8
JOURNAL OF RAMAN SPECTROSCOPY J. Raman Spectrosc. 2003; 34: 524–531 Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jrs.1032 High-pressure Raman spectroscopy of nano-structured ABO 3 perovskites: a case study of relaxor ferroelectrics J. Kreisel 1and P. Bouvier 2 1 Laboratoire des Mat ´ eriaux et du G ´ enie Physique, ENS de Physique de Grenoble, B.P. 46, 38402 St. Martin d’H ` eres, France 2 Laboratoire d’Electrochimie et de Physicochimie des Mat ´ eriaux et des Interfaces, ENSEEG, B.P. 75, 38402 St. Martin d’H ` eres Cedex, France Received 30 January 2003; Accepted 28 April 2003 A large number of outstanding physical properties observed in perovskite-type oxides, such as giant piezoelectricity, colossal magneto-resistance and high-T c superconductivity, are related to an intrinsic nano-scaled local structure. For instance, the presence of a nano-scaled structure is characteristic of so-called relaxor ferroelectrics (relaxors), which have attracted considerable attention since the recent discovery of ultra-high strain and giant piezoelectric properties. In this paper we review, through a case study of relaxors, how the combination of the external parameter high-pressure and Raman spectroscopy allows an instructive investigation of nano-scaled perovskites, which is often at best a difficult task for conventional ‘average-structure’ techniques such as diffraction. We show that perovskite-type relaxors present important pressure instabilities with respect to both cation sizes and we suggest that the latter instabilities might well be at the origin of the reduced dielectric performances that are observed for strained thin-film relaxors. Finally, we briefly illustrate how the approach might be used for nano-scaled manganite-type oxides. Copyright 2003 John Wiley & Sons, Ltd. KEYWORDS: ferroelectrics; perovskites; high pressure; nano-structure; relaxors INTRODUCTION Many of the remarkable physical properties observed in ABO 3 -type oxides are related to materials with an intrin- sic nano-scaled local structure, where the different regions are characterized by competing chemical, structural and/or physical properties (e.g. Refs 1–7). One of the major chal- lenges in the analysis of nano-scaled oxides is the experi- mental access to the local properties, which is often at best a difficult task, a fact that frequently inhibits the under- standing of properties such as colossal magneto-resistance, giant piezoelectricity and high-temperature superconduc- tivity. New approaches to the detailed characterization of nano-structured materials are therefore clearly of interest since they will provide an improved understanding which in turn should lead to the possibility of tuning of local properties to create new functional materials with superior properties. Furthermore, a fundamental understanding of the nano-scale features of materials including a good knowl- edge of the local structural properties is also an important prerequisite for promising ab initio calculations. 8–10 Ł Correspondence to: J. Kreisel, Laboratoire des Mat´ eriaux et du enie Physique (CNRS), ENS de Physique de Grenoble, Domaine Universitaire, B.P. 46, 38402 St. Martin d’H` eres, France. E-mail: [email protected] The presence of a nano-scaled structure is, for instance, characteristic of so-called relaxor ferroelectrics (relaxors), 1 materials that have attracted considerable attention since the recent discovery of ultra-high strain and giant piezoelectric properties in relaxor-based single crystals. 11,12 Nano-scaled structure is also an intrinsic key feature of other important classes of materials such as rare earth manganite-based oxides displaying giant magneto-resistance (suggested to be similar to relaxors 13,14 ) and high-T c superconductors or more general correlated electron systems. 2–5 In this paper we will mainly focus on relaxors, which we consider as a model system of nano-structured oxides and which is perhaps the most and best investigated among perovskite-type oxides. Our objective is to illustrate through a review of already published data on Na 1/2 Bi 1/2 TiO 3 (NBT) 15 and PbMg 1/3 Nb 2/3 O 3 (PMN) 16 and through preliminary results on Pb(Mg 1/3 Nb 2/3 1x Ti x O 3 17 and BaTi 0.65 Zr 0.35 O 3 18 that high-pressure Raman spectroscopy is a useful technique for the investigation of relaxor ferroelectrics and, more generally, the class of nano-scaled oxides. RELAXOR FERROELECTRICS AND RAMAN SPECTROSCOPY Most technologically important relaxors crystallize in the so- called perovskite-type structure. The ideal cubic structure in Copyright 2003 John Wiley & Sons, Ltd.

High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

Embed Size (px)

Citation preview

Page 1: High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

JOURNAL OF RAMAN SPECTROSCOPYJ. Raman Spectrosc. 2003; 34: 524–531Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jrs.1032

High-pressure Raman spectroscopy of nano-structuredABO3 perovskites: a case study of relaxor ferroelectrics

J. Kreisel1∗ and P. Bouvier2

1 Laboratoire des Materiaux et du Genie Physique, ENS de Physique de Grenoble, B.P. 46, 38402 St. Martin d’Heres, France2 Laboratoire d’Electrochimie et de Physicochimie des Materiaux et des Interfaces, ENSEEG, B.P. 75, 38402 St. Martin d’Heres Cedex, France

Received 30 January 2003; Accepted 28 April 2003

A large number of outstanding physical properties observed in perovskite-type oxides, such as giantpiezoelectricity, colossal magneto-resistance and high-Tc superconductivity, are related to an intrinsicnano-scaled local structure. For instance, the presence of a nano-scaled structure is characteristic ofso-called relaxor ferroelectrics (relaxors), which have attracted considerable attention since the recentdiscovery of ultra-high strain and giant piezoelectric properties. In this paper we review, through a casestudy of relaxors, how the combination of the external parameter high-pressure and Raman spectroscopyallows an instructive investigation of nano-scaled perovskites, which is often at best a difficult task forconventional ‘average-structure’ techniques such as diffraction. We show that perovskite-type relaxorspresent important pressure instabilities with respect to both cation sizes and we suggest that the latterinstabilities might well be at the origin of the reduced dielectric performances that are observed forstrained thin-film relaxors. Finally, we briefly illustrate how the approach might be used for nano-scaledmanganite-type oxides. Copyright 2003 John Wiley & Sons, Ltd.

KEYWORDS: ferroelectrics; perovskites; high pressure; nano-structure; relaxors

INTRODUCTION

Many of the remarkable physical properties observed inABO3-type oxides are related to materials with an intrin-sic nano-scaled local structure, where the different regionsare characterized by competing chemical, structural and/orphysical properties (e.g. Refs 1–7). One of the major chal-lenges in the analysis of nano-scaled oxides is the experi-mental access to the local properties, which is often at besta difficult task, a fact that frequently inhibits the under-standing of properties such as colossal magneto-resistance,giant piezoelectricity and high-temperature superconduc-tivity. New approaches to the detailed characterization ofnano-structured materials are therefore clearly of interestsince they will provide an improved understanding whichin turn should lead to the possibility of tuning of localproperties to create new functional materials with superiorproperties. Furthermore, a fundamental understanding ofthe nano-scale features of materials including a good knowl-edge of the local structural properties is also an importantprerequisite for promising ab initio calculations.8 – 10

ŁCorrespondence to: J. Kreisel, Laboratoire des Materiaux et duGenie Physique (CNRS), ENS de Physique de Grenoble, DomaineUniversitaire, B.P. 46, 38402 St. Martin d’Heres, France.E-mail: [email protected]

The presence of a nano-scaled structure is, for instance,characteristic of so-called relaxor ferroelectrics (relaxors),1

materials that have attracted considerable attention since therecent discovery of ultra-high strain and giant piezoelectricproperties in relaxor-based single crystals.11,12 Nano-scaledstructure is also an intrinsic key feature of other importantclasses of materials such as rare earth manganite-basedoxides displaying giant magneto-resistance (suggested tobe similar to relaxors13,14) and high-Tc superconductors ormore general correlated electron systems.2 – 5

In this paper we will mainly focus on relaxors, whichwe consider as a model system of nano-structured oxidesand which is perhaps the most and best investigated amongperovskite-type oxides. Our objective is to illustrate througha review of already published data on Na1/2Bi1/2TiO3 (NBT)15

and PbMg1/3Nb2/3O3 (PMN)16 and through preliminaryresults on Pb(Mg1/3Nb2/3�1�xTixO3

17 and BaTi0.65Zr0.35O318

that high-pressure Raman spectroscopy is a useful techniquefor the investigation of relaxor ferroelectrics and, moregenerally, the class of nano-scaled oxides.

RELAXOR FERROELECTRICS AND RAMANSPECTROSCOPY

Most technologically important relaxors crystallize in the so-called perovskite-type structure. The ideal cubic structure in

Copyright 2003 John Wiley & Sons, Ltd.

Page 2: High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

High-pressure Raman spectroscopy of ABO3 perovskites 525

perovskite-type oxides ABO3 consists of corner-linked anionoctahedra BO6, the B cations at the center of the octahedra andthe larger A cations in the space (coordination 12) betweenthe octahedra. A number of materials deviate slightly fromthis ideal perovskite, for instance by a rotation (tilting) ofBO6 octahedra or/and by cation displacements.19 – 22 It is the‘off-centering’ of the A and/or B cation which gives rise toferroelectric (polar) properties.

Relaxors form a special class of ferroelectrics character-ized by a dielectric response with a broad peak (anomaly)as a function of temperature and a frequency-dependentresponse, rather than a sharp frequency-independent peakas in classical ferroelectrics.1,7,20 Conventional x-ray/neutrondiffraction of most perovskite-type relaxors shows that thedielectric anomaly in relaxors is not accompanied by a crys-tallographic symmetry change but remains cubic (Pm3m) atany temperature.1,7 The latter is intriguing, since a paraelec-tric cubic symmetry cannot explain the observed anomalyin the dielectric constant, which rather points to ‘something’ferroelectric.

As a consequence of the latter, Smolenskii andAgranovskaya23 proposed in the very early days of thediscovery of relaxors that deviations on the local scale mightwell explain the relaxor-characteristic physical properties.Without entering into the details of the still controversialdiscussion about the true origin of the relaxor behavior, wemight draw the following simplified picture: the averagecrystallographic structure of most relaxors (prototyped byPbMg1/3Nb2/3O3, PMN), as observed by x-ray or neutrondiffraction, is the ideal cubic perovskite structure.1,7 How-ever, on a local scale, these relaxors are characterized bypolar entities, i.e. zones where the cations are displaced fromtheir high-symmetry positions giving rise to a local electricdipole.1,7 Such (slight) local deviations from the averagestructure exist only on the nano-scale, below the correla-tion length of diffraction, and it is often not trivial to proveexperimentally that they do exist.

Raman spectroscopy (RS) is well known to be anappropriate technique for the study of subtle structuralchanges and in particular for the detection of slight deviationsfrom the cubic structure that often occurs in the class ofperovskite-type oxides. The latter is one of the reasonswhy RS has been found to be an important tool for theinvestigation of temperature- or pressure-induced phasetransitions in ferroelectric oxides such as BaTiO3 (A D Ba,Pb) or KNbO3 (e.g. Refs 24–30). Furthermore, RS can beconsidered as a local probe, which allows the detection oflocal structural deviations on a much smaller length scalethan diffraction techniques. These and other characteristicsmake RS to stand out for the investigation of relaxors,where one major objective is the detection of distortion fromthe cubic structure on the nano-scale. A nice illustrationof the latter is that relaxor ferroelectrics display a Ramanspectrum, which directly implies that relaxors have localdeviations, because Raman bands would be forbidden for

the ideal cubic perovskite structure. In fact, the presence of aspectrum can be understood by the local breakdown of theRaman selection rules and is generally related to local polardeviations (symmetry-breaking defects) with respect to theaverage cubic structure.

Although RS has been known for some time to be a veryuseful technique for the study of relaxors, as illustrated byan important number of reported studies,31 – 41 its use was inthe past limited to temperature- or substitution-dependentinvestigations. On the other hand, for relaxors, the externalparameter high pressure was until recently never combinedwith RS. This is surprising since high-pressure Ramanspectroscopy has been known for a long time to be a powerfultechnique for the investigation of various perovskite-typeoxides owing to its interesting dielectric properties (see, forinstance, investigations of BaTiO3,27,42 PbTiO3,43 – 45 SrTiO3,46

CaTiO3,47 KNbO330 and, more recently, PbZr1�xTixO3

48 – 51).Furthermore, it has been repeatedly pointed out that theeffect of pressure is a ‘cleaner’ variable52 compared with otherparameters, since it acts only on interatomic interactions. Thelack of high-pressure Raman studies on relaxors might be dueto the fact that high-pressure investigations of relaxors evenby other techniques are in general scarce and, further, limitedto low-pressure investigations (<1 GPa) by diffraction ordielectric measurements only (e.g. Refs 52–58).

EXPERIMENTAL

Depolarized Raman spectra of powder samples wererecorded in backscattering geometry with a Dilor XY multi-channel spectrometer equipped with a microscope objective.Radiation of 514.5 nm from an argon ion laser was usedfor excitation. High-pressure experiments were performedin a diamond anvil cell using methanol–ethanol–water as apressure-transmitting medium. Powder samples were placedin a chamber 250 µm in diameter and 50 µm thick and thepressure was monitored by the shift of the 2Fg ! 4A2g fluo-rescence bands of Cr3C ions in a small ruby crystal placed inthe vicinity of the sample. The Raman spectra were system-atically decomposed into individual Lorentzian componentsusing JANDEL Peakfit software.

RESULTS: HIGH-PRESSURE RAMANSPECTROSCOPY OF RELAXORS

General trends and considerationsFigure 1 displays some representative Raman spectra ofPMN and NBT as a function of pressure. An exhaustive illus-tration and discussion of the pressure-dependent spectralevolution can be found in previous papers.15,16

The qualitative inspection of Fig. 1 reveals the important(seemingly simple) observation that the Raman scatteringpattern changes dramatically with pressure, even in themoderate 1–5 GPa pressure range. The latter observation is

Copyright 2003 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2003; 34: 524–531

Page 3: High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

526 J. Kreisel and P. Bouvier

0.42

1.75

13.313.8

11.0

9.15

7.256.40

5.55

4.083.30

2.21

10.0

4.90

0.74

19.017.015.6

(b) Na1/2Bi1/2TiO3

200 400 600 800 200 400 600 800

Pressure(GPa)

Pressure(GPa)

19.75

6.90

5.504.683.55

2.65

1.85

0.950.300.161 bar

6.40

14.75

10.80

9.35

7.99

7.00

(a) PbMg1/3Nb2/3O3

Wavenumber / cm-1 Wavenumber / cm-1

Inte

nsity

(a.

u)

Figure 1. Representative pressure-dependent Raman spectra for (a) PbMg1/3Nb2/3O3 and (b) Na1/2Bi1/2TiO3 (spectra taken fromRefs 15 and 16).

interesting and surprising if we remember that a character-istic reported feature of relaxors is that their Raman spectradisplay only small (if any) changes when the external param-eter temperature is used. In this sense, we should add that wehave made similar pressure observations for other relaxorssuch as Pb(Mg1/3Nb2/3)1�xTixO3 (Ref. 17) and BaTi0.65Zr0.35O3

(Ref. 18) and it thus appears that, in general terms, relaxorferroelectrics evolve differently under high pressure andtemperature. The latter conclusion is not trivial and wasnot straightforwardly expected before the experimental evi-dence, especially because we know that the Raman spectra ofclassic ferroelectrics in contrast evolve similarly under highpressure and temperature (e.g. BaTiO3,27,42 KNbO3

30).In addition to the above qualitative remark, Fig. 1

illustrates that for NBT and PMN pressure leads to bothnotable band changes (position, intensity, width) and newspectral features. Although the latter characteristics will bediscussed separately in the following sections, we mightalready note that their simple presence illustrates thatpressure alters fundamentally the structural and polarproperties of NBT and PMN, suggesting that intrinsic

pressure instabilities might well play an important role inthe observed59,60 reduction in dielectric properties in relaxor-related thin films.

Pressure-induced changes in the local polarityOne of the key points towards understanding relaxorferroelectrics under external parameters is the experimentalaccess to local polar characteristics. We will show in thefollowing sub-sections that different spectral domains giveinformation about different polar characteristics.

Low-wavenumber regionIn general, the low-wavenumber region, say below 150 cm�1,is known to be dominated by vibrations involving the Acations of the perovskite structure. For instance, the low-wavenumber region is dominated for NBT by a band at135 cm�1, which has been associated with Na–O vibrations,39

whereas the band at 140 cm�1 in PMN was reported tobelong to a Pb–O stretching mode of A1g symmetry.61 Asa consequence, such vibrational modes offer a probe ofpressure changes associated with the polar A cation.

Copyright 2003 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2003; 34: 524–531

Page 4: High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

High-pressure Raman spectroscopy of ABO3 perovskites 527

Figure 1 shows that the above-mentioned A site-associated bands shift first to lower wavenumbers whenthe pressure is increased before they start to increase againat a critical pressure pc (detailed fit results were presentedin previous papers15,16). According to the concept of hardmode spectroscopy,62 – 64 such a slope change indicates thatNBT and PMN undergo phase transitions. It is interestingthat one of the common opinions about relaxors is that theydo not present a soft mode, the latter so far being verified bythe absence of temperature-induced soft modes. Althoughwe should point that the wavenumber lowering observedhere is not a soft mode (since it displays only a slight down-ward shift and remains above 100 cm�1�, we can still speakin terms of a softening that indicates an anharmonicity neara phase transition. We propose that this softening resultsfrom a restoring of the A cation towards the high-symmetrycenter inside the AO12. The difference from classical fer-roelectrics lies in the fact that this restoration only occurswithin local polar clusters, and thus on a small length scale(correlation length), explaining that we observe a softeningrather than a soft mode. Finally, the hardening at pc can beunderstood in terms of a pressure-induced change in thedirection of the cation displacement or a change from paral-lel to anti-parallel displacements, i.e. for a pressure-inducedrelaxor ferroelectric ! anti-ferroelectric phase transition.Very recent high-pressure investigations via synchrotronradiation are in favor of the latter scenario.65,66

Mid-wavenumber regionLet us now consider the pressure-induced spectral changesin the Mid-wavenumber region (¾250–350 cm�1�, wherea broad and intense band is the dominant feature in theNBT and PMN Raman spectra under ambient conditions.For both materials this region is attributed to vibrationsassociated with the BO6 octahedra. As can be seen inFig. 1, this band undergoes important pressure changes inband position and wavenumber. In fact, a pressure-inducednegative wavenumber shift and a loss of intensity in thisregion are commonly used direct signatures to provideevidence of polar changes on the B site related to therestoration of the cation towards the center of the octahedron(e.g. Refs 27,42,43 and 45). Furthermore, studies of the Ag

band in similar RTiO3 (R D La, Ce, Pr, Nd, Sm, Gd),67

having an orthorhombically distorted perovskite structure,indicate a decrease in wavenumber from 385 to 287 cm�1

with decreasing orthorhombic distortion (octahedron tilt).In the light of the above remarks, we attribute the

important intensity breakdown in the mid-wavenumberregion to important polar changes of the B cation insidethe BO6 octahedra. Furthermore, the negative wavenumbershift suggests changes in the angle of the BO6 tilts. Moreprecisely, we propose for the example of NBT (with anorthorhombic high-pressure phase68) that the behavior ofthe 287 cm�1 band in NBT is probably related to a structuralrearrangement that is conditioned by both the restoration

of the Ti cation to the center of symmetry ([111]p ! [000])and a change in the tilt system (a�a�a� ! a�bCa�; Glazernotation21).

High-wavenumber regionAdditional support for polar changes on the B site comesfrom the evolution of the high-wavenumber modes, whichare dominated by vibrations of O2� anions involving themost rigid cation–oxygen bonds inside the BO6 octahedra(breathing and bending of BO6�. Although the polar B cationis not directly involved, such modes can give an indirectinsight into polar properties because polar B—O bondscondition a splitting into longitudinal (LO) and transversal(TO) components. In order to illustrate the latter, let usconsider, for instance, the rhombohedral phase of NBT whereeach of the A1 and E modes is Raman and infrared active15,39

and thus LO–TO splitting can be expected as a result of long-range electrostatic forces. Such LO–TO splitting has alreadybeen reported for BaTiO3

42 and PbTiO3.43 As a consequence,we attribute the observed split of the high-wavenumberbands around 650 and 800 cm�1 in PMN and NBT to LO–TOsplitting originating in the polar displacement of the polarcation inside the BO6 octahedron. As shown in Fig. 1, theinitial splitting under ambient conditions is reduced withincreasing pressure and leads finally to single a band.We believe that this pressure-induced band merging isdue to the loss of longitudinal (LO) and transversal (TO)character. Exemplifying again the case of NBT, we canunderstand the spectral changes by considering a high-pressure orthorhombic Pbnm symmetry (antiferroelectric Acation displacement, no B cation displacement68) where theAg, B1g, B2g and B3g modes are Raman active and, therefore, noLO–TO splitting is expected. Based on this consideration, theband merging can be interpreted as a reduction of the LO–TOsplitting on approaching the rhombohedral ! orthorhombicphase transition, i.e. restoration of the B cation towards itshigh-symmetry position in the BO6 octahedron (in agreementwith the above discussion of the mid-wavenumber region).Finally, the linear pressure shift of the single band makesit attractive for an estimate of the relaxor compressibility,but this is beyond the scope of this paper and the interestedreader is referred to Refs 69 – 71 where this approach is furtherdetailed.

Finally, from this and the above sections it becomesclear that NBT and PMN present structural and polarinstabilities towards external pressure. We emphasize thatthe observed pressure-induced fundamental modificationsof the polar properties take place at both cation sites Aand B of the perovskite structure, although presumablyat different critical pressures,15,16 suggesting a differentpressure-dependent behavior for different nano-regions. Oneof the structural mechanisms that we might well take intoconsideration to understand the latter point is based on theconsideration of chemically different nano-regions, which

Copyright 2003 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2003; 34: 524–531

Page 5: High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

528 J. Kreisel and P. Bouvier

have different compressibilities,71 thus leading to a differentbehavior with applied pressure.

A unique spectral signature?Let us now consider a peculiar spectral signature that wefirst reported for NBT.15 As shown in Fig. 2, NBT andPMN display a common pressure-induced signature thatis characterized by an intensity breakdown in the mid-wavenumber region occurring together with an almostunchanged intensity of the high-wavenumber bands. Such aspectral signature is very unusual and has, to the best of ourknowledge, no precedent among reported Raman studies onperovskites for pressure-dependent Raman spectra of clas-sical perovskite-type ferroelectrics, an intensity breakdownof the mid-wavenumber bands (revealing a change in long-range ferroelectric properties) is usually accompanied by anintensity loss for all other bands.

It is important to emphasize that such a pressure-dependent spectral similarity between NBT and PMN isat first sight surprising because PMN and NBT (i) are verydistinct from a chemical point of view, (ii) are B site (PMN)compared with A site (NBT) substituted and (iii) show acubic (PMN) compared with a rhombohedral (NBT) averagestructure under ambient conditions. On the other hand,PMN and NBT share the relaxor-characteristic property of anano-scaled structure, i.e. the presence of local, symmetry-breaking domains with polar cation displacements. It isnot straightforward to link the relaxor characteristics to thepeculiar signature, and we cannot yet propose a definitemodel explaining the observed behavior. However, wesuggest that the different behavior might well be linkedto the fact that the high-wavenumber bands are related tomore localized molecular vibrations of the oxygen network(e.g. breathing of BO6 octahedra) which are less sensitiveto disorder and nano-scale perturbations. On the otherhand, low-wavenumber (more cation-related) bands mightbe affected by local pressure-induced polar changes, leadingfor instance to increased polar disorder. In fact, observations

by Samara and co-workers52 – 55,72,73 suggesting a pressure-induced long-range to short-range transition in ferroelectricsgives some support to the latter scenario. Nevertheless, thefinal understanding of this peculiar signature will needfurther experimental and theoretical attention.

Whatever the true origin is, we believe that the commonrelaxor-like characteristics are at the origin of the similarbehavior and we propose that the Raman signature illus-trated in Fig. 2 might well be a unique pressure-dependentspectral signature for polar changes in relaxor ferroelectrics.The identification of a unique pressure-dependent spectralfeature in relaxors is interesting not only from a fundamentalpoint of view but also because the denomination ‘relaxor’(and thus interpretation) of a given material often leadsto long-standing and controversial discussions, and this isparticularly true in the case of thin films.74

A recent example that illustrates a potential problematicdenomination relaxor is the environmentally friendly high-strain material BaTi1�xZrxO3 (BTZ).75 – 81 An investigationby Ravez and Simon suggested that BTZ ceramics exhibitrelaxor-like properties (i.e. a relaxation in the dielectricresponse) for substitution rates higher than x D 0.25.However, it was realized that BTZ is an unusual relaxorcompared with classical relaxors such as PMN and NBT.Note that PMN and NBT share three characteristics: (i) thepresence of Pb2C or Bi3C, both showing a lone pair effect, (ii) aheterovalent disorder on at least one of the A and B sites and(iii) the formation of a chemical compound, i.e. a substanceof fixed composition in contrast to a solid solution. Thesolid solution BTZ, which involves a homovalent Ti4C/Zr4C

substitution on the B site, does not respond to any of thelatter characteristics, revealing that the observed relaxationin BTZ is at least intriguing. As a consequence, one questionthat arises is whether BTZ is a relaxor in the sense (theory,principle, mechanisms, etc.) of PMN and NBT or not. Afirst Raman scattering investigation by Farhi et al.77 showedthat the temperature-dependent Raman spectra of BTZ with

100 200 300 400 500 600 700

(a) PMN (b) NBT

Wavenumber / cm-1 Wavenumber / cm-1

Figure 2. Directly superimposed pressure-dependent Raman spectra for (a) PbMg1/3Nb2/3O3 (PMN) and (b) Na1/2Bi1/2TiO3

(NBT).15,16 It is a common characteristic feature of these two materials that with increasing pressure the intensity drops dramaticallyin the mid-wavenumber region whereas the features in the high-wavenumber region remain almost unchanged in its intensity. Sucha signature has no precedent among reported Raman studies on perovskites and might well be a unique and characteristic featureof relaxor ferroelectrics.

Copyright 2003 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2003; 34: 524–531

Page 6: High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

High-pressure Raman spectroscopy of ABO3 perovskites 529

x > 0.25 are characterized by a continuous variation of theoverall intensity of the spectra with temperature, withoutany change in the width or wavenumber of the bands.77

Such a temperature signature is generally considered to bea relaxor-characteristic feature (similarly for PMN and NBT)and as a consequence is was argued77 that BTZ is a relaxorin terms of PMN and NBT. In sharp contrast to the latter,our very recent high-pressure Raman experiment18 showsthat BTZ does not show the above-discussed peculiar relaxorfeature as was observed for PMN and NBT. The latter is, infact, the first real experimental evidence illustrating that BTZis different and illustrated that BTZ is not nano-structuredlike PMN- and NBT-type relaxors as one would intuitivelyexpect from the fact that BTZ is a solid solution but not acompound. In other words, BTZ shows a dielectric relaxation(and is thus somehow a relaxor) but is not a relaxor in thesense of NBT and PMN.

EXTENSION TO OTHER MATERIALS

We have already pointed out in the Introduction that relaxorswith their peculiar piezoelectric properties are not the onlymaterial showing nano-structural features. In fact, we believethat the approach proposed here is also useful for themore general case of nano-structured materials. In order toillustrate the latter, we will briefly discuss the case of so-calledmanganite-based perovskites, which have regained interestsince the discovery of their colossal magneto-resistance. Asa prototype example for our discussion we have chosen Sr-doped manganites La1�xSrxMnO3 (hereafter called LSMOx�.

Let us first set up a picture of the structural and phys-ical characteristics of LSMOx. La1�xSrxMnO3 is known toshow a complex composition–temperature (x–T) phasediagram that is characterized by various rhombohedral-to-orthorhombic (R ! O) phase transitions, which is oneof the common phenomena in manganite-based oxides. Forinstance, LSMOx undergoes at T D 300 K and x D 0.17 an x-dependent R ! O phase transition,82 which is accompaniedby important changes in the magnetic/electric properties andoutstanding magnetoresistive properties. From a crystallo-graphic point of view, the latter transition can be describedby a change in the tilt (rotation) of the MnO6 octahedra(a�a�a� ! a�bCa�; Glazer notation21), leading to a decreasein the Mn—O—Mn bond angle, which is the critical param-eter for the super-exchange-type magnetic interactions. Justas in other perovskites, the tilt angle changes, involvedaccompanying the R ! O transition, can be very smalland difficult to investigate by means of diffraction experi-ments. Furthermore, manganites are characterized by localintrinsic inhomogeneities (local nano-structure) that can beof different chemical (composition), structural (orthorhom-bic/rhombohedral), magnetic (ferro-/antiferro- or non-magnetic) or electric (insulator/conductor) nature.2 – 5,14 Theassociated structural deviations of such local domains areoften below the correlation length of x-ray and neutron

diffraction and only an average structure is thus observed.It is most likely that a number of manganites that haveso far been interpreted in the framework of an averagestructure only might well present in reality local inhomo-geneities/local structural transitions which have not beenobserved by using only average-structure techniques.

Raman spectroscopy offers a well-documented finger-print signature for the rhombohedral and orthorhombicphase of LSMOx.83 – 86 Furthermore, the R ! O transitionpresents a recognizable signature that can be described by aprogressive weakening of two broad and strong ‘orthorhom-bic peaks’ (500 and 600 cm�1� and the appearance of two nar-row ‘rhombohedral peaks’ (45 and 426 cm�1).84,86 Througha layer thickness-dependent investigation of LSMO0.3-containing multilayers, we have recently shown14 that bothtypes of characteristic bands might co-exist within a sample,attesting to the presence of rhombohedral and orthorhombicdomains (Fig. 3).14 The latter study has further shown thattensile stress (and somehow pressure) leads to a notableevolution of the different domains, which allowed a deeperunderstanding of the observed strain-dependent physicalproperties.14

We propose that the combination of RS and the externalparameter high pressure might well lead to more insightinto the presence or not of local structural deviations (mostlikely associated with chemical and/or magnetic–electricdeviations) in relation to macroscopically measured phys-ical properties. The feasibility of the latter is supportedby a recent layer thickness-dependent investigation ofLa0.7Sr0.3MnO3/SrTiO3 multilayers where we have shownthat tensile stress (and somehow pressure) allows the inves-tigation of the evolution of rhombohedral and orthorhombicclusters (see Ref. 14 for a more detailed discussion). Letus note, however, that Raman investigations of manganitesare not always straightforward, mainly owing to their lowRaman intensity combined with a high sensitivity to laserpower,85 but some first preliminary studies have illustratedthe feasibility.87,88

CONCLUSION

We have reviewed and discussed recent high-pressureRaman investigations of relaxor ferroelectrics, consideredas model nano-structured perovskites. Our results showthat relaxors present important pressure instabilities onboth cation sites of the perovskite structure within theorder of magnitude that one might encounter in thin-filmdevices. We have also discussed the occurrence of a pressure-induced peculiar spectral signature which we propose to becharacteristic for NBT- and PMN-type relaxors, thus offeringa new experimental way for discussing the denominationrelaxor for a given material.

Copyright 2003 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2003; 34: 524–531

Page 7: High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

530 J. Kreisel and P. Bouvier

200 400 600 800

LSMO36/STO90

LSMO100/STO90

LSMO210/STO90

O

Wavenumber / cm-1

LSMO100/STO124

R

O

*

**

Figure 3. Raman spectra of [(La0.7Sr0.3MnO3�m/(SrTiO3�n]15//LaAlO3 multilayers, m and n being the thickness of 15 timesalternated layers.14 The spectra have been shifted for clarity.Bands marked with R or O are characteristic of a rhombohedralor orthorhombic structural deformation, respectively, andbands marked with an asterisk belong to the LaAlO3 substrate.

However, whatever the specific results are, we believethat an important result is that one can probe relaxor-characteristic features with high-pressure Raman spec-troscopy. We should again point out that the latter is nottrivial since, for instance, temperature-dependent Ramanscattering does not offer the same, diffraction offers at bestmediocre access to relaxor-specific features for any externalparameter and dielectric measurements allow only a macro-scopic view of the dielectric properties. We have shownthat high-pressure Raman spectroscopy is an alternative,complementary and effective technique to handle the latter

deficiencies for relaxors but also, more generally, for thelarger class of nano-structured oxides for which we havediscussed the example of manganites.

We expect that further pressure-dependent investiga-tions by local probes, such as Raman but also EXAFS, NMRor diffuse scattering,65,66 will reveal insight into materialswith outstanding properties that are driven by a peculiarmicrostructure (relaxors, manganite-type oxides, etc.). Suchinvestigations are especially promising to guide us to theclever realization of thin-films and even more thin-film het-erostructures where strain plays an major role.

AcknowledgmentsThe authors thank G. Lucazeau for numerous instructive discus-sions on the Raman spectra of perovskite-type oxides. N. Rosman isacknowledged for invaluable technical assistance in the implemen-tation and realization of high-pressure Raman studies.

REFERENCES1. Cross LE. Ferroelectrics 1987; 76: 241.2. Tokura Y, Nagaosa N. Science 2000; 288: 462.3. Burgy J, Mayr M, Martin-Mayor V, Moreo A, Dagotto E. Phys.

Rev. Letters 2001; 87: 277 202.4. Moreo A, Yunoki S, Dagotto E. Science 1999; 283: 2043.5. Shibata T, Bunker B, Mitchell JF, Schiffer P. Phys. Rev. Lett. 2002;

88: 207 205.6. Dagotto E. Nanoscale Phase Separation and Colossal Magnetoresis-

tance. Springer Series in Solid-State Physics, vol. 136. Springer:Berlin, 2002.

7. Cross LE. Ferroelectrics 1994; 151: 305.8. Cockayne E, Rabe KM. Phys. Rev. 1998; 57: 13 973.9. Bellaiche L, Vanderbilt D. Phys. Rev. Lett. 1998; 81: 1318.

10. Iniguez J, Bellaiche L. Phys. Rev. Lett. 2001; 87: 095 503.11. Park SE, Shrout TR. J. Appl. Phys. 1997; 82: 1804.12. Fu H, Cohen RE. Nature (London) 2000; 403: 281.13. Kimura T, Tomioka Y, Kumai R, Okimoto Y, Tokura T. Phys.

Rev. Lett. 1999; 83: 3940.14. Kreisel J, Lucazeau G, Dubourdieu C, Rosina M, Weiss F. J.

Phys.: Condens. Matter 2002; 14: 5201.15. Kreisel J, Glazer AM, Bouvier P, Lucazeau G. Phys. Rev. B 2001;

63: 174 106.16. Kreisel J, Dkhil B, Bouvier P, Kiat JM. Phys. Rev. B 2002; 65:

172 101.17. Chaabane B, Kreisel J, Bouvier P, Lucazeau G. Phys. Rev. B 2003;

submitted.18. Kreisel J, Bouvier P, Maglione M, Simon A. Phys. Rev. B 2003;

submitted.19. Mitchell RH. Perovskites: Modern and Ancient. Almaz Press:

Ontario, 2002.20. Wadhawan VK. Introduction to Ferroic Materials. Gordon and

Breach: Amsterdam, 2000.21. Glazer AM. Acta Crystallogr., Sect. B 1972; 28: 3384.22. Glazer AM. Acta Crystallogr., Sect. A 1975; 31: 756.23. Smolenskii GA, Agranovskaya AI. Sov. Phys. Solid State 1959; 1:

1429.24. Burns G, Scott BA. Phys. Rev. Lett. 1970; 25: 167.25. Jiang YJ, Zeng LZ, Wang RP, Zhu Y, Liu TL. J. Raman Spectrosc.

1996; 27: 31.26. Li LM, Jiang YJ, Zeng LZ. J. Raman Spectrosc. 1996; 27: 503.27. Sood AK, Chandrabas N, Muthu DVS, Jayaraman A. Phys. Rev.

B 1995; 51: 8892.28. Sokoloff JP, Chase LL, Rytz D. Phys. Rev. B 1988; 38: 597–605.

Copyright 2003 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2003; 34: 524–531

Page 8: High-pressure Raman spectroscopy of nano-structured ABO3 perovskites: a case study of relaxor ferroelectrics

High-pressure Raman spectroscopy of ABO3 perovskites 531

29. Shen ZX, Hu ZP, Chong TC, Beh CY, Tang SH, Kuok MH. Phys.Rev. B 1995; 52: 3976–3980.

30. Gourdain D, Moya E, Chervin JC, Canny B, Pruzan P. Phys. Rev.B 1995; 52: 3108.

31. Setter N, Laulicht I. Appl. Spectrosc. 1987; 41: 526.32. Siny IG, Katiyar RS, Bhalla AS. J. Raman Spectrosc. 1998; 29: 385.33. Siny IG, Smirnova TA, Kruzina TV. Ferroelectrics 1991; 124: 207.34. Siny IG, Smirnova TA, Kruzina TV. Sov. Phys. Solid State 1991;

33: 61.35. Siny IG, Katiyar RS. Ferroelectrics 1998; 206: 307.36. Ching-Prado E, Cordero J, Katiyar RS, Bhalla AS. J. Vac. Sci.

Technol. A 1996; 14: 762.37. Jiang F, Kojima S, Zhao C, Feng C. J. Appl. Phys. 2000; 88: 3608.38. Farhi R, El Marssi M, Dellis JL, Yuzyuk YI, Ravez J,

Glinchuk MD. Ferroelectrics 1999; 235: 9.39. Kreisel J, Glazer AM, Jones G, Thomas PA, Abello L,

Lucazeau G. J. Phys.: Condens. Matter 2000; 12: 3267.40. Ohwa H, Iwata M, Yasuda N, Ishibashi Y. Ferroelectrics 1999; 229:

147.41. Hafid M, Kugel GE, Handerek J, Ujma Z. Ferroelectrics 1992; 135:

101.42. Venkateswaran UD, Naik VM, Naik R. Phys. Rev. B 1998; 58:

14 256.43. Sanjurjo JA, Lopez-Cruz E, Burns G. Phys. Rev. B 1983; 28: 7260.44. Cerdeira F, Holzapfel WB, Bauerle D. Phys. Rev. B 1975; 11: 1188.45. Burns G, Sanjurjo JA, Lopez-Cruz E. Phys. Rev. B 1984; 30: 7170.46. Grzechnik A, Wolf GH, McMillan PF. J. Raman Spectrosc. 1997;

28: 885.47. Gillet P, Guyot F, Price GD, Tournerie B, Cleach AL. Phys. Chem.

Miner. 1993; 20: 159.48. Souza Filho AG, Faria JLB, Freire PTC, Ayala AP, Sasaki JM,

Melo FEA, Mendes Filho J, Araujo EB, Eiras JA. J. Phys.: Condens.Matter 2001; 13: 7305.

49. Souza Filho AG, Freire PTC, Ayala AP, Sasaki JM, Guedes I,Mendes Filho J, Melo FEA, Araujo EB, Eiras JA. J. Phys.: Condens.Matter 2000; 12: 7295.

50. Souza Filho AG, Lima KCV, Ayala AP, Guedes I, Freire PTC,Mendes Filho J, Araujo EB, Eiras JA. Phys. Rev. B 2000; 61: 14 283.

51. Rouquette J, Haines J, Bornand V, Pintard M, Papet P, Astier R,Leger JM, Gorelli F. Phys. Rev. B 2002; 65: 214 102.

52. Samara GA. Phys. Rev. Lett. 1996; 77: 314.53. Samara GA. Fundamental Physics of Ferroelectrics 2000. American

Institute of Physics: New York, 2000; 344.54. Samara GA, Boatner LA. Phys. Rev. B 2000; 61: 3889.55. Samara GA, Venturini EL, Schmidt VH. Appl. Phys. Lett. 2000;

76: 1327.56. Yasuda N, Ohwa H, Oohashi J, Nomura K, Terauchi H,

Iwata M, Ishibashi Y. J. Phys. Soc. Jpn. 1997; 66: 1920.57. Yasuda N, Ohwa H, Oohashi J, Nomura K, Terauchi H,

Iwata M, Ishibashi Y. J. Phys. Soc. Jpn. 1998; 67: 3952.58. Nomura K, Shingai T, Yasuda N, Ohwa H, Terauchi H. J. Phys.

Soc. Jpn. 1999; 68: 866.

59. Nagarajan V, Ganpule CS, Nagaraj B, Aggarwal S, Alpay SP,Roytburd AL, Williams ED, Ramesh R. Appl. Phys. Lett. 1999;75: 4183.

60. Nagarajan VA, Alpay SP, Ganpule CS, Nagaraj BK, Aggarwal S,Williams ED, Roytburd AL, Ramesh R. Appl. Phys. Lett. 2000; 77:438.

61. Husson E, Abello L, Morell A. Mater. Res. Bull. 1990; 25: 539.62. Bismayer U. Phase Transitions 1990; 27: 211.63. Salje EKH, Yogal Y. J. Physics Chem. Solids 1996; 57.64. Salje EKH, Bismayer U. Phase Transitions 1997; 63: 1.65. Kreisel J, Bouvier P, Dkhil B, Thomas PA, Glazer AM,

Welberry TR, Chaabane B, Mezouar M. Phys. Rev. B 2003;accepted.

66. Chaabane B, Kreisel J, Dkhil B, Bouvier P, Mezouar M. Phys. Rev.Lett. 2003; in print.

67. Reedyk M, Crandles DA, Cardona M, Garret JD, Greedan JE.Physica B 1997; 55: 1442.

68. Jones GO, Kreisel J, Jennings V, Geday MA, Thomas PA,Glazer AM. Ferroelectrics 2002; 270: 191.

69. Loridant S, Lucazeau G. J. Raman Spectrosc. 1999; 30: 485.70. Chemarin C, Rosman N, Pagnier T, Lucazeau G. J. Solid State

Chem. 2000; 149: 298.71. Kreisel J, Glazer AM. J. Phys.: Condens. Matter 2000; 12: 9689.72. Samara GA. J. Appl. Phys. 1998; 84: 2538.73. Samara GA, Venturini EL, Schmidt VH. Phys. Rev. B 2001; 63:

184 104.74. Tyunina M, Levoska J. Phys. Rev. B 2001; 63: 224 102.75. Ravez J, Simon A. Eur. Phys. J.: Appl. Phys. 1997; 34: 1199.76. Ravez J, Simon A. J. Korean Phys. Soc. 1998; 32: S955.77. Farhi R, Marssi ME, Simon A, Ravez J. Eur. Phys. J. B 1999; 9:

559.78. Rehring PW, Park SE, Trolier-McKinstry S, Messing GL, Jones B,

Shrout TR. J. Appl. Phys. 1999; 86: 1657.79. Yu Z, Guo R, Bhalla AS. Appl. Phys. Lett. 2000; 77: 1535.80. Yu Z, Guo R, Bhalla AS. J. Appl. Phys. 2000; 88.81. Yu Z, Ang C, Guo R, Bhalla AS. J. Appl. Phys. 2002; 92: 1489.82. Urushibara A, Moritomo Y, Arima T, Asamitsu A, Kido G,

Tokura Y, Tomioka Y. Phys. Rev. B 1995; 51: 14 103.83. Podobedov VB, Weber A, Romero DB, Rice JP, Drew HD. Phys.

Rev. B 1998; 58: 43.84. Bjornsson P, Rubhausen M, Backstrom J, Kall M, Eriksson S,

Eriksen J, Borjesson L. Phys. Rev. B 2000; 61: 1193.85. Iliev MN, Abrashev MV, Lee HG, Popov VN, Sun YY,

Thomsen C, Meng RL, Chu CW. Phys. Rev. B 1998; 57: 2872.86. Granado E, Moreno NO, Garcia A, Sanjurjo JA, Rettori C,

Torriani I, Oseroff SB, Neumeier JJ, McClellan KJ, Cheong SW,Tokura Y. Phys. Rev. B 1998; 58: 11 435.

87. Congeduti A, Postorino P, Dore P, Lupi S, Mercone S, Nucara A,Nardone M. Int. J. Mod. Phys. B 2000; 14: 3418.

88. Lucazeau G, Bouvier P, Bihan TL, Kreisel J, Gorbenko OY,Bosak AA, Kaul A. Presented at the International WorkshopMSU-HTSC VI, Moscow, 2001.

Copyright 2003 John Wiley & Sons, Ltd. J. Raman Spectrosc. 2003; 34: 524–531