Magnetic Field-Induced Ferroelectric Switching in Multiferroic Aurivillius Phase Thin Films at Room Temperature

Magnetic Field-Induced Ferroelectric Switching in Multiferroic AurivilliusPhase Thin Films at Room Temperature

Lynette Keeney,‡,† Tuhin Maity,‡ Michael Schmidt,‡ Andreas Amann,‡,§ Nitin Deepak,‡

Nikolay Petkov,‡ Saibal Roy,‡ Martyn E. Pemble,‡ and Roger W. Whatmore,‡

‡Tyndall National Institute, University College Cork, Lee Maltings, Dyke Parade, Cork, Ireland

§School of Mathematical Sciences, University College Cork, Ireland

Single-phase multiferroic materials are of considerable interest

for future memory and sensing applications. Thin films of Auri-villius phase Bi7Ti3Fe3O21 and Bi6Ti2.8Fe1.52Mn0.68O18 (pos-

sessing six and five perovskite units per half-cell, respectively)

have been prepared by chemical solution deposition on c-planesapphire. Superconducting quantum interference device magne-

tometry reveal Bi7Ti3Fe3O21 to be antiferromagnetic (TN =190 K) and weakly ferromagnetic below 35 K, however,

Bi6Ti2.8Fe1.52Mn0.68O18 gives a distinct room-temperaturein-plane ferromagnetic signature (Ms = 0.74 emu/g, l0Hc =7 mT). Microstructural analysis, coupled with the use of a

statistical analysis of the data, allows us to conclude that

ferromagnetism does not originate from second phase inclu-sions, with a confidence level of 99.5%. Piezoresponse force

microscopy (PFM) demonstrates room-temperature ferroelec-

tricity in both films, whereas PFM observations on Bi6Ti2.8-Fe1.52Mn0.68O18 show Aurivillius grains undergo ferroelectric

domain polarization switching induced by an applied magnetic

field. Here, we show for the first time that Bi6Ti2.8-Fe1.52Mn0.68O18 thin films are both ferroelectric and ferromag-netic and, demonstrate magnetic field-induced switching of

ferroelectric polarization in individual Aurivillius phase grains

at room temperature.

I. Introduction

AS the use of computers continues to expand rapidly,there is increasingly a need for data storage technolo-

gies with higher densities, nonvolatility and lower power con-sumption.1 Single-phase, room-temperature magnetoelectricmultiferroic materials are of considerable interest for suchapplications.2–8 However, materials that are magnetoelectricat room temperature are very unusual9 (see Sidebar A). Theperovskite ferroelectric BiFeO3 exhibits antiferromagnetic

ordering at ambient temperature10 and its electric polariza-tion has been used to control antiferromagnetic ordering,11–13

but there is no evidence as of yet that its ferroelectric polari-zation can be switched by a magnetic field. There has, there-fore, been an intense search for room-temperaturemagnetoelectric multiferroics within which the coupling offerroelectric and ferromagnetic polarizations might bedemonstrated.

The ferroelectric Aurivillius layer structures,37 describedby general formula Bi2O2(Am � 1BmO3m + 1), are naturallytwo-dimensionally nanostructured with large c-axis parame-ters, high Curie temperatures (>600°C) and large in-planespontaneous polarizations. The number of ABO3 perovskiteunits (m) per half-cell can be changed within the range 2–9,depending on composition, and a wide variety of B-sitecations with +3 to +5 oxidation states accommodated.38–41

The system, discussed in greater detail in Sidebar B, offersthe potential for including substantial amounts of magneticcations within a strongly ferroelectric system, and hence, thepotential for the discovery of new room-temperature multif-erroics.

Lomanova et al.42 explored ceramics with general for-mula Bim � 1Fem � 3Ti3O3m + 1, and demonstrated the exis-tence of structures with m from 4 to 9, including some withfractional m. Weak room-temperature ferromagnetism hasbeen reported for m = 4,49 and antiferromagnetism (80–300 K) for m = 650 and m = 7.51 Zurbuchen et al.52 showedthat the manganese analogue of this system with m = 6 wasferromagnetic below 55 K, but not ferroelectric. Ferroelec-tricity and ferromagnetism above room temperature wasreported for cobalt-substituted, four-layered Bi5Ti3Fe0.5-Co0.5O15 ceramic,53 with a small remanent magnetization.Subsequent investigations23,34,54,55 of Bi5Ti3Fe0.5Co0.5O15

ceramics and Bi5Ti3Fe0.7Co0.3O15 films also demonstratedferroelectric and ferromagnetic behavior at room tempera-ture. However, detailed phase analyses detected trace levelsof CoFe2�xTixO4 second phase inclusions, not observed byX-ray diffraction (XRD), but which accounted for theobserved magnetization. Indeed, a remanent magnetizationof 7.8 memu/g as observed by Mao et al.53 would corre-spond to a trace CoFe2O4 second (or impurity)-phase levelof only 0.03 wt%, which would be very hard to see by any

D. Johnson—contributing editor

Manuscript No. 32991. Received April 3, 2013; approved May 24, 2013.†Author to whom correspondence should be addressed. e-mail: lynette.keeney@

tyndall.ie

J. Am. Ceram. Soc., 1–19 (2013)

DOI: 10.1111/jace.12467

© 2013 The American Ceramic Society

Journal

Feature

microanalytical method. This observation clearly demon-strates the difficulty of unambiguous assignment of magneticeffects to the parent Aurivillius phase. Other work56 hasreported ferroelectric and ferromagnetic behavior in thinfilms of Bi4.15Nd0.85Ti3Fe0.5Co0.5O15. Compounds with

higher values of m, such as Bi6Ti3Fe2O18 (B6TFO, m = 5),provide a means for increasing the proportion of magneticcations. Weak ferromagnetic/antiferromagnetic behavior wasreported in rare earth and co-doped B6TFO57–60 ceramicsand thin films, but none of the work presented phase analy-

Sidebar A. Multiferroics and Magnetoelectrics

Ferroelectric materials form a subset of the polar electrically polarizable materials for which the electrical dipole moments withintheir structure can be switched between at least two stable states (e.g., up and down) by an external electric field. Ferromagnetsform a subset of the magnetically polarizable materials (Sidebar D). Materials which demonstrate both ferroelectric andferromagnetic properties within the same phase are known as multiferroic materials (Fig. A.1).14–19 Magnetoelectric couplingrefers to the induction of magnetization by an electric field, or vice versa, and may arise through direct coupling betweenmagnetic and electric polarizations in a single material, or indirectly via strain-mediated coupling in a multiphase material.3,20–25

Such strain-mediated indirect magnetoelectric coupling in composite materials can occur, for example, through a magnetostric-tive strain (induced in one phase by a change in applied magnetic field) coupling to a piezoelectrically induced polarizationchange in a second phase mechanically coupled to the first. The SI unit of the magnetoelectric coupling coefficient, a, is (sm�1)which can be converted into the technical unit (V cm�1 Oe�1) if the permittivity (e) of the given material is known:(sm�1) = 1.1 9 10�11 e(V cm�1 Oe�1). A magnetoelectric coupling coefficient of 5.90 V cm�1 Oe�1 has been reported forlaminate complexes of lead zirconate titanate and Terfenol-D (TbDyFe2) by straining the magnetostrictive phase under a DCmagnetic bias of 4.2 T,26 which induces stress on the piezoelectric phase, generating an electric field in the piezoelectric phase.Magnetic force microscopy (MFM) imaging of (BiFeO3)0.65–(CoFe2O4)0.35 nanostructured composite heterostructures,27

demonstrated two electrically switchable perpendicular magnetic states at ambient conditions for ferromagnetic CoFe2O4

nanopillars embedded in BiFeO3. In addition, the magnetoelectric coupling effect becomes controllable in a weak perpendicularmagnetic field. Thin film heterostructures28 of CoFe–BaTiO3 grown by electron beam evaporation exhibit giant magnetoelectriccoupling coefficients (3 9 10�6 sm�1) at room temperature. Lahtinen et al. demonstrated that it is possible to precisely write anderase regular ferromagnetic domain patterns and to control the motion of magnetic domain walls in small electric fields over largeareas in these composites by strain-mediated correlations between ferromagnetic domain walls and ferroelastic domain boundaries.

Fig. A.1. The relationship between multiferroic andmagnetoelectric materials. (Redrawn from Ref. [23]).

Multiferroic coupling in a single-phase material can occur when the switching of one order parameter (e.g., ferromagneticpolarization) induces a switching of the other order parameter (in this example ferroelectric polarization). Examples of single-phase magnetoelectrics include: Cr2O3 (<260 K),29 CuO (<230 K),30 TbMnO3 (<27 K),31 Ni3B7O13I (<64 K),32 and DyMn2O5

(<43 K).33 The synthesis of novel room-temperature single-phase magnetoelectric multiferroic materials is particularlyappealing, not only because they have two sets of interesting physical properties with four polarization states (positive andnegative in both electrical and magnetic polarizations) but also because the multiferroic coupling interactions could lead to arange of potential applications. Such four-state multiferroic materials could potentially lead to a new generation of rapid,energy efficient magnetoelectric memory devices that can be electrically written and magnetically read, storage of multiple bitsper memory element,4 and magnetic field sensors where the ferromagnetic resonance could be tuned electrically instead ofmagnetically.21,34

However, single-phase magnetoelectric multiferroics are rare, especially at room temperature, due to the competingelectronic requirements for ferroelectricity and ferromagnetism (see Hill9). Hybridization between the cation and anion withinthe unit cell is essential for stabilizing ferroelectric distortion; therefore the cation driving ferroelectricity must formally be inthe d0 state. Conversely, d-orbital occupancy is a requirement for the existence of magnetic ordering. Hill has suggested thatthe conditions for obtaining ferroelectricity and ferromagnetism in a single phase can potentially be met by incorporating d0

and dn cations into the same structure.

Sr3Co2F24O21 demonstrates low-field magnetoelectric effects at room temperature, however, shows no polarization at zeromagnetic field and therefore is not a bilinear magnetoelectric.35 SrCo2Ti2Fe8O19 does exhibit spontaneous polarization at zeromagnetic field (~25 lC/cm)5 and MFM investigations of this ceramic under various electric fields (Edc = �20 to +20 kv/cm)demonstrated electric field control of magnetism at room temperature in the absence of a magnetic field bias (conversemagnetoelectric effect) and decreases in magnetization of up to 6.3% on application of a magnetic field of 46 mT and Edc of 22 kv/cm. Recently, a newly discovered single-phase multiferroic, [Pb(Zr0.53Ti0.47)O3]0.6–[Pb(Fe0.5Ta0.5)O3]0.4 has been shown to exhibitsignificant (~60% change in polarization) magnetoelectric coupling at room temperature (~1 9 10�7 sm�1),36 demonstrating thatwith materials development and design, the development of room-temperature multiferroic materials can be achieved.

2 Journal of the American Ceramic Society—Keeney et al.

ses at a level which would exclude the possibility that theferromagnetic responses were due to trace-level secondphases.

In this study, thin films of Bi7Ti3Fe3O21 (B7TFO) andBi6Ti2.8Fe1.52Mn0.68O18 (B6TFMO) with six and five perov-skite layers, respectively, were prepared by chemical solution

Sidebar B. The Aurivillius Family of Ferroelectric Oxides

Aurivillius37 bismuth-based compounds, sometimes referred to as the layered perovskites and described by the general formulaBi2O2(Am � 1BmO3m + 1), represent an important class of ferroelectric compounds. The materials are members of anhomologous series of Bi-layered oxides, where the structure is a naturally layered nanocomposite. The 2-D nanostructures havelarge c-axis lattice parameters, in the nanometer range, and consist of fluorite-structured (Bi2O2)

2+ layers of thickness f (typically~0.4 nm) lying in the (001) plane alternating with mABO3 perovskite units in a sandwich type arrangement. The averagethickness of the perovskite-type block, h, depends on the number of octahedral perovskite units (m) in the block: h = pm where pis the average thickness of the perovskite-like units (also typically ~0.4 nm).42 (Note that this is only an approximation, asoctahedral tilting, and choice of A & B cations will change the average height of each perovskite unit.43–45) The value of m can beinteger or fractional.14 Fractional values of m usually occur with “mixtures” between a pure Aurivillius phase compound and aperovskite end-member and are formed by recurrent intergrowth of the perovskite blocks of two Aurivillius end-members,forexample, BaBi8Ti7O27 (m = 3.5) is formed from (Bi4Ti3O12)0.75–(BaTiO3)0.25.

46,47 The values of f and h are related to the c-cellparameter by f + h = c/2 (Fig. B.1).

Fig. B.1. Projection [approximately down (101)] of half-unit cellsof (a) Bi7Ti3Fe3O21 and (b) Bi6Ti3Fe1.6Mn0.6O18 displaying the in-plane lattice directions (100) (dashed arrow) and (110) (yellowplane). Drawn using Crystallographica v1.60d62 and Mercury 3.0Crystal Structure Visualization software.95

The layered structured Aurivillius phase materials are a particularly attractive class of oxides as their structure allows thedesign and synthesis of new materials in thin film form with interesting electrical and magnetic properties. Between thebismuth oxide layers, the number of octahedral layers can be increased and a homologous series of compounds with thegeneral formula Bim + 1Fem � 3Ti3O3m + 3 (m = 4–9) has been realized by inserting bismuth ferrite units, BiFeO3, into three-layered bismuth titanate, Bi4Ti3O12. In Fig. B.1(a), three units of BiFeO3 have been inserted into Bi4Ti3O12 to form the six-layered material, Bi7Ti3FeO21.

For this homologous series, Lomanova et al.42 have pointed out that, as the number of perovskite-like layers increases, thecell c parameter rises almost linearly, implying that the perovskite-like units incorporated into the Aurivillius phase structureexperience only slight changes along the c-axis with increasing m. For this series, an average thicknesses of the perovskitelayers, p ffi 4.11 �A and the fluorite layers, f ffi 4.08 �A was estimated.42

The Fe distribution over the two nonequivalent octahedral B cation sites in the perovskite block (identified as B(1) insidethe block and B(2) for the octahedra on the outer sides of the block adjacent the Bi2O2

2� layers) has been investigated forthe Bim + 1Fem � 3Ti3O3m + 3 series.14 For m = 3.5–7, Fe3+ ions preferentially occupy the B(1) sites, however, the ordereddistribution of ions over B(1) and B(2) sites decreases with the increase in the perovskite-like block thickness. At m ≥ 7, thedistribution of Fe3+ and Ti4+ ions of the perovskite-type block tends to become more random and when the value of mincreases up to 8–9, concentrations of ions at B(1) and B(2) sites equalize.14

On increasing the number of perovskite layers (m), the microstructural, magnetic, and physical properties of the materialscan be altered significantly.38 The layered nature of these materials also allows for the incorporation of significant amounts ofmagnetic ions with +3 to +5 oxidation states48 within the mABO3 perovskite units. In this way, the normally conflictingelectronic structure requirements for ferroelectricity (unoccupied d-orbitals, d0) and ferromagnetism (partially filled d-orbitals,dn) in a single phase9 can potentially be circumvented and the fabrication of single-phase magnetoelectric multiferroicmaterials could conceivably be accommodated.

Magnetoelectric Switching in Multiferroic Thin Films 3

deposition on sapphire substrates to increase the content ofmagnetic cations within the Aurivillius structures Fig. B.1(b).Piezoresponse force microscopy (PFM) indicates that bothfilms are ferroelectric and SQUID (superconducting quantuminterference device magnetometry) magnetometry investiga-tions demonstrate the B6TFMO films are multiferroic atroom temperature. Careful microstructural analyses gives99.5% confidence that the ferromagnetic responses originatefrom the parent phase. PFM under magnetic fields has givendirect evidence that the ferroelectric and ferromagnetic orderparameters within the Bi6Ti2.6Fe1.77Mn0.63O18 grains aremagnetoelectrically coupled. This material is therefore an

exciting candidate for potential use in multiferroic, magneto-electric logic devices.

II. Experimental Procedure

(1) Thin Film SynthesisSolutions of Bi7Ti3Fe3O21 (m = 6; B7TFO) and Bi6Ti2.5-Fe1.75Mn0.75O18 (m = 5; B6TFMO) were prepared by dissolv-ing Bi(NO3)3�5H2O and Ti(OCH2CH2CH2CH3)4 in lactic acidat room temperature. Fe(NO3)3�9H2O and Mn(C5H7O2)3, asappropriate, were dissolved separately in acetylacetone. Whencomplete dissolution was achieved, the Fe3+ or Fe3+/Mn3+

Sidebar C. Piezoelectric Force Microscopy Applied to Ferroelectric Thin Films

Piezoresponse Force Microscopy has emerged as a powerful technique for locally probing nanoscale phenomena in piezoelectricand ferroelectric materials on the nanometer scale.67–69 PFM is based on the detection of a bias-induced piezoelectric surfacedeformation.70 A conductive tip is brought into contact with a piezoelectric sample surface, and the tip deflection resulting fromthe expansion or contraction of the sample due to the applied bias is measured [Fig. C.1(a)].

(a)

(b)

Fig. C.1. (a) Schematic representation of vertical PFM operationand (b) triangle step bias waveform applied to the sample duringswitching-spectroscopy PFM hysteresis loop acquisition.

Out-of-plane polarization is measured by recording the tip-deflection signal at the frequency of modulation. Therelationship between the strain and the applied electric field in piezoelectric materials is called the “converse piezoelectriceffect” and the vertical displacement under voltage can be expressed as follows:

Dz ¼ jðd33Vþ ðQ333=tÞV2Þ

where V is the applied voltage, t is the sample thickness, d33 and Q333 are the piezoelectric and electrostrictive coefficients,respectively, and j is a constant of proportionality, placed in the equation to indicate that the piezoelectric and electrostric-tive coefficients measured using PFM are not numerically identical with the coefficients measured using (for example) IEEE(Institute of Electrical and Electronics Engineers) Standard methods on bulk specimens because of effects such as electric fielddistortion and substrate clamping (on thin films). d33 is the most important component of the piezoelectric tensor for a typicalvertical PFM as it couples directly into the vertical motion of the cantilever. The electrostrictive effect is quadratic and doesnot depend on the sign of the electric field with respect to the specimen polarization and causes only a constant effect, whichis zero in the absence of an applied bias. On application of the external voltage from the tip, Vtip = Vdc + Vac cos(xt), drivenat frequency well below that of the contact resonance of the cantilever, the local piezoelectric response is detected as the firstharmonic response of the tip deflection:

Dz ¼ d33Vdc þ d33Vaccosðxtþ uÞ


ramesh

Inserted Text

(C.1) <insert equation no as per style)

ramesh

Inserted Text

(C.2) <insert equation no as per style)

solution was slowly dropped into the Bi3+/Ti4+ solutionunder constant stirring to prepare 0.03 mol dm�3 solutions.For all solutions, 17.5 mol% excess bismuth41 was used tocompensate for evaporation of bismuth during the annealingprocess. The films were spin coated on c-plane sapphire sub-strates by a commercial spinner (spin coater KW-4A, ChematTechnology) operating at 1000 rpm for 30 s. Residual organ-ics were removed from the films by baking on a calibratedhot plate at 300°C � 5°C for approximately 10 min. Filmswere annealed in ambient air for 1 h in a conventionalfurnace at temperatures of 850°C. Final thicknesses of ~100and ~200 nm were obtained for B7TFO and B6TFMO,respectively, as observed from cross-section high-resolutionscanning electron microscopy (HR-SEM) measurements.

(2) X-Ray DiffractionX-ray diffraction profiles were recorded at room temperatureusing a Philips Xpert PW3719 MPD diffractometer, equipped

with a CuKa radiation source (40 kV and 35 mA) and anickel filter on the incident beam over the range5° ≤ 2h ≤ 37.5°. To estimate the degree of c-axis orientation,the Lotgering factor,61 f, was calculated using theoretical(hkl) intensities of B7TFO and B6TFMO obtained fromCrystallographica62 software.

(3) Topography MappingHigh-resolution scanning electron microscopy images andenergy dispersive X-ray (EDX) analysis spectra wereobtained using a FEI Quanta 630 High-Resolution ScanningElectron Microscope with attached Oxford X-Max 20 detec-tor and Inca analysis software. A commercial atomic forcemicroscope (MFP–3DTM; Asylum Research) in AC mode,equipped with Olympus AC160TS silicon cantilevers(Olympus) (Al reflex coated, ~300 kHz resonant frequency),was used for topography mapping of the films. Cross sec-tions of the films were prepared for microstructural analysis

where u, the phase of the electromechanical response of the surface, yields information on the polarization direction belowthe tip. Both the magnitude and sign of the displacement can be translated into images of local piezoresponse and polariza-tion direction observed as amplitude and phase micrographs, respectively (e.g., Figs. 1(b) and (c) in the main document).When the polarization is parallel to the sample surface, the voltage applied to the tip in the lateral PFM mode couples to anin-plane surface displacement via one of the piezoelectric shear coefficients, which is translated into a torsional deflection ofthe cantilever. Due to the small radius of the PFM tip, the applied electric field spreads within the top layer of the sampleresulting in rapidly decreasing field strength at larger sample depths. A bottom electrode ensures a well-defined electric fielddistribution and thereby reproducible conditions for PFM imaging.71 In the absence of a conductive bottom electrode, partof the electric field decays in the substrate, whereas with a bottom electrode the electric field is fully applied to the film.Therefore, for thin film samples, a bottom electrode is more advantageous for PFM measurements, while larger bias valuesare required to image ferroelectric domains in the absence of a bottom electrode.72 As Soergel71 has explained, there is noabsolute need for a back electrode and PFM images have been recorded on films of 10 nm thickness without a bottom elec-trode. However, the stabilization of switched domains is greatly facilitated when a bottom electrode is present.72

The DART-PFM (Dual AC-Resonance Tracking-PFM) technique was developed by Rodriguez et al.64 to overcome prob-lems in electromechanical imaging of piezoelectric materials with relatively small vertical piezoresponses. The method uses thecantilever resonance frequency to boost the piezo signal in the vertical direction, while reducing cross talk between changes inthe sample-tip contact stiffness and the PFM signal by tracking the resonance frequency based on amplitude detection feed-back. The amplitude is measured at one drive frequency below the resonance frequency and another above it. The error sig-nal allows changes in the resonance frequency to be tracked, thereby reducing the effects of cross talk between the PFMsignal and changes in the sample-tip contact stiffness.

As well as local nanoscale imaging and polarization mapping, PFM spectroscopy modes can locally generate hysteresisloops and thereby provide information on local ferroelectric switching behavior.65,66,73 During acquisition of a hysteresis loopin switching spectroscopy PFM, the conducting PFM tip is fixed at a given location on the sample surface and a triangle stepbias waveform is applied as a function of time [Fig. C.1(b)]. The signal has two components, a DC bias voltage which isstepped up and down on a saw-tooth profile, plus an AC signal [at much higher frequency; refer to insets in Fig. C.1(b)]which is used to measure the displacement caused by the piezoelectric effect in the sample. The system records the amplitudeand phase of this displacement signal with the “field on” followed by the displacement signal with “field off”. Piezoelectric hysteresisloops are thus generated with the “field on” and “field off” [e.g., Figs. 2(a)–(d)] allowing characterization of electromechanicaland structural properties of a wide range of ferroelectric materials leading to a better understanding of material functionalitydown to the nanoscale level. Note that the electrostrictive term in Eq. (C.1). Above is zero in the “field off” condition. Theresponse in the “field on” state can also be affected by electrostatic interactions between the cantilever and the back electrode,which can be mistakenly interpreted as being due to electrostriction, and should thus be treated with caution.

In ferroelectric lithography, the external field is applied vertically to a ferroelectric thin film using the PFM probe, allowingcomplex ferroelectric patterns to be “written” without changing the surface topography. These patterns can be “read” afterthe electric field is removed [e.g., Figs. 2(e)–(f)], demonstrating that polarization information can be stored in the thin filmsand potentially encoded into rapid, energy efficient computer memory that persists even when powered-off (nonvolatile ran-dom access memory).

The VFM2-HV Variable Field Module74 allows simultaneous MFM and PFM under high tip-sample voltage bias and canprovide direct evidence of magnetoelectric coupling by allowing simultaneous piezoelectric force microscopy and MFM tolocally image the coupled piezoelectric–magnetic switching. The VFM2-HV can apply static magnetic fields up to �0.8 T(~1 G resolution), parallel to the sample plane, therefore allows observation of the effects of applying high magnetic fields in-plane while performing PFM experiments. It uses a unique design incorporating rare-earth magnets to produce the magneticfield. Because of this, there is no heating or drift as the field changes, providing low-noise, high-precision scanning probemeasurements. Rotation of the powerful rare-earth magnet allows the maximum magnetic field intensity at the sample to bevaried (maximum field when rotated at 90°, field is turned off at 0° or 180°). Once a field value is reached, the motor isturned off and the field is maintained without residual heat, thermal drift, or mechanical vibration.

Sidebar C. (Continued)


ramesh

Cross-Out

ramesh

Inserted Text

Figs. 8(a)–(d)

ramesh

Cross-Out

ramesh

Inserted Text

Figs. 8(e)–(f)

ramesh

Cross-Out

ramesh

Inserted Text

2

ramesh

Cross-Out

ramesh

Cross-Out

ramesh

Inserted Text

a

using a FEI DualBeam Helios NanoLab 600i Focussed IonBeam (FIB) (final thinning at 93 pA 30 kV, final polish 2 kV28 pA). Samples were gold coated to prevent charging.Microstructural analysis was performed on the B6TFMOfilms using high-resolution transmission electron microscopy(HR-TEM, Jeol 2100 transmission electron microscope;200 kV; double tilt holder) in conjunction with selected-areaelectron diffraction (SAED). Note that normally ~10% errorshould be accounted for when calculating d-spacings fromSAED due to electron optics of the instrumentation. Elemen-tal mapping using EDX (Oxford X-Max 80 detector andInca analysis software) over larger sample areas (3.99 lm2–1 mm2, medium spot size, X-ray generation area 10–30 nmin diameter, 200 nm thickness) was performed using the HR-SEM and scanning transmission electron microscopy(STEM) mode at the FEI Helios Nanolab. The EDX mea-surements were averaged over six to seven single measure-ments each containing up to 1100 frames. Hence, theprecision of the elemental ratios is based on the standarddeviation21 calculated, which ranges from 0.3% for Bi (heavyelement) to 2.4% for the Mn and Fe ratios. This is comparablewith literature under similar conditions.63

(4) Piezoresponse Force MicroscopyElectromechanical responses of the films were measured byPFM (see Sidebar C) using an Asylum Research MFP-3DTM

AFM in contact mode, equipped with a HVA220 Amplifier(Asylum Research) for PFM using Single Frequency (drivefrequency of 20 kHz) and Dual AC Resonance TrackingPiezoresponse Force Microscopy (DART-PFM)64 modes.Vertical hysteresis loop measurements were obtained byswitching spectroscopy PFM (SS-PFM)65,66 using a triangu-lar step waveform [comprised of pulse DC bias voltage(60–88 V) and an AC signal (5.5 V)]. The waveform wascycled twice at a frequency of 0.3 Hz with 68 AC steps perwaveform. PFM imaging under a magnetic field was per-

formed using the VFM2-HV [Asylum Research High VoltageVariable Field Module (Version 2)], where rotation of therare-earth magnet allowed the maximum magnetic fieldintensity at the sample to be varied. In this module, the mag-netic field lines are parallel along the short axis of the PFMcantilever. All images were conducted at a scan angle of 90°,where motion along this axis is parallel to the magnetic fieldlines and repeat measurements (~10 times) were performedto ensure imaging artifacts were not present. OlympusAC240TM Electrilevers, Ti/Pt-coated silicon cantilevers (Alreflex coated, 70 kHz resonant frequency, ~320 kHz contactresonance frequency) were used for PFM imaging. Note thatthese probes are conducting, but nonmagnetic. The InverseOptical Lever Sensitivity of the cantilevers was calibratedaccording to the MFP-3D Procedural Operation “Manua-lette”, the system inherent background was determined usinga nonpiezoelectric silicon wafer and the PFM was thencalibrated using a-quartz as a reference sample.

(5) SQUID MagnetometryRoom-temperature magnetic measurements of the thin filmswere carried out using a Quantum Design SQUID magne-tometer (Model- MPMS XL5; Quantum Design) under amaximum applied field of 5 T (see Sidebar D for a descrip-tion of the interpretation of SQUID magnetometerresponses). Before measuring any sample, an appropriatedemagnetization protocol was followed to remove any rema-nent magnetization. The field was set to zero field from ahigher field through an oscillating field sequence. The super-conducting coil of the SQUID was warmed up to roomtemperature to remove any trapped flux. The diamagneticcontribution from the quartz substrate was subtractedfrom the magnetic hysteresis loop measurements. The filmweight was estimated from sample area and film thicknessmeasurements, combined with the X-ray density to be1.02 9 10�4 g.62

(a) (b) (c) (d) (e)

(f) (g) (h) (i) (j)

Fig. 1. Representative (a) topography; (b) lateral single frequency PFM amplitude; (c) lateral single frequency PFM phase; (d) vertical singlefrequency PFM amplitude; and (e) vertical single frequency PFM phase images of B7TFO thin films on c-plane sapphire; and (f) topography; (g)lateral single frequency PFM amplitude; (h) lateral single frequency PFM phase; (i) vertical single frequency PFM amplitude; and (j) verticalsingle frequency PFM phase images of B6TFMO thin films on c-plane sapphire.

(a) (b) (c)

Fig. 2. Representative (a) topography; (b) vertical DART-PFM amplitude; and (c) vertical DART-PFM phase images of B6TFMO thin filmson c-plane sapphire.


Sidebar D. Oxide Magnetism and SQUID Magnetometry

Plots of the magnetic susceptibility of typical paramagnetic, ferromagnetic, and antiferromagnetic materials as a function oftemperature are strongly indicative of these different types ofmagnetic structures, and examples are shown inFig. D.1.Themagneticsusceptibility (v) is inversely proportional to temperature (T) for paramagneticmaterials [Fig. D.1(a)]. Ferromagnetic materials areparamagnetic above the Curie temperature (TC). However, below TC the relation between v and T is complex [green region inFig. D.1(b)]. In case of antiferromagnetic materials below the N�eel temperature (TN), which is recognized by a sharp peak in thesusceptibility curve [see Fig. D.1(c)], the spontaneous magnetic ordering opposes the natural tendency of the magnetic moments toalign parallel to the external applied field, which leads to a decrease in susceptibility as the temperature is reduced further.

(a) (b) (c)

Fig. D.1. Change in susceptibility (v) versus temperature (T) for different types of magnetic materials.

Measurement of magnetic susceptibilities with high sensitivity is possible with a Superconducting Quantum InterferenceDevices (SQUID) magnetometer. An equivalent SQUID circuit75 is shown in Fig. D.2. A closed superconducting loop, whichconsists of a pickup coil and input coil, is shown in the figure. A persistent current is generated in the superconducting loop dueto the magnetic flux field measured at the pickup coil. The two Josephson tunnel junctions in SQUID which are shunted withresistors eliminate hysteresis in tunnel junction current–voltage characteristics. The output voltage across the Josephson junctionappears due to the magnetic signal input at the pickup coil. The output voltage gradually changes due to the change in magneticfield for the quantum interference in Josephson junctions. Later, this output signal is refined through modulation coil andconverted into a magnetic moment.

Fig. D.2. Equivalent circuit of the SQUID magnetometer.

Different measurement protocols can be used. The zero field cooled (ZFC)–field cooled (FC) curve is one widely used protocolto investigate the magnetic properties as a function of temperature and bias field, and generally follows a particular procedureexplained below:

1. The sample is cooled down from a high starting temperature (normally room temperature) to low temperature (2–5 K)without applying any magnetic field.

2. A small magnetic field (chosen from the linear region of hysteresis loop of respective sample) is applied and maintainedand response moment from the sample is measured, while the temperature is swept up to the starting point and downagain with the same cooling rate and data accusing rate.

3. Finally, the field is removed and magnetization is measured with increasing temperature from lowest temperature to thehighest temperature.

Hence, the final curve is made of three different parts: ZFC, FC, and remanence curves [Fig. D.3(a)]. The point at which splittingbetween ZFC–FC curves occurs gives the transition temperatures, for example, N�eel (TN) or blocking (TB) temperatures, belowwhich, the material is antiferromagnetic or ferromagnetic, and will give small or large positive remanences, respectively. Anotherimportant measurement is the (M–H) measurement, where the magnetization (M) of the sample is measured as a function of appliedmagnetic field (H) [Fig. D.3(b)]. For paramagnetic and diamagnetic materials, the curves obtained are straight lines through the ori-gin with positive and negative slopes, respectively. For other types of materials (ferromagnetic/ferrimagnetic/antiferromagnetic/superparamagnetic, etc.) theM–Hmeasurement is nonlinear, for example, giving the schematic hysteresis loop in Fig. D.3(b).


III. Results and Discussion

(1) Structural and Compositional AnalysisChemical solution deposition processes were used to makeBiFeO3 and BiMnxFeyO3-doped Bi4Ti3O12 Aurivillius thinfilms on sapphire substrates. Two sols were formulated, withthe intention of producing materials with m = 6. Thesepossessed Ti:Fe:Mn ratios of 1:1:0 and 1:0.7:0.3, respectively.Considerable (17.5 mol%) excesses of Bi were used in bothcases to suppress pyrochlore formation.41 XRD [Fig. 3(a)]demonstrated that both sols produced Aurivillius phase typethin films, with the film lattice parameters being a = 5.468,b = 5.472, and c = 57.554 �A for the first sol, evidently pro-ducing an m = 6 film—Bi7Ti3Fe3O21 (B7TFO). The secondgave a = 5.497, b = 5.415, and c = 49.280 �A, clear evidencefor an m = 5 film, from an average sol composition ofBi6Ti2.5Fe1.75Mn0.75O18 (B6TFMO). (Note that as a consider-able excess of Bi is included in the sol, the minor changes inexcess given by producing an m = 5, rather than an m = 6film are not significant). All diffraction lines could be indexedeither to the relevant Aurivillius phase patterns, substratepeaks, or CuKb lines from the strongest main phase peaks.The thin films are predominantly c-axis oriented with aLotgering factors,61 f, of 0.977 for B7TFO and 0.997 for

B6TFMO. The five-layered structure for B6TFMO was con-firmed by HR-TEM [Fig. 3(b)] and electron diffraction[Fig. 3(c)]. Note that there are no detectable lines from spinel(or indeed any other) minor phases visible in the XRD pat-terns of the films—the strongest (311) spinel reflectionswould be expected at 2h = 35.4°. However, the noise level inany XRD scan places a limit on the detectability on suchminor phases and the method is intrinsically unable to detecttrace levels (typically 1–3 vol%, depending on the relativecompositions of secondary and parent phases) of stronglymagnetic secondary phases which may affect the overall mag-netization of the sample.54 Clearly, detailed microstructuralassessment is required for any sample of this type to excludethis as a possibility, and ideally to place a confidence levelon that exclusion.

High-resolution scanning electron microscopy images(Fig. 4) reveal the characteristic platelike grain morphologiesexpected from Aurivillius phase materials. MultipleHR-SEM-EDX surface scans (areas ranging from 900 lm2

to 1 mm2) showed an average film composition of Bi6Ti2.8-Fe1.52Mn0.68O18, which is slightly deficient in Fe and Mn rel-ative to the sol. Neither a 2 h HR-SEM-EDX area scan of a26 lm 9 22.6 lm area (120 lm3 volume) nor a STEM-EDXexamination of a 30 lm long cross section of thin film(1.2 lm3 volume) produced any evidence of Fe-rich regionsthat might indicate possible evidence of low-level minorphases. However, a 72 h-long HR-SEM-EDX data collectionover a 1600 lm2 area, followed by subtraction of the BiLafrom the FeKa and MnKa signals produced maps whichshowed areas of excess Fe and Mn for the B7TFO andB6TFMO films [Figs. 5(a)–(c)]. These maps showed extre-mely small amounts (~0.01 vol%) of FeOx oxide inclusionsin B7TFO and slightly larger amounts (~0.1 vol%) of aFexMnyOz phase in the B6TFMO with Mn:Fe ratio of 1.13:1and a size of ~350 nm. HR-STEM-EDX examination of theFexMnyOz inclusions demonstrated a composition ofMn0.53Fe0.47O. HR-TEM/SAED [Fig. 5(e)] indicates acubic structure with a lattice parameter of 4.4 �A, that closelycorresponds to that of a rock salt structure.76 Magnetite(Fe3O4) and Jacobsite (MnFe2O4) structures can be excludedas the measured Mn:Fe ratio of 1.13:1 does not fit the com-positions of these phases. In addition, the space group Fd-3m(227) for Magnetite and Jacobsite does not fit the electron-diffraction pattern obtained for the inclusions. The bixbyitephase (Mn2�2xFe2xO3 where x = 0.4–0.6) can also be dis-counted as the Ia-3 (206) spacegroup also demonstrates adifferent electron-diffraction pattern from that obtained.Neither the angles of the reflections nor the lattice parame-

(a) (b)(c)

Fig. 3. (a) XRD patterns from B7TFO and B6TFMO thin films,(b) HR-TEM image and (c) electron diffraction pattern of B6TFMO.Note that log(intensity) versus 2h is used here in the XRD pattern tomake-clear the weakest peaks relative to the strongest.

(a) (b)

Fig. D.3. Typical zero field cooled (ZFC)–field cooled (FC)–Remanence curve (a) and magnetic hysteresis (MH) loop (b) measured in SQUIDmagnetometer.

Sidebar D. (Continued)


ters of Fd-3m (227)/Ia-3 (206) fit in any direction to theelectron-diffraction pattern of the inclusions. A compositionclosely related to that of the inclusions, Mn0.56Fe0.44O(Mangano W€ustite), has a cubic lattice with the space groupFm-3m (225).76,77 It should be noted that this rock salt–struc-tured composition is nonferromagnetic, and antiferromag-netic at low temperatures with a N�eel point of ~150 K.77 Theelectron-diffraction pattern captured for the Mn0.53Fe0.47Oinclusions fits simulations of this space group along the (110)direction perfectly78 assuming micro/nanotwinning79,80

causes reflections at half positions. This assumption isstrengthened by the observations in TEM dark field mode(Fig. 6). In this technique, the unscattered or zero orderbeam is excluded. Only electrons diffracted on crystal planes

contribute to the formed image. Figure 6 shows clearly thatdifferent diffraction spots in the summary SAED [Fig. 6(a)]stem from different micro/nanocrystals. In particular, threeregions almost parallel to the substrate surface are discerniblein different distances. Viewed from the substrate, the bottompart is most visible in Fig. 6(a), the top part in Fig. 6(e), andthe interface in-between in Fig. 6(f). A second random twin-ning in horizontal direction is also observable within theinclusion which is illuminated by the complimentary high-lighting of different crystals in Figs. 6(d) and (g).

Also visible in these surface HR-SEM-EDX maps werelarger areas, similar in shape to the Aurivillius grains, wherethe Fe content slightly exceeded the surrounding grains.Detailed cross-sectional HR-TEM/SAED and HR-STEM-

(a) (b) (c)

(d) (e) (f) (g)

Fig. 5. Compositional maps produced by extended period (72 h) data collections from a 1600 lm2 sample area, followed by subtraction of theBiLa from the FeKa and MnKa signals. These show (a) regions of excess Fe in B7TFO; (b) regions of excess Fe in B6TFMO; and (c) regions ofexcess Mn in B6TFMO. Note the one-to-one correspondence between the small, bright (numbered) regions showing Fe in (b) and Mn excesses in(c). Note also the larger pale areas in (b) corresponding to areas where the Fe content slightly exceeds the surrounding grains. (d) Cross-sectionalHR-TEM image and (e) diffraction pattern taken from a single Mn0.53Fe0.47O inclusion; (f) Cross-sectional HR-TEM image and (g) diffractionpattern taken from a single higher-Fe content Aurivillius grain within B6TFMO.

(a) (b)

Fig. 4. Representative (a) AFM image of B7TFO thin films and (b) HR-SEM image of B6TFMO thin films on c-plane sapphire.


EDX examination of these grains [Figs. 5(f) and (g)] showedthat they were m = 5 Aurivillius-structured grains possessinga higher Fe content than the average of the film compositiondetermined by surface area HR-SEM-EDX. A compositionalsurvey by cross-section HR-STEM-EDX of 55 individualAurivillius grains within the B6TFMO sample, followed bynormalization of the Ti:Fe:Mn ratios to give five octahedralB-site cations, showed a strong relationship between the Feand Ti contents in these grains, with the Mn content beingless strongly dependent on the Ti B-site composition

[Fig. 7(a)]. The range of grain compositions (ranging fromTi:Fe:Mn = 3.38:1.14:0.48–2.18:2.16:0.66) is much greaterthan the observed precision of the EDX measurement tech-nique (0.3%–2.4%) and spans both the average compositionsdetermined from the sol and the area EDX scan notedabove. This graph helps to explain why a sol which wassetup to deliver an m = 6 structure could produce an m = 5structure without large amounts of second phase appearingin the film. Mn is well-known for taking variable valencyfrom three to four in perovskite oxide materials,81 and per-

(b) (c) (d)

(e)(a)

(f)

(g) (h) (i)

Fig. 6. Dark field TEM analysis of a single Mn0.53Fe0.47O inclusion. These show (a) SAED pattern (b)–(i) dark field images taken from thespots indicated in (a), crystal direction indicated in each image. The sapphire substrate is on the left side of the images.

Fig. 7. (a) Plot of the normalized B-site composition of Fe and Mn versus that for Ti, as determined by cross-section HR-STEM-EDX from 55Aurivillius grains within the B6TFMO sample; (b) Plot of the total B-site Mn composition and B-site Mn4+ composition, calculated to maintaincharge balance, versus the B-site Fe composition.


mitting nonstoichiometry to exist. As the Fe3+ replacesTi4+, it is suggested that Mn3+ is progressively oxidized toMn4+, to maintain charge balance. In this case, the propor-tion converted into the higher valence state is easily calcu-lated and is plotted in Fig. 7(b). It is interesting to note that,while the Fe and Ti B-site compositions are strongly interde-pendent, the Mn composition within these grains does notvary systematically with either, averaging around 0.63, andthe maximum Fe content of the B-site, at 1.77, occurs whenall of the Mn3+ is converted into Mn4+. It is believed thatthis has important consequences for the magnetic propertiesof the grains, as will be discussed below.

(2) Ferroelectric CharacterizationPiezoresponse force microscopy (PFM)68,73,82,83,84 was usedto investigate the local electromechanical properties of thefilms at room temperature. Single-frequency PFM imagingexperiments demonstrate that both sets of the films arepiezoelectric at room temperature, with higher piezorespons-es in the lateral direction (25 pm/V for B7TFO & 19 pm/Vfor B6TFMO) compared with those measured in the verticaldirection (3 pm/V for B7TFO & 8 pm/V for B6TFMO)(Fig. 1). As these films are preferentially c-axis oriented,most of the grains have their crystallographic a-axis lying inthe lateral plane of the film. The major polarization vectorfor Aurivillius phase materials is along the a-axis85 and as aresult, the single-frequency lateral PFM images demonstrategreater piezoresponses than the single-frequency vertical

PFM images, which given the low piezoresponses, topographycross talk is likely also to be contributing to the imagesobtained. As the films are comprised of some a-axis orientedgrains as well as c-axis oriented grains [Fig. 3(a)], a verticalPFM response arises from a-axis oriented grains which aretilted out-of-plane and therefore are accessible to probingby vertical PFM. There is a minor polarization along thec-axis for Aurivillius phases with odd numbers of perovskitelayers. Therefore, the difference between lateral and verticalPFM responses for the B6TFMO (m = 5) films is less thanthat of the B7TFO (m = 6) films. We employed the DART-PFM mode (Fig. 2) to intensify the imaging of the weakerout-of-plane component and reduce effects of topographycross talk. Investigations of the local room-temperatureferroelectric switching behavior in the films by VerticalDART-PFM switching spectroscopy measurements in theabsence of an applied DC bias are presented for B7TFO inFigs. 8(a) and (b), and for B6TFMO in Figs. 8(c) and (d),where 180° ferroelectric switching is clearly demonstratedfor both types of film. Ferroelectric polarization reversalover areas of the B6TFMO film could be achieved by apply-ing an applied voltage of up to 70 V vertically to an area ofthe samples via the PFM tip (in a “write” step). Given thatthe insulating substrate is 400 lm thick, this would corre-spond to an average field of 0.17 V/lm across the thin film,although the effects of nonuniform field spreading from thetip imply that the field within the film will be considerablyhigher than this. The written areas could be detected by asubsequent PFM scan (“read” step), as is demonstrated in

(a) (b)

(c) (d)

(e) (f)

Fig. 8. Vertical DART-PFM switching spectroscopy (a) phase and (b) piezoresponse loops of B7TFO and (c) phase and (d) piezoresponse loopsof B6TFMO thin films in the absence of an applied DC bias. Images of B6TFMO on c-plane sapphire: (e) out-of-plane PFM phase and (f) out-of-plane PFM amplitude after PFM lithography with an applied electric field of 70 V.


Figs. 8(e) and (f). Tests conducted over an 8 h period dem-onstrated that the films retained polarization for this periodof time.

(3) SQUID Magnetometry InvestigationsThe magnetic behavior of the B6TFMO and B7TFO thinfilms were investigated using SQUID magnetometry from 2to 300 K. A strong ferromagnetic signature was observed forthe B6TFMO samples, as is evident from SQUID magnetiza-tion measurements as a function of magnetic field [Fig. 9(a)]and temperature [Fig. 9(b)], whereas a clear antiferromag-netic behavior was observed in case of B7TFO [Fig. 9(d)].The average sample thickness was calculated after measuringat ~200 cross-section points across the sample piece. In addi-tion, we take in to account that 2.44% area of the substratewas not covered by the Aurivillius phase thin film due topore formation, [inset Fig. 9(a)]. Accordingly, the saturationmagnetization (MS) measured for Aurivillius phase B6TFMOis calculated to be 0.74 emu/gm with remanent magnetization(Mr) of 0.022 emu/g (0.18 emu/cm3) and coercivity (l0Hc) of7 mT at 300 K. The coercivity and remanence increase grad-ually as temperature decreases table within Fig. 9(c). ZFC–FC measurements [Fig. 9(b)] were performed to investigate

the magnetization behavior of the B6TFMO sample as afunction of temperature. A relatively low field of 10 mT wasapplied for these measurements. The clear split between theZFC–FC curves demonstrates the ferromagnetic nature ofthe sample as otherwise the ZFC–FC lines would normallycoincide58 with each other. The nonsubstituted compound,B7TFO (Bi7Ti3Fe3O21), demonstrates an antiferromagneticTN at 190 K and a magnetic transition to weak ferromagne-tism below 35 K [Fig. 9(d)] in accordance with earlier reports.50

The antiferromagnetic secondary phase Mn0.53Fe0.47O observedin B6TFMO by HR-SEM is reported to have a TN at~150 K.77 However, there is no 150 K magnetic transitiondetected in the measurement magnetization versus tempera-ture measurement (MT) for B6TFMO. Rather, the ZFC–FCcurves are well separated below 350 K which strengthens theevidence for B6TFMO being ferromagnetic, with a TC

greater than 350 K. Further it is observed that the FC curveof B6TFMO drops down at 190 K and again increases shar-ply below 35 K which is similar in nature with the MTbehavior of B7TFO [Fig. 9(d)]. This nonmonotonic behaviorof the FC curve for B6TFMO can be explained as follows. Itis most likely that a significant part of the parent B6TFMOphase has been modified to become ferromagnetic, with theremainder being antiferromagnetic in the same way as

(a) (b)

(c)(d)

Fig. 9. (a) M versus H and (b) M versus T measurements (ZFC & FC) for B6TFMO on sapphire. Inset image in (a) shows SEM image ofsample with 2.44% pores (red arrows) (c) Magnetic parameters for B6TFMO thin film on sapphire; (d) M versus T measurement for B7TFO.

(a) (b)

Fig. 10. The magnetic properties of B7TFO phase were investigated. (a) Magnetic hysteresis measured at 2 K and the inset shows hysteresisafter direct subtraction of the diamagnetic substrate contribution; (b) Zoomed hysteresis loop of B6TFMO measured at different temperatures.


unmodified B6TFO.57,59,86,87 The variation in magnetizationof B6TFMO as a function of temperature [Fig. 9(b)] is thusnonmonotonic in nature due to the influence of a proportionof grains behaving as would be expected from the unmodi-fied parent antiferromagnetic B6TFO phase.86,87 In the mag-netic hysteresis loop measurement (Fig. 10), the B6TFMOfilm shows a saturation magnetization (2.19 emu/gm at 2 Ktemperature-5 T field) which is substantially higher than theunsubstituted B7TFO phase (0.2 emu/gm at 5 K tempera-ture-5 T field). For antiferromagnetic materials, the rema-nence magnetization is naturally near zero (ideally zero) asthe opposite spins cancels out each other. The observation ofboth high remanence and an increase in remanence and coer-civity of B6TFMO with decrease in temperature in B6TFMOstrongly supports ferromagnetism in this material.

(4) Evidence for Magnetoelectric Multiferroic Coupling inB6TFMODirect evidence of magnetoelectric multiferroic coupling wassought by performing PFM under a variable magnetic fieldto locally image any coupled piezoelectric–magnetic switch-ing. Single-frequency lateral PFM was performed on the

B6TFMO sample, as shown in Figs. 11(c) and (d). On appli-cation of an in-plane magnetic field of +250 mT, which isabove that of the coercive field of the B6TFMO sample[Fig. 9(c)], two situations, (i) the emergence of in-planepiezoelectric domains (blue and green circles) and (ii) piezo-electric domain switching (red and orange circles) wereobserved by means of in-plane PFM imaging [Figs. 11(e) and(f)]. Vertical PFM in single-frequency mode was also per-formed in the same area; however, the vertical response ismuch lower than the in-plane response and mostly at noiselevel. This is because of the fundamental fact that most ofthe polarization (and hence piezoresponse) in these highlyoriented Aurivillius grains lies in the plane of the sample.Occasionally some vertical response is observed, which is dueto the presence of some grains which have a significant out-of-plane orientation. There are no obvious vertical PFMdomains in the regions where domain evolution was observedin the lateral PFM scans (green and blue circles). As the tor-sional twist of the cantilever used for PFM scanning is paral-lel to that of the magnetic field lines, piezoelectric domainformation and polarization switching is solely induced by theexternal magnetic field and a coupling of the electric andmagnetic order parameters within the B6TFMO grains.

(a) (b)

(c) (d)

(e) (f)

Fig. 11. Representative images of B6TFMO thin films: (a) topography; (c) lateral PFM amplitude; and (e) lateral PFM phase under 0 mT(�1.9 Oe) H field and (b) topography; (d) lateral PFM; and (f) lateral PFM phase under +250 mT (+2501 Oe) H field.


ramesh

Cross-Out

ramesh

Inserted Text

e

Polarization reversal due to the action of the imaging ac fieldis unlikely to occur at the imaging frequency.88 Inductionof piezoelectric polarization reversal by application of amagnetic field in the positive direction (+250 mT) was alsoobserved by vertical PFM imaging of the out-of-plane piez-oresponses. Vertical DART-PFM imaging [Figs. 12(e) and(f)] indicated (via piezoelectric signal-phase inversion) areasof local ferroelectric domain switching/polarization reversaldue to the magnetic field. When a magnetic field of 250 mTwas applied in the opposite (negative) direction, additionalareas exhibiting polarization inversion were obtained[Figs. 12(h) and (i)]. With the data obtained, it is not possi-ble to distinguish between 180° and 90° ferroelectric polariza-tion switching under the application of the magnetic field.Magnetoelectric switching has been observed in both lateral(Fig. 11) and vertical (Fig. 12) PFM experiments. Given thatthe orientations of the specific grains exhibiting the effectsare indeterminate, the details of the relative crystallographicorientation of ferroelectric and magnetic polarization direc-tions will be the subject of further work. Ideally, we needsingle-crystal epitaxial films to do that, which will be grown byatomic vapor deposition. The switching regions were approx-imately 250 nm in size, clearly related to the Aurivilliusgrains, of which they were a small fraction of the total num-ber (average change in polarization was 4% for Fig. 11 and7% for Fig. 1289) and widely dispersed throughout the film.The mechanism by which the coupling occurs is not obvi-ous from these experiments. Evans et al.36 put forward a

strain-mediated coupling mechanism for their observations offerroelectric domain switching in [Pb(Zr0.53Ti0.47)O3]0.6–[Pb(Fe0.5Ta0.5)O3]0.4 on application of a magnetic field. Thisdirect observation of the switching and formation of a ferro-electric polarization induced by a change in magnetic fieldwithin a single phase is significant as it provides strongevidence that within the B6TFMO thin film sample thereexists Aurivillius phases at the nanoscale, where ferromag-netic–ferroelectric order parameters are coupled within a sin-gle phase and are multiferroic at room temperature. It isproposed that the dispersion of the grains showing the mul-tiferroic switching reflects the fact that these are the grainswith the highest levels of Fe/Mn present, probably with acomposition at around Bi6Ti2.6Fe1.77Mn0.63O18, for which themajority of the Mn will be present as Mn4+. This conclusionis supported by the magnetic susceptibility measurements,where a proportion of the film is apparently ferromagneticand a proportion antiferromagnetic, as in unmodifiedB6TFO.

(5) Statistical Treatment of Microstructural AnalysisThe appearance of ferromagnetism in our B6TFMO sampleis intriguing. As described above, we have conducted extre-mely detailed microstructural analysis and have seen verysmall amounts (0.1%) of a Mn0.53Fe0.47O phase having therock salt structure, which is antiferromagnetic <150 K, buthave detected no ferromagnetic phases. However, it is

(a)

(d)

(g) (h) (i)

(e) (f)

(b) (c)

Fig. 12. Representative images of B6TFMO thin films: (a) topography; (b) vertical PFM amplitude; and (c) vertical PFM phase under 0 mT(�0.9 Oe) H field; (d) topography; (e) vertical PFM amplitude; and (f) vertical PFM phase under +250 mT (+2501 Oe) H field; and (g)topography; (h) vertical PFM amplitude; and (i) vertical PFM phase under �250 mT (�2501 Oe) H field.


impossible to be 100% certain that there are absolutely noferromagnetic spinel grains [e.g., Fe3O4 (Mr = 20 emu/g,Ms = 90 emu/g)90; MnFe2O4 (Mr = 18 emu/g, Ms = 80 emu/g)91] in the sample, which just did not happen to be seen inthe microstructural surveys. To check, if the ferromagneticresponse might be due to magnetic second (or impurity)phase inclusions, we have performed a statistical analysis ofthe upper bound of the density of magnetic inclusions, suchas Fe3O4 spinel particles, that might be expected in thesample.

A number of measurements were performed to search forpossible inclusions in the sample, and for each of them thescanned volume and the minimal detectable particle size wasrecorded. This data is summarized in Table I. The EDX sur-face scans have been recorded in 1024 9 886 pixels per map.For measurement 1 (area of 10 000 lm2), the detection limitfor inclusions is 1 9 1 pixel. For smaller areas, the smallestdetectable size increases to 2 9 2 and 3 9 3 pixels becausethe EDX interaction volume is mainly dependent on thebeam energy, and hence, stays constant. Its size is indepen-dent of how finely the area is rastered. Recording an averageof 4.4 9 109 counts over 72 h improved the signal-to-noiseratio by a factor of 67 compared with a standard scan withan average of 1 9 106 total counts. This was one of the twokey factors in the ability to record inclusions down to0.01 vol%. The other key factor was the subtraction of thebackground Fe that was bound to the main phase, hencemaking the excess Fe visible [Figs. 5(a)–(c)]. On the basis ofthe measurements performed, we can exclude the possibilitythat grains of diameters between 5 nm and about 5 lm areresponsible for the observed remanence.

Let us assume that in measurement number k a samplevolume Vk is scanned for inclusions with a diameter largerthan some minimal detectable diameter dk If no inclusionsare found in this measurement, then we can conclude thatthe density of inclusions qk larger than dk has an upperbound given by qk < 5.3/Vk with a confidence level of99.5%. This then allows us to put an upper limit on the vol-ume fraction /k of expected magnetic inclusions in the sam-ple within a diameter interval [dk, dk�1], where dk�1 is thesmallest diameter detectable in a previous measurement withless resolution. We calculate /k under the assumption thatinclusions are spherical (cylindrical) for diameters less thanthe sample thickness ds = 0.2 lm and the corresponding val-ues are given in Table I. On the basis of this analysis, we areable to calculate an upper bound for the contributions ofimpurity inclusions to the remanence magnetization Mr. Aswe do not observe such magnetic inclusions, and thereforedo not know their potential chemical properties, we assumeas a worst case scenario that their magnetic properties arecomparable to Fe3O4, which is the strongest known candi-date impurity in the material system under consideration. Wethus assume that inclusions with diameters <5 nm are notferromagnetic, inclusions of diameters between 5 and 20 nmcontribute <10 emu/g92 to the remanence of the material andall other inclusions are assumed to contribute ca 20 emu/g,comparable with the bulk value for Fe3O4.

93 Using theseupper estimates, we calculate an upper limit for the contribu-tion from undetected impurity inclusions to the remanence inour sample. This is done in the column denoted by Mr,k in

Table I. The maximum Mr,k is Mr,6 � 2.8 memu/g. Fromthis analysis, we therefore finally conclude that the contribu-tion to the remanence from unobserved inclusions in the sizerange below 5 lm is <2.8 memu/g with a confidence levelbetter than 99.5%. The details of the analysis and the mathe-matics behind this analysis will be the subject of a separatepublication. It is particularly interesting that the B7TFO filmshowed no room-temperature ferromagnetic response, despitebeing made in a similar way to the B6TFMO film, and pos-sessing a slightly larger number of magnetic cations/unit vol-ume (3.5 Fe ions/nm3 for B7TFO versus 3.3 Fe/Mn ions/nm3

for a B6TFMO grain with a composition Bi6Ti2.6-Fe1.77Mn0.63O18). The Mn ions are clearly making a majorcontribution to the ferromagnetic response in this material,and the fact that the multiferroic switchability is confined toa small number of grains strongly suggests that it is thosegrains with the highest Fe/Mn content that are responsiblefor the effect, and for these the manganese is probably pres-ent as Mn4+, which possesses three unpaired spins in its t2gd-orbitals, with no electrons in its eg orbitals when coordi-nated in the perovskite geometry. In contrast, high-spinFe3+ possesses two unpaired electrons in its eg orbitals. Theparent compound, B6TFO has been reported59,86,87 to beantiferromagnetic with a magnetic transition to weak ferro-magnetism below 65 K.

The mechanism involving ligand orbitals to facilitate cou-pling between metal electrons is referred to as superex-change. According to the Goodenough–Kanamori rule,94

superexchange interactions are antiferromagnetic where vir-tual electron transfer is between overlapping orbitals that areeach half-filled. The adjacent metal ions couple with theirspins antiparallel, with equal numbers of the two arrange-ments so that there is no resultant magnetization in theabsence of a magnetic field. On the other hand, the Gooden-ough–Kanamori rule predicts that superexchange interac-tions are strong and ferromagnetic when virtual electrontransfer is from a half-filled orbital to an empty orbital. Theelectron spins of each of the atoms couple strongly togetherto form a resultant unit cell magnetic moment in an appliedmagnetic field which remains when the external field isremoved. Hence, the Goodenough–Kanamori rule94 for su-perexchange between the half-filled eg orbitals of high-spinFe3+ via the oxygen p-orbitals in the perovskite blocks isexpected to give a strong antiferromagnetic interaction andwould explain the antiferromagnetic behavior we observed inB7TFO (and reported in the literature for B6TFO). How-ever, if we introduce Mn4+ into the structure, the eg orbitalis always empty. In this case, the Goodenough–Kanamorirule states that the superexchange interaction via the oxygenp-orbitals to an empty eg orbital should lead to ferromagne-tism. Therefore, a ferromagnetic interaction between thevacant Mn4+ eg orbital and the filled Fe3+ eg orbital94

within the Aurivillius phase structure is likely to cause theobserved ferromagnetism in the manganese-substitutedB6TFMO samples.

IV. Conclusions

In conclusion, thin film samples of the m = 6 and 5 Aurivil-lius compounds containing Fe and Fe/Mn ions were grown

Table I. Performed Volume Scans

Method k Volume Vk (lm3) Smallest diameter dk (nm) /k Mr,k (memu/g)

Surface EDX 1 2000 1000 (d0 = 5lm) 0.010% 2.1Surface EDX 2 450 350 0.006% 1.1Surface EDX 3 120 100 0.005% 1.0Surface EDX 4 28 60 0.001% 0.2Surface EDX 5 7.2 20 0.008% 1.7TEM 6 0.08 2 0.028% 2.8


on c-plane sapphire substrates by chemical solution deposi-tion displaying typical Aurivillius phase grain morphologiesand with no spinel secondary phases visible in the XRDpatterns. PFM images demonstrate that the films arepiezoelectric and ferroelectric at room temperature with themajor polarization vector in the lateral plane of the films.SQUID measurements demonstrated antiferromagnetism inB7TFO (TN = 190 K), but B6TFMO samples demonstratein-plane ferromagnetic hysteresis between 2 and 300 K. Athorough microstructural phase analysis performed on theB6TFMO thin films showed no traces of ferromagnetic inclu-sions and a statistical analysis based on the volumesinspected placed a confidence of 99.5% that the observed fer-romagnetism was not coming from unobserved ferromagneticgrains of spinel. Direct evidence for magnetic field-inducedferroelectric domain switching at the nanoscale in a single-phase magnetoelectric has been presented. The body of evi-dence reported here suggests that the higher Fe/Mn contentgrains with a composition of around Bi6Ti2.6Fe1.77Mn0.63O18

are single-phase room-temperature magnetoelectric multifer-roics. An explanation for the effect has been given basedupon the Goodenough–Kanamori rule for superexchangeand the Mn largely being present as Mn4+. As they stand,these materials have rather low ferromagnetic bulk magneti-zation values, but we have seen that the ferromagnetic com-ponent of the films is only a proportion of the whole, andwith further work they could be optimized to increase thetotal volume of ferromagnetic phase, in which case they mayfind application to a wide range of new or improved devicesand potentially meet future industry requirements in high-density memory applications. In any case, we believe theyprovide important pointers for the future development ofroom-temperature ferroelectric/ferromagnetic multiferroics.Clearly, further work is now required which will include thedirect measurement of the compositions of the multiferroical-ly switchable grains and X-ray photoelectron spectroscopy todetermine Mn oxidation states, as well as the development ofsynthetic techniques to develop thin films in which all grainspossess a composition around Bi6Ti2.6Fe1.77Mn0.63O18, forwhich it is expected that higher remanent magnetizations willbe achieved. Further work will also include the developmentof films onto back electrodes which can be used for directmagnetodielectric measurements. Part of this work was sup-ported by COST Action MP0904 [Single and multiphase fer-roics and multiferroics with restricted geometries(SIMUFER)].

Acknowledgments

The support of Science Foundation Ireland (SFI) under the FORME StrategicResearch Cluster Award number 07/SRC/I1172, Starting InvestigatorResearch Grant (09/SIRG/I1621), and SFI 09/SIRG/I1615 is gratefullyacknowledged. The authors acknowledge ICGEE (International Centre forGraduate Education in micro & nano Engineering) for funding Nitin Deepak’sPhD. The authors would like to thank Asylum Research for their complimen-tary temporary provision of High Voltage Variable Field Module. Part of thiswork was supported by COST Action MP0904 [Single and multiphase ferroicsand multiferroics with restricted geometries (SIMUFER)].

References

1Assessment of the Potential and Maturity of Selected Emerging ResearchMemory Technologies Workshop & ERD (Emerging Research Devices) /ERMWorking Group Meeting (April 6-7 2010) (2010).

2Emerging Research Materials, International Technology Roadmap for Semi-conductors. 2009 Edition, http://www.itrs.net/Links/2009ITRS/Home2009.htm.

3C. N. R. Rao, A. Sundaresan, and R. Saha, “Multiferroic and Magneto-electric Oxides: The Emerging Scenario,” J. Phys. Chem. Lett, 3 [16] 2237–46(2012).

4M. Bibes, “Nanoferronics is a Winning Combination,” Nat. Mater., 11 [5]354–7 (2012).

5L. Wang, D. Wang, Q. Cao, Y. Zheng, H. Xuan, J. Gao, and Y. Du,“Electric Control of Magnetism at Room Temperature,” Sci. Rep., 2, 223(2012).

6R. Ramesh, “Magnetoelectrics: Making Metallic Memories,” Nat. Nano-technol., 5 [11] 761–2 (2010).

7J. F. Scott, “Data Storage: Multiferroic Memories,” Nat. Mater., 6 [4]256–7 (2007).

8C. A. F. Vaz, J. Hoffman, C. H. Ahn, and R. Ramesh, “MagnetoelectricCoupling Effects in Multiferroic Complex Oxide Composite Structures,” Adv.Mater., 22 [26–27] 2900–18 (2010).

9N. A. Hill, “Why are There so Few Magnetic Ferroelectrics?” J. Phys.Chem. B, 104 [29] 6694–709 (2000).

10G. Catalan and J. F. Scott, “Physics and Applications of Bismuth Fer-rite,” Adv. Mater., 21 [24] 2463–85 (2009).

11Y.-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-Rodri-guez, A. Scholl, S. X. Wang, and R. Ramesh, “Electric-Field Control of LocalFerromagnetism Using a Magnetoelectric Multiferroic,” Nat. Mater., 7 [6]478–82 (2008).

12S. Lee, W. Ratcliff II, S.-W. Cheong, and V. Kiryukhin, “Electric FieldControl of the Magnetic State in BiFeO[sub 3] Single Crystals,” Appl. Phys.Lett., 92 [19] 192906 (2008).

13D. Lebeugle, D. Colson, A. Forget, M. Viret, A. M. Bataille, and A.Gukasov, “Electric-Field-Induced Spin Flop in BiFeO3 Single Crystals atRoom Temperature,” Phys. Rev. Lett., 100 [22] 227602 (2008).

14N. A. Lomanova, V. G. Semenov, V. V. Panchuk, and V. V. Gusarov,“Structural Changes in the Homologous Series of the Aurivillius PhasesBin + 1Fen � 3Ti3O3n + 3,” J. Alloy. Compd., 528 [0] 103–8 (2012).

15C. H. Macgillavry, G. D. Rieck, and K. Lonsdale, International Tablesfor Crystallography, Volume III, Physical and Chemical Tables, The Interna-tional Union of Crystallography by The Kynoch Press, Birmingham, UK,1968.

16J. S. Kasper and K. Lonsdale, International Tables for X-Ray Crystallogra-phy, Volume II, Mathematical Tables. The International Union of Crystallog-raphy by The Kynoch Press, Birmingham, UK, 1972.

17M. Bibes, J. E. Villegas, and A. Barth�el�emy, “Ultrathin Oxide Films andInterfaces for Electronics and Spintronics,” Adv. Phys., 60 [1] 5–84 (2011).

18A. Roy, R. Gupta, and A. Garg, “Multiferroic Memories,” Adv. inCondens. Matter Phys., 2012, 926290 (2012).

19K. E. Sickafus, J. M. Wills, and N. W. Grimes, “Structure of Spinel,”J. Am. Ceram. Soc., 82 [12] 3279–92 (1999).

20C. A. F. Vaz, “Electric Field Control of Magnetism in Multiferroic Het-erostructures,” J. Phys.: Condens. Matter, 24 [3] 333201 (2012).

21A. P. Pyatakov and A. K. Zvezdin, “Magnetoelectric and MultiferroicMedia,” Phys. Usp., 55 [6] 557–81 (2012).

22J. F. Scott, “Applications of Magnetoelectrics,” J. Mater. Chem., 22 [11]4567–74 (2012).

23L. W. Martin, S. P. Crane, Y. H. Chu, M. B. Holcomb, M. Gajek, M.Huijben, C. H. Yang, N. Balke, and R. Ramesh, “Multiferroics and Magneto-electrics: Thin Films and Nanostructures,” J. Phys.: Condens. Matter, 20,434220 (2008).

24W. Eerenstein, N. D. Mathur, and J. F. Scott, “Multiferroic and Magne-toelectric Materials,” Nature, 442 [7104] 759–65 (2006).

25M. Fiebig, “Revival of the Magnetoelectric Effect,” J. Phys. D: Appl.Phys., 38, R123–52 (2005).

26J. Ryu, S. Priya, A. V. Carazo, K. Uchino, and H.-E. Kim, “Effect of theMagnetostrictive Layer on Magnetoelectric Properties in Lead Zirconate Tita-nate/Terfenol-D Laminate Composites,” J. Am. Ceram. Soc., 84 [12] 2905–8(2001).

27F. Zavaliche, T. Zhao, H. Zheng, F. Straub, M. P. Cruz, P. L. Yang, D.Hao, and R. Ramesh, “Electrically Assisted Magnetic Recording in Multifer-roic Nanostructures,” Nano Lett., 7 [6] 1586–90 (2007).

28T. H. E. Lahtinen, K. J. A. Franke, and S. van Dijken, “Electric-FieldControl of Magnetic Domain Wall Motion and Local Magnetization Rever-sal,” Sci. Rep., 2, 258 (2012).

29V. J. Folen, G. T. Rado, and E. W. Stalder, “Anisotropy of the Magneto-electric Effect in Cr2O3,” Phys. Rev. Lett., 6 [11] 607–8 (1961).

30T. Kimura, Y. Sekio, H. Nakamura, T. Siegrist, and A. P. Ramirez, “Cup-ric Oxide as an Induced-Multiferroic with High-TC,” Nat. Mater., 7 [4] 291–4(2008).

31T. Kimura, T. Goto, H. Shintani, K. Ishizaka, T. Arima, and Y. Tokura,“Magnetic Control of Ferroelectric Polarization,” Nature, 426 [6962] 55–8(2003).

32E. Ascher, H. Rieder, H. Schmid, and H. Stossel, “Some Properties ofFerromagnetoelectric Nickel-Iodine Boracite, Ni3B7O13I,” J. Appl. Phys., 37

[3] 1404–5 (1966).33D. Higashiyama, S. Miyasaka, N. Kida, T. Arima, and Y. Tokura, “Con-

trol of the Ferroelectric Properties of DyMn2O5 by Magnetic Fields,” Phys.Rev. B, 70 [17] 174405 (2004).

34C.-W. Nan, M. I. Bichurin, S. Dong, D. Viehland, and G. Srinivasan,“Multiferroic Magnetoelectric Composites: Historical Perspective, Status, andFuture Directions,” J. Appl. Phys., 103 [3] 031101 (2008).

35Y. Kitagawa, Y. Hiraoka, T. Honda, T. Ishikura, H. Nakamura, and T.Kimura, “Low-Field Magnetoelectric Effect at Room Temperature,” Nat.Mater., 9 [10] 797–802 (2010).

36D. M. Evans, A. Schilling, A. Kumar, D. Sanchez, N. Ortega, M.Arredondo, R. S. Katiyar, J. M. Gregg, and J. F. Scott, “Magnetic Switchingof Ferroelectric Domains at Room Temperature in Multiferroic PZTFT,” Nat.Commun., 4, 1534 (2013).

37B. Aurivillius, “Mixed Bismuth Oxides with Layer Lattice II. Structure ofBi4Ti3O12,” Ark. Kemi., 1, 499–512 (1949).

38S. Patri, R. Choudhary, and B. Samantaray, “Studies of Structural,Dielectric and Impedance Properties of Bi9Fe5Ti3O27 Ceramics,” J. Electroce-ram., 20 [2] 119–26 (2008).


ramesh

Cross-Out

39P. F. Zhang, N. Deepak, L. Keeney, M. E. Pemble, and R. W. Whatmore,“The Structural and Piezoresponse Properties of c-Axis-Oriented AurivilliusPhase Bi5Ti3FeO15 Thin Films Deposited by Atomic Vapor Deposition,” Appl.Phys. Lett., 101 [11] 112903 (2012).

40S.-l. Ahn, Y. Noguchi, M. Miyayama, and T. Kudo, “Structural and Elec-trical Characterization of Bi5Ti3Fe1�xMnxO15 System,” Mater. Res. Bull., 35[6] 825–34 (2000).

41L. Keeney, C. Groh, S. Kulkarni, S. Roy, M. E. Pemble, and R. W. What-more, “Room Temperature Electromechanical and Magnetic Investigations ofFerroelectric Aurivillius Phase Bi5Ti3(FexMn1�x)O15 (x = 1 and 0.7) ChemicalSolution Deposited Thin Films,” J. Appl. Phys., 112 [2] 024101 (2012).

42N. A. Lomanova, M. I. Morozov, V. L. Ugolkov, and V. V. Gusarov,“Properties of Aurivillius Phases in the Bi4Ti3O12–BiFeO3 System,” Inorg.Mater., 42 [2] 189–95 (2006).

43D. Y. Suarez, I. M. Reaney, and W. E. Lee, “Relation between ToleranceFactor and Tc in Aurivillius Compunds,” J. Mater. Res., 16 [11] 3139–49 (2001).

44A. M. Glazer, “The Classification of Titled Octahedra in Perovskites,”Acta Crystallogr., 28, 3384–92 (1972).

45C. H. Hervoches, A. Snedden, R. Riggs, S. H. Kilcoyne, P. Manuel, andP. Lightfoot, “Structural Behavior of the Four-Layered Aurivillius-PhaseFerroelectrics SrBi4Ti4O14 and Bi5Ti3FeO15,” J. Solid State Chem., 164, 280–91 (2002).

46A. Sanson and R. W. Whatmore, “Properties of Bi4Ti3O12–(Na1/2Bi1/2)TiO3 Piezoelectric Ceramics,” Jpn. J. Appl. Phys., Part 1, 41 [11B] 7127–30(2002).

47A. Sanson and R. W. Whatmore, “Phase Diagram of the Bi4Ti3O12–BaTiO3–(Na1/2Bi1/2)TiO3 System,” J. Am. Ceram. Soc., 88 [11] 3147–53 (2005).

48P. Boullay, G. Trolliard, D. Mercurio, J. M. Perez-Mato, and L. Elcoro,“Toward a Unified Approach to the Crystal Chemistry of Aurivillius-TypeCompounds: I. The Structural Model,” J. Solid State Chem., 164 [2] 252–60(2002).

49X. Y. Mao, W. Wang, and X. B. Chen, “Electrical and Magnetic Proper-ties of Bi5FeTi3O15 Compound Prepared by Inserting BiFeO3 into Bi4Ti3O12,”Solid State Commun., 147 [5–6] 186–9 (2008).

50A. Srinivas, M. M. Kumar, S. V. Suryanarayana, and T. Bhimasankaram,“Investigation of Dielectric and Magnetic Nature of Bi7Fe3Ti3O21,” Mater.Res. Bull., 34 [6] 989–96 (1999).

51A. Srinivas, D.-W. Kim, K. S. Hong, and S. V. Suryanarayana, “Study ofMagnetic and Magnetoelectric Measurements in Bismuth Iron Titanate Cera-mic—Bi8Fe4Ti3O24,” Mater. Res. Bull., 39 [1] 55–61 (2004).

52M. A. Zurbuchen, R. S. Freitas, M. J. Wilson, P. Schiffer, M. Roeckerath,J. Schubert, M. D. Biegalski, G. H. Mehta, D. J. Comstock, J. H. Lee, Y. Jia,and D. G. Schlom, “Synthesis and Characterization of an n = 6 AurivilliusPhase Incorporating Magnetically Active Manganese, Bi7(Mn,Ti)6O21,” Appl.Phys. Lett., 91 [3] 033113 (2007).

53X. Mao, W. Wang, X. Chen, and Y. Lu, “Multiferroic Properties ofLayer-Structured Bi5Fe0.5Co0.5Ti3O15 Ceramics,” Appl. Phys. Lett., 95 [8]082901 (2009).

54L. Keeney, S. Kulkarni, N. Deepak, M. Schmidt, N. Petkov, P. F. Zhang,S. Cavill, S. Roy, M. E. Pemble, and R. W. Whatmore, “Room TemperatureFerroelectric and Magnetic Investigations and Detailed Phase Analysis ofAurivillius Phase Bi5Ti3Fe0.7Co0.3O15 Thin Films,” J. Appl. Phys., 112 [5]052010 (2012).

55M. Palizdar, T. P. Comyn, M. B. Ward, A. P. Brown, J. P. Harrington, S.Kulkarni, L. Keeney, S. Roy, M. E. Pemble, R. W. Whatmore, C. Quinne, S.H. Kilcoyne, and A. J. Bell, “Cystallographic and Magnetic Identification ofSecondary Phase in Oriented Bi5Fe0.5Co0.5Ti3O15 Ceramics,” J. Appl. Phys.,112 [7] 073919 (2012).

56F. Z. Huang, X. M. Lu, T. T. Xu, Y. Y. Liu, W. N. Su, Y. M. Jin, Y.Kan, and J. S. Zhu, “Multiferroic Properties of Co and Nd co-SubstitutedBi5Ti3FeO15 Thin Films,” Thin Solid Films, 520 [21] 6489–92 (2012).

57J. Yang, W. Tong, Z. Liu, X. B. Zhu, J. M. Dai, W. H. Song, Z. R. Yang,and Y. P. Sun, “Structural, Magnetic, and EPR Studies of the AurivilliusPhase Bi6Fe2Ti3O18 and Bi6FeCrTi3O18,” Phys. Rev. B, 86 [10] 104410 (2012).

58J. Yang, L. H. Yin, Z. Liu, X. B. Zhu, W. H. Song, J. M. Dai, Z. R. Yang,and Y. P. Sun, “Magnetic and Dielectric Properties of Aurivillius Phase Bi6Fe2-Ti3O18 and the Doped Compounds,” Appl. Phys. Lett., 101 [1] 012402 (2012).

59N. V. Prasad and G. S. Kumar, “Magnetic and Magnetoelectric Measure-ments on Rare-Earth-Substituted Five-Layered Bi6Fe2Ti3O12 Compound,” J.Magn. Magn. Mater., 213, 349–56 (2000).

60Z. Liu, J. Yang, X. W. Tang, L. H. Yin, X. B. Zhu, J. M. Dai, and Y. P.Sun, “Multiferroic Properties of Aurivillius Phase Bi6Fe2�xCoxTi3O18 ThinFilms Prepared by a Chemical Solution Deposition Route,” Appl. Phys. Lett.,101 [12] 122402 (2012).

61F. K. Lotgering, “Topotactical Reactions with Ferrimagnetic Oxides HavingHexagonal Crystal Structures—I,” J. Inorg. Nucl. Chem., 9 [2] 113–23 (1959).

62“Crystallographica – a software toolkit for crystallography,” J. Appl.Cryst., 30 418–9 (1997).

63P. Kuisma-Kursula and J. R€AIs€ANen, “Scanning Electron Microscopy-Energy Dispersive Spectrometry and Proton Induced X-ray Emission Analysesof Medieval Glass from Koroinen (Finland),” Archaeometry, 41 [1] 71–9 (1999).

64B. J. Rodriguez, C. Callahan, S. V. Kalinin, and R. Proksch, “Dual-Frequency Resonance-Tracking Atomic Force Microscopy,” Nanotechnology,18 [47] 475504 (2007).

65S. Jesse, A. P. Baddorf, and S. V. Kalinin, “Switching Spectroscopy Piez-oresponse Force Microscopy of Ferroelectric Materials,” Appl. Phys. Lett., 88[6] 062908 (2006).

66S. Jesse, H. N. Lee, and S. V. Kalinin, “Quantitative Mapping of Switch-ing Behavior in Piezoresponse Force Microscopy,” AIP, 77, 073702 (2006).

67A. Gruverman, O. Auciello, R. Ramesh, and H. Tokumoto, “ScanningForce Microscopy of Domain Structure in Ferroelectric Thin Films: Imagingand Control,” Nanotechnology, 8 [3A] A38–43 (1997).

68S. V. Kalinin, B. J. Rodriguez, S. Jesse, E. Karapetian, B. Mirman, E. A.Eliseev, and A. N. Morozovska, “Nanoscale Electromechanics of Ferroelectricand Biological Systems: A New Dimension in Scanning Probe Microscopy,”Annu. Rev. Mater. Res., 37 [1] 189–238 (2007).

69N. Balke, I. Bdikin, S. V. Kalinin, and A. L. Kholkin, “ElectromechanicalImaging and Spectroscopy of Ferroelectric and Piezoelectric Materials: Stateof the Art and Prospects for the Future,” J. Am. Ceram. Soc., 92 [8] 1629–47(2009).

70S. V. Kalinin, B. J. Rodriguez, S. Jesse, P. Maksymovych, K. Seal, M.Nikiforov, A. P. Baddorf, A. L. Kholkin, and R. Proksch, “Local Bias-Induced Phase Transitions,” Mater. Today, 11 [11] 16–27 (2008).

71E. Soergel, “Piezoresponse Force Microscopy (PFM),” J. Phys. D: Appl.Phys., 44, 464003 (2011).

72H. Chang, S. V. Kalinin, S. Yang, P. Yu, S. Bhattacharya, P. P. Wu, N.Balke, S. Jesse, L. Q. Chen, R. Ramesh, S. J. Pennycook, and A. Y. Borise-vich, “Watching Domains Grow: In Situ Studies of Polarization Switching byCombined Scanning Probe and Scanning Transmission Electron Microscopy,”J. Appl. Phys., 110 [5] 052014 (2011).

73S. V. Kalinin, A. Gruverman, and D. A. Bonnell, “Quantitative Analysisof Nanoscale Switching in SrBi2Ta2O9 Thin Films by Piezoresponse ForceMicroscopy,” Appl. Phys. Lett., 85 [5] 795–7 (2004).

74VFM2TM Variable Field Module for Magnetic AFM Applications. http://www.asylumresearch.com/Products/VFM2/VFM2.shtml (accessed November2011).

75A. Barone, Principles and Applications of Superconducting Quantum Inter-ference Devices. World Scientific Publishing Co., Singapore, 1992.

76P. K. Foster and J. E. Welch, “Metal-Oxide Solid Solutions. Partl.-Lattice-Constant and Phase Relationships in Ferrous Oxide (WUSTITE)and in Solid Solutions of Ferrous Oxide and Manganous Oxide,” Trans. Fara-day Soc., 52, 1626–35 (1956).

77D. A. Hope, A. K. Cheetham, and G. J. Long, “A Neutron Diffraction,Magnetic Susceptibility, and Mossbauer-Effect Study of the (MnxFe1�x)yOSolid Solutions,” Inorg. Chem., 21, 2804–9 (1982).

78Crystal Studio. Version 12 Quantum edition.79J. Reyes-Gasga, A. G�omez-Rodr�ıguez, X. Gao, and M. Jos�e-Yacam�an,

“On the Interpretation of the Forbidden Spots Observed in the Electron Dif-fraction Patterns of Flat Au Triangular Nanoparticles,” Ultramicroscopy, 108[9] 929–36 (2008).

80M. Retuerto, M. R. Li, Y. B. Go, A. Ignatov, M. Croft, K. V. Ramanuj-achary, R. H. Herber, I. Nowik, J. P. Hodges, W. Dachraoui, J. Hadermann,and M. Greenblatt, “High Magnetic Ordering Temperature in the PerovskitesSr4�xLaxFe3ReO12 (x = 0.0, 1.0, 2.0),” J. Solid State Chem., 194 [0] 48–58(2012).

81J. Mizusaki, N. Mori, H. Takai, Y. Yonemura, H. Minamiue, H. Tagawa,M. Dokiya, H. Inaba, K. Naraya, T. Sasamoto, and T. Hashimoto, “OxygenNonstoichiometry and Defect Equilibrium in the Perovskite-Type OxidesLa1 � xSrxMnO3 + d,” Solid State Ionics, 129 [1–4] 163–77 (2000).

82J. Varghese, S. Barth, L. Keeney, R. W. Whatmore, and J. D. Holmes,“Nanoscale Ferroelectric and Piezoelectric Properties of Sb2S3 NanowireArrays,” Nano Lett., 12 [2] 868–72 (2012).

83S. V. Kalinin, Z.-G. Ye, and A. L. Kholkin, “Preface to Special Topic:Piezoresponse Force Microscopy and Nanoscale Phenomena in Polar Materi-als,” J. Appl. Phys., 112 [5] 051901 (2012).

84S. V. Kalinin, N. Setter, and A. L. Kholkin, “Electromechanics on theNanometer Scale: Emerging Phenomena, Devices, and Applications,” MRSBull., 34 [09] 634–42 (2009).

85T. Watanabe and H. Funakubo, “Controlled Crystal Growth of Layered-Perovskite Thin Films as an Approach to Study Their Basic Properties,”J. Appl. Phys., 100 [5] 051602 (2006).

86J. Lu, L. J. Qiao, X. Q. Ma, and W. Y. Chu, “Magnetodielectric Effect ofBi6Fe2Ti3O18 Film under an Ultra-low Magnetic Field,” J. Phys.: Condens.Matter, 18, 4801–07 (2006).

87A. Srinivas, S. V. Suryanarayana, G. S. Kumar, and M. M. Kumar,“Magnetoelectric Measurements on Bi5FeTi3O15 and Bi6Fe2Ti3O18,” J. Phys.:Condens. Matter, 11, 3335–40 (1999).

88A. Gruverman and Y. Ikeda, “Characterization and Control of DomainStructure in SrBi2Ta2O9 Thin Films by Scanning Force Microscopy,” Jpn.J. Appl. Phys., Part 2, 37 [8A] L939–41 (1998).

89W. S. Rasband, Calculated Using Image J 1.44 (Image Processing andAnalysis in Java). ImageJ, U. S. National Institutes of Health, Bethesda, MD,1992–2012. http://imagej.nih.gov/ij/.

90S. Chikazumi and S. H. Charap, Physics of Magnetism. Wiley, New York,1964.

91R. S. Tebble and D. J. Craik, Magnetic Materials. Wiley, New York, 1969.92G. F. Goya, T. S. Berquo, F. C. Fonseca, and M. P. Morales, “Static and

Dynamic Magnetic Properties of Spherical Magnetite Nanoparticles,” J. Appl.Phys., 94 [5] 3520–8 (2003).

93N. N. Guan, Y. T. Wang, D. J. Sun, and J. Xu, “A Simple One-Pot Syn-thesis of Single-Crystalline Magnetite Hollow Spheres from a Single Iron Pre-cursor,” Nanotechnology, 20 [10] 105603 (2009).

94J. B. Goodenough, “Theory of the Role of Covalence in the Perovsjite-Type Manganites [La, M(II)]MnO3,” Phys. Rev., 100 [2] 564–73 (1955).

95C. F. Macrae, I. J. Bruno, J. A. Chisholm, P. R. Edgington, P. McCabe,E. Pidcock, L. Rodriguez-Monge, R. Taylor, J. van de Streek, and P. A.Wood, “Mercury CSD 2.0 - new features for the visualization and investiga-tion of crystal structures,” J. Appl. Cryst., 41 4664–70 (2008). h


Dr. Lynette Keeney received a B.Sc.in chemistry in 2001 from NationalUniversity of Ireland (NUI), Galwayand a Ph.D. in inorganic chemistry in2004 also from NUI, Galway whereshe won five University awards forher undergraduate and postgraduatestudies. Before joining Tyndall-UCCin 2008, she worked as a ResearchScientist in Charles River Laborato-ries, Preclinical Services, MontrealInc. in method development and vali-

dation of analytical research studies. Dr. Keeney has astrong background in the synthesis of complicated metaloxide thin film systems by chemical solution depositionmethods and is experienced in atomic vapour deposition andcharacterization of these systems for potential memory stor-age applications. Her research interests are the investigationof ferroelectric, ferromagnetic, and multiferroic materials andis one of the Irish representatives for the COST (EuropeanCooperation in Science and Technology) Action MP0904[Single and multiphase ferroics and multiferroics with restric-ted geometries (SIMUFER)]. Dr. Keeney is a member of theRoyal Society of Chemistry (MRSC) and is a StaffResearcher at Tyndall National Institute, University CollegeCork, Ireland.

Tuhin Maity received his M.Sc.degree in Physics in 2009 from Uni-versity of Pune, India. During hisMasters program he joined theNational Chemical Laboratory(NCL), Pune, India (May 2008–Janu-ary 2010) pursuing a research projectin nanophysics where he designed anddeveloped a system for measuring

temperature-dependent broadband dielectric spectroscopy ofmultiferroic nanoparticles. In February 2010 he moved toGermany and joined the Division of Superconductivity andMagnetism, Department of Experimental Physics in LeipzigUniversity, Leipzig, Germany as a scientific co-worker wherehe studied electric and magnetic properties of different oxidenanostructures. After gaining experience in the field of mag-netism, he then decided to undertake a course toward aPh.D. in Physics and joined Tyndall National Institute, Uni-versity College Cork, Ireland in February 2011. Presently, heis working in projects where the goal is to understand themagnetic properties of different nanostructured materials forminiaturized ICT devices. He has authored/co-authored 10journal papers to date.

Michael Schmidt obtained his M.Sc.degree in Physics in 1998 at theOtto-von-Guericke-University in Mag-deburg, Germany, working on“Catho-doluminescence microscopy onGaN and thereof based group-III-nit-rides.” From 1999 to 2003, he was aresearch assistant at the Otto-von-Gu-ericke-University in Magdeburg, at theMax Planck Institute of MicrostructurePhysics in Halle, mainly working on the

characterization of Er-doped Si nanocrystals embedded inSiOx, and at iLF e.V. in Magdeburg. After an excursion intoentrepreneurship in the field of cellular nutrition in 2003/2004,

he had been working as Research Lab Engineer with the Pho-tonic Nanostructures Group at Tyndall National Institute (for-merly NMRC) 2004–2008 and 2008–2009 with the Phononicand Photonic Nanostructures Group at the Catalan Institute ofNanotechnology (ICN) in Bellaterra, Spain. Now he holds aposition as an Electron Microscopist at Tyndall National Insti-tute where his main responsibilities include continual develop-ment, improvement and implementation of new samplepreparation and analysis methods and techniques. He is cur-rently working toward the Ph.D. degree on the application ofadvanced electron microscopy on nanostructured material atthe Tyndall National Institute, University College Cork, Cork,Ireland.

Dr. Andreas Amann received aDiploma degree in physics from theUniversity of Bonn, Germany, and aPh.D. degree in physics from theTechnical University of Berlin. Afterworking as a Visiting Researcher atthe National Institute of AppliedOptics, Florence, Italy, and as a StaffResearcher at the Tyndall NationalInstitute, Cork, Ireland, he has beenappointed Lecturer with the School ofMathematical Sciences, UniversityCollege Cork, in 2010. His research

focuses on the development and application of advancedmathematical methods to problems in physics and engineer-ing. His main interests are in the research areas of nonlineardynamics, photonics, and solid-state physics.

Nitin Deepak is a Ph.D. studentworking in the advanced materialsand surface group in TyndallNational Institute, Cork, Ireland. Hereceived his M.Sc. degree in physicsfrom the National Institute of Tech-nology, Jalandhar, India. His researchinterests include chemical vapor depo-sition synthesis of complex oxide epi-taxial thin films for ferroelectric andmultiferroic applications, probing thedomain structure of ferroelectric and

multiferroic materials using scanning probe microscopy tech-niques, investigating flexoelectric effects of the properties ofthin films, and X-ray diffraction studies of materials. Hereceived an International Centre for Graduate Education inMicro- & Nano-Engineering (ICGEE) fellowship for hisPh.D. studies at Tyndall National Institute.

Nikolay Petkov is a Staff Researcherat Tyndall National Institute, Univer-sity College Cork, where he is leadinga research area in advanced electronmicroscopy for the analysis of semi-conductor materials and devices. Hisresearch interests are in situ and cor-relative electron microscopy, materialsprocessing, and self-assembly. He hasauthored and co-authored over 60peer-review publications, several

reviews and book chapters.


Saibal Roy is a Group Head withinthe Microsystems Centre of TyndallNational Institute, Ireland. He is Sci-ence Foundation Ireland PrincipalInvestigator (SFI PI) in the Micro-power–Nanomagnetics research area.He obtained his M.Sc. degree inPhysics from the IIT and a Ph.D.working on advanced nanostructured

materials from IACS in India. Since receiving his Ph.D., hisprofessional experiences include 16 yr academic and 3 yrindustrial experience; particularly he has served both in aca-demia and industry while leading research groups consistingof substantial numbers of senior researchers. Dr. Roy’s pres-ent research interests at Tyndall include investigating howengineered nanostructures could be employed for the benefitsof micron scale devices from More than Moore to thebeyond Moore scenario. Since joining Tyndall-UCC, Dr.Roy has brought substantial (>€6 Million) government andcorporate competitive research funding. Dr. Roy has beenhonoured by the president of University College Cork forlicensing a patented technology to INTEL. Some of his pub-lished work featured widely in the media. Dr. Roy has super-vised several Ph.D./Postdocs at Tyndall-UCC. He has filed/published several international patents, book chapters and� 130 papers in Journals and peer reviewed conference pro-ceedings with a current h-index of 20.

Professer Martyn Pemble received hisB.Sc. degree in Chemistry in 1976 andhis Ph.D. in Spectroscopic Electro-chemistry in 1981, both from the Uni-versity of Southampton. He thenundertook postdoctoral studies at theUniversity of California at Irvine andthe University of East Anglia beforebeing appointed as a “New Blood”

Lecturer in Physical Chemistry at the University of Man-chester Institute of Science and Technology in 1984. In 1995,he was appointed to the Chair of Physical Chemistry at theUniversity of Salford. In 2004, he moved to Cork and

formed the Advanced Materials and Surfaces Group at Tyn-dall, see https://www.tyndall.ie/content/advanced-materials-and-surfaces-group-amsg. In 2008, he was appointed asStokes Professor of Materials Chemistry at UCC, see http://research.ucc.ie/profiles/D004/martynpemble/Research. Mar-tyn’s group studies the areas of advanced thin film technolo-gies with particular emphasis on chemical vapor depositionand atomic layer deposition. The group is also well-knownfor its work on the growth and characterisation of syntheticphotonic crystals. He has published over 250 papers in peer-reviewed scientific journals, currently has an h-index of over30, holds two patents and supervised some 40 Ph.D. studentsand 10 M.Sc. students. He is co-founder and Chairman of aUniversity spin-out company, CVD Technologies Ltd., seewww.cvdtechnologies.com.

Roger W. Whatmore graduated Ph.D.from Cambridge University (1977) &spent 18 yr with Plessey/GEC Mar-coni laboratories at Caswell (UK)working on the applications of ferro-electric materials. He was AwardedGEC’s Nelson Gold Medal (1993)and led the Prince of Wales’ Awardfor Innovation winning -team for

work on portable fire-fighting thermal imagers. He was aRoyal Academy of Engineering Professor in EngineeringNanotechnology & Head of Advanced Materials at CranfieldUniversity (1994–2005). He helped form IRISYS Ltd., (com-mercial infrared sensing) in 1996. He was CEO of TyndallNational Institute (Cork, 2006–2012). Now he is an EmeritusProfessor (University College Cork). He is a Fellow of theRoyal Academy of Engineering, Member of the Royal IrishAcademy, a Fellow of the Irish Academy of Engineering, aFellow of the Institute of Physics and a Fellow of the Insti-tute of Materials, Minerals, and Mining, who awarded himthe Griffith Medal and Prize for excellence in materials sci-ence in 2003. He has published ca300 papers & 40 patents,with an h-index of 42.


Magnetic Field-Induced Ferroelectric Switching in Multiferroic Aurivillius Phase Thin Films at Room Temperature

Documents