Metastable perovskite Bi1-xLaxFe0.5Sc0.5O3 phases in the range of the compositional crossover

ABSTRACT Perovskite ceramics of the Bi1-xLaxFe0.5Sc0.5O3 composition (0.30 ≤ x ≤ 0.35) that cannot be sintered in bulk form as a single phase using the conventional ceramic route were successfully prepared using the high-pressure/high-temperature technique. It has been shown that the room-temperature compositional crossover from the antipolar phase whose incommensurate modulation of displacements of Bi/La and oxygen is described by the Imma(00γ)s00 superspace group to the non-polar Pnma phase occurs in the narrow range between x = 0.33 and x = 0.34 with no phase coexistence. The features of this compositional crossover are discussed in comparison with that observed in the Bi1-xLaxFeO3 system.


Introduction
Perovskite bismuth ferrite is one of the most studied multiferroics since this compound can be obtained using the conventional preparation methods. BiFeO 3 is ferroelectric until T C D 1083 K while the magnetic-paramagnetic phase transition occurs at T N D 643 K [1]. The temperatures of both transitions are too high and far from each other which makes difficult a beneficial use of the lattice-magnetic coupling effect. Moreover, in bulk BiFeO 3 , the spins form a long-period cycloid that results in averaging of the local weak ferromagnetic components to zero [2]. A zero-field polar weak-ferromagnetic state in bismuth ferrite can be stabilized by means of chemical dopings. Besides, substitutions in the Bi 3C sites and in the Fe 3C sites modify the temperatures of the ferroelectric transition and the magnetic transition, respectively. A number of studies have been performed in this respect [3][4][5][6].
The BiFeO 3 -based solid solution systems where bismuth is substituted by a rear earth element are the most explored (Ref. [7] and references therein). Structural phases different from the parent rhombohedral R3c have been obtained and thoroughly studied [8][9][10][11][12]. Fe-site substitutions in bismuth ferrite are much less investigated since, in the majority of cases, small substitution rates (an order of 10 mol% when the resulting structure is still rhombohedral) are only achievable. As a rule, non-conventional preparation methods are needed to obtain the extended (entire) solid solution systems with substitution either in Fe 3C sites or in both cation sites. Polycrystalline solid solutions, Bi(Fe, B 3C )O 3 where B 3C is Co, Ga, Mn, Cr, Sc, were prepared using the high-pressure/ high-temperature method [13][14][15][16][17][18].
Recently, we initiated a systematic study of the quasi quadruple BiFeO 3 -BiScO 3 -LaFeO 3 -LaScO 3 perovskite system. Three end members of this system, BiFeO 3 , LaFeO 3 and LaScO 3 , can be obtained using the conventional methods (in particular, the standard ceramic technique), while a bulk perovskite BiScO 3 phase can be synthesized under the high-pressure conditions only [19]. The main idea of the exploration of the Bi 1-x La x Fe 1-y Sc y O 3 system is to control (adjust) the temperatures of the magnetic and the polar transitions. In this system, all the constituent cations are trivalent that makes possible to vary the parameters x and y independently. Hence, one can decrease the temperature of polar transition by means of a replacement of bismuth by lanthanum and decrease the temperature of magnetic transition through an iron-to-scandium substitution. Therein, the purpose of this study was to obtain the solid solution with the overlapping ranges of T C and T N aiming to achieve the maximal lattice-magnetic coupling effect.
We have reported three different structural phases observed in the as-prepared polycrystalline samples of the particular section of the system, when the Fe/Sc ratio is fixed to be 1:1. At room temperature, the perovskite phase of the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 ceramics with x D 0 was found to be antipolar Pnma with the x2a p £ 4a p £ 2x2a p superstructure (a p is the pseudocubic unit cell) [20]. In the ceramics with the x D 0.20 composition, an incommensurate modulation of displacements of Bi/La and oxygen is described by the Imma(00g)s00 superspace group (x2a p £ 2a p £ x2a p ) [21]. Non-polar Pnma phase (x2a p £ 2a p £ x2a p ) was detected in the samples with x D 0.35 [22]. In all the mentioned compositions, the magnetic transition was observed at about 220 K [20][21][22][23][24], i.e. slightly below room temperature. Below T N , these solid solutions were shown to exhibit long-range antiferromagnetic ordering with a weak ferromagnetic component.
Neutron diffraction study of the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 ceramics with x D 0.35 performed between room temperature and 1.5 K has revealed no transition from the non-polar Pnma phase [22]. Thus, at room temperature, the solid solution with x D 0.20 is an antipolar, while that with x D 0.35 is a non-polar. Hence, the composition of the solid solution whose transition between the antipolar phase and non-polar phase occurs in the vicinity of room temperature is certainly between 0.20 and 0.35.
In this work, the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 solid solutions in the range of 0.30 x 0.35 were prepared and thoroughly investigated aiming at determination of the compositional phase crossover and identify the composition for which T C is close to T N . The ceramics were synthesized/sintered under high pressure. The obtained samples were characterized using the methods of X-ray diffraction (XRD) including in situ temperature XRD and dielectric spectroscopy. The features of this compositional crossover are discussed in comparison with that observed in the Bi 1-x La x FeO 3 system.

Experimental
High-purity Bi 2 O 3 , La 2 O 3 , Fe 2 O 3 and Sc 2 O 3 were used as starting reagents to prepare the Bi 1-x La x-Fe 0.5 Sc 0.5 O 3 compositions with x D 0.30 and 0.35. Previously calcined oxides were mixed in the stoichiometric ratio, ball milled in acetone, dried and pressed into pellets. The pellets were heated in a closed alumina crucible at 1140 K for 10 min followed by a quenching down to room temperature to be then used for the high-pressure synthesis/sintering.
Since the crystal structure of the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 phases is certainly very sensitive to x, it is a fundamental issue to ensure the proper chemical composition in the precursors and, thereby, in the ceramics. The sol-gel method is known to offer opportunity to obtain ultra-homogeneous complex oxide materials. Therefore, the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 precursors (x D 0.30-0.35, step 0.01) were also prepared via a sol-gel combustion route using nitrates of the respective metals. The appropriate volumes of the nitrates were mixed together in a beaker, and complexing agent (citric acid monohydrate) was then added to the above solution in different molar ratios according to propellant chemistry. The resulted solution was mixed for 1 h at 333 K and then concentrated by evaporation of solvent at 373 K until it turned into a brown viscous gel. Finally, the resulted gel was heated up to about 670 K to initiate combustion reaction. The obtained ultra-fine powders were calcined in air at 870 K for 2 h to remove organic remains. High-power sonication of the powders was performed in ethanol in the cavitation regime using a water-cooled 30-ml cell. The ultrasonic generator power was 0.5 kW, and the treatment time was 20 min. The dried powder was pressed into pellets and used for the high-pressure synthesis/sintering without any additional thermal treatment.
High pressure was generated using an anvil press DO-138A with a press capacity up to 6300 kN. The ceramic samples were prepared at 6 GPa for 1-2 min. In order to produce ceramics from the thermally treated mixture of oxides, one needed the synthesis/sintering temperature to be in the range of 1420-1520 K, while in the case of the sol-gel prepared precursor, the temperature of 1020-1120 K was enough.
Powder XRD study of the precursors and the sintered samples was performed using a PANalytical XPert MPD PRO diffractometer (Ni-filtered CuKa radiation, PIXcel 1D detector, and the exposition corresponded to about 2 s per step of 0.02 o over the angular range of 15 o -100 o ) at room temperature. In situ XRD measurements were conducted in an Anton Paar high-temperature chamber (HTK 16N) in a temperature range between 300 and 870 K. Rietveld refinements of the diffraction data have been performed using JANA 2006 software [25]. For the x D 0.35 composition, the room temperature structural parameters were refined from both X-ray and neutron diffraction data (see Ref. [22] for details of the neutron diffraction experiment) revealing their excellent agreement.
The microstructure of the fractured surface and local chemical composition of the obtained ceramics was studied by scanning electron microscopy (SEM, Hitachi S-4100) equipped with an energy dispersive spectroscopy (EDS) detector.
Dielectric measurements were performed in the temperature range of 100-500 K and at frequencies from 20 Hz to 2 GHz using an LCR meter (HP 4284A) and a Vector Network Analyser (Agilent 8714ET). The ceramic samples were polished to obtain discs of about 0.5 mm thick and then electroded with silver paste.

Results and discussion
Attempts to prepare a single-phase Bi 1-x La x Fe 0.5 Sc 0.5 O 3 perovskite solid solution with x D 0.30 using the conventional ceramic route were unsuccessful. Two main phases were detected in the product of thermal treatment of the stoichiometric mixture of oxides, namely a perovskite phase derived from BiFeO 3 with the rhombohedral distortion of the crystal lattice and a phase based on either gor b-modifications of Bi 2 O 3 (depending on the treatment temperature) [18]. Direct synthesis under high pressure from the oxide mixture without thermal treatment has resulted in no single-phase product: along with the perovskite phase, a monoclinic phase based on bismuth oxide was present [18]. Single-phase perovskite ceramics with x D 0.30 have been prepared by means of high-pressure synthesis/sintering of the thermally treated oxide mixture as indicated in the 'Experimental' section.
It has been found from the analysis of the XRD pattern of these ceramics that the symmetry of the as-prepared phase with x D 0.30 is the same as that of the phase with x D 0.20, namely an incommensurately modulated antipolar phase that can be described using the Imma(00g)s00 superspace group [21]. Thus, the range of interest can be specialized: in the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 system, the crossover from antipolar to non-polar phase occurs in the range between x D 0. 30  In situ XRD study has revealed that the antipolar phase of the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 solid solution with x D 0.30 transforms on heating into the non-polar Pnma phase which is similar to that detected in the as-prepared samples of the x D 0.35 composition [22]. The transformation was found to be reversible. On heating, the Imma(00g)s00 ! Pnma transition begins at about 450 K; on cooling, the Pnma ! Imma(00g)s00 transition is complete at about 440 K. Beginning (ending) of the transition was determined by appearance (disappearance) of the weak commensurate reflections (111) and (210) characteristic of the non-polar phase (Figure 1).
Dielectric measurements confirmed the transition in the ceramics with the x D 0.30 composition (Figure 2). The obtained temperature dependences of complex dielectric permittivity for Bi 0.7 La 0.3-Fe 0.5 Sc 0.5 O 3 exhibit anomalous behaviour at about 460 K, clearly visible at frequencies above 20 MHz. This anomaly is associated with the antipolar-non-polar phase transition which was observed in the in situ XRD study. At lower frequencies, rather high dielectric losses indicate contribution of conductivity and no polar nature of the material can be observed.
It is well known that the materials based on BiFeO 3 are characterized by a high loss caused by the electronic structure of this compound (intrinsic loss). However, in our ceramics, an extrinsic contribution to the loss, i.e. loss associated with the microstructure is certainly considerable. It was found from the high-pressure experiments that homogeneous ceramics of the satisfactory quality can be prepared only in a very narrow temperature range. When the synthesis/sintering temperature is not high enough, the obtained material is very porous and poorly sintered. Application of the temperature that exceeds the range results in formation of a considerable amount of liquid phase in which the big grains grow up. Typical microstructure of the fractured surface of the high-pressure prepared ceramics is shown in Figure 3(a). The relatively thick inter-granular layers and the gaps between the grains are seen.
In order to produce more homogeneous and dense ceramics at lower synthesis/sintering temperature, the advanced methods of precursor preparation were used. XRD analysis of the Bi 1-x La x-Fe 0.5 Sc 0.5 O 3 powders with x from 0.30 to 0.35 (step 0.01) prepared using the sol-gel (combustion) method followed by annealing at 870 K revealed that the perovskite phase is the only phase present. However, it was impossible to determine the symmetry of the phase since the diffraction peaks were broadened (Figure 4). This broadening can be caused either by poor crystallization or small crystallite size.
It was found that a high-power sonication of the sol-gel prepared powders has a positive effect: the diffraction reflections of the perovskite phase became sharper and better resolved. Such an effect suggests that the observed peak broadening should be associated with poor (incomplete)   crystallization rather than with small size of the crystallites. High-power sonication (which is a kind of mechanoactivation) certainly promotes further crystallization of the sol-gel prepared powders. This phenomena deserves a particular study which is out of the scope of this paper. Although the obtained powders were single-phase perovskite, the attempts to sinter them into homogeneous ceramics under ambient pressure were unsuccessful. Different sintering temperatures above 870 K were tested; however, the obtained product was either not sintered or a non-single phase. The sintered material was found to have a similar phase composition as the product of synthesis from the mechanical oxide mixture (as mentioned at the beginning of 'Results and Discussion' section). It turned out that the powders obtained through a sol-gel route followed by a high-power sonication can be sintered into single-phase ceramics under high pressure at the temperatures that are 300-400 K lower that those applied when preparing ceramics from the thermally treated mixed oxide precursors. Comparison of XRD patterns showed that the perovskite phases of the ceramic samples of the same nominal composition (with either x D 0.30 or x D 0.35, see 'Experimental' section) are identical in terms of crystallinity and the lattice parameters in spite of both different ways of the precursor preparations and different temperatures of the high-pressure treatment. However, the ceramics produced from the sol-gel prepared precursors was found to be more homogeneous and denser: with thinner inter-granular layers and smaller gaps between the grains (cf. Figure 3(a,b)). An EDS mapping revealed a more homogeneous distribution of iron and scandium over the grains and the inter-granular layers.
It was found from the analysis of the XRD patterns of these ceramics that the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 perovskite phase is an incommensurate antipolar Imma(00g)s00 when x 0.33 and a non-polar Pnma when x 0.34. Figure 5 shows the compositional dependence of the normalized unit-cell volume in the range from 0.30 to 0.35. No region of phase coexistence has been detected. In this respect, the composition-phase behaviour of the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 perovskite solid solutions in the range of crossover from antipolar phase to non-polar one is a striking departure from that generally observed in the perovskite systems derived from bismuth ferrite. The BiFeO 3 -based solid solution systems where bismuth is substituted by a rear earth element demonstrate a wide-range phase coexistence at the antipolar to non-polar compositional crossover [26][27][28]. Particularly, in the Bi 1-x La x FeO 3 system, such a crossover occurs over the range of at least 10 mol% [29]. Although the lanthanum-substituted BiFeO 3 perovskites are the most studied among the system based on bismuth ferrite, there are several different opinions on nature of coexisting phases in this range [26,[29][30][31][32][33][34]. Overlapped diffraction peaks in XRD patterns result in difficulties/uncertainties in identification of the phases. The energy landscape of BiFeO 3 -based compositions consists of several almost degenerate phase states. That is why a phase coexistence is typical of these materials. However, the phase states can be switched by some perturbations. We assume that pressure acts as such a perturbation and makes the composition-phase dependence more certain. The Bi 1-x La x Fe 0.5 Sc 0.5 O 3 phases were sintered under high pressure followed by a quenching down to room temperature. It appears that under pressure, a difference in energies of the phases with different symmetries is more noticeable and one phase or another becomes certainly more favourable.
Detailed dielectric study and investigation of the magneto-electric effect in the Bi 1-x La x-Fe 0.5 Sc 0.5 O 3 ceramics with 0.30 x 0.35 are in progress and will be published elsewhere.

Conclusions
Single-phase perovskite ceramics of the Bi 1-x La x Fe 0.5 Sc 0.5 O 3 composition (0.30 x 0.35) cannot be prepared using a conventional mixed oxide route. A thermal treatment at ambient pressure of a stoichiometric mixture of the respective oxides results in formation of a BiFeO 3 -like rhombohedral perovskite and a phase based on the polymorphic modifications of Bi 2 O 3 .
Bi 1-x La x Fe 0.5 Sc 0.5 O 3 powders (0.30 x 0.35) prepared using the sol-gel combustion method followed by calcination are single-phase although poorly crystallized. High-power sonication of the powders was found to promote further crystallization. This phenomenon deserves a particular study. However, regardless of the precursor preparation method (either from the thermally treated oxide mixture or from the sol-gel derived powders), in order to sinter single-phase ceramics from these powders, the high-pressure/high-temperature technique is needed. At the same time, the powders obtained through a sol-gel route can be sintered into single-phase ceramics under high pressure at the temperatures that are 300-400 K lower that those applied when preparing ceramics from the thermally treated mixed oxide precursors.
The compositional crossover from the incommensurately modulated antipolar phase that can be described using the Imma(00g)s00 superspace group to the non-polar Pnma phase in the Bi 1-x La x-Fe 0.5 Sc 0.5 O 3 system at room temperature occurs between x D 0.33 and 0.34 with no range of phase coexistence. This very narrow compositional range is of a great interest since it corresponds to a solid solution with the maximal lattice-magnetic coupling effect expected.

Disclosure statement
No potential conflict of interest was reported by the authors.