Keith J.
Cordrey
a,
Magda
Stanczyk
a,
Charlotte A. L.
Dixon
a,
Kevin S.
Knight
b,
Jonathan
Gardner
a,
Finlay D.
Morrison
a and
Philip
Lightfoot
*a
aEaStCHEM and School of Chemistry, University of St Andrews, St Andrews, Fife KY16 9ST, UK. E-mail: pl@st-and.ac.uk
bISIS Facility, STFC Rutherford Appleton Laboratory, Harwell, Oxfordshire OX11 0QX, UK
First published on 19th January 2015
The layered perovskite LaTaO4 has been prepared in its polar orthorhombic polymorphic form at ambient temperature. Although no structural phase transition is observed in the temperature interval 25° C < T < 500 °C, a very large axial thermal contraction effect is seen, which can be ascribed to an anomalous buckling of the perovskite octahedral layer. The non-polar monoclinic polymorph can be stabilised at ambient temperature by Nd-doping. A composition La0.90Nd0.10TaO4 shows a first-order monoclinic-orthorhombic (non-polar to polar) transition in the region 250° C < T < 350 °C. Dielectric responses are observed at both the above structural events but, despite the ‘topological ferroelectric’ nature of orthorhombic LaTaO4, we have not succeeded in obtaining ferroelectric P–E hysteresis behaviour. Structural relationships in the wider family of AnBnX3n+2 layered perovskites are discussed.
In this paper we explore some of the crystal chemistry of another member of the AnBnX3n+2 family, viz. LaTaO4. This compound can exist in two polymorphic forms when prepared by ‘traditional’ high-temperature solid-state routes9–12 and a third polymorph can also be prepared via decomposition of a hydrated precursor.13 Of the former two phases, one is orthorhombic and polar and the other monoclinic and centrosymmetric (hereafter designated O-LaTaO4 and M-LaTaO4, respectively, Fig. 1). Earlier studies9–12 suggest M-LaTaO4 is the most stable form at ambient temperature, and that this phase undergoes a phase transition to O-LaTaO4 at ∼175 °C9 or 240 °C.10 In our study we have successfully prepared a near-pure sample of O-LaTaO4 and we study the structural behaviour of this system versus temperature using high-resolution powder neutron diffraction. We also show that the M-LaTaO4 phase can be stabilised further at ambient temperature by Nd-doping, and study the thermal evolution of this phase, and its transformation into the polar O-LaTaO4 polymorph. Finally, we carry out some general structural comparisons of these phases and related ones, which may be of relevance to their physical properties.
Variable temperature neutron powder diffraction (NPD) data on LaTaO4 and La0.9Nd0.1TaO4 were collected on the high-resolution instrument (HRPD) at ISIS. For LaTaO4 data were collected at ambient temperature and between 100 °C and 500 °C at 50 °C intervals; for La0.9Nd0.1TaO4 date were collected between 100 °C and 600 °C at 50 °C intervals. In each case, data collection times were approximately 40 minutes for ∼5 g samples. For LaTaO4, a preliminary dataset was collected at 25 °C on the GEM diffractometer.
All quantitative data analysis was carried out by Rietveld refinement using the GSAS program14 and its EXPGUI user interface. For each of the HRPD runs, detector banks at 2θ ∼ 168° and 90° were analysed simultaneously.
Fig. 2 Rietveld fit (NPD) for single phase O-LaTaO4 model at 100 °C. Overall χ2 = 6.67 for 37 refined variables. |
Fig. 3 Thermal evolution of lattice parameters and unit cell volume for O-LaTaO4, derived from Rietveld refinement of NPD data. |
The crystal structure (Fig. 1(b), 4) consists of perovskite-like blocks of corner-shared octahedra extending in the ac plane, separated by the La3+ cations, which may be regarded as nine-coordinated to oxide. In order to pinpoint the structural features leading to the anomalous expansivity behaviour, we can define three parameters within these two distinct blocks: the thickness of the inter-layer block, d1, the thickness of the perovskite block, d2, and the angle of tilt within the corrugated octahedral layers, ω. From the geometry of these definitions it can be seen that
b = 2(d1 + d2) |
Fig. 4 View of adjacent octahedral blocks in the O-LaTaO4 structure, with the definitions of the parameters, d1, d2 and ω shown. |
Fig. 5 Thermal evolution of the distortion parameters (defined in Fig. 4) for O-LaTaO4, (a) d1, (b) d2, (c) ω. |
A recent variable temperature (90–350 K) structural study of some higher n members of this family, Lan(Ti,Fe)nO3n+2 (n = 5 and 6)15 showed that in those cases the thermal variation in the interlayer distance parameter is also insignificant, whereas the thickness of the perovskite block showed a marked increase, with no observable anomalies.
For values of 0.05 ≤ x ≤ 0.40, preliminary analysis of XPD data suggested that the M-LaTaO4 could be prepared almost phase-pure. For x = 0.50 and 0.60 a significant additional phase (fergusonite type LnTaO4) was observed. Rietveld analysis of this series confirmed the extent of the M-LaTaO4 solid solution in La1−xNdxTaO4 as 0.05 ≤ x ≤ 0.40. Lattice parameters across the solid solution vary monotonically with x; these were determined by using a model with fixed atomic coordinates. An example Rietveld plot is given in Fig. 6, and further details are given in the ESI.†
A sample of M-La0.9Nd0.1TaO4 was studied versus temperature by NPD, up to 600 °C. As anticipated, a phase transition from the M- to the O-phase was observed within the temperature range studied. This transition is of a first-order nature, as shown by the co-existence of both phases with the region 250 < T < 350 °C (M/O phase fractions (%) at 250 °C, 300 °C and 350 °C are 90/10, 59/41 and 24/76, respectively). In fact, there is no simple group-subgroup relationship between the two crystal structures, so the transition is expected to be 1st order according to Landau theory. Rietveld fits are shown in Fig. 7(a) and (b) for single-phase fits to the M- and O-models at 100 and 500 °C, respectively. Attempts to refine the high-temperature phase in the centrosymmetric parent phase, space group Cmcm (Fig. 1(a)), led to significantly poorer fits and unrealistic atomic displacement parameters (Cmc21 model: 37 variables, χ2 = 3.26, Cmcm model: 30 variables, χ2 = 13.5). A similar contraction of the b-axis, as observed for the O-LaTaO4 phase, is seen here in the region 250 < T < 450 °C. Further details are given in ESI.†
Fig. 8 Dielectric data (obtained at 1 MHz) as a function of temperature for LaTaO4 (a) and La0.90Nd0.10TaO4 (b). |
Polarisation-field (P–E) measurements were carried out on O-LaTaO4 in an attempt to confirm the ferroelectric nature of this polymorph. P–E data obtained at room temperature under an applied field of 100 kV cm−1 at 1 kHz showed a linear dielectric response and no evidence of ferroelectric switching (see ESI†). The coercive field for isostructural BaMgF4 is known to increase dramatically with frequency,19,20 suggesting a high nucleation energy for switching. In order to investigate this, measurements were repeated at lower frequencies (down to 0.1 Hz). Unfortunately, however, under these conditions the samples became increasingly leaky (ESI†) and so no conclusive evidence for ferroelectric switching was obtained. The polar (orthorhombic) phase of La0.9Nd0.1TaO4 is only stable at temperatures above our measuring capabilities.
Using first-principles calculations, Ederer and Spaldin6 suggested that the topological ferroelectricity observed in the BaMF4 family was driven by a single polar soft-mode, essentially a ‘rigid’ octahedral tilt. However, employing similar methods, López-Pérez and Íñiguez7 found that in the n = 4 member of this layered perovskite family, La2Ti2O7, the cooperative interaction of two distinct soft-modes is fundamentally important. They suggest that although the tilt mode is the driver for polarity, coupling with an octahedral distortion mode is significant, and is actually the majority contributor to the overall calculated polarisation. This may suggest that the conventional ‘d0-ness’ criterion of ferroelectric activity does actually play a part in the oxide members of this family but not so much in the fluorides. In order to probe these effects further we can compare the degree of distortion within the octahedral units of several of these phases. Firstly, from the present work we can analyse the distortions present in the O-phase of LaTaO4 over a range of temperatures, and compare this to the O- and M-phases in La0.9Nd0.1TaO4. We can define simple ‘distortion indices’ for the octahedra, based on deviations of the six bond lengths from the mean, and the twelve bond angles from 90°:
Δ1 = 1/6∑|Rav − Ri| |
Δ2 = 1/12∑|90 − ϕi| |
These are presented in Table 1, and compared to selected examples from the BaMF4 series, based on well-determined single crystal X-ray studies.22,23 Also included for comparison are values for the related n = 4 phases La2Ti2O7 (theoretical study7) and Sr2Nb2O7 (single crystal X-ray study24), and for BiReO4 and NaAlF4, which both adopt the parent Cmcm structure for the n = 2 series.25,26 In the case of the n = 4 phases there are two distinct B sites (Fig. 9), representing ‘inner two’ and ‘outer two’ octahedra of each perovskite block. It has previously been noted8 that the distortion of the outer octahedra are greater than those of the inner ones; this is a natural consequence of the weaker inter-block versus intra-block bonding (i.e., the inner octahedra are directly linked to six neighbouring octahedra, whereas the outer ones are linked to four).
Fig. 9 Nature of the octahedral distortions in selected members of the AnBnX3n+2 family: (a) O-LaTaO4 at 100 °C (present study) (b) M-NaCrF4 at 25 °C (ref. 23) (c) O-Sr2Nb2O7 at 25 °C (ref. 24). Note that the latter contains two distinct octahedral sites. |
Phase | Δ1 (Å) | Δ2 (°) | Distortion type |
---|---|---|---|
O-LaTaO4 (100 °C) | 0.044 | 5.9 | Edge |
O-LaTaO4 (500 °C) | 0.073 | 5.0 | Edge |
M-La0.9Nd0.1TaO4 (100 °C) | 0.039 | 4.8 | Indistinct |
O-La0.9Nd0.1TaO4 (500 °C) | 0.053 | 5.1 | Edge |
O-BaMnF4 (25 °C)22 | 0.024 | 4.3 | Indistinct |
O-BaCoF4 (25 °C)22 | 0.047 | 3.3 | Axial |
O-BaNiF4 (25 °C)22 | 0.055 | 3.1 | Axial |
M-NaCrF4 (25 °C)23 | 0.031 | 2.0 | Edge |
NaAlF4 (25 °C)25 | 0.018 | 1.3 | Edge |
BiReO4 (25 °C)26 | 0.056 | 3.2 | Edge |
O-La2Ti2O7 (calc)7 | 0.088/0.071 | 6.2/5.6 | Apex |
O-Sr2Nb2O7 (25 °C)24 | 0.125/0.084 | 7.3/5.6 | Apex |
As a further comparison to these values, the corresponding Δ1 and Δ2 values for the tetragonal ferroelectric phase of BaTiO3 at room temperature are 0.067 Å and 2.5°, respectively.27
It can be immediately seen that all the examples given do have significant octahedral distortions. Moreover, these distortions have a significant component along the polar c-axis in the O-phases, and hence will contribute to any net polarisation. From the relatively limited data available it can also be suggested that (i) the oxide members of the n = 2 series generally have larger distortions than the fluorides, (ii) the n = 4 members of the series have enhanced octahedral distortions relative to the n = 2 members, even in the case of the ‘inner’ octahedral layers. However, the representation of the octahedral distortions in terms of these two numerical parameters does not necessarily tell the full story; the nature of the distortion differs, as represented in Fig. 9. In general, the B cation in a perovskite can be considered to be displaced towards an octahedral apex, edge or face (compare the tetragonal, orthorhombic and rhombohedral ferroelectric phases of BaTiO3, for example). Also shown in Table 1, and illustrated for selected examples in Fig. 9, is a rough classification of the different distortion types present in this family. As can be seen, for the fluorides there is no clear trend to the distortion types, whereas for the oxides there are two distinct types of displacement: edge-oriented for the n = 2 O-phases and apex-oriented for the n = 4 O-phases. Due to the limited data available, it is dangerous to draw any global conclusion from this, but it is clear that the d0 cations do display the types of distortion commonly seen in oxides; whether there is an inherent preference for a particular cation to show a particular type of distortion is unclear. Such effects have been discussed by Halasyamani and co-workers,28 who suggest that Ti4+, Nb5+ and Ta5+ show roughly similar preferences for edge- and apex-distortions. Therefore it is likely that the specific details of inter-block versus intra-block bonding and the nature of the A cation in these complex, highly distorted structures is also significant.
We suggest that, although rigid octahedral tilting in the O-phases intrinsically provides a polarisation, in reality the octahedra are far from rigid, and the complex nature and differences in these deviations from ideality are not necessarily straightforward to predict, depending on a subtle difference in compatibility of the perovskite blocks and separating layers, together with the size and electronic nature of the B cations.
Further, we have shown that the M-LaTaO4 polymorph can be stabilised to room temperature by a small amount of Nd doping at the La site: a consequence of the smaller A-cation size. In the series La0.9Nd0.1TaO4 we have shown that the stability range of the M-phase at ambient temperature extends to x ∼ 0.50. For the x = 0.1 case, we have shown that a first-order phase transition from the M- to the O-phase occurs in the region 250–350 °C. It is anticipated that the temperature of this phase transition will increase markedly for higher x-values, and further work on this is prompted. The dielectric data for x = 0.1 also suggests a further structural feature to be present at ca. 400 °C, and the intriguing possibility of antiferroelectric ordering; this also requires further study.
By comparing the present crystallographic data with selected previous examples of this structural family, we conclude that significant octahedral distortions are prevalent in all cases. This supports the previous theoretical studies, which suggest that although rigid octahedral tilting drives the transition into the polar O-phase, octahedral distortions are key in enhancing and stabilising the polarisation in this topological ferroelectric family. Although some broad general comments can be made on the nature and magnitude of these distortions, there is insufficient high quality structural data available to clarify specific ‘design’ principles, for the optimisation of net polarisation. The contributions of octahedral tilting versus octahedral distortion should therefore be considered on a case-by-case basis, and further systematic structural studies, on a more diverse range of compositions and structural variants, need to be carried out.
The transformation of an ambient temperature centrosymmetric phase into a higher temperature polar phase is highly unusual; indeed we are aware of no other examples of this amongst oxide ferroelectrics. It is anticipated that both the M- and O-polymorphs within this family will transform to an aristotype centrosymmetric phase (space group Cmcm) at high temperature. Although examples of the aristotype phase have been structurally characterised (e.g., BiReO4 and Sr2Ta2O7, which transforms to the polar O-phase at −107 °C), the nature of this transition has not been studied in detail previously. This would provide a very interesting study, in order to probe the relative importance of the octahedral tilt and distortion modes near TC.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4dt03721a |
This journal is © The Royal Society of Chemistry 2015 |