Home
Search
Collections
Journals
About
Contact us
My IOPscience
Neutron diffraction study of the (BiFeO3)1−x (PbTiO3)x solid solution: nanostructured multiferroic system
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 J. Phys.: Condens. Matter 27 046004 (http://iopscience.iop.org/0953-8984/27/4/046004) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 193.255.88.62 This content was downloaded on 08/01/2015 at 12:18
Please note that terms and conditions apply.
Journal of Physics: Condensed Matter J. Phys.: Condens. Matter 27 (2015) 046004 (8pp)
doi:10.1088/0953-8984/27/4/046004
Neutron diffraction study of the (BiFeO3)1−x (PbTiO3)x solid solution: nanostructured multiferroic system ˜ 4 , M Brunelli5 , I V Golosovsky1,2,3 , S B Vakhrushev2,3 , J L Garc´ıa-Munoz 6 6 7 W-M Zhu , Z-G Ye and V Skumryev 1
National Research Center “Kurchatov Institute”, B.P. Konstantinov Petersburg Nuclear Physics Institute, 188300 Gatchina, Russia 2 A.F. Ioffe Physico-Technical Institute RAS, 194021 St. Petersburg, Russia 3 St. Petersburg State Polytechnical University, 29 Politekhnicheskaya, 195251 St. Petersburg, Russia 4 Institut de Ci´encia de Materials de Barcelona, ICMAB-CSIC, E-08193 Bellaterra, Spain 5 Institut Laue Langevin, 6 rue Jules Horowitz, BP 156, F-38042 Grenoble, France 6 Department of Chemistry and 4D LABS, Simon Fraser University, Burnaby V5A 1S6, Canada 7 Instituci´o Catalana de Recerca i Estudis Avancats (ICREA), 08010 Barcelona, Spain E-mail: [email protected] Received 24 November 2014 Accepted for publication 9 December 2014 Published 8 January 2015 Abstract
Neutron diffraction studies performed on the solid solution of (BiFeO3 )1−x (PbTiO3 )x reveal a mixture of two nanoscale phases with different crystal structures: a rhombohedral BiFeO3 -based phase and a tetragonal PbTiO3 -based phase. The ratio of Fe3 + and Ti4 + ions in the two phases is practically constant; only the proportion of the phases changes. The magnetic moments in the BiFeO3 -based phase, in contrast to BiFeO3 , deviate from the basal plane. The temperature evolutions of the spin components along the hexagonal axis and within the perpendicular plane are different, leading to a spin re-orientation transition. The antiferromagnetic order in the PbTiO3 -based phase corresponds to a simple structure with the propagation vector (1/2, 1/2, 1/2). The temperature dependence of the antiferromagnetic moment in the tetragonal phase at x = 0.5 indicates a canted antiferromagnetic order and a net ferromagnetic moment. A strong magnetic coupling between the two constituting phases due to the nanoscale character of the phases and well-developed interface between nanoparticles has been observed. The system of (BiFeO3 )1−x (PbTiO3 )x demonstrates an interesting scenario, where the proximity effects in the unstable system play a crucial role in the appearance of the unusual magnetic properties. Keywords: multiferroics, neutron diffraction, nanostructured system (Some figures may appear in colour only in the online journal)
cancels the macroscopic magnetization and inhibits the linear magnetoelectric effect [1, 2]. This results in a weak response to the magnetic field, which is one of the main factors obstructing the practical applications of this material. BiFeO3 also has other drawbacks, in particular, a high electric conductivity, which causes strong leakage of charges, and a high electric coercive field. It had been realized that a solid solution with ferroelectric PbTiO3 can stabilize the distorted perovskite phase and forms a morphotropic phase boundary (MPB) due to the difference in
1. Introduction
The solid solution of (BiFeO3 )1−x (PbTiO3 )x (BF–PT) has recently attracted attention as a promising multiferroic material with strong coupling between magnetic and ferroelectric orders. Indeed, BiFeO3 is probably the most studied multiferroic, since it is the only known oxide which is antiferromagnetic and ferroelectric at room temperature, with the Ne´el temperature TN = 640 K and the Curie temperature TC = 1123 K. The spiral spin modulation in BiFeO3 0953-8984/15/046004+08$33.00
1
© 2015 IOP Publishing Ltd Printed in the UK
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
symmetry between BiFeO3 (R3c) and PbTiO3 (P4mm) [3–5]. Within the MPB, between the tetragonal and rhombohedral phases, the coercive field was expected to be reduced via ionic substitution of Ti4+ for Fe3+ and this is indeed what was found [6]. The dielectric and ferroelectric properties have been significantly improved in the BF–PT solid solution of the MPB compositions by appropriate aliovalent ionic substitution of Ti4+ for Fe3+ [7–9]. The electric polarisation appearing around 632 ◦ C [10] and the ferroelectric hysteresis loop makes the BF– PT system a high-temperature high-performance ferroelectric material. The BF–PT solid solution was first prepared by Y N Venevstev et al [11] and later by several other researchers who studied its crystallographic structure and electric properties [3–5, 12, 13]. Adding PbTiO3 to BiFeO3 not only changes the electric properties of the latter but also affects the magnetic properties due to the introduction of diamagnetic Ti4+ ions onto the perovskite structure and the subsequent change in crystallographic structure. This modifies the magnetic interaction by changing the geometric arrangement of the magnetic ions. Ti4+ dilution was expected to lead to a decrease in TN [6]. Structural investigations of the BF–PT within the MPB give rather controversial results. Initially it was proposed that the rhombohedral phase transforms to a tetragonal phase through a cubic phase [4, 14]. The phase transformations from a rhombohedral (BiFeO3 ) to cubic and finally to a tetragonal structure (BaTiO3 ) with increasing titanate concentration was reported for the nanostructured BF–PT [15]. In structural studies a 2-phase model was used, a rhombohedral phase with the space group R3c and a tetragonal phase with the space group P4mm [5, 16]. It was later proposed that there is a mixture of rhombohedral, orthorhombic and tetragonal phases within the MPB in the BF–PT [6, 8]. Recently it was reported that instead of an orthorhombic phase, there is in fact exists a monoclinic phase [17]. In contrast to the crystal structure the magnetic order in the BF–PT has not been investigated thoroughly and results are sketchy and controversial [18–20]. Here we report the results of a neutron powder diffraction study of temperature evolution of magnetic and atomic structure at some characteristic compositions of the (BiFeO3 )1−x (PbTiO3 )x phase diagram inside and outside the MPB.
of (BiFeO3 )1−x (PbTiO3 )x solid solution was obtained. In this paper the parameter x indicates the nominal composition. Neutron diffraction experiments were performed at D20 and D2B diffractometers of the Institute Laue–Langevin (Grenoble, France) with a neutron wavelength of 1.87 Å and 1.59 Å, respectively at temperatures up to about 560 K. Because of the limited beam-time we focused on samples of the most characteristic compositions, namely, x = 0.2; 0.24; 0.28; 0.3; 0.4 and 0.5. In the complementary magnetization study, a Quantum Design MPMS SQUID magnetometer was used. The neutron diffraction patterns were treated using the Fullprof package [21]. In some complex cases, the decomposition of the characteristic fragments of the diffraction pattern into independent peaks was used. In all cases the peak shape was described by the pseudo-Voigt function in the Thompson–Cox–Hastings approximation [22]. Profile analysis of the neutron diffraction patterns for all compositions showed that the peaks are substantially broadened in respect with the instrumental width, which is a signature of nanostructured objects. In such a case the physical resolution is defined by the size-effect, not the instrumental resolution. Large peak widths do not allow detecting of any crystal structure distortions [17] and the data treatment was carried out in the frame of a simplest model assuming a mixture of two nanoscale phases with rhombohedral and tetragonal symmetry. Estimations from the Scherer formula in isotropic approximation give characteristic sizes of the nanoparticles: 15–30 nm for the tetragonal phase and 30–50 nm for the rhombohedral phase. It is worth noting that the nanostructured form of (BiFeO3 )1−x (PbTiO3 )x system had been already reported, but with larger characteristic dimensions of grains— about a few hundred nm [26]. The refinement demonstrated that the full width at half maximum (FWHM) of the diffraction peaks does not follow the expected monotonic dependence on momentum transfer Q. To calculate this dependence we used an independent refinement of the FWHM of each peak. This approach is not supported by the standard Fullprof code therefore we used simple spectrum decomposition. It can be seen in figure 1 from the difference between observed and calculated profiles (blue line), that the independent refinement of the FWHM significantly improves the fit. Unfortunately such calculation is impossible to perform for multiphase compositions because of strong peak overlapping. Therefore we carried out it for only utmost compositions—BiFeO3 -based phase and PbTiO3 -based phase. The results were similar. In figure 2 the integral breadth β(Q), which is proportional to the FWHM [23, 24], is shown for BiFeO3 -based phase. Non-monotonic dependence of FWHM is characteristic of the so called ‘anisotropic size-effect’, which arise when a shape of nanoparticle cannot be approximated by a sphere [25]. In such a case, the ‘apparent size’—the nanoparticle dimension along the scattering vector [23]—differs for the reflections with the same inter-planar distance. Consequently a power averaging leads to an irregular behaviour of the FWHM with moment transfer Q.
2. Experimental details and data treatment
Polycrystalline samples of (BiFeO3 )1−x (PbTiO3 )x (with composition x 0.2) were synthesized by multi-step solid state reactions and a sintering process [6]. High-purity oxide reagents were mixed in stoichiometric amounts, pressed into a pellet and calcined at 800 ◦ C for 2 h. After calcination, the samples were reground and pressed into pellets in the presence of a few drops of the aqueous polyvinyl alcohol. The pellets were subsequently heated at 700 ◦ C for one hour to eliminate the PVA binder and then sintered in a sealed platinum crucible at temperatures ranging from 1000 ◦ C (for x = 0.20) to 1120 ◦ C (for x = 0.90), depending on the composition. Xray diffraction pattens confirmed that the pure perovskite phase 2
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
(a)
(b)
Figure 2. The refined peak broadening (integral breadth) β versus momentum transfer Q for the compound with x = 0.24. Red line is the instrumental resolution. The error bars correspond to the estimated standard deviation (e.s.d.); if not shown, they do not exceed the symbol size.
their intensities do not change with temperature. We did not observe any superstructure peaks due to the antiphase rotation of the octahedra reported in [13] most likely due to nanoscale character of the system. 3. Results 3.1. Structure and stoichiometry Figure 1. Profile of the sample with x = 0.24 measured at 2 K. This sample is practically single phase: weight fractions of the rhombohedral BiFeO3 -based phase is of 97.6 %. (a) isotropic model, (b) fit with individual FWHM for every reflection. In blue—the difference between observed and calculated profiles is shown.
Due to the opposite sign of the neutron scattering lengths of Fe and Ti their difference is very large, of 1.29·10−12 cm. This means large diffraction contrast. So in spite of zero contribution of Ti(Fe) to the total structure factor, the variation of Ti(Fe) in the latter is sensitive to the mutual content of Ti and Fe. As result, assuming that there are no defects on the Ti(Fe) site, Ti and Fe content were calculated with reasonable accuracy (see figure 3(a)). To maintain electroneutrality, the substitution of Ti4+ for 3+ Fe in the system (Bi3+ Fe3+ O3 )1−x (Pb2+ Ti4+ O3 )x requires the coupled substitution of Pb2+ for Bi3+ . However, the contrast between Bi and Pb is 0.09·10−12 cm, much smaller than between Ti and Fe. Refinement demonstrates the appearance of Pb ions on the Bi sites in the rhombohedral phase as well as the appearance of Bi ions on the Pb site in the tetragonal phase; however, independent determination of Bi and Pb was not possible. Refinement does not reveal oxygen vacancies within an accuracy of 3–4%. Assuming that the oxygen sites are fully occupied, which is expected from the nominal formula, and that there is no ‘position splitting’ for Fe and Ti at the B position of the pseudo-perovskite cell [27], the refined content of Ti(x) is shown in figure 3(a). Regarding the refined Ti amounts and assuming that there are no Fe4+ ions [28], from electroneutrality, the chemical formulas of the constituents could be written as (Bi0.8 Pb0.2 )(Fe0.8 Ti0.2 )O3 for the rhombohedral, BiFeO3 based phase, and (Pb0.5 Bi0.5 )(Fe0.5 Ti0.5 )O3 for the tetragonal, PbTiO3 -based phase.
It is well known that the peak broadening originates from a size effect as well as from inner stresses. The size-effect contribution is Q-independent, while the inner stresses give rise to a contribution proportional to Q. In figure 2, which full analogue of the Williamson–Hall plot, no systematic increase in β(Q) with Q is seen. This means that the peak broadening arises mainly from the size effect, while the inner stresses are small and unlikely to play a dominant role. We described the diffraction profiles for all compositions assuming anisotropic nanoparticle shape in the frame of an uniaxial model implemented in the Fullprof. In fact this model is rough, however, in contrast to the isotropic model it gives a better description of the peak profiles in the entire diffraction pattern, providing a better accuracy of the refined structural parameters. The goodness of fit (R-factor) is strongly dependent on the composition. For compositions with dominant BiFeO3 based phase the reduction by using the anisotropic shape of the nanoparticles is slight: 4.7 to 4.5, while for compositions with dominant PbTiO3 -based phase the reduction is substantial— from 12 to 8. There are some small unidentified reflections in the pattern, which most probably arise from impurities, since 3
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
(a)
Figure 4. Unit cell parameter ah for the rhombohedral phase (in hexagonal setting) (triangles); unit cell parameters for the tetragonal phase at (circles) and ct (squares) versus x. Closed symbols are from neutron diffraction; open symbols and dash lines, which indicate MPB, are from the x-ray measurements [6].
with the percentage of lead titanate up to ∼ 840 Å in the sample (BiFeO3 )0.75 (PbTiO3 )0.25 [30]. In our case the nanoparticles with rhombohedral structure have characteristic sizes of about 30–50 nm, so incommensurate magnetic structure cannot exist. However, in contrast with BiFeO3 , where the spins are aligned on the plane perpendicular to the hexagonal axis [29, 31], in the substituted BiFeO3 -based phase the spins are out of the basal plane. For the space group R3c with the propagation vector k = (0, 0, 0), the z-projections of the moments, along hexagonal axis, and the x- and y-projections in the perpendicular plane, correspond to different magnetic irreducible representations—one- and 2D representations, respectively [2, 18]. So, the observed magnetic structure includes the magnetic modes from different irreducible representations. In figure 5 the temperature dependences of the magnetic moment and its deviation from the basal plane for the BiFeO3 based samples with x = 0.20 and x = 0.24 are shown. It is seen that the temperature dependence of the total magnetic moment for the two samples is practically the same. However, in the sample with x = 0.24, a spin re-orientation takes place. As the temperature decreases, the component along hexagonal axis, which corresponds to a 90◦ out-of-plane deviation, appears first. When the temperature further decreases, the moment leans closer to the basal plane.The observed smooth spin re-orientation can be associated with the wide maximum in the magnetic susceptibility measurements [6]. No visible anomaly in the temperature dependence of the unit cell parameters for the sample with x = 0.24 was detected. Magnetic order in the tetragonal PbTiO3 -based phase was found to be a simple two-sublattice antiferromagnetic structure with the doubling of the chemical cell, similar to that reported for pure PbFeO3 [32] and for the BT–PT system [19, 20]. Statistical accuracy was not good enough to define with confidence the magnetic moment orientation. However, since in the neutron diffraction only the component of the moment perpendicular to the scattering vector is measured, the observation of the magnetic reflection (1/2 1/2 1/2) means
(b)
Figure 3. (a) The relative amount of Ti in the constituent phases versus x in the system (BiFeO3 )1−x (PbTiO3 )x . (b) Weight fractions of the rhombohedral, BiFeO3 -based phase and of the tetragonal, PbTiO3 -based phase. The densities of these phases are close: 8.3 g cm−3 and 7.9 g cm−3 , respectively, therefore the weight and volume ratios practically coincide. The error bars correspond to the estimated standard deviation (e.s.d.), if not shown do not exceed the symbol size.
It can be seen that the ratio of Fe over Ti (figure 3(a)), and, hence, that of Bi over Pb in the constituting phases are practically constant for all the studied compositions within our statistical accuracy. Only the relative fractions of these phases change, most significantly within the composition range of x = 0.28–0.30 (figure 3(b)), i.e. within the MPB. From figure 3(b) it follows that the MPB width is about x = 0.14. In figure 4 the unit cell parameters refined from neutron diffraction experiments are shown with those obtained from x-ray measurements [6]. It can be seen that the unit cell parameters in the range of x = 0.2–0.5 vary slightly. Therefore, one could conclude that the crystal structure of the two constituents changes insignificantly. 3.2. Magnetic order
Magnetic moments were obtained from the standard sequential Fullprof profile refinement. In this procedure the refinement at definite temperature starts from the parameters obtained at the previous temperature point. We fixed these parameters in which temperature changes were less than the statistical error. Our instrumental resolution was not high enough to detect any visible splitting of the antiferromagnetic reflections— characteristic of incommensurate magnetic structure in the BiFeO3 -based phase. It is known that the period of the incommensurate structure is ∼ 620 Å [29] and it increases 4
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
(a)
Figure 6. Low-angle fragment of the neutron diffraction pattern for the sample with x = 0.3, which is a mixture of 60% of BiFeO3 -based phase and 40% of PbTiO3 -based phase, measured at 2 K. Inset—the magnetic structure of PbTiO3 -based phase.
(b)
Figure 5. Temperature dependences of the magnetic moment (black circles). In red—the moment deviation from the basal plane for samples with x = 0.20 (a); x = 0.24 (b). The refined content of phases is shown in the panels. The solid line is a fit with a power law, which one should consider as a guide for the eye.
that magnetic moments cannot be aligned along the [1 1 1] axis. Indeed, the best fit corresponds to the model with the moments lying in a plane perpendicular to the [1 1 1] direction (inset in figure 6). In contrast with the single phase samples beyond the MPB, the extraction of the magnetic moments from the diffraction data for the two-phase samples within the MPB is a more delicate task because of the overlapping of the magnetic peaks of the different phases. In such cases, the magnetic moments are calculated from the decomposition of the low angle fragment of the neutron diffraction pattern, fixing some parameters refined from full profile analysis, in particular a lattice parameter, which gives a more reliable result (figure 6). The temperature dependences of the moments in the two coexisting phases for the sample with x = 0.30, within the MPB, which is a mixture of 60% rhombohedral and 40% tetragonal phases, are shown in figure 7. The beginning of the spin re-orientation in the BiFeO3 -based phase, when the magnetic moments start to tilt towards the hexagonal axis, is surprisingly close to the N´eel temperature of the PbTiO3 -based phase. Obviously these transitions are coupled to each other. Indeed, in contrast to the sample with x = 0.24, where no trace of spin re-orientation at the temperature dependence of the unit cell parameters was detected, in the sample within the MPB composition with x = 0.3 the corresponding peculiarities are clearly seen in both constituents (figure 8).
Figure 7. Temperature dependences of the total magnetic moment (in black), its deviation from the basal plane (in red) for BiFeO3 -based phase and the magnetic moment of PbTiO3 -based phase (in blue) for the sample with x = 0.3. The solid line is a fit with a power law, which one should consider as a guide for the eye.
The temperature dependence of the intensity of the magnetic reflection (1/2 1/2 1/2) for the sample with x = 0.5, on the right-side of the MPB, is shown in figure 9(a). In the temperature range of 40 K–220 K it can be well fitted by a straight line. The noticeable deviation from the straight line at the highest temperatures is explained by the contribution from overlapping magnetic reflections from the BiFeO3 -based phase. This phase has N´eel temperature of about 600 K, much higher than the N´eel temperature of PbTiO3 -based phase. In other words the magnetic order in the BiFeO3 -based phase still exists, when the magnetic order in the PbTiO3 -based phase disappears. Note that a similar ‘tail’ in the temperature dependence of magnetic moment had been reported for the tetragonal phase of (BiFeO3 )0.7 (PbTiO3 )0.3 with a 90% fraction of tetragonal phase, and it was tentatively attributed to the short range order [20]. 5
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
(a)
(a)
(b)
(b)
Figure 9. (a) Temperature dependence of the intensity of magnetic reflection (1/2 1/2 1/2) for the sample with x = 0.5. (b) Temperature dependence of susceptibility measured at 50 Oe after zero-field-cooling (ZFC) and field-cooling (FC) procedure. In inset—the magnetisation loop, measured at 2 K. The error bars do not exceed the symbol size.
Figure 8. The temperature dependences of the unit cell parameters in the BiFeO3 -based phase (a) and the PbTO3 -based phase (b) for the sample with x = 0.3. The error bars do not exceed the symbol size.
3.3. Ferromagnetic moment
It is interesting to note that the net ferromagnetic moment occurs at about 40 K where a maximum in the temperature dependence of the antiferromagnetic moment appears (figure 9(a)). This temperature is noticeably above the temperature of magnetic susceptibility anomaly of 23 K (figure 9(b)). Figure 10 shows a peculiarity in the temperature dependences of the unit cell parameters at 25 K, close to the maximum in the magnetic susceptibility. It seems that the emergence of the ferromagnetic moment is accompanied by a crystal structure distortion.Unfortunately, the errors of the refined oxygen positions appeared to be rather big, thus preventing the detection of any anomaly in the bond length/angles.
Magnetic measurements reveal net ferromagnetic moment only in the samples with x around 0.5, as shown by the magnetic hysteresis loop measurements [6]. Indeed, our magnetic measurements in the compound with x = 0.5 also demonstrate clear anomaly around 23 K (figure 9(b)) and hysteresis at low temperatures, indicating the existence of a net moment (inset in figure 9(b)). In neutron diffraction, the net ferromagnetic component caused by non-collinear structure gives contribution into the nuclear reflections. However, the refinement of the neutron diffraction patterns did not reveal any ferromagnetic moment within an accuracy of 0.5 µB in all samples studied. The temperature dependence of the integrated intensities of nuclear reflections does not demonstrate any anomalies, which could be thought as the emergence of a weak ferromagnetic moment. However, in the sample with x = 0.5, a decrease in the magnetic intensity of the reflection (1/2 1/2 1/2), which is proportional to the antiferromagnetic component of the total moment, was observed below about 40 K (figure 9(a)). Since the total magnetic moment must be preserved, the drop in the antiferromagnetic component means a spin canting, corroborates the occurrence of a spin canting and a small ferromagnetic component. It is worth noting, that in contrast to previous works [15, 33] in our experiments a ferromagnetic moment was detected reliably only for x = 0.5, but not in the whole system.
4. Discussion
Our neutron diffraction experiments demonstrated that the substitution of Ti4+ for Fe3+ leads to several apparent effects. First, it results in an expected decrease in the N´eel temperature, from 640 K for BiFeO3 down ∼610 K for BiFeO3 -based phase. The same was observed for the PbTiO3 -based phase; TN was estimated to be about 220–250 K, much lower than TN = 605 K reported for PbFeO3 [32]. Second, M¨ossbauer experiments on the sample (Bi0.8 Pb0.2 )FeO2.9 showed ferromagnetic order at room temperature [28]. In our BiFeO3 -based sample (Bi0.8 Pb0.2 ) 6
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
TN of the rhombohedral BiFeO3 -based phase. Therefore, at the transformation from the tetragonal to the rhombohedral phase under external pressure at corresponding temperature, the paramagnetic state of the system can change to an antiferromagnetic state. Such a transformation in our case should be expected for the sample with x = 0.5, which is paramagnetic at room temperature (see figure 9(a)), but not with x = 0.3–0.35. It follows from our experiments that the crystal structure and chemical formula of the BiFeO3 -based and PbTiO3 -based phases change slightly in the studied range of x = 0.2– 0.5 (figure 4). Nevertheless, these small variations result in remarkable changes in the magnetic properties. Thus, on the left-side of the MPB, the practically single phase samples with x = 0.20 and x = 0.24 have similar lattice parameters and temperature dependences of the magnetic moment. However, in the sample with x = 0.24, in contrast to the sample with x = 0.2, a new phenomenon—spin reorientation—appears (figures 5(a) and (b)). Most probably, the difference observed in the magnetic behavior is caused by small structure variations which are undetectable by neutron diffraction. Such changes may lead to different temperatures of ordering for spin components within the basal plane and along the hexagonal axis. The strong response to small variation in the structure means a high instability of the BiFeO3 -based phase left of the MPB. From the comparison of the N´eel temperatures for the single phase PbTiO3 -based sample with x = 0.5 and the twophase sample with x = 0.3, where the PbTiO3 -based phase is 40% only, it can be seen that the TN changes drastically: from ∼ 250 K to 550 K (see figures 7 and 9(a)). At the same time, profile refinement demonstrates that the unit cell parameters in these samples vary slightly: a changes from 3.860(5) Å to 3.844(5) Å, while b changes from 4.325(5) Å to 4.375(5) Å. We believe that such a large TN increase can be explained by the proximity effect only. The 60% of the rhombohedral BiFeO3 -based phase with a high N´eel temperature, biases the 40% of the tetragonal PbTiO3 -based phase with a low N´eel temperature. Another evidence for strong magnetic coupling between the constituting phases follows from the closeness of the temperature of the spin re-orientation transition in the BiFeO3 based phase and the magnetic ordering temperature in the PbTiO3 -based phase (figure 7). We argue that this is not a coincidence because anomalies in the temperature dependences of the unit cell parameters in both constituents (figure 8) exist at this particular temperature. Perhaps this manifests a simple fact, that the interaction is maximal at a specific mutual arrangement of the moments of the constituent phases. The observed proximity effects are not surprising, given the nature of the nanoscale phases, which involve a welldeveloped interface between nanoparticles with different magnetic orders. It is worth noting that one should expect the proximity effects, namely, the exchange bias in all systems where nanoscale regions with different magnetic orders coexist. For example, such effects were observed in phase-separated manganites and cobaltites (see work [36] and the references therein). In these compounds
(a)
(b)
Figure 10. Temperature dependence of the unit cell parameters (a)—(panel a) and (b)—(panel b) in the tetragonal phase of the sample with x = 0.5. The error bars do not exceed the symbol size.
(Fe0.8 Ti0.2 )O3 , which differs by the presence of Ti ions, net ferromagnetic moment was not detected. This means that the substitution of 20% of Ti4+ for Fe3+ suppresses any spin canting, i.e. the appearance of a net ferromagnetic moment. Third, the substitution of Ti4+ for Fe3+ stabilizes the tetragonal structure to the right of the MPB. It is worth noting, that the unstable oxide PbFeO3 , which can be synthesized under 7 GPa and at 1100–1200 ◦ C only, has an orthorhombic structure [34]. Analysis of the results demonstrates clear discrepancies in terms of structure and phase instability of the BF–PT solid solution system. First, the structures reported by different laboratories strongly differ. For example, the compound (BiFeO3 )0.7 (PbTiO3 )0.3 was reported to have single phase tetragonal structure [13, 16, 27], but also a mixture of two phases: tetragonal and monoclinic [17]. In our case we see as well two phases: but however they have been identified as tetragonal and rhombohedral ones. The same holds true for the width and the position of the MPB. The data on magnetic behaviour are partially contradictory as well. For example, in our study the spin reorientation occurs at the temperature 550 K, while in the work of Bhattacharjee et al [33] the spin re-orientation transition was observed around 460 K. It was recently reported that at room temperature on application of a moderate pressure (
Search
Collections
Journals
About
Contact us
My IOPscience
Neutron diffraction study of the (BiFeO3)1−x (PbTiO3)x solid solution: nanostructured multiferroic system
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 J. Phys.: Condens. Matter 27 046004 (http://iopscience.iop.org/0953-8984/27/4/046004) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 193.255.88.62 This content was downloaded on 08/01/2015 at 12:18
Please note that terms and conditions apply.
Journal of Physics: Condensed Matter J. Phys.: Condens. Matter 27 (2015) 046004 (8pp)
doi:10.1088/0953-8984/27/4/046004
Neutron diffraction study of the (BiFeO3)1−x (PbTiO3)x solid solution: nanostructured multiferroic system ˜ 4 , M Brunelli5 , I V Golosovsky1,2,3 , S B Vakhrushev2,3 , J L Garc´ıa-Munoz 6 6 7 W-M Zhu , Z-G Ye and V Skumryev 1
National Research Center “Kurchatov Institute”, B.P. Konstantinov Petersburg Nuclear Physics Institute, 188300 Gatchina, Russia 2 A.F. Ioffe Physico-Technical Institute RAS, 194021 St. Petersburg, Russia 3 St. Petersburg State Polytechnical University, 29 Politekhnicheskaya, 195251 St. Petersburg, Russia 4 Institut de Ci´encia de Materials de Barcelona, ICMAB-CSIC, E-08193 Bellaterra, Spain 5 Institut Laue Langevin, 6 rue Jules Horowitz, BP 156, F-38042 Grenoble, France 6 Department of Chemistry and 4D LABS, Simon Fraser University, Burnaby V5A 1S6, Canada 7 Instituci´o Catalana de Recerca i Estudis Avancats (ICREA), 08010 Barcelona, Spain E-mail: [email protected] Received 24 November 2014 Accepted for publication 9 December 2014 Published 8 January 2015 Abstract
Neutron diffraction studies performed on the solid solution of (BiFeO3 )1−x (PbTiO3 )x reveal a mixture of two nanoscale phases with different crystal structures: a rhombohedral BiFeO3 -based phase and a tetragonal PbTiO3 -based phase. The ratio of Fe3 + and Ti4 + ions in the two phases is practically constant; only the proportion of the phases changes. The magnetic moments in the BiFeO3 -based phase, in contrast to BiFeO3 , deviate from the basal plane. The temperature evolutions of the spin components along the hexagonal axis and within the perpendicular plane are different, leading to a spin re-orientation transition. The antiferromagnetic order in the PbTiO3 -based phase corresponds to a simple structure with the propagation vector (1/2, 1/2, 1/2). The temperature dependence of the antiferromagnetic moment in the tetragonal phase at x = 0.5 indicates a canted antiferromagnetic order and a net ferromagnetic moment. A strong magnetic coupling between the two constituting phases due to the nanoscale character of the phases and well-developed interface between nanoparticles has been observed. The system of (BiFeO3 )1−x (PbTiO3 )x demonstrates an interesting scenario, where the proximity effects in the unstable system play a crucial role in the appearance of the unusual magnetic properties. Keywords: multiferroics, neutron diffraction, nanostructured system (Some figures may appear in colour only in the online journal)
cancels the macroscopic magnetization and inhibits the linear magnetoelectric effect [1, 2]. This results in a weak response to the magnetic field, which is one of the main factors obstructing the practical applications of this material. BiFeO3 also has other drawbacks, in particular, a high electric conductivity, which causes strong leakage of charges, and a high electric coercive field. It had been realized that a solid solution with ferroelectric PbTiO3 can stabilize the distorted perovskite phase and forms a morphotropic phase boundary (MPB) due to the difference in
1. Introduction
The solid solution of (BiFeO3 )1−x (PbTiO3 )x (BF–PT) has recently attracted attention as a promising multiferroic material with strong coupling between magnetic and ferroelectric orders. Indeed, BiFeO3 is probably the most studied multiferroic, since it is the only known oxide which is antiferromagnetic and ferroelectric at room temperature, with the Ne´el temperature TN = 640 K and the Curie temperature TC = 1123 K. The spiral spin modulation in BiFeO3 0953-8984/15/046004+08$33.00
1
© 2015 IOP Publishing Ltd Printed in the UK
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
symmetry between BiFeO3 (R3c) and PbTiO3 (P4mm) [3–5]. Within the MPB, between the tetragonal and rhombohedral phases, the coercive field was expected to be reduced via ionic substitution of Ti4+ for Fe3+ and this is indeed what was found [6]. The dielectric and ferroelectric properties have been significantly improved in the BF–PT solid solution of the MPB compositions by appropriate aliovalent ionic substitution of Ti4+ for Fe3+ [7–9]. The electric polarisation appearing around 632 ◦ C [10] and the ferroelectric hysteresis loop makes the BF– PT system a high-temperature high-performance ferroelectric material. The BF–PT solid solution was first prepared by Y N Venevstev et al [11] and later by several other researchers who studied its crystallographic structure and electric properties [3–5, 12, 13]. Adding PbTiO3 to BiFeO3 not only changes the electric properties of the latter but also affects the magnetic properties due to the introduction of diamagnetic Ti4+ ions onto the perovskite structure and the subsequent change in crystallographic structure. This modifies the magnetic interaction by changing the geometric arrangement of the magnetic ions. Ti4+ dilution was expected to lead to a decrease in TN [6]. Structural investigations of the BF–PT within the MPB give rather controversial results. Initially it was proposed that the rhombohedral phase transforms to a tetragonal phase through a cubic phase [4, 14]. The phase transformations from a rhombohedral (BiFeO3 ) to cubic and finally to a tetragonal structure (BaTiO3 ) with increasing titanate concentration was reported for the nanostructured BF–PT [15]. In structural studies a 2-phase model was used, a rhombohedral phase with the space group R3c and a tetragonal phase with the space group P4mm [5, 16]. It was later proposed that there is a mixture of rhombohedral, orthorhombic and tetragonal phases within the MPB in the BF–PT [6, 8]. Recently it was reported that instead of an orthorhombic phase, there is in fact exists a monoclinic phase [17]. In contrast to the crystal structure the magnetic order in the BF–PT has not been investigated thoroughly and results are sketchy and controversial [18–20]. Here we report the results of a neutron powder diffraction study of temperature evolution of magnetic and atomic structure at some characteristic compositions of the (BiFeO3 )1−x (PbTiO3 )x phase diagram inside and outside the MPB.
of (BiFeO3 )1−x (PbTiO3 )x solid solution was obtained. In this paper the parameter x indicates the nominal composition. Neutron diffraction experiments were performed at D20 and D2B diffractometers of the Institute Laue–Langevin (Grenoble, France) with a neutron wavelength of 1.87 Å and 1.59 Å, respectively at temperatures up to about 560 K. Because of the limited beam-time we focused on samples of the most characteristic compositions, namely, x = 0.2; 0.24; 0.28; 0.3; 0.4 and 0.5. In the complementary magnetization study, a Quantum Design MPMS SQUID magnetometer was used. The neutron diffraction patterns were treated using the Fullprof package [21]. In some complex cases, the decomposition of the characteristic fragments of the diffraction pattern into independent peaks was used. In all cases the peak shape was described by the pseudo-Voigt function in the Thompson–Cox–Hastings approximation [22]. Profile analysis of the neutron diffraction patterns for all compositions showed that the peaks are substantially broadened in respect with the instrumental width, which is a signature of nanostructured objects. In such a case the physical resolution is defined by the size-effect, not the instrumental resolution. Large peak widths do not allow detecting of any crystal structure distortions [17] and the data treatment was carried out in the frame of a simplest model assuming a mixture of two nanoscale phases with rhombohedral and tetragonal symmetry. Estimations from the Scherer formula in isotropic approximation give characteristic sizes of the nanoparticles: 15–30 nm for the tetragonal phase and 30–50 nm for the rhombohedral phase. It is worth noting that the nanostructured form of (BiFeO3 )1−x (PbTiO3 )x system had been already reported, but with larger characteristic dimensions of grains— about a few hundred nm [26]. The refinement demonstrated that the full width at half maximum (FWHM) of the diffraction peaks does not follow the expected monotonic dependence on momentum transfer Q. To calculate this dependence we used an independent refinement of the FWHM of each peak. This approach is not supported by the standard Fullprof code therefore we used simple spectrum decomposition. It can be seen in figure 1 from the difference between observed and calculated profiles (blue line), that the independent refinement of the FWHM significantly improves the fit. Unfortunately such calculation is impossible to perform for multiphase compositions because of strong peak overlapping. Therefore we carried out it for only utmost compositions—BiFeO3 -based phase and PbTiO3 -based phase. The results were similar. In figure 2 the integral breadth β(Q), which is proportional to the FWHM [23, 24], is shown for BiFeO3 -based phase. Non-monotonic dependence of FWHM is characteristic of the so called ‘anisotropic size-effect’, which arise when a shape of nanoparticle cannot be approximated by a sphere [25]. In such a case, the ‘apparent size’—the nanoparticle dimension along the scattering vector [23]—differs for the reflections with the same inter-planar distance. Consequently a power averaging leads to an irregular behaviour of the FWHM with moment transfer Q.
2. Experimental details and data treatment
Polycrystalline samples of (BiFeO3 )1−x (PbTiO3 )x (with composition x 0.2) were synthesized by multi-step solid state reactions and a sintering process [6]. High-purity oxide reagents were mixed in stoichiometric amounts, pressed into a pellet and calcined at 800 ◦ C for 2 h. After calcination, the samples were reground and pressed into pellets in the presence of a few drops of the aqueous polyvinyl alcohol. The pellets were subsequently heated at 700 ◦ C for one hour to eliminate the PVA binder and then sintered in a sealed platinum crucible at temperatures ranging from 1000 ◦ C (for x = 0.20) to 1120 ◦ C (for x = 0.90), depending on the composition. Xray diffraction pattens confirmed that the pure perovskite phase 2
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
(a)
(b)
Figure 2. The refined peak broadening (integral breadth) β versus momentum transfer Q for the compound with x = 0.24. Red line is the instrumental resolution. The error bars correspond to the estimated standard deviation (e.s.d.); if not shown, they do not exceed the symbol size.
their intensities do not change with temperature. We did not observe any superstructure peaks due to the antiphase rotation of the octahedra reported in [13] most likely due to nanoscale character of the system. 3. Results 3.1. Structure and stoichiometry Figure 1. Profile of the sample with x = 0.24 measured at 2 K. This sample is practically single phase: weight fractions of the rhombohedral BiFeO3 -based phase is of 97.6 %. (a) isotropic model, (b) fit with individual FWHM for every reflection. In blue—the difference between observed and calculated profiles is shown.
Due to the opposite sign of the neutron scattering lengths of Fe and Ti their difference is very large, of 1.29·10−12 cm. This means large diffraction contrast. So in spite of zero contribution of Ti(Fe) to the total structure factor, the variation of Ti(Fe) in the latter is sensitive to the mutual content of Ti and Fe. As result, assuming that there are no defects on the Ti(Fe) site, Ti and Fe content were calculated with reasonable accuracy (see figure 3(a)). To maintain electroneutrality, the substitution of Ti4+ for 3+ Fe in the system (Bi3+ Fe3+ O3 )1−x (Pb2+ Ti4+ O3 )x requires the coupled substitution of Pb2+ for Bi3+ . However, the contrast between Bi and Pb is 0.09·10−12 cm, much smaller than between Ti and Fe. Refinement demonstrates the appearance of Pb ions on the Bi sites in the rhombohedral phase as well as the appearance of Bi ions on the Pb site in the tetragonal phase; however, independent determination of Bi and Pb was not possible. Refinement does not reveal oxygen vacancies within an accuracy of 3–4%. Assuming that the oxygen sites are fully occupied, which is expected from the nominal formula, and that there is no ‘position splitting’ for Fe and Ti at the B position of the pseudo-perovskite cell [27], the refined content of Ti(x) is shown in figure 3(a). Regarding the refined Ti amounts and assuming that there are no Fe4+ ions [28], from electroneutrality, the chemical formulas of the constituents could be written as (Bi0.8 Pb0.2 )(Fe0.8 Ti0.2 )O3 for the rhombohedral, BiFeO3 based phase, and (Pb0.5 Bi0.5 )(Fe0.5 Ti0.5 )O3 for the tetragonal, PbTiO3 -based phase.
It is well known that the peak broadening originates from a size effect as well as from inner stresses. The size-effect contribution is Q-independent, while the inner stresses give rise to a contribution proportional to Q. In figure 2, which full analogue of the Williamson–Hall plot, no systematic increase in β(Q) with Q is seen. This means that the peak broadening arises mainly from the size effect, while the inner stresses are small and unlikely to play a dominant role. We described the diffraction profiles for all compositions assuming anisotropic nanoparticle shape in the frame of an uniaxial model implemented in the Fullprof. In fact this model is rough, however, in contrast to the isotropic model it gives a better description of the peak profiles in the entire diffraction pattern, providing a better accuracy of the refined structural parameters. The goodness of fit (R-factor) is strongly dependent on the composition. For compositions with dominant BiFeO3 based phase the reduction by using the anisotropic shape of the nanoparticles is slight: 4.7 to 4.5, while for compositions with dominant PbTiO3 -based phase the reduction is substantial— from 12 to 8. There are some small unidentified reflections in the pattern, which most probably arise from impurities, since 3
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
(a)
Figure 4. Unit cell parameter ah for the rhombohedral phase (in hexagonal setting) (triangles); unit cell parameters for the tetragonal phase at (circles) and ct (squares) versus x. Closed symbols are from neutron diffraction; open symbols and dash lines, which indicate MPB, are from the x-ray measurements [6].
with the percentage of lead titanate up to ∼ 840 Å in the sample (BiFeO3 )0.75 (PbTiO3 )0.25 [30]. In our case the nanoparticles with rhombohedral structure have characteristic sizes of about 30–50 nm, so incommensurate magnetic structure cannot exist. However, in contrast with BiFeO3 , where the spins are aligned on the plane perpendicular to the hexagonal axis [29, 31], in the substituted BiFeO3 -based phase the spins are out of the basal plane. For the space group R3c with the propagation vector k = (0, 0, 0), the z-projections of the moments, along hexagonal axis, and the x- and y-projections in the perpendicular plane, correspond to different magnetic irreducible representations—one- and 2D representations, respectively [2, 18]. So, the observed magnetic structure includes the magnetic modes from different irreducible representations. In figure 5 the temperature dependences of the magnetic moment and its deviation from the basal plane for the BiFeO3 based samples with x = 0.20 and x = 0.24 are shown. It is seen that the temperature dependence of the total magnetic moment for the two samples is practically the same. However, in the sample with x = 0.24, a spin re-orientation takes place. As the temperature decreases, the component along hexagonal axis, which corresponds to a 90◦ out-of-plane deviation, appears first. When the temperature further decreases, the moment leans closer to the basal plane.The observed smooth spin re-orientation can be associated with the wide maximum in the magnetic susceptibility measurements [6]. No visible anomaly in the temperature dependence of the unit cell parameters for the sample with x = 0.24 was detected. Magnetic order in the tetragonal PbTiO3 -based phase was found to be a simple two-sublattice antiferromagnetic structure with the doubling of the chemical cell, similar to that reported for pure PbFeO3 [32] and for the BT–PT system [19, 20]. Statistical accuracy was not good enough to define with confidence the magnetic moment orientation. However, since in the neutron diffraction only the component of the moment perpendicular to the scattering vector is measured, the observation of the magnetic reflection (1/2 1/2 1/2) means
(b)
Figure 3. (a) The relative amount of Ti in the constituent phases versus x in the system (BiFeO3 )1−x (PbTiO3 )x . (b) Weight fractions of the rhombohedral, BiFeO3 -based phase and of the tetragonal, PbTiO3 -based phase. The densities of these phases are close: 8.3 g cm−3 and 7.9 g cm−3 , respectively, therefore the weight and volume ratios practically coincide. The error bars correspond to the estimated standard deviation (e.s.d.), if not shown do not exceed the symbol size.
It can be seen that the ratio of Fe over Ti (figure 3(a)), and, hence, that of Bi over Pb in the constituting phases are practically constant for all the studied compositions within our statistical accuracy. Only the relative fractions of these phases change, most significantly within the composition range of x = 0.28–0.30 (figure 3(b)), i.e. within the MPB. From figure 3(b) it follows that the MPB width is about x = 0.14. In figure 4 the unit cell parameters refined from neutron diffraction experiments are shown with those obtained from x-ray measurements [6]. It can be seen that the unit cell parameters in the range of x = 0.2–0.5 vary slightly. Therefore, one could conclude that the crystal structure of the two constituents changes insignificantly. 3.2. Magnetic order
Magnetic moments were obtained from the standard sequential Fullprof profile refinement. In this procedure the refinement at definite temperature starts from the parameters obtained at the previous temperature point. We fixed these parameters in which temperature changes were less than the statistical error. Our instrumental resolution was not high enough to detect any visible splitting of the antiferromagnetic reflections— characteristic of incommensurate magnetic structure in the BiFeO3 -based phase. It is known that the period of the incommensurate structure is ∼ 620 Å [29] and it increases 4
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
(a)
Figure 6. Low-angle fragment of the neutron diffraction pattern for the sample with x = 0.3, which is a mixture of 60% of BiFeO3 -based phase and 40% of PbTiO3 -based phase, measured at 2 K. Inset—the magnetic structure of PbTiO3 -based phase.
(b)
Figure 5. Temperature dependences of the magnetic moment (black circles). In red—the moment deviation from the basal plane for samples with x = 0.20 (a); x = 0.24 (b). The refined content of phases is shown in the panels. The solid line is a fit with a power law, which one should consider as a guide for the eye.
that magnetic moments cannot be aligned along the [1 1 1] axis. Indeed, the best fit corresponds to the model with the moments lying in a plane perpendicular to the [1 1 1] direction (inset in figure 6). In contrast with the single phase samples beyond the MPB, the extraction of the magnetic moments from the diffraction data for the two-phase samples within the MPB is a more delicate task because of the overlapping of the magnetic peaks of the different phases. In such cases, the magnetic moments are calculated from the decomposition of the low angle fragment of the neutron diffraction pattern, fixing some parameters refined from full profile analysis, in particular a lattice parameter, which gives a more reliable result (figure 6). The temperature dependences of the moments in the two coexisting phases for the sample with x = 0.30, within the MPB, which is a mixture of 60% rhombohedral and 40% tetragonal phases, are shown in figure 7. The beginning of the spin re-orientation in the BiFeO3 -based phase, when the magnetic moments start to tilt towards the hexagonal axis, is surprisingly close to the N´eel temperature of the PbTiO3 -based phase. Obviously these transitions are coupled to each other. Indeed, in contrast to the sample with x = 0.24, where no trace of spin re-orientation at the temperature dependence of the unit cell parameters was detected, in the sample within the MPB composition with x = 0.3 the corresponding peculiarities are clearly seen in both constituents (figure 8).
Figure 7. Temperature dependences of the total magnetic moment (in black), its deviation from the basal plane (in red) for BiFeO3 -based phase and the magnetic moment of PbTiO3 -based phase (in blue) for the sample with x = 0.3. The solid line is a fit with a power law, which one should consider as a guide for the eye.
The temperature dependence of the intensity of the magnetic reflection (1/2 1/2 1/2) for the sample with x = 0.5, on the right-side of the MPB, is shown in figure 9(a). In the temperature range of 40 K–220 K it can be well fitted by a straight line. The noticeable deviation from the straight line at the highest temperatures is explained by the contribution from overlapping magnetic reflections from the BiFeO3 -based phase. This phase has N´eel temperature of about 600 K, much higher than the N´eel temperature of PbTiO3 -based phase. In other words the magnetic order in the BiFeO3 -based phase still exists, when the magnetic order in the PbTiO3 -based phase disappears. Note that a similar ‘tail’ in the temperature dependence of magnetic moment had been reported for the tetragonal phase of (BiFeO3 )0.7 (PbTiO3 )0.3 with a 90% fraction of tetragonal phase, and it was tentatively attributed to the short range order [20]. 5
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
(a)
(a)
(b)
(b)
Figure 9. (a) Temperature dependence of the intensity of magnetic reflection (1/2 1/2 1/2) for the sample with x = 0.5. (b) Temperature dependence of susceptibility measured at 50 Oe after zero-field-cooling (ZFC) and field-cooling (FC) procedure. In inset—the magnetisation loop, measured at 2 K. The error bars do not exceed the symbol size.
Figure 8. The temperature dependences of the unit cell parameters in the BiFeO3 -based phase (a) and the PbTO3 -based phase (b) for the sample with x = 0.3. The error bars do not exceed the symbol size.
3.3. Ferromagnetic moment
It is interesting to note that the net ferromagnetic moment occurs at about 40 K where a maximum in the temperature dependence of the antiferromagnetic moment appears (figure 9(a)). This temperature is noticeably above the temperature of magnetic susceptibility anomaly of 23 K (figure 9(b)). Figure 10 shows a peculiarity in the temperature dependences of the unit cell parameters at 25 K, close to the maximum in the magnetic susceptibility. It seems that the emergence of the ferromagnetic moment is accompanied by a crystal structure distortion.Unfortunately, the errors of the refined oxygen positions appeared to be rather big, thus preventing the detection of any anomaly in the bond length/angles.
Magnetic measurements reveal net ferromagnetic moment only in the samples with x around 0.5, as shown by the magnetic hysteresis loop measurements [6]. Indeed, our magnetic measurements in the compound with x = 0.5 also demonstrate clear anomaly around 23 K (figure 9(b)) and hysteresis at low temperatures, indicating the existence of a net moment (inset in figure 9(b)). In neutron diffraction, the net ferromagnetic component caused by non-collinear structure gives contribution into the nuclear reflections. However, the refinement of the neutron diffraction patterns did not reveal any ferromagnetic moment within an accuracy of 0.5 µB in all samples studied. The temperature dependence of the integrated intensities of nuclear reflections does not demonstrate any anomalies, which could be thought as the emergence of a weak ferromagnetic moment. However, in the sample with x = 0.5, a decrease in the magnetic intensity of the reflection (1/2 1/2 1/2), which is proportional to the antiferromagnetic component of the total moment, was observed below about 40 K (figure 9(a)). Since the total magnetic moment must be preserved, the drop in the antiferromagnetic component means a spin canting, corroborates the occurrence of a spin canting and a small ferromagnetic component. It is worth noting, that in contrast to previous works [15, 33] in our experiments a ferromagnetic moment was detected reliably only for x = 0.5, but not in the whole system.
4. Discussion
Our neutron diffraction experiments demonstrated that the substitution of Ti4+ for Fe3+ leads to several apparent effects. First, it results in an expected decrease in the N´eel temperature, from 640 K for BiFeO3 down ∼610 K for BiFeO3 -based phase. The same was observed for the PbTiO3 -based phase; TN was estimated to be about 220–250 K, much lower than TN = 605 K reported for PbFeO3 [32]. Second, M¨ossbauer experiments on the sample (Bi0.8 Pb0.2 )FeO2.9 showed ferromagnetic order at room temperature [28]. In our BiFeO3 -based sample (Bi0.8 Pb0.2 ) 6
J. Phys.: Condens. Matter 27 (2015) 046004
I V Golosovsky et al
TN of the rhombohedral BiFeO3 -based phase. Therefore, at the transformation from the tetragonal to the rhombohedral phase under external pressure at corresponding temperature, the paramagnetic state of the system can change to an antiferromagnetic state. Such a transformation in our case should be expected for the sample with x = 0.5, which is paramagnetic at room temperature (see figure 9(a)), but not with x = 0.3–0.35. It follows from our experiments that the crystal structure and chemical formula of the BiFeO3 -based and PbTiO3 -based phases change slightly in the studied range of x = 0.2– 0.5 (figure 4). Nevertheless, these small variations result in remarkable changes in the magnetic properties. Thus, on the left-side of the MPB, the practically single phase samples with x = 0.20 and x = 0.24 have similar lattice parameters and temperature dependences of the magnetic moment. However, in the sample with x = 0.24, in contrast to the sample with x = 0.2, a new phenomenon—spin reorientation—appears (figures 5(a) and (b)). Most probably, the difference observed in the magnetic behavior is caused by small structure variations which are undetectable by neutron diffraction. Such changes may lead to different temperatures of ordering for spin components within the basal plane and along the hexagonal axis. The strong response to small variation in the structure means a high instability of the BiFeO3 -based phase left of the MPB. From the comparison of the N´eel temperatures for the single phase PbTiO3 -based sample with x = 0.5 and the twophase sample with x = 0.3, where the PbTiO3 -based phase is 40% only, it can be seen that the TN changes drastically: from ∼ 250 K to 550 K (see figures 7 and 9(a)). At the same time, profile refinement demonstrates that the unit cell parameters in these samples vary slightly: a changes from 3.860(5) Å to 3.844(5) Å, while b changes from 4.325(5) Å to 4.375(5) Å. We believe that such a large TN increase can be explained by the proximity effect only. The 60% of the rhombohedral BiFeO3 -based phase with a high N´eel temperature, biases the 40% of the tetragonal PbTiO3 -based phase with a low N´eel temperature. Another evidence for strong magnetic coupling between the constituting phases follows from the closeness of the temperature of the spin re-orientation transition in the BiFeO3 based phase and the magnetic ordering temperature in the PbTiO3 -based phase (figure 7). We argue that this is not a coincidence because anomalies in the temperature dependences of the unit cell parameters in both constituents (figure 8) exist at this particular temperature. Perhaps this manifests a simple fact, that the interaction is maximal at a specific mutual arrangement of the moments of the constituent phases. The observed proximity effects are not surprising, given the nature of the nanoscale phases, which involve a welldeveloped interface between nanoparticles with different magnetic orders. It is worth noting that one should expect the proximity effects, namely, the exchange bias in all systems where nanoscale regions with different magnetic orders coexist. For example, such effects were observed in phase-separated manganites and cobaltites (see work [36] and the references therein). In these compounds
(a)
(b)
Figure 10. Temperature dependence of the unit cell parameters (a)—(panel a) and (b)—(panel b) in the tetragonal phase of the sample with x = 0.5. The error bars do not exceed the symbol size.
(Fe0.8 Ti0.2 )O3 , which differs by the presence of Ti ions, net ferromagnetic moment was not detected. This means that the substitution of 20% of Ti4+ for Fe3+ suppresses any spin canting, i.e. the appearance of a net ferromagnetic moment. Third, the substitution of Ti4+ for Fe3+ stabilizes the tetragonal structure to the right of the MPB. It is worth noting, that the unstable oxide PbFeO3 , which can be synthesized under 7 GPa and at 1100–1200 ◦ C only, has an orthorhombic structure [34]. Analysis of the results demonstrates clear discrepancies in terms of structure and phase instability of the BF–PT solid solution system. First, the structures reported by different laboratories strongly differ. For example, the compound (BiFeO3 )0.7 (PbTiO3 )0.3 was reported to have single phase tetragonal structure [13, 16, 27], but also a mixture of two phases: tetragonal and monoclinic [17]. In our case we see as well two phases: but however they have been identified as tetragonal and rhombohedral ones. The same holds true for the width and the position of the MPB. The data on magnetic behaviour are partially contradictory as well. For example, in our study the spin reorientation occurs at the temperature 550 K, while in the work of Bhattacharjee et al [33] the spin re-orientation transition was observed around 460 K. It was recently reported that at room temperature on application of a moderate pressure (
