Novel Tetragonal-Like Phases of BiFeO3 Films Grown by Pulsed Laser Deposition, and their Characterisation
Thomas C. Young
A thesis in fulfilment of the requirements for the degree of
Master of Science
The University of New South Wales
Faculty of Science
School of Materials Science and Engineering
August 2019
Surname/Family Name : Young Given Name/s : Thomas Abbreviation for degree as give in the : Master of Science (MSc) University calendar Faculty : Science School : Materials Science and Engineering Novel Tetragonal-Like Phases of BiFeO Films Grown by Thesis Title : 3 Pulsed Laser Deposition, and their Characterisation
Abstract 350 words maximum: (PLEASE TYPE) This thesis presents an experimental study of multiferroic BiFeO3 thin films grown by pulsed laser deposition. Multiferroics – materials which possess two or more ferroic orders (most commonly ferroelectricity and magnetic order) – offer opportunities for novel devices in data storage and spintronics. Bismuth ferrite (BiFeO3 – BFO) is the only known room-temperature multiferroic, making it attractive for practical applications. This study considers the range of growth parameters available during pulsed laser deposition and their role in tuning the structure of so called ‘T’ phase’ BiFeO3 thin films, grown on (001) LaAlO3 substrates. To gain insight into the influence of growth parameters on the physical properties of the films, structural, electrical and magnetic characterisation was performed. X-ray diffraction techniques including 2θ-ω coupled scans, X-ray reflectivity (XRR) and reciprocal space mapping (RSM), along with transmission electron microscopy (TEM), were used for structural and chemical characterisation. Advanced modes of scanning probe microscopy (SPM) were used to probe surface quality, ferroelectric properties, and magnetic response of the films. It was found that by using growth conditions to modify film stoichiometry and structure in localised nano- regions, it was possible to stabilise the highly elongated out of plane T’ phase with high crystallinity and coherence to thicknesses up to 73 nm, far beyond the previously-reported ‘critical thickness’ of ~25 nm. It was also found that 2% cobalt doping on thick T’ phase BFO films had a negligible influence on the structural, electronic and magnetic properties; however the cobalt doping yielded films with functional response such as conductive domain walls. The results presented here provide new ways to probe the properties of ‘pure’ T’ phase BFO and demonstrate that by carefully controlling growth conditions, one can tailor structural parameters, phase fractions, and in turn the functional response of multiferroic thin films.
Declaration relating to disposition of project thesis/dissertation
I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.
I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).
…………………………………… ……………………………………..……………… ……….……………………...…….… Signature Witness Signature Date The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.
FOR OFFICE USE Date of completion of requirements
ONLY for Award:
i
ORIGINALITY STATEMENT
‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’
Signed ……………………………………………......
Date ……………………………………………......
ii
INCLUSION OF PUBLICATIONS STATEMENT
UNSW is supportive of candidates publishing their research results during their candidature as detailed in the UNSW Thesis Examination Procedure. Publications can be used in their thesis in lieu of a Chapter if: • The student contributed greater than 50% of the content in the publication and is the “primary author”, ie. the student was responsible primarily for the planning, execution and preparation of the work for publication • The student has approval to include the publication in their thesis in lieu of a Chapter from their supervisor and Postgraduate Coordinator. • The publication is not subject to any obligations or contractual agreements with a third party that would constrain its inclusion in the thesis
Please indicate whether this thesis contains published material or not. ☐ This thesis contains no publications, either published or submitted for publication
Some of the work described in this thesis has been published and it has ☒ been documented in the relevant Chapters with acknowledgement
This thesis has publications (either published or submitted for ☐ publication) incorporated into it in lieu of a chapter and the details are presented below
CANDIDATE’S DECLARATION I declare that: • I have complied with the Thesis Examination Procedure • where I have used a publication in lieu of a Chapter, the listed publication(s) below meet(s) the requirements to be included in the thesis. Name Thomas Young Signature Date (dd/mm/yy)
iii
COPYRIGHT STATEMENT ‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'
Signed ……………………………………………......
Date ……………………………………………......
AUTHENTICITY STATEMENT ‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’
Signed ……………………………………………......
Date ……………………………………………......
iv
Acknowledgements
A special thanks to my grandfather who fostered my interest in science from a young age and always encouraged me to follow my curiosity, you will be greatly missed.
To everyone who helped make this thesis possible. I would not have been able to make it this far without all the support you have given me.
v
Abstract
This thesis presents an experimental study of multiferroic BiFeO3 thin films grown by pulsed laser deposition. Multiferroics – materials which possess two or more ferroic orders (most commonly ferroelectricity and magnetic order) – offer opportunities for novel devices in data storage and spintronics. Bismuth ferrite (BiFeO3 – BFO) is the only known room-temperature multiferroic, making it attractive for practical applications. This study considers the range of growth parameters available during pulsed laser deposition and their role in tuning the structure of so called ‘T’ phase’ BiFeO3 thin films, grown on (001) LaAlO3 substrates. To gain insight into the influence of growth parameters on the physical properties of the films, structural, electrical and magnetic characterisation was performed. X-ray diffraction techniques including 2θ-ω coupled scans, X-ray reflectivity (XRR) and reciprocal space mapping (RSM), along with transmission electron microscopy (TEM), were used for structural and chemical characterisation. Advanced modes of scanning probe microscopy (SPM) were used to probe surface quality, ferroelectric properties, and magnetic response of the films. It was found that by using growth conditions to modify film stoichiometry and structure in localised nano-regions, it was possible to stabilise the highly elongated out of plane T’ phase with high crystallinity and coherence to thicknesses up to 73 nm, far beyond the previously-reported ‘critical thickness’ of ~25 nm. It was also found that 2% cobalt doping on thick T’ phase BFO films had a negligible influence on the structural, electronic and magnetic properties; however the cobalt doping yielded films with functional response such as conductive domain walls. The results presented here provide new ways to probe the properties of ‘pure’ T’ phase BFO and demonstrate that by carefully controlling growth conditions, one can tailor structural parameters, phase fractions, and in turn the functional response of multiferroic thin films.
vi
Contents
Acknowledgements ...... v
Abstract ...... vi
List of Figures ...... ix
Introduction ...... 1
1 Generalities ...... 2
1.1 Ferroelectricity ...... 2
1.2 Magnetism ...... 5
1.3 Multiferroics ...... 7
1.4 Bismuth Ferrite ...... 8
1.4.1 Bulk Properties ...... 9
1.4.2 Epitaxial thin films of BFO ...... 12
1.4.3 Tetragonal-like and mixed-phase BFO ...... 13
1.4.4 Applications of BFO ...... 15
2 Experimental Techniques ...... 17
2.1 Pulsed Laser Deposition ...... 17
2.1.1 Operating principle ...... 18
2.1.2 Experimental method ...... 20
2.1.3 Film Growth ...... 21
2.1.4 PLD growth conditions for BFO films ...... 21
2.2 Physical characterisation of epitaxial BiFeO3 thin films ...... 23
2.2.1 X-Ray diffraction ...... 23
2.2.2 X-ray Reflectometry ...... 28
2.2.3 Reciprocal Space Mapping ...... 31
2.2.4 Atomic Force Microscopy ...... 34
3 Using defects to stabilize novel phases in T’ like BFO ...... 38
3.1 Introduction and Motivation ...... 39
vii
3.2 Sample Growth ...... 40
3.3 Surface and structural characterisation ...... 41
3.4 Possible origins for stabilisation of T’ phase to ~80 nm ...... 47
3.4.1 Laser Fluence and Energy Density ...... 47
3.4.2 Deposition oxygen pressure...... 48
3.4.3 Substrate-target distance ...... 48
3.4.4 Temperature difference ...... 48
3.5 Inducing the mixed phase by intentional defects ...... 51
3.6 Discussion and Conclusion ...... 54
4 Cobalt Doping of BFO thin films ...... 55
4.1 Introduction and Motivation ...... 56
4.2 Sample Growth ...... 57
4.3 Structure and surface properties of Co-BFO films on LAO (001) ...... 58
4.4 Ferroelectric properties ...... 63
4.5 Functional response: electric-field interconversion between the phases and conductive domain walls ...... 66
4.6 Magnetic properties...... 67
4.7 Discussion, conclusions, and further work ...... 68
5 Conclusions and Perspectives ...... 70
5.1 Conclusions ...... 70
5.2 Perspectives ...... 71
6 References ...... 73
viii
List of Figures
Figure 1: Associated electronic properties of the crystal classes ...... 3
Figure 2: Structure of ABO3 perovskite: a) polarization up state, b) polarization down state, c)
Energy vs ionic displacement plots: below TC (blue) and above TC (red) showing the energy barrier (EB) and double well, d) typical ferroelectric hysteresis loop with coercive electric field
(Ec) and remnant polarisation (Pr) ...... 4 Figure 3: Hierarchy of magnetism [15] ...... 6 Figure 4: Magnetic orders with graphical depiction of magnetic dipole direction and strength.[16] ...... 6 Figure 5: The interactions between electric field, magnetic field and stress, and the control of polarisation, magnetism, and strain. In a multiferroic the presence of two or more ferroic orders can also exhibit coupling of the orders[1] ...... 7 Figure 6: Bismuth ferrite unit cell showing the alternating oxygen octahedra tilts in the pseudocubic perovskite cells [8] ...... 9 Figure 7: Schematic of G-type antiferromagnetic ordering. The arrows denote the direction of the magnetic moment at that atomic site...... 10
Figure 8: Polarization direction and cycloid plane for BiFeO3 bulk [15]. Note that if P is along [111] the cycloid plane is (-11-1), while when P is along [1-11], the cycloid plane is (111) [29] ...... 11 Figure 9: Strain engineering of perovskite oxides. (a) shows the unit cell with oxygen cages. In (b), the film material is grown on a substrate with a smaller in plane lattice constant, resulting in a compressive strain applied to the film. In (c) the substrate has a larger in plane lattice constant, imposing a tensile strain to the film (From Ref. [33])...... 12 Figure 10: Structural variants of BFO according to epitaxial strain state, from Ref. [13]...... 13 Figure 11: Schematic of structural variants in ‘mixed phase’ BFO, c/a ratios and octahedral rotation patterns, from Ref. [27] ...... 15 Figure 12: AFM topography of BFO film approximately 60 nm in thickness on (001) LAO substrate, showing typical striped 'mixed phase' topography (sample grown as part of study presented in Chapter 3)...... 15 Figure 13: Schematic of an MERAM device structure. Ferroelectric – antiferromagnetic multiferroic green layer, ferromagnetic blue layer. Interface exchange coupling allows low current switching of the magnetoresistive memory cell, from Ref. [49]...... 16 Figure 14: Typical PLD schematic ...... 17 Figure 15: Schematic diagram of Göbel mirror geometry ...... 25 Figure 16: 2-bounce monochromator schematic ...... 25 ix
Figure 17: 2ϴ-ω geometry schematic ...... 26 Figure 18: 2ϴ-ω 2theta-omega coupled scan of BFO on (001) LAO around the 001 film peak with simulated pattern (red)...... 27 Figure 19: Schematic of reflection at film interfaces ...... 29 Figure 20: XRR pattern of a ~55 nm thick BFO film on (001) LAO, with simulated pattern (red) 29 Figure 21: Schematic of ω, χ and φ geometry for in-plane lattice points ...... 31 Figure 22: Schematic of how angular space maps reciprocal space on the Ewald sphere ...... 32 Figure 23: Symmetric RSM of BFO on (001) LAO around the 001 reflection showing peaks tilted off the surface normal ...... 32 Figure 24: Schematic of a typical AFM system. a) tip and cantilever, commonly integrated, b) sample, c) sample stage with x,y motion, d) tip holder with z motion, e) piezoelectric transducer for applying tip oscillation, f) (laser diode) and g) (quadrant photodiode) for detecting tip position, h) feedback circuit...... 35 Figure 25: PFM schematic diagram, application of a bias through the tip and a bottom electrode, piezoelectric response can be measured locally. a) out of plane polarisation, b) in plane polarisation...... 37 Figure 26: AFM topography scans for PLD A and PLD B films. An increasing fraction of mixed phase is observed with increasing thickness in the PLD A films (from [70])...... 41 Figure 27: High angle 2ϴ/ω scans around the 001 and 002 reflections for BFO on (001) LAO grown in two different PLD chambers...... 42 Figure 28: Symmetric XRD reciprocal space maps (RSMs) near the 001 reflection for (a-d) films grown in PLD A; (e-h) films grown in PLD B, showing no tilted phases. Note that the thicker films from PLD A (g,h) show extra peaks corresponding to tilted S’ phases (from Ref. [70]) ...... 43 Figure 29: Energy dispersive X-ray (EDS) TEM analysis. (a) HRTEM image showing a defective region (circled). (b-e) show EDS maps at the same location for Bi, La, Fe, and Al respectively. Note that a slight increase in Bi brightness and corresponding decrease in Fe brightness may be surmised from these images; however conclusive evidence is lacking...... 44 Figure 30: STEM HAADF image processed by the Richardson-Lucy method [71], [72] showing defective nano-region, b) enlarged image of the nano-region, c) S’ domain observed only in TEM samples...... 44 Figure 31: Coupled 2θ/ω scan of a BFO film after the ‘breakdown’ to a rhombohedral phase has occurred. Some peaks appear to correspond to various orientations of rhombohedral BFO, while bismuth oxide phases are also found...... 45
x
Figure 32: a) RHEED specular intensity during growth of a PLD B film, showing steep drop in intensity around 30,000 pulses (or thickness of ~75 nm). b) AFM topography scan on the same film showing no evidence of mixed-phase stripes, nor stepped topography; instead an aligned needle-like structure is observed...... 46
Figure 33: Temperature-pressure phase diagram for the growth of epitaxial BiFeO3 thin films, from Bea et al. [59] ...... 49 Figure 34: Concept of epitaxy of Bi2O3 on SRO//STO as a buffer layer for T’ BFO films on top. From Ref. [80]...... 50 Figure 35: Schematic diagram of 'defect seeds' ...... 52 Figure 36: Formation of the mixed-phase regions in BFO//LAO films, using defect seeds. (a) AFM topography showing atomic steps and also existence of characteristic ‘striped’ regions. (b) XRD 2θ/ω scan, inset comparing with a similar film with no defect seeding, (c) STEM HAADF image showing the defect nano-pockets (inset shows enlarged view of a defect), (d) enlarged view of the amorphous interface region. (From Ref. [70]) ...... 53 Figure 37: AFM scans of Cobalt BFO films of various thicknesses on 001 LAO; a) 16k, b) 40k and c) 80k pulses of CoBFO. On 001 LAO with a ~2nm thick LSMO bottom electrode d) 16k, b) 40k and c) 80k pulses of CoBFO. All scans 3um x 3um, k denotes 1000 pulses...... 58 Figure 38: 1x1 µm AFM topography scan of a triangular outgrowth showing stepped structure, height scale is 4.81 nm...... 59 Figure 39: a) AFM scan showing stepped topography with voids and secondary phase particles. b) XRD 2theta-omega scan showing the T’ phase film peaks and no R’ or S’ peaks. c) - f) TEM and EDS scans showing that the secondary particles are iron rich...... 60 Figure 40: High angle XRD scans around the 001 and 002 reflections of Co doped BFO on 001 LAO substrates a) without a bottom electrode, blue 40k pulses of CoBFO, red 16k pulses of CoBFO, b) 40k pulses of CoBFO, blue with a ~20nm thick LSMO bottom electrode, red with a 2nm thick LSMO bottom electrode...... 61
Figure 41: Reciprocal space maps of a ~56nm BiFe0.98Co0.02O3//LaAlO3 film. a) 001 RSM, red circles denote expected position of tilted phase peaks. b),c) 103 and 113 RSMs respectively (from Young et al. 2018 [92])...... 61 Figure 42: PFM scan a) showing out of plane polarization direction (inset topography), b) in plane polarization, c) local PFM hysteresis loops...... 63 Figure 43: PFM box in box switching, a) Topography, b) amplitude and c) phase scans of a 2.5x2.5µm region ...... 63
xi
Figure 44: (a) AFM scan showing induced topography changes from a 12V applied bias at different time scales. Corresponding PFM amplitude and phase scans showing the induced ferroelectric switching (b) and (c) respectively. (d) Line profiles taken along the dashed lines. (e) Induced ferroelectric domain size with increasing pulse duration at 12V (black), 15V (blue) and 18V (red), logarithmic fit for the eye, pulse shape inset...... 64 Figure 45: Macro scale electronic properties of a Pt/Co-BFO/LSMO//LAO capacitor structure (Co- BFO 40k, LSMO 4.5k) a) Current - Voltage (I-V) response, b) Polarization – Voltage (P-V) response...... 65 Figure 46: AFM scans of the a) as-grown, b) mixed phase induced and c) after erasure of mixed phase topographies. d) line profiles of the surface along the indicated line showing the indicative saw tooth pattern of the mixed phase and the return to the original stepped surface upon reversal of bias...... 66 Figure 47: (a) PFM scan of a region with an artificially generated domain, (b) conductive AFM scan of the same region (DC bias -2.8V), showing the enhanced conductivity at the domain walls, (c) localised current-bias on the domain wall (blue) and off the domain wall (red), (d) localised current on the domain wall over 200s (DC bias -3V) (from Young et al. 2018 [92]) ...... 67 Figure 48: (a) M(H) loop indicating low saturation moment of ~4 emu/cm3. XAS and XMCD spectra for Fe (b) and Co (c) showing the L2,3 electron edges and no signal in XMCD...... 68
xii
Introduction
Multiferroics are a class of materials that exhibit two or more ferroic orders and have potential applications in devices such as actuators, sensors and data storage [1]. A multiferroic based device has potential to be multifunctional, a desirable property for the continued miniaturisation of electronics.
One such material is bismuth ferrite, BiFeO3 (BFO). First synthesised in the 1950’s [2], BFO is the only known multiferroic with its antiferromagnetic Néel [3] and ferroelectric Curie [4] transition temperatures above room temperature, making it a prime candidate for device applications.
The properties of BFO are intimately tied to the crystal structure, and many different polymorphs can be stabilised. A commonly employed technique to modify the crystal structure is growing epitaxial thin films [5]. Careful selection of the substrate crystal allows for fine tuning of material properties as well as potential integration into existing electronics manufacturing techniques.
When BFO is grown epitaxially under strong compressive (~4.5%) strain, a metastable phase with an elongated out of plane lattice parameter can be stabilised, referred to as the T’ phase [6], [7]. This change in structure is accompanied with an enhanced ferroelectric polarisation. Strain relaxation effects means that the T’ phase typically breaks down into a complex mixed phase structure in films thicker than ~20nm [8]. This limit on the thickness presents challenges for both analysis and application of the T’ phase.
It has also been shown that BFO can exhibit electrical conductivity in its ferroelectric domain walls [9], which can be utilised for nano-scale device operation [10].
In this thesis we investigate the growth and properties of thick T’ phase BFO films, and the effects of cobalt doping on the properties of BFO thin films.
The thesis is divided into five chapters:
Chapter 1 introduces the concepts of ferroelectricity, magnetism and multiferroics, an understanding of which is required for the remainder of the thesis. We then move to describe bismuth ferrite in more detail, first the bulk properties, and then explaining how the functional properties can be tuned by strain in thin film heterostructures. A detailed explanation of the so- called T’ phase of BFO follows.
1
Chapter 2 describes the experimental techniques used in this research work. Pulsed laser deposition (PLD), as the primary technique employed for sample preparation is described, and some of the factors that affect film quality are explored. Additionally, sample characterisation techniques such as X-ray diffraction and atomic force microscopy, and their derivatives are described and how they are useful for exploring thin films.
In Chapter 3, we present the results of an experimental study on the stabilisation of ultra-thick T’ phase BFO on LAO substrates. It is shown that by using ‘designer defects’, novel phases of BFO can be stabilised. We show that stoichiometric differences induced by specific growth conditions are the likely origin for such observations.
In Chapter 4, we explore the influence of Co doping on T’ BFO films, once again grown on LAO substrates. It is shown that 2% of cobalt substituted on the B site of the perovskite structure does not appreciably change the structural, symmetry, ferroelectric, or magnetic properties of T’ BFO films. We also show using scanning probe microscopy techniques that the domain walls in such films are conductive, offering perspectives for nanoscale domain wall electronics devices.
Finally, we conclude the thesis with future perspectives and possible further experiments with such BFO films.
1 Generalities
1.1 Ferroelectricity
Ferroelectricity is the presence of a spontaneous electric dipole polarization that can be switched by an electric field. For such a polarization to exist, the crystal lattice must have a non- centrosymmetric structure to allow the displacement of charge. Two (or more) orientations of the polarization are equivalent by symmetry, and it is possible through the application of electric field to switch between states. Most ferroelectric materials have a tetragonal structure and therefore the two polarization states can be used as the foundation of binary information storage. As a result, FE materials have been widely explored for application in random access memory (RAM). Since the FE polarization state is remanent (i.e. persists after the electric field is removed), the stored information is non-volatile, in contrast with transistor-based RAM. However, the requirement to destroy the polarization state during the readout process [11] has hindered large-scale development of Fe-RAM products. Nevertheless, there are commercial examples of Fe-RAMs, for example in mobile telephones or video game consoles. 2
Ferroelectrics can be characterised according to their crystal symmetry (Figure 1). From the 32 crystal classes, 21 are non-centrosymmetric, that is, they do not have a centre of inversion. All of these non-centrosymmetric crystal classes (except point group 432), exhibit the piezoelectric effect; that is, the formation of charge at the surface of a crystal when a physical strain is applied. The inverse piezoelectric effect is, on the other hand, the deformation of a crystal when an electric field is applied. The piezoelectric effect is widely used in sensors and the inverse piezoelectric effect in actuators and micromanipulators.
Among the 20 piezoelectric crystal classes, only 10 have an axis of high symmetry. Such crystals are considered polar and exhibit pyroelectricity. This effect manifests as a change in the charge distribution on the surface of a crystal when it undergoes a change in temperature, and is useful in infrared detectors.
Finally, if the polar state of the pyroelectric material can be switched by an applied electric field, it is ferroelectric. Note that the ability to switch the material is critical to its classification as a true ferroelectric; there are numerous examples of materials that have a strongly polar character but cannot be switched, and thus fall under the category of pyroelectric.
32 crystal classes
11 21 non centrosymmetric centrosymmetric classes classes
1 non- 20 piezoelectric piezoelectric classes
10 non- 10 pyroelectric pyroelectric
non- ferroelectrics ferroelectrics
Figure 1: Associated electronic properties of the crystal classes
3
The spontaneous dipole in ferroelectrics exists below the so-called Curie temperature (TC), above which temperature the material becomes paraelectric. At temperatures above TC the material is in a high-symmetry phase, while below TC the material is in a lower symmetry phase.
For perovskite materials with the simple ABO3 crystal structure [Figure 2 (a)], the microscopic origin of ferroelectricity can be divided into two broad categories; a) a reduction of symmetry due to ionic displacement; and b) the displacement of electrons [12]. Most common ferroelectric perovskites such as BaTiO3 and PbTiO3 have their ferroelectricity arising due to the displacement of the central B site cation towards the axial oxygen atoms where some bond hybridisation occurs [13], Figure 2 (a) and (b) show the two possible ionic configurations corresponding to the two polarization states.
Phenomenologically, for displacive ferroelectrics one can consider the existence of a ferroelectric distortion as related to a ‘double well’ in the free energy of the crystal as a function of the ionic displacements [see Figure 2 (c)]. At temperatures below TC, the free energy curve shows two minima, each corresponding to a possible configuration (and between which an electric field with energy sufficient to overcome the barrier EB can switch). On the other hand, at temperatures above TC the double well structure becomes a single minimum corresponding to a non-polar state. This is directly related to the crystal symmetry; for instance, for T > TC the crystal structure may be cubic.
a) b) A B O
c) E d) P T>Tc Pr
z E EB Ec T Figure 2: Structure of ABO3 perovskite: a) polarization up state, b) polarization down state, c) Energy vs ionic displacement plots: below TC (blue) and above TC (red) showing the energy barrier (EB) and double well, d) typical ferroelectric hysteresis loop with coercive electric field (Ec) and remnant polarisation (Pr) 4 1.2 Magnetism Every electron has a magnetic dipole arising from its spin, with the direction of the dipole corresponding to the spin direction. Due to energy minimisation effects electrons often form pairs with opposing spin known as Lewis pairs, with the net magnetic moment becoming zero [14]. For an atom to exhibit a magnetic moment it must possess unpaired electrons, and in order for the material to exhibit a spontaneous macroscopic magnetic moment the magnetic dipoles of all of the atoms must align. If we consider for simplicity magnetism arising from uncompensated orbital and angular momentum (i.e. uncompensated spins), we consider materials with permanent atomic moments. In this subcategory are four main types: paramagnetism, ferromagnetism, antiferromagnetism, and ferrimagnetism. These are described in Figures 3 & 4. Paramagnetic materials (e.g. at room temperature Al, Ti, FeO) have no long-range magnetic order. Ferromagnetic materials (e.g. Fe, Nd, Ni) have a spontaneous magnetic moment and as such are useful for magnetic memory (e.g. hard disk drives). Antiferromagnetic materials (e.g. Cr, NiO Fe2O3) have two sublattices in which the electron spins are oppositely oriented. The result is a zero net moment. As such antiferromagnetic materials are traditionally less technologically relevant (since they are more difficult to probe). Ferrimagnetic materials can be considered a combination of ferromagnetic and antiferromagnetic materials, where one sublattice has a lower magnetic moment, resulting in a (usually weak) net magnetic moment. 5 Diamagnetism: property of all matter Uncompensated orbital and spin Electron energy angular momenta in bands in metals all solid types Permanent Atomic Pauli spin moments paramagnetism Independent Cooperating atomic Band atomic moments moments antiferromagnetism Band Ideal paramagnetism Ferrimagnetism Ferromagnetism Antiferromagnetism ferromagnetism Figure 3: Hierarchy of magnetism [15] Figure 4: Magnetic orders with graphical depiction of magnetic dipole direction and strength.[16] 6 1.3 Multiferroics The preceding discussion presented two types of ferroic ordering in solids, namely ferroelectricity and magnetism. In certain materials, these orders can coexist, giving rise to complex and rich physical properties and applications potential. Such materials are called multiferroics, since they have multiple ferroic orders. Strictly speaking, multiferroic materials are defined as materials that exhibit at least two or more of ferroelectricity, ferromagnetism, and or ferroelasticity (recently another has been added known as ferrotoroidicity). Often in such materials a coupling of the order parameters exists (as first explored in Landau and Lifshitz in ‘Electrodynamics of continuous media’ [17]) Figure 5: The interactions between electric field, magnetic field and stress, and the control of polarisation, magnetism, and strain. In a multiferroic the presence of two or more ferroic orders can also exhibit coupling of the orders[1] Spontaneous ferromagnetism arises due to unbalanced electron spins, in the elemental ferromagnetic metals this is from the fermi energy level sitting within the 3d electron band. Due to the overlap with the 4s band and the 4s density of states being lower at the fermi level, the exchange interaction stabilizes more spins in the same direction. At the same time the atomic- level origin of ferroelectricity in perovskites is commonly related to the hybridisation of d shell electrons in the B site ion and the p shell electrons in the oxygen ion to generate ionic displacement. The transition metal d shell must be occupied for magnetism to arise, but this reduces the likelihood of ionic displacement occurring, supressing ferroelectricity. As a result there must be additional factors for magnetism and ferroelectricity to occur concurrently, and this even in perovskites containing magnetic transition metals, multiferroism is rare [1],[16]. 7 While multiferroics are extremely promising for next-generation memory applications, most multiferroic materials such as RMnO3 (R is rare earth) order well below room temperature, thus limiting their application potential. On the other hand, bismuth ferrite (BiFeO3 – BFO) is the only know example of a single phase multiferroic that orders above room temperature. As a result, significant research effort has been devoted to understanding this compound. 1.4 Bismuth Ferrite Bismuth ferrite (BiFeO3 – BFO) has attracted significant scientific interest in recent years. As the only known room temperature multiferroic material, it has been proposed as the foundation for a wide variety of devices exploiting its coupled ferroelectric and antiferromagnetic order [18], [19]. The concept of writing magnetic polarisation with an electric field offers power savings related to the energetic cost of generating a magnetic field. The extensive research interest in BFO since the report of epitaxial thin films in 2003 [5], has led to the discovery of a great many other properties of BFO containing thin film heterostructures, along with associated possible device applications. BFO was first synthesised in the 1950’s [2]. Initial bulk studies by neutron powder diffraction showed it to have a distorted perovskite structure with the rhombohedral space group R3c. The rhombohedral unit cell is comprised two pseudocubic perovskite cells as shown in Figure 6. Later it was shown to possess G-type antiferromagnetic order [3], [20]. At the time the measurable polarization of BFO was only 6 µC/cm2 [21], probably due to impurity phases present in the samples resulting in high leakage current hindering the measurements. First principles calculations, however, suggested a much higher intrinsic ferroelectric polarisation of the order ~100 µC/cm2 [7], [8], and the availability of high-purity single crystals in the 2000s allowed the measurement of BFO’s true, intrinsic polarization, with reported values as high as ~100 µC/cm2 [22], [23]. 8 Figure 6: Bismuth ferrite unit cell showing the alternating oxygen octahedra tilts in the pseudocubic perovskite cells [8] Despite the multiferroic nature of BFO and its high Curie and Néel temperatures (meaning that BFO is the only known multiferroic at room temperature [3], [4]), the apparently low polarisation reported in bulk samples [21] resulted in fairly low research activity until the 1990s. In 2003 the Ramesh group (Berkeley) reported work on epitaxial BFO thin film structures [5]. In this work, epitaxial films were grown on strontium titanate (SrTiO3 - STO) substrates with high purity, and most importantly, the FE polarization was found to be more than an order of magnitude higher than had been previously reported in the bulk. In addition to this, a magnetic moment (as high as 150 emu/cc) was measured on such films, apparently offering the dream of a strong ferroelectric coupled with ferromagnetism. Although the magnetism was later shown to be of extrinsic origin [24],[25], the original Science paper encouraged a new wave and renaissance in multiferroics and drove the community to explore BFO and multiferroic oxides in more detail. 1.4.1 Bulk Properties The origin of ferroelectricity in BFO is ionic displacement along the [111] pseudocubic direction, With the largest displacement being by the A site (bismuth ions) [8]. The 6s electrons on the bismuth ions, known as the ‘lone pair’, are not involved in chemical bonding and as such are able to move towards the surrounding oxygen atoms and undergo sp hybridization, facilitating the stabilization of the ionic displacement and giving rise to an electric dipole. This mechanism is also important for other materials with lone pairs, such as PbTiO3 [26]. The R3c space group - - - also allows antiphase octahedral rotations around the [111]pc direction (a a a in Glazer 9 notation,[27]). These rotations are induced in BFO due to the low Goldschmidt tolerance factor which means that the unit cell is too small and the octahedra buckle to fit inside the cell [19]. Given its rhombohedral structure, in the BFO pseudocubic unit cell the polarization can point along one of the 8 body diagonal directions. As a result, there are three possible types of domain walls separating regions of uniform polarization. These are commonly called 71°, 109° and 180° walls. The latter is a purely ferroelectric wall, while the first two are also ferroelastic, since there is a difference in the spontaneous strain in the domains on either side of the wall. The presence of Fe3+ in BFO gives rise to long range magnetic order. In this case the order is G- type antiferromagnetic; that is, every magnetic moment is aligned antiparallel with its nearest neighbour both in plane and intra-plane (Figure 7). The interaction responsible is the so called superexchange, where the interaction between Fe ions goes through the intermediary oxygen. Figure 7: Schematic of G-type antiferromagnetic ordering. The arrows denote the direction of the magnetic moment at that atomic site. The antiferromagnetic ordering results in a net zero magnetic moment, however in the case of BFO due to the symmetry being broken by the ferroelectric polarization, there is a slight spin canting, and this results in a weak ferromagnetic moment (~0.3 emu.g-1) via the Dzyaloshinskii- Moriya interaction [28]. Imprinted on top of the antiferromagnetic order is a long-distance spin cycloid with a periodicity ~ 62 nm, resulting in a net zero magnetic moment. The magnetic easy plane of the cycloid (i.e. the plane in which the spins rotate) is usually defined by the direction of polarization and the cycloid propagation direction, and in single crystal BFO, when the ferroelectric polarization direction is switched the cycloid propagation direction has been shown to change with it [29]. In thin films, strain effects play a critical role, with high levels of epitaxial strain destroying the cycloidal modulation. On the other hand, in thin films under low strain, the cycloid has been shown to exist [30]–[32]. 10 Figure 8: Polarization direction and cycloid plane for BiFeO3 bulk [15]. Note that if P is along [111] the cycloid plane is (-11-1), while when P is along [1-11], the cycloid plane is (111) [29] 11 1.4.2 Epitaxial thin films of BFO While in the bulk BFO is a remarkable compound, with considerable complexity in its multiferroic orders and the coupling between them, it has above all been the integration of BFO in epitaxial thin films that has shown the true versatility and multifunctional nature of this material. As with most perovskite oxides, the functional properties of BFO are intimately connected with its crystal structure. The interatomic distance and charge interactions dictate the ferroelectric and magnetic properties, and, as such it becomes apparent that modifying the crystal structure can tune the functional properties. This concept forms the basis of so-called ‘strain engineering’. One method of applying a strain to a crystal lattice is to grow the target crystalline material on another crystal with a different lattice parameter. The substrate crystal provides a template for the growth of the target material (Figure 9), and the film can be made to grow with perfect alignment to the substrate, and is one of the primary methods of producing high quality virtually defect-free samples. Figure 9: Strain engineering of perovskite oxides. (a) shows the unit cell with oxygen cages. In (b), the film material is grown on a substrate with a smaller in plane lattice constant, resulting in a compressive strain applied to the film. In (c) the substrate has a larger in plane lattice constant, imposing a tensile strain to the film (From Ref. [33]). If the film is much thinner than the substrate; i.e. tfilm << tsubstrate, the substrate can be considered to be virtually unstrained and all the strain applied to the film. The choice of single crystal substrate, as well as crystal orientation, can be used to vary the level of strain applied. As shown in Figure 9, a substrate with a smaller (larger) in-plane lattice parameter than the film will impart a compressive (tensile) strain to the film. Through this method strains that would result in fracture in bulk crystals is possible, large strains can result in changes in the crystal structure of the BFO film compared to bulk. BFO has been successfully grown on a wide range of substrates, with this flexibility being attributed to the small Goldschmid tolerance factor (~0.88) allowing 12 the oxygen octahedra a large degree of freedom to rotate and tilt to accommodate strain [34]. Figure 10: Structural variants of BFO according to epitaxial strain state, from Ref. [13]. When grown on strontium titanate (SrTiO3 – STO) (a popular perovskite oxide substrate) BFO is under approximately 1.5% compressive strain and a monoclinic MA phase similar to the bulk like rhombohedral structure is observed [13] [Figure 10 (c)]. In this work we refer to this as R’ BFO. In addition to the monoclinic MA phase, a range of other crystallographic structures are possible in epitaxial BFO thin films, as shown in Figure 10. Of particular note for this work is the monoclinic MC phase [Figure 10(b)], which can be stabilised by large compressive strains, for example by using a substrate of lanthanum aluminate (LaAlO3 – LAO) which imparts a compressive strain of ~4.5 %. This MC phase of BFO is sometimes called ‘super tetragonal’ or ‘tetragonal like’ BFO due to its very large (~1.23) axial ratio. The next section introduces this phase in more detail. 1.4.3 Tetragonal-like and mixed-phase BFO When BFO films are grown epitaxially under strong compressive strain (~ 4-5 %) a metastable phase of BFO can be formed. This ‘tetragonal like’ or T’ phase (thus referred to for the remainder of this work) typically has a Mc type monoclinic structure and a large axial ratio of ~1.23 [6], [7], as shown in Figure 10(b). The dramatic change in structure induces a shift in the Fe coordination from octahedral (in the R’ phase) to square pyramidal oxygen cage (in T’ phase) and a large displacement of the Fe3+ ion towards one of the apical oxygen ions. The large distortion of the unit cell in the T’ phase of BFO induces an apparent enhancement of the ferroelectric polarization up to 150 µC/cm2 [6], [35], [36], [37]; however this enhancement has not been shown unequivocally experimentally on pure T’ BFO samples. 13 A rather unique characteristic of the T’ phase of BFO is its strain relaxation mechanism. In thin film epitaxy, when a film is grown above its so-called critical thickness, strain relaxation typically occurs through the formation of misfit dislocations [38]. For T’ BFO, however, strain relaxation occurs through the transformation of the T’ phase into the more stable R’ phase. This strain relaxation mechanism results in a coexistence of the T’ and R’ phases [19], typically manifesting as striped topography on the film surface with a periodicity on the order of 10-100 nm (Figure 11) [39]. This ‘mixed phase’ BFO is remarkably similar to the situation in high performance piezoelectrics [40] in which a so-called morphotropic phase boundary (MPB) exists as a function of chemical composition. In the case of mixed phase BFO the MPB is induced by strain, resulting in a so-called ‘strain driven morphotropic phase boundary’. Similar to the situation in MPBs, external stimuli (such as electric field) can easily induce a phase transformation, resulting in giant responses. Since the axial ratio of the T’ and R’ like phases are 1.23 and 1.05 respectively, converting between them results in large change in the out-of-plane lattice parameter and therefore large electric-field-induced strains. This possibility was demonstrated by Zeches et. al. [39] who induced a large electric field induced local strain of up to 5 % in mixed-phase BFO films. This discovery triggered intense research interest in this material system as a potential lead-free piezoelectric. A remarkable characteristic of mixed-phase BFO is that the interface between the T’ and R’ phases is completely coherent (i.e. dislocation free), with a c/a ratio change of ~13 % over ~10 unit cells [39], resulting in a huge strain gradient [41]. The R’ phase present in the stripes of mixed-phase samples is compressively strained ~2.5% compared to bulk with a c-axis tilt of ~3° out of plane [42], as proposed previously, in this work we refer to this as the S’ phase, in order to differentiate it from the true R’ phase (which is under almost no compressive strain) [43]. The bulk R’ phase exhibits a weak magnetic moment due to spin canting of ~0.3 emu/cm2 [44], while the T’ phase and the reduction of symmetry in the c-axis oxygen octahedral tilt has been shown to contain no detectable magnetic moment [45]. The mixed phase exhibits an enhanced magnetic moment over the bulk R’ phase that vanishes at 175°C, this suggests that the enhanced magnetic moment arises due to the compressed rhombohedral S’ phase which has also been shown to vanish at 175°C [46]. The oxygen octahedral rotation around the c-axis increases with increasing compressive strain, before a suppression of rotation at ~4.5% compressive strain [43], [47], the presence of this tilt is a potential explanation for this enhanced magnetic moment in the S’ phase. 14 Figure 11: Schematic of structural variants in ‘mixed phase’ BFO, c/a ratios and octahedral rotation patterns, from Ref. [27] Figure 12: AFM topography of BFO film approximately 60 nm in thickness on (001) LAO substrate, showing typical striped 'mixed phase' topography (sample grown as part of study presented in Chapter 3). 1.4.4 Applications of BFO The functional properties of BFO and its wide compatibility with commercially available substrates make it appealing for application in devices [18]. Utilising its large ferroelectric polarisation, it has potential replacement for non-volatile memory applications such as FeRAM, where the large polarisation allows for easy reading of states. The antiferromagnetic ordering of BFO being coupled with the ferroelectric order at room temperature makes it a candidate for spintronic devices allowing the magnetic ordering to be switched with an electric field, typically at much lower power than required to switch a magnetic moment with a magnetic field. By growing heterostructures containing BFO and ferromagnetic layers, through interface exchange interactions the ferromagnetic material can be controlled by an electric field [48][49]. 15 Figure 13: Schematic of an MERAM device structure. Ferroelectric – antiferromagnetic multiferroic green layer, ferromagnetic blue layer. Interface exchange coupling allows low current switching of the magnetoresistive memory cell, from Ref. [49]. The discovery of electrical conductivity in the domain walls of BFO by Seidel et al. [9] led to proposed devices built around the nanoscale manipulation of domain walls, and their mobility and controllability with electric fields. Device prototypes showing good endurance, reliable switching and retention characteristics have recently been demonstrated [10]. The multi-functional nature of BFO and the wide range of crystal structures available in thin films, makes this material attractive for many device possibilities. The key to accessing this wide array properties on demand is carefully controlling the stoichiometry and structure of films. Fabrication methods and nuances of growth processes of these films, particularly in the T’ like phase of BFO is thus critical, and forms the foundation of this thesis. 16 2 Experimental Techniques There are many techniques available for growing oxide thin films, such as chemical routes (e.g. sol-gel), molecular beam epitaxy (MBE), sputtering, and atomic layer deposition (ALD). The most popular technique for oxide thin films is pulsed laser deposition (PLD). Physical vapor deposition techniques are probably the most popular methods for achieving epitaxial oxide growth. In this thesis, pulsed laser deposition (PLD) was extensively utilised. This technique is described in greater detail in this chapter. 2.1 Pulsed Laser Deposition Pulsed laser deposition (PLD) was pioneered in the 1960’s [50] and saw a surge in research interest in the 1980’s after it was reported as a method to produce epitaxial films of high- temperature superconductors such as Yttrium barium copper oxide (YBCO) [51]. The ability of PLD to grow complex multi-cation systems in an oxidising environment makes it a powerful method for producing oxide thin films [52], [53]. The PLD system consists of; 1) a deposition chamber containing the heated stage for mounting the substrate and the target holder, 2) laser and optics positioned such that it can be focused through a laser window onto the target surface, and 3) vacuum pump system and background gas introduction system. Figure 14: Typical PLD schematic 17 2.1.1 Operating principle When a highly focused laser pulse hits the target, the irradiated region is rapidly heated, and a small volume of material is ablated. The ablated material absorbs more of the laser pulse and is broken down into atomic and diatomic species, forming a plasma that travels to the substrate. The nonequilibrium nature of ablation is vital in ensuring cation stoichiometry in the plasma plume. Selection of laser wavelength is based on the absorption coefficient of the target material: absorption must be sufficient to produce localised heating much higher than that required for thermal evaporation. The material flux due to thermal evaporation would be dependent on the vapor pressure of each constituent of the target. By using nonequilibrium ablation all target cations can be vaporised simultaneously and have the same velocity, making the material flux composition identical to that of the target. An important aspect regarding ablation is that the target phase need not be the same as that of the film, only chemical composition is important. This can allow the epitaxial stabilisation of metastable phases that would normally not form in the bulk. Ultraviolet (UV) laser wavelength is suitable for most oxide thin film materials, which are usually also highly refractory. Krypton Fluorine (KrF, lambda = 248 nm) and frequency tripled Neodymium doper Yttrium Aluminium Garnet (Nd:YAG; lambda = 355 nm) lasers are commonly used due to their high energy output and suitable wavelength [53]. Target selection affects film growth in more than just cation stoichiometry. High target density is important for the suppression of particle formation during ablation, with incomplete sintering of precursor powders resulting in the release of large particles. In the specific case of BFO there is often high bismuth evaporation from the substrate surface (due to its low vapour pressure). To balance the bismuth loss, targets with a nominal 10 % bismuth excess (i.e. Bi1.1FeO3) are often employed. To prevent pitting into the target surface, the laser is rastered over the target surface to reduce uneven wear of the target. The plasma plume is strongly confined to the surface normal, regardless of the angle of incidence of the laser pulse due to collisions of particles within the plume [52]. The deposition pressure strongly influences the plume speed and energy by changing the mean free path before collision of the plasma. Growth can be performed in background pressures ranging from ultra-high vacuum (~10-9 Torr) to 1 Torr with the background gas being part of the film chemistry. In the case of oxides, oxygen can be introduced during deposition to supply the oxygen for film chemistry as well as to modulate the mean free path of the plasma [54]. To lower 18 the number of gas/plume interactions while maintaining oxygen chemistry, alternatively ozone can be used as the background gas [55]. Collisions with the background gas and the highly energetic plume results in the formation of molecules containing the background gas, assisting its incorporation into the film. Typical sticking coefficients (the ratio of atoms that stick to the surface compared to atoms that hit the surface [52][56]) for metal atoms are ~1 and gas atoms are ~0.1-0.5 at room temperature; molecules containing both metal and gas atoms have a typical sticking coefficient ~1 [52]. Angular variation of the plume due to variation in mass of the plume constituents and particle collisions [57] can lead to variations in stoichiometry across the film. Rotation of the substrate, as well as rastering the beam across the target surface during deposition can be used to minimise this effect. Background gas pressure and substrate-target distance has been shown to broaden the angular distribution of the plume and modify the angular distribution of individual components of the plume constituents [58]. Growth rate is dependent on a range of parameters, including laser repetition frequency (usually in the range of 1-20 Hz), volume of material evaporated, uniformity of plume flux, and plume mean free path till collision. Increasing the pulse rate increases the volume of material evaporated, however target heating due to decreased time for heat dissipation between pulses with higher pulse rates may lead to changes in plume chemistry due to preferential evaporation occurring as a result of variation of vapor pressures of the involved species. The introduction of larger particles into the plume may also occur due to the heated target. Higher pulse rates lead to a higher atomic flux and an increase in the number of collisions within the plume resulting in the formation of larger molecules within the plume. When these larger species reach the substrate in sufficient numbers they can provide additional nucleation points and thus change the so-called ‘growth mode’. The growth mode directly influences the crystallinity and/or surface morphology of the film. There are three growth modes, as determined by the balance of substrate surface free energy, substrate/film interface energy and film surface energy. Layer-by-layer growth occurs when the surface energy of the interface and film surface is lower than that of the substrate surface. Island growth occurs when the interfacial energy is higher than that of the bare substrate surface energy, preventing the film from wetting the substrate surface. In the case of vicinal substrate surface, step-flow growth can occur. The adatoms land on the substrate surface and diffuse to the step edge where they become trapped. 19 Surface mobility and the energy required for inter-layer diffusion determine which growth mode occurs. In layer by layer growth, surface mobility and inter-layer diffusion are high enough such that an entire layer forms before nucleation of new terraces. Island growth occurs when the inter-layer transport is not high enough and nucleation occurs on existing terraces. Step flow growth occurs on vicinal substrates where the step width is lower than the diffusion distance before the adatom becomes trapped, the step edge traps the adatom, preventing inter-layer transport. The average diffusion distance before the adatom becomes trapped is controlled by the surface diffusion coefficient Ds, given by: 2 퐸퐴 퐷푠 = 푣푎 ( ). 푘퐵푇 Here v is the attempt frequency, a is the characteristic jump distance and EA is the activation energy for diffusion. From this equation it is apparent that substrate temperature T has a huge influence on the surface diffusivity. The average diffusion distance is also affected by the time before re-evaporation which is also affected by substrate temperature. Species with a lower vapour pressure will have a different surface diffusion coefficient, implying that epitaxial growth occurs in a temperature window that is highly dependent on the material system. 2.1.2 Experimental method The majority of the films discussed in this thesis were prepared using a NEOCERA PLD system utilising a Coherent Compex Pro 102F KrF excimer laser with a 248 nm wavelength. Laser fluence was calculated to be ~3 J/cm2 from a measured laser energy before the focusing lens of ~115 mJ and an assumed energy loss of 10 % at each optics interface. Substrates were supplied by Shinkosha (Japan) as 15 x 15 x 0.5 mm3 single crystals, and were cleaved to the typical experimental size of approximately 5 x 5 mm2. As foreign particles on the substrate surface act as nucleation points for defects (and thus lower film quality), the cleaved substrates were thoroughly cleaned before being mounted for deposition. The substrate was submerged in isopropyl alcohol (IPA) in a clean beaker and sonicated for 60 seconds, then the top surface was flushed with IPA and then wiped down with a lint free laboratory wipe dampened with IPA. The substrate was then dried with nitrogen gas. 20 The substrate mounting plate and holder was sanded with 600 grit silicon carbide sandpaper and wiped down with IPA and dried with nitrogen gas; this is in order to remove any silver conductive paint left over from previous depositions as well as any oxides or cross contaminants. The substrate was adhered to the mounting plate with a film of conductive silver paint, and to remove any air bubbles and to ensure a good thermal interface, the substrate was pressed down. A final burst of nitrogen was used to remove any dust/substrate chips from the mounting procedure and the mounting plate (with substrate) loaded into the PLD chamber. 2.1.3 Film Growth The PLD chamber was first evacuated to a base pressure of approximately 5 x 10-5 Torr and the substrate heated to the deposition temperature at a ramp rate of 20 °C/min. The deposition pressure is attained by first slowing the turbo pump down to 300 Hz and oxygen flowed into the chamber through a mass flow controller, pressure was adjusted by controlling the speed of the turbo pump as well as adjusting the mass flow controller. This allows fine tuning of deposition pressure to within 0.1 mTorr. Prior to the deposition step, the target surface was first pre-ablated at growth conditions to remove contaminants and any dust remaining from the target grinding process. The substrate was covered by the shutter during pre-ablation and the target position, with rotation and raster swing angles ranges determined at this stage. Ablation was carried out at high frequency to evenly ablate the entire area to be used during film growth, and continued until the plume stabilised, indicating that steady state has been achieved. Typical pre-ablation conditions were 10 Hz for 3000 pulses. 2.1.4 PLD growth conditions for BFO films Deposition conditions for BFO were dependent on substrate and target composition. Typical growth conditions are summarised in Table 1. 21 Table 1: Typical PLD growth parameters in the Neocera system Deposition Oxygen Laser Laser Cooling Composition Substrate Temperature pressure Energy Frequency conditions (⁰C) (mTorr) (mJ) (Hz) Torr/oC/min BFO LAO (100) 700 23 235 5-15 450T/20C/min BFCO LAO (100) 650 13 185 3 450T/20C/min BFO STO (100) 600 10 225 5 450T/20C/min BFO STO (110) 650 15 200 5 450T/20C/min Within the temperature and pressure window for epitaxial growth of BFO, secondary phases can form during growth [59]. Even minute amounts of secondary phases can induce conductivity in BFO, giving rise to leakage currents that hinder measurements of the polarisation (i.e. the P-E loops show a leaky behaviour). The much higher vapour pressure of bismuth compared to iron can result in the formation of bismuth deficient regions while the film is at elevated temperature; this is exacerbated by the vacuum conditions of growth further increasing the bismuth vapour pressure gradient. Bismuth deficiencies can also give rise to the formation of iron oxides, not only providing conduction paths through the sample but also altering the magnetic properties of the film (iron oxides are typically strongly ferromagnetic [25]).To obtain nominally stoichiometric films and to reduce the presence of oxygen vacancies, the films were grown using targets with a nominal 10 % excess of bismuth, and after growth the films were cooled to room temperature in an oxygen pressure of 450 Torr. 22 2.2 Physical characterisation of epitaxial BiFeO3 thin films Subsequent to thin film growth, the structural and physical properties of the films were assessed. In this thesis, x-ray diffraction, atomic force microscopy, including piezoresponse force microscopy, ferroelectric, and magnetic characterization techniques were used. The following section explains these techniques in more detail. 2.2.1 X-Ray diffraction X-ray diffraction (XRD) is a versatile technique for non-destructive characterisation of crystalline materials [60], [61]. This is due to the relative ease of production of x-rays, the permeability of most materials to x-ray radiation, and the characteristic wavelength of x-rays (~angstroms - Å) being similar in length to the atomic bond distance. The various geometries available on common laboratory-based x-ray diffractometers allow a wide range of measurements on a single machine. These include phase identification, film thickness, density, surface and interface roughness, and crystallinity measurements. In this section the measurements most commonly used for thin film applications are described. 2.2.1.1 Theory of X-ray diffraction X-ray diffractometry is based on Bragg’s law, which describes at which incidence angles ϴ constructive interference will occur on a crystal lattice with interplanar spacing d for an electromagnetic wave with wavelength λ, where n is an integer and corresponds with the diffracting plane: 푛휆 = 2푑 ∙ 푠푖푛휃 Constructive interference occurs if the diffracted wave from one lattice plane has the same phase as the diffracted wave from the plane below. This occurs when the additional distance travelled by the wave that is diffracted off the lower plane is an integer multiple of the wavelength. This effect is cumulative over the many planes in a crystal, and thus sharp peaks in diffractive intensity are observed when the Bragg condition is met. 2.2.1.2 Generation of x-rays In the laboratory setup, X-rays are typically produced by firing an electron beam at a metal anode under vacuum. When the highly energetic electron beam collides with the metal lower valance orbital electrons are ejected, higher valence orbital electrons drop down to fill the lower orbital and a photon is produced in the process. The choice of cathode metal controls the characteristic wavelength of the generated X-ray based on the difference in energy of each 23 orbital. In metal oxide thin film applications, a copper anode is commonly used, when a K-orbital electron is ejected a 2p electron from the L-orbital or a 3p electron from the M-orbital can drop down to fill the vacancy. The X-rays arising from the L-orbital are called the Kα X-rays and the M-orbital X-rays Kβ, Kα intensity is roughly 5 times that of Kβ. The 2p electrons have 2 discrete orbital levels with a 0.020 keV energy difference, this results in X-rays with 2 different wavelengths that are very close together. Under high resolution conditions, peak splitting can be observed, giving so-called Kα1 and Kα2; however in most cases the mean wavelength of Kα1 and Kα2 is used, written K훼̅. Table 2: Copper Characteristic X-ray wavelengths Cu Radiation Wavelength K훼1 1.54056 Å K훼2 1.54439 Å K훼̅ 1.54184 Å K훽 1.39222 Å In addition to the generation of characteristic x-rays, as described above, the deceleration of the electrons as they hit the anode also produces x-rays. In accordance with the law of conservation of energy, the decelerating electron releases a photon with energy equivalent to the amount it slows, at the energies required for the generation of the characteristic X-rays the photons releases are in the X-ray pattern. This is known as Bremsstrahlung radiation and produces a broad continuous range of wavelengths. 2.2.1.3 Optics For meaningful experiments the X-rays must have the same wavelength (be monochromatic) and all moving in the same direction and closely parallel to each other. The choice of optics is dependent on the resolution required to make a meaningful measurement. Increasing resolution results in a loss of overall beam intensity and thus requires longer counting times. A combination of filters and monochromators are employed to achieve high resolution scans. From the source the X-rays are divergent from the point of focus of the electron beam. They are first made parallel by a parabolic mirror known as a Göbel mirror. The source is placed at the focus of the parabola and a combination of geometry and specialised coatings result in a beam 24 with angular divergence of ~0.04°. The material choice and geometry of the mirror is chosen to selectively reflect only the wavelengths to be used for measurement; for example, in the case of a Cu anode the Kα radiation. Figure 15: Schematic diagram of Göbel mirror geometry For epitaxial thin film applications where very small shifts in peak positions can be used to measure film strain, higher angular resolution is required. For this reason a channel-cut single crystal monochromator is used. Single crystals of Germanium are used for Cu Kα radiation due to the high atomic scattering factor, crystallinity and stability. The single crystal of germanium is cut such that two (220) planes are parallel. The incident X-ray beam enters the crystal, and only the X-rays that satisfy the Bragg condition for Cu Kα radiation are diffracted, (22.6° for Ge 220). This results in a highly collimated and monochromatic X-ray source. The drawback of using such a monochromator is that the intensity of the x-ray source is reduced significantly. Figure 16: 2-bounce monochromator schematic A nickel filter can be employed to selectively reduce the intensity of the Kβ peak. The X-ray absorption edge of nickel is 1.488 Å which is between that of Cu Kα and Kβ wavelengths. By inserting a nickel foil into the beam path the Kβ radiation can be attenuated. The application of the nickel filter increases the ratio of Kα radiation at the cost of lower overall intensity. The ratio of Kβ reduction is dependent on the foil thickness, typical thickness used is around 20 µm. 25 Once an acceptable beam quality is achieved with filters and attenuators, the beam shape must be selected for the sample to be measured. The goal is to have the largest possible beam size without straying off the edge of the sample. Larger beam spot results in higher intensity and shorter possible count times, or higher signal to noise ratio, however the beam spilling over the sample onto the backing plate will introduce background, reducing the signal to noise ratio. The selection of beam shapers must also take into consideration the angles used during measurement as the spot size will vary with incident angle. 2.2.1.4 2ϴ-ω coupled scans 2θ-ω coupled scans yield information for phase identification, out of plane (c-axis) spacing, confirmation of epitaxial growth, and in certain circumstances, can also give information on crystallinity and thickness. Figure 17: 2ϴ-ω geometry schematic The interplanar spacing of a material is determined by the atomic bonds and crystal structure and is characteristic of a material. Indexing the observed diffraction peaks using materials databases can be used for phase identification. In the case of thin film samples, often there is a strain introduced by the substrate and film having a different lattice parameter. This results in a change in the interplanar spacing and a corresponding shift in the angular position of the peak. By analysing the peak position, with the x-ray wavelength known it is possible to calculate the c- axis parameter. Considering the a and b-axis which is known for a single crystal substrate, the amount of shift of the film peak with respect to the substrate peak can be used to determine if the film is in compression or tension, or if epitaxial growth has occurred at all. 26 The full width at half maximum (FWHM) of a peak gives information about the crystallinity of the phase. For a perfect or nearly perfect crystal the FWHM is very narrow, as observed for a commercial substrate. The FWHM will increase with low angle grain boundaries, as the range of angles of the crystallites is averaged out under the area irradiated by the beam. For films with high crystallinity fringing may be observed around the film peak. These fringes are known as Laue fringes, and arise due to the constructive interference of X-rays as the beam is partially reflected and partially transmitted at each lattice layer of the X-ray beam as it travels [62]. The spacing of the fringes can be related to the thickness of the film using Bragg’s law where d is now the film thickness and n is the order of the fringe. The fringe spacing is identical to that of the Kiessig fringes observed in X-ray Reflectivity (XRR) (discussed below). Laue fringes due to arising due to high crystallinity are less sensitive to interface roughness than XRR and give complimentary information about the film quality. Figure 18 shows a 2θ scan exhibiting Laue fringes along with a simulated pattern. Figure 18: 2ϴ-ω 2theta-omega coupled scan of BFO on (001) LAO around the 001 film peak with simulated pattern (red). 2.2.1.5 Experimental Method used for this thesis A Phillips MRD with a Cu Kα tube source and a 2 bounce Ge-220 monochromator, as well as a Bruker D8 discover with a Cu Kα rotating anode source and a Hybrid Göbel mirror and 2 bounce Ge-220 monochromator were used to perform high resolution XRD. The choice of masks and 27 slits was determined by the size of the sample, with beam size maximised without being larger than the sample size. The sample was mounted to the goniometer stage using double sided tape on a glass microscopy slide. The use of an amorphous backing plate prevents any sharp peaks being introduced if the beam goes over the edge of the sample during scanning. Beam alignment is performed with all the source optics in place to ensure an accurate zero position. Continuing with the substrate alignment the z-axis of the goniometer is adjusted till half intensity is recorded at the detector, placing the substrate half way through the beam. A rocking curve or omega scan is performed to make the substrate surface parallel with zero degrees. The z and omega alignment is repeated until there are no further changes. Crystal alignment is then performed to place the crystal plane orthogonal to the scattering vector. The theoretical 2ϴ and ω positions for the peak with the highest intensity for the substrate used is moved to (e.g. the 002 peak for an 001 oriented STO substrate), and an omega scan performed to account for any substrate miss-cut or unevenness in sample mounting. Phi and Psi scans are performed to further refine the alignment, as well as positional x, y and z refinement. The refined positions are then used to reset the zero point of the goniometer. Once the sample is fully aligned the sample is scanned with a symmetric geometry. The detector settings are selected based on required resolution and reflected intensity. Typical settings were 0.01 degrees and 0.3 seconds per step. 2.2.2 X-ray Reflectometry X-ray reflectometry (XRR) is a glancing angle technique that uses internal reflection between interfaces to give information on film thickness, density, and interface roughness. Below a critical angle determined by the materials electron density, the beam is completely reflected, beyond the critical angle refraction begins to occur and the X-rays penetrate the material. At the air/film interface the incident beam undergoes partial reflection and partial transmittance. The transmitted portion of the beam undergoes a further partial reflection and transmittance at the film/substrate interface, and the reflected portion is partially transmitted at the air/film interface but with a phase difference corresponding to the film thickness. This process of partial reflection and transmittance is repeated until absorption of the X-rays brings the intensity down to zero. 28 Figure 19: Schematic of reflection at film interfaces Performing a coupled 2θ-ω scan yields an angle-dependent phase shift of the transmitted X-rays and constructive interference gives rise to an oscillating pattern known as Kiessig fringes, with a rapid drop off in intensity due to absorption. Thicker layers result in a shorter period of oscillation. Figure 20 shows a typical XRR pattern for a single layer film of BFO on LAO. Figure 20: XRR pattern of a ~55 nm thick BFO film on (001) LAO, with simulated pattern (red) The period of the oscillations and the refractive index of the materials in the system can be used to calculate the film thickness, though typically the oscillations are compared to a simulated pattern to determine thickness. In the case of multiple layers the oscillations are overlayed and a Fourier analysis and simulation allows determination of the thickness of such films. As XRR relies on the presence of interfaces the quality of the interface will influence the resulting pattern, non-specular reflection will occur when the interfaces are not perfectly flat, and this results in a loss of intensity. The extent of signal drop off if proportional to interface roughness, typically a film roughness less than 1 nm RMS is required for a signal to be obtained. The 29 film/substrate interface roughness is often neglected as commercial substrates usually possess very low roughness (~0.1 nm RMS), though poor substrate cleaning may result in a rough interface. The amplitude of the oscillations is dependent on the difference in density on either side of the interface, with a larger amplitude observed for a larger difference in density. 2.2.2.1 Experimental Method Choice of optics is dependent on the film thickness, with the smaller period of oscillation in a thicker film, higher resolution may be required. In the case of a very thin film, the oscillation period will be very wide so lower resolution optics may be selected. Thinner films will have a lower signal, so the increased intensity afforded by lower resolution optics can compensate for this. The choice of masks must be considered to avoid introducing background noise, with the glancing angle condition resulting in wider beam spread in the beam direction. The sample is mounted to the goniometer stage as per a coupled 2ϴ-ω scan and beam alignment and halving is identical. X and Y refinement cannot be performed at the critical angle due to the glancing angle geometry, to overcome this the theoretical peak is moved to confirmed with an omega scan, and x and y refined. The goniometer is then moved to around the critical angle for reflection (typically 2ϴ = 0.7° and ω = 0.35° for most oxides with a refractive index just below 1), and an omega scan performed to find the highest intensity. Phi and Psi are then refined. The reflectivity scan range is dependent on the film thickness, with multiple oscillations required for thickness calculation it is important to scan for a longer range if the film is very thin and the oscillation period is very long. As intensity drop off is rapid, it is recommended that the initial peak intensity at the critical intensity be in the order of a million counts, as such comparatively long step times are used compared to a normal scan, often in the order of seconds. For a roughly 50 nm thick film, a range of 0.3-5° with an angular resolution of 0.002° per step is typical. For thinner films in order to avoid excessively long scan times the angular resolution can be lowered to compensate for the longer step times and wider range required. 30 2.2.3 Reciprocal Space Mapping The XRD techniques discussed thus far have concerned probing the crystal lattice in the out of plane direction only. A reciprocal space map (RSM) is a way to visualise the entire crystal structure at once, both in and out of plane. Reciprocal space is effectively the Fourier transform of real space, allowing the 3-dimensional crystal lattice to be represented as a 2-dimensional projection. Reciprocal space is defined by the reciprocal space vector Q with the units of inverse length, related to angular space by: 2휋 4휋 sin 휃 푄 = = 푑 휆 Q can be broken down into orthogonal vectors qx, qy and qz which are analogous to the miller indices h, k and l with: 2sin(휃) sin(휃 − 휔) 푞 = 푥 휆푑 2sin(휃) cos(휃 − 휔) 푞 = 푧 휆푑 Diffraction of in-plane lattice points is possible by tilting the scattering plane away from the sample surface normal by changing the χ angle and the addition of an ω offset as well as changing the azimuthal angle φ. The in-plane vectors qx and qy are sometimes instead labelled q∥ for parallel to the surface, and the out of plane vector qz as q⊥ for perpendicular. To access o qy, the sample can be rotated about φ (e.g. 90 ), in which case the formula for qx can be used with qx = qy. Figure 21: Schematic of ω, χ and φ geometry for in-plane lattice points RSMs are taken by performing multiple scans with the sample tilt adjusted with each successive scan, allowing a region of the sectioned Ewald sphere to be mapped. 31 Figure 22: Schematic of how angular space maps reciprocal space on the Ewald sphere Analysis of peak positions and splitting gives information on crystal structure and symmetry and strain state. Maps around the symmetric axis can also yield information on tilted phases that would otherwise not be observed by a conventional coupled scan, seen for example in Figure 23. The peaks at positions qx ≠ 0 would not be seen in a 2θ/ω scan. Figure 23: Symmetric RSM of BFO on (001) LAO around the 001 reflection showing peaks tilted off the surface normal 2.2.3.1 Experimental Method Selection of sample optics and crystal alignment are identical to that of a coupled scan. The most commonly used triple axis settings are the 2θ-ω scan against ω offset for each line scan, and uncoupled 2θ scan against ω, different scan options will map a different shape in Q space and selection is based on minimisation of scanning empty Q space. Due to the 2-dimensional nature of an RSM, scanning times are naturally much longer than that of a coupled scan, in order to 32 minimise time spent scanning empty Q space, a low-resolution map of the desired peak area allows for more precise identification of the peak position and a high-resolution scan with refined range can then be performed. 33 2.2.4 Atomic Force Microscopy First developed in 1986 by Binning et al. [63] as a modification of the Scanning Tunnelling Microscope (STM), Atomic Force Microscopy (AFM) is a form for Scanning Probe Microscopy (SPM). SPM images are formed by rastering a physical probe across the sample surface, and the interactions between the probe and the surface form the basis of the image. By careful selection of probe properties and operation mode, many material surface properties can be visualised, such as topography, electrical conductivity, piezoresponse, and magnetism [64][65]. Modern probe tips dimensions are in the order of tens of nanometres, resulting in very small tip/sample interaction volumes, and resolution limits an order of magnitude lower than conventional (diffraction-limited) light microscopy. The use of a physical probe to achieve this resolution also means that the AFM can operate in atmospheric conditions and without the need for a conductive sample or additional sample preparation when compared electron microscopy techniques [65]. 2.2.4.1 Operating Principle The AFM consists of a cantilever with the sharp tip on one end (typically integrated), a support which holds the cantilever and has z-axis motion, often with a piezoelectric transducer for oscillating the tip, the detector, and the sample stage with x-y motion. A schematic of such a system is shown in Figure 24. The tip, mounted on a cantilever, is rastered across the surface of the sample and tip-surface interactions generate a deflection in the tip, which is then used to generate a 3D map of the surface. The deflection of the tip is commonly measured by beam deflection. A laser (with wavelength typically in the visible range) is reflected from the back of the probe, and onto a quadrant photodiode sensor. The intensity of the signal from the photodiode sensor in the vertical direction corresponds to the vertical deflection of the tip, while any shear forces which generate torsion in the cantilever result in displacement of the laser spot in the horizontal axis. The signal from the photodiodes can then be fed into a feedback system to control the tip positioning to maintain a constant force. 34 g) f) d) a) e) b) c) h) Figure 24: Schematic of a typical AFM system. a) tip and cantilever, commonly integrated, b) sample, c) sample stage with x,y motion, d) tip holder with z motion, e) piezoelectric transducer for applying tip oscillation, f) (laser diode) and g) (quadrant photodiode) for detecting tip position, h) feedback circuit 2.2.4.2 Topography measurement Topography can be measured through contact and non-contact ‘modes’. As the name suggests, contact mode has the tip in contact with the sample surface while the feedback loop ensures that the tip maintains contact with a constant force, the variation in z displacement forms the image. Due to tip wear, as well as potential sample surface damage (for soft samples), non-contact methods were developed. First demonstrated in 1987 by Martin et al. [66] dynamic AFM techniques involve inducing a vibration in the tip, typically with a piezo transducer. The interaction of the tip and the surface via long range forces such as van der Waals forces modify the resonant frequency of the cantilever and the change in resonant frequency is used both to control the feedback loop as well as to generate the image. This change in resonant frequency has a much higher sensitivity to forces on the tip than purely tip displacement. 2.2.4.3 Piezoresponse Force Microscopy The use of electrically conductive tips allows for simultaneous electrical measurements of conductive samples, as well as localised application of an electric field for investigating electronic properties. 35 Piezoresponse force microscopy (PFM) is a specialised mode of AFM that allows ferroelectric characterisation of a sample. Contact mode is used to apply a highly localised voltage through a conductive tip. The principle of PFM is based on the inverse piezoelectric effect, as shown in Figure 25. Since every ferroelectric is piezoelectric (as explained in Section 1.1) the application of a voltage bias to a ferroelectric thin film will induce an inverse piezoelectric response in the film and in turn induce a deflection to the cantilever. Use of a quadrant photodiode detector allows for detection of in-plane and out-of-plane polarisation. The response can be significantly improved by application of an AC voltage to the cantilever/film system. The film/cantilever system will have a resonant frequency, and this oscillation can be utilised for imaging like in dynamic AFM. Since the piezoelectric deformation is in the order of picometers, the application of lock-in techniques and signal processing allows for sensitivity high enough to measure this. Another functionality of the PFM setup is the possibility to electrically pole regions of the FE film by applying a constant bias to the AFM tip while the sample is grounded (or vice versa). If the applied electric field is sufficiently high (i.e. higher than the coercive field for the ferroelectric) then the polarization in well-defined regions can be switched from ‘UP’ to ‘DOWN’ and vice versa, depending on the polarity of the applied bias. 2.2.4.4 Conductive-tip AFM Finally, a further mode of AFM is called conductive-tip AFM, or C-AFM. In this mode, the conductive AFM tip is rastered across the surface in contact mode with a small bias applied to the tip, while the lower electrode of the sample is grounded. A high sensitivity current amplifier is used to detect the current flowing through the sample and allows the creation of a ‘resistance map’ as well as topography information. Using this technique allows studying of conductive domain walls as shown in Chapter 4 of this thesis. 36 a) Lock-in amplifier P Ferroelectric Electrode b) Lock-in amplifier P Ferroelectric Electrode Figure 25: PFM schematic diagram, application of a bias through the tip and a bottom electrode, piezoelectric response can be measured locally. a) out of plane polarisation, b) in plane polarisation. 37 3 Using defects to stabilize novel phases in T’ like BFO DECLARATION REGARDING PUBLISHED MATERIAL The content presented in this chapter has in part been previously published in the following journal article: “Designer defect stabilization of the super tetragonal phase in >70-nm- thick BiFeO3 films on LaAlO3 substrates” Daniel Sando, Thomas Young, Ralph Bulanadi, Xuan Cheng, Yanyu Zhou, Matthew Weyland, Paul Munroe and Nagarajan Valanoor Japanese Journal of Applied Physics, Volume 57, Number 9, 0902B2 (2018). Author contributions: Thomas Young grew the films, performed AFM, XRD measurements, analysed data, prepared figures and wrote the paper. Daniel Sando participated in film growth, XRD characterisation, data analysis and paper writing. Ralph Bulanadi participated in thin film growth. Xuan Cheng performed TEM characterisation and analysis. Yanyu Zhou performed TEM characterisation and analysis. Matthew Weyland supervised TEM characterisation and analysis. Paul Munroe supervised TEM characterisation and analysis. Nagarajan Valanoor supervised the study. All authors participated in manuscript preparation. Name: Thomas Young Signed Date 38 3.1 Introduction and Motivation As mentioned in Section 1.4, when BFO is grown epitaxially on (001) LaAlO3 (LAO) substrates (compressive strain of ~4.5 %), a metastable polymorph can be formed, which we call the T’ phase. Usually, under such epitaxial conditions, films thicker than ~25 nm undergo a strain relaxation whereby a complex mixture of polymorphs is formed. This mixed phase typically manifests as regions with a striped topography consisting of S’tilt and a T’ phase with an out of plane axis tilt of ~1.4° (hereafter referred to as T’tilt), with the proportion of mixed phase increasing with increasing film thickness until ~200 nm where pure R’ phase is observed [43]. The T’ phase has a MC monoclinic structure and is so named due to its tetragonal-like giant axial ratio (c/a ~1.23). The R’ phase has an MA monoclinic structure that is similar to the bulk rhombohedral structure. The S’tilt phase is a compressively strained R’ phase an MA monoclinic structure and a tilt along the out-of-plane axis of ~2.5-3°. The mixed phase has garnered intense research interest due to its enhanced piezo electric response [67] as well as its photovoltaic [68], electrochromic [69], and magnetic [45] properties. Over the course of this study, it was observed that films grown using the Neocera PLD chamber did not exhibit such mixed phase regions. This result was surprising, given the consensus in the literature showing that mixed-phase BFO appears for thicknesses above ~25 nm. Careful analysis was required to determine that this unexpected result was indeed ‘pure’ T’ phase, with XRD, SPM and TEM techniques being used to characterise the films. This chapter is devoted to understanding the mechanisms behind the formation of thick T’ BFO films up to 80 nm, without the formation of the typical mixed phase regions. By manipulating local chemistry, it becomes possible to unlock phases not available, just considering the macroscale mixed phase system. To systematically study the influence of PLD growth conditions on the stabilisation of the T’ phase, a thickness series of BFO on LAO was grown in two PLD systems using the same target and identical substrates. The first chamber, Pascal (PLD A) has a setup which can be considered ‘standard’ and thus similar to many other PLD chambers in other groups around the world. The second chamber, Neocera (PLD B) has a larger substrate-target distance, along with other slight variations compared with more ‘standard’ setups. Here we consider four films from each chamber for comparison, based on their similar thicknesses. In PLD B, pure T’ phase films with thicknesses up to ~70 nm were possible, while films grown in PLD A showed mixed phase at t > 30 nm, with the fraction of the mixed phase 39 increasing with thickness. As mentioned above, for T’ BFO films grown on LAO substrates the critical thickness before strain relaxation through the formation of mixed phase regions is typically ~30-35 nm [43]. Throughout this chapter, we use films growth in PLD A as the ‘control’ samples, as they exhibit the thickness-dependent mixed phase structure as shown almost exclusively in the literature. 3.2 Sample Growth The samples presented in this chapter were grown on the same batch of (001) oriented LAO, with pieces cleaved from the very same crystal being used in both chambers to minimise variation in substrate crystallinity and surface properties. The same Bi1.1FeO3 target was used for all films. Set A was prepared using a Pascal PLD system (since its parameters are similar to ‘typical’ PLD chambers), while set B was prepared using the Neocera system as outlined in Chapter 2. Both systems used a Coherent Compex Pro 102F KrF excimer laser with operating wavelength of = 248 nm. Growth parameters for each chamber are compiled in Table 3. Table 3: Growth Parameters for T’ BFO films grown on LaAlO3 (001). Parameter PLD A (Pascal) PLD B (Neocera) Temperature (°C) 590 700 Heater type Laser diode on SiC Coil heater on steel Oxygen pressure (mTorr) 100 23 Substrate-target distance 5 10.5 (cm) Laser fluence J/cm2) 2.2 4.1 Laser repetition rate (Hz) 5-10 5-15 Cooling rate (°C /min) 20 20 Cooling pressure (Torr) 5 450 Note that the most critical difference between the two chambers is the much larger working (substrate-target) distance for PLD B. This in turn requires lower operating pressure, which will be discussed in more detail later in the chapter. The apparent large difference in temperature is due to variation in how temperature is measured between chambers. In PLD A a laser pyrometer reads the temperature of the SiC mounting block, while PLD B has a K-type thermocouple near the heating coil. 40 The growth rate did show some variation between depositions and with increasing thickness in both chambers, with typical growth rates of ~200 pulses/nm in PLD A and ~400 pulses/nm in PLD B. As such, for films of comparable thickness, the growth time was doubled (an additional ~45 min for a 60 nm film) in PLD B. This observation is important as we will see later. 3.3 Surface and structural characterisation Atomic force microscopy (AFM) topography scans were performed with an Asylum Labs Cypher Scanning Probe Microscope with Budget sensors platinum-chromium coated tips in non-contact mode. Films grown in PLD A had a clear atomic-stepped topography for thinner films, but exhibited mixed-phase stripes for thicknesses above ~20 nm, with the fraction of mixed phase increasing with thickness. Films grown by PLD B typically showed a smooth surface with atomic steps, with no signs of the characteristic striped topography of mixed phase, even up to a thickness of 73 nm. Multiple areas were scanned to confirm that the pure T’ phase structure was consistent throughout the film and not just a localised region for the PLD B set. AFM is more sensitive to small volume fractions of mixed phase than XRD techniques due to the long count times required to detect the S’ phase [43]. (a) (b) (c) (d) 11 nm 29 nm 64 nm 94 nm PLD A (e) (f) (g) (h) 14 nm 31 nm 58 nm 73 nm PLD B Figure 26: AFM topography scans for PLD A and PLD B films. An increasing fraction of mixed phase is observed with increasing thickness in the PLD A films (from [70]). High angle XRD 2θ- coupled scans were measured for the films, the results of which are presented in Figure 27. For PLD B films, Laue (finite thickness) fringes are observed on the (001) and (002) film peaks, indicative of high crystallinity and low interface roughness. There is no 41 evidence of extra peaks that would correspond to S’ or R’ phases. Interestingly, the thicker films exhibit a broad peak around 31°, which we return to later. On the other hand, films grown by PLD B exhibit Laue fringes for the thinner films, while the thicker films do not show fringes, and extra peaks (likely corresponding to the S’ and R’ phases) appear. The absence of Laue fringes is understandable given the significant surface roughening that occurs with the formation of the mixed phase regions [43] [see for instance Figure 26 (d)]. (a) (b) PLD A PLD B LAO 001 LAO 002 LAO 001 LAO 002 LAO T-BFO 001 T-BFO 002 T-BFO 001 T-BFO 002 k k 11 nm 14 nm 29 nm 31 nm 48 nm 58 nm Intensity (arb. Intensity units) Intensity (arb. Intensity units) 94 nm 73 nm 15 20 25 30 35 40 45 50 15 20 25 30 35 40 45 50 2 (degree) 2 (degree) Figure 27: High angle 2ϴ/ω scans around the 001 and 002 reflections for BFO on (001) LAO grown in two different PLD chambers. To confirm the absence of the tilted S’ phases in the PLD B films, reciprocal space maps (RSMs) around the 001 reflection were carried out, see Figure 28. Using RSMs around a symmetric (i.e. 001) peak allows to search for phases that might have their crystallographic c-axis tilted away from the surface normal, and would thus be missed in 2θ/ω scans (a standard 2θ/ω scan traces a straight line through reciprocal space, i.e. in Figure 28 from Qz = 0.78 to Qz = 1.02, with Qx = 0). As can be seen in Figure 28, the RSMs of the thicker PLD A films exhibit extra peaks corresponding to S’tilt phases, as evidenced by the diffuse peaks appearing at Qx = ± 0.05 and Qy = 0.91 r.l.u. The ‘wings’ (labelled T’tilt) that appear beside the T’ BFO peak for the same PLD A films are attributed to the T’tilt phase which forms in conjunction with the S’tilt phase [43]. On the other hand, the PLD B films show no such extra peaks, even for the thickest film of 73 nm. These RSM results are thus consistent with the results of AFM topography scans; that is, films in PLD A show the typical thickness-dependent mixed-phase evolution, while for PLD B films, the films remain pure T’ phase up to a thickness of at least 73 nm. 42 Figure 28: Symmetric XRD reciprocal space maps (RSMs) near the 001 reflection for (a-d) films grown in PLD A; (e-h) films grown in PLD B, showing no tilted phases. Note that the thicker films from PLD A (g,h) show extra peaks corresponding to tilted S’ phases (from Ref. [70]) To try to understand why the films grown in PLD B showed very high crystallinity and an apparent lack of strain relaxation, even up to ~70 nm thicknesses, we used transmission electron microscopy (TEM) to inspect the nanoscale structure of the films. High angular annular dark- field (HAADF) scanning transmission electron microscopy (STEM) imaging on a 58 nm thick PLD B film was performed by Maggie Zhou (at UNSW) using a JEOL JEM ARM200F. Figure 29 shows the coherent interface with the LAO substrate as well as regions of half plane stacking defects [Figure 29(b)]. Energy Dispersive X-ray spectrometry (EDX) scans were performed to try to detect changes in stoichiometry (that would be present for example in the case of Bi2O3 nano- regions); see [Figure 29 (b) and (d)]. Note that one may infer a local slight increase in Bi intensity, and an apparent decrease in Fe intensity, the signal-to-noise ratio was too low to conclusively confirm the stoichiometry of the nano-regions. 43 (a) (b) (c) (d) (e) Figure 29: Energy dispersive X-ray (EDS) TEM analysis. (a) HRTEM image showing a defective region (circled). (b-e) show EDS maps at the same location for Bi, La, Fe, and Al respectively. Note that a slight increase in Bi brightness and corresponding decrease in Fe brightness may be surmised from these images; however conclusive evidence is lacking. a) b) c) BFO BFO 5nm 20 nm LAO LAO Figure 30: STEM HAADF image processed by the Richardson-Lucy method [71], [72] showing defective nano-region, b) enlarged image of the nano-region, c) S’ domain observed only in TEM samples. It is interesting to note the presence of regions that apparently correspond to S’ domains in the TEM sample [Figure 30 c)], which were not observed by XRD measurements in the as-grown sample. These two types of structures are commonly observed in ‘mixed phase’ samples [36]. Therefore the existence of both types of domains in our samples is completely at odds with the AFM and XRD measurements which showed an absence of any relaxed S’ or R’ phases. 44 We propose that the mixed phases in our TEM samples arise from the sample preparation process itself. Since mechanical stress can nucleate the mixed phase in the BFO//LAO system [73], we attribute the existence of the mixed phase in our TEM sample to the mechanical stresses imparted on the film/crystal heterostructure during TEM sample preparation. Also important to recall is that the typical TEM specimen thickness is in the order of 30 nm, thus the in-plane strain can no longer be considered the same as in the bulk sample, changing the strain state of the crystal. The measured thickness from the TEM analysis being thinner than that coming from XRD analysis may be attributed to a localised thinner region, and strain relaxation. Based on this work, it appears that the maximum attainable thickness for the pure T’ phase films grown in PLD B is ~80 nm. For thicknesses above these values, the films were observed to ‘break down’ directly to an R’ like derivative of BFO. Indeed, XRD measurements of thicker films with a rough topography showed the existence of extra peaks corresponding to the bulk rhombohedral phase of BFO (Figure 31). Further, for such a film, the reflection high energy electron diffraction (RHEED) specular spot intensity [Figure 32(a)] (which gives a measure during growth, of the surface quality and crystalline structure), evidenced a drop in intensity after ~30,000 pulses (or ~70 nm thickness), which would be consistent with a loss of the layer-by-layer growth mode. AFM images taken ex-situ [Figure 32(b)] show that the surface topography is completely modified relative to the pure T’ films, consistent with the formation of a film comprised of phase separated Bi2O3, Fe2O3 and/or the bulk BFO phase. 109 35k BFO // LAO (001) 107 LAO 002 LAO LAO 001 LAO 105 021 3 O 2 3 ? BiOx 10 -Bi BFO110 BFO012 BFO024 BFO202 Intensity (a.u.) Intensity 101 16 20 24 28 32 36 40 44 48 52 2 (deg) Figure 31: Coupled 2θ/ω scan of a BFO film after the ‘breakdown’ to a rhombohedral phase has occurred. Some peaks appear to correspond to various orientations of rhombohedral BFO, while bismuth oxide phases are also found. 45 Figure 32: a) RHEED specular intensity during growth of a PLD B film, showing steep drop in intensity around 30,000 pulses (or thickness of ~75 nm). b) AFM topography scan on the same film showing no evidence of mixed-phase stripes, nor stepped topography; instead an aligned needle-like structure is observed. 46 3.4 Possible origins for stabilisation of T’ phase to ~80 nm In this study, the films grown in PLD A exhibit structure similar to those commonly reported in the literature [43], [45], [67]. In comparison, for PLD B films the thickness to which the T’ phase is stabilised highlights the strong impact of growth conditions on the structure of the resultant films. We suggest that the half-plane stacking-fault-like defects observed in TEM studies are tetragonal β-Bi2O3 corresponding to the broad peak at ~31° in XRD. Indeed the 002 reflection of β-Bi2O3, if this was strained on LAO, would appear at 2θ ~ 30°. These defects could act as a strain- preserving mechanism, thus increasing the critical thickness before which mixed phase nucleation occurs. If indeed the nano-regions are Bi2O3, then this suggests that for films grown in PLD B, the film stoichiometry is different, i.e. the films contain a slight bismuth excess. If we now focus on the growth process in the PLD B films, we first consider the influence of conditions on the film stoichiometry. Film stoichiometry is controlled by three main factors: a) the stoichiometry of the plume as it leaves the target; b) gas phase interactions that prevent all of the ablated species reaching the substrate; and c) re-evaporation of the film species from the substrate [52], [55]. Next, the differences in deposition parameters are examined, and their potential influence on film stoichiometry is considered. 3.4.1 Laser Fluence and Energy Density The stoichiometry of the ablated species is controlled by the target composition and laser conditions. The same Bi1.1FeO3 target was used across both chambers, and both lasers are KrF excimer lasers with a wavelength of 248 nm. The laser energy density of PLD B is estimated to be double that of PLD A, since in the latter the laser is slightly defocused. During the laser pulse the irradiated region is heated to temperatures in the range of 5000 K within a few nanoseconds [53], [74] resulting in the production of a plasma with composition matching that of the target. For a focused pulse, the laser penetration is deeper resulting in the formation of a crater with a higher vapor density than for a defocused pulse. The density of the evaporated phase greatly affects the amount of laser energy absorbed, and the resultant energy distribution in the plume, changing the interactions in the gas phase [52], [75]. Secondary thermal evaporation is also a consideration as the target surface can stay above the evaporation temperature of different species after the laser pulse has ended [75]–[77]. The influence of secondary evaporation on film stoichiometry is dependent on many factors; laser operation affecting the degree of heating, target thermal conductivity, growth duration, and vapour pressure of the constituent species resulting in incongruent evaporation [52], [53]. 47 3.4.2 Deposition oxygen pressure Gas phase interactions are dependent on the mean free path before collision. These interactions are controlled by two factors: the deposition gas pressure, and the substrate-target distance. Gas phase collisions result in the formation of larger oxide molecules within the plume, which can aid the adhesion of oxygen at the substrate surface, where the sticking coefficient of atomic or molecular oxygen is orders of magnitude lower than that of metal atoms [55]. Gas phase collisions also average out the plume kinetic energy by the time it reaches the substrate, with overly high plume kinetic energy resulting in the formation of defects [54], [78]. Re-evaporation of film species from the substrate is also affected by the deposition pressure, due to the effect of the kinetic energy of the plume on the sticking coefficient. 3.4.3 Substrate-target distance Substrate-target distance is a rarely reported parameter for PLD growth. We believe however that the PLD A system is more representative of what is commonly employed in practice. The larger substrate-target distance in PLD B can affect the film stoichiometry in multiple ways. The longer distance to travel increases the likelihood of gas phase collisions, requiring the background pressure to be lowered in order to compensate. This in turn changes sticking coefficients of the different film forming species at the substrate surface [52], [53], [55]. 3.4.4 Temperature difference Crystal quality is highly dependent on substrate temperature due to surface diffusion [53], which determines growth mode. Higher substrate temperature also results in higher re-evaporation of atoms from the film surface, however in this case we do not believe this to be a major factor as the apparently large disparity in substrate temperature is in part due to the placement of the thermocouple located on the coil heater in PLD B, while a laser pyrometer is aimed at the back of the SiC substrate mounting plate in PLD A. As a result the expected temperature at the substrate surface for PLD B is ~100°C lower. Furthermore, if the films grown in PLD B were at a higher temperature they would be expected to have a bismuth deficiency due to the higher bismuth vapour pressure and the longer time spent at elevated temperature compared to PLD A films. This discussion highlights that indeed, in PLD B the primary factor that influences film stoichiometry is the larger substrate-target distance. This in turn requires a lower growth pressure (e.g. 23 mTorr vs 100 mTorr in PLD A). According to the phase diagram for BiFeO3 thin films in Figure 33 [59], a lower growth pressure will in turn require a lower growth temperature, 48 to be able to obtain the BFO phase. This implies that although the perceived temperature in PLD B is rather high, it is in fact possible that the temperature is lower than that of PLD A. Such a lower growth temperature will clearly favour a higher Bi stoichiometry in such growth conditions. Figure 33: Temperature-pressure phase diagram for the growth of epitaxial BiFeO3 thin films, from Bea et al. [59] Interestingly, if we consider the literature, it appears that Bi2O3 and T’ BFO are rather chemically compatible. Bi2O3 has an in-plane lattice constant of 7.741Å [79] which would result in a compressive strain of 2.1% when constrained across 2 unit cells of LAO with its 3.791Å in-plane lattice constant, suggesting that a Bi2O3 nano-pocket would impart compressive strain on a T’ BFO matrix. It is suggested in Ref. [43] that an excess of bismuth is required for stabilization of T’ phase – this is apparent for some groups as a requirement to lower the growth temperature for the T’ phase as compared to the R’ phase. Further, there have been some reports showing that T’ BFO can be grown on STO substrates using Bi2O3 as a buffer layer [80]. Such a result is surprising since the compressive strain nominally imparted by STO is only 1.5 %, too low for the 4.5 % normally-required for the T’ phase of BFO. In that study, the Bi2O3 layer was used as a ‘template’, in effect changing the lattice parameter of the substrate and thus promoting T’ BFO growth; this is shown schematically in Figure 34. That report, and the present study thus both highlight that T’ BFO and Bi2O3 are both chemically and structurally compatible. 49 Figure 34: Concept of epitaxy of Bi2O3 on SRO//STO as a buffer layer for T’ BFO films on top. From Ref. [80]. 50 3.5 Inducing the mixed phase by intentional defects Now that we have proposed the mechanism behind the stabilisation of ultra-thick T’ BFO films using stoichiometry to induce nano-scale bismuth rich regions that resemble stacking faults. It is interesting to consider if through extrinsic effects, films in PLD B can exhibit, in particular conditions, the mixed phase regions. Recall that the formation of the mixed phase is attributed to a peculiar strain relaxation mechanism whereby more stable R’-derived phases are formed. Presumably the location for the nucleation of such regions is at defects, and in fact occurs during the cool down from high temperature during growth [81]. It is therefore interesting to consider if intentionally placed ‘defect seeds’ can indeed induce the mixed phase topography in PLD B films, shown schematically in Figure 35. To test this prediction, films were grown in PLD B with a pre-growth step whereby the substrate was heated to a temperature 200 °C lower than the growth temperature, i.e. 500 °C. At this temperature, a small number of pulses were applied to the target (just as would occur for growth); however at the lower temperature the film is unlikely to crystallize. Instead the atoms arriving at the surface would simply form a defective surface. Subsequent to the pre-growth step, the substrate was heated to the growth temperature and a standard T’ BFO growth was performed, for a film thickness of approximately 50 nm. AFM topography scans show that indeed the use of such ‘defect seeds’ has had the desired effect: the film surface shows evidence of mixed-phase regions, as shown in Figure 36 a). Note that the fraction of mixed phase is rather low when compared to films grown in the more conventional PLD A at the same thickness. This can be attributed to the two competing effects: the first being the formation of the (possibly Bi2O3) nano-regions which preserve strain coherence to high thicknesses, and the second being the defect seeds. From the preliminary observations it seems that the nano-pockets have a stronger influence on the stabilization of the T’ phase, as evidenced by the very low fraction of the mixed phase regions. Also note that the broad peak at 2θ ~30° in the 2θ/ω scan is also observed for this film [Figure 36 b)] To gain more information about the competing mechanisms in the film growth with such ‘defect seeds’, HAADF-STEM imaging was performed by Heidi Cheng at Monash University. The results are shown in Figure 36. Note that the presence of defective nano-regions is observed throughout the film thickness. Remarkably, the film is observed to be coherent with the substrate, despite the fact that at the interface appears a region that is almost completely amorphous [Figure 36 d)]. This amorphous region is induced by the defect seeds during growth. 51 Note that in this TEM specimen we do not observe the mixed T’ and S’ regions; this may be attributed to the fact that the specimen was prepared by FIB rather than by the traditional polishing techniques. Further work is required to determine the clear causality of the relaxation of the T’ phase to the R’ derived phases, during TEM specimen preparation. No defect seeds BFO LAO LAO Mixed phase striations Defect Seeds With defect seeds BFO LAO LAO LAO Film growth Figure 35: Schematic diagram of 'defect seeds' 52 Figure 36: Formation of the mixed-phase regions in BFO//LAO films, using defect seeds. (a) AFM topography showing atomic steps and also existence of characteristic ‘striped’ regions. (b) XRD 2θ/ω scan, inset comparing with a similar film with no defect seeding, (c) STEM HAADF image showing the defect nano-pockets (inset shows enlarged view of a defect), (d) enlarged view of the amorphous interface region. (From Ref. [70]) 53 3.6 Discussion and Conclusion This chapter has described a method for stabilising pure T’ like BFO films on LAO substrates up to thicknesses of 70 nm. Such films continue to be routinely obtained in the laboratory, even with LSMO underlayers. Detailed XRD, TEM and AFM studies have shown that the strain coherence required to stabilise the T’ phase up to such high thicknesses is provided by nano-regions throughout the film. These defective nano- regions may be comprised of a secondary Bi2O3 phase. It was not possible, due to the very low volume, to obtain clear evidence of stoichiometry changes in the nano-regions using EDS or EELs-based analyses. However, if it is indeed Bi2O3, this phase would be tetragonal and therefore can apply a compressive stress to the film. This can preserve strain coherence up to the very high thicknesses as observed here. The addition of ‘defect seeds’ deposited onto the substrate before growth of the film is proposed as a method of inducing the mixed phase, with the ‘defect seeds’ acting as a nucleation point for the mixed phase. The scope of further possibilities of such films is rather wide. Since traditionally it has been very difficult, if not impossible, to stabilize the pure T’ phase to such high thicknesses, the present study allows to open new doors. A few examples are: i) measuring the true FE polarization of the T’ phase (is it indeed giant at 150 µC/cm2 as postulated?), ii) studies of optical switching of ferroelectric thin films, iii) incorporation of the thick T phase samples with other interesting systems such as graphene (for gating the transport characteristics), and iv) studies of magnetic structure using neutron diffraction, where ultrathin films are usually not possible for such studies. 54 4 Cobalt Doping of BFO thin films DECLARATION REGARDING PUBLISHED MATERIAL The content presented in this chapter has in part been previously published in the following journal article: “Structural, magnetic, and ferroelectric properties of T-like cobalt-doped BiFeO3 thin films” Thomas Young, Pankaj Sharma, DoHyung Kim, Thai Duy Ha, Jenh-Yih Juang, Ying-Hao Chu, Jan Seidel, Nagarajan Valanoor, Shintaro Yasui, Mitsuru Itoh, and Daniel Sando APL Materials 6, 026102 (2018). Author contributions: Thomas Young grew the films, performed AFM, XRD measurements, analysed data, prepared figures and wrote the paper. Pankaj Sharma performed AFM, PFM characterisation and analysed data. DoHyung Kim performed SPM characterisation. Thai Duy Ha performed XAS and XMCD characterisation. Jenh-Yih Juang performed XAS and XMCD characterisation. Ying-Hao Chu supervised XAS and XMCD characterisation. Jan Seidel supervised AFM and PFM characterisation and analysis. Nagarajan Valanoor supervised the study. Shintaro Yasui performed XRD and RSM characterisation and data analysis. Mitsuru Itoh supervised XRD and RSM characterisation. Daniel Sando participated in XRD measurement, analysed data, prepared figures, participated in paper writing, and supervised the study. All authors participated in manuscript preparation. Name: Thomas Young Signed Date 55 4.1 Introduction and Motivation The low value of the ferromagnetic moment in BFO has been a problem for the direct implementation of BFO as a multiferroic. It has been reported that substitution at the B-site with transition metals such as cobalt may improve the magnetic properties of BFO by altering the spin canting [82], [83]. Further, the introduction of cobalt into the BFO structure has been predicted to induce discrete high and low spin states [84] and through this mechanism increase the magnitude of the weak magnetic moment. Also, since the end member BiCoO3 is tetragonal [85]–[87], substitution of cobalt in to the BiFeO3 structure has been proposed as a mechanism to improve the piezoelectric response of BFO through a morphotropic phase boundary (MPB) between rhombohedral (BFO) and tetragonal (BCO) structures. In this chapter, we explore the influence of cobalt doping on the structural, ferroelectric, magnetic response of T’ like BFO films. It is found that 2 % of cobalt doping has a very weak influence of the aforementioned properties with a very low magnetic moment measured. This chapter addresses the structural and ferroic characteristics of 2% cobalt-doped T’ like Co- BFO films. A thickness series of Co-BFO films on LAO was prepared with and without conductive oxide bottom electrodes. Structural and surface characterisation was carried out with XRD, SPM and TEM techniques. Ferroelectric response, as well as magnetic characteristics was investigated using magnetometry, PFM and XMCD techniques. Properties applicable to potential nanoscale device applications were also investigated. 56 4.2 Sample Growth Epitaxial films of BiFe0.98Co0.02O3 (BFCO) were grown on 001 oriented LAO using the Neocera PLD system detailed in Chapter 2. To facilitate electrical measurements, a 2 nm thick La0.67Sr0.33MnO3 (LSMO) bottom electrode was grown at a temperature of 800°C, oxygen partial pressure of 100 mTorr, using a KrF excimer laser (wavelength 248 nm, fluence ~3 J/cm2) at a repetition rate of 5 Hz with an estimated deposition rate of 0.01 Å per pulse. The BFCO was grown at a substrate temperature of 650°C, oxygen partial pressure of 15 mTorr, 4.1 J/cm2 at 3-7 Hz using a 2% Cobalt doped target with 10% excess bismuth, resulting in a deposition rate of ~0.015 Å per pulse. Conductive oxide buffer layers are commonly employed to provide a bottom electrode for electrical testing and are critical for device design. Bottom electrode buffer layers have a strong impact on strain relaxation and can thus be used to tune the strain on the film. Strontium ruthenate SrRuO3 (SRO) and lanthanum strontium manganate La0.67Sr0.33MnO3 (LSMO) are common choices for their conductivity and compatibility with BFO and common commercially available substrates. By growing multiple layers of different materials it is possible to create functional heteroepitaxial structures for device applications. In the following, thin lower electrode layers of LSMO were grown on LAO. LSMO was chosen for its structural compatibility with the LAO substrate (thus allowing T’ Co-BFO to be grown on top). In addition, it was found that ultrathin LSMO layers of ~2 nm, allowed high quality pure T’ Co-BFO films to subsequently be grown. 57 4.3 Structure and surface properties of Co-BFO films on LAO (001) A series of films with various thicknesses was prepared. The films presented have all shown a stepped topography and T’ phase structure. Examples of the AFM topography are shown in Figure 37. a) b) c) Without LSMO d) e) f) With LSMO Increasing thickness Figure 37: AFM scans of Cobalt BFO films of various thicknesses on 001 LAO; a) 16k, b) 40k and c) 80k pulses of CoBFO. On 001 LAO with a ~2nm thick LSMO bottom electrode d) 16k, b) 40k and c) 80k pulses of CoBFO. All scans 3um x 3um, k denotes 1000 pulses. The data in Figure 37 show that even with a lower LSMO electrode, the high crystalline and surface quality of the T’ like BFO films are preserved. On some samples triangular outgrowths with a stepped atomically smooth structure were observed, see Figure 38. It is important to note that these outgrowths were not observed in films grown using undoped BFO targets, suggesting that cobalt plays some role in their formation. There was no clear influence of growth parameters on the formation of these outgrowths. 58 Figure 38: 1x1 µm AFM topography scan of a triangular outgrowth showing stepped structure, height scale is 4.81 nm. Possible explanations are a change in growth mode from step flow to Stranski-Krastanov growth, triggered by a change in adatom mobility of cobalt or the slight experimental variations in deposition conditions. Target density variation may go towards explaining the variability of the presence of the outgrowths, with regions of lower density in the target resulting in the formation of particles during ablation that act as nucleation points for island growth when they land on the sample surface. The thickest films (~70 nm) exhibited secondary phase particles and voids, TEM energy- dispersive X-ray spectrometry (EDS) analysis revealed that the secondary phases were iron rich (Figure 39). AFM and XRD analysis showed no mixed phase structures present. The presence of iron oxides can be explained by the greater than 3 hours growth duration, it is likely that the extended time at the growth temperature resulted in the re-evaporation of bismuth from the film surface generating iron rich regions [88]. Note that although AFM and TEM analyses evidence the presence of a secondary FeOx phase, XRD does not. This could be due to the very low volume fraction of the Fe-rich phase, thus not being detectable in standard scans. 59 a) b) T’ BFO 001 LAO LAO 001 T’ BFO 002 LAO LAO 002 β pc - pc pc Bi pc 2 O 3 002 pc c) d) e) f) Figure 39: a) AFM scan showing stepped topography with voids and secondary phase particles. b) XRD 2theta-omega scan showing the T’ phase film peaks and no R’ or S’ peaks. c) - f) TEM and EDS scans showing that the secondary particles are iron rich. High resolution 2θ- coupled scans (Figure 40) indicate a pure T’ phase structure with Laue fringes indicating high crystallinity and coherent interfaces with a very slight increase in c-axis spacing (4.66Å) as compared with undoped samples (cf. Chapter 3). Of note is the presence of the broad peak at 2θ ≈ 30°. Recall from Chapter 3 that the nano-pockets that are suggested to preserve strain coherence and stabilize the T’ like phase up to high thicknesses. It thus appears that such nano-pockets are also present in the Co-BFO films. Such an observation suggests that, as expected, it is indeed the growth parameters in PLD B that result in the formation of the nano- pockets and it is transferable to different BFO targets. The presence of a 20 nm thick LSMO bottom electrode destroyed the Laue fringes of the CoBFO layer and interestingly the broad peak at 2θ ≈ 30° [Figure 40 b)]. The reason for this remains unknown. 60 Co-BFO // LAO (001) 55 nm Co-BFO / LSMO // LAO (001) a) 109 b) 109 107 107 BFO001 BFO002 LAO 001 LAO LAO 002 LAO BFO002 BFO001 LAO 001 LAO LSMO 002 LSMO LAO 002 LAO 105 105 001 LSMO 20 nm 103 40k 103 LSMO BiOx ? BiOx ? Intensity (a.u.) Intensity 16k (a.u.) Intensity 2 nm LSMO 101 101 20 30 40 50 20 30 40 50 2 (deg) 2 (deg) Figure 40: High angle XRD scans around the 001 and 002 reflections of Co doped BFO on 001 LAO substrates a) without a bottom electrode, blue 40k pulses of CoBFO, red 16k pulses of CoBFO, b) 40k pulses of CoBFO, blue with a ~20nm thick LSMO bottom electrode, red with a 2nm thick LSMO bottom electrode. To ascertain the structural symmetry, reciprocal space maps (RSMs) were acquired around the 001, 103, and 113 pseudocubic reflections (Figure 41 a), b) and c) respectively) [89], [90]. The maps around the (103) and (113) reflections showed splitting of the film peaks is consistent with an MC type monoclinic structure expected of T’-phase BFO [91]. To search for the R’, S’ and T’tilt phases commonly observed in BFO films on LAO substrates greater than ~35 nm [43] (as discussed in Chapter 3), an RSM around the symmetric 001 reflection was performed. As no extra peaks other than the substrate and BFO peak are detected, we conclude that there are no tilted phases in this film. a) b) c) Figure 41: Reciprocal space maps of a ~56nm BiFe0.98Co0.02O3//LaAlO3 film. a) 001 RSM, red circles denote expected position of tilted phase peaks. b),c) 103 and 113 RSMs respectively (from Young et al. 2018 [92]). 61 There XRD-based analyses thus suggests that the triangular outgrowths observed in the AFM topography are formed of T’ phase CoBFO and not a secondary phase, or a tilted phase. 62 4.4 Ferroelectric properties Piezoresponse Force Microscopy (PFM) analysis was carried out on an Asylum Labs Cypher by Pankaj Sharma and Peggy Zhang (UNSW). Scans indicate that the as-grown polarization direction is down [Figure 42 a)], while lateral PFM scans show a stripe-like domain structure with domain walls oriented along the (110) plane [Figure 42 b)], consistent with the MC symmetry [93], [94]. Local PFM hysteresis measurements carried out between the conductive AFM tip and the lower LSMO electrode, show a coercive field of 2Ec = 12 V [Figure 42 c)]. The clear ‘butterfly’ shape of the amplitude curve [Figure 42 c), bottom], and the ~180° switching in the phase curve [Figure 42 c), top] attest to the ferroelectric character of this film. a) b) c) 1.0 20 0.8 20 0.6 10 µm 180 0 0.4 Deg -10 0.2 -20 0.0 -20 0.0 0.2 0.4 0.6 0.8 1.0 -180 µm Figure 42: PFM scan a) showing out of plane polarization direction (inset topography), b) in plane polarization, c) local PFM hysteresis loops. a) b) c) Figure 43: PFM box in box switching, a) Topography, b) amplitude and c) phase scans of a 2.5x2.5µm region Further evidence for ferroelectricity is presented in Figure 43 by PFM box in box switching. Ferroelectric switching begins to induce the mixed phase regions as evidenced by the topography change. Strangely, the PFM amplitude in the converted regions is lower than the regions surrounding it. This could be related to the fact that the mixed phase regions have a lot of phase boundaries. This is rather strange since in normal mixed phase regions the piezoresponse is actually enhanced, possibly due to domain wall movement. The origin of the 63 contrary response here is unknown. A more controlled way to convert between the R’/T’ and T’ phases is shown later in this chapter. The PFM images in Figure 43 also show that the triangular outgrowth regions (that rather surprisingly preserve the stepped topography) remain ferroelectric, and therefore are most likely still the BFO phase. This is consistent with XRD analysis presented earlier. Figure 44: (a) AFM scan showing induced topography changes from a 12V applied bias at different time scales. Corresponding PFM amplitude and phase scans showing the induced ferroelectric switching (b) and (c) respectively. (d) Line profiles taken along the dashed lines. (e) Induced ferroelectric domain size with increasing pulse duration at 12V (black), 15V (blue) and 18V (red), logarithmic fit for the eye, pulse shape inset. Time dependent domain nucleation was carried out by Pankaj Sharma (at UNSW) on an Asylum Research Cypher, application of a set bias at different pulse widths via the PFM tip. Analysis of the resulting topography, PFM amplitude and PFM phase scans showed that in the -12V DC applied bias case, below 50 ms only mixed phase transformation occurs as evidenced by the saw tooth pattern induced in the topography and the nucleation of the mixed phase before onset of ferroelectric switching. After ferreoelectric switching begins domain size increased with increasing pulse duration in a logarithmic relation with full reversal of phase from 90° to -90° at 64 0.8 s. At higher applied biases ferroelectric switching was always observed in the time scales measured. For biases larger than -15 V and pulse widths larger than 50 ms a secondary annular topography developed directly beneath the PFM tip. The domain switching dynamics presented here are rather similar to that observed for other ferroelectric thin films [95]. Attempts to measure the macroscopic ferroelectric polarization of T’ Co-BFO films are shown in Figure 45. Platinum top electrodes were sputter deposited to make a Pt/Co-BFO/LSMO//LAO capacitor structure (Co-BFO 40k, LSMO 4.5k). I-V and P-V curves show a very large leakage current, meaning that reliable hysteresis loops were not possible and thus showed ‘leaky’ loops as seen in Figure 45 b). Note that this is a common problem for T’ BFO films [37]. While P-E loops have been reported on mixed T/R BFO films [37], and on Mn-doped BFO films [35], in the present work it is presumed that the leakage current is much too high to obtain clean loops. It is also possible that the thin lower electrode also caused complications in the measurements process. (a) (b) Figure 45: Macro scale electronic properties of a Pt/Co-BFO/LSMO//LAO capacitor structure (Co- BFO 40k, LSMO 4.5k) a) Current - Voltage (I-V) response, b) Polarization – Voltage (P-V) response. 65 4.5 Functional response: electric-field interconversion between the phases and conductive domain walls In this section, we present some evidence for functionality of these Co-BFO films. These features may be attractive for nanoscale devices. AFM studies of the as-grown surface of a 55 nm Co- BFO/LAO film [Figure 46 a)] show a smooth (RMS ~0.25 nm) stepped structure, with line trace (Figure 46 d) black) indicating the steps are unit cell height. Application of a +6V DC bias via the tip along a line scan resulted in development of a region of striped contrast characteristic of mixed phase, with the line trace confirming the characteristic saw tooth pattern and a 1-1.5 nm change in height. Application of a -7V DC bias to the same region resulted in a reversion to an almost identical state as the as-grown surface [Figure 46 c)]. The profiles have been offset vertically for clarity. This completely reversible phase transformation makes the material a promising candidate for nanoscale devices, utilising the highly localised changes in optical properties [69], magnetism [45] and piezoreponse [96]. a) d) 15 12 As grown 300 nm b) 9 6 After writing (nm) height c) 3 After erasure 0 0.0 0.2 0.4 0.6 0.8 1.0 distance (µm) Figure 46: AFM scans of the a) as-grown, b) mixed phase induced and c) after erasure of mixed phase topographies. d) line profiles of the surface along the indicated line showing the indicative saw tooth pattern of the mixed phase and the return to the original stepped surface upon reversal of bias. A further attractive functionality of such films is demonstrated in the conductivity of the domain walls as shown in Figure 47. By locally writing regions of reversed polarisation with the PFM tip, 66 domain walls can be controllably created and erased. A greatly enhanced conductivity compared to the surrounding domains was observed in these domain walls using conductive AFM. Current flow onset was observed at ~3V of applied bias, with excellent stability of domain wall current. Such an observation makes these films good candidates for domain wall electronics [10]. (a) (b) PFM amplitude C-AFM a.u. pA 1.0 0 0.3 μm 0 -30 (c) (d) 0 Stability of domain wall 0 -5 current -10 Domain Domain Wall -15 -50 Current(pA) Current(pA) -20 At -3.0 V -25 -100 -6 -3 0 3 6 0 50 100 150 200 Bias (V) time (s) Figure 47: (a) PFM scan of a region with an artificially generated domain, (b) conductive AFM scan of the same region (DC bias -2.8V), showing the enhanced conductivity at the domain walls, (c) localised current-bias on the domain wall (blue) and off the domain wall (red), (d) localised current on the domain wall over 200s (DC bias -3V) (from Young et al. 2018 [92]) 4.6 Magnetic properties A representative 55 nm thick Co-BFO// LAO (001) film was sent for magnetic characterisation at the lab of Y-H Chu, Taiwan. Magnetometry and XMCD measurements were performed. The summary of results is shown in Figure 48. First, the magnetic moment measured by SQUID is very low, as shown in Figure 48 (a), thus confirming that the film contains nominally zero parasitic iron oxide type phases. Note that this is consistent with the formation of Bi2O3 nano- regions since such a sample would be formed in bismuth rich conditions. 67 XMCD shows that the magnetic moment is negligible (within detection limits), once again evidencing a very low density of iron based parasitic phases. Finally, XMCD on the Co L edges shows a signature of cobalt in these films, as expected for a small 2 % doping level. In summary, these magnetic measurements are consistent with high quality, T’ like BFO films with a small amount of cobalt substitution. Figure 48: (a) M(H) loop indicating low saturation moment of ~4 emu/cm3. XAS and XMCD spectra for Fe (b) and Co (c) showing the L2,3 electron edges and no signal in XMCD. Magnetic hysteresis measurements, X-ray absorption spectroscopy (XAS) and X-ray magnetic circular dichroism (XMCD) taken on a similar film grown without LSMO buffer layer shows a low magnetic moment confirming the absence of parasitic iron oxide phases. This result is consistent with previous studies [35] indicating that the magnetic moment is enhanced in the R’-phase and not in the T’-phase. To see if the magnetic moment in cobalt doped films was enhanced in the mixed phase, Magneto Force Microscopy (MFM) techniques were applied. Regions of mixed phase stripes were first nucleated via electrical bias with the SPM tip, and then the sample scanned in MFM mode. No magnetic contrast was detected. This is consistent with the weak ferromagnetism expected of BFO being below the detection limit of MFM. 4.7 Discussion, conclusions, and further work In this chapter we have presented a detailed study of structural, symmetry, ferroelectric and magnetic properties of Co-BFO films on LAO substrates. Growth conditions and structural characteristics are comparable to that of undoped BFO samples. Crystal structure was tetragonal like T’ phase, Mc monoclinic with an elongated c-axis of 4.66Å. Triangular outgrowths were observed with a stepped surface, XRD and PFM characterisation suggests that these outgrowths are T’ phase like the rest of the sample. 68 The electronic properties of 2% Co doped BFO is unchanged compared to undoped BFO, with the mixed phase structure capable of being induced via application of electrical bias. Erasure of the mixed phase with reversed bias was possible, with a very clean return to the stepped surface with little to no traces of the S’ phase was observed. Further studies with higher percentage Co substitution are required to conclusively determine if the addition of cobalt is stabilising the tetragonal structure. Due to the low percentage of cobalt doping it is unclear if the cobalt addition had any effect on the magnetic properties of BFO. Due to the reduced spin canting in the T’ phase structure the ferromagnetic moment is removed, while the pure T’ phase structure of the films makes it unclear if the ferromagnetic moment was increased in the mixed phase due to the difficulty in measuring the small magnetic moment of such a small fraction of the film. Cobalt doping at 2% does not appear to have adversely affected the ferroelectric properties of BFO, however it has also had a negligible effect on the magnetic properties. Further work with increased cobalt doping is required to conclusively determine the effect of cobalt on the magnetic properties of BFO. The almost completely perfectly reversible switching of localised polarization and phase offers perspectives for nanoscale devices in which modulation of electronic, magnetic[97], elastic[98] and optical [99] properties can be utilised. 69 5 Conclusions and Perspectives 5.1 Conclusions In this work, films of multiferroic BiFeO3 have been grown by pulsed laser deposition (PLD) and characterised by various techniques. The focus was on the tetragonal-like (T’) structure of BFO commonly obtained when the films are grown epitaxially on LaAlO3 (001) substrates. The thesis has focused on two main aspects: the influence of growth conditions on the stability of the T’ phase of BFO, particularly for very high thicknesses; and the influence of cobalt doping on the physical characteristics of T’ BFO films. X-ray diffraction (2θ/ω scans, reciprocal space mapping), x-ray reflectometry (XRR), atomic force microscopy (AFM), piezoresponse force microscopy (PFM), conductive AFM (C-AFM), and macroscopic ferroelectric measurements were all used extensively throughout this work. The first part of the study showed that using specific growth conditions afforded by notably a much larger-than-normal substrate-target distance during PLD growth, allowed the stabilisation of the T’ phase of BFO in thicknesses up to 73 nm. This observation was rather surprising since typically in this system, films above ~25 nm in thickness undergo strain relaxation and then comprise of ‘mixed-phase’ regions of T’ and R’ BFO. Detailed X-ray diffraction and transmission electron microscopy characterisation showed that nano-pockets of ‘stacking-fault’ structure are likely the reason for strain preservation in such films. Attempts to measure local changes in stoichiometry using EDS were unsuccessful, most likely due to the nominal change being beyond the detection limit of such techniques. However, the out of plane lattice spacing of the nano- pockets is consistent with the structure of Bi2O3. It was therefore suggested that the lower growth temperature induced a slight increase in bismuth content, therefore making possible the formation of such bismuth-rich nano-pockets. Finally, it was shown that by using ‘defect seeds’ at the substrate-film interface, the mixed-phase of BFO can be induced; albeit with a very small phase fraction. In the next part, thin films of 2 % cobalt doped BFO, grown by PLD on LAO were studied. The structure of such films was found to be comparable to undoped films, with the out of plane lattice constant the same, the symmetry being of monoclinic MC (as commonly reported in literature), and a lack of the mixed phase regions, just as was observed in Chapter 2. This in particular showed that indeed the growth conditions are responsible for the preservation of the T’ phase up to high thicknesses, since it was shown with different targets in the same chamber. 70 Ferroelectric characterisation by local techniques (PFM) showed a clear ferroelectric behaviour, with well-aligned stripe domains in the as-grown state. Attempts to measure macroscopic P-E hysteresis loops were unsuccessful, due to the high leakage current that is often inherent to this phase of BFO. The functional response of the T’ Co-BFO films was demonstrated by the electric-field induced reversible interconversion between the T’ and T’/R’ mixed phases. Remarkably, recovery of the smooth atomic steps after this process was possible. A further functionality of such films was demonstrated by PFM-based lithography of up and down FE domains. The domain walls between these regions of different polarization were shown to be conductive. Finally, magnetic characterisation (performed by colleagues in Taiwan) showed that while the presence of cobalt was detectable, the influence on the macroscopic magnetization M-H loops and the local magnetic moment by x-ray circular magnetic dichroism (XMCD) was negligible. This is consistent with a rather low fraction of cobalt doping. 5.2 Perspectives The research into multiferroic materials, and, particularly in the form of epitaxial thin films, is currently very active. This is primarily driven by the desire to design ultra-low energy nanoscale memory devices and spintronics architectures. In this context, the present work offers new routes to nanoscale control of strain ferroelectric and multiferroic thin films. The stabilization of the novel T’ phase of BFO up to higher thicknesses than conventional possible, enables measurements of the giant polarization (probably at low temperature), magnetic measurements (such as neutron diffraction which requires larger sample volumes and is thus not suited to ultrathin films), among others. It is noted that some of these experiments are presently underway as part of the continuing research work in Valanoor’s group. The inclusion of cobalt into the T’ BFO structure appeared not to strongly influence the physical properties of the material. This is likely due to the fraction of cobalt (2 %) being too low to induce an morphotropic phase boundary-type behaviour (usually ~10-15 % has been shown to cause significant structural changes in T’ BFO films [87]). The high film quality allowed to show almost perfect interconversion between the T’ phase and the mixed T’/R’ striped regions. In addition, the films showed measurable conductivity at the domain walls, opening up perspectives for domain-wall based nanoelectronic devices based on T’ BFO. 71 Despite the maturity of PLD as a growth technique, the fundamental interactions during growth are not fully understood. In particular the influence of substrate-target distance and secondary thermal evaporation of the target on growth are underreported factors in PLD growth. High speed cameras combined with strobe photography techniques could provide a window into these mechanisms in situ during growth. Computer modelling could also be employed to analyse the effects of a thermally evaporated cloud on the kinetic energy distribution of the plume, and any interactions between the thermally evaporated material and directly ablated plume. Understanding how these gas phase interactions in the plume affect the composition of the film forming species could open the door to finer control of film stoichiometry and structure, and in addition provide a model for other PVD based fabrication techniques. 72 6 References [1] N. A. Spaldin and M. Fiebig, “The Renaissance of Magnetoelectric Multiferroics,” Science (80-. )., vol. 309, p. 391, 2005. [2] P. Royen and K. Swars, “Das System Wismutoxyd-Eisenoxyd im Bereich von 0 bis 55 Mol{%} Eisenoxyd,” Angew. Chemie, vol. 69, no. 24, p. 779, Dec. 1957. [3] S. V Kiselev, R. P. Ozerov, and G. S. Zhdanov, “Detection of magnetic order in ferroelectric BiFeO3 by neutron diffraction,” Sov. Phys. Dokl., vol. 7, p. 742, 1963. [4] N. N. Krainik, N. P. Khuchua, V. V Zhdanova, and V. A. Evseev, “Phase transitions in BiFeO3,” Sov. Phys. Solid State, vol. 8, p. 654, 1966. [5] J. Wang et al., “Epitaxial BiFeO3 Multiferroic Thin Film Heterostructures,” Science (80-. )., vol. 299, no. 5613, p. 1719 LP-1722, Mar. 2003. [6] C. Ederer and N. A. Spaldin, “Effect of Epitaxial Strain on the Spontaneous Polarization of Thin Film Ferroelectrics,” Phys. Rev. Lett., vol. 95, no. 25, p. 257601, 2005. [7] D. Ricinschi, K.-Y. Yun, and M. Okuyama, “A mechanism for the 150 µC cm-2 polarization of BiFeO3 films based on first-principles calculations and new structural data.,” J. Phys. Condens. Matter, vol. 18, no. 6, pp. L97--L105, Feb. 2006. [8] J. B. Neaton, C. Ederer, U. V Waghmare, N. A. Spaldin, and K. M. Rabe, “First-principles study of spontaneous polarization in multiferroic BiFeO3,” Phys. Rev. B, vol. 71, no. 1, p. 14113, Jan. 2005. [9] J. Seidel et al., “Conduction at domain walls in oxide multiferroics,” Nat Mater, vol. 8, no. 3, pp. 229–234, 2009. [10] P. Sharma et al., “Nonvolatile ferroelectric domain wall memory,” Sci. Adv., vol. 3, no. 6, 2017. [11] J. F. Scott and C. A. Paz de Araujo, “Ferroelectric Memories,” Science (80-. )., vol. 246, no. 4936, p. 1400 LP-1405, Dec. 1989. [12] P. Barone and S. Picozzi, “Mechanisms and origin of multiferroicity,” Comptes Rendus Phys., vol. 16, no. 2, pp. 143–152, Mar. 2015. [13] D. Sando, A. Barthélémy, and M. Bibes, “BiFeO 3 epitaxial thin films and devices: past, present and future,” J. Phys. Condens. Matter, vol. 26, no. 47, p. 473201, Nov. 2014. 73 [14] E. Du Trémolet de Lacheisserie, D. Gignoux, and M. Schlenker, Magnetism, no. 1. Springer, 2005. [15] H. P. Myers, Introductory Solid State Physics, Second Edition. Taylor & Francis, 1997. [16] N. A. Hill, “Why Are There so Few Magnetic Ferroelectrics?,” J. Phys. Chem. B, vol. 104, no. 29, pp. 6694–6709, Jul. 2000. [17] L. D. Landau, E. M. Lifshitz, and A. L. King, “Electrodynamics of Continuous Media,” American Journal of Physics, vol. 29, no. 9. pp. 647–648, 1961. [18] D. Sando, A. Barthélémy, and M. Bibes, “BiFeO{<}sub{>}3{<}/sub{>} epitaxial thin films and devices: past, present and future,” J. Phys. Condens. Matter, vol. 26, no. 47, p. 473201, Nov. 2014. [19] G. Catalan and J. F. Scott, “Physics and Applications of Bismuth Ferrite,” Adv. Mater., vol. 21, no. 24, pp. 2463–2485, Jun. 2009. [20] J. M. Moreau, C. Michel, R. Gerson, and W. J. James, “Ferroelectric BiFeO3 X-ray and Neutron Diffraction Study,” J. Phys. Chem. Solids, vol. 32, p. 1315, 1971. [21] J. R. Teague, R. Gerson, and W. J. James, “Dielectric Hysteresis in Single Crystal BiFeO3,” Solid State Commun., vol. 8, p. 1073, 1970. [22] D. Lebeugle, D. Colson, A. Forget, and M. Viret, “Very large spontaneous electric polarization in BiFeO3 single crystals at room temperature and its evolution under cycling fields,” Appl. Phys. Lett., vol. 91, no. 2, 2007. [23] V. V Shvartsman, W. Kleemann, R. Haumont, and J. Kreisel, “Large bulk polarization and regular domain structure in ceramic BiFeO3,” Appl. Phys. Lett., vol. 90, no. 17, p. 172115, 2007. [24] W. Eerenstein, F. D. Morrison, J. Dho, M. G. Blamire, J. F. Scott, and N. D. Mathur, “Comment on ‘Epitaxial BiFeO3 multiferroic thin film heterostructures,’” Science, vol. 307, p. 1203, Feb. 2005. [25] H. Béa et al., “Investigation on the origin of the magnetic moment of BiFeO3 thin films by advanced x-ray characterizations,” Phys. Rev. B, vol. 74, no. 2, p. 20101, Jul. 2006. [26] R. E. Cohen, “Origin of ferroelectricity in perovskite oxides,” Nature, vol. 358, p. 136, Jul. 1992. 74 [27] A. M. Glazer, “The classification of tilted octahedra in perovskites,” Acta Crystallogr. Sect. B, vol. 28, no. 11, p. 3384, 1972. [28] A. M. Kadomtseva, A. K. Zvezdin, Y. F. Popov, A. P. Pyatakov, and G. P. Vorob’ev, “Space- time parity violation and magnetoelectric interactions in antiferromagnets,” J. Exp. Theor. Phys. Lett., vol. 79, no. 11, p. 571, Jun. 2004. [29] D. Lebeugle, D. Colson, A. Forget, M. Viret, A. M. Bataille, and A. Gukasov, “Electric-Field- Induced Spin Flop in BiFeO3 Single Crystals at Room Temperature,” Phys. Rev. Lett., vol. 100, no. 22, p. 227602, Jun. 2008. [30] D. Sando et al., “Crafting the magnonic and spintronic response of BiFeO 3 films by epitaxial strain,” Nat. Mater., vol. 12, no. 7, pp. 641–646, 2013. [31] X. Ke, P. P. Zhang, S. H. Baek, J. Zarestky, W. Tian, and C. B. Eom, “Magnetic structure of epitaxial multiferroic BiFeO3 films with engineered ferroelectric domains,” Phys. Rev. B, vol. 82, no. 13, p. 134448, Oct. 2010. [32] W. Ratcliff et al., “Neutron Diffraction Investigations of Magnetism in BiFeO3 Epitaxial Films,” Adv. Funct. Mater., vol. 21, no. 9, p. 1567, May 2011. [33] D. G. Schlom et al., “Elastic strain engineering of ferroic oxides,” MRS Bull., vol. 39, no. 02, pp. 118–130, Feb. 2014. [34] Z. Chen et al., “Low-Symmetry Monoclinic Phases and Polarization Rotation Path Mediated by Epitaxial Strain in Multiferroic BiFeO3 Thin Films,” Adv. Funct. Mater., vol. 21, no. 1, p. 133, Jan. 2011. [35] H. Béa et al., “Evidence for Room-Temperature Multiferroicity in a Compound with a Giant Axial Ratio,” Phys. Rev. Lett., vol. 102, no. 21, p. 217603, 2009. [36] M. D. Rossell et al., “Atomic Structure of Highly Strained BiFeO{_}{{}3{}} Thin Films,” Phys. Rev. Lett., vol. 108, no. 4, p. 47601, Jan. 2012. [37] J. X. Zhang et al., “Microscopic Origin of the Giant Ferroelectric Polarization in Tetragonal- like BiFeO3,” Phys. Rev. Lett., vol. 107, no. 14, p. 147602, Sep. 2011. [38] S. J. Pennycook et al., “Misfit accommodation in oxide thin film heterostructures,” Acta Mater., vol. 61, no. 8, p. 2725, May 2013. [39] R. J. Zeches et al., “A strain-driven morphotropic phase boundary in BiFeO{<}sub{>}3{<}/sub{>},” Science (80-. )., vol. 326, no. 5955, pp. 977–980, Nov. 2009. 75 [40] H. Fu and R. E. Cohen, “Polarization rotation mechanism for ultrahigh electromechanical response in single-crystal piezoelectrics,” Nature, vol. 403, no. 6767, p. 281, Jan. 2000. [41] C.-E. Cheng et al., “Revealing the flexoelectricity in the mixed-phase regions of epitaxial BiFeO{<}sub{>}3{<}/sub{>} thin films,” Sci. Rep., vol. 5, p. 8091, 2015. [42] A. R. Damodaran et al., “Nanoscale structure and mechanism for enhanced electromechanical response of highly Strained BiFeO3 thin films.,” Adv. Mater., vol. 23, no. 28, p. 3170, Jul. 2011. [43] D. Sando, B. Xu, L. Bellaiche, and V. Nagarajan, “A multiferroic on the brink: Uncovering the nuances of strain-induced transitions in BiFeO3,” Appl. Phys. Rev., vol. 3, no. 1, p. 11106, 2016. [44] C. Ederer and N. A. Spaldin, “Weak ferromagnetism and magnetoelectric coupling in bismuth ferrite,” Phys. Rev. B, vol. 71, no. 6, p. 60401, Feb. 2005. [45] Q. He et al., “Electrically controllable spontaneous magnetism in nanoscale mixed phase multiferroics,” Nat Commun, vol. 2, p. 225, 2011. [46] W. Siemons, M. D. Biegalski, J. H. Nam, and H. M. Christen, “Temperature-Driven Structural Phase Transition in Tetragonal-Like BiFeO{$}{_}{{}3{}}{$},” Appl. Phys. Express, vol. 4, no. 9, p. 95801, Aug. 2011. [47] Y. Yang, M. Stengel, W. Ren, X. H. Yan, and L. Bellaiche, “Epitaxial short-period PbTiO{_}{{}3{}}/BiFeO{_}{{}3{}} superlattices studied by first-principles calculations,” Phys. Rev. B, vol. 86, no. 14, p. 144114, Oct. 2012. [48] J. Allibe et al., “Coengineering of ferroelectric and exchange bias properties in BiFeO{<}sub{>}3{<}/sub{>} based heterostructures,” Appl. Phys. Lett., vol. 95, no. 18, p. 182503, 2009. [49] M. Bibes and A. Barthélémy, “Towards a magnetoelectric memory,” Nat. Mater., vol. 7, p. 425, 2008. [50] H. M. Smith and A. F. Turner, “Vacuum Deposited Thin Films Using a Ruby Laser,” Appl. Opt., vol. 4, no. 1, pp. 147–148, Jan. 1965. [51] A. Inam et al., “As‐deposited high Tc and Jc superconducting thin films made at low temperatures,” Appl. Phys. Lett., vol. 53, no. 10, pp. 908–910, 1988. [52] L. R. Shaginyan, “Chapter 13 - Pulsed laser deposition of thin films: Expectations and 76 reality,” in Handbook of Thin Films, H. S. Nalwa, Ed. Burlington: Academic Press, 2002, pp. 627–673. [53] R. Eason, Pulsed Laser Deposition of Thin Films: Applications-Led Growth of Functional Materials. Wiley, 2007. [54] K. R. Chen et al., “Mechanisms affecting kinetic energies of laser‐ablated materials,” J. Vac. Sci. Technol. A, vol. 14, no. 3, pp. 1111–1114, 1996. [55] I. T. and S. Higuchi, “Pulsed-Laser Deposition of LiNbO 3 in Low Gas Pressure Using Pure Ozone,” Jpn. J. Appl. Phys., vol. 43, no. 8R, p. 5307, 2004. [56] L. I. Maissel and R. Glang, Handbook of thin film technology. McGraw-Hill, 1970. [57] K. L. Saenger, “On the origin of spatial nonuniformities in the composition of pulsed‐laser‐ deposited films,” J. Appl. Phys., vol. 70, no. 10, pp. 5629–5635, 1991. [58] M. C. Foote, B. B. Jones, B. D. Hunt, J. B. Barner, R. P. Vasquez, and L. J. Bajuk, “Composition variations in pulsed-laser-deposited Y-Ba-Cu-O thin films as a function of deposition parameters,” Phys. C Supercond., vol. 201, no. 1, pp. 176–182, 1992. [59] H. Béa et al., “Influence of parasitic phases on the properties of BiFeO 3 epitaxial thin films,” Appl. Phys. Lett., vol. 87, no. 7, p. 72508, 2005. [60] Y. Waseda, E. Matsubara, and K. Shinoda, X-Ray Diffraction Crystallography: Introduction, Examples and Solved Problems. Springer Berlin Heidelberg, 2011. [61] B. D. Cullity, Elements of X Ray Diffraction - Scholar’s Choice Edition. Scholar’s Choice, 2015. [62] A. Authier, “Dynamical theory of X-ray diffraction,” in International Tables for Crystallography Volume B: Reciprocal space, U. Shmueli, Ed. Dordrecht: Springer Netherlands, 2001, pp. 534–551. [63] G. Binnig, C. F. Quate, and C. Gerber, “Atomic Force Microscope,” Phys. Rev. Lett., vol. 56, no. 9, pp. 930–933, Mar. 1986. [64] R. Garcıá and R. Pérez, “Dynamic atomic force microscopy methods,” Surf. Sci. Rep., vol. 47, no. 6, pp. 197–301, 2002. [65] S. Morita, R. Wiesendanger, and E. Meyer, Noncontact Atomic Force Microscopy, 1st ed. Springer-Verlag Berlin Heidelberg, 2002. 77 [66] Y. Martin, C. C. Williams, and H. K. Wickramasinghe, “Atomic force microscope–force mapping and profiling on a sub 100‐Å scale,” J. Appl. Phys., vol. 61, no. 10, pp. 4723–4729, May 1987. [67] R. J. Zeches et al., “ Morphotropic Phase Boundary in BiFeO3
[68] S. N. and T. U. and K. D. and K. S. and H. F. and O. S. and Y. K. and N. T. and M. Shimizu, “Strain evolution of epitaxial tetragonal-like BiFeO 3 thin films on LaAlO 3 (001) substrates prepared by sputtering and their bulk photovoltaic effect,” Jpn. J. Appl. Phys., vol. 55, no. 10, p. 101501, 2016.
[69] D. Sando et al., “Large elasto-optic effect and reversible electrochromism in multiferroic BiFeO3,” Nat. Commun., vol. 7, 2016.
[70] D. Sando et al., “Designer defect stabilization of the super tetragonal phase in >70-nm- thick BiFeO3 films on LaAlO3 substrates,” Jpn. J. Appl. Phys., vol. 57, no. 9, p. 0902B2, 2018.
[71] W. H. Richardson, “Bayesian-Based Iterative Method of Image Restoration$\ast$,” J. Opt. Soc. Am., vol. 62, no. 1, pp. 55–59, Jan. 1972.
[72] L. B. Lucy, “An iterative technique for the rectification of observed distributions,” Astron. J., vol. 79, p. 745, 1974.
[73] Y. Heo, B.-K. Jang, S. J. Kim, C.-H. Yang, and J. Seidel, “Nanoscale Mechanical Softening of Morphotropic BiFeO3,” Adv. Mater., vol. 26, no. 45, pp. 7568–7572, 2014.
[74] R. Kelly, “On the dual role of the Knudsen layer and unsteady, adiabatic expansion in pulse sputtering phenomena,” J. Chem. Phys., vol. 92, no. 8, pp. 5047–5056, 1990.
[75] E. Fogarassy, C. Fuchs, J. P. Stoquert, P. Siffert, J. Perrière, and F. Rochet, “High Tc YBaCuO and BiSrCaCuO superconducting thin films deposited by pulsed excimer laser evaporation,” J. Less Common Met., vol. 151, pp. 249–256, 1989.
[76] C. Cali, V. Daneu, A. Orioli, and S. Riva-Sanseverino, “Flash evaporation of compounds with a pulsed-discharge CO2 laser,” Appl. Opt., vol. 15, no. 5, pp. 1327–1330, 1976.
[77] A. Gupta, B. Braren, K. G. Casey, B. W. Hussey, and R. Kelly, “Direct imaging of the fragments produced during excimer laser ablation of YBa 2 Cu 3 O 7−δ,” Appl. Phys. Lett.,
78
vol. 59, no. 11, pp. 1302–1304, Sep. 1991.
[78] D. H. Lowndes, D. B. Geohegan, A. A. Puretzky, D. P. Norton, and C. M. Rouleau, “Synthesis of Novel Thin-Film Materials by Pulsed Laser Deposition,” Science (80-. )., vol. 273, no. 5277, pp. 898–903, 1996.
[79] S. K. Blower and C. Greaves, “The structure of β-Bi2O3 from powder neutron diffraction data,” Acta Crystallogr. Sect. C, vol. 44, no. 4, pp. 587–589, Apr. 1988.
[80] H. Liu, P. Yang, K. Yao, K. P. Ong, P. Wu, and J. Wang, “Origin of a Tetragonal BiFeO3 Phase with a Giant c/a Ratio on SrTiO3 Substrates,” Adv. Funct. Mater., vol. 22, no. 5, p. 937, Mar. 2012.
[81] C. Beekman et al., “Phase Transitions, Phase Coexistence, and Piezoelectric Switching Behavior in Highly Strained BiFeO3 Films,” Adv. Mater., vol. 25, no. 39, pp. 5561–5567, 2013.
[82] H. Naganuma et al., “Enhancement of ferroelectric and magnetic properties in BiFeO3 films by small amount of cobalt addition,” J. Appl. Phys., vol. 103, no. 7, p. 07E314, 2008.
[83] H. Naganuma, J. Miura, and S. Okamura, “Ferroelectric, electrical and magnetic properties of Cr, Mn, Co, Ni, Cu added polycrystalline BiFeO3 films,” Appl. Phys. Lett., vol. 93, no. 5, p. 052901, Aug. 2008.
[84] I. B. Bersuker, “Pseudo Jahn-Teller Origin of Perovskite Multiferroics, Magnetic- Ferroelectric Crossover, and Magnetoelectric Effects: The d0−d10 Problem,” Phys. Rev. Lett., vol. 108, no. 13, p. 137202, 2012.
[85] A. A. Belik et al., “Neutron powder diffraction study on the crystal and magnetic structures of BiCoO3,” Chem. Mater., vol. 18, p. 798, 2006.
[86] Y. Nakamura et al., “Enhanced Piezoelectric Constant of (1- x )BiFeO 3 – x BiCoO 3 Thin Films Grown on LaAlO 3 Substrate,” Jpn. J. Appl. Phys., vol. 50, p. 31505, Mar. 2011.
[87] K. Shimizu, H. Hojo, Y. Ikuhara, and M. Azuma, “Enhanced Piezoelectric Response due to Polarization Rotation in Cobalt-Substituted BiFeO3 Epitaxial Thin Films,” Adv. Mater., p. n/a-n/a, 2016.
[88] K. A. Bogle et al., “Synthesis of Epitaxial Metal Oxide Nanocrystals via a Phase Separation Approach,” ACS Nano, vol. 4, no. 9, pp. 5139–5146, Sep. 2010.
[89] D. Sando, B. Xu, L. Bellaiche, and V. Nagarajan, “A multiferroic on the brink: uncovering 79
the nuances of strain-induced transitions in BiFeO{<}sub{>}3{<}/sub{>},” Appl. Phys. Rev., vol. 3, p. 11106, 2016.
[90] H. M. Christen, J. H. Nam, H. S. Kim, A. J. Hatt, and N. A. Spaldin, “Stress-induced R-MA- MC-T symmetry changes in BiFeO3 films,” Phys. Rev. B, vol. 83, no. 14, p. 144107, Apr. 2011.
[91] Z. Chen et al., “Low symmetry monoclinic MC phase in epitaxial BiFeO3 thin films on LaSrAlO4 substrates,” Appl. Phys. Lett., vol. 97, no. 24, p. 242903, 2010.
[92] T. Young et al., “Structural, magnetic, and ferroelectric properties of T-like cobalt-doped BiFeO3 thin films,” APL Mater., vol. 6, no. 2, p. 26102, Feb. 2018.
[93] P. Chen et al., “Optical properties of quasi-tetragonal BiFeO[sub 3] thin films,” Appl. Phys. Lett., vol. 96, no. 13, p. 131907, 2010.
[94] R. K. Vasudevan, Y. Liu, J. Li, and W. I. Liang, “Nanoscale Control of Phase Variants in Strain-Engineered BiFeO{<}sub{>}3{<}/sub{>},” Nano Lett., vol. 11, p. 828, 2011.
[95] P. Sharma, T. Nakajima, S. Okamura, and A. Gruverman, “Effect of disorder potential on domain switching behavior in polymer ferroelectric films,” Nanotechnology, vol. 24, no. 1, p. 15706, 2013.
[96] J. X. Zhang et al., “Large field-induced strains in a lead-free piezoelectric material.,” Nat. Nanotechnol., vol. 6, no. 2, p. 98, Feb. 2011.
[97] W. Siemons et al., “Strain dependence of transition temperatures and structural symmetry of BiFeO3 within the tetragonal-like structure,” Appl. Phys. Lett., vol. 101, no. 21, p. 212901, 2012.
[98] P. Sharma et al., “Morphotropic Phase Elasticity of Strained BiFeO3,” Adv. Mater. Interfaces, vol. 3, no. 12, p. 1600033, Mar. 2016.
[99] H. Dong, Z. Wu, S. Wang, W. Duan, and J. Li, “Improving the optical absorption of BiFeO3 for photovoltaic applications via uniaxial compression or biaxial tension,” Appl. Phys. Lett., vol. 102, no. 7, p. 72905, 2013.
80