ABSTRACT 
 
 
 
 
Title of Dissertation: SYNTHESIS AND CHARACTERIZATION OF 
MULTIFERROIC THIN FILMS 
 
 
 Sung Hwan Lim, Doctor of Philosophy, 2008 
Dissertation directed By: Professor Lourdes Salamanca-Riba and  
Associate Professor Ichiro Takeuchi 
Department of Materials Science and Engineering 
 
 
Multiferroic materials and multiferroic materials systems which 
simultaneously exhibit ferroelectricity and magnetism have attracted great attention 
because of their exotic physical properties and their potential applications which 
utilize coupling of magnetism and ferroelectricity. The goal of this thesis was to study 
multiferroic materials systems in thin film and multilayer forms in order to explore 
the possibility of fabricating room temperature thin film devices. 
In particular, we have focused on two types of multiferroic materials systems: 
1) intrinsic multiferroic/magnetoelectric thin film materials and 2) magnetostrictive/ 
piezoelectric bilayer systems for investigation of the strain-mediated magnetoelectric 
(ME) effect. 
BiFeO
3
 is an intrinsic multiferroic which displays ferroelectricity and 
antiferromagnetism at room temperature, and thus of strong interest for ambient 
  
device applications. In this thesis, we have extensively investigated the role of 
microstructure on the properties of BiFeO
3
 thin films. We studied multiphase 
formation in Bi-Fe-O thin films, and found that formation of secondary phases such 
as ?-Fe
2
O
3
, ?-Fe
2
O
3
, and Fe
3
O
4
 increased overall saturation magnetization and 
released the misfit strain of the BiFeO
3
 grains in the films. We also observed large 
polarization in Bi-Fe-O thin films containing secondary phases that have almost fully 
relaxed the misfit strain. 
We have studied several aspects of the ME effect which are directly relevant 
to possible novel device applications. Electric field tunable spintronic devices using 
the ME effect have been proposed. In one such device configuration, the desired 
effect is electric field tuning of giant magnetoresistance or tunnel magnetoresistance 
through control of exchange bias via the ME effect. We have investigated the 
feasibility of such a device using exchange-biased Co/Pt multilayers on Cr
2
O
3
 thin 
films. 
The strain-mediated ME effect at the interface of magnetostrictive/ 
piezoelectric bilayers has been widely used to demonstrate magnetic field detection 
with extremely high sensitivity. Although the overall mechanism of such an effect is 
known, the details of the bilayer interfaces and how they affect the coupling is not 
understood. In order to directly observe the strain-mediated ME coupling effect, we 
fabricated bilayer thin film structures and performed in-situ dynamic observation of 
magnetic domains while an electric-field was being applied using Lorentz 
transmission electron microscopy. Electric-field induced motion of magnetic domain 
boundaries in the magnetostrictive layer was observed for the first time. 
  
 
 
 
 
 
 
 
 
 
SYNTHESIS AND CHARACTERIZATION OF MULTIFERROIC THIN FILMS  
 
 
 
By 
 
 
Sung Hwan Lim 
 
 
 
 
 
Dissertation submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 
Doctor of Philosophy 
2008 
 
 
 
 
 
 
 
 
 
Advisory Committee: 
 
     Professor Lourdes Salamanca-Riba, Chair / Advisor 
     Associate Professor Ichiro Takeuchi, Co-Advisor 
   Professor Manfred Wuttig 
   Assistant Professor John Cumings 
     Professor Richard Greene 
 
 
 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
? Copyright by 
Sung Hwan Lim 
2008 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
Acknowledgements 
I would like to express my sincere gratitude to my advisors, Prof. Lourdes 
Salamanca-Riba and Prof. Ichiro Takeuchi, for giving me the opportunity to work on 
exciting projects. I would especially like to thank both of them for their guidance and 
support throughout my Ph.D research at the University of Maryland. I have been 
working with Prof. Salamanca-Riba since the first year I came to the U.S. I benefited 
from her guidance in every aspect during my Ph.D study, including the discussions 
we held and intelligent suggestions she made regarding my research. I deeply 
appreciate all the time and effort she spent discussing my projects and advising me. I 
also appreciate all corrections to my papers and thesis, attentive consideration and 
encouragement she gave me. It would not have been possible to accomplish my 
scientific goals without her guidance. Also, the same gratitude goes to Prof. I. 
Takeuchi, who served as my co-advisor. It was a great honor to join his group and 
learn thin film synthesis and characterization. I have great respect for his knowledge 
and passion for research. He always encouraged me to go forward to achieve my 
academic goals. 
I am also grateful for the insights and advice Prof. John Cumings shared with 
me. His suggestions and guidance were a great help for performing electron beam 
lithography and Lorentz TEM as related to the experiments in Chapter 5. 
I would like to thank Prof. Manfred Wuttig for his kind assistance and advice 
on the theoretical calculations and understanding of the magnetoelectric coupling 
effect, also in Chapter 5. 
 
ii
 
I am also grateful to Prof. Samuel E. Lofland, Prof. Nagarajan Valanoor, Prof. 
Nigel Browning, and Prof. S. Sundar Manoharan for their valuable assistance in my 
research. 
I would like to thank Prof. Richard Greene for taking time out of his busy 
schedule to serve on my committee and evaluate my work.  
I would like to give my special acknowledgements to Dr. Todd Brintlinger as 
a collaborator. All the ideas we shared and discussed have proven very fruitful. 
Without his contribution this project would not have gone so well.  
I would like to thank Dr. Makoto Murakami for his help and advice at the 
beginning of my studies. I would like to express my gratitude for the helping hands of 
my colleagues and collaborators: Anbusathaiah Varatharajan, Shenqiang Ren, Yi Qi, 
Volkan Ortalan, Dr. Daisuke Kan, Kamal Hussain Baloch, Dr. Ranjan K. Sahu, 
Dwight Hunter, and all other members in Prof. Takeuchi group and Prof. Salamanca-
Riba group. 
 I would like to give my special thanks to Sheng-Yu Young and Kwan Lee as  
collaborators and also as best friends. All the help and cheer they showed have been 
very rewarding. All the friendship and delightful memories we share will be kept in 
my mind with the greatest care. 
I especially appreciate my father, my mother, my brothers and my sisters for 
their understanding, encouragement, and endless love. 
Finally, I want to thank my lovely wife, Mi-Sun Kim, and my son, Hyun-Bin 
Lim. I would never achieve anything without their deep love. 
 
iii
 
Table of Contents 
 
 
Abstract 
Acknowledgements??????????????..??????????ii 
List of Tables???????????????????????????..vi 
List of Figures???????????????????????????..vii 
List of Abbreviations????????????????????????xx 
Chemical Nomenclature??????????????????????xxi 
 
Chapter 1. Introduction????????????????????????..1 
1.1 Background????????????????????????.1 
1.2 Intrinsic magnetoelectric materials and intrinsic multiferroic materials?.4 
1.3 Composite multiferroic materials???????????????.12 
 
Chapter 2. Experimental approach???????????????????..16 
2.1 Thin films fabrication???????????????????.16 
2.2 Characterization techniques????????????????..?19 
2.2.1 Transmission electron microscopy (TEM)????????19 
2.2.2 Lorentz transmission electron microscopy (LTEM)????.20 
2.2.3 Atomic force microscopy (AFM)?????????????22 
2.2.4 Magnetic and electrical properties measurements??????.25 
 
Chapter 3. Multiphase formation in Bi-Fe-O thin films???????????.26 
            3.1 Introduction????????????????????...???.26 
            3.2 Oxygen pressure dependent multiphase formation?????????27 
            3.3 Change of magnetic properties of Bi-Fe-O by multiphase formation?..33 
                        3.3.1 Formation of gamma Fe
2
O
3
 and Fe
3
O
4
????????..33 
                        3.3.2 Tuning of magnetic property of Bi-Fe-O thin film????..36 
            3.4 Change of misfit strain in BiFeO
3
 thin films by multiphase formation..39 
                        3.4.1 Gradient thickness samples?????????????..39 
                        3.4.2 Combinatorial approach??????????????..42 
            3.5 Change of ferroelectric properties by multiphase formation????..46 
                        3.5.1 Ferroelectric properties in columnar BiFeO
3
-Fe
2
O
3
???...?47 
                        3.5.2 Ferroelectric domains switching analysis using Piezo force 
                                  Microscopy (PFM)????????????????. ...49 
                        3.5.3 Chemical analysis on columnar BiFeO
3
-Fe
2
O
3
 thin films using  
                                  EDS and EELS in TEM?.. ?????????????.52 
            3.6 Multiferroic properties in polycrystalline BiFeO
3
 thin films????.53 
                        3.6.1 Size effects on electrical polarization and leakage current?...56 
                        3.6.2 Switching of ferroelectric nano-domains ???????....60 
                        3.6.3 Annealing effects on the microstructure and electrical properties 
                                  of polycrystalline BiFeO
3
 films????????????.63 
           3.7 Enhancement of dielectric properties of BiFeO
3
 films by Flux-mediated 
                  epitaxy (FME) method??????????????????....67 
 
iv
 
                        3.7.1 Introduction of FME method?????????????..67 
                        3.7.2 Combinatorial approach for optimum growth conditions??...69 
                                   3.7.2.1 Temperature gradient libraries?????????69 
                                   3.7.2.2 Pseudo Ternary composition libraries??????72 
                        3.7.3 Enhanced structural and dielectric properties??????.....74 
 
Chapter 4. Application the ME effect for spintronic device application??..........79 
            4.1 Introduction of exchange bias?????????????????79 
            4.2 Control of magnetic moment and exchange bias in Cr
2
O
3
 thin films??81 
            4.3 Magnetoelectric effect on exchange bias in Cr
2
O
3
 thin films??.??89 
 
Chapter 5. Dynamic observation of ME effect using Lorentz TEM?????........90 
            5.1 Strain-mediated ME effect using Fe-Ga/BaTiO
3
 bilayer??????91 
            5.2 Magnetic domain structure of Fe-Ga by MFM and Lorenz TEM??.94 
            5.3 TEM sample preparation using electron beam lithography and Focused- 
                   Ion Beam (FIB) milling????????????????.??100 
            5.4 Switching magnetic domains by applying a magnetic field??.?...102 
            5.5 Switching magnetic domains by applying an electric field (ME Effect) 
                   ????????????????????????.?110 
 
Chapter 6. Summary and future work?????????????????.114 
 
 
References???????????????????????????.?.117 
 
v
 
List of Tables 
 
Table 1 Relationship between the current through the objective lens in the 
LaB
6
 TEM and the corresponding out-of-plane and in-plane 
components (by 9 degree tilting) of the magnetic field on the 
sample.???????????????????????107 
 
vi
 
List of Figures 
 
Figure 1.1 Example of magnetoelectric coupling in Ni
3
B
7
O
13
I. The magneto- 
electric (ME)
H
 voltage signal (polarization) was reversed in magnetic 
field (- 0.6 T < ?
0
H < 0.6 T) at 46 K ?????????????4 
Figure 1.2 Schematic of the unit cell of BFO
3
 which has ABO type perovskite 
structure????????????????????????5 
Figure 1.3 Time-reversal and spatial-inversion symmetry in multiferroic 
materials????????????????????????6 
Figure 1.4 Schematic of a GMR device and its magnetoresistance curve involving 
an ME film as a pinning layer. Blue and red represent two states (of 
applied voltage)?????????????????????11 
Figure 1.5 Schematic of a multiferroic composite for strain-mediated ME 
coupling.????????.???????????????12 
Figure 1.6 (a) Induced voltage by an applied ac magnetic field on an ME sensor 
(metglas + polyvinylidene fluoride).[41] The inset shows a calibration 
scan of a ME device. (b) Large sharp change of magnetization 
measured by VSM due to an applied electric field in La
0.67
Sr
0.33
MnO
3
-
BaTiO
3
.[42]??????????????????????13 
Figure 2.1 Schematics of (a) combinatorial pulsed laser deposition and (b) PLD 
system, which consists of (c) the chamber, (d) moving shadow mask, 
sample stage, and (e) multiple target stage.??????????18 
 
vii
 
Figure 2.2 Two different methods to observe magnetic domains in Lorentz TEM. 
Schematics for Fresnel imaging (a) and Foucault imaging (b)???21 
Figure 2.3 Schematic of an Atomic Force Microscope??????????23 
Figure 2.4 A SQUID device, which consists of two superconductors separated by 
thin insulating layers to form two parallel Josephson junctions?25 
Figure 3.1 XRD spectrum from BiFeO
3
 films grown at different deposition 
oxygen????????????????????????27 
Figure 3.2 a) Cross-sectional and b) plan-view TEM images taken from BiFeO
3
 
film grown at 1 mTorr oxygen pressure, and the c) diffraction patterns 
and d) indexing results of plan-view diffraction pattern. e) Cross-
sectional TEM image and electron diffraction (inset) from the BiFeO
3
 
film grown at 0.1 mTorr oxygen pressure??????????30 
Figure 3.3 TEM bright field images and SAD pattern of bismuth iron oxide thin 
films fabricated on (001) SrTiO
3
 substrates at 1 mTorr Po
2
. (a) 
Magnified cross-section (left) and (b) plan-view (right) images 
showing ?-Fe
2
O
3
 precipitates and (c) SAD pattern and corresponding 
schematic of ?-Fe
2
O
3
 and ?-Fe
2
O
3 
phases in the film deposited at 1 
mTorr. (d) High resolution cross sectional TEM image of ?-Fe
2
O
3
 in 
?-Fe
2
O
3
 matrix?????????????????????32 
Figure 3.4 Oxygen pressure dependence of saturation magnetization in Bi-Fe-O 
films. As oxygen pressure decreases, magnetization increases??33 
 
viii
 
Figure 3.5 Room temperature saturation magnetization (emu/cc) of Bi-Fe-O films 
as a function of thickness for films grown at 5 mTorr and 20 
mTorr.????????????????????????34 
Figure 3.6 Top left image is a cross-sectional image of a Bi-Fe-O film grown at 5 
mTorr oxygen pressure with schematic of FE and FM regions before 
(right) and after (left) annealing. Right schematic is the transformation 
of ?-Fe
2
O
3
 to ?-Fe
2
O
3
 upon annealing. Annealing procedures to 
transform ?-Fe
2
O
3
 into ?-Fe
2
O
3
 includes hydrogen flowing to 
transform ?-Fe
2
O
3
 to Fe
3
O
4
, which is followed by oxidation to 
transform Fe
3
O
4
 to ?-Fe
2
O
3
.????????????????37 
Figure 3.7 Magnetic force microscopy (MFM) images of an annealed multiphase 
film (a) before and (c) after magnetizing the film perpendicular to the 
surface by applying 20 kOe magnetic field. (b) Hysteresis loops of as 
grown and annealed multiphase Bi-Fe-O films. (d) Schematic of the 
microstructure of Bi-Fe-O multiferroic nanocomposites?????38 
Figure 3.8 (a) X-ray diffraction spectra (in log scale) and (b) TEM cross-sectional 
images (left) and EDS spectra (right) of Bi-Fe-O films grown at 5 
mTorr with different thickness. Nucleation of impurity phases starts at 
~ 50 nm. The intensity of the peak labeled Fe
2
O
3
 in (a) corresponding 
to the impurity phases increases with increasing film thickness..?40 
Figure 3.9 Out-of-plane lattice parameters in pure BiFeO
3
 and BiFeO
3
 with 
secondary phases as a function of film thickness???????41 
 
ix
 
Figure 3.10 Superpositions of x-ray microdiffraction scans from regions of films 
with a thickness gradient grown at (a) 20 mTorr and (b) 5 mTorr. The 
thickness varied from 30 to 300 nm. Schematics of the misfit strain 
relaxation mechanism in (c) pure BiFeO
3
 film and (d) multiphase Bi-
Fe-O film are suggested. Multiphase film shows coexistence of highly 
strained region (A) and fully relaxed region (B)????????43 
Figure 3.11 Schematics of two different strain relaxation mechanisms. In 
multiphase BiFeO
3
, in-plane compressive stress of BiFeO
3
 columns is 
relaxed by the formation of Fe
2
O
3
 phase with smaller volume. AFM 
image from the surface (upper right) also confirmed smaller volume of 
Fe
2
O
3
 phase from height difference between columns??????45 
Figure 3.12 Electrical polarization vs. applied electric field from the samples 
grown at (a) 20 mTorr and (b) 2 mTorr. (c) Leakage current density 
measured from the two samples...??????????????47 
Figure 3.13 Bright field TEM images from BiFeO
3
 films grown at (a) 20 mTorr 
and (b) 2 mTorr. Surface morphology of the films grown at (c) 20 
mTorr and (d) 2 mTorr measured by AFM..??????????48 
Figure 3.14 (a) 5 ? 5 ?m
2
 height image and (b) ferroelectric domain switching 
images from the BiFeO
3
 thin film grown at 20 mTorr of oxygen 
pressure. To obtain image showing switching of ferroelectric domains 
a + 10 V and ? 10 V bias were alternately applied on 3 ? 3 ?m
2
 and 1 
? 1 ?m
2
 area, respectively. (c) The piezoelectric coefficient (d
33
) was 
measured from strain loop...????????????????49 
 
x
 
Figure 3.15 (a) 5 ? 5 ?m
2
 AFM Height image and (b) switching ferroelectric 
domain image from the BiFeO
3
 film grown at 2 mTorr. Magnified (c) 
AFM height image and (d) corresponding PFM image of switching 
domains from the region marked in (b) d
33
 versus applied voltage 
obtained from (e) middle and (f) edge of a BiFeO
3
 column... 
???????????????????????????51 
Figure 3.16 (a) Scanning TEM dark field image and (b) EDS elemental mapping of 
Bi-L edge. (c) Magnified dark field STEM image and (d) EELS line 
scan for Fe
2+
/Fe
3+
.????????????????????53 
Figure 3.17 Bright field TEM images and corresponding electron diffraction 
patterns from the polycrystalline BiFeO
3
 films grown at (a) 300 
o
C, (b) 
400 
o
C, and (c) 500 
o
C. As the deposition temperature increases, the 
grain size increases from 40 to 150 nm. (d) The diffraction pattern 
from the film grown at 400 
o
C indicates randomly oriented 
polycrystalline BiFeO
3
 with no impurities...??????57 
Figure 3.18 Electrical polarization hysteresis loops from the samples grown at (a) 
300 
o
C (b) 400 
o
C, and (c) 500 
o
C. Each loop was acquired at different 
applied voltage (38 ? 55 V for (a) and 37 ? 45 V in (b). (d) Leakage 
current density of each sample as function of applied electric field, and 
(e) relationship between saturation polarization and leakage current on 
deposition temperature. The film grown at 400 
o
C showed the largest 
polarization (Ps ~ 106 ?C/cm
2
) with 800 kV/cm of coercive field as 
well as highest leakage current density???????????58 
 
xi
 
Figure 3.19 AFM height images (top) and PFM images (bottom) of switching 
electrical domain from the films grown at 300 
o
C (a, d), 400 
o
C (b, e), 
and 500 
o
C (c, f). For switching domains, a positive followed by a 
negative bias (? 15 V) were applied in 3 ? 3, 1.6 ? 1.6 and 0.8 ? 0.8 
?m
2
 area, alternately. Only the film grown at 400 
o
C showed change of 
contrast with switching the applied bias???????????61 
Figure 3.20 (a) Bight-field TEM image and (b) high resolution TEM image 
showing pure BiFeO
3
 nanograins with size of 5 ? 10 nm diameter after 
annealing at 500 
o
C for 8 hours.  (c) Changes of leakage current and 
(d) electrical polarization upon annealing. Nucleation of nanodomain 
structure of polycrystalline BiFeO
3
 enhanced the polarization, but also 
increased the leakage current density mainly due to an increase in the 
area of domain boundaries and evaporation of Bi
3+
 and O
2-
 upon 
annealing.???????????????????????64 
Figure 3.21 (a) Bright field and (b) dark field Scanning TEM (STEM) images from 
the same area in the annealed poly BiFeO
3
 film. The inset to (b) 
represents the chemical composition through one of the grains with ~ 
20 nm diameter in size using EDS line scanning??????65 
Figure 3.22 FME for BiFeO
3
 growth. First SrRuO
3
 bottom electrode (for electrical 
measurements) (50 ? 80 nm), After the temperature is increased to 
650
o
C, BiFeO
3
 seed layer (~ 10 nm) is deposited on (001) oriented 
SrTiO
3
 by PLD. Then the flux layer of a mixture of (Bi
2
O
3
 + CuO) is 
deposited on the seed layer. The temperature is then increased to 
 
xii
 
730
o
C, and BiFeO
3
 is deposited on the flux layer. SEM image shows 
uniformly distributed flux droplets on the BiFeO
3
 seed layer with a 
size ranging 2 ? 4 ?m diameter. Once the desired thickness of BiFeO
3
 
is deposited the remaining flux is etched off.?????????68 
Figure 3.23 Schematic (top left) of the deposition procedure for temperature-
gradient combinatorial analysis. The temperatures are varied from 580 
to 810 
o
C with compositional spread of Bi
2
O
3
 (400 ?) and CuO (0 ~ 
400 ?). (a) XRD map of combinatorial library shows the intensity of 
the BiFeO
3
 (001) peak as a function of substrate temperature and flux 
CuO composition to decide the optimum condition for FME. (b) 2? x-
ray map in the region marked I in (a) for 0 % CuO. (c) Addition of 10 
? 30 vol% Cu (XRD mapping result labeled II in (a)) shows highest 
intensity of the BiFeO
3
 (001) peak enlarging the growth window of 
BiFeO
3
 compared to the film deposited without Cu (labeled (I))??71 
Figure 3.24 Schematic (top left) of the pseudo ternary composition spread 
procedure. (a) the picture and (b) schematic of the shadow mask used 
in combinatorial PLD system..???????????????72 
Figure 3.25 (a) The specimen of Bi
2
O
3
-CuO-BiFeO
3
 ternary system. (b) 2-
dimensional and c) 3-dimensional x-ray maps for the diffraction 
intensity and 2 theta angles of BiFeO
3
 peak ?????????73 
Figure 3.26 Atomic force microscopy images taken from 5 ?m ? 5 ?m area of 
BiFeO
3
 thin films grown (a) by conventional PLD and (b) by FME 
method. (c) The 3-D image of magnified BiFeO
3
 domain in (b) shows 
 
xiii
 
atomic steps during growth of the film with grain size of ~ 2 ?m (~ 10 
times larger than that of BiFeO
3
 grown by normal PLD).????74 
Figure 3.27 (a) Cross-sectional bright field TEM image showing smooth surface of 
the BiFeO
3
 film which was grown on SrRuO
3
 buffered (001) oriented 
SrTiO
3
. 1000 ? thick palladium (Pd) was used as the top electrode. 
The high resolution lattice images of (b) interface between SrTiO
3
 and 
SrRuO
3
 and (c) interface between SrRuO
3
 and BiFeO
3
 show epitaxial 
growth. The corresponding diffraction pattern from each layer is 
shown as inset in (b) and (c)????????????????75 
Figure 3.28 (a) Dielectric constant of BiFeO
3
 thin films fabricated by typical PLD 
method on SrTiO
3
 (100) and (111), and by flux method on SrTiO
3
 
(100) as a function of frequency (10 kHz ? 10 MHz). (b) Leakage 
current density (A/cm2) vs. electric field (MV/cm) curves of BiFeO
3
 
thin films fabricated by typical PLD and flux method. (c) Magnetic 
hysteresis curve (in-plane) of BiFeO
3
 thin film fabricated by flux 
method. (d) Electric field dependence of polarization loops of BiFeO
3
 
thin film fabricated by flux method measured at 1 kHz. The inset in (d) 
is the P-E loop from a PLD grown film???????????77 
Figure 4.1 (a) Typical shape of M-H hysteresis loop of ferromagnetic materials. 
b) Schematic diagram of the spin configuration of FM-AFM bilayer 
???????????????????????????80 
 
xiv
 
Figure 4.2 (a) XRD 2-theta scan of a Co/Pt/Al-ZnO/Cr
2
O
3
 film. (b) Leakage 
current density of Cr
2
O
3
 film measured using Pt (top) and Al-ZnO 
(bottom) electrodes at room temperature..??????????83 
Figure 4.3 (a) High-resolution TEM image of Co/Pt layer/Cr
2
O
3
 interface. Co 
particles were embedded in a Pt matrix. The thickness of Co/Pt layer 
was ~ 10 nm. (b) Electron diffraction pattern and (c) schematic of the 
crystal orientation relation of the Cr
2
O
3
/Al-ZnO/Al
2
O
3
. The film 
showed epitaxial growth of Al-ZnO and Cr
2
O
3
 layers on c-axis 
sapphire substrate????????????????????84 
Figure 4.4 (a) and (b) show the out-of-plane hysteresis loops of our (Co/Pt)/Cr
2
O
3
 
structure at room temperature after magnetic-field cooling (MFC) 
through the N?el temperature following heating of the sample to 325 K, 
where H
C
 of the Co/Pt layer was 15 Oe. We found clear exchange bias 
with cooling in a field of only 20 Oe. As the temperature decreased 
from 325 K to 298 K, H
E
 was seen to gradually increase to ~ 170 Oe, 
and the sign of H
E
 was consistent with the direction of the applied field. 
Even at 298 K, with the exchange bias present, H
C
 of Co remained 
narrow at 15 Oe????????????????????86 
Figure 4.5 (a) The schematic of magnetic field cooling process with sweeping the 
magnetic field with various offsets and (b) Control of magnetic 
hysteresis by sweeping the magnetic field with various offsets during 
magnetic field cooling. The hysteresis loops gradually change from ? 
 
xv
 
H
E
 to + H
E
 and also from single loop to dual loop by changing offset 
from + 0.2 kOe to ? 0.2 kOe...???????????????88 
Figure 5.1 (a) X-ray diffraction spectrum showing epitaxial growth of BaTiO
3
 
thin film on (001) oriented SrTiO
3
. (b) Electrical polarization versus 
applied electric field showing ferroelectric hysteresis curve of BaTiO
3
. 
(c) Bright field TEM image of polycrystalline Fe
0.7
Ga
0.3
 thin film. (d) 
M-H hysteresis curves obtained from the 600nm thick Fe-Ga thin film 
along the out-of-plane and in-plane directions..???????92 
Figure 5.2 (a) Schematic for Lorentz TEM sample preparation. After deposition 
of FE and FM bilayer thin film structure, FM layer was patterned 
using optical lithography and electron beam lithography to apply an 
electric field. Schematics of (b) randomly oriented magnetic domains 
in the initial state, (c) induced tensile strain along field direction after 
applying an electric field (E), and (d) alignment of magnetic domains 
along tensile strain direction...???????????????93 
Figure 5.3 (a) Fe-Ga patterned on BaTiO
3
/SrTiO
3
. (b) Height image acquired 
from 1.5 ?1.5 ?m
2
 area in the middle pad. The data scale was 0 ~ 5 nm 
(c) MFM image from the same region. The data scale was 0 ~ 6 degree 
??????????????????????????? 94 
Figure 5.4 Fresnel images of Fe-Ga thin film obtained at various defocusing 
values of (a) over focus (+ 1.2 mm), (b) in focus (zero), and (c) under 
focus (? 1.2 mm). (d) Phase map from the area marked in (c) 
reconstructed by transport of intensity equation (TIE)??????96 
 
xvi
 
Figure 5.5 Foucalt images of Fe-Ga thin film. By changing the aperture position 
along x direction [(a) and (b)] and y direction [(c) and (d)], magnetic 
domains which are oriented along x direction or y direction show as 
bright and dark areas...? ? ? ? ?.???????????97 
Figure 5.6 Change of magnetic domains in an Fe-Ga thin film under an applied 
external field. The magnetic field was applied along the in-plane 
direction (red arrow) and increased by increasing the tilt angle from 0 
to 9 degrees. The number from 1 to 9 represents the sequence. The 
small arrows represent the direction of the magnetic domains???98 
Figure 5.7 Change of magnetic domains in Fe-Ga thin film. The magnetic field 
(yellow arrow) was applied along opposite direction (compare to Fig. 
5. 6) and increased by increasing the tilt angle from 0 ? 9 degrees. The 
number from 1 to 9 represents the sequence?????????99 
Figure 5.8 Process for deposition of Fe-Ga electrodes. First the electrodes were 
patterned using photo lithography. After developing the pattern Fe-Ga 
was deposited by magnetron sputtering at room temperature. Then, 
electron-beam lithography was used to pattern a gap on the middle Fe-
Ga electrode. Finally, the TEM sample was prepared by mechanical 
polishing and FIB milling..????????????????100 
Figure 5.9 SEM images obtained at different stages of FIB milling. (a) after 
milling using a beam with largest beam current (11500 pA) and (b) 
second largest current (6600 pA). SEM images number from 1 to 6 
show gradual change of contrast by decreasing thickness. (c) and (d) 
 
xvii
 
are SEM images from the bottom view and end view, respectively, 
after milling was completed...???????????????101 
Figure 5.10 Magnetic domain structure of fully unconstrained Fe-Ga/BaTiO
3
 thin 
film. Fresnel images obtained at (a) over focus (?F = + 4500 nm) and 
(b) under focus (?F= ? 4500 nm). (c) and (d) are Foucalt images 
obtained by moving the aperture along ?y axis..???????103 
Figure 5.11 Increase of magnetic field component in-plane by tilting the sample to 
+ 9.0 
o
 and increasing the current though the objective lens (numbers 
in each figure)????????????????????104 
Figure 5.12 Decrease of magnetic field component in-plane by tilting the sample 
(+ 9.0 
o
) and decreasing the current through the objective lens 
(numbers in each figure)????????????????105 
Figure 5.13 Increase and decrease of magnetic field along in-plane by tilting the 
sample (+ 9.0 
o
) and changing the current through the objective lens 
(numbers in each figure). (a) Fresnel mode at 0 
o
 tilt. (b) Foucalt mode 
at 0 
o
 tilt. (c) ? (h) Foucalt images corresponding to an in-plane 
component of the magnetic domains. (i) The new magnetic domains 
after reducing the magnetic field back to zero Red and light blue 
arrows indicate magnetic domains orientation????????106 
Figure 5.14 Change of tilting angle and focal length from (a) + 9.0 (F = ? 6000 
nm) to (b) ? 9.0 degree (F = + 6000 nm)??????????108 
Figure 5.15 Change of magnetic domain structure according to applied magnetic 
field direction (indicated in the images)??????????109 
 
xviii
 
Figure 5.16 Switching of magnetic domains by the application of an electric field. 
(a) E = 0 V, (b) E = 7.5 V, and (c) E= 10 V. (d) Acquired leakage 
current versus applied electric field (voltage) The arrows indicate the 
changed regions????????????????????112 
Figure 5.17 Comparison of the magnetic domains before and after applying an 
electric field. Red-colored region represents the magnetic domains 
with magnetization along x before applying an electric field (E = 0 V) 
and the blue-colored region represents the increased area of magnetic 
domains after applying an electric field (E = 10 V)?113 
 
xix
 
 
List of Abbreviations 
 
 
ME                         Magnetoelectric 
PLD                       Pulsed laser deposition 
PVD                       Physical vapor deposition 
FME                       Flux-mediate epitaxy 
P-E                         Electrical polarization-electric field 
I-V                          Current-voltage 
MOKE                   Magneto-optical Kerr effect 
VSM                      Vibrating sample magnetometer 
AFM                      Atomic force microscopy 
XRD                      X-ray diffraction 
TEM                      Transmission electron microscopy 
EDS                      Energy-dispersive X-ray spectroscopy 
EELS                     Electron energy loss spectroscopy 
EB                         Exchange bias 
MFC                      Magnetic field cool 
MEFC                    Magnetic and electric field cool 
 
 
xx
 
Chemical Nomenclature 
 
 
BiFeO
3
                          Bismuth ferrite 
SrTiO
3
                           Strontium titanate 
?-Fe
2
O
3
                         Hematite 
?-Fe
2
O
3
                          Maghemite 
Cr
2
O
3
                            Chromium oxide 
FeGa                               Iron gallium 
BaTiO
3
                           Barium titanate 
 
 
Greek Letters 
 
 
2?                                  Scattering angle 
?                                      Strain 
?                                      Stress 
?                                   X-ray wavelength 
?                                      Susceptibility, 
 
 
xxi
 
Chapter 1  
1. Introduction 
 
1.1 Background 
 
The focus of my study was fundamental materials science and device physics 
aimed at exploring possible room temperature device application of multiferroic/ 
magnetoelectric thin film materials and multiferroic composite thin films. I studied 
multiferroic properties of intrinsic multiferroic BiFeO
3
 thin film which has a weak 
ferromagnetism and ferroelectricity at room temperature. I have explored the 
possibility of fabricating tunable spintronic devices using magnetoelectric Cr
2
O
3
 thin 
films which had been predicted to show electric-field tunable antiferromagnetism at 
temperatures near room temperature. I also studied direct observation of strain-
mediated magnetoelectric (ME) coupling effect using multiferroic composite thin films. 
Multiferroics are materials that exhibit two or three of the ferroic properties: 
ferroelectricity, ferromagnetism, and ferroelasticity.[1] Ferroelectricity is the 
phenomenon that involves electrical polarization. A ferroelectric material undergoes a 
phase transition from a high-temperature phase (paraelectric) state above the Curie 
temperature (T
C
) to a low temperature phase (ferroelectric) that has a spontaneous 
polarization along a preferred direction. As electric field is applied, ferroelectric 
materials exhibit a saturation polarization (P
S
) along the direction of the applied 
electric field. Once saturation is reached, remnant polarization (P
r
) remains in the 
absence of an external electric field. This polarization direction can be reversed by 
 1 
 
switching the external electric field. At the atomic scale, the existence or absence of 
ferroelectricity in an ionic material is determined by a balance between the short-
range repulsions, which prefer the nonferroelectric (paraelectric) symmetric structure 
and long-range Coulomb forces, which prefer the ferroelectric state. The most widely 
studied and used ferroelectric materials are oxides with a perovskite structure, which 
has the form ABO
3
. Ferroelectrics find a large range of applications, such as 
nonvolatile random access memories, capacitors, and transducers. Because many 
ferroelectrics also exhibit large electromechanical coupling, they are also widely used 
for piezoelectric applications. 
Ferromagnetism is a phenomenon which involves spontaneous magnetization 
below a critical temperature. The spontaneous magnetization can be switched and 
saturated (M
s
) along the direction of external magnetic field (H). There is a remanent 
magnetization (M
r
) once the field is removed. The applications of ferromagnetic 
materials are strongly determined by the values of the remanent magnetization and 
coercivity (H
C
). For example, a square-shaped hysteresis loop with high remanent 
magnetization and high coercive field is needed for permanent magnet applications, 
while materials with narrow hysteresis loops are needed for recording media.  
Ferroelasticity is a phenomenon which involves a spontaneous deformation (?). 
In ferroelastic materials, by applying a stress (?) to a ferroelastic phase, a phase can be 
changed from one phase to an equally stable structure with different orientations, and 
this process involves formation motion of twin boundaries. This stress-induced phase 
change results in a spontaneous strain in the material. The shape memory effect and 
superelasticity are important phenomena which involve ferroelasticity. 
 2 
 
The projects carried out in this thesis mainly focused on multiferroic materials 
in which ferroelectricity and ferromagnetism coexist. The magnetoelectric (ME) 
coupling effect is manifested in the control of spontaneous polarization by applying a 
magnetic field and/or control of spontaneous magnetization by applying an electric 
field. The multiferroic coupling effects are often related to strain in the materials. For 
example, coupling between ferroelectricity and ferroelasticity is known as 
piezoelectricity. Piezoelectricity gives rise to a change of polarization (P) by an 
applied stress (?), or a change of strain (?) by an applied electric field (E). Similarly, 
the coupling between ferromagnetism and ferroelasticity is called piezomagnetism, 
through which a change of strain can take place by an applied magnetic field, and 
vice versa. Also, this coupling is referred to as magnetostriction which corresponds to 
a change of strain as a quadratic function of the applied magnetic field. 
Magnetoelectric materials which show ME coupling effects had been 
investigated earlier from the 1950s to 1970s.[4-6] However, recently there has been a 
surge of activities in this area because of discoveries of new materials and 
development of better understanding of the multiferroic properties. Magnetoelectric 
materials are also being pursued because the ability to couple the two order 
parameters (simultaneous ferroelectric and magnetic properties) allows an additional 
degree of freedom in device design. 
All the magnetoelectric materials that have been studied to date can be 
classified into two categories: single phase multiferroics which show ferroelectricity 
and ferromagnetism simultaneously in one phase and multiferroic composites which 
are combinations of ferromagnetic and ferroelectric phases. 
 3 
 
1.2 Intrinsic magnetoelectric materials and intrinsic multiferroic 
materials 
 
There are various magnetoelectric materials and multiferroic materials. 
Magnetoelectric materials are not necessarily multiferroic since they can show the 
ME coupling effect without robust simultaneous presence of ferroelectricity. In 1960s, 
a linear magnetoelectric coupling coefficient was observed in Cr
2
O
3
, which is not a 
ferroelectric material.[11,12] In 1966, a boracite Ni
3
B
7
O
12
I, which has weak 
ferromagnetic and weak ferroelectric ordering  below 60 K was first used to 
demonstrate the magnetic reversing of electric polarization.[13] 
 
 
 
 
 
 
 
 
Figure 1.1 Example of magnetoelectric coupling in Ni
3
B
7
O
13
I. The 
magnetoelectric (ME)
H
 voltage signal (polarization) was reversed in a magnetic field 
(- 0.6 T < ?
0
H < 0.6 T) at 46 K.[13] 
 
As shown in Fig. 1.1, a weak magnetic-field induced reversal of the 
magnetization was observed to flip the polarization. The magnetic field induced effect 
gave rise to switching of the polarization (~ 0.076 ?Ccm
-2
), and the ME voltage 
 4 
 
signal was flipped from about -100 mV to 100 mV by switching the magnetic field 
from -0.6 T to 0.6 T. These were some of the first investigation of ME effects. 
Some of the more recently studied multiferroics include hexagonal rare-earth 
manganites with formula MMnO
3
 (M = Y, Ho, Er, Tm, Yb, Lu, or Sc) which are 
intrinsic multiferroics which exhibit ferroelectricity and frustrated magnetic 
ordering.[14-17] The compounds BaMF
4
 (M = Ho, Fe, Co, Ni) have an orthorhombic 
crystal structure at high temperatures with antiferromagnetic (weak ferromagnetic) 
ordering and antiferroelectric (ferro-electro-elastic) properties at low 
temperature.[18,19] Also, there are other perovskite multiferroic materials. A 
perovskite compound contains magnetic ions which fully (or partially) occupy the 
octahedral positions in the perovskite structure. For example, the compound BiFeO
3
 
has a rhombohedral, pseudoperovskite, structure with alternating layers of bismuth 
and oxygen and iron and oxygen. The octahedral positions contain Fe
3+
 ions which 
cause exchange interaction between the Fe
3+
 ions along the Fe
3+
 ? O
2-
 ? Fe
3+
 chains 
(see Fig. 1.2). 
 
 
 
 
 
 
 
Figure 1.2 Schematic of the unit cell of BiFeO
3
 which has the ABO
3
 type 
perovskite structure. 
 5 
 
Fig. 1.3 summarizes the occurrence of coexistence of ferromagnetism and 
ferroelectricity in a single phase. Figure 1.3(a) depicts a ferromagnet by a spin 
orientation of a charge and local magnetic moment m. Time reversal switches both 
the direction of orbit and moment, while spatial inversion leaves both orbit and 
moment unchanged. Figure 1.3 (b) represents ferroelectrics with local dipole moment 
P from asymmetrical displacement of positive and negative charges in the unit cell. 
Spatial inversion switches P, while time inversion shows no change. Fig. 1.3(c) 
represents multiferroics that have both ferroelectric and ferromagnetic properties with 
time reversal of magnetization and spin orbit and spatial inversion of polarization. 
 
 
 
 
 
 
 
 
Figure 1.3 Time-reversal and spatial-inversion symmetry in multiferroic 
materials [W. Eerenstein, et al., Nature 442, 17, 759 (2006)] 
he possible coexistence of the spontaneous magnetization and polarization in 
the same phase does not contradict the general criteria for the appearance of 
ferroma netism and ferroelectricity taken separately. Magnetic ordering is governed 
by the exchange interaction of the electron spins, while ferroelectric ordering is 
 
T
g
 6 
 
govern
number of multiferroics exist in nature, or have been synthesized in the laboratory, 
and the
iFeO
3
, which make them a strong candidate 
for roo
ed by the redistribution of charge density in the lattice. However, a limited 
 reasons for the scarcity of ferromagnetic/ferroelectric coexistence has been 
explored by Nicola A. Hill.[7] She found that, in general, the d electrons in transition 
metals, which are essential for magnetism, reduce the tendency for off-center 
ferroelectric distortion. Consequently, an additional electronic or structural driving 
force must be present for ferromagnetism and ferroelectricity to coexist. In addition to 
the fact that such materials are rare, these materials exhibit multiferroic properties and 
magnetoelectric coupling at low temperatures (below room temperature) reducing the 
feasibility for practical applications. For example, TbMn
2
O
5
 is an intrinsic 
multiferroic material which shows magnetism-induced ferroelectricity and switchable 
polarization with magnetic field, but its polarization is very small (~ 80 nCcm
-1
) and 
it is at a very low temperature (3 K).[8] 
Single crystal BiFeO
3
 has been known as an intrinsic multiferroic material 
which is a canted antiferromagnet (weak ferromagnetism) below its N?el temperature 
(T
N
 = ~ 640 K), and a ferroelectric below its Curie temperature (T
C
 = ~ 1100 K). 
However, recently a remarkable discovery has been reported in this material by J. 
Wang et al.[20] They found significant increase in its magnetization and electrical 
polarization in thin film structures of B
m-temperature operated multiferroic device applications. In thin film structures, 
P
S
 as high as ~ 55 ?C/cm
2
 was observed, compared to ~ 3.5 ?C/cm
2
 observed in 
single crystals earlier. In Wang?s thin film materials, M
S
 was also found to be as high 
as 150 emu/cc at room temperature compared to ~ 2 emu/cc in single crystals.[21-24] 
 7 
 
In subsequent reports, the misfit strain in the film induced from the lattice mismatch 
with the substrate as well as structural distortions from the transformation of 
rhombohedral to tetragonal crystal structures in BiFeO
3
 thin films were attributed to 
as the dominant reason for enhancements of both ferroelectric and ferromagnetic 
properties.[20] However, clear evidences for the origin of increased intrinsic 
magnetization and polarization were not found. The observed properties were often 
not reproducible, and there had been a number of reports on the properties of BiFeO
3
 
thin films with conflicting results.  
In the present study, we have studied the origin of enhanced ferromagnetism 
and ferroelectric properties in BiFeO
3
 thin films. The results are presented in Chapter 
3. To understand the enhancement of ferromagnetism, we studied multiphase 
formation in Bi-Fe-O thin film. We found that formation of secondary phases such as 
?-Fe
2
O
3
, ?-Fe
2
O
3
, and Fe
3
O
4
 takes place at relatively high growth temperature at 
lower oxygen pressure. The formation of ferromagnetic ?-Fe
2
O
3
 and Fe
3
O
4
 increased 
the ove
 the role of strain and 
grain s
rall saturation magnetization of the Bi-Fe-O thin films. 
To understand the large polarization reported in BiFeO
3
 films, we investigated 
the columnar structure of BiFeO
3
 and Fe
2
O
3
 which showed large polarization and 
found that misfit strain in these films is remarkably relaxed by the formation of Fe
2
O
3
. 
Our results suggest that the origin of large polarization of the BiFeO
3
 films is not 
misfit strain in the thin film structure (as had been reported prior to this study) but 
that it is an intrinsic property of BiFeO
3
. To better understand
ize on domain switching, we also studied the ferroelectric properties of 
polycrystalline BiFeO
3
 thin films grown at relatively low temperatures. These films 
 8 
 
are less sensitive to misfit strain from the substrate and possess less impurities and 
defects due to the relatively low growth temperature. In these polycrystalline BiFeO
3
 
thin films, large polarization and effective switching of ferroelectric domains were 
observed in the films with nano-sized grains. The effect of grain size on ferroelectric 
properties and annealing effect was investigated. These results are presented in 
Chapter 3.6. 
We have also explored tuning of the structural, electrical and magnetic 
properties in BiFeO
3
 thin films. In order to improve the crystallinity of the films, the 
flux-mediate epitaxy method (FME) was introduced. Using the FME method, single 
crystal-like thin films with large grains and atomically smooth surface were 
synthesized. These films showed improvement of the dielectric properties compared 
to regular films. Also, phase transformation from antiferromagnetic ?-Fe
2
O
3
 to 
ferrom
magnetic field sensor application, the modulation of 
electric
agnetic ?-Fe
2
O
3
 by annealing at hydrogen atmosphere was investigated in 
order to explore tuning of the ferromagnetic properties in films containing columnar 
structure of BiFeO
3
-Fe
2
O
3
.  
A possible device application of magnetoelectric materials using exchange 
bias is discussed in Chapter 4. Novel memory devices, magnetic read-head, and 
magnetic field sensors have been suggested as possible applications of the ME effect. 
In the memory device schemes, ferroelectric and ferromagnetic random access 
memories can be combined which can result in an increase in memory storage ability 
and functionalities. In the 
al polarization is utilized, and this effect had been found to show extremely 
high sensitivities. 
 9 
 
Magnetic read-head and some sensor schemes are based on giant 
magnetoresistance (GMR)[43-46] or tunnel magnetoresistance (TMR) effects[47], 
which show change of resistance when the magnetic configuration between two 
neighboring ferromagnetic layers is modified. Typically, one magnetic layer is pinned 
through exchange bias by an adjacent antiferromagnetic layer. Recently, there has 
been a growing interest in a direct method for magnetization reversal including spin 
transfe
 
r from spin-polarized current in the device.[48,49] An alternative is to combine 
GMR or a TMR device with a magnetoelectric thin film where the ME thin film is 
used as the antiferromagnetic pinning layer providing exchange bias.[50] In such a 
ME spintronic device, antiferromagnetic ordering in the ME thin film is not affected 
by a moderate external magnetic field. However, the magnetic configuration of a 
spintronic device can be tuned in the presence of an applied electric field by changing 
the antiferromagnetic property of the ME thin film layer. In theory, the 
magnetoresistance can then be controlled by changing the antiferromagnetic property 
of the ME thin film which in turn tunes the exchange bias to the adjacent 
ferromagnetic layer. 
One embodiment of such a device scheme is depicted in Fig. 1.4. Here, the 
magnetic properties of ME layer are controlled by an applied voltage. This layer 
controls switching of the bottom ferromagnetic pinned layer. This in turn controls the 
magnetoresistance behavior of the device.  
 
 
 10 
 
 
 
 
 
 
 
 
Figure 1.4 gnetoresistance curve 
involving an ME film d red represent two states (of applied 
voltage).[119] 
 
 an effect, we have investigated the 
behavior of exchange co ilms. The idea is to 
see if we ca e used Cr
2
O
3
 
instead of BiFeO
3
established to date in BiFeO
3 2 3
ll known as the first material which 
interface between Co/Pt and the Cr
2
O
3
 layer (RMS ~ 0.3 nm) and applied a 
sufficie
 Schematic of a GMR device and its ma
 as a pinning layer. Blue an
To explore the possible occurrence of such
upled Co/Pt multilayers on Cr
2
O
3
 thin f
n observe tunable exchange bias using the ME coupling. W
, because the ME coupling effects have not been unambiguously 
, while Cr O  is we
showed the magnetoelectric coefficient.[11,12] We achieved an extremely smooth 
ntly high electric field of 77 MV/m after the structure is field cooled through 
the N?el temperature of Cr
2
O
3
 (T
N
 = 308 K). This electric field value is ~ 150 times 
higher than the applied field on an earlier reported exchange biased Cr
2
O
3
 single 
crystal experiment.[109] Despite the high quality of our samples, the thin film device 
exhibited no significant ME effect on exchange bias with in the measured sensitivity 
range even at the high electric field. We discuss the possible reasons for this null 
observation in Chapter 4. 
 11 
 
Using the CoPt/Cr
2
O
3
 device configuration, we found an efficient technique 
for controlling the exchange bias field and its sign by sweeping a magnetic field with 
offset during cooling. We found that the sign of the exchange bias and the shape of 
the hysteresis loops, i.e. single/dual hysteresis loop of the out-of-plane magnetized 
Co/Pt layers can be delicately controlled by adjusting the magnetic field during 
cooling through the N?el temperature of Cr
2
O
3
. 
 
1.3 Co
 which consist of a ferroelectric 
(piezoelectric/electrostrict magnetic/magnetostrictive) 
ia strain mediation at their interface as 
schema cally shown in Fig. 1.5.[9,25]  
ltiferroic composite for strain-mediated ME 
coupling.  
mposite multiferroic materials 
 
As mentioned above, intrinsic (single phase) multiferroics are relatively rare 
in nature, and a weak ME coupling effect usually can only be observed at low 
temperatures and high magnetic fields. In contrast, artificial multiferroic materials can 
be achieved by synthesis of multiferroic composites
ive) phase and a ferromagnetic (piezo
phase, and these two parameters are coupled v
ti
 
 
 
 
 
 
Figure 1.5 Schematic of a mu
 12 
 
In such a geometry, the distortion of the magnetostrictive phase by an applied 
magnetic stimulus is transferred to the piezoelectric phase and results in an electrical 
response. Conversely, by applying an electric stimulus, a magnetic response can be 
generated in the magnetostrictive layer. Such multiferroic composites[26] can be 
achieved in the form of laminates,[27,28] or epitaxial multilayers.[29] To date, 
CoFe
2
O , NiFeO
3
, Terfenol-D, and La
1-x
Sr
x
MnO
3
 have been used for the 
ferroma
used fo
Figure 1. 6 (a) Induced voltage by an applied ac magnetic field on an ME 
sensor (metglas + polyvinylidene fluoride).[41] The inset shows a calibration scan of 
a ME device. (b) Large sharp change of magnetization measured by VSM due to an 
applied electric field in La
0.67
Sr
0.33
MnO
3
-BaTiO
3
.[42] 
4
gnetic phase, and BaTiO
3
, Pb(Zr,Ti)O
3
 (PZT), BiFeO
3
, PbTiO
3
 have been 
r the ferroelectric phase.[30-36] The magnetoelectric coefficient (?
ME
 = 
dP/dH) in PZT /Terfenol-D composite material was found to be as high as 2 Vcm
-
1
Oe
-1
 = 5 ? 10
?8
 sm
?1
 (or Cm
-2
Oe
-1
).[10] In a La
0.67
Sr
0.33
MnO
3 
/ BaTiO
3
 system, the 
converse effect was observed with ?
0
(dM/dE) = 2.3 ? 10
?7
 sm
?1
.[41] These values are 
substantially larger than the highest ME coefficient observed in single-phase 
magnetoelectric material, Cr
2
O
3
 (?
ME
 = dE/dH = 0.01 Vcm
-1
Oe
-1
 =  2.67 ? 10
?12
 sm
?1
 
at near N?el temperature of 307 K)[37, 40]. 
 
 
 
 
 
 
 
 13 
 
Fig. 1.6 (a) and (b) show methods for measuring ME effects using multiferroic 
composites. Using a piezoelectric polyvinylidene fluoride (PVDF) layer and a 
magnetostrictive metglas layer, J. R. Hattrick-Simpers et al. measured voltage in the 
polyvinylidene layer induced by an applied magnetic field [Fig. 1.6 (a)].[41] The 
slope of the curve indicates that the sensor has a sensitivity of 467 ?V/Oe over an ac 
field range of 10
?6
 ~ 10
3
 T. In Fig. 1.6(b), W. Eerenstein et al. measured the change of 
magnetization using a vibrating sample micrometer (VSM) during the application of 
an elec
domain
tric field (6 kVcm
-1
) on a La
0.67
Sr
0.33
MnO
3
/BaTiO
3
 heterostructure. After 
applying an electric field to the unpoled BaTiO
3
 substrate, the population of 90 
o
 
s was altered due to displacement of the Ti cation in the BaTiO
3
 unit cell as 
determined from the X-ray diffraction spectra (not shown), and the sharp switching of 
magnetic moments was observed by applying an electric field on the sample.[42] 
While such structures are beginning to be common for exploring strain-mediated ME 
devices, the details of the state of the interface between the two layers are not well 
understood. In fact, it is generally accepted that the formation of the interface which 
shows good coupling effect is non-trivial and the mechanism of the coupling through 
a typically used adhesive agent such as silver paste is not understood. 
There had been no direct and dynamic visual observations of strain-mediated 
ME coupling effect because such a study is non-trivial. In Chapter 5, we describe the 
direct observation of strain-mediated ME coupling using in-situ Lorentz transmission 
electron microscopy (TEM) in an unconstrained double layer of ferromagnetic iron 
gallium (Fe-Ga) and ferroelectric barium titanate (BaTiO
3
) film. The bilayer coupling 
is due to the natural adhesion of room temperature deposited Fe-Ga layer on the 
 14 
 
BaTiO
3
 thin film. The strain propagation, domain boundary movement, and reversible 
switching of magnetic domains in the ferromagnetic layer were dynamically observed 
by applying an electric field to the ferroelectric layer in Lorentz TEM. 
 15 
 
Chapter 2 
2. Experimental Approach 
 
To synthesize the films investigated in the work, pulsed laser deposition 
(PLD) was used. X-ray diffraction (XRD) and transmission electron microscopy 
(TEM) were used for structural analysis. Atomic force microscopy (AFM) and 
secondary electron microscopy (SEM) were used to investigate surface morphology 
such as roughness and/or grain size. The magnetic properties were measured using a 
vibrating sample magnetometer (VSM), a magneto-optical Kerr effect (MOKE) set-
up, and a superconducting quantum interference device (SQUID). The electrical 
properties were measured using an RT6000 test system, and HP 4192 impedance/gain 
analyzer 
 
2.1 Thin film fabrication 
 
For the thin film fabrication, pulsed laser deposition (PLD) was used for the 
growth of the oxide thin films, and physical vapor deposition (PVD) was used for the 
growth if the magnetic or conducting metal thin films. As shown in Fig. 2.1, a PLD 
system includes a laser source such as an ArF, KrF, or XeF excimer laser or a 
Nd:YAG laser, an optical system which contains an aperture, a mirror, and a focusing 
lens. It also includes a vacuum chamber which has a heater, a heating stage, a rotating 
target holder, and a moving shadow mask inside. For this study, a KrF excimer laser 
 16 
 
with a 248 nm wavelength and 30 ns pulse width was used. During the deposition 
process, the pulsed laser beam is focused onto a target to increase the laser energy 
density (energy/unit area at target surface) by a focusing lens. Typically the laser 
energy is in the range between 0.01 to 1.2 J at a frequency of 1 ? 20 Hz during 
deposition. The focused laser beam which has higher energy than the ablation 
threshold value dissociates and ablates the target material and forms a plasma plume. 
The plume is directed normal to the target surface and collected on the heated 
substrate. High film quality can be achieved through the optimization of deposition 
parameters which include substrate temperature, base pressure, deposition gas 
pressure, laser energy density, and frequency. For the deposition of oxide thin films, 
oxygen must be introduced into the chamber in order to assist in the formation of the 
desired phase and obtain the correct film composition. The most important advantage 
of PLD is its versatility. Using the PLD technique, various materials such as oxides, 
semiconductors, and polymers can be easily grown. In general, the conditions are 
optimized to achieve stoichiometric transfer of the target composition to the film. 
Another advantage of PLD is the easy optimization process compared to other 
deposition techniques. Because of the short duration and high energy of the laser 
pulse, the target material plume is directed toward the substrate and different elemental 
components can have similar deposition rates, which help to make films with the same 
composition as the target. Also, PLD can be used to make multilayer structures by 
using multiple targets. PLD is primarily used in research laboratories to investigate new 
materials onto relatively small area substrate (up to 1 cm ? 1 cm) because the highly 
directional plume makes it difficult to obtain uniform films over a large area. 
 17 
 
In a combinatorial PLD chamber (Fig 2.1), the most important feature is the 
moving shadow mask located in front of a sample stage which allows the growth of 
films with gradients in the composition and/or thickness. The mask has a 10 ? 10 mm 
square hole which is typically the same size as most commercial oxide substrates. By 
moving the shadow mask to cover and expose the substrate to the deposition plume, 
each position in the sample can have different exposure time to the plume, and this 
results in continuously varied compositions across one sample. Such a sample is 
called a composition spread. Also using a single target, one can easily make samples 
with a thickness gradient which is a very useful technique to study microstructural 
evolution during film growth or the change of physical properties as a function of 
heteroepitaxial constraint. 
 
 
 
 
 
 
 
 
 
 
Figure 2.1 Schematics of (a) combinatorial pulsed laser deposition and (b) 
PLD system, which consists of (c) the chamber, (d) moving shadow mask, sample 
stage, and (e) multiple target stage. 
 
 18 
 
2.2 Characterization techniques 
 
2.2.1 Transmission electron microscopy (TEM) 
 
In this thesis, transmission electron microscopy (TEM) was performed 
extensively to characterize the microstructure of the films. The morphological, 
crystallographic features as well as chemical information in the films have been 
obtained using TEM. The microscopes that were used in this study include a JOEL 
4000FX TEM operated at 300 KV, a JEOL 2100F field emission TEM and a JEOL 
2100 LaB
6
 TEM operated at 200 KV. 
A TEM operates similarly to an optical microscope, except that optical 
microscopes use light sources and focus light beams with glass lenses, while TEMs 
use electron sources and focus electron beams with electromagnetic lenses. The 
electrons emitted from the filament are accelerated by a high voltage (100 kV ? 1000 
kV) and focused through a set of condenser lenses onto the specimen. The electron 
beam is scattered by the specimen. The diffracted beams are then brought to focus by 
the objective lens on its back focal plane and form a diffraction pattern. A final TEM 
image or diffraction pattern can be produced on a fluorescent viewing screen by a 
series of objective lens, intermediate lens and projector and magnifying lenses. The 
main imaging and diffraction techniques include: 1. conventional imaging (bright-
field and dark-field TEM); 2. electron diffraction (selected area electron diffraction, 
SAD); 3. convergent-beam electron diffraction (CBED); 4. phase-contrast imaging 
(high-resolution TEM, HRTEM); and 5. Z-contrast imaging. Besides diffraction and 
 19 
 
spatial imaging, the high-energy electrons in TEM cause electronic excitations of the 
atoms in the specimen. Two important spectroscopic techniques make use of these 
excitations, i.e. energy-dispersive x-ray spectroscopy (EDS) and electron energy-loss 
spectroscopy (EELS). In this study, the conventional bright-field and dark-field 
imaging, SAD, and HRTEM are used to study the microstructures and EDS and 
EELS are used to study the chemical compositions of the films.  
 
2.2.2 Lorentz transmission electron microscopy (LTEM) 
 
We can also use TEM to study magnetic materials. If a sample is in-plane 
magnetized and has magnetic domains, the electron beam will feel a Lorentz force 
from the magnetic domains as it propagates through the sample. This force results in 
change of electron beam path. This deflection in the beam path allows the observation 
of magnetic domain by detecting the tilting of the electron beam. However, the 
electron beam can only be tilted by the component of the magnetic moments aligned 
perpendicular to the electron beam. Therefore, we can only see magnetic domains 
which are aligned along the in-plane direction in the thin film. 
However, there is a limitation for magnetic domain observation in TEM.  To 
tilt the electron beam and magnify the image, magnetic lenses are used in a TEM. To 
obtain relatively high magnification, the objective lens, which is the nearest lens to 
the sample, has to be used. However, when the objective lens is turned on, a few tesla 
of magnetic field along the out of plane direction is applied in the region where the 
sample is. This strong field results in switching of magnetic domains along the beam 
 20 
 
direction. Therefore, we cannot see the magnetic domains at high magnification. 
Consequently, in Lorentz TEM, there is an upper limit in the magnification that can 
be used to observe magnetic domains. This magnification does not use the objective 
lens and only utilizes the lenses in low magnification mode. 
As shown in Figure 2.2, there are two different methods to observe magnetic 
domain structures. One is called Fresnel imaging and the other is Foucalt imaging. In 
the Fresnel mode, by adjusting the focal length between under focus and over focus, 
the domain boundaries have more or less electron density and will appear as 
alternating dark and bright lines (by Lorentz force) in the TEM image [Fig. 2.2 (a)]. 
The contrast of the domain boundaries reverses as we change from underfocus  to 
overfocus. In the Foucault imaging mode [Fig. 2.2 (b)], by selecting the tilted beam in 
the diffraction pattern using an aperture, one can obtain a dark field image of four 
differently oriented (? x, ? y) domains. Also in Lorentz TEM, one can apply a 
magnetic field along the in-plane and out-of-plane direction by tilting the sample and 
increase the current though the objective lens. 
 
 
 
 
 
 
 
Figure 2.2 Two different methods to observe magnetic domains in Lorentz 
TEM. Schematics for Fresnel imaging (a) and Foucault imaging (b). 
 21 
 
2.2.3 Atomic force microscopy (AFM) 
 
AFM was also used to evaluate the surface morphology by measuring the 
forces between a cantilever tip and the sample surface. The atomic force microscope 
(AFM) has become a very important characterization tool in many fields since G. 
Binnig and co-workers invented it in 1986. The AFM scans over a surface by using a 
fine ceramic or semiconductor tip. Typically, the tip is located at the end of a 
cantilever with a laser beam focused at the end of the cantilever. As the tip is either 
repelled by or attracted to the sample surface, the cantilever deflects which results in 
a change of the laser beam position on a photodiode (see Figure 2.3). The photodiode 
detects the relative position of the laser on the photodiode face. The photodiode is 
composed of two major parts which separately detect normal motions and shear 
motions. The top half and bottom half of the photodiode detect normal and shear 
motions, respectively by the laser displacement along the x axis of the photodiode. A 
plot of the laser deflection versus tip position in the x, y, and z directions on the 
sample surface provides 2 or 3 dimensional surface images. A sensitive piezoelectric 
element controls the position and motion of the tip over the surface of the substrate.  
There are three different modes in AFM. These are the contact mode, the non-
contact mode, and the tapping mode. In the contact mode, the tip responds to 
Coulomb forces with the surface. An image of the surface topography is obtained by 
tracking the deflection of the cantilever as it is scanned over the surface. Maintaining 
the same force between tip and sample maintains the same distance between the tip 
and the surface. The Hooks law and Coulomb forces can be written, respectively. 
 22 
 
H
Fkx=? ,                 (Hook?s Law) 
12
2
0
4
C
qq
F
r??
= ,                        (Coulomb force) 
where k represents the elastic (spring) constant and q
1
, q
2
 are the charges of 
the tip and the sample and r is the distance between the charges. However, in the 
contact mode, the tip can cause damage to the sample. So the non-contact mode was 
developed to investigate samples with soft surfaces. In the non-contact mode, the 
AFM tip is held at a distance in the range 3-15 nm above a surface of the sample, and 
this distance is maintained constant by using the force signal. As the tip is scanned 
over a sample surface, long range attractive forces acting between the tip and the 
sample are detected, and a topographic image is recorded. 
 
 
 
 
 
 
 
Figure 2.3 Schematic of an Atomic Force Microscope 
 
In the non-contact mode, AFM measures van der Waals forces from the 
vibrating tip at a resonance frequency with nano-scale amplitude by detecting the 
change of amplitude. However, in the non-contact mode the data could be inaccurate 
due to weak van der Waals forces. Therefore, the tapping mode, another method, was 
 23 
 
developed and has come to enjoy widespread use. This mode has advantages over the 
two modes mentioned above. In this method, the tip vibrates at a resonance frequency 
with an amplitude of 20 ? 100 nm and taps the surface and detects the change of 
frequency due to the Coulomb force. A cantilever is oscillated at its resonant 
frequency and is in contact with a surface for only a very small fraction of its 
oscillation period. 
In our experiments, AFM was also used to measure magnetic and electrical 
properties. To measure the electrical properties, the tip detects the piezoresponse from 
the surface of a ferroelectric material. This technique is called piezo force microscopy 
(PFM). In the PFM mode, an external voltage with a frequency? ? is applied through 
the tip, which causes the ferroelectric sample under the electric field to vibrate at the 
same frequency due to the converse piezoelectric effect. This vibration then forces the 
AFM tip to oscillate, and the modulated deflection signal is detected using a lock-in 
amplifier. Using PFM, we can obtain quantitative values of d
33
 from the strain loop as 
a function of applied voltage as well as piezoelectric imaging. Also, using a magnetic 
tip, AFM can map local magnetization near the sample surface. This technique is 
known as magnetic force microscopy (MFM). In the MFM mode, magnetic domains 
are imaged from the contour of the magnetic force. 
 
2.2.4 Magnetic and electrical properties measurements 
 
Magnetic properties were also measured by a vibrating sample magnetometer 
(VSM) and a superconducting quantum interference device (SQUID) magnetometer. 
 24 
 
In VSM, a sample is placed in a homogenous magnetic field, and then a magnetic 
moment is induced in the sample. If this sample is made to undergo a sinusoidal 
motion, the vibration induces a magnetic flux change. This in turn induces a voltage 
in a set of pick-up coils. So the magnetic moment determined by a VSM is related to 
the magnetization of the sample by its susceptibility. 
 
 
 
 
 
 
 
 
Figure 2.4 A SQUID device, which consists of two superconductors separated 
by thin insulating layers to form two parallel Josephson junctions. 
 
The SQUID is the most sensitive device which can detect magnetic flux with 
sensitivity as high as ~ 10
-14
 T in terms of magnetic field. A SQUID consists of a 
Josephson junctions which have two superconductors separated by a thin insulating 
layer and acts as a flux-to-voltage transducer (schematically shown in Fig. 2.4). 
For the electrical properties measurements, polarization ? electric field (P-E) 
hysteresis loops were performed using a Radiant Precision LC ferroelectric probe 
station. An HP 4275A impedance/gain analyzer was also used to measure the 
dielectric properties. 
 25 
 
Chapter 3 
3. Multiphase formation in Bi-Fe-O thin films 
 
3.1 Introduction 
 
Although BiFeO
3
 has the advantage that its magnetic and ferroelectric 
transitions are above room temperature, the small magnetization (from canted 
antiferromagnetism) and significant leakage current density are significant problems 
for practical applications. In addition, the high volatility of bismuth gives rise to 
bismuth deficient impurity phases and results in a narrow growth window for pure 
stoichiometric BiFeO
3
. In the Bi-Fe-O system, there are several phases of different 
structures, compositions, and electrical and magnetic properties which can co-
exist.[51,52] The presence of these phases can substantially affect the structural, 
electrical and magnetic properties of the overall material. This chapter presents 
growth conditions that give rise to multiphase formation and the effect of secondary 
phases on the properties of BiFeO
3
 films. Films grown at low oxygen partial pressure 
show the canted antiferromagnetic phase ?-Fe
2
O
3
 embedded in a matrix of BiFeO
3
. 
The ferromagnetic phase, ?-Fe
2
O
3
 is found to precipitate inside the ?-Fe
2
O
3
 grains. 
The magnetization of the Bi-Fe-O films is controlled by the presence of the ?-Fe
2
O
3 
phase.  
The formation of these phases not only changes the magnetic properties of the 
films, it also reduces the misfit strain. The precipitation of ?-Fe
2
O
3 
regions within the 
BiFeO
3 
film reduces the compressive stress due to the smaller volume of ?-Fe
2
O
3 
 26 
 
compared to that of BiFeO
3
. These stain accommodation mechanisms affect the 
electrical properties such as leakage current and electrical polarization. 
 
3.2 Oxygen pressure dependent multiphase formation 
 
We have studied the coexistence of BiFeO
3
 and Fe
2
O
3
 phases (? and ?) with a 
systematically varying degree of mixture controlled by the oxygen partial pressure 
during deposition. In order to fabricate bismuth iron oxide thin films by pulsed laser 
deposition, we used a stoichiometric BiFeO
3
 target and SrTiO
3
 (STO) (001) as the 
substrate, and the oxygen pressure was varied in the range of 0.025 ? 100 mTorr 
while the temperature was kept at 650 
o
C.  
 
 
 
 
 
 
 
 
 
 
Figure 3.1 XRD spectra from BiFeO
3
 films grown at different deposition 
oxygen pressures. 
 
 27 
 
Figure 3.1 shows x-ray diffraction (XRD) spectra of a series of bismuth iron 
oxide thin films deposited under different oxygen pressures. Epitaxial and pure 
BiFeO
3
 films were obtained at oxygen pressures higher than 5 mTorr, and the 
intensity of the (002) diffraction peaks of the BiFeO
3
 increases as the pressure 
increases. At the oxygen pressure of 20 mTorr, single phase BiFeO
3
 shows a 
monoclinically distorted (by about 0.5?) tetragonal structure. The lattice constants of 
the BiFeO
3
 calculated from electron diffraction data (not shown) are a = 0.394 nm, c 
= 0.398 nm. As the oxygen pressure is reduced below 5 mTorr, peaks corresponding 
to the Fe
2
O
3
 phases [(012) of ?-Fe
2
O
3
 and (024) of ?-Fe
2
O
3
] appeared, and the 
intensity of these peaks increases as the oxygen pressure decreases. This transition in 
the dominant phase reflects the change in the amount of Bi incorporation in the films 
due to the high volatility of Bi. 
Fig. 3.2 shows TEM images of 200 nm thick bismuth iron oxide thin films 
grown at a low oxygen partial pressure. BiFeO
3
 and Fe
2
O
3
 phases were found to have 
grown epitaxially with respect to the substrate as well as with respect to each other in 
the lateral direction in a columnar manner. We have found that the Fe
2
O
3
 domains 
grow with a mixture of ? and ? phases, with ? being the predominant phase. Fig. 3.2 
(a) is a cross-sectional bright field image of a film showing the presence of BiFeO
3
 
and ?-Fe
2
O
3
. This film was grown at 1 mTorr oxygen partial pressure. The 
corresponding SAD pattern from the sample was identified as the superposition of the 
(100) pattern from BiFeO
3
 and that of the rhombohedral crystal structure of ?-Fe
2
O
3
 
(space group: R3
_
c) with lattice constants of a = b = 0.505 nm, c = 1.319 nm, which 
are in good agreement with the published structure (i.e. JCPDS #33-0664). We have 
 28 
 
also confirmed this result with energy dispersive x-ray spectroscopy (EDS). The 
epitaxial relationship of BiFeO
3
 and ?-Fe
2
O
3
 was determined from the electron 
diffraction patterns of the cross-sectional view (not shown here) and the plan view 
(Fig. 3.2 (c)): [001]
BiFeO3
 is seen to be almost parallel (within 0.2?) to [001]
STO
 
(corresponding to the out-of-plane direction). However, [012] of ?-Fe
2
O
3
 deviates by 
2.1? from [001]
BiFeO3
. Fig. 3.2 (b) is a plan view TEM image taken from the same 
sample grown at 1 mTorr. The interface between BiFeO
3
 and ?-Fe
2
O
3
 is parallel to 
the [102
_
] and [11
_
2] directions of ?-Fe
2
O
3
 [see Fig. 3.2 (b)]. From the electron 
diffraction patterns, [100]
BiFeO3
 is seen to be almost parallel to [100]
STO
 
(corresponding to the in-plane direction), with the in-plane lattice parameters of 0.394 
nm and 0.3905 nm, respectively (see Fig. 3.2 (d)). However, [102
_
] of ?-Fe
2
O
3
 makes 
an angle of ~2.1? with respect to the [100]
BiFeO3.
 The measured interplanar distance 
between (000) and (102
_
) of ?-Fe
2
O
3
 is 0.370 nm, which is slightly larger than a 
published result (0.3684 nm, JCPDS #33-0664) for bulk ?-Fe
2
O
3
. These results can 
be explained as a consequence of an in-plane tensile stress imposed by bismuth iron 
oxide on the ?-Fe
2
O
3
. The increase of the (102
_
) interplanar distance gives rise to a 
decrease in the (012) interplanar distance, which was measured as 0.364 nm. The 
lattice mismatch strain between BiFeO
3
 and ?-Fe
2
O
3
 is 6.4 %. After relaxation, a 
residual strain of 4.9 % in the Fe
2
O
3
 regions was calculated from our TEM and XRD 
results. 
 
 
 29 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.2 (a) Cross-sectional and (b) plan-view TEM images taken from a 
BiFeO
3
 film grown at 1 mTorr oxygen pressure, and (c) diffraction pattern and (d) 
indexing of plan-view diffraction pattern. e) Cross-sectional TEM image and electron 
diffraction (inset) from a BiFeO
3
 film grown at 0.1 mTorr oxygen pressure. 
 
Fig. 3.2 (e) shows a BiFeO
3
 thin film fabricated at 0.1 mTorr. At this low 
pressure, the dominant phase is ?-Fe
2
O
3 
which forms a continuous layer on a thin 
layer (? 5 nm) of BiFeO
3
 above the substrate. The fact that ?-Fe
2
O
3
 appears to have 
 30 
 
grown above BiFeO
3
 indicates that it nucleated at the top surface of the film as Bi 
atoms vaporized following deposition. The SAD pattern in the inset of Fig 3.2 (d) 
taken from the iron oxide area was indexed to ?-Fe
2
O
3
, and the measured angle (~ 
86.5?) between (012) and (102
_
) agrees well with the calculated angle (86.0?) in 
rhombohedral bulk ?-Fe
2
O
3
. 
The magnified TEM images from cross-sectional and plan-view specimens in 
Figs. 3.3 (a) and (b) show ?-Fe
2
O
3
 in the form of precipitates with a plate-like shape 
embedded in the ?-Fe
2
O
3
 phase. From indexing the SAD patterns (see Fig. 3.3 (c)), 
we obtained the following preferred in-plane epitaxial relations among BiFeO
3
, ?-
Fe
2
O
3
, and ?-Fe
2
O
3
; [100]
BiFeO3 
// [102]
?
 // [120]
?
, and [010]
BiFeO3
 // [11
_
2]
?
 // [21
_
1
_
]
?
. 
The ?-Fe
2
O
3
 also showed other orientations with respect to BiFeO
3
 and ?-Fe
2
O
3
 in 
other areas of the sample. This is probably due to the fact that PLD is a non-
equilibrium growth technique. 
The SAD pattern and the schematic in Fig. 3.3 (c) are from one of the Fe
2
O
3
 
regions where ?-Fe
2
O
3
 and ?-Fe
2
O
3 
coexist. The out-of-plane and the in-plane 
parameters in ?-Fe
2
O
3
 as measured from a high resolution TEM image (see Fig. 3.3 
(d)) were d(120) = 0.37 nm and d(21
_
1
_
) = 0.34 nm, respectively, which are in good 
agreement with published values (i.e. JCPDS #39-1346). The lattice mismatch 
between ?- and ?-Fe
2
O
3
 is 8.1 %. Some distortion of the ?-Fe
2
O
3
 lattice and strain 
concentration around ?-Fe
2
O
3
 was observed due to the large lattice mismatch. 
  
 
 31 
 
 
 
 
 
Figure 3.3 TEM bright field images and SAD pattern of bismuth iron oxide 
thin films fabricated on (001) SrTiO
3
 substrates at 1 mTorr Po
2
. (a) Magnified cross-
section (left) and (b) plan-view (right) images showing ?-Fe
2
O
3
 precipitates and (c) 
SAD pattern and corresponding schematic of ?-Fe
2
O
3
 and ?-Fe
2
O
3 
phases in the film 
deposited at 1 mTorr. (d) High resolution cross sectional TEM image of ?-Fe
2
O
3
 in ?-
Fe
2
O
3
 matrix.   
 
 
 32 
 
3.3 Change of magnetic properties of Bi-Fe-O by multiphase 
formation 
 
3.3.1 Formation of ?-Fe
2
O
3
 and Fe
3
O
4
 
 
In some BiFeO
3
 thin films, improved magnetic properties were observed in an 
earlier study caused by a change of spiral magnetic spin through epitaxial 
constraint.[48] However, in our study (see section 3.2), we found that the formation 
of ferri (ferro) magnetic phase, ?-Fe
2
O
3
 embedded in antiferromagnetic ?-Fe
2
O
3
 
grains in films grown at relatively low oxygen pressure or higher temperature is the 
dominant reason for the observed enhanced magnetization rather than epitaxial 
constraint by the substrate.[47,52] The saturation magnetization of various BiFeO
3 
? 
Fe
2
O
3
 films were measured by a SQUID magnetometer at room temperature.  
 
 
 
 
 
 
 
 
Figure 3.4 Oxygen pressure dependence of saturation magnetization in Bi-Fe-
O films. As oxygen pressure decreases, magnetization increases.  
 33 
 
As shown in Fig. 3.4, the magnetization observed in the BiFeO
3
 films grown at 
oxygen partial pressures higher than 5 mTorr was about 2 emu/cc, which is consistent 
with the value reported for bulk BiFeO
3
. However, as the deposition pressure 
decreases, the magnetization increased up to ~ 80 emu/cc. The enhancement was 
caused by the formation of ?-Fe
2
O
3
 at relatively low oxygen partial pressure condition 
as shown below. ?-Fe
2
O
3 
(hematite)
 
is known to exhibit weak ferromagnetism below 
the N?el temperature[53] (T
N
 ? 950 K). In contrast, ?-Fe
2
O
3
 is known to be a 
ferro(ferri)magnetic phase with a strong magnetization at room temperature.[54] 
 
 
 
 
 
 
 
 
 
Figure 3.5 Room temperature saturation magnetization (emu/cc) of Bi-Fe-O 
films as a function of thickness for films grown at 5 mTorr and 20 mTorr. 
 
Figure 3.5 shows the change in magnetization with thickness for films grown at 
5 mTorr and 20 mTorr oxygen partial pressures. Pure BiFeO
3
 films grown at 20 mTorr 
having thickness of 35 nm showed magnetization of about 5 emu/cc, which is smaller 
than the results of Wang et al.,[20], but is in agreement with other recent 
 34 
 
measurements.[65] This slightly enhanced magnetization could result from the 
heteroepitaxial strain effect in the canted antiferromagnetic film. The magnetic moments 
of Fe in BiFeO
3
 are ferromagnetically aligned on the (111) plane and 
antiferromagnetically switched along the [111] direction. Canting of the Fe magnetic 
moment allows weak ferromagnetism in BiFeO
3
. Ederer and Spaldin[66] found through 
density functional theory calculations that weak ferromagnetic moments (0.1 ?
B
 per unit 
cell for about 1
o
 canting from the collinear direction) are induced by a structural 
distortion. Also, as the thickness of the film increases, magnetization in the films with 
multiple phases is expected to decrease faster than for pure BiFeO
3
, due to the 
nucleation of ?-Fe
2
O
3
. However, this reduction of magnetization might be partially 
compensated by the nucleation of ?-Fe
2
O
3
. The saturation magnetization decreases 
with increasing oxygen partial pressure up to 20 mTorr and remains constant above 
this pressure when the films become pure BiFeO
3
. 
The formation of Fe
2
O
3
 in BiFeO
3
 films affected the electrical properties such 
as leakage current density and electrical polarization as discussed in Section 3.5. 
Measurements of P-E hysteresis loops were carried out from films grown on SrRuO
3
 
buffered SrTiO
3
 or 5 % Nb-doped SrTiO
3
 substrates. After growing BiFeO
3
 films, 
optical lithography and deposition of Pt were carried out to deposit a top electrode. 
Electrical hysteresis loops were measured from the pure BiFeO
3
 and multiphase Bi-
Fe-O films, and the multiphase film showed larger polarization (~ 65 ?C/cm
2
) than 
the pure BiFeO
3
 film (~ 2 ?C/cm
2
) (as shown in section 3.5 below). Also, lower 
leakage currents were obtained from the multiphase films than from pure BiFeO
3
. 
The observed different electrical properties between the pure BiFeO
3
 thin film and the 
 35 
 
multiphase BiFeO
3
 thin films are somehow related to the different strain relaxation 
mechanism given in section 3.4 and/or difference in stoichiometry presented in 
section 3.5.3. 
 
3.3.2 Tuning of magnetic properties of Bi-Fe-O thin film 
 
To enhance the magnetic properties of multiphase Bi-Fe-O film with the 
columnar structure, we explored the possibility of transforming ?-Fe
2
O
3
 
(antiferromagnetic phase) grains to ?-Fe
2
O
3
 (ferromagnetic phase) grains. Typically, 
Fe
2
O
3
 has two different crystal structures. One is a spinel structure (maghemite, ?-
Fe
2
O
3
), and the other is a corundum structure (hematite, ?-Fe
2
O
3
). In bulk, it is 
known that there is an irreversible transformation from ?-Fe
2
O
3
 to ?-Fe
2
O
3
 occurring 
at temperatures above 400 
o
C via restacking of the oxygen layers and displacement of 
the Fe ions located in the interstices rather than recrystallization. This transformation 
occurs on the close-packed planes of ?- and ?-Fe
2
O
3
 without diffusion, and thus the 
crystal orientation relationship between ?- and ?-Fe
2
O
3
 in the bulk was reported as 
(0001)
?
 // (111)
?
, [011
_
0]
?
 // [11
_
0]
?
. In our experiments, however, we observed the 
coexistence of ?-Fe
2
O
3
 and ?-Fe
2
O
3
 at room temperature, even though ?-Fe
2
O
3
 is the 
stable low temperature phase. As described in Fig. 3. 6, we have devised a two-step 
process to enhance the transformation of ?-Fe
2
O
3
 to ?-Fe
2
O
3
.[67] The first stage is 
hydrogen reduction to transform ?-Fe
2
O
3
 into Fe
3
O
4
, and the second stage is 
oxidation to transform Fe
3
O
4
 to ?-Fe
2
O
3
. The reduction was conducted by flowing a 
 36 
 
mixture gas of Ar and H
2
 (10
Ar
:1
H2
) for 1 hour while the film was kept at 300 
o
C. This 
was followed by oxidation at 300 
o
C in an oxygen ambient for 1 hour. 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.6 Top left image is a cross-sectional image of a Bi-Fe-O film grown 
at 5 mTorr oxygen pressure with schematic of FE and FM regions before (right) and 
after (left) annealing. Right schematic is the transformation of ?-Fe
2
O
3
 to ?-Fe
2
O
3
 
upon annealing. Annealing procedures to transform ?-Fe
2
O
3
 into ?-Fe
2
O
3
 includes 
hydrogen flowing to transform ?-Fe
2
O
3
 to Fe
3
O
4
, which is followed by oxidation to 
transform Fe
3
O
4
 to ?-Fe
2
O
3
. 
 
The change of magnetization after the annealing process is shown in the Fig. 
3.7. The annealed film showed increased magnetization (~ 6 emu/cc) due to the 
increase of the fraction of ?-Fe
2
O
3
 (less than 10 %). The magnetic domain image of 
the annealed multiphase Bi-Fe-O film was acquired using MFM to confirm the 
existence of magnetic domains (?-Fe
2
O
3
) after annealing. The MFM image in Fig. 3.7 
(a) shows the randomly oriented magnetic spins which originate from the ?-Fe
2
O
3
 
 37 
 
phase in the absence of an external magnetic field, and the MFM image in Fig. 3.7 (c) 
shows the switching of magnetic spins along the out-of plane direction after poling 
with a field of 20 kOe. From these results, we conclude that the magnetic properties 
of the film can be controlled by converting ?-Fe
2
O
3
 into ?-Fe
2
O
3
 through the 
described procedure. In fact, we found that by adjusting the annealing time and the 
concentration of hydrogen in the gas mixture, the fraction of ?-Fe
2
O
3
 and 
magnetization of the film can be controlled. The magnetization of Bi-Fe-O film can 
be increased up to ~ 160 emu/cc when all of ?-Fe
2
O
3
 transforms to ?-Fe
2
O
3
. 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.7 Magnetic force microscopy (MFM) images of an annealed 
multiphase film (a) before and (c) after magnetizing the film perpendicular to the 
surface by applying 20 kOe magnetic field. (b) Hysteresis loops of as grown and 
annealed multiphase Bi-Fe-O films. (d) Schematic of the microstructure of Bi-Fe-O 
multiferroic nanocomposites. 
 
 38 
 
3.4 Change of misfit strain in BiFeO
3
 thin films by multiphase 
formation 
 
3.4.1 Thickness dependent misfit strain 
 
To investigate the microstructural evolution in the Bi-Fe-O system and its 
effect on the structural and magnetic properties of the films, Bi-Fe-O thin films of 
different thicknesses (15 ? 500 nm) were epitaxially grown on SrTiO
3
 (001) 
substrates by pulsed laser deposition under various oxygen partial pressures.  
Figures 3.8 (a) and (b) show, respectively, x-ray diffraction spectra obtained 
from films of different thicknesses grown at an oxygen partial pressure of 5 mTorr 
and corresponding TEM cross-sectional images. The 35 nm thick film grown at 5 
mTorr oxygen partial pressure shows the pure epitaxial BiFeO
3
 phase with the [001] 
orientation and a slightly larger lattice parameter (c = 0.407 nm) than that of the pure 
300 nm thick BiFeO
3
 film grown at 20 mTorr. As the film thickness increases, the 
intensity of the (002) diffraction peaks of the BiFeO
3
 increases, and the peak shifts to 
a higher 2? angle. This indicates that the out-of-plane lattice parameters of BiFeO
3
 
decreases with increasing thickness. We also observed peaks corresponding to the 
(024) plane of the ?-Fe
2
O
3
 phase and the (240) plane of the ?-Fe
2
O
3
 phase. The ?-
Fe
2
O
3 
phase has a rhombohedral crystal structure (space group: R3
_
c) with lattice 
constants of a = b = 0.505 nm, c = 1.319 nm, and the ?-Fe
2
O
3
 phase has the cubic (or 
inverse spinel) crystal structure that belongs to the space group P4132 (or Fd3
_
m) with 
 39 
 
lattice parameter a = 0.83515 nm. The intensity of the peaks corresponding to ?- and 
?-Fe
2
O
3
 increases with increasing film thickness. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.8 (a) X-ray diffraction spectra (in log scale) and (b) TEM cross-
sectional images (left) and EDS spectra (right) of Bi-Fe-O films grown at 5 mTorr 
with different thickness. Nucleation of impurity phases starts at ~ 50 nm. The 
intensity of the peak labeled Fe
2
O
3
 in (a) corresponding to the impurity phases 
increases with increasing film thickness. 
 
From the TEM cross-sectional images shown in Fig. 3.8 (b), we can see that 
the secondary phase, ?-Fe
2
O
3
, forms in the surface region of films of thickness of 50 
nm, and then it grows to have a columnar structure with continued deposition. The 
EDS spectra in Fig. 3.8 (b) obtained from the bright and dark columns of the film 
 40 
 
confirmed the chemical composition of the separated phases. The EDS spectrum from 
the bright columns shows little bismuth content (5.5 at%), and the atomic ratio 
between iron (38.5 at%) and oxygen (56 at%) is 1:1.45, which is consistent with 
Fe
2
O
3
. The EDS spectrum from the dark columns, shows a 1: 1.1 ratio between 
bismuth (24 at%) and iron (26.3 at%), which is consistent with BiFeO
3
. 
To further understand the misfit strain relaxation mechanism, multiphase Bi-
Fe-O films with different thickness (35 ? 500 nm) were compared to a single phase 
BiFeO
3
 film with a gradual change in thickness. The change of out-of plane lattice 
parameter with different thicknesses was obtained from the pure BiFeO
3
 film and 
multiphase Bi-Fe-O film (see Fig. 3. 9).  
 
 
 
 
 
 
 
 
 
 
 
Figure 3.9 Out-of-plane lattice parameters in pure BiFeO
3
 and BiFeO
3
 with 
secondary phases as a function of film thickness. 
 
 41 
 
The region of the film with 35 nm thickness showed a highly strained area. As 
the film thickness increases from 30 to 300 nm, the out-of-plane lattice parameter of 
pure BiFeO
3
 film gradually decreases from 0.406 to 0.398 nm. However, in the 
multiphase films, the out-of-plane lattice parameter decreases from 0.407 to 0.396 nm 
with increasing film thickness from 35 to 500 nm. This result confirms that the ?-
Fe
2
O
3
 domains that form on the surface region of the film for films of thickness 
above 50 nm (grown at low oxygen partial pressure) enhance the relaxation of misfit 
strain in BiFeO
3
 columns. As a result, the total misfit strain reduction is from 0.51 % 
(pure BiFeO
3
) to 0.25 % (mixture of BiFeO
3
 and Fe
2
O
3
) for films with thickness of 
300 nm.  
 
3.4.2 Gradient thickness samples 
 
The existence of secondary phases can affect the relaxation of misfit strain in 
BiFeO
3
 and result in significant changes of the ferroelectric and ferromagnetic 
properties of BiFeO
3
 films. To compare the out-of-plane lattice parameter of 
multiphase Bi-Fe-O and pure BiFeO
3
 with different thicknesses, BiFeO
3
 films with 
gradual change in thickness were fabricated by the gradient deposition technique 
using a shadow mask. Oxygen partial pressure of: 20 mTorr was selected to fabricate 
pure BiFeO
3
 films, and 5 mTorr to make multiphase Bi-Fe-O films. The thickness of 
these films varied in the range of 30 ? 300 nm.  
 
 
 42 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.10 Superpositions of x-ray microdiffraction scans from regions of 
films with a thickness gradient grown at (a) 20 mTorr and (b) 5 mTorr. The thickness 
varied from 30 to 300 nm. Schematics of the misfit strain relaxation mechanism in (c) 
pure BiFeO
3
 film and (d) multiphase Bi-Fe-O film are suggested. Multiphase film 
shows coexistence of highly strained region (A) and fully relaxed region (B). 
 
Figures 3.10 (a) and (b) show superposition of x-ray microdiffraction scans 
from different regions of the pure BiFeO
3
 film (a) and multiphase Bi-Fe-O film (b) 
with different thicknesses. In Fig. 3.10 (a), the position of the (002) peak of BiFeO
3
 
shifts from 2? = 44.6
o
 to 45. 6
o
 as the film thickness increases. In contrast, the spectra 
in Fig. 3.10 (b) show a broad peak at ~ 2? = 44.5
o
 in the thin regions of the sample 
which evolves into two peaks as the film thickness increases; the thicker regions 
 43 
 
show a peak at ~ 45.7
o
 which corresponds to BiFeO
3
. However, a second peak at ~ 
44.5? remains with weaker intensity even for thickness up to 300 nm. The observation 
of two peaks in Fig. 3.10 (b) is consistent with the coexistence of a strained region 
and a relaxed region in the film. More evidence of the two regions was obtained from 
TEM images as shown below. These results indicate that the multiphase film has a 
different relaxation mechanism than the pure BiFeO
3
 films. Typically, the misfit 
strain between a film and a substrate is relaxed by the generation of misfit 
dislocations at the interface with increasing film thickness.[68,69] As shown in Fig 
3.10 (c), pure BiFeO
3
 showed typical relaxation mechanisms in which the strain 
gradually relaxes as the film thickness increases. The multiphase Bi-Fe-O film with 
small thickness showed a highly strained film as expected. However, the thicker 
region of the film showed both a relaxed region [represented by the x-ray peaks 
labeled ?B? in Fig. 3.10 (b)] and a highly strained region [labeled ?A? in Fig. 3.10 
(b)] in the film even though the thickness of the film was above the critical thickness 
for relaxation to take place [see also Fig. 3.10 (d)]. This observation can be explained 
by the formation of the ?-Fe
2
O
3
 regions, which appeared in the surface region of the 
film with thickness around 50 nm. The transformation to ?-Fe
2
O
3
 in the surface 
region of the film reduces the thickness of the BiFeO
3
 underneath. Therefore, the area 
below ?-Fe
2
O
3
 [marked A in Fig. 3.8 (b)] showed higher misfit strain and a lower 
misfit dislocation density than the region with pure BiFeO
3
 [marked B in Fig. 3.8 (b)] 
as determined by high resolution TEM images (not shown). In addition, the formation 
of ?-Fe
2
O
3
 regions enhances the relaxation of misfit strain in the BiFeO
3
 columns 
between ?-Fe
2
O
3
 columns due to the reduction of unit volume as explained below. 
 44 
 
The reduction of volume partially releases the compressive strain along the in-plane 
direction. 
The observation of a broad peak in the XRD spectra in Fig. 3.10 (b) for 
thicknesses above ? 50 nm indicates that the bottom portion of the films under the 
Fe
2
O
3
 region (region A) in Fig. 3.10 (d) remains strained. In order to understand the 
higher degree of stain relaxation in the films with Fe
2
O
3
, we estimate the change in 
volume in these regions. 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.11 Schematics of two different strain relaxation mechanisms. In 
multiphase BiFeO
3
, in-plane compressive stress of BiFeO
3
 columns is relaxed by the 
formation of Fe
2
O
3
 phase with smaller volume. AFM image from the surface (upper 
right) also confirmed smaller volume of Fe
2
O
3
 phase from height difference between 
columns 
 45 
 
A reduction in volume of ? -19 % by the formation of the Fe
2
O
3
 phase allows 
expansion of BiFeO
3
 (which is under compressive strain with the substrate) which in 
turn reduces the misfit strain. Therefore, the formation of ?-Fe
2
O
3
 by evaporation of 
the more volatile Bi metal at relatively low oxygen partial pressure enhances the 
relaxation of the compressive stress in BiFeO
3
. The relaxation is accelerated as the ?-
Fe
2
O
3
 volume fraction increases (see Fig. 3.11). 
 
3.5 Change of ferroelectric properties by multiphase formation 
 
After the initial discovery of the increase in electrical polarization (~ 55 
?C/cm
2
) in BiFeO
3
 thin films, the misfit strain and structural distortion of the BiFeO
3
 
lattice were suggested as the reasons for the large polarization.[20] The crystal 
structure in the films changed from the bulk rhombohedral to monoclinic due to the 
epitaxial strain induced by the substrate. There have also been a number of reports on 
thin films of BiFeO
3
 with large electrical polarization, but the origin of the large 
polarization in these cases has not been unambiguously established. We also have 
fabricated numerous BiFeO
3
 films with different conditions, and found that the 
BiFeO
3
 films which are grown at relatively lower oxygen partial pressure and include 
Fe
2
O
3
 phase show large polarization. These films also show clear contrast change in 
switching of ferroelectric domains in PFM images as shown below. To further 
understand the role of multiphase formation on the electrical properties of BiFeO
3
 films, 
we performed a systematic study of the electrical properties of BiFeO
3
 films to 
understand the origin of the observed large polarization in an attempt to control and 
 46 
 
improve the properties as shown below. 
 
3.5.1 Ferroelectric properties in columnar BiFeO
3
 ? Fe
2
O
3
 
 
To study the origin of the larger polarization in BiFeO
3
, we compared two 
different types of BiFeO
3
 films which were grown at 2 mTorr and 20 mTorr, because 
the films grown 20 mTorr are yellow in color and show weak polarization (less than 2 
?C/cm
2
), while the films grown at 2 mTorr are red in color and exhibit large 
polarization (~ 72 ?C/cm
2
) [see Fig. 3.12]. 
 
 
 
 
 
 
 
 
 
Figure 3.12 Electrical polarization vs. applied electric field from the samples 
grown at (a) 20 mTorr and (b) 2 mTorr. (c) Leakage current density measured from 
the two samples. 
 
Typically, the films grown at low oxygen pressure have more oxygen 
vacancies and bismuth deficiencies in the films and result in poor insulating 
properties, but as shown in Fig. 3.12 (c), the columnar BiFeO
3
 showed robust 
 47 
 
insulating property even though the film was synthesized at low oxygen partial 
pressure. 
To characterize the structure of the films, TEM and AFM were used. From 
our plan-view TEM images shown in Figs. 3.13 (a) and (b), the film grown at 20 
mTorr oxygen pressure [Fig. 3.13 (a)] is pure BiFeO
3
, but the film grown at 2 mTorr 
oxygen pressure [Fig. 3.13 (b)] has columnar structure of Bi deficient phase, Fe
2
O
3
 
(bright region) and BiFeO
3
 (dark region) as was presented in sections 3.1, 3.2, and 3.3. 
 
 
 
 
 
 
 
 
 
 
Figure 3.13 Bright field TEM images from BiFeO
3
 films grown at (a) 20 
mTorr and (b) 2 mTorr. Surface morphology of the films grown at (c) 20 mTorr and 
(d) 2 mTorr measured by AFM. 
 
AFM images from both samples showed similar square shaped grains. 
However, the squares in Fig. 3.13 (d) represent the columnar structure of multiphase 
BiFeO
3
, while the squares in Fig. 3.13 (c) correspond to roughness or texture of pure 
BiFeO
3
 grains. 
 48 
 
3.5.2 Ferroelectric domain switching analysis using piezo force 
microscopy (PFM) 
 
Figures 3.14 and 3.15 show switching domain images and piezoelectric 
response from the BiFeO
3
 films grown at 20 mTorr and 2 mTorr, respectively, using 
PFM. The multiphase BiFeO
3
 film grown at 2 mTorr [Fig. 3.15 (b)] shows more clear 
contrast of switching in polarization than the pure BiFeO
3
 film grown at 20 mTorr 
[Fig. 3.14 (b)]. 
 
 
 
 
 
 
 
 
 
Figurer 3.14. (a) 5 ? 5 ?m
2
 height image and (b) ferroelectric domain 
switching images from the BiFeO
3
 thin film grown at 20 mTorr oxygen pressure. To 
obtain image showing switching of ferroelectric domains a + 10 V and ? 10 V bias 
were alternately applied on 3 ? 3 ?m
2
 and 1 ? 1 ?m
2
 area, respectively. (c) The 
piezoelectric coefficient (d
33
) was measured from strain loop. 
 
As shown in the AFM image in Fig. 3.14 (a), the BiFeO
3
 film grown at 20 
mTorr showed texture along [100] and [010] orientation of SrTiO
3
. From the 
 49 
 
switching ferroelectric domain image in Fig. 3.14 (b), when the electric field in 
reversed from + 10 V to ? 10 V and ? 10 V to + 10V, the pure BiFeO
3
 film showed 
non-uniform contrast and poor switching probably due to the large leakage current in 
the sample [see Fig. 3.12 (c)]. Also the d
33
 loop measurement showed asymmetric 
shape indicative of a poor ferroelectric. 
However, the columnar BiFeO
3
 films which have secondary phase (Fe
2
O
3
) 
exhibit clear change of contrast with reversing dc bias [see Fig. 3.15 (b)]. Also 
interestingly, the switching of ferroelectric domains show more clearly in the 
interface regions between BiFeO
3
 and Fe
2
O
3
 columns than in the BiFeO
3
 regions [see 
Fig. 3.15 (d)]. Furthermore, the strain loops obtained from the edge region showed a 
more symmetric shape than that from the middle of the BiFeO
3
 domains. These facts 
indicate that the BiFeO
3
 in the middle and edge of column might have different state 
of misfit strain, and/or chemical composition. 
As shown in section 3.4.1, multiphase formation aids in the strain relaxation 
mechanism in Bi-Fe-O films. Furthermore, the formation of Fe
2
O
3
 induced non-
uniform strain relaxation of the surrounding BiFeO
3
 phase. The strain in the BiFeO
3
 
columns between the Fe
2
O
3
 columns was reduced due to the reduction in volume 
upon the formation of Fe
2
O
3
. Therefore, strain relaxation can be one reason for the 
large polarization. Furthermore, the observed large polarization might be an intrinsic 
property of BiFeO
3
. Recently, there has been an attempt to reduce the strain in 
SrRuO
3
 films and obtain a single-domain variant of the SrRuO
3
 by using DyScO
3
 
substrates which have almost the same lattice constant as SrRuO
3
,[55] These films 
showed increase in polarization. Their results also suggest that strain relaxation could 
 50 
 
play a role in influencing the polarization. In addition, large polarization has been 
reported from bulk single crystal BiFeO
3
 in recent papers.[56] 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.15. (a) 5 ? 5 ?m
2
 AFM height image and (b) switching ferroelectric 
domain image from the BiFeO
3
 film grown at 2 mTorr. Magnified (c) AFM height 
image and (d) corresponding PFM image of switching domains from the region 
marked in (b) d
33
 versus applied voltage obtained from (e) middle and (f) edge of a 
BiFeO
3
 column. 
 
However, there still exists another possible reason for the large polarization in 
films containing secondary phases, which is the local change of stoichiometry in the 
film. Volatile Bi can easily evaporate from the surface of BiFeO
3
 during growth, and 
result in the formation of Bi deficient Fe
2
O
3 
phase. In this procedure, the BiFeO
3
 
region surrounding the Fe
2
O
3
 regions could also be off-stoichiometric and change the 
 51 
 
volume of the unit cell and the Fe ion valence from 2+ to 3+. This possibility is 
investigated by electron energy loss spectroscopy as discussed in the next section. 
 
3.5.3 Chemical analysis on columnar BiFeO
3
 ? Fe
2
O
3
 thin films using 
EDS and EELS in TEM 
 
We used EDS and EELS, for the chemical analysis on the columnar structure 
of a BiFeO
3
 thin film grown at 2 mTorr. The dark field scanning TEM (STEM) image 
in Fig. 3.16 (a) and EDS mapping image for Bi-L peak in Fig. 3.16 (b) identified the 
bright columns in an STEM image to be BiFeO
3
. Therefore, an EELS line scanning 
was performed on a bright column in an STEM image for chemical information of 
BiFeO
3
 across the interface between BiFeO
3
 and Fe
2
O
3
. The Fe valence state (3+ and 
2+) was investigated from EELS spectra acquired at ? 3 M magnification measured 
point by point across the interface between BiFeO
3
 and Fe
2
O
3
 columns. The ratio of 
Fe
3+
 and total Fe was calculated from the ratio between Fe L
3
 and Fe L
2
 peaks (Fe
3+
/ 
?Fe) in the EELS spectra and are given in Fig. 3.16 (d). The values show the position 
at which each spectrum was collected. Higher values indicate a higher amount of Fe
3+
 
state. The Fe
3+
/ ?Fe ratio decreases near the interface indicating an increase of Fe
2+
 
and O
2-
 vacancies in the boundary area between BiFeO
3
 and Fe
2
O
3
. The variation in 
Fe
2+
, Fe
3+
 and O
2-
 could be responsible for the increase in polarization in BiFeO
3
 thin 
films. A more detailed study of EELS should be followed to understand the role of 
Fe
2+
 and O
-
 vacancies in the ferroelectric properties of BiFeO
3
 as discussed in the 
future work section of this thesis. 
 52 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.16. (a) Scanning TEM dark field image and (b) EDS elemental 
mapping of Bi-L edge. (c) Magnified dark field STEM image and (d) EELS line scan 
for Fe
2+
/?Fe
3+
.[118] 
 
3.6 Multiferroic properties in polycrystalline BiFeO
3
 thin films 
 
In order to obtain the ferroelectric properties of unstrained BiFeO
3
 films, the 
ferroelectric properties of polycrystalline multiferroic bismuth ferrite thin films were 
investigated. These films have no strain effects or impurities which can be easily 
introduced in epitaxial single crystal thin film growth due to the relatively high 
growth temperature and the epitaxial constraint from the substrate. We observed 
relatively large polarization (P
S
 = 106 ?C/cm
2
) in polycrystalline thin films grown at 
400 
o
C which suggests an intrinsically excellent ferroelectric property of BiFeO
3
. In 
 53 
 
particular, effective switching of ferroelectric domains during poling of 
polycrystalline BiFeO
3
 was observed only in films with nano-grains probably 
containing single domains. The low density of domain boundaries facilitates domain 
switching. Also, post growth annealing of films grown at 300 
o
C produced nucleation 
of nano-sized grains from non-crystallized bismuth ferrite with increase in saturation 
polarization (P
S
) from ~ 8 (as grown) to 82 (after annealing) ?C/cm
2
. 
The recent discovery of increase in polarization and magnetization in BiFeO
3
 
thin films [20] and bulk [117] have given BiFeO
3
 a new outlook, especially since 
bulk BiFeO
3
 had been known to exhibit weak ferroelectricity (P
S
 ~ 3.5 ?C/cm
2
 below 
T
C
 = ~ 1100 K) and canted antiferromagnetism (~ 2 emu/cc below T
N
 = ~ 640 K) at 
room temperature.[21-24] In recent reports, the misfit strain in the film induced from 
the lattice mismatch with the substrate as well as structural distortions from the 
transformation of rhombohedral to tetragonal crystal structure in BiFeO
3
 thin films 
were believed to be responsible for the increase in both electrical polarization and 
magnetization. However, we observed that the increased magnetization is due to the 
presence of ferromagnetic phases (Fe
3
O
4
 and ?-Fe
2
O
3
) in the films resulting from the 
high volatility of bismuth.[76-78] The actual cause for the increase in polarization in 
BiFeO
3
 films is still not fully understood, although very recent reports of bulk BiFeO
3
 
suggested that the high dielectric constant and polarization are intrinsic to 
BiFeO
3
.[79] 
There have also been reports of weak ferroelectricity of BiFeO
3
 [23,80,81] 
based on unsaturated P-E hysteresis curves (P
S
 ~ 3.5 ?C/cm
2
) measured from single 
crystalline BiFeO
3
. These investigations have attempted to solve the problem of 
 54 
 
applying insufficient electric field to bulk BiFeO
3
 and the issue of large leakage 
current found in earlier studies of bulk BiFeO
3
. Recent studies reported larger 
polarizations in single crystalline films with reduced misfit strain and in single 
crystalline bulk BiFeO
3
.[82,83]  
In spite of much recent studies of the ferroelectric properties of BiFeO
3
, its 
intrinsic properties are not well understood because of the complex dependence of its 
properties on the presence of impurities, strain and/or defects. In the current study, we 
systematically investigated the structural evolution of polycrystalline BiFeO
3
 films 
and their intrinsic properties. The films used for our study were non-epitaxial to 
reduce the effects of strain with the substrate and grown at low temperatures to avoid 
the formation of impurity phase. More importantly, the work in this section explores 
the effects of grain size on ferroelectric domain switching which is crucial for 
practical applications. Moreover, polycrystalline BiFeO
3
 thin films grown at low 
temperature have the advantage over single crystalline epitaxial films because they 
are not clamped by the substrate and are less likely to contain defects such as, oxygen 
vacancies produced to compensate for the evaporation of volatile bismuth atoms 
when the films are grown at high growth temperatures. In this current work, we 
observed switching of ferroelectric domains as well as large polarization in nano-
grain polycrystalline BiFeO
3
 films. The nano-grain structure was specially chosen in 
an attempt to achieve a statistically uniform domain size distribution and ultra high 
density which are requirements in the ferroelectric memory storage devices, such as 
the ferroelectric random access memory (FeRAM) or the probe base data storage 
(PBDS) applications.[84-86] 
 55 
 
 
3.6.1 Size effects on electrical polarization and electrical properties 
 
For the systematic study of the relationship between the domain size and 
ferroelectric properties of polycrystalline BiFeO
3
, we deposited a series of films using 
PLD at low temperatures in the range 300 ? 500 
o
C. The microstructure of the films 
was characterized by X-ray diffraction and transmission electron microscopy (TEM). 
Figure 3.17 shows bright field TEM images and diffraction patterns acquired from 
samples grown at different temperatures. As shown in Fig. 3.17 (a), the film grown at 
300 
o
C has a semicrystalline or amorphous-like structure due to the low growth 
temperature. The high resolution TEM analysis from the sample (not shown) showed 
no contrast from grain boundaries or lattice fringes indicating that the film is 
amorphous-like. However, the films grown at 400 
o
C showed nano-sized (~ 40 nm) 
grains which nucleated and crystallized from the amorphous matrix. The diffraction 
patterns from the sample grown at 300 
o
C show continuous rings which partially 
change to spots due to grain growth at 400 
o
C. Based on X-ray diffraction spectra (not 
shown) and indexing of the electron diffraction patterns obtained from the films 
grown at 300 
o
C and 400 
o
C [shown in Fig. 3.17 (d)], the films show polycrystalline 
randomly oriented rhombohedral BiFeO
3
 structure with no evidence of secondary 
phases. As the growth temperature increases further to 500 
o
C, the grain size 
increased to about 100 ? 150 nm diameter [Fig. 3.17 (c)]. 
 
 
 56 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.17. Bright field TEM images and corresponding electron diffraction 
patterns from the polycrystalline BiFeO
3
 films grown at (a) 300 
o
C, (b) 400 
o
C, and 
(c) 500 
o
C. As the deposition temperature increases, the grain size increases from 40 
to 150 nm. (d) The diffraction pattern from the film grown at 400 
o
C indicates 
randomly oriented polycrystalline BiFeO
3
 with no impurities. 
 
The diffraction pattern shown in Fig. 3.17 (c) indicates more nucleation and 
grain growth based on a dimmer continuous ring pattern with superimposed strong 
spots. However, even at 500 
o
C no evidence for secondary phases was observed. 
The electrical polarizations corresponding to the films with different grain 
sizes were measured. As shown in the polarization-electric field (P-E) hysteresis 
 57 
 
curves in Fig. 3.18, the semicrystalline film grown at 300 
o
C [Fig. 3.18 (a)] exhibits 
low saturation polarization (P
S
 ~ 8 ?C/cm
2
) with unsaturated hysteresis loops. 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.18. Electrical polarization hysteresis loops from the samples grown at 
(a) 300 
o
C (b) 400 
o
C, and (c) 500 
o
C. Each loop was acquired at different applied 
voltage (38, 44 and 55 V for (a) and 37, 41, 43 and 45 V in (b). (d) Leakage current 
density of each sample as function of applied electric field, and (e) relationship 
between saturation polarization and leakage current on deposition temperature. The 
film grown at 400 
o
C showed the largest polarization (Ps ~ 106 ?C/cm
2
) with 800 
kV/cm of coercive field as well as highest leakage current density. 
 
However, the film grown at 400 
o
C which has more crystallized area than the 
film grown at 300 
o
C consisting of nano-grains shows a dramatic increase in 
saturation polarization (P
s
 ~ 106 ?C/cm
2
) and a decrease of coercive field (~ 800 
kV/cm) at an applied electric field of 750 kV/cm as shown in Fig. 3.18 (b). This value 
was the largest polarization observed in the series of polycrystalline BiFeO
3
 films 
implying that the low polarization in bulk BiFeO
3
 in early reports does not represent 
 58 
 
the intrinsic saturation polarization of BiFeO
3
. Our results also indicate that the 
observed increase in epitaxial single-crystalline thin film is not caused by the misfit 
strain and structural distortions from the substrate. Compared to bulk, smaller 
remanent polarization, reduced by ~ (2
-1/2
)P
r
 is typically predicted in thin films.[87] 
However, the P
r
 obtained from our polycrystalline BiFeO
3
 deposited at 400 
o
C was 
much higher (91 ?C/cm
2
) than the reported value for bulk (~ 3.5 ?C/cm
2
) as well as 
for thin epitaxial films and it is comparable to the recent reports from bulk 
BiFeO
3
.[117] 
As the growth temperature increases further to 500 
o
C, the films show a 
decrease in polarization without saturation [Fig. 3.18 (c)]. Even at the electric field of 
165 MV/m, the hysteresis loop was not saturated. Typically, as the grain size 
decreases in submicron scale (d < 1 ?m), the saturation polarization as well as the 
remanent polarization decrease because domain wall mobility decreases by a reduced 
number of domain variants in the fine grains. The leakage current density and 
saturation polarization as a function of growth temperature plotted in Figs 3.18 (d) 
and (e) show that as the growth temperature increases, the leakage current density 
also increases. This increase is probably due to the increase in grain boundaries in the 
films at higher growth temperature.[88] It is interesting to notice that the film grown 
at 400 
o
C showed the highest leakage current (0.91 ? 10
2
 A/cm
3
 at an applied electric 
field of 50 MV/cm
3
) as well as the largest polarization. As the growth temperature 
increased further, there was a rapid decrease of leakage current density due to grain 
growth and reduction of the grain boundary area. The large polarization in the film 
grown at 400 
o
C suggests that nano-grain formation might facilitate the ferroelectric 
 59 
 
domain mobility by decreasing the domain wall density within the nano-grains. It is 
well known that the domain wall mobility is closely related to the relative 
crystallographic orientation between neighboring domains. In micron-size grains (d > 
150 nm), each grain contains multi-domains and the average domain wall density (N
d
) 
is known to be proportional to N
d
 ~ d
1/2
. However, as the diameter of the grains 
approaches the nano-scale, each grain prefers to have single domain and the domain 
wall density equation follows N
d
 ~ d
-1/2
. Therefore, a smaller domain wall density in a 
nano-crystalline structure might reduce pinning from domains oriented along 
different directions and result in the switching of ferroelectric domains. 
 
3.6.2 Switching of ferroelectric nano-domains 
 
Further investigation of the polarization reversal in the nano-sized grains 
was performed using piezo force microscopy (PFM). The BiFeO
3
 with rhombohedral 
crystal structure is known to have a spontaneous polarization along the [111] 
direction below Tc, with either 70 
o
 or 109 
o
 ferroelectric domain switching during 
poling.[90,91] Figure 3.19 shows the height images and corresponding images of 
switching of polarization domains acquired from the samples grown at different 
temperatures. From the height images in Figs. 3.19 (a), (b) and (c), the films showed 
increase of grain sizes and roughness as the growth temperature increased from 300 
to 500 
o
C. As shown in Fig. 3.19 (a), the film grown at 300 
o
C has very smooth 
surface (root mean square roughness ~ 0.15 nm) without clear grain boundaries. The 
PFM images in Figs. 3.19 (d), (e) and (f) were obtained after applying a voltage of + 
 60 
 
15 V to an area of 3 ? 3 ?m
2
 and then ? 15 V to an area of 1.6 ? 1.6 ?m
2
 within the 
original applied voltage. Fig. 3 19(d) shows no contrast which would indicate change 
in polarization under the applied electric field. However, the film grown at 400 
o
C 
which showed the largest polarization with nano-grains also exhibited clear domain 
switching after writing [see Fig. 3.19 (e)]. As the applied bias field reverses from + 15 
V in 3 ? 3 ?m
2
 area to ? 15 V in 1.6 ? 1.6 ?m
2
 area to + 15 V in 0.8 ? 0.8 ?m
2
 area, 
the contrast changed due to the change in orientation of the polarized ferroelectric 
domains [see Fig. 3.19 (e)]. Figure 3.19 (c) shows larger grains in the film grown at 
500 
o
C. A higher contrast of the ferroelectric domains [Fig. 3. 19(f)] is observed due 
to the enhanced crystallinity of the film. However, the direction of polarization of the 
domains did not reverse after switching the dc bias from + 15 V to ? 15 V alternately. 
 
 
 
 
 
 
 
 
 
Figure 3.19. AFM height images (top) and PFM images (bottom) of switching 
electrical domain from the films grown at 300 
o
C (a, d), 400 
o
C (b, e), and 500 
o
C (c, 
f). For switching domains, a positive followed by a negative bias (? 15 V) were 
applied in 3 ? 3, 1.6 ? 1.6 and 0.8 ? 0.8 ?m
2
 area, alternately. Only the film grown at 
400 
o
C showed change of contrast with switching the applied bias. 
 
 61 
 
Generally, in ceramic bulk and powder ferroelectric materials, it is observed 
that as the grain size decreases from micron to submicron scale (d < 1 ?m), the 
saturation polarization as well as the remanent polarization decrease. This decrease 
can be explained from the decrease in domain wall mobility by a reduced number of 
domain variants in the fine grains. Therefore, switching domains in this system 
requires higher electric field because the domains have to switch against their 
preferred crystallographic orientation relation. This explanation can help understand 
our observation of paraelectric hysteresis curve with zero remanent polarization and 
no domain switching in PFM from samples with relatively large grain size grown at 
500 
o
C. These results are in agreement with previous studies.[92] In general, there is 
an increase of free energy (?F) associated with the formation of nuclei polarized 
along antiparallel directions which can be expressed by the following equation, 
2
1
2
2
SW S
F EP V A NP V??=? + +  
where, P
S
, ?
w
, A, V, and N represent respectively, the spontaneous polarization, 
surface energy associated with domain wall, surface area, average volume of nuclei, 
and the depolarizing factor. When the grains are small and contain single domains, 
the depolarizing factor of each nuclei is negligible. Therefore, the critical nucleation 
size is determined by the first two terms in the equation for ?F. 
In the case of BiFeO
3
, when the grain size approaches nano-scale 
dimensions (d < 150 nm), the films showed different tendency from the films with 
micron to submicron-sized grains. The films with less than 50 nm grains showed a 
dramatic increase in polarization as well as uniform domain switching, and as the 
microstructure approaches the critical domain size, the nano-grains tend to have a 
 62 
 
single-domain and the domain wall density is greatly reduced within a grain.[89] This 
reduction in domain density might lead to a reduction of constraint in switching of 
the polarization by domains of different orientation.  
From the results of electrical polarization and switching of domains, the 
observed large polarization in crystalline BiFeO
3
 might be an intrinsic property of 
BiFeO
3
 and the switching of ferroelectric domain significantly depends on the grain 
size when the grains are in the nano-scale. The domain boundaries inside the grains 
of submicron size might, more significantly, constrain the switching of domain by 
crystallographically fastening neighboring domains within a grain. Therefore, when 
grains are in the nano-scale surrounded by amorphous matrix and contain single 
domain, each domain can switch independently without interference from 
neighboring domains. 
 
3.6.3 Annealing effects on the microstructure and electrical 
properties of polycrystalline BiFeO
3
 films 
 
To understand the effect of nano-grain size on the electrical properties of 
BiFeO
3
 films and acquire uniform size and distribution of nanodomains, an annealing 
process was conducted on the films grown at 300 
o
C, which have semicrystalline 
microstructure. The films were annealed at 500 
o
C for a period of time in the range 2 
to 8 hours at atmosphere to promote nucleation and grain growth. 
Figure 3.20 (a) shows an increase in crystallinity with grains of ~ 5 ? 10 nm 
in diameter upon annealing. The grains nucleated in different parts of the film and not 
 63 
 
from the interface with the substrate. The grains have random orientation and are 
fairly uniformly distributed within an amorphous-like matrix [compare with Fig. 3.17 
(a)]. As shown in Fig. 3.20 (c), the annealed films show an increase of leakage 
current density from 5.3 ? 10 
-2
 A/cm
2
 to 1.4 ? 10 
-1
 A/cm
2
 after annealing. This 
increase is mainly due to an increase of grain boundary area and the possibly the 
formation of oxygen vacancies upon annealing. However, the smaller grains (also 
domains) increased the electrical polarization compared to the as grown amorphous-
like films, by about 10 times (~ 82 ?C/ cm
2
) 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.20. (a) Bight-field TEM image and (b) high resolution TEM image 
showing pure BiFeO
3
 nanograins with size of 5 ? 10 nm diameter after annealing at 500 
o
C for 8 hours.  (c) Changes of leakage current and (d) electrical polarization upon 
annealing. Nucleation of nanodomain structure of polycrystalline BiFeO
3
 enhanced the 
polarization, but also increased the leakage current density mainly due to an increase in 
the area of domain boundaries and evaporation of Bi
3+
 and O
2-
 upon annealing. 
 64 
 
 
Therefore, our annealing results confirmed that the domain size is the most 
dominant factor for switching the electrical polarization, and the increase in boundary 
area is responsible for an increase in leakage current in polycrystalline BiFeO
3
 thin 
films. Also the observed large polarization and switching in BiFeO
3
 films with 
uniform nanoscale-grains (also nanoscale-domains) is important in the context that 
the nanosize domain structure enables the increase in data storage ability for 
ferroelectric memory storage devices, which require uniform domain size and shape 
for reliable writing and reading. However, the domain size for device application still 
has to be optimized together with the thickness of the film that will require lower 
operation voltage and leakage current. 
 
 
 
 
 
 
Figure 3.21. (a) Bright field and (b) dark field Scanning TEM (STEM) images 
om the same area in the annealed poly BiFeO
3
 film. The inset to (b) represents the 
ical composition through one of the grains with ~ 20 nm diameter in size using 
DS line scanning. 
 
We performed chemical analysis on the annealed BiFeO
3
 thin film to 
etric phases formed upon annealing. The bright field [Fig. 
 
fr
chem
E
investigate if off-stoichiom
 65 
 
3.21 (a)] and dark field [Fig. 3.21 (b)] Scanning TEM images were acquired from the 
same a
STEM image in Fig. 3.21 (b), this region is brighter in contrast compared to the 
matrix 
 analysis on the 
semicry
rea to analyze the uniformity of the composition distribution. The dark area in 
the bright field image represents crystallized BiFeO
3
 nanograins. From the dark field 
area, which indicates that this region contains more heavy elements, such as 
bismuth. An EDS line can was acquired across a grain which is about 10 nm in 
diameter. The EDS spectra show that the Fe element was fairly uniform both in the 
grain and the matrix. However, Bi showed slightly higher concentration in the central 
region of the grain, and the ratio of Bi:Fe was about 1:1 in this region. These results 
indicate that the matrix region is slightly Bi deficient and contains oxygen vacancies 
after annealing. The bismuth atoms might be easier to evaporate from the amorphous 
matrix than in the nucleated grains which fasten bismuth in their crystalline sites in 
the structure. This could be the reason for the observed higher leakage current 
together with an increase of grain boundaries in the annealed sample. 
The electrical polarization and switching polarization in polycrystalline 
BiFeO
3
 thin film with nanodomain size were investigated systematically. The domain 
size was controlled by varying the growth temperature in the PLD chamber. For small 
grain size at lower deposition temperature, the leakage current density is higher due to 
the increase in grain boundary area. Also larger polarization was observed when the 
BiFeO
3
 grains are in the nano-scale. From the annealing and chemical
stalline BiFeO
3
 thin film grown at 300 
o
C, an increase in electrical 
polarization was observed because of the nucleation and growth of nano-grains. An 
increase in grain boundaries coupled with the evaporation of Bi and O during 
 66 
 
annealing caused high leakage current density in the annealed films. Based on our 
results, the domain size is the most dominant factor for switching of electrical 
polarization and leakage current in polycrystalline BiFeO
3
 thin films. Our results also 
suggest that the large polarization observed in polycrystalline BiFeO
3
 films free of 
strain and of secondary phases is not induced by strain but it is an intrinsic property of 
BiFeO
3
. 
 
3.7 Enhancement of dielectric properties of BiFeO
3
 films by flux-
mediated epitaxy (FME) method 
 
We have fabricated single crystal-like BiFeO  (BFO) thin films by flux-
mediated epitaxy using pulsed laser deposit
3
ion (PLD). The Bi-Cu-O flux composition 
ickness were optimized using composition spread, thickness gradient and 
2 ?m and higher dielectric constant, in the 
range 260 ? 340, than those for standard PLD grown films. In addition, the leakage 
current
studied because of their antiferromagnetic and ferroelectric ordering well above room 
and its th
temperature gradient libraries. The optimized BFO thin films grown with this 
technique showed larger grain size of ~ 
 density of the films was reduced by two orders of magnitude compared to that 
of standard PLD grown films. 
 
3.7.1 Introduction of FME method 
 
As mentioned in chapter 1, multiferroic BiFeO
3
 (BFO) thin films are widely 
 67 
 
temperature.[57-60] In pursuing their device applications, it is critical to address and 
control their microstructural variation and the relatively large leakage current 
primarily due to the high volatility of Bi and heteroepitaxial constraints imposed by 
muth can 
roduce Bi deficient secondary phases such as antiferromagnetic ?-Fe
2
O
3
, and 
ferroma
 bottom 
easurements) (50 ? 80 nm). After the temperature is 
creased to 650
o
C, BiFeO
3
 seed layer (~ 10 nm) is deposited on (001) oriented 
3
 by PLD. Then the flux layer of a mixture of (Bi
2
O
3
 + CuO) is deposited on 
e seed layer. The temperature is then increased to 730
o
C, and BiFeO
3
 is deposited 
age shows uniformly distributed flux droplets on the 
iFeO
3
 seed layer with a size ranging 2 ? 4 ?m diameter. Once the desired thickness 
3
 is deposited the remaining flux is etched off. 
the substrate. The strain in the films together with the evaporation of bis
p
gnetic ?-Fe
2
O
3
 and Fe
3
O
4
 which form to compensate the local imbalance in 
chemical composition and structural distortion.[61-64] In this section we present the 
successful use of FME to grow high quality BiFeO
3
 films by preventing the 
evaporation of Bi through the use of a CuO-Bi
2
O
3
 flux. With this technique, we 
obtained a substantial improvement of leakage current and dielectric properties (? ? 
260 ? 340) in single crystal-like BFO films grown by flux-mediated epitaxy (FME). 
 
 
 
 
 
 
 
 
Figure 3.22. FME for BiFeO
3
 growth. First deposition of SrRuO
3
electrode (for electrical m
in
SrTiO
th
on the flux layer. SEM im
B
of BiFeO
 68 
 
As shown in Fig. 3. 22, in the FME method, an oxide flux on the surface of 
e film plays a key role in improving the crystallinity of complex oxide films.[95] 
he vapor ablated from the target material by a pulsed laser meets the flux on the 
surface
atoms in
mperature and thickness of 
e Bi
2
O
3
-CuO flux. The oxygen partial pressure for the growth of the FME-BFO 
m) was 
eposited at 650 
o
C and 30 mTorr oxygen partial pressure before depositing the flux 
r increasing the temperature to 730 
o
C, a 
i
2
O
3
 layer (~ 1 nm) and a CuO layer (~ 1 nm) were deposited in an alternating 
manner 
th
T
 of the substrate, diffuses into the liquid intermediate phase and finally 
solidifies on a predeposited seed layer. The liquid phase (flux) prevents direct 
deposition of the vapor phase on the solid surface, and improves crystallinity. 
Compared to PLD, in the FME method, there is enhanced diffusion of the surface 
 the flux above the quasi-melting temperature (~ 0.5 T
m
) which leads to film 
growth in equilibrium even at high growth rates.[99-102] 
 
3.7.2 Combinatorial approach for optimum growth conditions 
 
3.7.2.1 Temperature gradient libraries 
 
The schematic in Fig. 3.23 shows the deposition procedure for the 
combinatorial library used to optimize the composition, te
th
film was varied in the range of 0.05 to 1 Torr. A BFO seed layer (~ 10 n
d
material to lead to growth of BFO. Afte
B
to induce interdiffusion between the layers and to prepare a layer with a 
gradient in composition of the flux. The total thickness of Bi
2
O
3
 was fixed to 40 nm. 
 69 
 
Using a moving shadow mask in the PLD chamber during the deposition of CuO, a 
layer with a thickness gradient was achieved. These conditions gave rise to regions 
with different ratios of Bi
2
O
3
/CuO in the flux which varied from pure Bi
2
O
3
 (without 
CuO) to (Bi
2
O
3
)
0.5
?(CuO)
0.5
. In addition, a temperature gradient from 560 
o
C to 805 
o
C was achieved on the substrate over a distance of 9 mm by heating one side of the 
sample stage using a semiconductor laser. Following the deposition of the flux, a 300 
nm thick BFO thin film was deposited on the flux layer. 
Fig. 3.23 (a) shows the intensity map of the (001) BFO diffraction peak as a 
function of position of the sample which corresponds to changes in CuO layer 
thickness (0 ? 40 nm) and the sample growth temperature. The highest intensity of the 
(001) BFO peak was observed in the region with CuO thickness of 4.5 ? 17 nm and 
temperature of 710 ? 750 
o
C. To understand the effect of CuO on the Bi based flux 
growth, temperature dependent XRD spectra in the regions with 0 vol% CuO [line I 
in Fig. 3.23 (a)] and ~ 4.5 nm thick CuO (10 vol% CuO) [line II in Fig. 3.23 (a)] are 
compared in Figs. 3. 23(b) and (c), respectively. The regions with pure Bi
2
O
3
 flux 
show a strong Bi
2
O
3
 XRD intensity peak over a large range of temperatures below 
730 
o
C. This region overlaps with the temperature range for Fe
2
O
3
 growth [see Fig. 
3.23 (b)]. Fig. 3.23 (c) shows that addition of 10 vol% CuO in the flux plays an 
important role: namely, CuO decreases the melting temperature of the Bi
2
O
3
?CuO 
flux material thereby suppressing the precipitation of Bi
2
O
3
. Furthermore, the amount 
of Bi
2
O
3
 melt in the flux compensates for evaporated Bi thus suppressing the 
formation of Fe
2
O
3
.[103] The comparison of the shadow areas in Figs. 3.23 (b) and 
(c) illustrates the enlarged growth window of BFO for a flux with 10 vol% CuO. 
 70 
 
Following the combinatorial optimization, a BFO thin film was grown by FME using 
the optimum flux composition (10 vol% CuO) and the optimized growth temperature 
(730 
o
C). 
 
 
 
 
 
 
 
 
 
 
Figure 3.23. Schematic (top left) of the deposition procedure for te
ysis. The temperatures are varied from 580 to 810 
mperature-
gradient combinatorial anal
o
C with 
ompositional spread of Bi
2
O
3
 (400 ?) and CuO (0 ~ 400 ?). (a) XRD map of 
combinatorial library shows the intensity of the BiFeO
3
 (001) peak as a function of 
bstrate temperature and flux CuO composition to decide the optimum condition for 
FME. (b) 2? x-ray map in the region marked I in (a) for 0 % CuO. (c) Addition of 10 ? 
0 vol% Cu (XRD mapping result labeled II in (a)) shows highest intensity of the 
BiFeO
3
 (001) peak enlarging the growth window of BiFeO
3
 compared to the film 
eposited without Cu (labeled (I)) 
 
3.7.2.2 Pseudo Ternary composition libraries 
 
c
su
3
d
To find the optimum amount of flux for the growth of BiFeO
3
 by FME, we 
studied a pseudo ternary combinatorial system. The growth temperature of BiFeO
3
 
was fixed at 730 
o
C as obtained from temperature gradient combinatorial study 
 71 
 
presented in the previous section. As shown in Fig. 3.24, two different shapes of 
moving shadow masks were used to make composition gradient along three different 
orientations in a triangle. The schematic in the top left of Fig. 3.24 shows the 
deposition procedure for the pseudo ternary combinatorial library. First, a 10 nm thick 
BiFeO
3
 seed layer was deposited on SrTiO
3
 at 650 
o
C at 30 mTorr oxygen partial 
pressure using standard PLD growth. Then, alternating layers of a Bi
2
O
3
 layer (~ 10 
?) and a CuO layer (~ 10 ?) were deposited to induce interdiffusion between the 
different layers and to make variant composition of flux. The total thickness of Bi
2
O
3
 
and CuO varies from 0 to 400 ?, respectively. Then, the temperature was increased to 
730 C, and a thicker layer of BiFeO
3
 with constant thickness (3000 ?) was deposited 
on the flux. 
 
 
 
 
 
 
 
 
mposition spread 
rocedure. (a) the picture and (b) schematic of the shadow mask used in 
combinatorial PLD system. 
Figure 3.25 (a) shows an optical image of the real surface taken from the 
specimen of Bi
2
O
3
-CuO-BiFeO
3
 ternary system. Scanning x-ray micro-diffraction 
o
Figure 3.24 Schematic (top left) of the pseudo ternary co
p
 
 72 
 
maps were obtained from the region inside the white triangle in (a). The 2-
imensional and 3-dimensional graphic images of the x-ray diffraction intensity of 
e (001) BiFeO
3
 peak shown in Figs. 3.25 (b) and (c) indicate that the highest 
tensity for the 2 theta angle of BiFeO
3
 is found in the region with high Bi
2
O
3
. 
Therefo
ith the results obtained from the temperature gradient 
combinatorial study. From these combinatorial approaches, the optimum growth 
conditions were decided. We selected 730 C for the growth temperature and ~ 45 ? 
thick CuO (10 vol% CuO) and 400 ? thick Bi
2
O
3
 for flux, and deposited a 3000 ? 
BiFeO
3
 thin film on the flux. 
 
 
 
 
 
 
 
 
 
 
 
ary system. (b) 2-
imensional and c) 3-dimensional x-ray maps for the diffraction intensity and 2 theta 
angles of BiFeO
3
 peak. 
 
d
th
in
re, addition of small amount of CuO improved the crystallinity of BiFeO
3
 thin 
film, which is coincident w
o
Figure 3.25. (a) The specimen of Bi
2
O
3
-CuO-BiFeO
3
 tern
d
 73 
 
3.7.3 Enhanced structural and dielectric properties 
 
 
 
 
 
 
 
 
m area of 
BiFeO
3
hin films grown (a) by conventional PLD and (b) by FME method. (c) The 3-
D image of magnified BiFeO
3
 domain in (b) shows atomic steps during growth of the 
film with grain size of ~ 2 ?m (~ 10 times larger than that of BiFeO
3
 grown by 
normal PLD). 
he AFM image in Fig. 3.26 (b) shows much increased grain size (~ 2 ?m) 
for the FME grown BFO film, which is more than 10 times larger than the grain size 
for film  grown by standard PLD [shown in Fig. 3.26 (a) with typical grain size of 50 
? 200 nm]. The height profile of the FME grown BFO film [Fig. 3.26 (c)] shows a 
fairly smooth surface (compared to the standard PLD grown BFO films) with an 
average step height of approximately 1.2 nm, which corresponds to 3 unit cells of 
 
 
 
 
Figure 3.26. Atomic force microscopy images taken from 5 ?m ? 5 ?
 t
 
T
s
 74 
 
BFO [see inset to Fig. 3. 26 (c)]. The width of the terraces varies in the range from 70 
to 100 
 
 
 
 
 
 
 
 
 
SrTiO
3
resolu
diffraction p
 
ages 
(bottom)
Fe
2
O
3
 (?
nm. 
 
 
Figure 3.27. (a) Cross-sectional bright field TEM image showing smooth 
surface of the BiFeO
3
 film which was grown on SrRuO
3
 buffered (001) oriented 
. 1000 ? thick palladium (Pd) was used as the top electrode. The high 
tion lattice images of (b) interface between SrTiO
3
 and SrRuO
3
 and (c) 
interface between SrRuO
3
 and BiFeO
3
 show epitaxial growth. The corresponding 
attern from each layer is shown as inset in (b) and (c) 
A cross-sectional bright field TEM image (top) and high resolution im
 of the FME grown BFO film in Fig. 3.27 exhibit epitaxial growth with very 
sharp interfaces and smooth surface of the BFO film. In the large area view [Fig. 3.27 
(a)], the film grown by FME shows single crystal-like BFO film without grain 
boundaries, and we see no evidence of bismuth deficient secondary phases such as 
- or ?-), or Fe
3
O
4
 as we have observed in films grown by PLD at low oxygen 
 75 
 
partial 
FME still suffers from some residual compressive strain (? 0.9 %), originating from 
the latt
 (~ 
2 ?m) 
pressure (see section 3.2).[62,63] The lattice constants of the BFO film 
determined from XRD and electron diffraction patterns are a = b = 0.394 nm (in-
plane), and c = 0.397 nm (out-of-plane). These values indicate that the film grown by 
ice mismatch (? 1.15%) between the substrate and the film, even though the 
film was grown under a quasi-equilibrium condition facilitated by the flux. 
Figure 3.28 (a) shows the dielectric constant (?) of BFO thin films grown by 
FME method (labeled ?FME-BFO?) and by the standard PLD technique (labeled 
?PLD-BFO?) as a function of frequency. The dissipation factor (tan ?) values of the 
FME-BFO film and (100) oriented PLD-BFO film were 0.011 ? 0.009 and 0.012 ? 
0.01, respectively, in the frequency range of 10 kHz ? 10 MHz at room temperature. 
The dielectric constant values of the PLD-BFO thin films varied from 80 ? 130 
depending on the frequency and crystalline orientation of the films. These values are 
slightly higher than the values reported for bulk BFO.[104,105] In contrast, the 
dielectric constant of FME-BFO films is 3 times as large (260 ? 340) as that of PLD-
BFO films. This may be due to the enhanced crystallinity with the large grain size
in the FME grown films. The leakage current density of FME-BFO thin films 
is also improved by two orders of magnitude compared to that of PLD-BFO films at 
applied electric fields of up to 100 MV/m as shown in Fig. 3.28 (b). X-ray 
photoelectron spectroscopy analysis (not shown here) from a PLD-BFO film did not 
show any significant amount of Fe
2+
 ions which could explain leakage by double 
exchange interaction among Fe
2+
?O
2-
?Fe
3+ 
ions. Therefore, the concentration of Fe
2+
 
ions cannot be the dominant reason for the high leakage current in the PLD-BFO 
 76 
 
films. One possible reason for the leakage current in PLD-BFO films could be the 
larger fractional area of grain boundary regions in these films, which allow current to 
pass through with less resistance than in the bulk of the grains. Another possibility is 
that oxygen vacancies and Bi deficiency in PLD-BFO induce impurity energy levels 
in the band gap that increase free carrier density by hopping of electrons to these 
defect levels.[106] 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.28. (a) Dielectric constant of BiFeO
3
 thin films fabricat
thod on SrTiO3 (100) and (111), and by flux method on SrTiO3 (100) as a 
ed by typical 
PLD me
nction of frequency (10 kHz ? 10 MHz). (b) Leakage current density (A/cm2) vs. 
electric field (MV/cm) curves of BiFeO
3
 thin films fabricated by typical PLD and 
lux method. (c) Magnetic hysteresis curve (in-plane) of BiFeO
3
 thin film fabricated 
by flux method. (d) Electric field dependence of polarization loops of BiFeO
3
 thin 
lm fabricated by flux method measured at 1 kHz. The inset in (d) is the P-E loop 
from a PLD grown film. 
 
fu
f
fi
 77 
 
In summary, we improved the crystallinity of BiFeO
3
 films by flux-mediated 
pitaxy. For the FME growth, we optimized the growth conditions by the 
ombinatorial method. The FME grown films exhibited single crystal quality with 
rge grain size (~ 2 ?m) and a flat surface as well as higher dielectric constant and 
wer leakage current than films grown by standard PLD. However, the P-E loop 
showed
e
c
la
lo
 unsaturated polarization even for field of 25 ? 10
6
 V/m. (Fig. 3.28(d)). Also, 
the magnetization in the FME-BFO films was low as shown in Fig. 3.28(c). 
 78 
 
Chapter 4 
4. Application the ME effects for spintronic device 
application 
 
4.1 Introduction of exchange bias 
 
Many multiferroic materials exhibit antiferromagnetic (AFM) 
properties.[70,71] Therefore, one important application of these materials is related to 
exchange bias.[72] Exchange bias is the phenomenon associated with the exchange 
anisotropy created at the interface between an AFM layer and a ferromagnetic (FM) 
layer. Possible applications of exchange bias are domain stabilizer in recording heads 
based on anisotropic magnetoresistance or magnetic recording. The combination of 
exchange bias and magnetoelectric (ME) effects could allow control of exchange bias 
by an electric field. The ME controlled exchange bias was first reported in a Cr
2
O
3
 
crystal and Co bilayer system.[102] 
Exchange bias was discovered in 1956 by Meiklejohn and Bean in FM Co 
particles embedded in a CoO (AFM) matrix.[103] Generally, exchange bias results in 
the horizontal shift in the magnetization-magnetic field (M-H) hysteresis loop due to 
the pinned FM phase at the interface between a FM and an AFM phase layer when 
the materials are cooled in the presence of a static magnetic field through the N?el 
temperature (T
N
) of the AFM layer which is lower than the Curie temperature (T
C
) of 
the FM material. After the field cool procedure, the M-H loop is shifts along the field 
 79 
 
axis in the opposite direction to the cooling field. Also the coercive field of the 
system increases after this cooling procedure.  
 
 
 
 
 
 
 
 
 
 
 
Figure 4.1. (a) Typical shape of M-H hysteresis loop of ferromagnetic 
materials. (b) Schematic diagram of the spin configuration of FM-AFM bilayer 
 
Figure 4.1 shows a schematic diagram of the spin configuration in a FM-AFM 
bilayer structure. When a field is applied in the temperature range T
N
 < T < T
C
, the 
FM spins align with the external field, while the AFM spins remain random. When 
cooling to T < T
N
, the AFM spins next to the FM interface align along the direction of 
ferromagnetic spins due to the interaction at the interface. The other spins in the AFM 
material follow AFM order giving zero net magnetization.  
When the field is reversed, the spins in the AFM try to keep the spins in the 
FM layer in their original direction (ferromagnetically aligned at the interface). 
 80 
 
Because of this interfacial interaction, an extra field (H
ex
) is needed to reverse spins in 
the FM layer. However, when the field is applied in the original direction, the spins in 
the FM would start to rotate at a lower field due to the interaction with the spin in the 
AFM. From these interfacial interactions between AFM and FM in the presence of an 
external field, the M-H hysteresis loop of the bilayer is shifted from the origin and has 
increased coercivity. 
In the opposite case, when the coupling of FM-AFM at the interface has an 
antiferromagnetic order at the initial stage, the magnetic spins in the FM layer can 
change direction when the field is reversed below T
N
. However, when the field is 
reversed to the original direction, a higher field is needed to switch the spins due to 
the ferromagnetic alignment at the interface. This behavior is called the positive 
exchange bias. In positive exchange bias, the M-H hysteresis loop shifts along the 
field axis in the same direction as the cooling field. 
 
4.2 Control of magnetic moment and exchange bias in Cr
2
O
3
 
 
Cr
2
O
3
 is a Heisenberg antiferromagnet (AFM) showing antiferromagnetic 
ordering of Cr
3+
 spins aligned along the c axis of its rhombohedral unit cell below the 
N?el temperature (T
N
 = 308 K).[104] Cr
2
O
3
 was the first compound in which the 
magnetoelectric (ME) effect was theoretically predicted and experimentally 
observed.[105-108] Recently, the ME effect on exchange bias was observed on a 
Co/Pt multilayer on a c-axis oriented Cr
2
O
3
 single crystal.[109,110] Such an effect 
 81 
 
can be the basis for new type of memory and logic devices where both electric and 
magnetic fields serve as external control parameters. 
Thin film multilayer devices are desirable over bulk since rather large 
electric fields can be applied to the AFM layer, and thus the ME effect on exchange 
bias is expected to be larger. To this end, in our study, we have fabricated Co/Pt 
multilayers on Cr
2
O
3
 thin films with extremely smooth interfaces. Out-of-plane 
magnetized exchange-biased dual hysteresis loops were clearly observed. Detailed 
features in the loops were found to be delicately controlled by demagnetizing ac 
magnetic field with different amplitude offset during magnetic field cooling (MFC). 
The effect of high electric field (~ 77 MV/m) cooling (EFC) as well as direct electric 
field E at a fixed temperature after magnetic field cooling were investigated. 
130-nm thick Cr
2
O
3
 thin films were grown on Al-doped ZnO (Al-ZnO) 
buffered c-axis oriented sapphire (Al
2
O
3
) substrates by pulsed laser deposition. Cr
2
O
3
 
and Al-ZnO targets were ablated using a KrF excimer laser with a typical fluence of 2 
J/cm
2
. The epitaxial Al-ZnO (170 nm) layer was used for the bottom electrode to 
apply an electric field on the Cr
2
O
3
 thin film, and [Co (~ 0.3 nm)/Pt (~ 1.5 nm)]
3
 
layers were deposited alternately on the Cr
2
O
3
 layer by electron-beam evaporation at 
500 K. Pt (~ 0.5 nm) was the first layer deposited to prevent oxygen diffusion from 
Cr
2
O
3
 and subsequent potential formation of antiferromagnetic CoO.[111] To 
characterize the microstructure of the multilayer, phi (?) scans and theta (?) scans 
were performed with a 4-circle x-ray diffractometer (Bruker D8 Discover). 
Transmission electron microscopy (TEM) images and selected area diffraction (SAD) 
patterns of the films were obtained at an accelerating voltage of 200 keV with a JEOL 
 82 
 
2100F field emission TEM. The Co/Pt multilayer atop the Cr
2
O
3
 layer was patterned 
into 200 ?m ? 200 ?m square dots by ion milling to make capacitor devices. A 
magneto-optical Kerr effect (MOKE) system was used to detect the Kerr rotation of 
Co spins in the Co/Pt multilayer and measure the hysteresis loops directly on 
individual capacitor devices. 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 4.2. (a) XRD 2-theta scan of a Co/Pt/Al-ZnO/Cr
2
O
3
 film. (b) Leakage 
current density of Cr
2
O
3
 film measured using Pt (top) and Al-ZnO (bottom) 
electrodes at room temperature. 
 
The ? ? 2? x-ray diffraction (XRD) spectrum in Fig. 4.2 (a) shows that the 
Al-ZnO and Cr
2
O
3
 layers, which have the hexagonal and rhombohedral crystal 
structures, respectively, grew c-axis oriented on the (001) oriented Al
2
O
3
 substrate, 
 83 
 
which has a rhombohedral crystal structure. The sheet resistance of the Al-ZnO 
bottom electrode layer was less than 10 k?. The Cr
2
O
3
 film showed sufficient 
insulation for applying a high electric field. As shown in Fig. 4.2 (b), the leakage 
current density of the Cr
2
O
3
 film measured at room temperature was ~ 10
-2
 A/cm
2
 
even for electric fields as high as 380 MV/m at room temperature. This allowed us to 
apply up to ? 100 MV/m continuously without breakdown. 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 4.3. (a) High-resolution TEM image of Co/Pt layer/Cr
2
O
3
 interface. 
Co particles were embedded in a Pt matrix. The thickness of Co/Pt layer was ~ 10 nm. 
(b) Electron diffraction pattern and (c) schematic of the crystal orientation relation of 
the Cr
2
O
3
/Al-ZnO/Al
2
O
3
. The film showed epitaxial growth of Al-ZnO and Cr
2
O
3
 
layers on c-axis sapphire substrate. 
 
 84 
 
As shown in the electron diffraction pattern in Fig. 4.3 (b) and the schematic 
in Fig. 4.3 (c), the Cr
2
O
3
 and Al-ZnO layers grew epitaxially on the (001) oriented 
Al
2
O
3
 substrate. The growth orientation relationship between the layers indexed from 
the diffraction pattern is [001]
Al2O3
 || [001]
Al-ZnO
 || [001]
Cr2O3
 along the out-of-plane 
direction and [100]
Al2O3
 || [110]
Al-ZnO
 || [100]
Cr2O3
 along the in-plane direction [Fig. 4.3 
(c)]. 
The interface roughness between the ferromagnetic (FM) and AFM layers is 
known to influence the AFM spin configuration and cause complex exchange bias 
behavior affecting the exchange bias field (H
E
) and the coercive field (H
C
).[112] In 
our sample, the Co/Pt layer on the Cr
2
O
3
 layer showed a continuous structure with a 
sharp interface over the whole lateral range (several hundred microns) visible in the 
TEM sample of the film which we attribute to the extremely smooth surface of the 
Cr
2
O
3
 layer. The surface roughness (root mean square) of a Cr
2
O
3
 layer deposited on 
an Al-ZnO film (170 nm) measured by atomic force microscopy was ~ 0.3 nm over a 
10 ? 10 (?m)
2
 area. The selected area diffraction pattern (not shown here) obtained 
from the Co/Pt multilayer [Fig. 4.3 (a)] indicated that the multilayer was 
polycrystalline. 
Figure 4.4 (a) shows the out-of-plane hysteresis loop of our (Co/Pt)/Cr
2
O
3
 
structure at room temperature after magnetic-field cooling (MFC) through the N?el 
temperature following heating of the sample to 325 K, where H
C
 of the Co/Pt layer 
was 15 Oe. We found clear exchange bias with cooling in a field of only 20 Oe. As 
the temperature decreased from 325 K to 298 K, H
E
 was seen to gradually increase to 
~ 170 Oe, and the sign of H
E
 was consistent with the direction of the applied field and 
 85 
 
the AFM layer and the FM layer having ferromagnetic coupling. Even at 298 K, with 
the exchange bias present, H
C
 of Co remained narrow at 15 Oe. As the magnetic field 
was reversed (? 20 Oe) during MFC, H
C
 of the Co/Pt layer switched sign from the 
negative to positive side as shown in Fig. 4.4 (b). These results indicate that the 
applied magnetic field aligned the soft (Co) layers during MFC, and the configuration 
of the magnetic moments in this layer duplicated the pattern of the very top layer of 
Cr
2
O
3
 layer (see schematics in Fig. 4.4). 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 4.4. (a) and (b) show the out-of-plane hysteresis loops of our 
(Co/Pt)/Cr
2
O
3
 structure at room temperature after magnetic-field cooling (MFC) 
through the N?el temperature following heating of the sample to 325 K, where H
C
 of 
the Co/Pt layer was 15 Oe. We found clear exchange bias with cooling in a field of 
only 20 Oe. As the temperature decreased from 325 K to 298 K, H
E
 was seen to 
gradually increase to ~ 170 Oe, and the sign of H
E
 was consistent with the direction 
of the applied field. Even at 298 K, with the exchange bias present, H
C
 of Co 
remained narrow at 15 Oe.  
 86 
 
As shown in the schematic in Fig. 4.5 (a), we found that we could control the 
overall configuration of the magnetic moments of the interface Co and Cr
2
O
3
 layers 
by an applied magnetic field during the MFC process for tunable exchange bias and 
hysteresis loops. Dual hysteresis loops result from an uneven up/down spin fraction 
set up in the AFM layer at the interface.[113] To achieve this, we have superimposed 
a slow (0.02 Hz) ac field with a constant amplitude of 0.7 kOe with dc offsets of 
either 0, ? 0.02, or ? 0.2 kOe [see Fig. 4.5 (a)]. The ac field during the cooling 
process (~ 30 minutes) provided a demagnetizing effect on the Co/Pt layer, and the dc 
offset created an uneven up/down spin fraction depending on its sign and amplitude. 
Figure 4.5 (b) shows that for the + 0.2 kOe offset case, the ac magnetic field 
was made to oscillate between + 0.9 kOe and ? 0.5 kOe. The net positive magnetic 
field swing (+ 0.4 kOe) in this case should be sufficient to saturate and align all 
magnetic spins of the Co layer along the positive field direction. This leads to the Co 
magnetic moments imprinting their homogenous magnetic spin orientation into the 
very top layer of spins of the AFM Cr
2
O
3
 during cooling and results in a fully 
uncompensated interface and a negative H
E
. Similarly, in the ? 0.2 kOe offset case, 
the magnetic field is swept between + 0.5 and ? 0.9 kOe with a net negative magnetic 
field (? 0.4 kOe) which aligns all magnetic spins along the negative field direction 
resulting in a positive value of H
E
. 
For the case of zero offset, the magnetic field oscillates between + 0.7 kOe 
and ? 0.7 kOe. After cooling, a dual hysteresis loop is observed. This is marked as 
?zero offset? in Fig. 4.5 (b). In this case, the demagnetizing process is consistent with 
domains in the Co/Pt multilayers such that there are equal numbers of up and down 
 87 
 
spins. Upon cooling, these domains are imprinted onto the very top layer of Cr
2
O
3
. 
They in turn result in exchange biasing ?half? of the Co spins in one direction while 
the other half is biased in the opposite direction,[113] giving rise to the double loop. 
For the intermediate offset values [? 0.02 kOe in Fig 4.5 (b)], uneven distributions of 
up and down magnetic domains become imprinted on the top layer of Cr
2
O
3
 resulting 
in vertically asymmetric dual hysteresis loops reflecting the distributions. Thus, by 
adjusting the dc field offset, we can tune the shape of the dual hysteresis loop. 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 4.5 (a) The schematic of magnetic field cooling process with sweeping 
the magnetic field with various offsets and (b) Control of magnetic hysteresis by 
sweeping the magnetic field with various offsets during magnetic field cooling. The 
hysteresis loops gradually change from ? H
E
 to + H
E
 and also from single loop to dual 
loop by changing offset from + 0.2 kOe to ? 0.2 kOe. 
 
 88 
 
4.3 Magnetoelectric effect on exchange bias in Cr
2
O
3
 
 
To investigate the ME effect on exchange bias in the thin-film devices, the 
hysteresis loops were compared after different (simultaneous) magnetic and electric 
field cooling (MEFC) conditions (magnetic field = 0.7 kOe and E = ? 77 MV/m). 
Based on previous measurements of the ME susceptibility (tensor) in Cr
2
O
3
, we have 
performed the MEFC down to 250 K in order to maximize the ME effect in 
Cr
2
O
3
.[114] Typical H
E
 was 550 Oe in the Co/Pt multilayer at this temperature. 
However, no difference in the shape of the dual hysteresis was detected for the two 
MEFC conditions. Given the sensitivity of our measurement set up, we place the 
upper bound in the shift of the H
E
 to be less than 5 Oe. In addition, we have also 
looked for direct electric field effect under different constant magnetic field values (0, 
? 0.13 kOe) at 298 K. We applied an ac electric field and used lock-in detection of 
the Kerr signal and found no detectable changes in H
E
 (sensitivity ~ 1 Oe) or 
magnetic moment (sensitivity better than 1% of total Co/Pt moment). 
Thus, despite the fact that we have been able to apply electric field about 150 
times higher than that used in the previous experiment on a Cr
2
O
3
 single crystal,[110] 
the apparent ME effect on the exchange bias appears to be minimal. One possible 
reason for the lack of electric field effect is that the magnetic spin orientations of the 
very top AFM layer in Cr
2
O
3
 are mostly affected by the magnetic spin configuration of 
the Co layer through exchange interaction rather than the ME effect in Cr
2
O
3
. 
Investigation of the possible ME effect on other exchange-biased AFM-FM multilayer 
multiferroic systems is also currently underway. 
 89 
 
Chapter 5 
5. Dynamic observation of ME effect using Lorentz TEM 
 
The mutual control of magnetic and electric ordering in multiferroic (or 
magnetoelectric) materials recently has been reinvestigated because of its increasing 
potential for the applications in magnetoelectric (ME) or spintronic memory devices 
and sensors.[72-75] The keystone of these applications is based on the combination of 
magnetism and electricity using, for example, spin-polarized carrier effects and giant 
magnetoresistance (GMR) effects. Another possible application is to further increase 
the storage ability by tuning magnetic or electric properties in a multiferroic material 
by ME coupling effects. 
In most cases the correlations between magnetism and electricity in ME 
materials have been evaluated from either magnetic field-induced voltage / electrical 
polarization (P = ?H) or electric field-induced magnetization (M = ?E). There has 
also been a report on the observation of the change in magnetization in a 
ferromagnetic layer upon an applied electric field on a multiferroic layer.[115] This 
report demonstrated change of magnetization in a CoFe layer due to an exchange type 
of interaction with BiFeO
3
 which is antiferromagnetic. The change was produced 
when an electric field changed the direction of polarization in the BiFeO
3
 layer. This 
change in polarization induced a change in the orientation of the magnetic moments 
in the BiFeO
3
 layer. This last change gave rise to an exchange interaction with the 
CoFe layer. However, there has been no evidence, so far, for the direct observation of 
strain-mediated ME coupling, especially, control of magnetic domains by strain 
 90 
 
induced from both magnetic and/or electric fields. This is a very important 
mechanism for ME memory storage devices. There are several difficulties that have 
prevented the observation of strain-mediated ME coupling. The main problems are 
the difficulty in applying a strong enough electric field for the bulk samples and 
clamping effect from the substrate that exists in thin film structures. Here, we 
describe direct observation of strain-mediate ME coupling in an unconstrained bilayer 
using in-situ Lorentz transmission electron microscopy. The strain propagation and 
reversible switching of magnetic domains in a ferromagnetic layer were dynamically 
observed under an applied electric field to a ferroelectric layer. 
 
5.1 Strain-mediated ME effect using Fe-Ga/BaTiO
3
 bilayer 
 
For this experiment, we used a bilayer thin film structure of a ferromagnetic 
(magnetostrictive) phase and a ferroelectric (piezoelectric) phase. Firstly, for the 
ferroelectric layer, a 200 nm thick BaTiO
3
 thin film was deposited on a SrTiO
3
 
substrate at 760 
o
C and 100 mTorr oxygen pressure using PLD. As shown in the 
X-ray diffraction spectrum in Fig. 5.1 (a), the BaTiO
3
 film was heteroepitaxially 
grown on (001) SrTiO
3
. BaTiO
3
 is a ferroelectric material with a perovskite structure. 
The saturation polarization measured from a separately made Fe
0.7
Ga
0.3
(top 
electrode)/BaTiO
3
/SrRuO
3
(bottom electrode)/SrTiO
3
 was P
S
 ~ 17 ?C/cm
2
, and the 
remanent polarization was Pr ~ 7 ?C/cm
2
 [see Fig. 5.1 (b)]. The piezoelectric 
coefficient measured for these films was d
33
 = ~ 50 pm/V.[118] 
 
 91 
 
 
 
 
 
 
 
 
 
 
Figure 5.1 (a) X-ray diffraction spectrum showing epitaxial growth of BaTiO
3
 
thin film on (001) oriented SrTiO
3
. (b) Electrical polarization versus applied electric 
field showing ferroelectric hysteresis curve of BaTiO
3
. (c) Bright field TEM image of 
polycrystalline Fe
0.7
Ga
0.3
 thin film. (d) M-H hysteresis curves obtained from the 
600nm thick Fe-Ga thin film along the out-of-plane and in-plane directions. 
 
Over the BaTiO
3
 film we deposited a 60 nm thick polycrystalline film of 
Fe
0.7
Ga
0.3
 for the FM layer by magnetron sputtering at room temperature. This layer 
was also used as top electrode for the application of voltage to the BaTiO
3
 layer. As 
shown in the bright field TEM image in Fig. 5.1 (c), the grain size of the 
polycrystalline Fe-Ga film was about 50 nm in diameter. The in-plane and out-of-
plane magnetization were obtained from a thicker Fe-Ga film (600 nm) using a 
vibrating sample magnetometer (VSM). The film showed strong magnetic anisotropy 
along the in-plane direction, and the anisotropy increased for films with smaller 
thickness. Fe
0.7
Ga
0.3
 is known to show relatively large magnetostriction. The 
magnetostrictive coefficient (?
eff
 = ~ 150 ppm for thin film) was measured at room 
temperature by measuring the change of length along the magnetic field after 
 92 
 
applying a magnetic field and saturating the magnetic moments from a film deposited 
on a cantilever. 
 
 
 
 
 
 
 
 
 
Figure 5.2 (a) Schematic for Lorentz TEM sample preparation. After 
deposition of FE and FM bilayer thin film structure, FM layer was patterned using 
optical lithography and electron beam lithography to apply an electric field. 
Schematics of (b) randomly oriented magnetic domains in the initial state, (c) induced 
tensile strain along field direction after applying an electric field (E), and (d) 
alignment of magnetic domains along tensile strain direction. 
 
As shown in Fig. 5.2 (a), to apply an electric field along an in-plane direction, 
the conducting Fe-Ga layer was used as electrodes after patterning using photo- 
lithography and electron-beam lithography. To eliminate the clamping effect from the 
substrate, we removed the SrTiO
3
 substrate from a 10 ? 10 ?m
2
 window area using 
focused ion beam (FIB) milling and prepared a thin TEM sample to apply an electric 
field in the Lorentz TEM. BaTiO
3
 and Fe-Ga are known to have positive values in 
piezoelectric and magnetostriction coefficients, respectively. Therefore, as shown in 
the schematics in Figs. 5.2 (b), (c) and (d), if we apply an electric field on this sample 
 93 
 
along the in-plane direction, the BaTiO
3
 will be polarized along the field direction 
and the displacement of Ti
4+
 cation along the field direction is accompanied by a 
tetragonal distortion of the perovskite structure also along the field direction. This 
distortion induces tensile strain along the same direction. This distortion and tensile 
strain will be maximized in this structure because there is no clamping effect from the 
substrate. The induced strain will be transferred to the upper Fe-Ga layer giving rise 
to alignment of the magnetic domains along the tensile strain direction [Fig 5.2 (d)]. 
 
5.2 Magnetic domain structure of Fe-Ga by MFM and LTEM 
 
We performed further characterization of the Fe-Ga magnetic domain 
structure of a 60 nm thick Fe-Ga patterned by photolithography [see Fig. 5.3 (a)] 
using magnetic force microscopy (MFM). MFM is more sensitive to the magnetic 
moments which are oriented along the perpendicular direction to the surface (out-of-
plane direction). 
 
 
 
 
 
Figure 5.3 (a) Fe-Ga patterned on BaTiO
3
/SrTiO
3
. (b) Height image acquired 
from 1.5 ?1.5 ?m
2
 area in the middle pad. The data scale was 0 ~ 5 nm (c) MFM 
age from the same region. The data scale was 0 ~ 6 degree. 
 
 
im
 94 
 
From the AFM height image in Fig. 5.3 (b), the grain size of Fe-Ga film was 
bout 50 nm in diameter, which is in agreement with our TEM analysis from the 
mple. However, the magnetic domain image in Fig. 5.3 (c) shows very weak 
intensit
ned along the in-plane 
direction using Lorentz TEM. In LTEM, we can only observe contrast from the 
magnetic domains which have easy axis in-plane by detecting the Lorentz force on 
the incident electron beam from the domains. As discussed in section 2.2.2, there are 
two different techniques for magnetic domain imaging in Lorenz TEM. One is 
Fresnel imaging and the other is Foucault imaging. 
 
Fresnel image of polycrystalline Fe-Ga thin film 
 
In the Fresnel imaging mode, by changing the defocusing value in the TEM, the 
domain boundaries can have more or less electrons, which give rise to bright or dark 
contrast of the domain boundaries, respectively. As shown in Fig. 5.4, clear contrast 
from the magnetic domains in Fe-Ga was observed using Fresnel imaging. From the in-
focus image, we cannot see contrast from the magnetic domain walls, but by changing 
the defocusing value, bright and dark lines appear at the domain boundaries. The 
contrast from the domain boundaries changes from dark to bright and bright to dark by 
changing the focal length from under to over focus. From the magnified domain image 
and phase map reconstructed by transport of intensity equation,[116] we see 180 degree 
domain walls, cross-tie walls, and magnetization ripple in the polycrystalline Fe-Ga. 
a
sa
y indicating that the magnetic domains are preferentially aligned in-plane. We 
also investigated the magnetic domains which are mostly alig
 95 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 5.4. Fresnel images of Fe-Ga thin film obtained at various defocusing 
values of (a) over focus (+ 1.2 mm), (b) in focus (zero), and (c) under focus (? 1.2 
mm). (d) Phase map from the area marked in (c) reconstructed by transport of 
intensity equation (TIE).[116] 
oucalt image of polycrystalline Fe-Ga thin films 
llow one of the four (000) deflected electron beams in the diffraction pattern to form 
mains. The dark and bright areas 
 each image in Fig. 5.5 exhibit magnetic domains oriented antiparallel to each other. 
 
F
 
We also investigated the magnetic domain structure using Foucalt imaging, as 
shown in the schematic of Fig 2.2 (b). Using an objective aperture, we can selectively 
a
a dark field image of the corresponding magnetic do
in
 96 
 
By cha
position agnetic 
omains which are oriented along x direction or y direction show as bright and dark 
areas.  
To further understand the magnetization of the domains, a magnetic field was 
pplied along the in-plane direction. In Lorentz TEM, the current through the 
objectiv
e in-plane direction of the sample as well as to control the applied field, one can tilt 
the sample to an arbitrary degree and then control the current through the objective 
nging the aperture from positive x (or y) position to negative ? x (or ? y) 
positions, the contrast of the magnetic domains changes from dark to bright and 
bright to dark. 
 
 
 
 
 
 
 
 
 
 
 
Figure 5.5. Foucalt images of Fe-Ga thin film. By changing the aperture 
 along x direction [(a) and (b)] and y direction [(c) and (d)], m
d
 
a
e lens was adjusted to control the magnetic field on the sample. This field is 
perpendicular to the surface of the sample. Therefore, to apply a magnetic field along 
th
 97 
 
lens. O
 under an 
applied ne direction 
ed arrow) and increased by increasing the tilt angle from 0 to 9 degrees. The number 
from 1 to 9 represents the sequence. The small arrows represent the direction of the 
agnetic domains 
 
As shown in the Fig. 5.6, the magnetic domain boundaries moved in direction 
perpendicular to the field direction (towards the left of the image) by increasing the 
magnetic field component on the sample by tilting the sample from 0 ? 9 degrees. 
ne can also fix the current and change the tilting angle of the sample. Here, we 
used the latter method and increased the component of the magnetic field parallel to 
the film surface by increasing the tilting angle while keeping the current through the 
objective lens constant. 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 5.6. Change of magnetic domains in an Fe-Ga thin film
 external field. The magnetic field was applied along the in-pla
(r
m
 98 
 
Finally,
ing the in-plane component of the magnetic field, the magnetic 
domain boundaries formed again (not shown here). Also, by tilting the sample in the 
opposite direction, the magnetic domains along the opposite direction (yellow arrow 
in Fig. 5.7) increased and the domain wall moved to the right hand side of the images. 
 
 
 
 
 
 
 
 
 
 
 
 
 
agnetic 
eld (yellow arrow) was applied along opposite direction (compare to Fig. 5. 6) and 
increased by increasing the tilt angle from 0 ? 9 degrees. The number from 1 to 9 
presents the sequence. 
 all magnetic moments were aligned along the field direction and the magnetic 
domain boundaries disappeared from the field of view. After tilting the sample back 
to zero, i. e., reduc
Figure 5.7.  Change of magnetic domains in Fe-Ga thin film. The m
fi
re
 99 
 
5.3 TEM sample preparation using electron beam lithography and 
ocused-Ion Beam (FIB) milling 
After cleaning the surface of BaTiO
3
 using acetone, S1813 photo-resist was spin 
coated. A spin speed of 4000 rpm was used for 40 seconds for coating, and then the 
electrode was patterned by exposing it to ultraviolet light for 11 seconds in an MJB-3 
mask aligner. After developing the pattern by soaking the sample into the CD 30 
solution for 45 seconds, Fe-Ga was deposited by magnetron sputtering. After 
stripping, to make a gap in the middle of the pad, we spin coated alternating layers of 
PMMA 950 and PMMA 495 electron beam resists. 
 
 
 
 
 
 
 
 
Figure 5.8 Process for deposition of Fe-Ga electrodes. First the electrodes 
were patterned using photo lithography. After developing the pattern Fe-Ga was 
deposited by magnetron sputtering at room temperature. Then, electron-beam 
lithography was used to pattern a gap on the middle Fe-Ga electrode. Finally, the 
TEM sample was prepared by mechanical polishing and FIB milling. 
 
F
 
To apply an electric field on the BaTiO
3
 layer, the Fe-Ga layer was patterned 
to form electrodes by photo lithography. Figure 5.8 shows the process for patterning. 
 100 
 
The multilayer of resist was used to protect the Fe-Ga layer from ion milling. 
After soft baking, a 1 ?m wide gap was patterned using a JEOL JSM 840 scanner. 
ple in MIBK-IPE for 40 seconds, and 
then duplicated on the F
 
 
 
 
 
 
 
Figure 5.9 SEM images obtained at different stages of FIB milling. (a) after 
illing using a beam with largest beam current (11500 pA) and (b) second largest 
current (6600 pA). SEM images number from 1 to 6 show gradual change of contrast 
y decreasing thickness. (c) and (d) are SEM images from the bottom view and end 
view, respectively, after milling was completed. 
 
The pattern was developed by soaking the sam
e-Ga using ion milling. The ion milling time to remove 60 nm 
thick Fe-Ga layer was about 210 seconds at 3.5 kV and 1.3 mA. After ion milling, the 
thickness and magnetic properties of the Fe-Ga layer were measured by AFM and 
MFM, respectively, and then oxygen plasma cleaning was performed to clean the 
surface and improve conductivity.  
 
 
 
 
m
b
 101 
 
To prepare the TEM samples, mechanical polishing and FIB milling (Dual 
eam 620, FEI) were used. By mechanical polishing, the size of the sample was 
duced to less than 3 mm in diameter, which is the maximum sample size for TEM 
ecimen holder. A slope with an angle of 45 degrees was made on the backside 
(substra
ig. 5.9 (b)]. As 
the SrTiO
3
 thickness decreased more, we decreased the beam current, in steps of 6600, 
2700, 1000, 350 and 157 pA, and the milled facet of SrTiO
3
 became uniform and flat. 
The figures numbered 1 to 6 in Fig. 5.9, show the change of contrast due to the 
decrease of thickness of the SrTiO
3
 and BaTiO
3
 layers. After finishing the FIB 
milling [Fig. 5.9 (c) and (d)], the 1 ?m size gap was clearly visible in the window area 
(15 ?m ? 10 ?m). The SrTiO
3
 was completely removed from this area and the total 
thickness of the remaining BaTiO
3
 and the Fe-Ga was less than 200 nm. 
 
5.4 Switching magnetic domains by applying a magnetic field 
 
The magnetic domain structure of the unconstrained Fe-Ga/BaTiO
3
 thin film 
was investigated using a JEOL 2100 LaB
6
 TEM. Figures 5.10 (a) and (b) are Fresnel 
images of magnetic domain boundaries from as-grown sample. By changing the focal 
length from over (?F = + 4500 nm) to under focus (?F = ? 4500 nm), the domain 
boundaries appeared and changed their contrast from dark line to bright line and vice 
versa. Figures 5.10 (c) and (d) are Foucalt images of the magnetic domains acquired 
b
re
sp
te side) of the sample to avoid blocking the transmitted beam, and then the 
substrate was milled to form a window region of  15 ? 15 ? 15 (?m)
3
 using focused 
ion beam with 11500 pA beam current [Figs. 5.9 (a)] and 6600 pA [F
 102 
 
by placing the aperture along the y axis, which is parallel to the gap (trench) direction 
in the Fe-Ga layer. By changing the position from + y to ? y, the contrast of magnetic 
domains which are oriented along the ? x direction (perpendicular to the gap) 
changed. Analysis of the Foucalt images indicates that there coexist 90 
o
 domains 
(mainly on the left side of the gap) and 180 
o
 domains on the right side of the gap 
which show clear boundaries. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 5.10 Magnetic domain structure of fully unconstrained Fe-Ga/BaTiO
3
 
thin film. Fresnel images obtained at (a) over focus (?F = + 4500 nm) and (b) under 
focus (?F= ? 4500 nm). (c) and (d) are Foucalt images obtained by moving the 
aperture along ?y axis. 
 
 103 
 
To further investigate the behavior of the magnetic domain structure of 
polycry
 
stalline Fe-Ga, a magnetic field was applied in the sample area. To apply a 
magnetic field, the sample was tilted by + 9 degree, and then the focal length was 
changed to over focus. The current in the objective lens was increased in steps to 
increase the magnetic field on the sample and Fresnel images were obtained at every 
step as indicated in Fig 5.11. In this fashion, the component of the magnetic field 
along the in-plane direction of the sample was increased. 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 5.11 Increase of magnetic field component in-plane by tilting the 
sample to + 9.0 
o
 and increasing the current though the objective lens (numbers in 
each figure). 
 
 104 
 
As shown in Figs. 5.11, the magnetic domain boundaries moved in the 
direction parallel to the magnetic field direction and eventually all domain boundaries 
disappeared indicating that all magnetic moments were aligned along the direction of 
the field. After all the magnetic domains were aligned along the field direction, the 
magnetic field was decreased back to zero, and a multidomain structure of magnetic 
domains reformed with similar configuration to the initial state (see Fig. 5.12). 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 5.12 Decrease of magnetic field component in-plane by tilting the 
sample (+ 9.0 
o
) and decreasing the current through the objective lens (numbers in 
each figure). 
To ascertain that the effect observed was due to domain wall motion, we 
pplied in-plane magnetic field and observed the change of domain configuration in 
oucalt imaging mode. After tilting the sample to zero and applying a magnetic field 
 
a
F
 105 
 
along the out-of-plane direction through the objective lens current, we tilted the 
mple back to 9 degrees to induce an in-plane component of the magnetic field on 
e sample.  
 
 
Figure 5.13 Increase and decrease of magnetic field along in-plane by tilting 
the sample (+ 9.0 
o
) and changing the current through the objective lens (numbers in 
each figure). (a) Fresnel mode at 0 
o
 tilt. (b) Foucalt mode at 0 
o
 tilt. (c) ? (h) Foucalt 
images corresponding to an in-plane component of the magnetic domains. (i) The 
new magnetic domains after reducing the magnetic field back to zero. Red and light 
blue arrows indicate magnetic domains orientation. 
 
sa
th
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 106 
 
The first two images in Fig. 5.13 represent the initial magnetic domain 
ructure objective aperture 
round + x spot) imaging modes, respectively. By increasing the magnetic field [Figs. 
.13 (c ? h)] along the y direction, the magnetic moments gradually switch to the 
irection of the field and eventually all magnetic domains were aligned along that 
irection. This observation corroborated the results obtained in Fresnel imaging mode. 
fter reducing the magnetic field back to zero [Fig. 5.13 (i)], a multi domain structure 
milar to the initial state was obtained. From the relationship between electric current 
 the objective lens and the in-plane component of the magnetic field calibrated in 
able 1, we established the maximum value of the applied in-plane magnetic field to 
e ~ 196 Oe which is close to the magnetic field required to saturate the 
agnetization of Fe-Ga thin film (~ 300 Oe). 
 
 
in the 
LaB
6
 TEM and the corresponding out-of-plane and in-plane components (by 9 degree 
lting) of the magnetic field on the sample. 
 
st  at 0 degree tilt obtained by Fresnel and Foucalt (with 
a
5
d
d
A
si
in
T
b
m
 
 
 
 
 
 
 
Table 1. Relationship between the current through the objective lens 
ti
 107 
 
 
 
 
 
 
 
 
 
 
Figure 5.14 Change of tilting angle and focal length from (a) + 9.0 (F = ? 
000 nm) to (b) ? 9.0 degree (F = + 6000 nm). 
Figure 5.14 shows Fresnel images at different tilting angles and focal lengths. 
o differentiate between the magnetic domain structure and bend contours, the 
mple was tilted from + 9 [Fig. 5.14 (a)] to ? 9 degrees [Fig. 5.14 (b)] in field-free 
onditions. Tilting the sample did not change the contrast of the magnetic domain 
alls. However, the bend contours were completely changed. Furthermore, the 
hange of focal length from ? 6000 nm [Fig. 5.14 (a)] to + 6000 nm [Fig. 5.14 (b)], 
learly distinguished the magnetic domain walls because of the change in contrast 
from b
 
 
6
 
T
sa
c
w
c
c
right to dark and dark to bright from the bend contours which did not change 
contrast. 
 
 108 
 
 
plied 
ma
The evaluation of magnetic domain structure with magnetic field direction is 
own in Fig. 5.15. A magnetic field was applied along the in-plane direction [Figs. 
.15 (b), (d), and (f)] followed by an out-of-plane [(c) and (e) images in Fig. 5.15] 
irection of the magnetic field alternately to see the change of magnetic domain 
onfiguration with field direction. After removing the magnetic field, some magnetic 
oments switched back to the original direction (x-direction) and formed new 
omains which are aligned along the x-direction as shown in Fig. 5.15 (b), (d), and (f). 
his preferred orientation of the domains is related to the misfit strain relaxation of 
the unc
 
 
 
 
 
 
 
 
 
 
 
Figure 5.15 Change of magnetic domain structure according to ap
gnetic field direction (indicated in the images). 
onstrained BaTiO
3
 and clamping at the edges of the FIB prepared window 
 
sh
5
d
c
m
d
T
 109 
 
region. These complex interactions cause the asymmetric shape of BaTiO
3
, i.e., 
elongation of BaTiO
3
 along the x direction in the edge region and result in non-
uniform strain distribution and the bend contour. We also applied a magnetic field 
along t
3
 the electric field would be 
on should give preferred orientation of the 
agnetic domains along the y-direction. Once the electric field is applied, we expect 
ore dom ic field direction). 
the electric field 
 zero [Fig. 5.16 (a)] to 10 V [5.16 (c)], and we observed changes in the magnetic 
ains which had been pre-aligned along the y-axis (in-plane component of the 
agnetic field). The magnetic domain structure created by the H-field along the y-
tion is less stable than the magnetic domain structure created by applying an out-
he direction perpendicular to the surface and then removed the field. As shown 
in Fig. 5.15 (c) and (e), the magnetic domains under this condition were similar to the 
as grown state. The bend contours also showed similar configuration before and after 
the application of the magnetic field. 
 
5.5 Switching magnetic domains by applying an electric field (ME 
effect) 
 
In order to investigate if we could observed the strain-mediated ME effect 
between BaTiO  and Fe-Ga, we analyzed the magnetic domain configuration under 
an applied electric field. For this purpose, we first applied a magnetic field in the y-
direction (along the gap) perpendicular to the direction that
applied next (x-direction). This configurati
m
to see m ain re-orientation along the x-direction (the electr
After bringing the magnetic field back to zero, we increased 
from
dom
m
direc
 110 
 
of-plane magnetic field. By applying an electric field, a tensile strain is induced in the 
BaTiO
3
ains from the y-direction to the x-
irection. During the application of voltage, there was no change in the magnetic 
omain configuration of the Fe-Ga layer until the voltage was raised to 7.5 volts [Fig. 
.16 (b)]. At this point, there was a sudden increase in volume of the domains with 
agnetic moment along the x-axis. With continued increase in voltage to 10 volts, 
ere was another sudden change in the magnetic domain structure [Fig. 5.16 (c)]. 
hese results indicate that there is a resistance of the domains to switch and only 
hen enough strain energy has built up, the domains suddenly grow. The magnetic 
omains changed at an applied voltage of 7.5 V [Fig. 5.16 (b)] and 10 V [Fig. 5.16 (c)]. 
he leakage current versus applied voltage during this observation was measured and 
 shown in Fig. 5.16 (d). This curve shows a slight reduction in current at 7.5 and 10 
olts when the domain boundaries moved. At a higher voltage of 11 volts the leakage 
current
reak down. 
 
 thin layer along the x-direction. This tensile strain is transferred to the Fe-Ga 
layer and results in a change of the magnetic dom
d
d
5
m
th
T
w
d
T
is
v
 density increased dramatically and the sample experienced an irreversible 
b
 
 
 
 
 
 
 111 
 
 
 
 
 
Figure 5.16 Switching of magnetic domains by the application of an electric 
eld. (a) E = 0 V, (b) E = 7.5 V, and (c) E = 10 V. (d) Acquired leakage current 
versus applied electric field (voltage). The arrows indicate the changed regions. 
The volume of the domains with magnetic moment along the x-axis was 
creased after applying an electric field. As shown in Fig. 5.17, the newly formed 
fi
 
in
 112 
 
domains (red-colored region) after removing the magnetic field (y-direction) are 
mainly
rain along the x-direction, these domains merged and increased in volume (blue-
colored
 
 
 
agnetic domains before and after applying an 
 represents the magnetic domains with 
magnetization along x before applying an electric field (E = 0 V) and the blue-colored 
region represents the increased area of magnetic domains after applying an electric 
 oriented along x-direction. After applying an electric field and induce the 
st
 region) by switching the magnetic moment of neighboring domains from the 
y-direction to the x-direction 
Therefore, our results show dynamic observation in Lorentz TEM of strain-
mediated magnetoelectric effect by the application of an electric field on the sample. 
 
 
 
 
 
Figure 5.17 Comparison of the m
electric field. Red-colored region
field (E = 10 V). The arrows indicate magnetic domains orientations. 
 113 
 
Chap
6. Summary and future work 
changes to the m
 
switching of ferroelectric domains in columnar structure of BiFeO
3
-Fe
2
O
3
 films, the 
ter 6 
 
In this thesis, the ferroelectric and ferromagnetic properties in multiferroic 
BiFeO
3
 thin films were investigated. The heteroepitaxial constraint created in a film 
structure by the substrate and the volatile nature of bismuth cause complicated 
icrostructure and physical properties of BiFeO
3
 films. Secondary 
phases, such as, ?-Fe
2
O
3
, ?-Fe
2
O
3
, and Fe
3
O
4
 formed spontaneously in the films 
grown at either relatively high growth temperature or low oxygen pressure. The 
formation of ferromagnetic ?-Fe
2
O
3
 and Fe
3
O
4
 significantly increased the saturation
magnetization of the films. To enhance the ferromagnetism of columnar structure of 
the sample with BiFeO
3
-Fe
2
O
3
 thin films, a phase transformation from 
antiferromagnetic ?-Fe
2
O
3
 to ferromagnetic ?-Fe
2
O
3
 was induced by annealing under 
a hydrogen atmosphere followed by annealing in oxygen atmosphere. 
The ferroelectricity of BiFeO
3
 was studied in films with columnar structure of 
BiFeO
3
-Fe
2
O
3
 as well as films with polycrystalline BiFeO
3
. Both of them showed 
larger electrical polarization as well as less or no misfit strain compared to single 
crystalline epitaxial films with single BiFeO
3
 phase. From our results, the observed 
increased polarization in thin film structures probably originated from an intrinsic 
property of BiFeO
3
 single crystal rather than the misfit strain and/or tetragonal 
distortion in heteroepitaxial thin film. 
To further understand the origin of the ferroelectric properties and enhanced 
 114 
 
relative concentrations of  Fe
2+
, Fe
3+
 and O
2-
 have to be measured by EELS across the 
two phases to obtain any compositional variations that might exist among the two 
phases. Also, convergent beam electron diffraction (CBED) is required to investigate 
the strain distribution and difference in strain relaxation near the interfacial area and 
center of BiFeO
3
 columns.  
In an effort to enhance the structural, magnetic, and electrical properties of the 
iFeO
3
 films, we investigated various ways for improving the crystallinity of the 
lms using a new technique called flux-mediate epitaxy (FME), and achieved single 
rystal-like thin films with large grains, smooth surfaces and without impurity phases.. 
ompared to PLD grown films FME films showed improved dielectric properties 
igher dielectric constant and lower leakage current). 
The present studies also include the possible applications of magnetoelectric 
oupling effects, in noble spintronic devices in which magnetoresistance and 
xchange bias are controlled by ME effects. We investigated this effect using a Co/Pt 
multilayer on Cr
2
O
3
 thin films. We achieved an extremely smooth interface between 
Co/Pt and the Cr
2
O
3
 layer (RMS ~ 0.3 nm) to reduce unexpected effects from 
roughness, and more importantly, enhanced resistance of the Cr
2
O
3
 film. We were 
able to apply sufficiently high electric field 150 times higher than the applied field on 
earlier single crystal experiments. Despite our high quality sample and the high field, 
the thin film device exhibited no significant ME effect on exchange bias in the 
sensitivity range of our equipment. However, in this work, we showed an efficient 
way for controlling the exchange bias field and its sign by sweeping the magnetic 
field with an offset during cooling. We controlled the sign of the exchange bias and 
B
fi
c
C
(h
c
e
 115 
 
the shape of the hysteresis loops, i.e., single/dual hysteresis loops of the out-of-plane 
magnetized Co/Pt layers by sweeping a magnetic field with various offsets during 
ma
 
 
gnetic 
dom  
stra
ma
A mo
obtain an estim ate the amount of 
strain on the sam should also be able 
to estimate the dom ry to move it. Also, 
dditional attempts can be challenging, for example, the shape dependence on 
switchi
ence, 
electrodes with circular, oval and square shape could be compared. 
gnetic field cooling. 
In addition, we demonstrated for the first time the dynamic observation of 
strain-mediated ME coupling effect using ferromagnetic Fe-Ga and ferroelectric
BaTiO
3
 bilayer thin film structure. The reversible switching of magnetic domains in 
the ferromagnetic layer was observed using in-situ Lorenz TEM while increasing an
applied electric field and inducing strain on the ferroelectric layer. This is the first 
direct observation of strain-mediated ME coupling, especially, control of ma
ains by strain induced from an electric field. We observed direct evidence for
in-mediate ME coupling from the strain propagation and reversible switching of 
gnetic domains by an applied electric field in Lorenz TEM. 
re detailed analysis of the Lorentz TEM images should be carried out to 
ate for ?, the magnetoelectric coupling, and to estim
ple after the application of the electric field. We 
ain wall energy by estimating the work necessa
a
ng of Fe-Ga magnetic single domains or the control of magnetic domains by 
applying an electric field along two different orientations. For the shape depend
 116 
 
References 
1 W. Eerenstein, N. D. Mathur, and J. F. Scott, Nature 442, 759 (2006). 
2 P. Weiss, J. Phys. 6, 66
3 E. C. Stoner, Philos. Mag. 
1 (1907). 
15, 1080 (1933). 
4 
5 
Hill, N. A., J. Phys. Chem. B 104, 6694 (2000). 
 N. Hur, S. Park, P. A. Sharma, J. S. Ahn, S. Guha and S. -W. Cheong, Nature 
429, 392 (2004). 
 Ryu, J., V?squez Carazo, A., Uchino, K.& Kim, H. ?E. Jpn. J. Appl. Phys. 40, 
4948 (2001). 
S. X. Dong, J. F. Li, and D. Vieland, Appl. Phys. Lett. 83, 2265 (2003.) 
1 V. J. Folen, G. T. Rado and E. W. Stalder, Phys. Rev. Lett. 6, 607 (1961). 
12 G. T. Rado and V. J. Folen, Phys. Rev. Lett. 7, 310 (1961).  
3 E. Ascher, H. Rieder, H. Schimid, and H. St?ssel, J. Appl. Phys. 37, 1404 
(1966). 
14 F. Bertaut, F. forrat, and P. Fang, C. R. Acad. Sci. 256, 1958 (1963).  
5 V. A. Bokov, G. A. Smolenski?, S. A. Kizhaev, and I. E. MyI'nikova, Sov. 
Phys. Solid State 5, 2646 (1964).  
I. G. Ismailzade and S. A. Kizhaev, Sov. Phys. Solid State 7, 236 (1965). 
7 J. Chappert, Phys. Lett. 18, 229 (1965). 
18 M. Eibsch?tz and H. J. Guggenheim, Solid State Commun. 6, 737 (1968). 
9 L. Holmes, M. Eibsh?tz, and H. J. Guggenheim, ibid. 7, 973 (1969). 
0 J. Wang, J. B. Neaton, H. Zheng, V. Nagarajan, S. B. Ogale, B. Liu, D. 
Viehland, V. Vaithyanathan, D. G. Schlom, U. V. Waghmare, N. A. Spaldin, 
K. M. Rabe, M. Wuttig, R. Ramesh, Science 299, 1719 (2003). 
K.Y. Yun, M. Noda, M. Okuyama, H. Seki, H. Tabata and K. Saito, J. Appl. 
25 Van Run, A. M. J. G., Terrell, D. R., and Scholing, J. H., J. Mater. Sci. 9, 
1710 (1974). 
26 Nan, C. ?W, et al. Appl. Phys. Lett. 81, 3831 (2002). 
27 Cai, N., Nan, C. ?W, Zhai, J. and Lin, Y. Appl. Phys. Lett. 84, 3516 (2004). 
28 Srinvasan, G. et al. Phys. Rev. B 65, 134402 (2002). 
29 Lee, M. K. et al. Appl. Phys. Lett. 77, 3547 (2000). 
30 J. Van Suchtelen, Philips Res. Rep. 27, 28 (1972). 
31 J. Van Den Boomgaard, D. R. Terrelll, R. A. J. Born, and H. F. J. I. Giller, J. 
Mater. Sci. 9, 1705 (1974). 
G. A. Smolenskii and I. E. Chupis, Sov. Phys. Usp. 25, 475 (1962). 
S. Alexander and S. Shtrikman, Solid State Commun. 4, 115 (1966).  
6 E. Fischer, G. Gorodetsky and R. M. Hornreich, Solid State Commun. 10, 
1127 (1972). 
7 
8
9
10 
1
1
1
16 
1
1
2
21 
Phys. 96, 3399 (2004). 
22 Yu. F. Popov, A. M. Kadomtseva, G. P. Vorobev, A. K. Zvezdin, 
Ferroelectrics, 162, 135 (1994). 
23 J. R. Teague, R. Gerson, W. J. James, Solid State Commun. 8, 1073 (1970) 
24 V. A. Murashav, D. N. Rakov, V. M. Ionov, I. S. Dubenko, Y. U. Titov, 
Ferroelectrics, 162, 11 (1994). 
 117 
 
32 A. Hanu
Bull. Mater. Sci. 
maiah, T. Bhimasankaram, S. V. Suryanaryana, and G. S. Kumar, 
17, 405 (1994). 
34 W. E. Kramer, R. H. Hopkins, and M. R. Danel, J. Mater. Sci. Lett. 12, 409 
(1977). 
G. Harshe, J. P. Dougherty, and R. E. Newnham, Int. J. Appl. Electromagn. 
36 
50 
. Mater. 18, 2307 (2006). 
91 
57 
33 J. Van Den Boomgaard, A. M. J. G. Van Run, and J. Van Suchtelen, 
Ferroelctrics 14, 727 (1976). 
35 
Mater. 4, 145 (1993). 
M. Avellaneda and G. Harshe, J. Intell. Mater. Syst. Struct. 5, 501 (1994). 
37 D. N. Astrov, Sov. Phys. JETP 11, 708?709 (1960). 
38 G. A. Smolenski? and I. E. Chupis, Sov. Phys. Usp. 25, 475 (1982). 
39 G. A. Smolenski?, Fizika Tverdogo Tela 4, 1095 (1962). 
40 D. N. Astrov, Sov. Phys. JEPT 13, 729 (1961). 
41 J. R. Hattrick-Simpers, L. Dai, M. Wuttig, I. Takeuchi, and E. Quandt, Rev. 
Sci. Instrum. 78 106103 (2007) 
42 W. Eerenstein M. Wiora, J. L. Prieto, J. F. Scott, and N. D. Mathur, Nature. 
Mat. 6, 348-351 (2007). 
43 P. Gr?nberg, . R. Schreiber, Y. Pang, M. B. Brodsky, and H. Sowers, Phys. 
Rev. Lett. 57 2442 (1986). 
44 M. N. Baibich, J. M. Broto, A. Fert, F. Nguyen Van Dau, F. Petroff, P. 
Eitenne, G. Creuzet, A. Friederich, and J. Chazelas Phys. Rev. Lett. 61 2472 
(1988). 
45 M. N. Baibich et al. Phys. Rev. B 39 4828 (1989). 
46 J. S. Moodera, Lisa R. Kinder, Terrilyn M. Wong, and R. Meservey, Phys. 
Rev. Lett. 74 3273 (1995). 
47 G.A. Prinz, Science 282, 1660 (1998). 
48 J. Grollier, V. Cros, A. Hamzic, J. M. George, H. Jaffr?s, A. Fert, G. Faini, J. 
Ben Youssef, and H. Legall, Appl. Phys. Lett. 78 3663 (2001). 
49 S. Urazhdin, Norman O. Birge, W. P. Pratt, Jr., and J. Bass, Appl. Phys. Lett. 
84, 1516 (2004). 
Ch Binek and B Doudin, Magnetoelectrics with magnetoelectrics. J. Phys.: 
Condens. Matter 17 (2005) L39. 
51 H. B?a, M Bibes, A. Barth?l?my, K. Bouzehouane, E. Jacquet, A. Khodan, J.-
P. Contour, S. Wyczisk, A. Forget, D. Lebeugle, D. Colson and M. Viert, 
Appl. Phys. Lett. 87, 072508 (2005). 
52 M. Murakami, S. Fujino, S.-H. Lim, L. G. Salamanca-Riba, M. Wuttig, I. 
Takeuchi, H. Sugaya, T. Hasegawa and S. E. Lodland, Appl. Phys. Lett. 88, 
112505 (2006). 
53 Y. ?Y. Li Phys. Rev. 101, 1450 (1956). 
54 A. J. Koch, J. J. Becker, A. E. Berkowitz, W. J. Schuele and P. J. Flanders, J. 
Appl. Phys. 39, 1261 (1968). 
55 Y. ?H. Chu, Q. Zhan, et al. Adv
56 D. Lebeugle, D. Colson, A. Forget, and M. Viret, Appl. Phys. Letts. 
022907 (2007). 
K.Y. Yun, M. Noda, M. Okuyama, H. Seki, H. Tabata and K. Saito, J. Appl. 
Phys. 96, 3399 (2004). 
 118 
 
58 J. R. Teague, R. Gerson, W. J. James, Solid State Commun. 8, 1073 (1970). 
Ferroelectrics 162, 135 (1994). 
Ferroelectrics, 162, 11 (1994). 
K. Zvezdin, D. Viehland Appl. Phys. Lett. 84, 255261 (2004). 
S.E. Lofland, L. G. S
59 Yu. F. Popov, A. M. Kadomtseva, G. P. Vorobev, A. K. Zvezdin, 
60 V. A. Murashav, D. N. Rakov, V. M. Ionov, I. S. Dubenko, Y. U. Titov, 
61 J. Li, J. Wang, M. Wuttig, R. Ramesh, N. Wang, B. Ruette, A. P. Pyatakov, A. 
62 M. Murakami, S. Fujino, S.-H. Lim, B. Varughese, H. Sugaya, T. Hasegawa, 
alamanca-Riba, M. Wuttig, I. Takeuchi, Appl. Phys. Lett. 
88, 112505 (2006). 
63 
Nagarajan , S. Fujino, M. Wuttig, I. Takeuchi, L. Salamanca-Riba, Adv. Funct. 
64 H. B?a, M. Bibes, A. Barth?l?my, K. Bouzehouane, E. Jacquet, A. Khodan, J. 
Viret, Appl. Phys. Lett. 87, 072508 (2005). 
Mathur, Science 307, 1203a (2005). 
67 B. J. Kim, E. T. Lee and G. E. Jang, Thin Solid films 341, 79 (1999). 
15, 595 (1967). 
69 J. W. Matthews, A. E. Blakeslee and S. Mader, Thin Solid Films 33, 253 
70 G. A. Smolenskii and I. E. Chupis, Sov. Phys. Usp. 25, 475 (1982). 
72 Ch Binek and B. Doudin, J. Phys.: Condens. Matter 17, L39 (2005). 
Mater. 18, 1445 (2006). 
75 N. Hur, S. Park, P. A. Sharma, J. S. Ahn, S. Guha and S.-W. Cheong, Nature 
76 M. Murakani, S. Fujino, S. -H. Lim, B. Varughese, H. Sugaya, T. Hasegawa, 
Lett. 88, 112505 (2006). 
P. Contour, S. Wyczisk, A. Forget, D. Lebeugle, D. Colson, and M. Viert, 
78 S. H. Lim, M. Murakami, W. L. Sarney, S. Q. Ren, A. Varatharajan, V. 
Mater. 17 2594 (2007). 
(2007). 
81 X. Qi, J. Dho, M. Blamire, Q. Jia, J. S. Lee, S. Foltyn, J. L. MacManus-
S. H. Lim, M. Murakami, W. L. Sarney, S. Q. Ren, A. Varatharajan, V. 
Mater. 17, 2594 (2007). 
-P. Contour, S. Fusil, F. Wyczisk, A. Forget, D. Lebeugle, D. Colson, M. 
65 W. Eerenstein, F. D. Morrison, J. Dho, M. G. Blamire, J. F. Scott and N. D. 
66 C. Ederer and N. A. Spaldin, Phys. Rev. B 71, 060401 (2005). 
68 J. W. Matthews and W. A. Jesser, Acta Metall. 
(1976). 
71 W. Eerenstein, N. D. Mathur, and J. F. Scott, Nature 442, 759 (2006). 
73 J. Dho, X. Qi, H. Kim, J. L. MacManus-Driscoll, and Mark G. Blamire Adv. 
74 J. F. Scott, Science 315, 954 (2007). 
(London) 429, 392 (2004). 
S.E. Lofland, L. G. Salamanca-Riba, M. Wuttig, and I. Takeuchi, Appl. Phys. 
77 H. B?a, M. Bibes, A. Barth?l?my, K. Bouzehouane, E. Jacquet, A.Khodan, J.-
Appl. Phys. Lett. 87, 072508 (2005). 
Nagarajan, S. Fujino, M. Wuttig, I. Takeuchi, L. Salamanca-Riba, Adv. Funct. 
79 D. Lebeugle, D. Colson, A. Forget, and M. Viret Appl. Phys. Lett. 91 022907 
80 Ederer and N. A. Spaldin, Phys. Rev. B 71, 060401 (2005). 
Driscoll, J. Magnetism and Magnetic Materials, 283 415 (2004).  
 119 
 
82 K. Y. Yun, D. Ricinschi, T. Kanashima, M. Okuyama, Appl. Phys. Lett. 89 
83 J. Dho, X. Qi, H. Kim, J. L. MacManus-Driscoll, and Mark G. Blamire Adv. 
84 J. F. Scott et al., 
192902 (2006). 
Mater. 18 1445 (2006.) 
Science 246, (1989). 
85 H. Park, J. Jung, D. Min, S. Kim, and S. Hong, Appli. Phys. Lett. 84 1734 
86 Y. Kim, Y. Cho, S. Hong, S. B?hlmann, H. Park, D. ?K. Min, S. ?H. Kim, 
87 J. Y. Li, R. C. Rogan, E. Ustundag and K. Bhattacharya, Nature Mat. 4, 776 
88 H. F. Kay and J. W. Dunn, Philo. Mag. 7, 2027 (1962). 
14337 (1996). 
H. Chu, C. Ederer, N. A. Spaldin, R. R. Das, D. M. Kim, S. H. Baek, C. B. 
91 R. Kretshmer and K. Binder, Phys. Rev. B 20, 1065 (1979.) 
93 A. Gruverman, A. Kholkin, A. Kingon, and H. Tokumoto, Appl. Phys. Lett. 
94 A. Gruverman, Appl. Phys. Lett. 75, 1452 (1999). 
. Nakajima, T. 
Chikyow, H. Koinuma, and Y. Matsumoto, Adv. Funct. Mater. 16, 485 (2006) 
96 
Growth 262, 308 (2004). 
Koinuma, Appl. Surf. Sci. 252, 2477 (2006). 
Matsumoto, J. Appl. Phys. 101, 033511 (2007). 
Petrov, and G. S. Zhdanov, Sov. Phys. JEPT 23, 47 (1996). 
Lett. 76, 2764 (2000). 
102 Ch. Binek, A. Hochstrat, X. Chen, P. Borisov, W. Kleemann, and B. Doudin, 
Condens. Matter 17, L39 (2005) 
104 D. N. Astrov. Sov. Phys. JETP 11, 708 (1960). 
105 I. E. Dzyaloshinskii, Sov. Phys. JEPT 10, 628 (1960). 
106 D. N. Astrov, Soviet Phys. JETP 13, 729 (1961). 
107 V. J. Folen, G. T. Rado, and E. W. Stalder, Phys. Rev. Lett. 6, 607 (1961). 
108 S. Shtrikman and D. Treves, Phys. Rev. 130, 986 (1963). 
109 P. Borisov, A. Hochstrat, X. Chen, W. Kleemann, and Ch. Binek, Phys. Rev. 
Lett. 94 117203 (2005). 
(2004). 
and K. No, Appl. Phys. Letts. 89, 162907 (2006). 
(2005) 
89 S. B. Ren, C. J. Lu, J. S. Liu, H. M. Shen, and Y. N. Wang, Phys. Rev. B. 54 
90 T. Zhao, A. Scholl, F. Zavaliche, K. Lee, M. Barry, A. Doran, M. P. Cruz, Y. 
Eom, and R. Ramesh, Nature Mat. 5, 823 (2006) 
92 P. Wurfel, I. P. Batra, and J. T. Jacobs, Phys. Rev. Lett. 30, 1218 (1973). 
78, 2751 (2001). 
95 R. Takahashi, Y. Yonezawa, M. Ohtani, M. Kawasaki, K
R.Takahashi, Y. Matsumoto, T. Kohno, M. Kawasaki, H. Koinuma, J. Cryst. 
97 R. Takahashi, Y. Yonezawa, M. Ohtani, M. Kawasaki, Y. Matsumoto, H. 
98 R. Takahashi, Y. Tsuruta, Y. Yonezawa, T. Ohsawa, H. Koinuma, and Y. 
99 Yu. E. Rogomskaya, Yu. Ya. Tomashpol?skii, Yu. N. Venevtsev, V. M. 
100 M. Mahesh Kumar, V. R. Palkar, K. Srinvas, S. V. Suryanarayana, Appl. Phys. 
101 M. Li, J. L. MacManus-Driscoll, Appl. Phys. Lett. 87, 252510 (2005). 
J. Appl. Phys. 97, 10C514 (2005); Ch. Binek and B. Doudin, J. Phys.: 
103 W. H. Meiklejohn, C. P. Bean, Phys. Rev. 102 1413 (1956). 
 120 
 
110 hstrat, P. Borisov, and W. Kleemann, Appl. Phys. Lett. 89, 
2 
3  Mu?oz, B. Dieny, M. D. Baro, 
4 
5 . 
7 478 (2008). 
6 
7 
9 
X. Chen, A. Hoc
202508 (2006). 
111 W. H. Meiklejohn and C. P. Bean, Phys. Rev. 102, 1413 (1956). 
11 M. Kiwi, J. of Magn. Magn. Mater. 234, 584 (2001). 
11 S. Br?ck, J. Sort, V. Baltz, S. Suri?ach, J. S.
and J. Nogu?s, Adv. Mater. 17 2978 (2005). 
11 S. Shtrikman and D. Treves, Phys. Rev. 130 986 (1963). 
11 Y. -H. Chu, L. W. Martin, M. B. Holcomb, M. Gajek, S. -J. Han, Q. He, N
Balke, C. -H. Yang, D. Lee, W. Hu, Q. Zhan, P. -L. Yang, A. Fraile-
Rodr?guez, A. Scholl, S. X. Wang & R. Ramesh, Nature. Mat., 
11 Marc De Graef and Yimei Zhu, J. of Appl. Phys. 89, 7177 (2001). 
11 D. Lebeugle et al.  Appl. Phys. Letts. 92 022907 (2007). 
H. Zheng et. al, Sc118 ience 303, 661 (2004) 
11 [Ch Binek and B. Doudin, J. Phys.: Condens. Matter 17 L39 (2005). 
 121