also the step-and-terrace structure
will not be so clear compare to STO single crystalline substrate due to the fact that there will be lattice
misfit, therefore a strain created on the interface [47].
Figure 4.6 AFM image of surface topography Si-STO substrate
43
As expected, Figure 4.6 shows the scanned image of as-received Si-STO substrate which only has
cleaning treatment. Although there is a gradient of color which shows the terrace-and step structure,
but the fact that it is STO deposited on top of Silicon, no clear step-and-terrace structure is observed as
compared to Figure 4.5 above. In addition, small spots indicate that the surface is rough. The RMS value
is obtained at 0.4 nm with peak to peak value of 3.8 nm.
Several scans are also done in order to check what the obtained scan from previously was accurate or
not. The average RMS value of 5 scanned AFM images for Si-STO as-received substrate treated with
cleaning procedure is 0.24 nm (not shown here). It is important to notice that most of the AFM images
taken could also have a noise from many contributions, also setting the right parameters, such as
imaging gains or bandwidth is not taken into account, the standard number of integral grain is 0.4 and
1.2 for proportional grain.
The surface morphology of STO films deposited on Si surface was characterized using same parameters,
scan size of 2 x 2 µm2, 512 samples/lines, the height of z-axis is changed to 2 µm. This chapter will also
discuss further the 'before and after' deposition image of three different treatments. It is expected that
surface will gets smoother after the annealing procedure to 950oC. In addition, etching procedure will
create etch pit. Surface characterization using AFM is further used in order to check the surface
morphology before and after deposition of 10 unit cells of LAO. As it has been reported before, the color
stretch range of every figure are different, since it is the optimal color stretch to obtain a high quality
AFM scanned image.
In the annealing treatment (Figure 4.6 b and 4.7a, it is indeed that the crystallinity of the surface is
enhanced. Previously on the as-received substrate from figure 4.6 or 4.6a is observed that the surface is
very rough with no clear color differences, in addition, white spots are observed throughout the whole
area. After the annealing treatment, the white spots are disappeared. The RMS value is decreased from
an average value of 0.24 nm to 0.008 nm before the deposition and 0.009 nm after the deposition. The
decreased value of RMS can be explained by annealing. Annealing to a certain temperature, in this case
950oC give pressures energy to the atoms, in addition, the mobility of the atoms increased, and atoms
move in such a way to an optimum position, where it is energetically favorable than previous position.
Peak to peak values are observed, reduced as well after the annealing treatment, which indicate the
segregation of bulk elements are reduced as well.
44
Figure 4.6a, 4.6b, 4.6c and 4.6d present the LAO AFM images at before deposition of 10 unit cells. Figure
4.7a, 4.7b and 4.7c present LAO AFM images at after deposition of 10 unit cells with keep in mind that
as-received Si//STO substrate was lost during the deposition, so only three pictures are presented here.
Figure 4.6a. The 2nd AFM image of surface topography Si//STO substrate with 0.240 nm RMS value and 1.6 nm peak to peak value (left) and Figure 4.6b with 0.08 nm and 0.5 nm peak to peak value shows annealed treatment of Si//STO substrate (right).
Figure 4.6c AFM image of surface topography with 0.3 nm RMS value 2.7 nm peak to peak value shows annealed and etched treatment (left) and Figure 4.6d with 0.8 nm RMS value and 7.3 nm peak to peak value shows etched and annealed treatment (right).
45
Figure 4.7 a AFM image of surface topography of Si//STO/LAO with annealing treatment with RMS value of 0.09 nm and peak to peak value of 0.9 nm (left) and with annealing and then etching treatment with RMS value of 0.36 nm and peak to peak value of 2.4 nm (right).
Figure 4.7 c AFM image of surface topography of Si//STO/LAO with etching and annealing treatment with RMS value of 0.7 nm and 4.3 nm peak to peak value.
In general, annealing a substrate can induce three kinds of process, recovery or intrinsic stress
liberation, recrystallization or structural improving, and grain growth or control over the growth in both
microstructure of a metal or thin film metal [48][49]. Qualitatively, annealing enhances surface
morphology of a crystal or any material depends with temperature and time. This can be explained by
several phenomena, for example, relaxation process driven by surface diffusion, or atomic
46
rearrangement driven by thermally activated energy as shown before in different annealing treatment
of Si-STO substrates as reported by Suwardy et al [50]. The small increase in RMS value after the
deposition of 10 unit cells of LAO indicate the increase in roughness due to the deposition itself, surface
reconstruction or manipulation of charge as has been explained in the possible phenomena behind
2DEGs as is discussed in Chapter 2.
The other two combinations of treatments were the etched and then annealed, and last one is annealed
and then etched. Wang et al proved that SrTiO3 treated with pre-annealing, etching, and then post-
annealing has smoothest surface, it should be expected that the surface roughness of treatments with
annealing and then etched should be smooth as well in this our case [46]. Wang et al mentioned that if
chemical etching is combined with thermal annealing, and then controlling the BHF pH value, and
etching time is controlled very well as well, substrate with nearly perfect and atomically flat surface
would have been obtained [51].
Figure 4.6c and 4.7b represents the Si//STO substrate treated with annealing and then etching before
the deposition. It shows a rough interface. It is expected that the surface is indeed rough. This is due to
the time length of annealing process that is not long enough, which was around 10 hours or so as has
been done by Wang et al [46], therefore holes are created. The pillar that is observed is Ti oxides, since
the Sr oxides is etched and then evaporated, which can be indicated by the increase in peak to peak
value. The increase in RMS value is caused by the etching procedure, which create holes in the surface.
The RMS value, however, increase to 0.36 nm due to the deposition of LAO.
Figure 4.6d and 4.7c shows a scanned AFM image of Si/STO substrate treated with etching and then
annealing procedure. The surface itself looks comparably smooth with other figures, unfortunately the
holes created after the etching procedure is quite deep, which is around 2.5 nm long and 0.4 nm step is
not preserved in this combination of treatments. The RMS value of both before and after the deposition
increased significantly, with a decrease after the deposition. Etching procedure increases the surface
roughness prior to the thermal annealing procedure. This is one of the disadvantages of wet etching,
since the system cannot be controlled during the etching procedure. An increase of some amount of
time could increase the roughness of the surface than it is expected. The expectation was that the
etching was mainly done to strontium hydroxide, although Ti ions also could be etched along the way
without proper control of time and temperature. It is also observed different shape of etch pit, the
annealing and etching procedure gives a more circular pit, while etching and annealing procedure gives
more square shaped. This depends mostly on the crystal structure of the substrate itself, it could be that
there are more open spaces, which is energetically more favorable with respect to the underlying crystal
structure, in this case Si substrate.
Due to large lattice mismatch difference between Si and STO, and STO and LAO, there are a strain build
up in the interfaces. This strain is created at the interface of Si and STO, and STO and LAO relaxes when
the so-called critical thickness is surpassed [52]. The critical thickness for Si and STO has not a clear value
of how many unit cells of STO needed to surpass the critical thickness for strain relaxation, but Bhuiyan
et al [53] showed that the flatness and crystallinity of STO films with thickness of around 100 to 300 Å
were better than other samples that has <100Å and around 1000Å. In our assignment, the STO has a
47
thickness of 80 Monolayers (ML) or approximately 370 Å, which can be concluded that our surface is
smooth enough. However, critical thickness of LAO that needs to be deposited on top of STO is 25 unit
cells, which is way more than this assignment had. In general, strain relaxation of films with a rough
surface occurs by twinning or by formation of dislocations, while smooth films, like the ones of our
samples, relax by cracking [54] . Although such cracks are not observed in our Si/STO/LAO
heterostructures since the critical thickness is not achieved.
However, this is the case when one talks about a nearly-perfect single crystalline substrate which
deposited with another nearly-perfect single crystalline structure in case of STO deposited with LAO. Si
(001) on the other hand is not perfect crystal, where it is expected to have lots of defects when it is
deposited with STO on top. There is no clear step-and-terrace structure as can be seen in Figure 4.5
compared to all figures in both Figure 4.6 and 4.7. In nearly-perfect single crystalline substrate like STO,
an etching procedure that produce a single-termination process to TiO2 stacking only, generates 2DEG
easily when it is deposited with LAO. On the other hand, the Si//STO substrate has a lot of defect that
also can be a line defect in the surface. The defect even worse when it got etched, it could produce a
large etch pit or trench that could not produce 2DEGs as can be indicated by a large value of peak to
peak. Annealing on the other hand, reduced the defect density which creates a smooth surface, able to
create conduction pathways rapidly. Interestingly, the etch pit becomes square if the annealing
treatment is done first before etching treatment.
4.3.2 X-Ray Diffraction
X-ray diffraction is used to check the crystallinity of the substrates achieved before and after the
deposition of 10 unit cells of LAO. It is important to notice that the diffractogram shown in this section is
not a complete picture of different treatments that will affect the crystallinity.
The substrates that are used in this assignment are thin films substrates. The substrate peak that were
used mostly is the Silicon substrate which has (004) peak at 69.171o. The program used to scan STO
substrate was Gonio scan. Gonio scan is also known as θ-2θ scan. The diffraction profile or diffractogram
is measured using 0D detector, rotated around the sample and keeping in the same angle of ω=θ. The
first scan was STO substrate in general, to serve as a basis for comparison. The scan range was from 16o
to 85o, with a step size of 0.0016 for around 30 minutes.
Figure 4.8 shows the typical STO substrate where the observed peaks for (001), (002), and (003) peaks
are present. There is little broadening in (002) peak where it indicates defect that might be present, it
could be that the miscut from the original orientation causes a broadening. FWHM is in the order of less
than 0.05, which indicates very low FWHM, which means it was single crystalline substrate. The out of
place lattice constant showed in STO (001) peak showed a number of 3.905 Å. To verify this number,
Bragg's equation is used where 2θ = 22.7516, and θ = 11.3758, and n=1, rewriting equations:
The dhkl that is observed is ≈3.906Å.
48
Figure 4.8 shows the scan of STO substrate, which shows the (100) peak at ≈22.7516o, (200) peak at ≈46.5096o, and (300)
peak at ≈72.5891o which can be validated by the literature[1] taken using logarithmic scale.
By thermodynamic stability, certain amount of free energy has to be achieved in order for layer to be
deposited onto another. In principle, layer by layer energy minimization according to the Gibbs Free
Energy, , induce a deviation from the real crystal structure. The implication is that crystal
containing defects have a lower free energy, and therefore, easier to deposit energetically. This
assignment deals with hetero-epitaxial grown films. It means that the substrate and epitaxial layer are
different materials with different lattice parameter as well as crystal symmetry. To calculate the
difference between the lattice constant, lattice mismatch is defined as follow [37] :
In this assignment, the substrate that is deposited is Silicon, with 80 Monolayers of STO crystal, and on
top of STO, LAO is deposited as many as 10 unit cells. Lattice constant of Silicon is 5.431 Å, STO has 3.905
Å, and LAO has 3.79 Å, which means that Si to STO has 30% lattice misfit, and STO to LAO has 3% lattice
misfit.
For a small lattice mismatch like the case in LAO deposited on top of STO, the epitaxial film can grow
pseudo morphical, or in other words, LAO tries to adapt the in plane lattice constant of STO, hence,
accumulating biaxial strain [37]. For this case of Si that is deposited with STO, the deposition method is
not directly form deposition of STO single crystal, instead, layer-by-layer of elements constituting STO,
49
or by co-evaporation of Sr molecular and Ti molecular inside effusion cells. With oxygen background
pressure as has been explained before in chapter 3.2, quantitatively, the peak shift of STO due to Silicon
will be very big if the process was direct deposition, but as can be seen from Figure 4.9a and 4.9b, the
lattice constant at (001) peak of STO shifted to the left to a value of ≈ 3.88 Å. However, this is also
caused by less experience of the author, since when taking the alignment of the (004) peak Silicon
substrate, it was not exactly at 69.171o, instead it was exactly at 69o. There could be slight shift peak due
to the wrong alignment. In addition, the peaks that are observed at ≈28.2537o is not due to the STO, or
some pollutant, it was because the background was not set to quartz zero background. In other words,
the extra peaks are due to the noise from the background, since when the quartz zero background is put
again under the substrate, there were no extra peaks seen in Figure 4.9b. There should also be
contribution from thermal coefficient of expansion, but the extension of this assignment is not yet
reach that point.
Figure 4.9 a and b corresponds respectively to cleaned Si-STO as-received substrate with non-quartz zero background plate and cleaned-and-annealed Si-STO substrate with quartz zero background.
As can be seen from Figure 4.9a and 4.9b, there are no big differences of STO peaks and Si peaks aside
from the peaks due to the non-quartz zero background plate. These figures are made before the
deposition of 10 unit cells of LAO. The quartz zero background reduced the noise to a very low level
near zero. It is known that the intended wavelength that is used is CuKα1, however there were also
other wavelength that is possible went through the diffraction that is not reduced by the non-zero
quartz background, such as CuKα2 (1.544426 Å), or even Kα(1.541874 Å) and Kβ (1.392250 Å) as well
(the wavelength is specified in the device). The second peaks that is observed in Figure 4.9a maybe
related to the STO (001) peak as well, at ≈28.2922o and STO(002) at ≈56.0366o, it could be that the
second peaks correspond to the incompletely suppressed Kβ component in the characteristic spectrum
of the Cu anode. This can proven by the example in Si (004) peaks that is observed experimentally
before, that has three different peaks observed for (004) reflection [55]. Si (003) or Si (002) is not
observed in this case due to zero value of structure factor (although the extend of this assignment has
not reach yet that point, it is worth mentioning why only Si (004) is observed there). In powder scanning,
there is not enough resolution to see the destructive interference caused by the other single crystalline
phase of other elements. It could be that if Si has random orientation, and the other peaks could show
as well.
50
After the deposition of 10 unit cells of LAO, our observation found there is a small side peaks are around
(001) and (002) STO peaks. At a glance, it might not be so clear enough to observe it, but when the
diffractograms are combined together, small side peaks can be seen.
As can be seen in the combined diffractogram, there is a small peaks that broaden the original STO (001)
peaks. This can be that with many reasons, one of the examples is shifting of the peaks due to different
treatments. However, previously, different treatments from as-received substrate and the one with
annealing treatment is basically the same thing. The observation is only differs from the peaks which is
observed due to non-quartz zero background.
Figure 4.10 Diffractogram of Si//STO/LAO 10uc, the side peaks seems to be increased and broadened.
One could also argue that the side peaks are due to the deposition of 10 unit cells of LAO. Let us further
discuss the observation here. The position of the main reflection determines the lattice constant, such
as the lattice constant of STO (001) after the deposition of 10 unit cells is ≈ 3.86 Å, which differs by
around ≈ 0.02 Å. This means that STO surface undergo another compressive strain due to lattice
mismatch with LAO. Previously, Figure 4.9b does not have a broadening of the peaks and increase
numbers of side peaks, as also can be seen from the combined diffractogram. After the deposition,
there are small side peaks observed in the diffractogram in both Figure 4.10 and 4.11. The side peaks
can also be called as Laue oscillations, but one could hardly argue this because the resolution of the
diffractogram is barely shows a so-called Laue-fringes.
51
By definition, Laue oscillations are always observed on both sides of the main peak, that can be used to
quantify numbers of contributing layers. As it is known, Laue oscillations emerge due to crystalline
surface, and Laue fringe can still be observed weakly from the deposition of LAO [Unfortunately for the
resources, I could not find the title of the book nor the author, it was simply a part chapter of a book
that does not specify the authors, the link as follows: [56]]. In order to investigate further whether LAO
peak is really there or not, a detailed scan is done under the scan range from 21o to 28o, under 0.002
step size.
4.11 Combined diffractogram of Si-STO substrate before and after the deposition of 10 unit cells of LAO. The picture is not so clear, but blue diffractogram is before deposition, treated with annealing procedure, and red one is after the deposition, with etching and annealing procedure.
Figure 4.12 LAO peak observed around STO (001) peak.
Using high score software, the background is reduced further, and the wavelength of Kα2 was stripped
off the diffractogram. The previously, LAO peak is still present in one of the peak which could be the
52
indicator that LAO is indeed there. The peak parameters show the out-of-plane lattice constant of
around 3.799 Å corresponds to 2θ=23.2954. Let's take a look at Figure 4.13.
Figure 4.13 Diffractogram of Si//STO/LAO 10 uc after reducing the background noise much further and stripping of Kα2 wavelength.
Let's try quantitative approach where :
2θhkl=23.2954, λ=1.540598Å, which then the dhkl ≈ 3.816 Å, which corresponds to c ≈ 3.816 Å. The out-of-
plane lattice constant of LAO could be that value, which corresponds to a strain when LAO is deposited
to Si//STO. Unfortunately, this is done under Powder Diffraction, it is better for the next scan to do it
under X'PERT MRD for a better scan resolution, and to determine the in-plane lattice constant of LAO.
4.4 Electronic Transport Properties
The hetero-structure of LAO/STO platform is successfully made and the process is reproducible, varying
different treatments as well as other parameters in both the treatments and deposition. This chapter
will mainly discuss about the electronic properties of the heterostructure (Si/STO/LAO), and how
different parameters is given, expected to alter its electrical transport properties.
The interface between STO and LAO is well known recently due to the conducting behavior, although its
bulk properties is insulating. By exceeding thickness of 4 unit cell, thickness that is required for charge
reconstruction at the interface, the properties become conducting.
LAO(001)
53
The sample is not of patterned geometry, and the conventional resistivity measurement won't work.
Therefore, the van der pauw method is used. The contact is successfully made with conventional wire
bonding using thermosonic. The contact is set one in each of the corner boundary. Van der Pauw
method allows for obtaining material properties such as, resistivity, the major dopants, its sheet carrier
density and mobility. Another measurement done is Hall measurements, where the voltage contact is
now perpendicular with the current contact. As such, in our case, it was set diagonally, where one
voltage contact reach the other diagonal end, and so is for the current contact.
There were 4 samples originally intended to be measured their transport properties. However, there
was only one sample that has enough data to provide the measurement of transport properties. The
first sample, as-received Si//STO substrate was intended to replace the initial first sample that is
dropped off during deposition, however, due to some reason, it could be that the first sample lost its
contact, at some point during measurement, the signal is lost. It could also be that the measurement
cannot read due to resistance limit, since the first sample is expected to be insulating due to not having
2DEGs that is generated from the interface (lack of LAO).
The second sample, which is treated with etched and annealed treatment, is indeed insulating as has
been confirmed in the AFM image, due to the impurities that are presents and etch pit that could
hamper the mobility of the charge densities. In appendix it can be seen that indeed the sample treated
with etched and annealed lost contact at some point. The third sample has some interesting properties
that could be promising, although lots of noises were present in the plot. This chapter will mainly discuss
on the annealed sample which provides enough data to be analyzed.
54
Figure 4.14 Typical Van der Pauw measurements configurations, specifically this model is designed for the annealed sample. The longitudinal measurements that were used, are vertically or horizontally going from one edge to the other, while hall measurement that was used, going diagonally.
The focus on the electrical transport properties only on the fourth sample, treated with annealing
procedure, which has enough data to be analyzed. The configuration for sample 4 can be seen in Figure
4.14. The hall measurement is carried out under the presence of a magnetic field goes perpendicular
through the sample. So in RH1 for example, if current is flowed from 12 to 11, the charge is flowed in
opposite direction from 11 to 12. Since the magnetic field is flowed perpendicular in the sample, by right
hand rule, the charges will move to 14, and there will be accumulation of negative charges in 14, and the
depletion charge in 13, causing a negative potential difference since V+ to V- is going from 14 to 13,
which results in negative resistance, and the other way around for RH2 (which is positive potential and
positive resistance), this in agreement later with the data that is obtained.
4.4.1 Sample 4 - Deposited 10 uc LAO - Annealing Treatment
The annealed sample, which was placed in the same resistivity puck with as-received substrate, shows a
metallic behavior, which can be clearly seen in Figure 4.15. The first measurement was set to see the
resistivity with respect to temperature change, and the offset magnetic field is -0.28 Oe. In Figure 4.15,
the anisotropy can be observed indeed the resistance value in Rx is different with Ry.
It is also observed that T-dependent resistivity as well as the sheet resistance shows an odd behavior,
showing a metallic behavior, and then having an increase in the resistance at low T, the insulator to
metal transition can be observed in this case. In order to investigate this behavior much further, the
55
magnetic field is varied starting from a temperature of 100K down to 2K since the resistivity went up
after 100K. By varying the magnetic field from -1 to 1 Tesla, it can be seen that the asymmetrical Hall
Resistance Rxy displays linearity in Hall effect, the resistance goes down as magnetic field is varied from
positive to negative value.
Sample 4 is the only sample that gives the best results of 4 samples that are measured. Even though
noises are present in a lot of graphs in Hall measurement, the expected behavior for Hall resistivity
versus Magnetic Field is present. The anti-symmetry Hall Resistance Rxy displays linearity in Hall Effect at
temperature is varied from 2K to 100K.
According to Hall measurements, at temperatures of 2K and 150K, the carriers are dominated by
negative charge, it means that the interface is indeed n-type. In order to calculate the sheet densities,
one has to understand the hall effects measurement. The hall effects explain the phenomenon behind a
conducting material such as semi-conductor or metal, in which if the current flowed through the
material, is exposed with a varying magnetic field. The geometry that is explained before using 4 probe
measurement apply van der pauw method, where the hall resistance measured across the sample
diagonally has been explained before in experimental procedure.
According to Hall measurements, from varied temperatures of 150K down to 2K, the carriers are
negatively charged with sheet densities at 2K is ≈ 7.34 x 1013 cm-2 and at 150K is ≈ 1.53 x 1014 cm-2. It is
not possible to get the sheet density value at 300K since the hall measurement for sample 4 is not done
from 300 K. The corresponding mobility µ are ≈ 13.3 cm2/Vs and 9.8 cm2/Vs at 2K and 150K respectively.
These values somewhat has quite small gap, since it is expected that the mobility of charge carriers at 2K
has to be some number of magnitude higher than 150K. Nevertheless, since the sheet density at 300K is
not calculated, hence there is no mobility at 300K that can give a complete picture of how fast the
mobility of charge carriers in this system of heterostructure (Table 4.1 is provided for a complete values
varied from 150K down to 2K), the plot can be seen in figure 4.16. Furthermore, the activation energy
seen in figure 4.17 in which the charge carrier seems frozen out is 13 meV, which is pretty big, twice the
results from the system of only STO//LAO[57].
Figure 4.15a shows temperature dependent resistivity of annealed sample from 300K down to 2K and
figure 4.15b shows sheet resistance against temperature from 250K. Resistance of the sample goes
down as temperature decrease, and at some point, around 100K, it increases again. This model is in
agreement with Brinkman et al. which had a sheet resistance of around ≈ 5000 Ω/□ that starts from
100K above, corresponding to oxygen pressure of 3 x 10-5 mbar, while this system has 5 x 10-5 mbar. The
upturn slope was not observed in linear scale, but when the scale is changed into log scale, it is clearly
seen that there is an upturn slope at <100K. In addition, the deviation on the value might be due to this
system that is not perfect crystal system like STO//LAO, instead, STO is a film grown on Si substrate,
defect plays an important role in this case (mostly not by oxygen vacancy since the system is grown on
low pressure regime where oxygen vacancy introduced is not significant) as has been explained in AFM
image.
56
As a comparison, it is possible to say that the sample has metallic properties, in which the resistivity
decrease monotonically by decreasing temperature which can be explained by electron-phonon
scattering [36]. Unfortunately, this theory could not really described what is happening when the
temperature reach 100K, where the resistance is increasing to almost the same value in 250K. This is
especially happened in low region liquid helium temperature of 2K.
This can be intuitively understood by the principal of electron-lattice scattering that determine the
electrical conductivity, hence the electrical resistance (keep in mind that this model is based on metallic
materials). For a relatively high T, let's say a room temperature, the dominant mechanism is due to
scattering of electron by phonons. One way to understand this temperature-dependent effect of
resistance is by relating the magnitude of phonon scattering with mean square displacement of the
molecules in lattice <x2>[58]. Since it is not possible to make an assumption whether this is a simple
harmonic solid or not, the quantity is not exactly kbT, there should be another model that can explain
the temperature dependence of sheet resistance.
Brinkman, et al. explained further of the model (not explained further in this report since the scope of
the knowledge is not reached yet, but it is worth noting that this model serves the best for this system
too) of this temperature dependence of sheet resistance where:
where T2 and T5 corresponds to suggestive of electron-electron and electron phonon scattering
respectively, relevant at higher T. Teff ≈ 70 K[59], whereas in this system, it is approximately Teff ≈ 100 K.
Teff is the effective temperature in which a crossover is happening, when the transition is happening
from metal to insulator as can be seen in figure 4.15a and 4.15b.
For low temperatures, phonon scattering effect decreases and T allows for scattering dominated by
lattice imperfection, or defects in the interface. In low-T domain, where the transition occurs from 100 K
down to 2K, the resistance is significantly increased. The value of such resistance is mainly determined
by the amount of impurities and imperfections in the interface of the system. As has been mentioned
before, the sheet density value is at ≈ 7.34 x 1013 cm-2, whereas at 150K, the sheet density value is ≈1.53
x 1014 cm-2. This means that the mobile electrons or charge carriers is really hampered by the defects
present at the interface, thus the decrease in the sheet density. In metallic elements, few part of million
impurities can affect greatly the transport properties, and so is the amount of work generated by the
low Temperature regime [58].
At this point of transition, the explanation from normal scattering between particles as explained in the
model of metallic materials cannot be used. To further investigate what was happening down from
temperature of 100K to 2K, the magnetic field dependence of resistance is measured. The original
measurement was intended to vary from 300K down to 2K, but the actual measurement was varied
from 150K down to 2K, and since the transition occurs at ≈ 100K or less, the data that is analyzed is
starting from 100K which is the most interesting. The hall measurement is done here, with the
57
configuration shown in Figure 4.14 for RH1 and RH2. Unfortunately, RH2 measurement does not give
very useful information compare to RH1, so the main discussion is in RH1 measurement.
The magnetic field is varied from 10T back and forth to -10T (the actual measurement was in Oe, and
conversion is multiplying by 10-4). As mentioned before in Brinkman. et al [59] and Huijben. et al[6], the
individual property does not have magnetic properties, as magnetic field is varied from 100K to 2K,
linearity of the magneto-resistance is observed in Figure 4.16 where the resistance in proportionally
increasing by increasing magnetic field. This is interesting since this can explain what happened at the
sudden up-turn in the temperature dependent resistance. One intuitive way to understand this is by
make normalization of the plot, where the magneto-resistance is now defined as the change in
resistance relative to zero-field resistance[59]:
A large number of magneto-resistance’s effect at 9T were observed in the annealed sample at 26%
when varying the negative field, and 44% when varying the positive field at 2K. Magnetoresistance
hysteresis is observed clearly in 2K region, but the other temperature range were not clear enough to be
interpreted. The plot in figure 4.19 showed only from 0 to 9 T, because from 0 to -9 T, a lot of noises
were observed. The hysteresis that was observed could be the indicator of ferromagnetic domain
formation, where in principle, symmetry can be observed when crossing zero field for 0 to 9T and 0 to -
9T.
The magneto-resistance can be intuitively understood by the spin of the electrons. The scattering
phenomenon that happens down at this temperature when varying magnetic field is due to the
scattering of spins of electrons. Electrons can be assumed likes a tiny magnets, when a current is flowed
through a sample, most of the orientation of the spin electron is randomly oriented, but in average, the
spin of electrons is directed by the direction of the current. If an electron is scattered, the direction will
be changed by the affecting magnetic field, thus increasing the magneto-resistance of a material.
This depends on the probability of the scattering effect, in which it increases by the availability of the
quantum states that are present for the electron to scatter into. More states means more scattering,
hence, an increase in the magneto-resistance. In other words, the spins that in phase or parallel to the
direction of the magnetic field will scatter less, and thus small increase in the magneto-resistance, and
so is the other way around. Quantitatively, this can be understood by Born-Ockhur approximation[60].
Another phenomenon that could be the reason behind this is due to Kondo effect, suggested by
Brinkman. et al[59]. The Kondo’s effect is a different kind of scattering, in which the mechanism of
conduction electron in a metallic material is due to magnetic impurities, which then contributes to the
electrical resistivity, and decreasing as a function of log (T)[61]. Further investigation is needed to see
the behavior of the magneto-resistance in even lower temperature than 2K. The odd behavior where in
the gap of resistance value in between 2K and 5K also may be understood by the Kondo’s effect.
58
Much of the theoretical background and experimental part is still needed to understand this behavior.
But this suggestion explains a bit more in intuitive way the behavior of the upturn slope observed in
temperature <100K.
Figure 4.15 Temperature Dependent Resistance of Annealed sample from 300 K (up-left) and Sheet Resistance from 250K (up-right), serves as a comparison is system of STO//LAO presented by Brinkman et al (bottom-left and right).
59
Figure 4.16 Plot o the sheet carrier density as a function of temperature for a sample treated with pre-annealing on the Si//STO substrate. According to the sign of the Hall voltage the charge carriers are electrons.
Figure 4.17 Plot of Temperature dependence of ln(ns-n0) for Si//STO/LAO 10 unit cells where ns is defined as = -1/RH*e and n0 is the temperature limit of ns
60
Figure 4.18 Hall Resistance vs T annealed sample. Linearity is observed throughout the whole temperature range
Figure 4.19 Magnetoresistance hysteresis of Annealed sample
61
Chapter 5 Conclusions and Suggestions
The generation of two dimensional electron gases (2DEGs) on Silicon in (001) orientation has been
successfully made in this assignment. The heterostructure used in the system was two perovskite oxides,
namely SrTiO3 and LaAlO3. This heterostructure has been well made before in an atomic layer-by-layer
stacking sequence. The substrate however, was Si//STO, in which the STO comprised of 80 Monolayers
was deposited using Molecular Beam Epitaxy (not done in University of Twente). Different combination
of pre-treatments on the substrate was also done, in order to achieve an atomically smooth surface on
the substrate, thereby enhancing the transport properties in the system later. The first treatment was
the annealing only, second treatment was etching and then annealing, the third treatment was
annealing and then etching, and one last substrate was left alone as a base comparison. Atomic Force
Microscopy images shown that annealed sample has the lowest RMS value and peak to peak value,
thereby concluding the smoothest sample. X-Ray Diffraction shown small side peaks observed, which
could be the indicator of successful 10uc deposition of LAO, this could further be improved by using the
MRD machine instead the Powder one. Further investigations need to be done. The deposition process
for the STO to Si substrate needs to get more controlled in order to have an atomically smooth STO film
grown. The wet chemical etching process is very hard to control, it could also use dry etching for more
controllable process.
However, as-received substrate was lost during deposition, and only annealing treatment substrate gave
enough data to be analyzed. Indeed 2DEGs was generated in this system, the sweep measurement in
Quantum Design - Physical Property Measurement System from 300K down to 2K showed an insulating-
metallic transition, however, starting from 100K, the resistance increased again. Furthermore, to
investigate what was happening, hall measurement is done from 150K down to 2K. According to hall
measurement, all charge carriers is negatively charged with sheet densities at 2K is ≈ 7.34 x 1013 cm-2
and at 150K is ≈ 1.53 x 1014 cm-2, and its corresponding mobilities are ≈ 13.3 cm2/Vs and 9.8 cm2/Vs at 2K
and 150K respectively. Linearity is observed in the hall measurements. Furthermore, magnetoresistance
is also observed, where the magnetoresistance hysteresis is observed clearly in 2K, this could be the
indicator of ferromagnetic domain formation.
62
References
[1] S. P. Thiel, "Study of the Interface Properties in LaAlO3/SrTiO3 Heterostructures," Augsburg, 2009,
pp. 5-6.
[2] R. Hueting, A. Mouthaan and J. Schmitz, Semiconductor Devices Explained More, Enschede:
University of Twente, 2012.
[3] M. Sparks and T. G.K., "Method of Making P-N Junctions in Semiconductor Materials". United States
of America Patent U.S. Patent 2,631,356, 17 March 1953.
[4] E. Schubert, in Doping in III-V Semiconductors, Cambridge, Cambridge University Press, 2005, pp.
241-243.
[5] E. Jin, L. Kornblum, K. Zou, K. Broadbridge, J. Ngai, C. Ahn and F. Walker, "A high density two-
dimensional electron gas in an oxide heterostructure on Si (001)," APL MATERIALS 2, vol. 2, pp.
116109 1-6, 2014.
[6] M. Huijben, A. Brinkman, G. Koster, G. Rijnders, H. Hilgenkamp and D. Blank, "Structure-Property
Relation of SrTiO3/LaAlO3 Interfaces," Advanced Materials, no. 21, pp. 1665-1677, 2009.
[7] M. Norton and C. Carter, "Chapter 11 - Point Defects, Charge, and Diffusions," in Ceramic Materials
- Science and Engineering, New York, Springer Science+Business Media, 2013, pp. 187-206.
[8] A. Gozar, G. Logvenov, L. Kourkoutis, A. Bollinger, L. Giannuzzi, D. Muller and I. Bozovic, "High-
temperature interface superconductivity between metallic and insulating copper oxides," NATURE,
vol. 455, pp. 782-785, 2008.
[9] N. Ganguli and P. Kelly, "Tuning Ferromagnetism at Interfaces between Insulating Perovskite
Oxides," Physical Review Letters, vol. 113, no. 12, pp. 127201 1-5, 2014.
[10] J. Haeni and E. AL, "Room-Temperature ferroelectricity in strained SrTiO3," NATURE, vol. 430, pp.
758-761, 2004.
[11] F. Dovland, "Investigation of Pulsed Laser Deposition Growth Parameters and their influence on the
sheet resistance of a Complex Oxide Heterointerface," Norwegian University of Science and
Technology, Trondheim, 2011.
[12] A. Ohtomo and H. Hwang, "A high-mobility electron gas at the LaAlO3/SrTiO3 heterointerface,"
NATURE, vol. 427, pp. 423-426, 2004.
63
[13] Wikipedia, "Lev Perovski," Wikimedia Foundation, 18 May 2015. [Online]. Available:
https://en.wikipedia.org/wiki/Lev_Perovski. [Accessed 15 June 2015].
[14] Wikipedia, "Perovskite," Wikimedia Foundation, 1 June 2015. [Online]. Available:
https://en.wikipedia.org/wiki/Perovskite. [Accessed 15 June 2015].
[15] M. Levy, "Crystal Structure and Defect Properties in Ceramic Materials," London, Imperial College
London, 2005, pp. 80-87.
[16] C. Carter and M. Norton, "Complex Crystals and Glass Structure," in Ceramic Materials Science and
Engineering, New York, Springer Science+Business Media New York, 2013, p. 106.
[17] C. Ziyong and J. Lin, "Layered organic–inorganic hybrid perovskites: structure, optical properties,
film preparation, patterning, and templating engineering.," CrystEngComm, no. 12, pp. 2646-2662,
2010.
[18] K. e. a. Chang, "High-Performance SrTiO3 MIM Capacitors for Analog Applications," IEEE
Transactions On Electron Devices, vol. 53, no. 9, pp. 2312-2319, 2006.
[19] C. P. e. al, "ritical-Current Measurements in Epitaxial Films of YBa2Cu3O7-x Compound," Phys. Rev.
Lett., no. 58, p. 2684, 1987.
[20] MediaWiki.org, "http://en.wikipedia.org/," MediaWiki.org, 7 April 2015. [Online]. Available:
http://en.wikipedia.org/wiki/Strontium_titanate. [Accessed 25 May 2015].
[21] MediaWiki.org, "http://en.wikipedia.org/," MediaWiki.org, 25 April 2015. [Online]. Available:
http://en.wikipedia.org/wiki/Direct_and_indirect_band_gaps. [Accessed 25 May 2015].
[22] M. A. Hein, Supercond. Sci. Technol, no. 10, p. 867, 1997.
[23] H. Boschker, Perovskite oxide heteroepitaxy : strain and interface engineering, Enschede:
Wohrmann Print Service, 2011.
[24] N. Nakagawa, H. Hwang and D. Muller, "Why Some Interfaces Cannot be Sharp," Nature Materials,
vol. 5, pp. 204-209, 2006.
[25] J. E. Kleibeuker, Reconstructions at complex oxide interfaces, Enschede: Wohrmann Print Service,
Zutphen, The Netherlands, 2012.
[26] H. Chen, H. Park, A. J. Millis and C. A. Marianetti, "Charge transfer across transition-metal oxide
interfaces : Emergent conductance and electronic structures," Physical Review B, no. 90, pp.
245138-1 - 245138-11, 2014.
64
[27] Wikipedia, "Wikipedia, the free encyclopedia," Wikimedia Foundation, 13 March 2015. [Online].
Available: http://en.wikipedia.org/wiki/Surface_second_harmonic_generation. [Accessed 7 May
2015].
[28] A. Savoia, D. Paparo, P. Perna, Z. Ristic, M. Saluzzo, F. M. Granozio and e. al, "Polar catastrophe and
electronic reconstructions at the LaAlO3/SrTiO3 interface:evidence from optical second harmonic
generation," University of Augsburg, Augsburg, 2009.
[29] R. Pentcheva, M. Huijben, K. Otte, W. Pickett, J. Kleibeuker, J. Huijben, H. Boschker and e. al,
"Parallel Electron-Hole Bilayer Conductivity from Electronic Interface Reconstruction," Physical
Letter Review, vol. 104, pp. 166804 1-4, 2010.
[30] W. Siemons, G. Koster, H. Yamamoto, W. Harrison, G. Lucovsky, T. Geballe, D. Blank and M. Beasley,
"Origin of Charge Density at LaAlO3 on SrTiO3 Heterointerfaces: Possibility of Intrinsic Doping,"
Physical Review Letters, vol. 98, no. 19, pp. 196802 1-4, 2007.
[31] S.-H. Baek and C.-B. Eom, "Epitaxial Integration of Perovskite-based Multifunctional Oxides on
Silicon," Acta Mater, vol. 9, no. 73, pp. 1-17, 2012.
[32] S. Y, R. Rholdan, M. I. Katsnelson and F. Guinea, "Effect of point defects on the optical and transport
properties of MoS2 and WS2," Physical Review, vol. 90, pp. 041402 1-5, 2014.
[33] L. Li, C. Richter, J. Mannhart and R. Ashoori, "Coexistence of magnetic order and two-dimensional
superconductivity at LaAlO3/SrTiO3 interfaces," Nature Physics, vol. 7, no. 10, pp. 762 1-4, 2011.
[34] R. McKee, F. Walker and M. Chisholm, "Crystalline Oxides on Silicon : The First Five Monolayers,"
Physical Review Letters, vol. 81, no. 14, pp. 3014-3017, 1998.
[35] G. Binnig and C. Quate, "Atomic Force Microscope," Physical Review Letter, vol. 56, no. 9, pp. 930-
934, 1986.
[36] J. Hook and H. Hall, Solid State Physics, Manchester: John Wiley & Sons, 1991.
[37] Walter Schottky Institute; Technische Universitat Munchen, "Fortgeschrittenpraktikum - High
Resolution X-Ray Diffraction," Technische Universitat Munchen, Garching, 2009.
[38] A. Kalabukhov, R. Gunnarsson, J. Borjesson, E. Olsson, D. Winkler and T. Claeson, "Effect of Oxygen
Vacancies in the SrTiO3 substrate on the electrical properties of the LaAlO3/SrTiO3 interface,"
Physical Review B, vol. 75, p. 121404, 2007.
[39] M. Huijben, "Interface Engineering for Oxide Electronics:Tuning electronic properties by atomically
controlled growth," University of Twente, the Netherlands, Enschede, 2006.
65
[40] Wikipedia, "Van der Pauw Method," Wikimedia Foundation, 29 April 2015. [Online]. Available:
https://en.wikipedia.org/wiki/Van_der_Pauw_method. [Accessed 6 June 2015].
[41] Quantum Design, PPMS Hardware and Options Manuals, Quantum Design, 2007.
[42] F. Dovland, "Contacting the conductive LaAlO 3 / SrTiO 3 interface," NTNU, Trondheim, 2010.
[43] G. Rijnders, G. Koster, D. Blank and H. Rogalla, "In situ monitoring during pulsed laser deposition of
complex oxides using reflection high energy electron diffraction under high oxygen pressure,"
Applied Physics Letter, vol. 70, no. 14, pp. 1888-1890, 1997.
[44] M. Lagally and D. Savage, MRS Bull, vol. 18, no. 1, p. 24, 1993.
[45] Wikipedia, "Surface roughness," Wikimedia foundation, 17 May 2015. [Online]. Available:
https://en.wikipedia.org/wiki/Surface_roughness. [Accessed 18 June 2015].
[46] W. Xu, F. Yiyan, L. Huibin, J. Kui-juan, Z. Xiangdong, C. Zhenghao, Z. Yueliang and Y. Guozhen,
"Atomic Force Microscopy studies of SrTiO3 (001) substratres treated by chemical etching and
annealing in oxygen.," Science in China, vol. 48, no. 4, pp. 1-10, 2005.
[47] A. Gangulee, "Strain-Relaxation in thin films on substrates," Acta Metallurgica, vol. 22, pp. 177-183,
1974.
[48] G. Alonzo-Medina, A. Gonzalez-Gonzalez, J. Sacedon and A. Oliva, "Understanding the thermal
annealing process on metallic thin films," IOP Conference Series : Materials Science and Engineering,
vol. 45, pp. 1-6, 2013.
[49] P. Adegbuyi and A. Atiri, "The Effect of Annealing on the Microstructure of Mechanical Properties of
a Rolled steel product," The Pacific Journal of Science and Technology, vol. 10, no. 2, pp. 149-162,
2009.
[50] J. Suwardy and Y. Darma, "Thermal Annealing Effects of SrTiO3 Film on Si(100)," in Padjajaran
International Physics Simposium, Bandung, 2015.
[51] G. Koster, B. Kropman, G. Rijnders and e. al., "Quasi Ideal Strontium Titanate surface through
formation of strontium hydroxide," Applied Physics Letter , vol. 73, pp. 2920-2922, 1998.
[52] Y. Li, C. Weatherly and M. Niewczas, "TEM Studies of stress relaxation in GaAsN and GaP thin films,"
Philosophical Magazines, vol. 85, no. 3073, 2005.
[53] M. Bhuiyan, A. Matsuda, T. Yasumura, T. Tambo and C. Tatsuyama, "Study of epitaxial SrTiO3 (STO)
thin films grown on Si (001)-2x1 substrates by Molecular Beam Epitaxy," Applied Surface Science,
vol. 216, pp. 590-595, 2003.
66
[54] H. Yamada, M. Kawasaki and Y. Tokura, "Epitaxial growth and valence control of strained perovskite
SrFeO3 films," Applied Physics Letter, vol. 80, p. 622, 2002.
[55] V. Pecharsky and P. Zavalij, Fundamentals of Powder Diffraction and Structural Characterization of
Materials, New York: Springer Science and Business Media, 2009.
[56] Frei Universitate Berlin, "www.diss.fu-berlin.de," [Online]. Available: http://www.diss.fu-
berlin.de/diss/servlets/MCRFileNodeServlet/FUDISS_derivate_000000002709/06_ch3.pdf?hosts=.
[Accessed 21 June 2015].
[57] M. Huijben, G. Rijnders, D. Blank, S. Bals, S. Van Aert, J. Verbeeck, G. Van Tendeloo, A. Brinkman
and H. Hilgenkamp, "Electronically coupled complementary interfaces between perovskite band
insulators," Nature Materials, vol. 5, pp. 556-560, 2006.
[58] S. Van Sciver, "Chapter 2 - Low-Temperature Materials Properties," in Helium Cyrogenics, New York,
Springer Science+Business Media, LLC, 2012, pp. 17-58.
[59] A. Brinkman, M. Huijben, M. Van Zalk, J. Huijben, U. Zeitler, J. Maan, W. Van der Wiel, G. Rijnders,
D. Blank and H. Hilgenkamp, "Magnetic effects at the interface between non-magnetic oxides.,"
Nature Materials, vol. 6, pp. 493-496, 2007.
[60] J. Matthew, "Spin dependence of the electron inelastic mean free path and the elastic scattering
cross section - a high energy atomic approximation," Surface Science Divison, National Bureau of
Standards, Washington, D.C., 1981.
[61] J. Kondo, "Resistance minimum in dilute magnetic alloys," Progress of Theoretical and Experimental
Physics, vol. 32, pp. 37-49, 1964.
[62] P. Willmott, S. Pauli, R. Herger, S. C.M., D. Martoccia, B. Patterson, B. Delley, R. Clarke, D. Kumah, C.
Cionca and Y. Yacoby, "Structural Basis for the Conducting Interface between LaAlO3 and SrTiO3,"
Physical Review Letters, vol. 99, no. 15, pp. 155502 1-4, 2007.
[63] G. Niu, G. Saint-Girons, B. Vilquin, G. Delhaye, J.-L. Maurice, C. Botella, Y. Robach and G. Hollinger,
"Molecular beam epitaxy of SrTiO3 on Si (001): Early stages of the growth and strain relaxation,"
Applied Physics Letters, vol. 95, pp. 06292 1-3, 2009.
[64] A. (. O. I. Company), "Pulsed Laser Depostion - An Introduction to Pulsed Laser Deposition," 1989.
[Online]. Available: http://www.andor.com/learning-academy/pulsed-laser-deposition-an-
introduction-to-pulsed-laser-deposition. [Accessed 1 June 2015].
67
Appendices
1. As-received Si//STO substrate
2. Pre-treatment: Etching and Annealing Si//STO/LAO
68
3. Pre-treatment: Annealing and Etching Si//STO/LAO
69
4. Table 4.1: Sheet Resistances, Hall Coefficient, Sheet Densities, and Mobilities varied around
temperatures from 150K down to 2K