ArticlePDF Available

Electric in-plane polarization in multiferroic CoFe2O4/BaTiO3 nanocomposite tuned by magnetic fields

Authors:

Abstract and Figures

Ferrimagnetic CoFe2O4 nanopillars embedded in a ferroelectric BaTiO3 matrix are an example for a two-phase magnetoelectrically coupled system. They operate at room temperature and are free of any resource-critical rare-earth element, which makes them interesting for potential applications. Prior studies succeeded in showing strain-mediated coupling between the two subsystems. In particular, the electric properties can be tuned by magnetic fields and the magnetic properties by electric fields. Here we take the analysis of the coupling to a new level utilizing soft X-ray absorption spectroscopy and its associated linear dichroism. We demonstrate that an in-plane magnetic field breaks the tetragonal symmetry of the (1,3)-type CoFe2O4/BaTiO3 structures and discuss it in terms of off-diagonal magnetostrictive-piezoelectric coupling. This coupling creates staggered in-plane components of the electric polarization, which are stable even at magnetic remanence due to hysteretic behaviour of structural changes in the BaTiO3 matrix. The competing mechanisms of clamping and relaxation effects are discussed in detail.
Content may be subject to copyright.
ARTICLE
Received 15 Aug 2012 | Accepted 23 May 2013 | Published 25 Jun 2013
Electric in-plane polarization in multiferroic
CoF e
2
O
4
/BaTiO
3
nanocomposite tuned by
magnetic fields
Carolin Schmitz-Antoniak
1
, Detlef Schmitz
2
, Pavel Borisov
1,w
, Frank M. F. de Groot
3
, Sven Stienen
1
,
Anne Warland
1
, Bernhard Krumme
1
, Ralf Feyerherm
2
, Esther Dudzik
2
, Wolfgang Kleemann
1
& Heiko Wende
1
Ferrimagnetic CoFe
2
O
4
nanopillars embedded in a ferroelectric BaTiO
3
matrix are an example
for a two-phase magnetoelectrically coupled system. They operate at room temperature and
are free of any resource-critical rare-earth element, which makes them interesting for
potential applications. Prior studies succeeded in showing strain-mediated coupling between
the two subsystems. In particular, the electric properties can be tuned by magnetic fields and
the magnetic properties by electric fields. Here we take the analysis of the coupling to a new
level utilizing soft X-ray absorption spectroscopy and its associated linear dichroism. We
demonstrate that an in-plane magnetic field breaks the tetragonal symmetry of the (1,3)-type
CoFe
2
O
4
/BaTiO
3
structures and discuss it in terms of off-diagonal magnetostrictive-piezo-
electric coupling. This coupling creates staggered in-plane components of the electric
polarization, which are stable even at magnetic remanence due to hysteretic behaviour of
structural changes in the BaTiO
3
matrix. The competing mechanisms of clamping and
relaxation effects are discussed in detail.
DOI: 10.1038/ncomms3051
OPEN
1
Fakulta
¨
tfu
¨
r Physik and Center for Nanointegration Duisburg-Essen (CENIDE), Universita
¨
t Duisburg-Essen, Duisburg 47048, Germany.
2
Helmholtz-Zentrum
Berlin fu
¨
r Materialien und Energie, Albert-Einstein-Strasse 15, Berlin 12489, Germany.
3
Department of Inorganic Chemistry and Catalysis, Universiteit
Utrecht, Universiteitsweg 99, CG Utrecht 3584, The Netherlands. w Present address: Department of Physics, West Virginia University, 135 Willey Street,
Morgantown 26506 WV, USA. Correspondence and requests for materials should be addressed to C.S.-A. (email: carolin.antoniak@uni-due.de).
NATURE COMMUNICATIONS | 4:2051 | DOI: 10.1038/ncomms3051 | www.nature.com/naturecommunications 1
& 2013 Macmillan Publishers Limited. All rights reserved.
M
ultiferroic materials showing both magnetic and electric
ordering allow an additional degree of freedom in the
design of actuators, transducers and storage devices and
thus have attracted scientific interest from the technological
perspective as well as from basic research
1
. Because the choice of
single-phase multiferroic materials being suitable at room
temperature is limited, the use of magnetoelectric two-phase
composites has proven to be more promising
2
. Recently, even
classic materials such as ferromagnetic Fe and ferroelectric
BaTiO
3
(BTO) have been shown as potentially useful
3
. They
reveal electric field-controllable magnetoelectric coupling via
interfacial spin polarization of Ti
4 þ
. Research on core-shell
nanoparticles had shown that the spin polarization of Ti
4 þ
at the
interfaces between BTO and Fe and its oxides remains virtually
negligible
4
. Hence, two-phase composites of a piezoelectric and a
magnetostrictive phase, which can be magnetoelectrically coupled
via stress mediation
5
, still remain favourites of the current
discussion
6
. One of the prime candidates of such systems is the
BTO/CoFe
2
O
4
(CFO) system.
CFO is a ferrimagnet below T
c
¼ 790 K with an inverse spinel
structure that exhibits a remarkable magnetostriction of up to
l
||
E 2 10
4
along the direction of the applied magnetic field
(longitudinal magnetostriction) and l
>
E1 10
4
for transverse
magnetostriction at room temperature
7
. The strength of
magnetostriction strongly depends on the cation distribution in
CFO. The inverse spinel structure (space group Fd
3m; no. 227) is
a cubic crystal system, with oxide anions arranged in a cubic
close-packed lattice and metal cations tetragonally or octahedrally
surrounded by the oxygen. Each unit cell consists of 32 oxygen
ions, 16 octahedral sites and 8 tetrahedral sites. For CFO, Fe
3 þ
ions occupy the tetrahedral sites and half of the octahedral sites,
whereas Co
2 þ
ions are located at the remaining octahedral sites.
Three crystallographic phases of BTO are ferroelectric:
rhombohedral o190 K, orthorhombic for 190 KoTo278 K and
tetragonal for 278 KoTo395 K. At higher temperatures, BTO is
paraelectric. In this work, all measurements were performed at
room temperature, where BTO has a tetragonal unit cell (space
group P4mm; no. 99) with the lattice parameters a ¼ b ¼ 0.3992
nm and c ¼ 0.4036 nm. The occurence of the spontaneous electric
polarization is connected to a displacement of the Ti
4 þ
ion along
the crystallographic c axis and a sensitive charge hybridization of
the 3d states of the Ti cation with the 2p states of the surrounding
O anions. By density functional theory within the local spin
density approximation, it has been shown that the off-centre
displacement is connected to a substantial charge rearrangement
changing the occupation of the d
z
2
, d
xz
, and d
yz
orbitals of Ti to
slightly higher values at the expense of electron occupation of the
d
xy
orbital of Ti and p orbitals of the oxygen ions located on the z
axis yielding a destabilization of the centrosymmetric configura-
tion
8
. On a macroscopic scale, piezoelectric coefficients of
p
||z
E150 pm V
1
along the z axis and p
>z
E 80 pm V
1
along x and y axis were found
9
.
As previously shown
10
, structural compatibility of CFO with
BTO allows growth of different two-phase heterostructures when
using SrTiO
3
(STO) as a substrate. The resulting combination of
magnetostrictive CFO with piezoelectric BTO presents an
artificial multiferroic composite whose electric polarization P
can by tuned by a magnetic field and whose magnetic properties
can be altered by an electric field
11
, respectively, by strain
mediation. In the literature (for example, Zheng et al.
10
),
multilayered (2,2)-structures of CFO and BTO are reported, as
well as vertically aligned (1,3)-structures (Fig. 1a). The latter
configuration minimizes clamping effects to the substrate that
substantially reduce the possible structural deformations and is
also used in this work. To achieve complete regularity of the
nanopillar pattern, a recently experienced stencil-derived direct
epitaxy technique
12
might eventually be chosen for pertinent
applications. From application’s point of view, our quasi-two-
dimensional assembly of nanopillars of CFO embedded in BTO
as introduced by Zheng et al.
10
is a prototype system for
electrically addressable magnetic data-storage devices working at
room temperature. However, for all future applications, an
understanding of the coupling mechanism between the two
constituents is essential. Although in phase-field simulations,
electrostriction (strain pP
2
) in BTO has often exclusively been
taken into account
13
, piezoelectricity (strain pP) is rarely
discussed
14
.
In this work, hitherto neglected non-diagonal piezoelectricity
components are taken into account for the first time. They are
shown to be crucial for the observed changes in local charge
symmetry and analysed to gain a more detailed understanding of
the microscopic coupling effect between CFO and BTO. In this
regard, the soft X-ray absorption spectroscopy and its associated
circular and linear dichroism offer the unique possibility to study
both spin and charge anisotropy element specifically. Thus, by
applying magnetic fields, it is possible not only to measure the
magnetic response of the CFO but also to monitor changes in
the electronic structure of the BTO host matrix, especially the
changes around the Ti
4 þ
ions as done in this work.
Results
Phase-separated growth and pre-characterization. The sample
was prepared by pulsed laser deposition from a target consisting
of 65% BTO and 35% CFO. Details of the sample preparation can
be found in the Methods section. A spontaneous phase separation
between the BTO and CFO makes the growth of self-assembled
CFO pillars in a BTO matrix possible. Scanning electron micro-
scopy (Fig. 1b) images reveal a quasi-periodic assembly of CFO
pillars with rectangular cross-sections, the mean edge-length in
the sample plane being about 70 nm and the length around
400 nm. A scanning electron microscopy image over a large
sample area is shown in Supplementary Fig. S1. From atomic
force microscopy, we found that the length of the CFO pillars is
about 10% larger than the thickness of the BTO matrix. X-ray
diffraction (Fig. 1c) shows a clear phase separation of the con-
stituents and a (001) texture. For the crystallographic in-plane
axes, we have CFO[010]||BTO[010]||STO[010] that have been
chosen here as y axis and CFO[100]||BTO[100]||STO[100] that
have been chosen as x axis. Please note, that the BTO/CFO
interfaces at the sides of the nanopillars are {110}-type planes
similarly as verified for the related BFO( ¼ BiFeO
3
)/CFO sys-
tem
15
. To avoid the impression that the in-plane crystallographic
axes of the CFO nanopillars could be rotated with respect to the
ones of BTO, the unit cell orientation of CFO is depicted in the
schematic figures in this work.
The in-plane mosaicity is well o1° for both CFO and BTO as
discussed in the Supplementary Methods. The corresponding
X-ray diffraction peaks are shown in Supplementary Fig. S2, and
the results are summarized in Supplementary Table S1. This high
quality of epitaxial growth is crucial for the proper analysis of the
X-ray absorption spectroscopy measurements.
Magnetometry using a superconducting quantum interference
device (SQUID) clearly shows that the easy axis of magnetization
points along z as expected because of the shape anisotropy of the
nanopillars (Fig. 1d). The magnetization is normalized to
the CFO fraction in the sample, and its saturated value of
M
s
¼ (3.15
±
0.4) 10
5
Am
1
is in agreement with the one
reported in the literature
10
. The anisotropy field can be
estimated by linear extrapolation of the field-dependent
magnetization along the hard axis yielding about (2.5
±
0.5) T.
Calculating the shape anisotropy within the Stoner Wohlfarth
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3051
2 NATURE COMMUNICATIONS | 4:2051 | DOI: 10.1038/ncomms3051 | www.nature.com/naturecommunications
& 2013 Macmillan Publishers Limited. All rights reserved.
model for a slender prolate ellipsoid
16
with an aspect ratio of
about r ¼ 5.7, the shape anisotropy field of the nanopillars
amounts to m
0
M
s
(1 þ 1/r
2
[1 ln(2r)])/2E0.19 T in agreement to
the value reported by Zheng et al.
10
who explained the difference
between the estimated anisotropy from experimental
magnetometry data and the calculated shape anisotropy due to
demagnetizing fields by the occurence of the magnetoelastic
coupling.
Crystal field parameters of BTO in magnetic remanence. Soft
X-ray absorption spectra were taken at the L
3,2
absorption edges
of Ti and Co at which electrons are excited from the 2p
3/2
and
2p
1/2
core levels into the 3d states and at the M
5,4
absorption
edges (3d
5/2
and 3d
3/2
into the 4f states) of Ba, respectively.
An overview spectrum over the elements is presented in
Supplementary Fig. S3. For linearly polarized light, the electric
field vector of the X-rays acts like a search light for the unoc-
cupied density of states. As an example, let us consider the d
z
2
orbital: If the electric field vector is parallel to the z axis, the
absorption is maximum, whereas for the case of the electric field
vector lying in the xy plane or including any non-vanishing angle
with the z axis, the absorption will be lower. In Fig. 2, the relation
between the X-ray absorption at the Ti L
3
edge and the 3d orbitals
is depicted: the two absorption maxima occur due to electronic
transitions towards the lower-energetic d
xz
, d
yz
and d
xy
orbitals on
the one hand and the higher-energetic d
z
2
and d
x
2
y
2
on the other
hand. Because of hybridization effects, for example, orbitals
pointing towards the oxygen ligands yield broader absorption
lines in the absorption spectrum. The additional energy splitting
due to the Ti displacement is too small to be directly visible in the
absorption spectrum. However, it can be analysed by measuring
the dichroism between two different directions of linear polar-
ization. For a quantitative analysis of the X-ray linear dichroism
(XLD), the sample has been measured under grazing incidence
(y
k
¼ 60°), so the main component of the electric field vector of
the X-rays is either perpendicular to the z axis (vertical polar-
ization) or along the z axis (horizontal polarization) as sketched
in Supplementary Fig. S4. Because the angle of incident X-ray
cannot be y
k
¼ 90° for technical reasons, the spectral intensities
have been recalculated according to our measurement geometry
to obtain the full linear dichroism. As also the in-plane compo-
nent of the wave vector k of incident X-rays may be important
when analysing the XLD (ref. 17), k was kept in the yz plane for
all measurements presented here. The same holds for the external
magnetic field. Measurements in normal incidence (y
k
¼ 0) do
not require any correction. In that case, the electric field vector of
the vertically and horizontally polarized X-rays is along the x and
y direction, respectively. Details on both exact measurement
geometry and corrections, respectively, can be found in the
Methods section and Supplementary Fig. S4.
In Fig. 3, the experimental soft X-ray absorption near-edge
structure (XANES) and the associated XLD of Ti in magnetic
remanence after magnetizing along the z direction ( m
0
H ¼ 3T)
are compared with charge-transfer multiplet calculations using
the CTM4XAS program
18
. The following discussion of the
influence of the Ti
4 þ
ion displacement on the X-ray absorption
spectra is based on crystal field parameters. A short introduction
to crystal field theory and its connection to XAS can be found, for
example, in the work of de Groot
19
. We would like to mention
that the value of the crystal field parameter 10D
q
denotes the
energy splitting between the t
2g
and e
g
states in a cubic symmetry
as depicted in Fig. 2, whereas D
s
takes into account the deviation
CFO (004)
BTO (002)
STO (002)
BTO (003)
STO (003)
BTO (004)
CFO (008)
STO (004)
Magnetization (10
3
Am
–1
= emu c
m
–3
)
x ⎜⎜ [100]
y ⎜⎜ [010]
20
Diffraction intensity (arb. units)
1
10
10
2
10
3
10
4
–200
–400
200
400
0
–4
Magnetic field (T )
CFO
ac d
b
BTO
STO
x
y
H
= 0
H
= 90°
z ⎜⎜ [001]
30 40 50 60 70
Momentum transfer (nm
–1
)
–2 0 2 4
Figure 1 | Sample pre-characterization. (a) Illustration of nanopillar structure and introduction of sample coordinate system. Scale bar, 200 nm.
(b) Scanning electron microscopy image of the sample corresponding to a top view on the nanopillar structure. (c) X-ray diffraction evidencing the strongly
texturized growth and phase separation between BTO and CFO. The red line corresponds to the measurement of the sample in total, and the black line
corresponds to the substrate only. (d) SQUID magnetometry at room temperature reveals the easy direction of magnetization along the sample normal.
461
462
460
459
458
0
4
8
Photon energy (eV)
e
g
t
2g
6D
q
2D
s
6D
q
+ 2D
s
–4D
q
D
s
–4D
q
+ 2D
s
d
x
2
y
2
d
z
2
d
xy
d
xz
d
yz
Absorption (arb.units)
Figure 2 | Relating XAS and atomic orbitals. The soft X-ray absorption
spectrum at the L
3
edge of Ti (left) is related to the energy levels of the Ti
ion in the crystal field for positive values of 10D
q
and negative values
of D
s
.(D
t
is neglected.) In the tetragonal phase of BTO, the energy levels
of the cubic e
g
and t
2g
states further split into sublevels. The energy
scale for this splitting is stretched.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3051 ARTICLE
NATURE COMMUNICATIONS | 4:2051 | DOI: 10.1038/ncomms3051 | www.nature.com/naturecommunications 3
& 2013 Macmillan Publishers Limited. All rights reserved.
from a cubic symmetry including the displacement of the
Ti
4 þ
ion. The best fit to experimental data was obtained for
the crystal field parameters 10D
q
¼ 2.0 eV, in agreement with de
Groot et al.
20
and Chasse
´
et al.
21
, and D
s
¼0.04 eV. Note that
the XANES is not very sensitive to changes of D
s
, whereas the
XLD strongly depends on this value and vanishes for D
s
¼ 0. The
negative value of D
s
ensures the correct energy splitting of t
2g
and
e
g
states into sublevels in accordance with band structure
calculations reported in the literature for tetragonal BTO
8
.In
that reference, Filippetti and Hill calculated the density of states
for Ti
4 þ
in BTO for a cubic and tetragonal unit cell by a plane-
wave pseudopotential implementation of density functional
theory within the local spin density approximation (LSDA)
22
,
and assigned it to atomic orbitals. In the tetragonal phase, the
displacement of the Ti
4 þ
from its centrosymmetric position
along the z direction yields a splitting between the d
xy
orbital and
the orbitals with a significant z component, that is, d
xz
and d
yz
,
which are at higher energies. From the view of crystal field
parameters, this is equivalent to the introduction of a negative
value of D
s
as depicted in Fig. 2. Note that from the parameter D
s
,
the geometric structure cannot be deduced directly
23
. Here the
negative value of D
s
is associated with an elongation of the BTO
unit cell along the z direction. This is in agreement to XRD data
revealing that the the elongated axis of the tetragonal BTO is
parallel to z (cf. Fig. 1c). Using these values for the crystal field,
not only the spectral shape of XANES and XLD are reasonably
reproduced by our simulations but also the XLD amplitude fits
well. Thus, one can conclude that the electric polarization in the
sample points along z as it was assumed in the simulation.
Electric in-plane polarization in magnetic remanence.
Measuring the soft X-ray absorption in magnetic remanence after
magnetizing along z (y
H
¼ 0, m
0
H ¼ 3 T) in normal incidence,
that is, with the electric field vector of incident soft X-rays per-
pendicular to z for both horizontal and vertical polarization, does
not give rise to a dichroic signal as can be seen in Fig. 4a, left
panel. This is because of the fact that the d
z
2
, d
x
2
y
2
and d
xy
, and
the sum of the energetically degenerated d
xz
and d
yz
, have the
same 90° symmetry in the xy plane as the electric field vectors for
horizontally and vertically polarized X-rays probing these orbi-
tals. Let us consider the 90° symmetry of the BTO orbitals
observed after magnetizing along the z direction in terms of the
ferroelectric polarization. In a simple picture, the magnetic field
gives rise to a contraction of the CFO nanopillars (l
||
o0).
Because of Poisson’s law, the pillars have the tendency to expand
in the xy plane (l
>
40). Close to the substrate, this expansion is
hindered by clamping effects through the substrate, whereas near
the surface, they have the possibility to expand in both x and y
directions by compressing the BTO matrix. By virtue of the non-
400
500
c = 0.403 nm
H
= 0
H
= 90°
XANES and XLD intensity
(arb. units)
–0.02
0.02
0.04
0.06
0.08
ad
bc
0
456 460 464 468
XLDx20
XLDx20
30.8 31.2 31.6
c = 0.404 nm
100
200
300
STO substrate
+P
x
+P
y
P
x
P
y
STO substrate
+P
x
P
y
P
x
P
x
+P
y
H
= 90°,
k
= 0
H
= 0,
k
= 0
456 460 464
Difference x 10
–100
0
BTO (002)
Photon energy (eV) Photon energy (eV)
XRD intensity (arb. units)
Momentum transfer (nm
–1
)
(
H
= 90°)
H ⎜⎜y
H ⎜⎜z
(
H
= 0)
Figure 4 | Experimental soft X-ray absorption and X-ray diffraction in magnetic remanence. (a) Soft XANES and XLD in normal incidence (y
k
¼ 0) in
magnetic remanence after magnetically saturating the sample along the z directio n (y
H
¼ 0) and y direction (y
H
¼ 90°), respectively. (b,c) Illustration of different
experimental geometries and (defor med) shapes of CFO nanopillars’ unit cells with apical polarization vectors of the BT O environment
±
P
x,y
.(d)X-raydiffraction
at the BT O (002) peak in magnetic remanence after magnetically satura ting the sample along the z direction (black line, y
H
¼ 0) and y direction (green line,
y
H
¼ 90°), respectivel y . Symbols refer to the differ ence of the two diffraction signals (enlarg ed by a factor of 10), and the red line refers to a fit using two Gaussian
distributions.
456
Photon energy (eV)
–0.02
0.02
0
0.04
0.06
0.08
XANES and XLD intensity (arb. units)
H ⎜⎜z (
H
= 0),
k
= 60°
XANES
XLD x 4
460 464 468
Figure 3 | Simulation and experimental data. Soft XANES and its
associated XLD from experiment (symbols) and simulation (lines).
Measurements were performed in magnetic remanence after magnetically
saturating the sample along the z direction (y
H
¼ 0) under grazing
incidence (y
k
¼ 60°) and were corrected for the angular misalignment to
obtain full dichroism.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3051
4 NATURE COMMUNICATIONS | 4:2051 | DOI: 10.1038/ncomms3051 | www.nature.com/naturecommunications
& 2013 Macmillan Publishers Limited. All rights reserved.
vanishing piezoelectric shear coefficients
24
, this gives rise to the
evolution of in-plane components P
x
and P
y
of the electric
polarization, as schematically shown in Fig. 4b. In this geometry,
x and y directions are equivalent in agreement to the 90° in-plane
symmetry obtained in the experiment (|P
x
| ¼ |P
y
|). Formally, this
can be written as
P
x
¼d
15
s
13
; P
y
¼d
15
s
23
ð1Þ
where d
15
is the decisive and quite sizable off-diagonal
piezoelectric coefficient
24
, and s
13
and s
23
are components of
the stress tensor causing the shear in xz and yz planes,
respectively. It should be emphasized, again, that the non-zero
shear stress components s
13
and s
23
are most important
consequences of the nanopillar clamping at the STO substrate
(Fig. 4b,c). The quadratic deformation of the CFO unit cell cross-
sections is by symmetry characterized by |s
13
| ¼ |s
23
|.
In contrast, application of the external magnetic field in the
sample plane, for example, along the y direction (y
H
¼ 90°,
m
0
H ¼ 3 T), breaks this symmetry because the nanopillars
contract in y direction and expand in x (and z) direction yielding
a rectangular deformation of the unit cell cross-sections of the
nanopillars with |s
13
|a|s
23
|. This leads to a compression of the
BTO along the x direction only and to directly opposed in-plane
polarization components as sketched in Fig. 4c and in more detail
in Supplementary Fig. S5. As a consequence, a measurable XLD
signal evolves under normal incidence of the X-rays (Fig. 4a, right
panel), reflecting the spatial in-plane anisotropy of the electric
polarization (|P
x
|a|P
y
|). It can be switched on and off by
application of an in-plane external magnetic field or a magnetic
field perpendicular to the sample plane, respectively, and is stable
in magnetic remanence. By comparison with the XLD presented
in Fig. 3, its amplitude clearly exceeds the value of 0.01 P
z
estimated in the Supplementary Methods. This may be due to a
reduced value of P
z
because we did not apply an electric field to
obtain the XLD presented in Fig. 3. However, the large in-plane
XLD signal indicates an induced charge anisotropy on a
macroscopic length scale that is not restricted to the CFO/BTO
interface. Comparing the sign of the XLD signals shown in Figs 3
and 4a, respectively, one can conclude that the more elongated
in-plane axis of BTO after magnetizing along the y direction is
indeed the y axis supporting the model presented above. From
our measurements, one can conclude that an in-plane component
of the electric polarization of the BTO can be tuned by
application of an external magnetic field. This change of the
electric polarization towards x and y directions
24
is quite
significant for BTO. It can be modelled within the Ginzburg
Landau theory
25
and has been assigned to the vicinity of the
phase transition from the tetragonal phase at room temperature
to the orthorhombic phase of BTO at about 278 K (ref. 26).
To confirm that the XLD is caused by an in-plane component
of the electric polarization due to different remanent mechanical
stresses on the BTO matrix, we performed X-ray diffraction
(XRD) measurements under the same conditions as for the
previous XLD measurements. That is, XRD scans were taken in
magnetic remanence after application of a magnetic field
(m
0
H ¼ 3 T) either along the z direction or along the y direction.
We focused on the (00L) reflections, that is (002) for BTO and
STO substrate and (004) for CFO (cf. Supplementary Fig. S2). The
diffraction peak for BTO is presented in Fig. 4d. After
magnetizing along the z direction, the BTO (002) peak is
asymmetric that is related to the gradient of the magnetostriction
of CFO due to the clamping to the substrate as depicted in Fig. 4b.
The position of the peak maximum corresponds to a lattice
constant of cE0.404 nm in agreement to the expected bulk value
for the long axis of the BTO unit cell. Applying the magnetic field
in the sample plane yields the evolution of a second peak at
higher momenta corresponding to a reduced lattice constant of
about 0.402 nm (not shown here). After removing the magnetic
field, the second peak is smaller, but still visible, and corresponds
to an intermediate reduced lattice constant c
0
E0.403 nm
(Fig. 4d). This behaviour can be understood as the response of
the matrix to different in-plane stresses by the CFO pillars. When
the external magnetic field was applied perpendicularly to the
sample plane, we have a compression of the BTO in both x and y
directions as explained above, reducing the lattice constants a
and b. The lattice constant c will be an equilibrium value
determined by two competing mechanisms: on the one hand, the
unit cell volume should be conserved, which favours an
0
0.05
0.10
0.15
–0.10
–0.05
XMCD asymmetry
Co L
3,2
1.0 2.0 3.00–1.0–2.0–3.0
0
0.05
0.10
0.15
–0.10
–0.05
–0.15
XMCD asymmetry
Co L
3,2
XLD asymmetry
0.04
0
0.06
a b
0.02
Ti L
3,2
XLD asymmetry
0.04
0
0.06
1.0 2.0 3.0
0
0.02
Ti L
3,2
H
= –30°,
k
= 60°
H
= –30°,
k
= 60°
H
= 60°,
k
= 60°
H
= 60°,
k
= 60°
H
main
||y
H
main
|| z
H
main
|| y
External magnetic field (T )External magnetic field (T )
H
main
|| z
Figure 5 | Changing the Ti XLD by external magnetic fields. (a) Field-dependent soft XLD at the Ti L
3,2
absorption edges measured under grazing
incidence (y
k
¼ 60°) for different directions of the magnetic field with the main component H
main
either parallel to z or perpendicular to z. Error bars are
estimated by comparison of 2–5 pairs of spectra to extract the XLD asymmetry and slight energy variation. (b) Field-dependent soft X-ray magnetic circular
dichroism at the Co L
3
absorption edge as a measure of magnetization for the same geometries as in a. Grey dotted lines correspond to raw data, and green
solid lines correspond to the XMCD signal after elimination of field-dependent rotation of the magnetization, assuming an anisotropy field of 2.5 T.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3051 ARTICLE
NATURE COMMUNICATIONS | 4:2051 | DOI: 10.1038/ncomms3051 | www.nature.com/naturecommunications 5
& 2013 Macmillan Publishers Limited. All rights reserved.
expansion of the c axis, and on the other hand, the BTO is
coupled to the CFO pillars, which shows a contraction along that
direction. This competition can also yield the asymmetric line
shape of the BTO diffraction peak obtained here because it is
reasonable that close to the BTO/CFO interface, the coupling
between the constituents is dominant, whereas it is weaker at
larger distances from the interface. When the sample is
magnetized in-plane along the y direction, the BTO matrix is
allowed to expand in this direction. To conserve the unit cell
volume, BTO should exhibit a reduced value of the c axis.
However, because CFO shows a lattice expansion in this case, we
end up with two competing contributions again. But this time, the
positive magnetostriction of CFO along the z axis for H||y is only
half the absolute value of the magnetostriction for H||z because
the longitudinal magnetostriction is up to l
||
E 2 10
4
,
whereas the transverse magnet ostri ction is l
>
E1 10
4
(ref. 7).
Therefore, the structu ral relaxation of the BTO when distancing the
BTO/CFO interface may appear sooner. In addition, especially at
thesurfaceofthesamplewhereclampingisnegligible,thein-plane
contraction of the nanopillars along y and, consequently, the
expansion of the BTO along y may be significant. That means, parts
ofthesamplehavedifferentin-planelatticeconstants,aob,anda
reduced c lattice constant compared with the bulk value as obtained
in the X-ray diffraction data. As already mentioned above, the
phase-transition temperature from the tetragonal phase to the
orthorhombic phase of BTO (at 278 K) is quite close, facilitating the
occurence of in-plane components of the electric polarization as well
as structural deformations. Similarly to the broad thermal hysteresis
of the BTO structure (about 10 K (ref. 26)), a field-dependent
hysteretic behaviour of the structure as evidenced here appears
reasonable. The obtained distortion of the former tetragonal unit cell
in combination with the shear forces depicte d in Fig. 4c makes it
possible to tune an in-plane component of the electric polarization
of the BTO host matrix.
Magnetic field dependence of induced electric polarization.To
prove that there is a direct coupling between CFO and BTO, the
XLD signals at the Ti ions have been measured as a function of
external magnetic field in two different geometries: with the main
component of the magnetic field either along the z axis or per-
pendicular, that is, along the y axis (Supplementary Fig. S4). The
resulting Ti XLD asymmetries, that is, the XLD divided by the
sum of absorption spectra measured with vertical and horizontal
polarization of X-rays (I
ver
I
hor
)/(I
ver
þ I
hor
), are plotted in
Fig. 5a. For a reasonable analysis of the small dichroic signals,
only amplitudes of the XLD asymmetry averaged over both
absorption edges are presented, as shown in Supplementary Fig.
S6 and discussed in the Methods section.
We start the discussion with the common geometry of the
magnetic field’s main component applied parallel to the long axis
of the CFO nanopillars, that is, the z axis. The XLD of the Ti ions
is increasing with increasing field by a factor of 3 to about 6.5% at
an external magnetic field of 1.5 T. For higher external magnetic
fields, a slight decrease is obtained, and the XLD is asymptotically
approaching a value of about 5% for higher fields. This decrease is
due to the fact that in our measurement geometry, the magnetic
field could not be applied exactly parallel or perpendicular to the
CFO nanopillars. As at about 1.5 T the CFO saturates, the effect
of pulling the magnetization from its easy direction towards the
direction of the external magnetic field significantly influences the
field dependence of the XLD. In Fig. 5b, the raw data of the field-
dependent magnetization are shown that exhibit an additional
linear slope of the signal (grey dotted lines). This slope can be
easily eliminated (green solid lines) by correcting the different
angles between magnetization and external magnetic field,
assuming an effective anisotropy field of 2.5 T in agreement to
the anisotropy estimated by extrapolating SQUID magnetometry
data, as discussed in the Supplementary Methods and sketched in
Supplementary Fig. S7. The comparison of the Ti XLD and the
Co X-ray magnetic circular dichroism (XMCD) as a measure of
the magnetization clearly evidences the correlation between these
two values. As the XLD is a measure of the anisotropy of
electronic states, it indicates on a microscopic scale the
modification of the electric polarization of BTO by a magnetic
field. When the external magnetic field is applied with its main
component in-plane, the change of the XLD is quite small and
increases from 2.2 to 3.5% with increasing field. This corresponds
to our findings presented above: when the magnetic field is
applied in-plane, that is, along the y direction, the CFO contracts
in this direction, the surrounding BTO is allowed to expand while
the c lattice parameter in z direction is slightly smaller, reducing
the tetragonal distortion and the electric polarization along z
consequently. This corresponds to a reduced XLD signal. In x
direction, the BTO is compressed as for the case of a magnetic
field applied along the z direction, yielding an enhanced XLD. On
average, a slightly enhanced XLD is obtained. We conclude that
the observed magnetic field dependences within our model clearly
prove the direct coupling between the CFO nanopillars and the
BTO matrix.
Discussion
The possibility of tuning the electric polarization of the BTO
matrix hosting CFO nanopillars by an external magnetic field has
been evidenced on a microsopic scale by correlating the charge
anisotropies of the Ti ions (measured by XLD) with the spin
anisotropies of Co ions (measured by XMCD). Although the
electric polarization of the BTO matrix spontaneously points
along the z direction in the as-grown state, it achieves shear
stress-induced transverse components
±
P
x
and
±
P
y
, both under
out-of-plane and in-plane magnetic fields. Field-induced quad-
ratic and rectangular deformations, respectively, of the CFO unit
cell cross-sections give rise to characteristic staggered modula-
tions of the in-plane component of the electric polarization in
BTO, reflecting the quasi-periodic distribution of the nanopillars
and their deformation patterns. This has been verified by the
spatial anisotropy of the averaged Ti XLD. The in-plane
components of the electric polarization are still observed in
magnetic remanence and are accompanied by a hysteretic
behaviour in structural changes of BTO.
The observed coupling yields a sizeable effect on the averaged
electronic structure of BTO, much too large to be assigned to an
interfacial effect like a charge transfer between BTO and CFO and
a spin coupling between the polarization of Ti
4 þ
in BTO and Fe
ions in CFO as in the work of Valencia et al.
3
Thus, the effect in
the electric in-plane polarization of BTO obtained in this work
extends over a large area. Under the constraint of completely
regular arrays of CFO nanopillars
12
, one might even envisage
data-storage concepts by encoding the local polarization patterns
of single nanopillars (Fig. 4c) via external current-controlled local
magnetic fields H
x
or H
y
.
Methods
Sample preparation. The sample was prepared by pulsed laser deposition from a
commercially available target (Kurt J. Lesker company) containing 65 mol% BTO
and 35 mol% CFO. Ultraviolet radiation was provided by a 248 nm KrF excimer
laser (Lambda Physik) with a laser fluence of about 1 J cm
2
and a pulse repetition
rate of 10 Hz. The 400 nm thick film was grown on STO(001)-oriented single-
crystal substrate (CrysTec GmbH), which was kept at 950 °C during the growth in
an oxygen atmosphere at 15 Pa ( ¼ 0.15 mbar).
A scanning electron microscopy image of a large area of the sample is shown in
Supplementary Fig. S1, revealing that the CFO nanopillars are quite homo-
geneously spread over the whole sample. The topography of the sample was
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3051
6 NATURE COMMUNICATIONS | 4:2051 | DOI: 10.1038/ncomms3051 | www.nature.com/naturecommunications
& 2013 Macmillan Publishers Limited. All rights reserved.
investigated by atomic force microscopy and was found to sho w maximum
variations in height of 10%, that is, about 40 nm.
Sample pre-characterization by magnetometry and diffraction methods
.
Magnetometry measurements were performed at 300 K using a commercial SQUID
MPMS from Quantum Design providing magnetic fields up to
±
5 T. The tex-
turized growth was confirmed in conventional y–2y geometry on a Philips PW
1730 diffractometer using Cu K
a
radiation at the University of Duisburg-Essen. For
a more detailed analysis of the in-plane crystal order, X-ray diffraction has been
carried out at the Helmholtz Center Berlin (HZB) on a conventional laboratory
four-circle X-ray diffractometer using Mo K
a
radiation.
Soft X-ray absorption spectroscopy
. For soft X-ray absorption experiments, the
high-field endstation at beamline UE46-PGM1 of the HZB has been used at the
HZB-BESSYII synchrotron radiation source. This endstation offers an in-plane
rotable field up to
±
6 T, and the beamline is well suited for the soft X-ray range
with variable polarization, for example, linearly polarized light with the electric field
vector lying either horizontal or vertical, as well as circularly polarized light. The
measurement geometries are schematically shown in Supplementary Figs S4 and S7.
All measurements presented here were performed in total electron yield (TEY)
mode by measuring the sample drain current at room temperature where BTO
exhibits its tetragonal phase. Before the measurements, the time-dependent stability
of the TEY has been monitored to exclude the occurrence of charging effects.
Elemental analysis
. An additional chemical overview of the elements expected in
the BTO/CFO sample, that is, Ba, Ti, O, Co and Fe, is given by the soft X-ray
absorption spectra shown in Supplementary Fig. S3. As the spectra were obtained
in the TEY mode by measuring the sample drain current, which is very surface
sensitive due to the small electron escape depth of a few nanometres, no signal
from the STO substrate can be detected through the 400-nm thick BTO/CFO
composite film. This ensures a reliable analysis also for the case of Ti, which is
present in both STO substrate and BTO matrix. In Supplementary Fig. S3, the soft
XANES and its XLD are plotted for the elements. Here we present the measured
TEY normalized to the incoming photon intensity without any background
removal or any further treatment. The spectra were measured under grazing X-ray
incidence (y
k
¼ 60°) without a magnetic field applied, and the XLD reflects the
charge anisotropy. For BTO, a clear XLD signal could be obtained at the Ti L
3,2
and
Ba M
5,4
absorption edges, respectively. At the oxygen K edge, only a small
dichroism could be obtained after measuring a smaller energy range with higher
resolution. However, because oxygen is present in both BTO and CFO, an analysis
of the XLD seems ambiguous. From the CFO nanopillars, a clear XLD signal is
visible at both Fe and Co L
3,2
absorption edges.
Soft XLD
. Soft X-ray absorption spectra have been measured under normal or
grazing incidence as depicted in Supplementary Fig. S3. In the former case, the
electric field vector of the soft X-rays propagating along z is parallel to the x axis,
and for horizontal polarization, it is parallel to the y axis. The linear dichroism
obtained in this geometry reflects the charge anisotropy between x and y direction
in the sample. To obtain the spectra in magnetic remanence shown in Fig. 5 of the
main text, the magnetic field was applied either along z before the measurement or
the sample was rotated by 90° around the x axis, the magnetic field was applied in y
direction and the sample was rotated back to the former position after removal of
the magnetic field. For grazing incidence, some corrections have to be done to
overcome the angular misalignment. As can be seen in Supplementary Fig. S4b, the
electric field vector for vertical polarization is again parallel to the x axis. For
horizontal polarization, the electric field vector includes an angle y 90° ¼ 30°
with the z axis. Thus, the spectrum contains not only the pure spectrum of the
electronic structure parallel to z but also a contribution of the electronic structure
parallel to y. For the sake of simplification, the electric field vector components
yielding spectra of the electronic structure along x and y are denoted E
>z
in
contrast to E
||z
. The related absorption intensities that are proportional to the
electric field vector squared are denoted I
>z
and I
||z
, respectively. Hence, the
measured dichroism can be deduced as follows:
E
ver
E
hor
¼ E
?z
E
kz
sin y E
?z
cos y
I
ver
I
hor
¼ I
?z
I
kz
sin
2
y I
?z
cos
2
y
) I
ver
I
hor
¼ I
?z
3
4
I
kz
1
4
I
?z
¼
3
4
ðI
?z
I
kz
ð2Þ
That means, the dichroism obtained by simply subtracting the spectrum
measured with horizontal polarization from the one measured with vertical
polarization is smaller by a factor of 3
=
4, with respect to the real linear dichroism
between the two directions along z and perpendicular to z, respectively. As a
consequence, the linear dichroism measured under grazing incidence has been
multiplied by 4=3. Note that for the case of an distorted BTO unit cell with the
electric polarization not pointing along the z direction, the correction factor is
larger. In our work, in some cases an in-plane component of the BTO electric
polarization evolves as discussed in the main text. As the in-plane component
cannot be quantified, we restrict to the correction presented above. In the unlikely
case of a pure in-plane electric polarization of the BTO (with a fourfold symmetry),
this procedure would underestimate the XLD by a factor of 2.
The XLD asymmetries for Ti, Co and Ba are shown in Supplementary Fig. S6
for different magnetic fields. The XLD asymmetry refers to the difference between
the absorption measured with either horizontal or vertical polarization of X-rays
divided by the sum of the absorption measured with the two different polarizations,
(I
ver
I
hor
)/(I
ver
þ I
hor
). This value is useful because it does not depend on the
number of unoccupied final states but reflects changes in the polarization per
unoccupied d state.
It can clearly be seen that at the Co and Ba sites, only small magnetic field-
induced changes occur, whereas at the Ti sites, the changes are significant. For a
quantification of the XLD from measured spectra, the XLD amplitude has been
analysed at various energies to minimize the error caused by noise. These XLD
amplitudes used here are depicted in Supplementary Fig. S6 for the case of Ti. To
obtain the averaged XLD, not only the occupation numbers for the initial states are
taken into account, but also for the different t
2g
and e
g
-like final d states. The Ti
4 þ
ion is characterized by totally empty d states; only a small fraction of the d states is
occupied via hybridization with the electronic states of the surrounding oxygen
atoms. Neglecting this hybridization effect yields fractions of 3/5 t
2g
and 2/5 e
g
states resulting in IðTiÞ¼ð2I
t
2g
L
3
þ 3I
e
g
L
3
þ 4I
t
2g
L
2
þ 6I
e
g
L
2
Þ=15, and for the XLD
asymmetry, I(Ti) ¼ (a þ b)/3 þ (2c þ 2d)/3 with a–d being defined as sketched in
Supplementary Fig. S6 (left).
X-ray diffraction using synchrotron radiation
. X-ray diffraction measurements
beyond the structural pre-characterization using a conventional laboratory X-ray
source have been performed at the MAGS endstation of the 7T-MPW beamline at
the HZB-BESSYII synchrotron radiation source. This endstation is composed of a
Huber 5021 six-circle diffractometer and offers the possibility to apply magnetic
fields up to
±
5 T perpendicular to the scattering plane.
Simulations
. Simulations were performed using the CTM4XAS
charge-transfer multiplet software. This semi-empirical program is based on a
Hartree–Fock method corrected for correlation effects to solve the atomic
Hamiltonian
27
. It includes the core and valence spin-orbit coupling, the
core-valence two-electron integrals (multiplet effects) and the effects of strong
correlations within the charge-transfer model. In the case of metal ions in
oxidic environment, this approach is known to give reliable results. More details
about the program and some examples of applications can be found, for
example, in Stavitski and de Groot
18
and de Groot et al.
28
To simulate the
experimental data, the Slater integrals have been reduced to 0.81, and an individual
Lorentzian peak broadening for the four peaks has been included similar to
other reported values
20
. The peaks have been broadened by 0.17 eV; additional
broadenings have been included for the peaks number 2–4: 0.62 eV for the
second peak, 0.65 eV for the third peak and 0.80 eV for the fourth peak.
Instrumental broadening was taken into account by an additional Gaussian
broadening of 0.05 eV.
For the case of XMCD measured at the Co L
3,2
absorption edges, the
simulations can also be used to analyse possible cation disorder in CFO. However,
comparison of our experimental data with simulated spectra of Co
2 þ
and Co
3 þ
as
presented in Supplementary Fig. S8 clearly indicates that the majority of Co ions is
in the 2 þ state as expected for CFO.
References
1. Fiebig, M. Revival of the magnetoelectric effect. J. Phys. D: Appl. Phys. 38,
R123–R152 (2005).
2. Wang, Y., Hu, J., Lin, Y. & Nan, C. -W. Multiferroic magnetoelectric composite
nanostructures. NPG Asia Mater. 2, 61–68 (2010).
3. Valencia, S. et al. Interface-induced room-temperature multiferroicity in
BaTiO
3
. Nat. Mater. 10, 753–758 (2011).
4. Kim, D. H. et al. Interface electronic structures of BaTiO
3
@X nanoparticles
(X ¼ g-Fe
2
O
3
,Fe
3
O
4
, a-Fe
2
O
3
, and Fe) investigated by XAS and XMCD, Phys.
Rev. B 79, 033402 (2009).
5. van Suchtelen, J. Product properties: a new application of composite materials.
Philips Res. Rep. 27, 28–37 (1972).
6. Liu, G., Nan, C. -W., Xu, Z. K. & Chen, H. Coupling interaction in multiferroic
BaTiO
3
–CoFe
2
O
4
nanostructures. J. Phys. D: Appl. Phys. 38, 2321–2326 (2005).
7. Zheng, Y. X. et al. Study of uniaxial magnetism and enhanced magnetostriction
in magnetic-annealed polycrystalline CoFe
2
O
4
. J. Appl. Phys. 110, 043908
(2011).
8. Filippetti, A. & Hill, N. A. Coexistence of magnetism and ferroelectricity in
perovkites. Phys. Rev. B 65, 195120 (2002).
9. Tazaki, R., Fu, D., Itoh, M., Daimon, M. & Koshihara, S. Lattice distortion
under an electric field in BaTiO
3
piezoelectric single crystal. J. Phys.: Condens.
Matter 21, 215903 (2009).
10. Zheng, H. et al. Multiferr oic BaTiO
3
-CoFe
2
O
4
nanostructures. Science 303,
661–663 (2004).
11. Zavaliche, F. et al. Electrically assisted magnetic recording in multiferroic
nanostructures. Nano Lett. 7, 1586–1590 (2007).
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3051 ARTICLE
NATURE COMMUNICATIONS | 4:2051 | DOI: 10.1038/ncomms3051 | www.nature.com/naturecommunications 7
& 2013 Macmillan Publishers Limited. All rights reserved.
12. Lu, X. et al. Magnetoelectric coupling in ordered arrays of multilayered
heteroepitaxial BaTiO
3
/CoFe
2
O
4
nanodots. Nano Lett. 11,
3202–3206 (2011).
13. Zhang, J. X. et al. Phase-field model for epitaxial ferroelectric and magnetic
nanocomposite thin films. Appl. Phys. Lett. 90, 052909 (2007).
14. Nan, C. -W., Liu, G., Lin, Y. & Chen, H. Magnetic-field-induced electric
polarization in multiferroic nanostructures. Phys. Rev. Lett. 94, 197203 (2005).
15. Zheng, H. et al. Controlling self-assembled perovskite-spinel nanostructures.
Nano Lett. 6, 1401–1407 (2006).
16. Osborn, J. A. Demagnetizing factors of the general ellipsoid. Phys. Rev. 67,
351–357 (1945).
17. van der Laan, G., Arenholz, E., Chopdekar, R. V. & Suzuki, Y. Influence of
crystal field on anisotropic X-ray magnetic linear dichroism at the Co
2 þ
L
2,3
edges. Phys. Rev. B 77, 064407 (2008).
18. Stavitski, E. & de Groot, F. M. F. The CTM4XAS program for EELS and XAS
spectral shape analysis of transition metal L edges. Micron. 41, 687–694 (2010).
19. de Groot, F. M. F. Multiplet effects in x-ray spectroscopy, Coord. Chem. Rev.
249, 31–63 (2005).
20. de Groot, F. M. F., Fuggle, J. C., Thole, B. T. & Sawatzky, G. A. L
2,3
absorption
edges of d
0
compounds: K
þ
,Ca
2 þ
,Sc
3 þ
, and Ti
4 þ
in O
h
(octahedral)
symmetry. Phys. Rev. B 41, 928–937 (1990).
21. Chasse
´
,A.et al. High-resolution x-ray absorption spectroscopy of BaTiO
3
:
Experiment and first-principles calculations. Phys. Rev. B 84, 195135, 2011).
22. Perdew, J. P. & Zunger, A. Self-interaction correction to density-functional
approximations for many-electron systems. Phys. Rev. B 23, 5048–5079 (1981).
23. Donini, J. C., Hollebone, B. R. & Lever, A. B. P. Existence of
3
E
g
Ground terms
in tetragonal d
8
complexes and the possibility of high-spin square-planar
nickel(II). J. Am. Chem. Soc. 93, 6455–6462 (1971).
24. Budimir, M., Damjanovic, D. & Setter, N. Piezoelectric anisotropy-phase transition
relations in perovskite single crystals. J. Appl. Phys. 94, 6753–6761 (2003).
25. Damjanovic, D. Contributions to the piezoelectric effect in ferroelectric single
crystals and ceramics. J. Am. Ceram. Soc. 88, 2663–2676 (2005).
26. von Hippel, A. Ferroelectricity, domain structure, and phase transitions of
barium titanate. Rev. Mod. Phys. 22, 221–237 (1950).
27. Cowan, R. D. The Theory of Atomic Structure and Spectra (University of
California Press, Berkeley, 1981).
28. de Groot, F. M. F., Fuggle, J. C., Thole, B. T. & Sawatzky, G. A. 2p x-ray
absorption of 3d transition-metal compounds: an atomic multiplet description
including the crystal field. Phys. Rev. B 42, 5459–5468 (1990).
Acknowledgements
We would like to thank the HZB-BESSY II staff for their kind support during beamtimes. We
gratefully acknowledge U. v. Ho
¨
rsten from the University of Duisburg-Essen for help in the
structural pre-chara cteriz ation . This work was funded by the BMBF (05 ES3XBA/5) and the
DFG (SFB491).
Author contributions
C.S.-A., H.W., W.K., P.B., S.S. and B.K. proposed beamtime. C.S.-A., D.S., S.S., A.W. and
B.K. performed the experiment on X-ray linear dichroism, and R.F., E.D., C.S.-A. and
D.S. on X-ray diffraction at the synchrotron radiation facility. C.S.-A. and D.S. analysed
the data. D.S., C.S.-A. and F.M.F d.G. performed the simulations. S.S. and P.B. prepared
the sample, P.B. performed SQUID magnetometry, and S.S. performed the scanning
electron microscopy imaging. C.S.-A. and W.K. wrote the paper. All authors discussed
the results and commented on the manuscript.
Additional information
Supplementary Information accompanies this paper at http://www.nature.com/
naturecommunications
Competing financial interests: The authors declare no competing financial interests.
Reprints and permission information is available online at http://npg.nature.com/
reprintsandpermissions/
How to cite this article: Schmitz-Antoniak, C. et al. Electric in-plane polarization in
multiferroic CoFe
2
O
4
/BaTiO
3
nanocomposite tuned by magnetic fields. Nat. Commun.
4:2051 doi: 10.1038/ncomms3051 (2013).
This work is licensed under a Creative Commons Attribution-
NonCommercial-NoDerivs 3.0 Unported License. To view a copy of
this license, visit http://creativecommons.org/licenses/by-nc-nd/3.0/
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3051
8 NATURE COMMUNICATIONS | 4:2051 | DOI: 10.1038/ncomms3051 | www.nature.com/naturecommunications
& 2013 Macmillan Publishers Limited. All rights reserved.
... Magnetoelectric multiferroics are a fascinating class of materials, in which not only magnetic and ferroelectric orders coexist, but also a cross-coupling occurs between magnetic and electric degrees of freedom. This cross-coupling, i.e., the magnetoelectric effect, allows the application of these materials for the development of weak magnetic field sensors, low-power consuming magnetic-read/electric write memory elements, or energy harvesting devices [1][2][3][4][5][6]. ...
... On the contrary, composite multiferroics, where chemically different magnetic and ferroelectric phases are ''artificially'' connected, offer a convenient way to achieve sizable magnetoelectric coupling at room temperature [3]. These twophase magnetoelectrics can be fabricated with a variety of configurations, including particulate composites (with 3-3 or 3-0 connectivity) [4], pillar-matrix (1-3 connectivity) [5], and the layer-by-layer (2-2 connectivity) type [6]. ...
... When looking for new composite multiferroics, the proper choice of the ferroelectric/piezoelectric and the magnetostrictive material is very important. While it is natural that the mechanical properties of the composites are critical to achieving large magnetoelectric coupling [5,8,9], they have not been studied as intensively as ferroelectric or magnetic characteristics of the composites. ...
Article
Full-text available
We report on temperature-dependent studies of ultrasonic and dielectric properties of (x)0.5(Ba0.7Ca0.3)Ti03–0.5Ba(Ti0.8Zr0.2)O3(BCZT)/(1 − x)NiFe2O4 (BCZT/NFO) composite multiferroics and their relationship to the magnetoelectric (ME) effect in these materials. The most decisive factor in the maximization of the ME effect is the strong elastic softening of the BCZT phase at the phase transition between its ferroelectric phases with orthorhombic and tetragonal symmetry. The proximity of this phase transition to room temperature makes the system promising for practical applications of the ME effect. The magnetostrictive phase does not play any direct role in the determination of the ME temperature dependence because of its weakly temperature-dependent mechanical properties.
... The presence of partially filled d-orbitals is responsible for magnetic behavior, while the presence of empty dorbitals contributes to the ferroelectric behavior of multiferroics [7,8]. The multifunctionality of these materials leads to their significant potential for applications such as memory devices (FeRAMs, MRAMs) [9,10], spintronics, sensors and biomedical applications [11][12][13][14]. Mixed perovskite (doping d n ion in ferroelectric material) is one of the widely used approach to synthesize high performance multiferroic materials [15]. ...
Preprint
Full-text available
Present work investigates the detailed multiferroic properties of Pb 1 − x Sm x Ti 1− x Fe x O 3 , ( x = 0.21, 0.22, 0.23, 0.24 and 0.25) synthesize through solid state reaction route. The structural, dielectric, ferroelectric and magnetic properties were measured. A sincere study is carried out to detect the magneto-electric coupling in all the samples through magneto-dielectric, magnetization after electric poling and magneto-pe response. The maximum value of coupling coefficient (γ) and magneto-dielectric response (MDR) of sample x = 0.24 has shown by magneto-dielectric properties and magnetization after electric poling whereas magneto-pe has shown that all the samples possess multiferroic nature.
... In order to examine the enhanced multiferroic magnetoelectric coupling and to induce the electric nature inside the magnetic spinel ferrite CFO, there is a need of some modifications. Significant efforts for improving the properties of CFO have been focused on partially substituting the Fe 3? ions [9,10], composites with ferroelectric/piezoelectric materials [11], nanoparticles [12] and core/shell structures [13]. Among existing multiferroic magnetoelectric systems, CoFe 2 O 4 -BaTiO 3 /PbTiO 3 composite systems are extensively reconnoitered [2,4,6,14]. ...
Article
Full-text available
Magnetoelectric bulk composites of Co0.5Ni0.5Fe2O4–BaTiO3 (CNFO–BT) were synthesized employing solid-state reaction method. The structural properties of CNFO–BT composites as discussed by X-ray diffraction method confirm lattice distortion and enlarged strain owing to BT substitution in CNFO. The dielectric and impedance measurements exhibit conventional Maxwell–Wagner polarization and confirm the existence of grain dominated non-Debye relaxation phenomena in CNFO–BT composites. The magnetic hysteresis curves reveal strong ferromagnetic behavior in all composites. The maximum energy storage density and an efficiency achieved for 0.8CNFO–0.2BT composite are 4.25 mJ/cm3 and 31.6%, respectively. The variation of polarization with magnetic field confirmed the highest magnetoelectric coefficient of 5 mV/cm/Oe for 0.8CNFO–0.2BT composite. The variation of dielectric permittivity and ferroelectric polarization with magnetic field reveals lattice distortion, interfacial charge polarization and restricted ferromagnetic domain wall rotation arising from substitution of BT in CNFO. These structure-dependent results suggest potential application of CNFO–BT composites in magnetoelectric sensors and energy storage devices.
Article
Electrical control of magnetization or magnetic control of polarization offers an extra degree of freedom in materials possessing both electric and magnetic dipole moments, viz., magnetoelectric (ME) multiferroics. A microstructure with polycrystalline configurations that enhances the overall polarization/magnetization and that outperforms single crystalline configurations is identified in a 1–3 CoFe[Formula: see text]O[Formula: see text]–BaTiO[Formula: see text] (or CFO–BTO) composite. The characterization of local fields corresponding to the polycrystal configuration underlines a nontrivial role played by randomness in better cross coupling mediated by anisotropic and asymmetric strains. The microscopic field ( local field) profile of the composite provides rich information regarding the distribution of key parameters central to the magnetoelectric effect. The differential contractual stress level observed in the local stress profile of CFO–BTO composite upon applying an external magnetic field conforms with the previous experimental magnetostriction observed in CFO. The role played by residual stresses stemming from misalignment of the polarization in the neighboring grains in enhancing the ME coupling is briefly discussed.
Article
Piezoelectric semiconductors can be polarized and used in mechanoredox systems and in photoredox catalysis. Conventional non-piezoelectric semiconductors have limitations when it comes to charge carrier recombination and slow transport rates in catalytic reactions, which can be overcome by piezoelectric polarization processes in piezoelectric semiconductors. Heterostructures based on semiconducting piezoelectrics often offer enhanced catalytic reactivities that are related to their mechanical, piezoelectric, optical, and electronic characteristics. We review how to use such heterostructures to convert mechanical energy into chemical energy, and how the related piezoelectric polarization tunes the band structures and provides advantages in piezophotocatalysis over regular photocatalysis. We discuss fundamental concepts of piezoelectricity, piezoelectric potential, and examine different piezoelectric heterostructures for piezo- and piezophotocatalysis. A review of dynamic investigations of piezo- and piezophotocatalytic processes is presented. The complementary developments in the understanding of the piezotronic and piezophototronic effects are described, which include the induced charge-transfer mechanisms for piezo- and piezophotocatalytic reactions that can occur with piezoelectric heterostructures. Finally, we derive design principles and suggest future research directions in the emerging field of piezo- and piezophotocatalysis employing semiconductive heterostructures.
Article
Present work proposes new insights in the field of multiferroics and its properties. The substitution of SmFeO3 in PbTiO3 (Pb1-xSmxTi1-xFexO3, x = 0.21, 0.22, 0.23, 0.24 and 0.25) over a range of composition has been synthesized and explored for multiferroicity. The properties such as ferroelectric, magnetic as well as coupling between these two have been investigated at room temperature. An in-depth study is carried out to detect the magneto-electric coupling in all the samples through magneto-dielectric, magnetization after electric poling and magneto-pe. The magneto- pe measurement gives the indication of multiferroic nature of all the samples. The maximum value of coupling coefficient (γ) and magneto-dielectric response (MDR) of about 10.3 g²/emu² and 13.09 at 1.5T respectively for sample x = 0.24 has been achieved by magneto-dielectric properties. Moreover, a comparative higher shift (∼0.061 emu/g) in magnetization has been observed after electric poling.
Article
In this research work, the structural and electrical properties of La0.8Pb0.2Fe0.75Mg0.25O3, elaborated by sol–gel process, are studied. The X-ray diffraction profile refinement shows that the structure of this perovskite is orthorhombic. DC conductivity (σDC) reveals the semi-conductor behavior and the activation energy is estimated to be 547 meV. It is also obvious that the conductivity AC follows Jonsher's universal power-law and the exponent “s” decreases with temperature increase, showing that the phenomena of conduction is related to Correlated Barrier Hopping (CBH). The plots of Nyquist demonstrate the existence of two different relaxation contributions linked to grain boundaries and grains.
Article
Utilizing the magnetostrictive properties of CoFe2O4, we demonstrate reversible room temperature control of the Ti electronic structure in SrTiO3-CoFe2O4 heterostructures, by inducing local and reversible strain in the SrTiO3. By means of X-ray absorption spectroscopy, we have ascertained the changes that take place in the energy levels of the Ti 3d orbitals under the influence of an external magnetic field. The observed Ti electronic state when the sample is subjected to moderately large external magnetic fields and the disappearance of the induced phase upon their removal indicates lattice distortions that are suggestive of the development of a net electric polarization.
Article
Full-text available
In this report, the multiferroic composites consisting of Yb-doped PbZrTiO3–Nd-doped CoFe2O4 of varying concentrations were synthesized by solid-state reaction method. The XRD studies of all the prepared samples confirm their crystallographic phases. The Williamson–Hall (W–H) approach was employed to obtain the average crystallite size and the strain in the composites. The SEM analysis ascertained the non-uniform distribution with agglomerated grains in the composites. The ferroelectric nature of the composites was confirmed from the P-E loops traced out. The M-H hysteresis loops obtained confirmed the ferromagnetic nature of the composites. The magneto-dielectric studies of the composites manifest the strong coupling between the ferrite and ferroelectric phases. The magneto-capacitance MC (%) revealed improvement at the higher ferrite content of the composites. The simultaneous occurrence of both the magnetic and ferroelectric (M-H and P-E) hysteresis loops and strong magneto-electric coupling signifies the multifunctionality and applicability of the prepared composite multiferroics in modern technology.
Chapter
Preparations, characterization, and applications of multiferroic nanocomposites will be discussed in this chapter. The chapter starts with a brief introduction on multiferroic nanocomposites and preparation, followed by a discussion on characterization of multiferroic nanocomposites, and ends with the applications of multiferroic nanocomposites.
Article
Full-text available
The electronic structures of BaTiO3@X (core@shell) nanoparticles ( X=gamma-Fe2O3 , Fe3O4 , and Fe) have been investigated by employing soft x-ray-absorption spectroscopy and x-ray magnetic circular dichroism (XMCD). It is found that the valence states of Ti ions near interfaces are formally tetravalent (Ti4+:3d0) and that the valence states of Fe ions in shells are essentially the same as those of the corresponding bulk materials, with some disorder in the site occupations for X=gamma-Fe2O3 and Fe3O4 . The negligible Ti2p XMCD signals were observed, indicating that the induced spin polarization of the interface Ti3d electrons is negligible in BaTiO3@X nanoparticles.
Article
Full-text available
Recent research activities on the linear magnetoelectric (ME) effect---induction of magnetization by an electric field or of polarization by a magnetic field---are reviewed. Beginning with a brief summary of the history of the ME effect since its prediction in 1894, the paper focuses on the present revival of the effect. Two major sources for 'large' ME effects are identified. (i) In composite materials the ME effect is generated as a product property of a magnetostrictive and a piezoelectric compound. A linear ME polarization is induced by a weak ac magnetic field oscillating in the presence of a strong dc bias field. The ME effect is large if the ME coefficient coupling the magnetic and electric fields is large. Experiments on sintered granular composites and on laminated layers of the constituents as well as theories on the interaction between the constituents are described. In the vicinity of electromechanical resonances a ME voltage coefficient of up to 90 V cm-1 Oe-1 is achieved, which exceeds the ME response of single-phase compounds by 3-5 orders of magnitude. Microwave devices, sensors, transducers and heterogeneous read/write devices are among the suggested technical implementations of the composite ME effect. (ii) In multiferroics the internal magnetic and/or electric fields are enhanced by the presence of multiple long-range ordering. The ME effect is strong enough to trigger magnetic or electrical phase transitions. ME effects in multiferroics are thus 'large' if the corresponding contribution to the free energy is large. Clamped ME switching of electrical and magnetic domains, ferroelectric reorientation induced by applied magnetic fields and induction of ferromagnetic ordering in applied electric fields were observed. Mechanisms favouring multiferroicity are summarized, and multiferroics in reduced dimensions are discussed. In addition to composites and multiferroics, novel and exotic manifestations of ME behaviour are investigated. This includes (i) optical second harmonic generation as a tool to study magnetic, electrical and ME properties in one setup and with access to domain structures; (ii) ME effects in colossal magnetoresistive manganites, superconductors and phosphates of the LiMPO4 type; (iii) the concept of the toroidal moment as manifestation of a ME dipole moment; (iv) pronounced ME effects in photonic crystals with a possibility of electromagnetic unidirectionality. The review concludes with a summary and an outlook to the future development of magnetoelectrics research.
Article
Full-text available
A phase-field model was developed for studying the magnetoelectric coupling effect in epitaxial ferroelectric and magnetic nanocomposite thin films. The model can simultaneously take into account the ferroelectric and magnetic domain structures, the electrostrictive and magnetostrictive effects, substrate constraint, as well as the long-range interactions such as magnetic, electric, and elastic interactions. As an example, the magnetic-field-induced electric polarization in BaTiO3-CoFe2O4 nanocomposite film was analyzed. The effects of the film thickness, morphology of the nanocomposite, and substrate constraint on the degree of magnetoelectric coupling were discussed.
Article
Recent research activities on the linear magnetoelectric (ME) effect—induction of magnetization by an electric field or of polarization by a magnetic field—are reviewed. Beginning with a brief summary of the history of the ME effect since its prediction in 1894, the paper focuses on the present revival of the effect. Two major sources for 'large' ME effects are identified. (i) In composite materials the ME effect is generated as a product property of a magnetostrictive and a piezoelectric compound. A linear ME polarization is induced by a weak ac magnetic field oscillating in the presence of a strong dc bias field. The ME effect is large if the ME coefficient coupling the magnetic and electric fields is large. Experiments on sintered granular composites and on laminated layers of the constituents as well as theories on the interaction between the constituents are described. In the vicinity of electromechanical resonances a ME voltage coefficient of up to 90 V cm−1 Oe−1 is achieved, which exceeds the ME response of single-phase compounds by 3–5 orders of magnitude. Microwave devices, sensors, transducers and heterogeneous read/write devices are among the suggested technical implementations of the composite ME effect. (ii) In multiferroics the internal magnetic and/or electric fields are enhanced by the presence of multiple long-range ordering. The ME effect is strong enough to trigger magnetic or electrical phase transitions. ME effects in multiferroics are thus 'large' if the corresponding contribution to the free energy is large. Clamped ME switching of electrical and magnetic domains, ferroelectric reorientation induced by applied magnetic fields and induction of ferromagnetic ordering in applied electric fields were observed. Mechanisms favouring multiferroicity are summarized, and multiferroics in reduced dimensions are discussed. In addition to composites and multiferroics, novel and exotic manifestations of ME behaviour are investigated. This includes (i) optical second harmonic generation as a tool to study magnetic, electrical and ME properties in one setup and with access to domain structures; (ii) ME effects in colossal magnetoresistive manganites, superconductors and phosphates of the LiMPO4 type; (iii) the concept of the toroidal moment as manifestation of a ME dipole moment; (iv) pronounced ME effects in photonic crystals with a possibility of electromagnetic unidirectionality. The review concludes with a summary and an outlook to the future development of magnetoelectrics research.
Article
A summarizing account is given of the research on barium titanate in progress at the Laboratory for Insulation Research at M. I. T. since 1943. The investigations have led to an understanding of the mechanism of ferroelectricity in the titanates and to discoveries such as the piezoelectric effect in the ceramics and the domain structure of the single crystals of BaTiO3. The high dielectric constant, field strength and temperature sensitivity, and piezo-response of the barium titanate dielectrics make them useful for numerous technical applications.
Article
We expand the previous theoretical treatment for the strong anisotropy of the x-ray magnetic linear dichroism (XMLD) in a crystal field of cubic point-group symmetry to the more general case of tetragonal point-group symmetry. For the cubic symmetry, there are only two fundamental spectra, which have the same shape for rotation of either linear light polarization E or magnetization direction H . For the tetragonal symmetry, the XMLD is a linear combination of four fundamental spectra, with a different shape for linear dichroism (rotation of E ) and magnetic dichroism (rotation of H ). However, only one extra spectrum is required to relate the linear and magnetic dichroism. The validity of the theory is demonstrated using a CoFe2O4(011) thin film on SrTiO3 , which has both tetrahedrally distorted symmetry and large magnetic anisotropy. The XMLD at the Co L2,3 edges was found to exhibit a strong dependence on the relative orientation of external magnetic field, x-ray polarization, and crystalline axes. The large variations in the peak structure as a function of angle are not caused by the spin-orbit-induced magnetocrystalline anisotropy but arise from the symmetry of the measurement geometry. The results are compared with calculated spectra using atomic multiplet theory for Co2+ d7-->2p5d8 in octahedral and tetragonal crystal field symmetry. Although the magnitude of the dichroism is strongly influenced by the temperature, its spectral shape remains largely unaffected. The measured fundamental spectra are also robust against incomplete magnetization. The influence of the tetragonal distortion is revealed by small differences between the linear and magnetic dichroism. It is shown that the magnetic dichroism spectra can be transferred from CoFe2O4 to CoO. Therefore, the rich structure in the Co2+ L3 XMLD provides a sensitive probe to determine the orientation of the spin axis with respect to the crystalline axes, hence offering a valuable tool for experimentalists for the study of exchange bias in Co oxides. In contrast, the Co2+ L2 edge, where the fundamental spectra have similar spectral shape but with opposite sign, does not allow an unambiguous determination.
Article
High resolution x-ray absorption spectroscopy (XAS) affords new insight into the microscopic properties of perovskite transition metal oxides. Interpretation of XAS spectra in transition metal oxides requires theoretical tools capable of describing relativistic and many-body effects. In this work, full relativistic (SPR-KKR) and multiplet calculations (CTM4XAS) are carried out and compared to experimental multiedge XAS spectra of BaTiO3 single crystals. The impact of relativistic and many-body effects on the calculated density of states and x-ray absorption near edge structure spectra are individually considered.
Article
We present a first-principles comparison of BaTiO3,CaMnO3, and YMnO3 which reveals the fundamental role of the d electron occupation in preventing or favoring the simultaneous presence of magnetic and electric polarization. The ferroelectric compounds are distinguished from the nonferroelectrics by an ultrasensitivity of the p-d charge hybridization to the atomic displacements. To be effective these hybridization changes require the d orbitals oriented in the direction of the ferroelectric distortion to be formally empty. This orbital directionality can explain the coexistence of antiferromagnetism and ferroelectricity in the hexagonal phase of YMnO3, which is ferroelectric along the c axis: in the Mn3+ ion, four filled d orbitals provide a magnetic moment, and one empty d orbital (dz2) can participate in hybridization changes and induce a ferroelectric distortion. The Born effective charges are found to be highly anomalous in CaMnO3, which is far from being ferroelectric.